From 3b14459612d078f64872f480cd05f1eb92d70101 Mon Sep 17 00:00:00 2001
From: gmillot <gael.millot@pasteur.fr>
Date: Fri, 3 Feb 2023 20:45:14 +0100
Subject: [PATCH] release v11.6: bug fixed in fun_comp_2d, since
identical(data1, data2) returns FALSE if both are identical but data1 is
typeof double and data2 is typeof integer
---
README.md | 5 +
check/TODO.txt | 3 +-
cute_little_R_functions.R | 26643 +++++++++++++++++----------------
cute_little_R_functions.docx | Bin 499127 -> 499498 bytes
fun_gg_boxplot.docx | Bin 115871 -> 116015 bytes
fun_gg_scatter.docx | Bin 121152 -> 121237 bytes
6 files changed, 13334 insertions(+), 13317 deletions(-)
diff --git a/README.md b/README.md
index ee7618e..f934bd2 100755
--- a/README.md
+++ b/README.md
@@ -170,6 +170,11 @@ Gitlab developers
## WHAT'S NEW IN
+### v11.6.0
+
+1) bug fixed in fun_comp_2d, since identical(data1, data2) returns FALSE if both are identical but data1 is typeof double and data2 is typeof integer
+
+
### v11.5.0
1) modification of ggplot2::guides() in fun_gg_boxplot() and fun_gg_scatter() to do not have the warning message about FALSE replaced by "none"
diff --git a/check/TODO.txt b/check/TODO.txt
index ce76e92..68e93d7 100644
--- a/check/TODO.txt
+++ b/check/TODO.txt
@@ -84,7 +84,8 @@ Argument and return : leave it to me
For packages :
See mails with cuteathon
-
+For speed, use rbenchmark :
+https://towardsdatascience.com/r-is-slow-and-its-your-fault-2fcedacc7abb
diff --git a/cute_little_R_functions.R b/cute_little_R_functions.R
index 35485dd..0ca1133 100755
--- a/cute_little_R_functions.R
+++ b/cute_little_R_functions.R
@@ -20,7 +20,7 @@
# https://stackoverflow.com/questions/17552917/merging-existing-pdf-files-using-r
# https://www.r-bloggers.com/2019/04/join-split-and-compress-pdf-files-with-pdftools/
# https://rdrr.io/cran/staplr/man/staple_pdf.html
-
+# For speed, use rbenchmark : https://towardsdatascience.com/r-is-slow-and-its-your-fault-2fcedacc7abb
@@ -98,343 +98,343 @@
# -> transferred into the cute package
# Do not modify this function in cute_little_R_function anymore. See the cute repo
fun_check <- function(
-data,
-class = NULL,
-typeof = NULL,
-mode = NULL,
-length = NULL,
-prop = FALSE,
-double.as.integer.allowed = FALSE,
-options = NULL,
-all.options.in.data = FALSE,
-na.contain = FALSE,
-neg.values = TRUE,
-print = FALSE,
-data.name = NULL,
-fun.name = NULL
+ data,
+ class = NULL,
+ typeof = NULL,
+ mode = NULL,
+ length = NULL,
+ prop = FALSE,
+ double.as.integer.allowed = FALSE,
+ options = NULL,
+ all.options.in.data = FALSE,
+ na.contain = FALSE,
+ neg.values = TRUE,
+ print = FALSE,
+ data.name = NULL,
+ fun.name = NULL
){
-# AIM
-# Check the class, type, mode and length of the data argument
-# Mainly used to check the arguments of other functions
-# Check also other kind of data parameters, is it a proportion? Is it type double but numbers without decimal part?
-# If options == NULL, then at least class or type or mode or length argument must be non-null
-# If options is non-null, then class, type and mode must be NULL, and length can be NULL or specified
-# WARNINGS
-# The function tests what is written in its arguments, even if what is written is incoherent. For instance, fun_check(data = factor(1), class = "factor", mode = "character") will return a problem, whatever the object tested in the data argument, because no object can be class "factor" and mode "character" (factors are class "factor" and mode "numeric"). Of note, length of object of class "environment" is always 0
-# If the tested object is NULL, then the function will always return a checking problem
-# Since R >= 4.0.0, class(matrix()) returns "matrix" "array", and not "matrix" alone as before. However, use argument class = "matrix" to check for matrix object (of class "matrix" "array" in R >= 4.0.0) and use argument class = "array" to check for array object (of class "array" in R >= 4.0.0)
-# ARGUMENTS
-# data: object to test
-# class: character string. Either one of the class() result (But see the warning section above) or "vector" or "ggplot2" (i.e., objects of class c("gg", "ggplot")) or NULL
-# typeof: character string. Either one of the typeof() result or NULL
-# mode: character string. Either one of the mode() result (for non-vector object) or NULL
-# length: numeric value indicating the length of the object. Not considered if NULL
-# prop: logical. Are the numeric values between 0 and 1 (proportion)? If TRUE, can be used alone, without considering class, etc.
-# double.as.integer.allowed: logical. If TRUE, no error is reported in the cheking message if argument is set to typeof == "integer" or class == "integer", while the reality is typeof == "double" or class == "numeric" but the numbers strictly have zero as modulo (remainder of a division). This means that i <- 1, which is typeof(i) -> "double" is considered as integer with double.as.integer.allowed = TRUE. WARNING: data%%1 == 0L but not isTRUE(all.equal(data%%1, 0)) is used here because the argument checks for integers stored as double (does not check for decimal numbers that are approximate integers)
-# options: a vector of character strings or integers indicating all the possible option values for the data argument, or NULL. Numbers of type "double" are accepted if they have a 0 modulo
-# all.options.in.data: logical. If TRUE, all of the options must be present at least once in the data argument, and nothing else. If FALSE, some or all of the options must be present in the data argument, and nothing else. Ignored if options is NULL
-# na.contain: logical. Can the data argument contain NA?
-# neg.values: logical. Are negative numeric values authorized? Warning: the default setting is TRUE, meaning that, in that case, no check is performed for the presence of negative values. The neg.values argument is activated only when set to FALSE. In addition, (1) neg.values = FALSE can only be used when class, typeof or mode arguments are not NULL, otherwise return an error message, (2) the presence of negative values is not checked with neg.values = FALSE if the tested object is a factor and the following checking message is returned "OBJECT MUST BE MADE OF NON NEGATIVE VALUES BUT IS A FACTOR"
-# print: logical. Print the message if $problem is TRUE? Warning: set by default to FALSE, which facilitates the control of the checking message output when using fun_check() inside functions. See the example section
-# data.name: character string indicating the name of the object to test. If NULL, use what is assigned to the data argument for the returned message
-# fun.name: character string indicating the name of the function checked (i.e., when fun_check() is used to check the arguments of this function). If non-null, the value of fun.name will be added into the message returned by fun_check()
-# RETURN
-# A list containing:
-# $problem: logical. Is there any problem detected?
-# $text: message indicating the details of the problem, or the absence of problem
-# $object.name: value of the data.name argument (i.e., name of the checked object if provided, NULL otherwise)
-# REQUIRED PACKAGES
-# None
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# None
-# EXAMPLE
-# test <- matrix(1:3) ; fun_check(data = test, print = TRUE, class = "vector", mode = "numeric")
-# see http
-# DEBUGGING
-# data = mean ; class = NULL ; typeof = NULL ; mode = NULL ; length = NULL ; prop = FALSE ; double.as.integer.allowed = FALSE ; options = "a" ; all.options.in.data = FALSE ; na.contain = FALSE ; neg.values = TRUE ; print = TRUE ; data.name = NULL ; fun.name = NULL
-# function name
-# no used in this function for the error message, to avoid env colliding
-# end function name
-# required function checking
-# end required function checking
-# reserved words
-# end reserved words
-# fun.name checked first because required next
-if( ! is.null(fun.name)){ # I have to use this way to deal with every kind of class for fun.name
-if(all(base::class(fun.name) == "character")){ # all() without na.rm -> ok because class(NA) is "logical"
-if(base::length(fun.name) != 1){
-tempo.cat <- paste0("ERROR IN fun_check(): THE fun.name ARGUMENT MUST BE A CHARACTER VECTOR OF LENGTH 1: ", paste(fun.name, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(any(is.na(fun.name))){ # normally no NA with is.na()
-tempo.cat <- paste0("ERROR IN fun_check(): NO ARGUMENT EXCEPT data AND options CAN HAVE NA VALUES\nPROBLEMATIC ARGUMENT IS fun.name")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-tempo.cat <- paste0("ERROR IN fun_check(): THE fun.name ARGUMENT MUST BE A CHARACTER VECTOR OF LENGTH 1") # paste(fun.name, collapse = " ") removed here because does not work with objects like function
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end fun.name checked first because required next
-# arg with no default values
-mandat.args <- c(
-"data"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
-# end argument primary checking
-# second round of checking and data preparation
-# management of special classes
-basic.class <- c(
-"NULL", # because class(NULL) is "NULL". The NULL aspect will be dealt later
-"logical",
-"integer",
-"numeric",
-# "complex",
-"character"
-# "matrix",
-# "array",
-# "data.frame",
-# "list",
-# "factor",
-# "table",
-# "expression",
-# "name",
-# "symbol",
-# "function",
-# "uneval",
-# "environment",
-# "ggplot2",
-# "ggplot_built",
-# "call"
-)
-tempo.arg.base <-c( # no names(formals(fun = sys.function(sys.parent(n = 2)))) used with fun_check() to be sure to deal with the correct environment
-"class",
-"typeof",
-"mode",
-"length",
-"prop",
-"double.as.integer.allowed",
-"options",
-"all.options.in.data",
-"na.contain",
-"neg.values",
-"print",
-"data.name",
-"fun.name"
-)
-tempo.class <-list( # no get() used to be sure to deal with the correct environment
-base::class(class),
-base::class(typeof),
-base::class(mode),
-base::class(length),
-base::class(prop),
-base::class(double.as.integer.allowed),
-base::class(options),
-base::class(all.options.in.data),
-base::class(na.contain),
-base::class(neg.values),
-base::class(print),
-base::class(data.name),
-base::class(fun.name)
-)
-tempo <- ! sapply(lapply(tempo.class, FUN = "%in%", basic.class), FUN = all)
-if(any(tempo)){
-tempo.cat1 <- tempo.arg.base[tempo]
-tempo.cat2 <- sapply(tempo.class[tempo], FUN = paste0, collapse = " ")
-tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": ANY ARGUMENT EXCEPT data MUST HAVE A BASIC CLASS\nPROBLEMATIC ARGUMENT", ifelse(base::length(tempo.cat1) > 1, "S", ""), " AND ASSOCIATED CLASS", ifelse(base::length(tempo.cat1) > 1, "ES ARE", " IS"), ":\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n")) # normally no NA with is.na()
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of special classes
-# management of NA arguments
-if(any(is.na(data.name)) | any(is.na(class)) | any(is.na(typeof)) | any(is.na(mode)) | any(is.na(length)) | any(is.na(prop)) | any(is.na(double.as.integer.allowed)) | any(is.na(all.options.in.data)) | any(is.na(na.contain)) | any(is.na(neg.values)) | any(is.na(print)) | any(is.na(fun.name))){ # normally no NA with is.na()
-tempo <- c("data.name", "class", "typeof", "mode", "length", "prop", "double.as.integer.allowed", "all.options.in.data", "na.contain", "neg.values", "print", "fun.name")[c(any(is.na(data.name)), any(is.na(class)), any(is.na(typeof)), any(is.na(mode)), any(is.na(length)), any(is.na(prop)), any(is.na(double.as.integer.allowed)), any(is.na(all.options.in.data)), any(is.na(na.contain)), any(is.na(neg.values)), any(is.na(print)), any(is.na(fun.name)))]
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": NO ARGUMENT EXCEPT data AND options CAN HAVE NA VALUES\nPROBLEMATIC ARGUMENT", ifelse(length(tempo) > 1, "S ARE", " IS"), ":\n", paste(tempo, collapse = "\n")) # normally no NA with is.na()
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"prop",
-"double.as.integer.allowed",
-"all.options.in.data",
-"na.contain",
-"neg.values",
-"print"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){ # normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN fun.check():\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT BE NULL:\n", paste0(tempo.arg[tempo.log], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# dealing with logical
-# tested below
-# end dealing with logical
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-# end warning initiation
-# other checkings
-if( ! is.null(data.name)){
-if( ! (base::length(data.name) == 1L & all(base::class(data.name) == "character"))){ # all() without na.rm -> ok because class(NA) is "logical"
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": data.name ARGUMENT MUST BE A SINGLE CHARACTER ELEMENT AND NOT ", paste(data.name, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(is.null(options) & is.null(class) & is.null(typeof) & is.null(mode) & prop == FALSE & is.null(length)){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": AT LEAST ONE OF THE options, class, typeof, mode, prop, OR length ARGUMENT MUST BE SPECIFIED (I.E, TRUE FOR prop)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(options) & ( ! is.null(class) | ! is.null(typeof) | ! is.null(mode) | prop == TRUE)){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE class, typeof, mode ARGUMENTS MUST BE NULL, AND prop FALSE, IF THE options ARGUMENT IS SPECIFIED\nTHE options ARGUMENT MUST BE NULL IF THE class AND/OR typeof AND/OR mode AND/OR prop ARGUMENT IS SPECIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (all(base::class(neg.values) == "logical") & base::length(neg.values) == 1L)){ # all() without na.rm -> ok because class(NA) is "logical"
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE neg.values ARGUMENT MUST BE TRUE OR FALSE ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(neg.values == FALSE & is.null(class) & is.null(typeof) & is.null(mode)){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE neg.values ARGUMENT CANNOT BE SWITCHED TO FALSE IF class, typeof AND mode ARGUMENTS ARE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(class)){ # may add "formula" and "Date" as in https://renenyffenegger.ch/notes/development/languages/R/functions/class
-if( ! all(class %in% c("vector", "logical", "integer", "numeric", "complex", "character", "matrix", "array", "data.frame", "list", "factor", "table", "expression", "name", "symbol", "function", "uneval", "environment", "ggplot2", "ggplot_built", "call") & base::length(class) == 1L)){ # length == 1L here because of class(matrix()) since R4.0.0 # all() without na.rm -> ok because class cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT MUST BE ONE OF THESE VALUE:\n\"vector\", \"logical\", \"integer\", \"numeric\", \"complex\", \"character\", \"matrix\", \"array\", \"data.frame\", \"list\", \"factor\", \"table\", \"expression\", \"name\", \"symbol\", \"function\", \"environment\", \"ggplot2\", \"ggplot_built\", \"call\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(neg.values == FALSE & ! any(class %in% c("vector", "numeric", "integer", "matrix", "array", "data.frame", "table"))){ # no need of na.rm = TRUE for any() because %in% does not output NA
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT CANNOT BE OTHER THAN \"vector\", \"numeric\", \"integer\", \"matrix\", \"array\", \"data.frame\", \"table\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(typeof)){ # all the types are here: https://renenyffenegger.ch/notes/development/languages/R/functions/typeof
-if( ! (all(typeof %in% c("logical", "integer", "double", "complex", "character", "list", "expression", "symbol", "closure", "special", "builtin", "environment", "S4", "language")) & base::length(typeof) == 1L)){ # "language" is the type of object of class "call" # all() without na.rm -> ok because typeof cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT MUST BE ONE OF THESE VALUE:\n\"logical\", \"integer\", \"double\", \"complex\", \"character\", \"list\", \"expression\", \"name\", \"symbol\", \"closure\", \"special\", \"builtin\", \"environment\", \"S4\", \"language\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(neg.values == FALSE & ! typeof %in% c("double", "integer")){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT CANNOT BE OTHER THAN \"double\" OR \"integer\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(mode)){ # all the types are here: https://renenyffenegger.ch/notes/development/languages/R/functions/typeof
-if( ! (all(mode %in% c("logical", "numeric", "complex", "character", "list", "expression", "name", "symbol", "function", "environment", "S4", "call")) & base::length(mode) == 1L)){ # all() without na.rm -> ok because mode cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT MUST BE ONE OF THESE VALUE:\n\"logical\", \"numeric\", \"complex\", \"character\", \"list\", \"expression\", \"name\", \"symbol\", \"function\", \"environment\", \"S4\", \"call\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(neg.values == FALSE & mode != "numeric"){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT CANNOT BE OTHER THAN \"numeric\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(length)){
-if( ! (is.numeric(length) & base::length(length) == 1L & all( ! grepl(length, pattern = "\\.")))){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": length ARGUMENT MUST BE A SINGLE INTEGER VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! (is.logical(prop) & base::length(prop) == 1L)){ # is.na() already checked for prop
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": prop ARGUMENT MUST BE TRUE OR FALSE ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(prop == TRUE){
-if( ! is.null(class)){
-if( ! any(class %in% c("vector", "numeric", "matrix", "array", "data.frame", "table"))){ # no need of na.rm = TRUE for any() because %in% does not output NA
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT CANNOT BE OTHER THAN NULL, \"vector\", \"numeric\", \"matrix\", \"array\", \"data.frame\", \"table\" IF prop ARGUMENT IS TRUE") # not integer because prop
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(mode)){
-if(mode != "numeric"){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT CANNOT BE OTHER THAN NULL OR \"numeric\" IF prop ARGUMENT IS TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(typeof)){
-if(typeof != "double"){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT CANNOT BE OTHER THAN NULL OR \"double\" IF prop ARGUMENT IS TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}
-if( ! (all(base::class(double.as.integer.allowed) == "logical") & base::length(double.as.integer.allowed) == 1L)){ # all() without na.rm -> ok because class() never returns NA
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE double.as.integer.allowed ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(double.as.integer.allowed, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (is.logical(all.options.in.data) & base::length(all.options.in.data) == 1L)){
-tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": all.options.in.data ARGUMENT MUST BE A SINGLE LOGICAL VALUE (TRUE OR FALSE ONLY): ", paste(all.options.in.data, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (all(base::class(na.contain) == "logical") & base::length(na.contain) == 1L)){ # all() without na.rm -> ok because class() never returns NA
-tempo.cat <- paste0("ERROR IN fun_check(): THE na.contain ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(na.contain, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (all(base::class(print) == "logical") & base::length(print) == 1L)){ # all() without na.rm -> ok because class() never returns NA
-tempo.cat <- paste0("ERROR IN fun_check(): THE print ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(print, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# data.name and fun.name tested at the beginning
-# end other checkings
-# end second round of checking and data preparation
-# package checking
-# end package checking
-# main code
-if(is.null(data.name)){
-data.name <- deparse(substitute(data))
-}
-problem <- FALSE
-text <- paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT")
-if(( ! is.null(options)) & (all(base::typeof(data) == "character") | all(base::typeof(data) == "integer") | all(base::typeof(data) == "double"))){ # all() without na.rm -> ok because typeof() never returns NA
-if(all(base::typeof(data) == "double")){
-if( ! all(data %% 1 == 0L, na.rm = TRUE)){
-problem <- TRUE
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nBUT IS NOT EVEN TYPE CHARACTER OR INTEGER")
-}
-}else{
-text <- ""
-if( ! all(data %in% options)){ # no need of na.rm = TRUE for all() because %in% does not output NA
-problem <- TRUE
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nTHE PROBLEMATIC ELEMENTS OF ", data.name, " ARE: ", paste(unique(data[ ! (data %in% options)]), collapse = " "))
-}
-if(all.options.in.data == TRUE){
-if( ! all(options %in% data)){ # no need of na.rm = TRUE for all() because %in% does not output NA
-problem <- TRUE
-text <- paste0(ifelse(text == "", "", paste0(text, "\n")), ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE MADE OF ALL THESE OPTIONS: ", paste(options, collapse = " "), "\nTHE MISSING ELEMENTS OF THE options ARGUMENT ARE: ", paste(unique(options[ ! (options %in% data)]), collapse = " "))
-}
-}
-if( ! is.null(length)){
-if(base::length(data) != length){
-problem <- TRUE
-text <- paste0(ifelse(text == "", "", paste0(text, "\n")), ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE LENGTH OF ", data.name, " MUST BE ", length, " AND NOT ", base::length(data))
-}
-}
-if(text == ""){
-text <- paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT")
-}
-}
-}else if( ! is.null(options)){
-problem <- TRUE
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nBUT IS NOT EVEN TYPE CHARACTER OR INTEGER")
-}
-arg.names <- c("class", "typeof", "mode", "length")
-if( ! is.null(class)){
-if(class == "matrix"){ # because of class(matric()) since R4.0.0
-class <- c("matrix", "array")
-}else if(class == "factor" & all(base::class(data) %in% c("factor", "ordered"))){ # to deal with ordered factors # all() without na.rm -> ok because class(NA) is "logical"
-class <- c("factor", "ordered")
-}
-}
-if(is.null(options)){
-for(i2 in 1:base::length(arg.names)){
-if( ! is.null(get(arg.names[i2], env = sys.nframe(), inherit = FALSE))){
-# script to execute
-tempo.script <- '
+ # AIM
+ # Check the class, type, mode and length of the data argument
+ # Mainly used to check the arguments of other functions
+ # Check also other kind of data parameters, is it a proportion? Is it type double but numbers without decimal part?
+ # If options == NULL, then at least class or type or mode or length argument must be non-null
+ # If options is non-null, then class, type and mode must be NULL, and length can be NULL or specified
+ # WARNINGS
+ # The function tests what is written in its arguments, even if what is written is incoherent. For instance, fun_check(data = factor(1), class = "factor", mode = "character") will return a problem, whatever the object tested in the data argument, because no object can be class "factor" and mode "character" (factors are class "factor" and mode "numeric"). Of note, length of object of class "environment" is always 0
+ # If the tested object is NULL, then the function will always return a checking problem
+ # Since R >= 4.0.0, class(matrix()) returns "matrix" "array", and not "matrix" alone as before. However, use argument class = "matrix" to check for matrix object (of class "matrix" "array" in R >= 4.0.0) and use argument class = "array" to check for array object (of class "array" in R >= 4.0.0)
+ # ARGUMENTS
+ # data: object to test
+ # class: character string. Either one of the class() result (But see the warning section above) or "vector" or "ggplot2" (i.e., objects of class c("gg", "ggplot")) or NULL
+ # typeof: character string. Either one of the typeof() result or NULL
+ # mode: character string. Either one of the mode() result (for non-vector object) or NULL
+ # length: numeric value indicating the length of the object. Not considered if NULL
+ # prop: logical. Are the numeric values between 0 and 1 (proportion)? If TRUE, can be used alone, without considering class, etc.
+ # double.as.integer.allowed: logical. If TRUE, no error is reported in the cheking message if argument is set to typeof == "integer" or class == "integer", while the reality is typeof == "double" or class == "numeric" but the numbers strictly have zero as modulo (remainder of a division). This means that i <- 1, which is typeof(i) -> "double" is considered as integer with double.as.integer.allowed = TRUE. WARNING: data%%1 == 0L but not isTRUE(all.equal(data%%1, 0)) is used here because the argument checks for integers stored as double (does not check for decimal numbers that are approximate integers)
+ # options: a vector of character strings or integers indicating all the possible option values for the data argument, or NULL. Numbers of type "double" are accepted if they have a 0 modulo
+ # all.options.in.data: logical. If TRUE, all of the options must be present at least once in the data argument, and nothing else. If FALSE, some or all of the options must be present in the data argument, and nothing else. Ignored if options is NULL
+ # na.contain: logical. Can the data argument contain NA?
+ # neg.values: logical. Are negative numeric values authorized? Warning: the default setting is TRUE, meaning that, in that case, no check is performed for the presence of negative values. The neg.values argument is activated only when set to FALSE. In addition, (1) neg.values = FALSE can only be used when class, typeof or mode arguments are not NULL, otherwise return an error message, (2) the presence of negative values is not checked with neg.values = FALSE if the tested object is a factor and the following checking message is returned "OBJECT MUST BE MADE OF NON NEGATIVE VALUES BUT IS A FACTOR"
+ # print: logical. Print the message if $problem is TRUE? Warning: set by default to FALSE, which facilitates the control of the checking message output when using fun_check() inside functions. See the example section
+ # data.name: character string indicating the name of the object to test. If NULL, use what is assigned to the data argument for the returned message
+ # fun.name: character string indicating the name of the function checked (i.e., when fun_check() is used to check the arguments of this function). If non-null, the value of fun.name will be added into the message returned by fun_check()
+ # RETURN
+ # A list containing:
+ # $problem: logical. Is there any problem detected?
+ # $text: message indicating the details of the problem, or the absence of problem
+ # $object.name: value of the data.name argument (i.e., name of the checked object if provided, NULL otherwise)
+ # REQUIRED PACKAGES
+ # None
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # None
+ # EXAMPLE
+ # test <- matrix(1:3) ; fun_check(data = test, print = TRUE, class = "vector", mode = "numeric")
+ # see http
+ # DEBUGGING
+ # data = mean ; class = NULL ; typeof = NULL ; mode = NULL ; length = NULL ; prop = FALSE ; double.as.integer.allowed = FALSE ; options = "a" ; all.options.in.data = FALSE ; na.contain = FALSE ; neg.values = TRUE ; print = TRUE ; data.name = NULL ; fun.name = NULL
+ # function name
+ # no used in this function for the error message, to avoid env colliding
+ # end function name
+ # required function checking
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # fun.name checked first because required next
+ if( ! is.null(fun.name)){ # I have to use this way to deal with every kind of class for fun.name
+ if(all(base::class(fun.name) == "character")){ # all() without na.rm -> ok because class(NA) is "logical"
+ if(base::length(fun.name) != 1){
+ tempo.cat <- paste0("ERROR IN fun_check(): THE fun.name ARGUMENT MUST BE A CHARACTER VECTOR OF LENGTH 1: ", paste(fun.name, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(any(is.na(fun.name))){ # normally no NA with is.na()
+ tempo.cat <- paste0("ERROR IN fun_check(): NO ARGUMENT EXCEPT data AND options CAN HAVE NA VALUES\nPROBLEMATIC ARGUMENT IS fun.name")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN fun_check(): THE fun.name ARGUMENT MUST BE A CHARACTER VECTOR OF LENGTH 1") # paste(fun.name, collapse = " ") removed here because does not work with objects like function
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end fun.name checked first because required next
+ # arg with no default values
+ mandat.args <- c(
+ "data"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of special classes
+ basic.class <- c(
+ "NULL", # because class(NULL) is "NULL". The NULL aspect will be dealt later
+ "logical",
+ "integer",
+ "numeric",
+ # "complex",
+ "character"
+ # "matrix",
+ # "array",
+ # "data.frame",
+ # "list",
+ # "factor",
+ # "table",
+ # "expression",
+ # "name",
+ # "symbol",
+ # "function",
+ # "uneval",
+ # "environment",
+ # "ggplot2",
+ # "ggplot_built",
+ # "call"
+ )
+ tempo.arg.base <-c( # no names(formals(fun = sys.function(sys.parent(n = 2)))) used with fun_check() to be sure to deal with the correct environment
+ "class",
+ "typeof",
+ "mode",
+ "length",
+ "prop",
+ "double.as.integer.allowed",
+ "options",
+ "all.options.in.data",
+ "na.contain",
+ "neg.values",
+ "print",
+ "data.name",
+ "fun.name"
+ )
+ tempo.class <-list( # no get() used to be sure to deal with the correct environment
+ base::class(class),
+ base::class(typeof),
+ base::class(mode),
+ base::class(length),
+ base::class(prop),
+ base::class(double.as.integer.allowed),
+ base::class(options),
+ base::class(all.options.in.data),
+ base::class(na.contain),
+ base::class(neg.values),
+ base::class(print),
+ base::class(data.name),
+ base::class(fun.name)
+ )
+ tempo <- ! sapply(lapply(tempo.class, FUN = "%in%", basic.class), FUN = all)
+ if(any(tempo)){
+ tempo.cat1 <- tempo.arg.base[tempo]
+ tempo.cat2 <- sapply(tempo.class[tempo], FUN = paste0, collapse = " ")
+ tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": ANY ARGUMENT EXCEPT data MUST HAVE A BASIC CLASS\nPROBLEMATIC ARGUMENT", ifelse(base::length(tempo.cat1) > 1, "S", ""), " AND ASSOCIATED CLASS", ifelse(base::length(tempo.cat1) > 1, "ES ARE", " IS"), ":\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n")) # normally no NA with is.na()
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of special classes
+ # management of NA arguments
+ if(any(is.na(data.name)) | any(is.na(class)) | any(is.na(typeof)) | any(is.na(mode)) | any(is.na(length)) | any(is.na(prop)) | any(is.na(double.as.integer.allowed)) | any(is.na(all.options.in.data)) | any(is.na(na.contain)) | any(is.na(neg.values)) | any(is.na(print)) | any(is.na(fun.name))){ # normally no NA with is.na()
+ tempo <- c("data.name", "class", "typeof", "mode", "length", "prop", "double.as.integer.allowed", "all.options.in.data", "na.contain", "neg.values", "print", "fun.name")[c(any(is.na(data.name)), any(is.na(class)), any(is.na(typeof)), any(is.na(mode)), any(is.na(length)), any(is.na(prop)), any(is.na(double.as.integer.allowed)), any(is.na(all.options.in.data)), any(is.na(na.contain)), any(is.na(neg.values)), any(is.na(print)), any(is.na(fun.name)))]
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": NO ARGUMENT EXCEPT data AND options CAN HAVE NA VALUES\nPROBLEMATIC ARGUMENT", ifelse(length(tempo) > 1, "S ARE", " IS"), ":\n", paste(tempo, collapse = "\n")) # normally no NA with is.na()
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "prop",
+ "double.as.integer.allowed",
+ "all.options.in.data",
+ "na.contain",
+ "neg.values",
+ "print"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){ # normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN fun.check():\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT BE NULL:\n", paste0(tempo.arg[tempo.log], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # dealing with logical
+ # tested below
+ # end dealing with logical
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ # end warning initiation
+ # other checkings
+ if( ! is.null(data.name)){
+ if( ! (base::length(data.name) == 1L & all(base::class(data.name) == "character"))){ # all() without na.rm -> ok because class(NA) is "logical"
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": data.name ARGUMENT MUST BE A SINGLE CHARACTER ELEMENT AND NOT ", paste(data.name, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(is.null(options) & is.null(class) & is.null(typeof) & is.null(mode) & prop == FALSE & is.null(length)){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": AT LEAST ONE OF THE options, class, typeof, mode, prop, OR length ARGUMENT MUST BE SPECIFIED (I.E, TRUE FOR prop)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(options) & ( ! is.null(class) | ! is.null(typeof) | ! is.null(mode) | prop == TRUE)){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE class, typeof, mode ARGUMENTS MUST BE NULL, AND prop FALSE, IF THE options ARGUMENT IS SPECIFIED\nTHE options ARGUMENT MUST BE NULL IF THE class AND/OR typeof AND/OR mode AND/OR prop ARGUMENT IS SPECIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (all(base::class(neg.values) == "logical") & base::length(neg.values) == 1L)){ # all() without na.rm -> ok because class(NA) is "logical"
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE neg.values ARGUMENT MUST BE TRUE OR FALSE ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(neg.values == FALSE & is.null(class) & is.null(typeof) & is.null(mode)){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE neg.values ARGUMENT CANNOT BE SWITCHED TO FALSE IF class, typeof AND mode ARGUMENTS ARE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(class)){ # may add "formula" and "Date" as in https://renenyffenegger.ch/notes/development/languages/R/functions/class
+ if( ! all(class %in% c("vector", "logical", "integer", "numeric", "complex", "character", "matrix", "array", "data.frame", "list", "factor", "table", "expression", "name", "symbol", "function", "uneval", "environment", "ggplot2", "ggplot_built", "call") & base::length(class) == 1L)){ # length == 1L here because of class(matrix()) since R4.0.0 # all() without na.rm -> ok because class cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT MUST BE ONE OF THESE VALUE:\n\"vector\", \"logical\", \"integer\", \"numeric\", \"complex\", \"character\", \"matrix\", \"array\", \"data.frame\", \"list\", \"factor\", \"table\", \"expression\", \"name\", \"symbol\", \"function\", \"environment\", \"ggplot2\", \"ggplot_built\", \"call\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(neg.values == FALSE & ! any(class %in% c("vector", "numeric", "integer", "matrix", "array", "data.frame", "table"))){ # no need of na.rm = TRUE for any() because %in% does not output NA
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT CANNOT BE OTHER THAN \"vector\", \"numeric\", \"integer\", \"matrix\", \"array\", \"data.frame\", \"table\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(typeof)){ # all the types are here: https://renenyffenegger.ch/notes/development/languages/R/functions/typeof
+ if( ! (all(typeof %in% c("logical", "integer", "double", "complex", "character", "list", "expression", "symbol", "closure", "special", "builtin", "environment", "S4", "language")) & base::length(typeof) == 1L)){ # "language" is the type of object of class "call" # all() without na.rm -> ok because typeof cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT MUST BE ONE OF THESE VALUE:\n\"logical\", \"integer\", \"double\", \"complex\", \"character\", \"list\", \"expression\", \"name\", \"symbol\", \"closure\", \"special\", \"builtin\", \"environment\", \"S4\", \"language\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(neg.values == FALSE & ! typeof %in% c("double", "integer")){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT CANNOT BE OTHER THAN \"double\" OR \"integer\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(mode)){ # all the types are here: https://renenyffenegger.ch/notes/development/languages/R/functions/typeof
+ if( ! (all(mode %in% c("logical", "numeric", "complex", "character", "list", "expression", "name", "symbol", "function", "environment", "S4", "call")) & base::length(mode) == 1L)){ # all() without na.rm -> ok because mode cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT MUST BE ONE OF THESE VALUE:\n\"logical\", \"numeric\", \"complex\", \"character\", \"list\", \"expression\", \"name\", \"symbol\", \"function\", \"environment\", \"S4\", \"call\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(neg.values == FALSE & mode != "numeric"){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT CANNOT BE OTHER THAN \"numeric\" IF neg.values ARGUMENT IS SWITCHED TO FALSE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(length)){
+ if( ! (is.numeric(length) & base::length(length) == 1L & all( ! grepl(length, pattern = "\\.")))){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": length ARGUMENT MUST BE A SINGLE INTEGER VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! (is.logical(prop) & base::length(prop) == 1L)){ # is.na() already checked for prop
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": prop ARGUMENT MUST BE TRUE OR FALSE ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(prop == TRUE){
+ if( ! is.null(class)){
+ if( ! any(class %in% c("vector", "numeric", "matrix", "array", "data.frame", "table"))){ # no need of na.rm = TRUE for any() because %in% does not output NA
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": class ARGUMENT CANNOT BE OTHER THAN NULL, \"vector\", \"numeric\", \"matrix\", \"array\", \"data.frame\", \"table\" IF prop ARGUMENT IS TRUE") # not integer because prop
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(mode)){
+ if(mode != "numeric"){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": mode ARGUMENT CANNOT BE OTHER THAN NULL OR \"numeric\" IF prop ARGUMENT IS TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(typeof)){
+ if(typeof != "double"){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": typeof ARGUMENT CANNOT BE OTHER THAN NULL OR \"double\" IF prop ARGUMENT IS TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ }
+ if( ! (all(base::class(double.as.integer.allowed) == "logical") & base::length(double.as.integer.allowed) == 1L)){ # all() without na.rm -> ok because class() never returns NA
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": THE double.as.integer.allowed ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(double.as.integer.allowed, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (is.logical(all.options.in.data) & base::length(all.options.in.data) == 1L)){
+ tempo.cat <- paste0("ERROR IN fun_check()", ifelse(is.null(fun.name), "", paste0(" INSIDE ", fun.name)), ": all.options.in.data ARGUMENT MUST BE A SINGLE LOGICAL VALUE (TRUE OR FALSE ONLY): ", paste(all.options.in.data, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (all(base::class(na.contain) == "logical") & base::length(na.contain) == 1L)){ # all() without na.rm -> ok because class() never returns NA
+ tempo.cat <- paste0("ERROR IN fun_check(): THE na.contain ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(na.contain, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (all(base::class(print) == "logical") & base::length(print) == 1L)){ # all() without na.rm -> ok because class() never returns NA
+ tempo.cat <- paste0("ERROR IN fun_check(): THE print ARGUMENT MUST BE TRUE OR FALSE ONLY: ", paste(print, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # data.name and fun.name tested at the beginning
+ # end other checkings
+ # end second round of checking and data preparation
+ # package checking
+ # end package checking
+ # main code
+ if(is.null(data.name)){
+ data.name <- deparse(substitute(data))
+ }
+ problem <- FALSE
+ text <- paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT")
+ if(( ! is.null(options)) & (all(base::typeof(data) == "character") | all(base::typeof(data) == "integer") | all(base::typeof(data) == "double"))){ # all() without na.rm -> ok because typeof() never returns NA
+ if(all(base::typeof(data) == "double")){
+ if( ! all(data %% 1 == 0L, na.rm = TRUE)){
+ problem <- TRUE
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nBUT IS NOT EVEN TYPE CHARACTER OR INTEGER")
+ }
+ }else{
+ text <- ""
+ if( ! all(data %in% options)){ # no need of na.rm = TRUE for all() because %in% does not output NA
+ problem <- TRUE
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nTHE PROBLEMATIC ELEMENTS OF ", data.name, " ARE: ", paste(unique(data[ ! (data %in% options)]), collapse = " "))
+ }
+ if(all.options.in.data == TRUE){
+ if( ! all(options %in% data)){ # no need of na.rm = TRUE for all() because %in% does not output NA
+ problem <- TRUE
+ text <- paste0(ifelse(text == "", "", paste0(text, "\n")), ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE MADE OF ALL THESE OPTIONS: ", paste(options, collapse = " "), "\nTHE MISSING ELEMENTS OF THE options ARGUMENT ARE: ", paste(unique(options[ ! (options %in% data)]), collapse = " "))
+ }
+ }
+ if( ! is.null(length)){
+ if(base::length(data) != length){
+ problem <- TRUE
+ text <- paste0(ifelse(text == "", "", paste0(text, "\n")), ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE LENGTH OF ", data.name, " MUST BE ", length, " AND NOT ", base::length(data))
+ }
+ }
+ if(text == ""){
+ text <- paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT")
+ }
+ }
+ }else if( ! is.null(options)){
+ problem <- TRUE
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE SOME OF THESE OPTIONS: ", paste(options, collapse = " "), "\nBUT IS NOT EVEN TYPE CHARACTER OR INTEGER")
+ }
+ arg.names <- c("class", "typeof", "mode", "length")
+ if( ! is.null(class)){
+ if(class == "matrix"){ # because of class(matric()) since R4.0.0
+ class <- c("matrix", "array")
+ }else if(class == "factor" & all(base::class(data) %in% c("factor", "ordered"))){ # to deal with ordered factors # all() without na.rm -> ok because class(NA) is "logical"
+ class <- c("factor", "ordered")
+ }
+ }
+ if(is.null(options)){
+ for(i2 in 1:base::length(arg.names)){
+ if( ! is.null(get(arg.names[i2], env = sys.nframe(), inherit = FALSE))){
+ # script to execute
+ tempo.script <- '
problem <- TRUE ;
if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": THE ", data.name, " OBJECT MUST BE ") ;
@@ -443,75 +443,75 @@ text <- paste0(text, " AND ") ;
}
text <- paste0(text, toupper(arg.names[i2]), " ", if(all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) %in% c("matrix", "array"))){"matrix"}else if(all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) %in% c("factor", "ordered"))){"factor"}else{get(arg.names[i2], env = sys.nframe(), inherit = FALSE)})
' # no need of na.rm = TRUE for all() because %in% does not output NA
-# end script to execute
-if(base::typeof(data) == "double" & double.as.integer.allowed == TRUE & ((arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "integer")) | (arg.names[i2] == "typeof" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "integer")))){ # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names # typeof(data) == "double" means no factor allowed
-if( ! all(data %% 1 == 0L, na.rm = TRUE)){ # to check integers (use %%, meaning the remaining of a division): see the precedent line. isTRUE(all.equal(data%%1, rep(0, length(data)))) not used because we strictly need zero as a result. Warning: na.rm = TRUE required here for all()
-eval(parse(text = tempo.script)) # execute tempo.script
-}
-}else if( ! any(all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) %in% c("vector", "ggplot2"))) & ! all(eval(parse(text = paste0(arg.names[i2], "(data)"))) %in% get(arg.names[i2], env = sys.nframe(), inherit = FALSE))){ # test the four c("class", "typeof", "mode", "length") arguments with their corresponding function. No need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for any() because get get(arg.names) does not contain NA
-eval(parse(text = tempo.script)) # execute tempo.script
-}else if(arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "vector") & ! (all(base::class(data) %in% "numeric") | all(base::class(data) %in% "integer") | all(base::class(data) %in% "character") | all(base::class(data) %in% "logical"))){ # test class == "vector". No need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names
-eval(parse(text = tempo.script)) # execute tempo.script
-}else if(arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "ggplot2") & ! all(base::class(data) %in% c("gg", "ggplot"))){ # test ggplot object # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names # no need of na.rm = TRUE for all() because %in% does not output NA
-eval(parse(text = tempo.script)) # execute tempo.script
-}
-}
-}
-}
+ # end script to execute
+ if(base::typeof(data) == "double" & double.as.integer.allowed == TRUE & ((arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "integer")) | (arg.names[i2] == "typeof" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "integer")))){ # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names # typeof(data) == "double" means no factor allowed
+ if( ! all(data %% 1 == 0L, na.rm = TRUE)){ # to check integers (use %%, meaning the remaining of a division): see the precedent line. isTRUE(all.equal(data%%1, rep(0, length(data)))) not used because we strictly need zero as a result. Warning: na.rm = TRUE required here for all()
+ eval(parse(text = tempo.script)) # execute tempo.script
+ }
+ }else if( ! any(all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) %in% c("vector", "ggplot2"))) & ! all(eval(parse(text = paste0(arg.names[i2], "(data)"))) %in% get(arg.names[i2], env = sys.nframe(), inherit = FALSE))){ # test the four c("class", "typeof", "mode", "length") arguments with their corresponding function. No need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for any() because get get(arg.names) does not contain NA
+ eval(parse(text = tempo.script)) # execute tempo.script
+ }else if(arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "vector") & ! (all(base::class(data) %in% "numeric") | all(base::class(data) %in% "integer") | all(base::class(data) %in% "character") | all(base::class(data) %in% "logical"))){ # test class == "vector". No need of na.rm = TRUE for all() because %in% does not output NA # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names
+ eval(parse(text = tempo.script)) # execute tempo.script
+ }else if(arg.names[i2] == "class" & all(get(arg.names[i2], env = sys.nframe(), inherit = FALSE) == "ggplot2") & ! all(base::class(data) %in% c("gg", "ggplot"))){ # test ggplot object # no need of na.rm = TRUE for all() because == does not output NA if no NA in left of ==, which is the case for arg.names # no need of na.rm = TRUE for all() because %in% does not output NA
+ eval(parse(text = tempo.script)) # execute tempo.script
+ }
+ }
+ }
+ }
if(prop == TRUE & all(base::typeof(data) == "double")){ # all() without na.rm -> ok because typeof(NA) is "logical"
-if(is.null(data) | any(data < 0 | data > 1, na.rm = TRUE)){ # works if data is NULL # Warning: na.rm = TRUE required here for any() # typeof(data) == "double" means no factor allowed
-problem <- TRUE
-if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
-}else{
-text <- paste0(text, " AND ")
-}
-text <- paste0(text, "THE ", data.name, " OBJECT MUST BE DECIMAL VALUES BETWEEN 0 AND 1")
-}
+ if(is.null(data) | any(data < 0 | data > 1, na.rm = TRUE)){ # works if data is NULL # Warning: na.rm = TRUE required here for any() # typeof(data) == "double" means no factor allowed
+ problem <- TRUE
+ if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
+ }else{
+ text <- paste0(text, " AND ")
+ }
+ text <- paste0(text, "THE ", data.name, " OBJECT MUST BE DECIMAL VALUES BETWEEN 0 AND 1")
+ }
}else if(prop == TRUE){
-problem <- TRUE
-if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
-}else{
-text <- paste0(text, " AND ")
-}
-text <- paste0(text, "THE ", data.name, " OBJECT MUST BE DECIMAL VALUES BETWEEN 0 AND 1")
+ problem <- TRUE
+ if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
+ }else{
+ text <- paste0(text, " AND ")
+ }
+ text <- paste0(text, "THE ", data.name, " OBJECT MUST BE DECIMAL VALUES BETWEEN 0 AND 1")
}
if(all(base::class(data) %in% "expression")){ # no need of na.rm = TRUE for all() because %in% does not output NA
-data <- as.character(data) # to evaluate the presence of NA
+ data <- as.character(data) # to evaluate the presence of NA
}
if(na.contain == FALSE & (base::mode(data) %in% c("logical", "numeric", "complex", "character", "list"))){ # before it was ! (class(data) %in% c("function", "environment"))
-if(any(is.na(data)) == TRUE){ # not on the same line because when data is class envir or function , do not like that # normally no NA with is.na()
-problem <- TRUE
-if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
-}else{
-text <- paste0(text, " AND ")
-}
-text <- paste0(text, "THE ", data.name, " OBJECT CONTAINS NA WHILE NOT AUTHORIZED")
-}
+ if(any(is.na(data)) == TRUE){ # not on the same line because when data is class envir or function , do not like that # normally no NA with is.na()
+ problem <- TRUE
+ if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
+ }else{
+ text <- paste0(text, " AND ")
+ }
+ text <- paste0(text, "THE ", data.name, " OBJECT CONTAINS NA WHILE NOT AUTHORIZED")
+ }
}
if(neg.values == FALSE & all(base::mode(data) %in% "numeric") & ! any(base::class(data) %in% "factor")){ # no need of na.rm = TRUE for all() because %in% does not output NA
-if(any(data < 0, na.rm = TRUE)){ # Warning: na.rm = TRUE required here for any()
-problem <- TRUE
-if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
-}else{
-text <- paste0(text, " AND ")
-}
-text <- paste0(text, "THE ", data.name, " OBJECT MUST BE MADE OF NON NEGATIVE NUMERIC VALUES")
-}
+ if(any(data < 0, na.rm = TRUE)){ # Warning: na.rm = TRUE required here for any()
+ problem <- TRUE
+ if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
+ }else{
+ text <- paste0(text, " AND ")
+ }
+ text <- paste0(text, "THE ", data.name, " OBJECT MUST BE MADE OF NON NEGATIVE NUMERIC VALUES")
+ }
}else if(neg.values == FALSE){
-problem <- TRUE
-if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
-text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
-}else{
-text <- paste0(text, " AND ")
-}
-text <- paste0(text, "THE ", data.name, " OBJECT MUST BE MADE OF NON NEGATIVE VALUES BUT IS ", ifelse(any(base::class(data) %in% "factor"), "A FACTOR", "NOT EVEN MODE NUMERIC"))
+ problem <- TRUE
+ if(identical(text, paste0(ifelse(is.null(fun.name), "", paste0("IN ", fun.name, ": ")), "NO PROBLEM DETECTED FOR THE ", data.name, " OBJECT"))){
+ text <- paste0(ifelse(is.null(fun.name), "ERROR", paste0("ERROR IN ", fun.name)), ": ")
+ }else{
+ text <- paste0(text, " AND ")
+ }
+ text <- paste0(text, "THE ", data.name, " OBJECT MUST BE MADE OF NON NEGATIVE VALUES BUT IS ", ifelse(any(base::class(data) %in% "factor"), "A FACTOR", "NOT EVEN MODE NUMERIC"))
}
if(print == TRUE & problem == TRUE){
-cat(paste0("\n\n================\n\n", text, "\n\n================\n\n"))
+ cat(paste0("\n\n================\n\n", text, "\n\n================\n\n"))
}
# output
output <- list(problem = problem, text = text, object.name = data.name)
@@ -525,104 +525,104 @@ return(output)
fun_secu <- function(pos = 1, name = NULL){
-# AIM
-# Verify that variables in the environment defined by the pos parameter are not present in the above environment (following R Scope). This can be used to avoid R scope preference of functions like get()
-# ARGUMENTS
-# pos: single integer indicating the position of the environment checked (argument n of parent.frame()). Value 1 means one step above the fun_secu() local environment (by default). This means that when fun_secu(pos = 1) is used inside a function A, it checks if variables in the local environment of this function A are also present in above environments (following R Scope). When fun_secu(pos = 1) is used in the Global environment, it checks the objects of this environment
-# name: single character string indicating the name of the function checked. If NULL, fun_secu() checks all the variables of the environment indicated by pos, as explained in the pos argument description. If non-null, fun_secu() checks all the variables presents in the local env of the function will be checked in the above envs (which includes the working environment (Global env)
-# RETURN
-# A character string of the local variables that match variables in the different environments of the R scope, or NULL if no match
-# REQUIRED PACKAGES
-# None
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_secu()
-# fun_secu(pos = 2)
-# mean <- 0 ; fun1 <- function(){sd <- 1 ; fun_secu(name = as.character(sys.calls()[[length(sys.calls())]]))} ; fun2 <- function(){cor <- 2 ; fun1()} ; fun1() ; fun2() ; rm(mean) # sys.calls() gives the function name at top stack of the imbricated functions, sys.calls()[[length(sys.calls())]] the name of the just above function. This can also been used for the above function: as.character(sys.call(1))
-# test.pos <- 2 ; mean <- 0 ; fun1 <- function(){sd <- 1 ; fun_secu(pos = test.pos, name = if(length(sys.calls()) >= test.pos){as.character(sys.calls()[[length(sys.calls()) + 1 - test.pos]])}else{search()[ (1:length(search()))[test.pos - length(sys.calls())]]})} ; fun2 <- function(){cor <- 2 ; fun1()} ; fun1() ; fun2() ; rm(mean) # for argument name, here is a way to have the name of the tested environment according to test.pos value
-# DEBUGGING
-# pos = 1 ; name = NULL # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument primary checking
-# arg with no default values
-# end arg with no default values
-# using fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = pos, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(name)){
-tempo <- fun_check(data = name, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <- c(
-"pos"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT BE NULL:", paste0(tempo.arg[tempo.log], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# end second round of checking and data preparation
-# main code
-# match.list <- vector("list", length = (length(sys.calls()) - 1 + length(search()) + ifelse(length(sys.calls()) == 1L, -1, 0))) # match.list is a list of all the environment tested (local of functions and R envs), length(sys.calls()) - 1 to remove the level of the fun_secu() function, sys.calls() giving all the names of the imbricated functions, including fun_secu, ifelse(length(sys.calls()) == 1L, -1, 0) to remove Global env if this one is tested
-tempo.name <- rev(as.character(unlist(sys.calls()))) # get names of frames (i.e., enclosed env)
-tempo.frame <- rev(sys.frames()) # get frames (i.e., enclosed env)
-# dealing with source()
-# source() used in the Global env creates three frames above the Global env, which should be removed because not very interesting for variable duplications. Add a <<-(sys.frames()) in this code and source anova_contrasts code to see this. With ls(a[[4]]), we can see the content of each env, which are probably elements of source()
-if(any(sapply(tempo.frame, FUN = environmentName) %in% "R_GlobalEnv")){
-global.pos <- which(sapply(tempo.frame, FUN = environmentName) %in% "R_GlobalEnv")
-# remove the global env (because already in search(), and all the oabove env
-tempo.name <- tempo.name[-c(global.pos:length(tempo.frame))]
-tempo.frame <- tempo.frame[-c(global.pos:length(tempo.frame))]
-}
-# end dealing with source()
-# might have a problem if(length(tempo.name) == 0L){
-match.list <- vector("list", length = length(tempo.name) + length(search())) # match.list is a list of all the environment tested (local of functions and R envs), length(sys.calls()) - 1 to remove the level of the fun_secu() function, sys.calls() giving all the names of the imbricated functions, including fun_secu, ifelse(length(sys.calls()) == 1L, -1, 0) to remove Global env if this one is tested
-ls.names <- c(tempo.name, search()) # names of the functions + names of the search() environments
-ls.input <- c(tempo.frame, as.list(search())) # environements of the functions + names of the search() environments
-names(match.list) <- ls.names #
-match.list <- match.list[-c(1:(pos + 1))] # because we check only above pos
-ls.tested <- ls.input[[pos + 1]]
-ls.input <- ls.input[-c(1:(pos + 1))]
-for(i1 in 1:length(match.list)){
-if(any(ls(name = ls.input[[i1]], all.names = TRUE) %in% ls(name = ls.tested, all.names = TRUE))){
-match.list[i1] <- list(ls(name = ls.input[[i1]], all.names = TRUE)[ls(name = ls.input[[i1]], all.names = TRUE) %in% ls(name = ls.tested, all.names = TRUE)])
-}
-}
-if( ! all(sapply(match.list, FUN = is.null))){
-output <- paste0("SOME VARIABLES ", ifelse(is.null(name), "OF THE CHECKED ENVIRONMENT", paste0("OF ", name)), " ARE ALSO PRESENT IN :\n", paste0(names(match.list[ ! sapply(match.list, FUN = is.null)]), ": ", sapply(match.list[ ! sapply(match.list, FUN = is.null)], FUN = paste0, collapse = " "), collapse = "\n"))
-}else{
-output <- NULL
-}
-return(output)
+ # AIM
+ # Verify that variables in the environment defined by the pos parameter are not present in the above environment (following R Scope). This can be used to avoid R scope preference of functions like get()
+ # ARGUMENTS
+ # pos: single integer indicating the position of the environment checked (argument n of parent.frame()). Value 1 means one step above the fun_secu() local environment (by default). This means that when fun_secu(pos = 1) is used inside a function A, it checks if variables in the local environment of this function A are also present in above environments (following R Scope). When fun_secu(pos = 1) is used in the Global environment, it checks the objects of this environment
+ # name: single character string indicating the name of the function checked. If NULL, fun_secu() checks all the variables of the environment indicated by pos, as explained in the pos argument description. If non-null, fun_secu() checks all the variables presents in the local env of the function will be checked in the above envs (which includes the working environment (Global env)
+ # RETURN
+ # A character string of the local variables that match variables in the different environments of the R scope, or NULL if no match
+ # REQUIRED PACKAGES
+ # None
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_secu()
+ # fun_secu(pos = 2)
+ # mean <- 0 ; fun1 <- function(){sd <- 1 ; fun_secu(name = as.character(sys.calls()[[length(sys.calls())]]))} ; fun2 <- function(){cor <- 2 ; fun1()} ; fun1() ; fun2() ; rm(mean) # sys.calls() gives the function name at top stack of the imbricated functions, sys.calls()[[length(sys.calls())]] the name of the just above function. This can also been used for the above function: as.character(sys.call(1))
+ # test.pos <- 2 ; mean <- 0 ; fun1 <- function(){sd <- 1 ; fun_secu(pos = test.pos, name = if(length(sys.calls()) >= test.pos){as.character(sys.calls()[[length(sys.calls()) + 1 - test.pos]])}else{search()[ (1:length(search()))[test.pos - length(sys.calls())]]})} ; fun2 <- function(){cor <- 2 ; fun1()} ; fun1() ; fun2() ; rm(mean) # for argument name, here is a way to have the name of the tested environment according to test.pos value
+ # DEBUGGING
+ # pos = 1 ; name = NULL # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument primary checking
+ # arg with no default values
+ # end arg with no default values
+ # using fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = pos, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(name)){
+ tempo <- fun_check(data = name, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <- c(
+ "pos"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT BE NULL:", paste0(tempo.arg[tempo.log], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # end second round of checking and data preparation
+ # main code
+ # match.list <- vector("list", length = (length(sys.calls()) - 1 + length(search()) + ifelse(length(sys.calls()) == 1L, -1, 0))) # match.list is a list of all the environment tested (local of functions and R envs), length(sys.calls()) - 1 to remove the level of the fun_secu() function, sys.calls() giving all the names of the imbricated functions, including fun_secu, ifelse(length(sys.calls()) == 1L, -1, 0) to remove Global env if this one is tested
+ tempo.name <- rev(as.character(unlist(sys.calls()))) # get names of frames (i.e., enclosed env)
+ tempo.frame <- rev(sys.frames()) # get frames (i.e., enclosed env)
+ # dealing with source()
+ # source() used in the Global env creates three frames above the Global env, which should be removed because not very interesting for variable duplications. Add a <<-(sys.frames()) in this code and source anova_contrasts code to see this. With ls(a[[4]]), we can see the content of each env, which are probably elements of source()
+ if(any(sapply(tempo.frame, FUN = environmentName) %in% "R_GlobalEnv")){
+ global.pos <- which(sapply(tempo.frame, FUN = environmentName) %in% "R_GlobalEnv")
+ # remove the global env (because already in search(), and all the oabove env
+ tempo.name <- tempo.name[-c(global.pos:length(tempo.frame))]
+ tempo.frame <- tempo.frame[-c(global.pos:length(tempo.frame))]
+ }
+ # end dealing with source()
+ # might have a problem if(length(tempo.name) == 0L){
+ match.list <- vector("list", length = length(tempo.name) + length(search())) # match.list is a list of all the environment tested (local of functions and R envs), length(sys.calls()) - 1 to remove the level of the fun_secu() function, sys.calls() giving all the names of the imbricated functions, including fun_secu, ifelse(length(sys.calls()) == 1L, -1, 0) to remove Global env if this one is tested
+ ls.names <- c(tempo.name, search()) # names of the functions + names of the search() environments
+ ls.input <- c(tempo.frame, as.list(search())) # environements of the functions + names of the search() environments
+ names(match.list) <- ls.names #
+ match.list <- match.list[-c(1:(pos + 1))] # because we check only above pos
+ ls.tested <- ls.input[[pos + 1]]
+ ls.input <- ls.input[-c(1:(pos + 1))]
+ for(i1 in 1:length(match.list)){
+ if(any(ls(name = ls.input[[i1]], all.names = TRUE) %in% ls(name = ls.tested, all.names = TRUE))){
+ match.list[i1] <- list(ls(name = ls.input[[i1]], all.names = TRUE)[ls(name = ls.input[[i1]], all.names = TRUE) %in% ls(name = ls.tested, all.names = TRUE)])
+ }
+ }
+ if( ! all(sapply(match.list, FUN = is.null))){
+ output <- paste0("SOME VARIABLES ", ifelse(is.null(name), "OF THE CHECKED ENVIRONMENT", paste0("OF ", name)), " ARE ALSO PRESENT IN :\n", paste0(names(match.list[ ! sapply(match.list, FUN = is.null)]), ": ", sapply(match.list[ ! sapply(match.list, FUN = is.null)], FUN = paste0, collapse = " "), collapse = "\n"))
+ }else{
+ output <- NULL
+ }
+ return(output)
}
@@ -638,273 +638,273 @@ return(output)
# -> transferred into the cute package
# Do not modify this function in cute_little_R_function anymore. See the cute repo
fun_info <- function(
-data,
-n = NULL,
-warn.print = TRUE
+ data,
+ n = NULL,
+ warn.print = TRUE
){
-# AIM
-# Provide a broad description of an object
-# WARNINGS
-# None
-# ARGUMENTS
-# data: object to analyse
-# n: positive integer value indicating the n first number of elements to display per compartment of the output list (i.e., head(..., n)). Write NULL to return all the elements. Does not apply for the $STRUCTURE compartment output
-# warn.print: logical. Print potential warnings at the end of the execution? If FALSE the warning messages are added in the output list as an additional compartment (or NULL if no message).
-# RETURN
-# A list containing information, depending on the class and type of data. The backbone is generally:
-# $NAME: name of the object
-# $CLASS: class of the object (class() value)
-# $TYPE: type of the object (typeof() value)
-# $LENGTH: length of the object (length() value)
-# $NA.NB: number of NA and NaN (only for type "logical", "integer", "double", "complex", "character" or "list")
-# $HEAD: head of the object (head() value)
-# $TAIL: tail of the object (tail() value)
-# $DIMENSION: dimension (only for object with dimensions)
-# $SUMMARY: object summary (summary() value)
-# $STRUCTURE: object structure (str() value)
-# $WARNING: warning messages (only if the warn.print argument is FALSE)
-# If data is made of numerics, provide also:
-# $INF.NB: number of Inf and -Inf
-# $RANGE: range after removing Inf and NA
-# $SUM: sum after removing Inf and NA
-# $MEAN: mean after removing Inf and NA
-# If data is a 2D object, provide also:
-# $ROW_NAMES: row names
-# $COL_NAMES: column names
-# If data is a data frame, provide also:
-# $COLUMN_TYPE: type of each column (typeof() value)
-# If data is a list, provide also:
-# $COMPARTMENT_NAMES: names of the comprtments
-# $COMPARTMENT_TYPE: type of each compartment (typeof() value)
-# REQUIRED PACKAGES
-# None
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# fun_check()
-# fun_get_message()
-# EXAMPLE
-# fun_info(data = 1:3)
-# see http
-# DEBUGGING
-# mat1 <- matrix(1:3) ; data = env1 ; n = NULL ; warn.print = TRUE # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check",
-"fun_get_message"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words
-# end reserved words
-# arg with no default values
-mandat.args <- c(
-"data"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if( ! is.null(n)){
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = n, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){ # normally no NA
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"data",
-# "n", # because can be NULL
-"warn.print"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-# warn.count <- 0 # not required
-# end warning initiation
-# other checkings
-if( ! is.null(n)){
-if(n < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A POSITIVE AND NON NULL INTEGER")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(is.finite(n)){
-# warn.count <- warn.count + 1
-tempo.warn <- paste0("SOME COMPARTMENTS CAN BE TRUNCATED (n ARGUMENT IS ", n, ")")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end other checkings
-# reserved word checking
-# end reserved word checking
-# end second round of checking and data preparation
-# package checking
-# end package checking
-# main code
-# new environment
-env.name <- paste0("env", as.numeric(Sys.time()))
-if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-assign(env.name, new.env())
-assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
-}
-# end new environment
-data.name <- deparse(substitute(data))
-output <- list("NAME" = data.name)
-tempo.try.error <- fun_get_message(data = "class(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("CLASS" = class(data))
-output <- c(output, tempo)
-}
-tempo.try.error <- fun_get_message(data = "typeof(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("TYPE" = typeof(data))
-output <- c(output, tempo)
-}
-tempo.try.error <- fun_get_message(data = "length(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("LENGTH" = length(data))
-output <- c(output, tempo)
-}
-if(all(typeof(data) %in% c("integer", "numeric", "double")) & ! any(class(data) %in% "factor")){ # all() without na.rm -> ok because typeof(NA) is "logical" # any() without na.rm -> ok because class(NA) is "logical"
-tempo <- list("INF.NB" = sum(is.infinite(data)))
-output <- c(output, tempo)
-tempo <- list("RANGE" = range(data[ ! is.infinite(data)], na.rm = TRUE))
-output <- c(output, tempo)
-tempo <- list("SUM" = sum(data[ ! is.infinite(data)], na.rm = TRUE))
-output <- c(output, tempo)
-tempo <- list("MEAN" = mean(data[ ! is.infinite(data)], na.rm = TRUE))
-output <- c(output, tempo)
-}
-if(all(typeof(data) %in% c("logical", "integer", "double", "complex", "character", "list"))){ # all() without na.rm -> ok because typeof(NA) is "logical"
-tempo.try.error <- fun_get_message(data = "is.na(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("NA.NB" = sum(is.na(data)))
-output <- c(output, tempo)
-}
-}
-tempo.try.error <- fun_get_message(data = "head(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("HEAD" = head(data))
-output <- c(output, tempo)
-tempo <- list("TAIL" = tail(data)) # no reason that tail() does not work if head() works
-output <- c(output, tempo)
-}
-tempo.try.error <- fun_get_message(data = "dim(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-if(length(dim(data)) > 0){
-tempo <- list("DIMENSION" = dim(data))
-if(length(tempo[[1]]) == 2L){
-names(tempo[[1]]) <- c("NROW", "NCOL")
-}
-output <- c(output, tempo)
-}
-}
-if(all(class(data) == "data.frame") | all(class(data) %in% c("matrix", "array")) | all(class(data) == "table")){ # all() without na.rm -> ok because typeof(NA) is "logical"
-if(length(dim(data)) > 1){ # to avoid 1D table
-tempo <- list("ROW_NAMES" = dimnames(data)[[1]])
-output <- c(output, tempo)
-tempo <- list("COLUM_NAMES" = dimnames(data)[[2]])
-output <- c(output, tempo)
-}
-}
-tempo.try.error <- fun_get_message(data = "summary(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- list("SUMMARY" = summary(data))
-output <- c(output, tempo)
-}
-tempo.try.error <- fun_get_message(data = "noquote(matrix(capture.output(str(data))))", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
-if(is.null(tempo.try.error)){
-tempo <- capture.output(str(data))
-tempo <- list("STRUCTURE" = noquote(matrix(tempo, dimnames = list(rep("", length(tempo)), "")))) # str() print automatically, ls.str() not but does not give the order of the data.frame
-output <- c(output, tempo)
-}
-if(all(class(data) == "data.frame")){ # all() without na.rm -> ok because class(NA) is "logical"
-tempo <- list("COLUMN_TYPE" = sapply(data, FUN = "typeof"))
-if(any(sapply(data, FUN = "class") %in% "factor")){ # if an ordered factor is present, then sapply(data, FUN = "class") return a list but works with any(sapply(data, FUN = "class") %in% "factor") # any() without na.rm -> ok because class(NA) is "logical"
-tempo.class <- sapply(data, FUN = "class")
-if(any(unlist(tempo.class) %in% "ordered")){ # any() without na.rm -> ok because class(NA) is "logical"
-tempo2 <- sapply(tempo.class, paste, collapse = " ") # paste the "ordered" factor" in "ordered factor"
-}else{
-tempo2 <- unlist(tempo.class)
-}
-tempo[["COLUMN_TYPE"]][grepl(x = tempo2, pattern = "factor")] <- tempo2[grepl(x = tempo2, pattern = "factor")]
-}
-output <- c(output, tempo)
-}
-if(all(class(data) == "list")){ # all() without na.rm -> ok because class(NA) is "logical"
-tempo <- list("COMPARTMENT_NAMES" = names(data))
-output <- c(output, tempo)
-tempo <- list("COMPARTMENT_TYPE" = sapply(data, FUN = "typeof"))
-if(any(unlist(sapply(data, FUN = "class")) %in% "factor")){ # if an ordered factor is present, then sapply(data, FUN = "class") return a list but works with any(sapply(data, FUN = "class") %in% "factor") # any() without na.rm -> ok because class(NA) is "logical"
-tempo.class <- sapply(data, FUN = "class")
-if(any(unlist(tempo.class) %in% "ordered")){ # any() without na.rm -> ok because class(NA) is "logical"
-tempo2 <- sapply(tempo.class, paste, collapse = " ") # paste the "ordered" factor" in "ordered factor"
-}else{
-tempo2 <- unlist(tempo.class)
-}
-tempo[["COMPARTMENT_TYPE"]][grepl(x = tempo2, pattern = "factor")] <- tempo2[grepl(x = tempo2, pattern = "factor")]
-}
-output <- c(output, tempo)
-}
-if( ! is.null(n)){
-output[names(output) != "STRUCTURE"] <- lapply(X = output[names(output) != "STRUCTURE"], FUN = head, n = n, simplify = FALSE)
-}
-# output
-if(warn.print == FALSE){
-output <- c(output, WARNING = warn)
-}else if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-return(output)
-# end output
-# end main code
+ # AIM
+ # Provide a broad description of an object
+ # WARNINGS
+ # None
+ # ARGUMENTS
+ # data: object to analyse
+ # n: positive integer value indicating the n first number of elements to display per compartment of the output list (i.e., head(..., n)). Write NULL to return all the elements. Does not apply for the $STRUCTURE compartment output
+ # warn.print: logical. Print potential warnings at the end of the execution? If FALSE the warning messages are added in the output list as an additional compartment (or NULL if no message).
+ # RETURN
+ # A list containing information, depending on the class and type of data. The backbone is generally:
+ # $NAME: name of the object
+ # $CLASS: class of the object (class() value)
+ # $TYPE: type of the object (typeof() value)
+ # $LENGTH: length of the object (length() value)
+ # $NA.NB: number of NA and NaN (only for type "logical", "integer", "double", "complex", "character" or "list")
+ # $HEAD: head of the object (head() value)
+ # $TAIL: tail of the object (tail() value)
+ # $DIMENSION: dimension (only for object with dimensions)
+ # $SUMMARY: object summary (summary() value)
+ # $STRUCTURE: object structure (str() value)
+ # $WARNING: warning messages (only if the warn.print argument is FALSE)
+ # If data is made of numerics, provide also:
+ # $INF.NB: number of Inf and -Inf
+ # $RANGE: range after removing Inf and NA
+ # $SUM: sum after removing Inf and NA
+ # $MEAN: mean after removing Inf and NA
+ # If data is a 2D object, provide also:
+ # $ROW_NAMES: row names
+ # $COL_NAMES: column names
+ # If data is a data frame, provide also:
+ # $COLUMN_TYPE: type of each column (typeof() value)
+ # If data is a list, provide also:
+ # $COMPARTMENT_NAMES: names of the comprtments
+ # $COMPARTMENT_TYPE: type of each compartment (typeof() value)
+ # REQUIRED PACKAGES
+ # None
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # fun_check()
+ # fun_get_message()
+ # EXAMPLE
+ # fun_info(data = 1:3)
+ # see http
+ # DEBUGGING
+ # mat1 <- matrix(1:3) ; data = env1 ; n = NULL ; warn.print = TRUE # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check",
+ "fun_get_message"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # arg with no default values
+ mandat.args <- c(
+ "data"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if( ! is.null(n)){
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = n, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){ # normally no NA
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "data",
+ # "n", # because can be NULL
+ "warn.print"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ # warn.count <- 0 # not required
+ # end warning initiation
+ # other checkings
+ if( ! is.null(n)){
+ if(n < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A POSITIVE AND NON NULL INTEGER")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(is.finite(n)){
+ # warn.count <- warn.count + 1
+ tempo.warn <- paste0("SOME COMPARTMENTS CAN BE TRUNCATED (n ARGUMENT IS ", n, ")")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end other checkings
+ # reserved word checking
+ # end reserved word checking
+ # end second round of checking and data preparation
+ # package checking
+ # end package checking
+ # main code
+ # new environment
+ env.name <- paste0("env", as.numeric(Sys.time()))
+ if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ assign(env.name, new.env())
+ assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
+ }
+ # end new environment
+ data.name <- deparse(substitute(data))
+ output <- list("NAME" = data.name)
+ tempo.try.error <- fun_get_message(data = "class(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("CLASS" = class(data))
+ output <- c(output, tempo)
+ }
+ tempo.try.error <- fun_get_message(data = "typeof(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("TYPE" = typeof(data))
+ output <- c(output, tempo)
+ }
+ tempo.try.error <- fun_get_message(data = "length(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("LENGTH" = length(data))
+ output <- c(output, tempo)
+ }
+ if(all(typeof(data) %in% c("integer", "numeric", "double")) & ! any(class(data) %in% "factor")){ # all() without na.rm -> ok because typeof(NA) is "logical" # any() without na.rm -> ok because class(NA) is "logical"
+ tempo <- list("INF.NB" = sum(is.infinite(data)))
+ output <- c(output, tempo)
+ tempo <- list("RANGE" = range(data[ ! is.infinite(data)], na.rm = TRUE))
+ output <- c(output, tempo)
+ tempo <- list("SUM" = sum(data[ ! is.infinite(data)], na.rm = TRUE))
+ output <- c(output, tempo)
+ tempo <- list("MEAN" = mean(data[ ! is.infinite(data)], na.rm = TRUE))
+ output <- c(output, tempo)
+ }
+ if(all(typeof(data) %in% c("logical", "integer", "double", "complex", "character", "list"))){ # all() without na.rm -> ok because typeof(NA) is "logical"
+ tempo.try.error <- fun_get_message(data = "is.na(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("NA.NB" = sum(is.na(data)))
+ output <- c(output, tempo)
+ }
+ }
+ tempo.try.error <- fun_get_message(data = "head(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("HEAD" = head(data))
+ output <- c(output, tempo)
+ tempo <- list("TAIL" = tail(data)) # no reason that tail() does not work if head() works
+ output <- c(output, tempo)
+ }
+ tempo.try.error <- fun_get_message(data = "dim(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ if(length(dim(data)) > 0){
+ tempo <- list("DIMENSION" = dim(data))
+ if(length(tempo[[1]]) == 2L){
+ names(tempo[[1]]) <- c("NROW", "NCOL")
+ }
+ output <- c(output, tempo)
+ }
+ }
+ if(all(class(data) == "data.frame") | all(class(data) %in% c("matrix", "array")) | all(class(data) == "table")){ # all() without na.rm -> ok because typeof(NA) is "logical"
+ if(length(dim(data)) > 1){ # to avoid 1D table
+ tempo <- list("ROW_NAMES" = dimnames(data)[[1]])
+ output <- c(output, tempo)
+ tempo <- list("COLUM_NAMES" = dimnames(data)[[2]])
+ output <- c(output, tempo)
+ }
+ }
+ tempo.try.error <- fun_get_message(data = "summary(data)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- list("SUMMARY" = summary(data))
+ output <- c(output, tempo)
+ }
+ tempo.try.error <- fun_get_message(data = "noquote(matrix(capture.output(str(data))))", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))
+ if(is.null(tempo.try.error)){
+ tempo <- capture.output(str(data))
+ tempo <- list("STRUCTURE" = noquote(matrix(tempo, dimnames = list(rep("", length(tempo)), "")))) # str() print automatically, ls.str() not but does not give the order of the data.frame
+ output <- c(output, tempo)
+ }
+ if(all(class(data) == "data.frame")){ # all() without na.rm -> ok because class(NA) is "logical"
+ tempo <- list("COLUMN_TYPE" = sapply(data, FUN = "typeof"))
+ if(any(sapply(data, FUN = "class") %in% "factor")){ # if an ordered factor is present, then sapply(data, FUN = "class") return a list but works with any(sapply(data, FUN = "class") %in% "factor") # any() without na.rm -> ok because class(NA) is "logical"
+ tempo.class <- sapply(data, FUN = "class")
+ if(any(unlist(tempo.class) %in% "ordered")){ # any() without na.rm -> ok because class(NA) is "logical"
+ tempo2 <- sapply(tempo.class, paste, collapse = " ") # paste the "ordered" factor" in "ordered factor"
+ }else{
+ tempo2 <- unlist(tempo.class)
+ }
+ tempo[["COLUMN_TYPE"]][grepl(x = tempo2, pattern = "factor")] <- tempo2[grepl(x = tempo2, pattern = "factor")]
+ }
+ output <- c(output, tempo)
+ }
+ if(all(class(data) == "list")){ # all() without na.rm -> ok because class(NA) is "logical"
+ tempo <- list("COMPARTMENT_NAMES" = names(data))
+ output <- c(output, tempo)
+ tempo <- list("COMPARTMENT_TYPE" = sapply(data, FUN = "typeof"))
+ if(any(unlist(sapply(data, FUN = "class")) %in% "factor")){ # if an ordered factor is present, then sapply(data, FUN = "class") return a list but works with any(sapply(data, FUN = "class") %in% "factor") # any() without na.rm -> ok because class(NA) is "logical"
+ tempo.class <- sapply(data, FUN = "class")
+ if(any(unlist(tempo.class) %in% "ordered")){ # any() without na.rm -> ok because class(NA) is "logical"
+ tempo2 <- sapply(tempo.class, paste, collapse = " ") # paste the "ordered" factor" in "ordered factor"
+ }else{
+ tempo2 <- unlist(tempo.class)
+ }
+ tempo[["COMPARTMENT_TYPE"]][grepl(x = tempo2, pattern = "factor")] <- tempo2[grepl(x = tempo2, pattern = "factor")]
+ }
+ output <- c(output, tempo)
+ }
+ if( ! is.null(n)){
+ output[names(output) != "STRUCTURE"] <- lapply(X = output[names(output) != "STRUCTURE"], FUN = head, n = n, simplify = FALSE)
+ }
+ # output
+ if(warn.print == FALSE){
+ output <- c(output, WARNING = warn)
+ }else if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ return(output)
+ # end output
+ # end main code
}
@@ -912,62 +912,62 @@ return(output)
fun_head <- function(
-data1,
-n = 6,
-side = "l"
+ data1,
+ n = 6,
+ side = "l"
){
-# AIM
-# as head() but display the left or right head of big 2D objects
-# ARGUMENTS
-# data1: any object but more dedicated for matrix, data frame or table
-# n: as in head() but for for matrix, data frame or table, number of dimension to print (10 means 10 rows and columns)
-# side: either "l" or "r" for the left or right side of the 2D object (only for matrix, data frame or table)
-# BEWARE: other arguments of head() not used
-# RETURN
-# the head
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# obs1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:6], LETTERS[1:5])) ; obs1 ; fun_head(obs1, 3)
-# DEBUGGING
-# data1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = side, options = c("l", "r"), length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
-return(head(data1, n))
-}else{
-obs.dim <- dim(data1)
-row <- 1:ifelse(obs.dim[1] < n, obs.dim[1], n)
-if(side == "l"){
-col <- 1:ifelse(obs.dim[2] < n, obs.dim[2], n)
-}
-if(side == "r"){
-col <- ifelse(obs.dim[2] < n, 1, obs.dim[2] - n + 1):obs.dim[2]
-}
-return(data1[row, col])
-}
+ # AIM
+ # as head() but display the left or right head of big 2D objects
+ # ARGUMENTS
+ # data1: any object but more dedicated for matrix, data frame or table
+ # n: as in head() but for for matrix, data frame or table, number of dimension to print (10 means 10 rows and columns)
+ # side: either "l" or "r" for the left or right side of the 2D object (only for matrix, data frame or table)
+ # BEWARE: other arguments of head() not used
+ # RETURN
+ # the head
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # obs1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:6], LETTERS[1:5])) ; obs1 ; fun_head(obs1, 3)
+ # DEBUGGING
+ # data1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = side, options = c("l", "r"), length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
+ return(head(data1, n))
+ }else{
+ obs.dim <- dim(data1)
+ row <- 1:ifelse(obs.dim[1] < n, obs.dim[1], n)
+ if(side == "l"){
+ col <- 1:ifelse(obs.dim[2] < n, obs.dim[2], n)
+ }
+ if(side == "r"){
+ col <- ifelse(obs.dim[2] < n, 1, obs.dim[2] - n + 1):obs.dim[2]
+ }
+ return(data1[row, col])
+ }
}
@@ -975,62 +975,62 @@ return(data1[row, col])
fun_tail <- function(
-data1,
-n = 6,
-side = "l"
+ data1,
+ n = 6,
+ side = "l"
){
-# AIM
-# as tail() but display the left or right head of big 2D objects
-# ARGUMENTS
-# data1: any object but more dedicated for matrix, data frame or table
-# n: as in tail() but for for matrix, data frame or table, number of dimension to print (10 means 10 rows and columns)
-# side: either "l" or "r" for the left or right side of the 2D object (only for matrix, data frame or table)
-# BEWARE: other arguments of tail() not used
-# RETURN
-# the tail
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# obs1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:6], LETTERS[1:5])) ; obs1 ; fun_tail(obs1, 3, "r")
-# DEBUGGING
-# data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = side, options = c("l", "r"), length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
-return(tail(data1, n))
-}else{
-obs.dim <- dim(data1)
-row <- ifelse(obs.dim[1] < n, 1, obs.dim[1] - n + 1):obs.dim[1]
-if(side == "l"){
-col <- 1:ifelse(obs.dim[2] < n, obs.dim[2], n)
-}
-if(side == "r"){
-col <- ifelse(obs.dim[2] < n, 1, obs.dim[2] - n + 1):obs.dim[2]
-}
-return(data1[row, col])
-}
+ # AIM
+ # as tail() but display the left or right head of big 2D objects
+ # ARGUMENTS
+ # data1: any object but more dedicated for matrix, data frame or table
+ # n: as in tail() but for for matrix, data frame or table, number of dimension to print (10 means 10 rows and columns)
+ # side: either "l" or "r" for the left or right side of the 2D object (only for matrix, data frame or table)
+ # BEWARE: other arguments of tail() not used
+ # RETURN
+ # the tail
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # obs1 = matrix(1:30, ncol = 5, dimnames = list(letters[1:6], LETTERS[1:5])) ; obs1 ; fun_tail(obs1, 3, "r")
+ # DEBUGGING
+ # data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = side, options = c("l", "r"), length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
+ return(tail(data1, n))
+ }else{
+ obs.dim <- dim(data1)
+ row <- ifelse(obs.dim[1] < n, 1, obs.dim[1] - n + 1):obs.dim[1]
+ if(side == "l"){
+ col <- 1:ifelse(obs.dim[2] < n, obs.dim[2], n)
+ }
+ if(side == "r"){
+ col <- ifelse(obs.dim[2] < n, 1, obs.dim[2] - n + 1):obs.dim[2]
+ }
+ return(data1[row, col])
+ }
}
@@ -1038,12371 +1038,12377 @@ return(data1[row, col])
fun_comp_1d <- function(data1, data2){
-# AIM
-# compare two 1D datasets (vector or factor or 1D table, or 1D matrix or 1D array) of the same class or not. Check and report in a list if the 2 datasets have:
-# same class
-# common elements
-# common element names (except factors)
-# common levels (factors only)
-# ARGUMENTS
-# data1: vector or factor or 1D table, or 1D matrix or 1D array
-# data2: vector or factor or 1D table, or 1D matrix or 1D array
-# RETURN
-# a list containing:
-# $same.class: logical. Are class identical?
-# $class: class of the 2 datasets (NULL otherwise)
-# $same.length: logical. Are number of elements identical?
-# $length: number of elements in the 2 datasets (NULL otherwise)
-# $same.levels: logical. Are levels identical? NULL if data1 and data2 are not factors
-# $levels: levels of the 2 datasets if identical (NULL otherwise or NULL if data1 and data2 are not factors)
-# $any.id.levels: logical. Is there any identical levels? (NULL if data1 and data2 are not factors)
-# $same.levels.pos1: positions, in data1, of the levels identical in data2 (NULL otherwise or NULL if data1 and data2 are not factors)
-# $same.levels.pos2: positions, in data2, of the levels identical in data1 (NULL otherwise or NULL if data1 and data2 are not factors)
-# $same.levels.match1: positions, in data2, of the levels that match the levels in data1, as given by match(data1, data2) (NULL otherwise or NULL if data1 and data2 are not factors)
-# $same.levels.match2: positions, in data1, of the levels that match the levels in data2, as given by match(data1, data2) (NULL otherwise or NULL if data1 and data2 are not factors)
-# $common.levels: common levels between data1 and data2 (can be a subset of $levels or not). NULL if no common levels or if data1 and data2 are not factors
-# $same.names: logical. Are element names identical? NULL if data1 and data2 have no names
-# $name: name of elements of the 2 datasets if identical (NULL otherwise)
-# $any.id.name: logical. Is there any element names identical ?
-# $same.names.pos1: positions, in data1, of the element names identical in data2. NULL if no identical names
-# $same.names.pos2: positions, in data2, of the elements names identical in data1. NULL if no identical names
-# $same.names.match1: positions, in data2, of the names that match the names in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.names.match2: positions, in data1, of the names that match the names in data2, as given by match(data1, data2) (NULL otherwise)
-# $common.names: common element names between data1 and data2 (can be a subset of $name or not). NULL if no common element names
-# $any.id.element: logical. is there any identical elements ?
-# $same.elements.pos1: positions, in data1, of the elements identical in data2. NULL if no identical elements
-# $same.elements.pos2: positions, in data2, of the elements identical in data1. NULL if no identical elements
-# $same.elements.match1: positions, in data2, of the elements that match the elements in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.elements.match2: positions, in data1, of the elements that match the elements in data2, as given by match(data1, data2) (NULL otherwise)
-# $common.elements: common elements between data1 and data2. NULL if no common elements
-# $same.order: logical. Are all elements in the same order? TRUE or FALSE if elements of data1 and data2 are identical but not necessary in the same order. NULL otherwise (different length for instance)
-# $order1: order of all elements of data1. NULL if $same.order is FALSE
-# $order2: order of all elements of data2. NULL if $same.order is FALSE
-# $identical.object: logical. Are objects identical (kind of object, element names, content, including content order)?
-# $identical.content: logical. Are content objects identical (identical elements, including order, excluding kind of object and element names)?
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# none
-# EXAMPLES
-# obs1 = 1:5 ; obs2 = 1:5 ; names(obs1) <- LETTERS[1:5] ; names(obs2) <- LETTERS[1:5] ; fun_comp_1d(obs1, obs2)
-# obs1 = 1:5 ; obs2 = 1:5 ; names(obs1) <- LETTERS[1:5] ; fun_comp_1d(obs1, obs2)
-# obs1 = 1:5 ; obs2 = 3:6 ; names(obs1) <- LETTERS[1:5] ; names(obs2) <- LETTERS[1:4] ; fun_comp_1d(obs1, obs2)
-# obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[1:5]) ; fun_comp_1d(obs1, obs2)
-# obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[10:11]) ; fun_comp_1d(obs1, obs2)
-# obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[4:7]) ; fun_comp_1d(obs1, obs2)
-# obs1 = factor(c(LETTERS[1:4], "E")) ; obs2 = factor(c(LETTERS[1:4], "F")) ; fun_comp_1d(obs1, obs2)
-# obs1 = 1:5 ; obs2 = factor(LETTERS[1:5]) ; fun_comp_1d(obs1, obs2)
-# obs1 = 1:5 ; obs2 = 1.1:6.1 ; fun_comp_1d(obs1, obs2)
-# obs1 = as.table(1:5); obs2 = as.table(1:5) ; fun_comp_1d(obs1, obs2)
-# obs1 = as.table(1:5); obs2 = 1:5 ; fun_comp_1d(obs1, obs2)
-# DEBUGGING
-# data1 = 1:5 ; data2 = 1:5 ; names(data1) <- LETTERS[1:5] ; names(data2) <- LETTERS[1:5] # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# argument checking
-if( ! any(class(data1) %in% c("logical", "integer", "numeric", "character", "factor", "table"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A NON NULL VECTOR, FACTOR OR 1D TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(all(class(data1) %in% "table")){
-if(length(dim(data1)) > 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A 1D TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ # AIM
+ # compare two 1D datasets (vector or factor or 1D table, or 1D matrix or 1D array) of the same class or not. Check and report in a list if the 2 datasets have:
+ # same class
+ # common elements
+ # common element names (except factors)
+ # common levels (factors only)
+ # ARGUMENTS
+ # data1: vector or factor or 1D table, or 1D matrix or 1D array
+ # data2: vector or factor or 1D table, or 1D matrix or 1D array
+ # RETURN
+ # a list containing:
+ # $same.class: logical. Are class identical?
+ # $class: class of the 2 datasets (NULL otherwise)
+ # $same.length: logical. Are number of elements identical?
+ # $length: number of elements in the 2 datasets (NULL otherwise)
+ # $same.levels: logical. Are levels identical? NULL if data1 and data2 are not factors
+ # $levels: levels of the 2 datasets if identical (NULL otherwise or NULL if data1 and data2 are not factors)
+ # $any.id.levels: logical. Is there any identical levels? (NULL if data1 and data2 are not factors)
+ # $same.levels.pos1: positions, in data1, of the levels identical in data2 (NULL otherwise or NULL if data1 and data2 are not factors)
+ # $same.levels.pos2: positions, in data2, of the levels identical in data1 (NULL otherwise or NULL if data1 and data2 are not factors)
+ # $same.levels.match1: positions, in data2, of the levels that match the levels in data1, as given by match(data1, data2) (NULL otherwise or NULL if data1 and data2 are not factors)
+ # $same.levels.match2: positions, in data1, of the levels that match the levels in data2, as given by match(data1, data2) (NULL otherwise or NULL if data1 and data2 are not factors)
+ # $common.levels: common levels between data1 and data2 (can be a subset of $levels or not). NULL if no common levels or if data1 and data2 are not factors
+ # $same.names: logical. Are element names identical? NULL if data1 and data2 have no names
+ # $name: name of elements of the 2 datasets if identical (NULL otherwise)
+ # $any.id.name: logical. Is there any element names identical ?
+ # $same.names.pos1: positions, in data1, of the element names identical in data2. NULL if no identical names
+ # $same.names.pos2: positions, in data2, of the elements names identical in data1. NULL if no identical names
+ # $same.names.match1: positions, in data2, of the names that match the names in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.names.match2: positions, in data1, of the names that match the names in data2, as given by match(data1, data2) (NULL otherwise)
+ # $common.names: common element names between data1 and data2 (can be a subset of $name or not). NULL if no common element names
+ # $any.id.element: logical. is there any identical elements ?
+ # $same.elements.pos1: positions, in data1, of the elements identical in data2. NULL if no identical elements
+ # $same.elements.pos2: positions, in data2, of the elements identical in data1. NULL if no identical elements
+ # $same.elements.match1: positions, in data2, of the elements that match the elements in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.elements.match2: positions, in data1, of the elements that match the elements in data2, as given by match(data1, data2) (NULL otherwise)
+ # $common.elements: common elements between data1 and data2. NULL if no common elements
+ # $same.order: logical. Are all elements in the same order? TRUE or FALSE if elements of data1 and data2 are identical but not necessary in the same order. NULL otherwise (different length for instance)
+ # $order1: order of all elements of data1. NULL if $same.order is FALSE
+ # $order2: order of all elements of data2. NULL if $same.order is FALSE
+ # $identical.object: logical. Are objects identical (kind of object, element names, content, including content order)?
+ # $identical.content: logical. Are content objects identical (identical elements, including order, excluding kind of object and element names)?
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # none
+ # EXAMPLES
+ # obs1 = 1:5 ; obs2 = 1:5 ; names(obs1) <- LETTERS[1:5] ; names(obs2) <- LETTERS[1:5] ; fun_comp_1d(obs1, obs2)
+ # obs1 = 1:5 ; obs2 = 1:5 ; names(obs1) <- LETTERS[1:5] ; fun_comp_1d(obs1, obs2)
+ # obs1 = 1:5 ; obs2 = 3:6 ; names(obs1) <- LETTERS[1:5] ; names(obs2) <- LETTERS[1:4] ; fun_comp_1d(obs1, obs2)
+ # obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[1:5]) ; fun_comp_1d(obs1, obs2)
+ # obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[10:11]) ; fun_comp_1d(obs1, obs2)
+ # obs1 = factor(LETTERS[1:5]) ; obs2 = factor(LETTERS[4:7]) ; fun_comp_1d(obs1, obs2)
+ # obs1 = factor(c(LETTERS[1:4], "E")) ; obs2 = factor(c(LETTERS[1:4], "F")) ; fun_comp_1d(obs1, obs2)
+ # obs1 = 1:5 ; obs2 = factor(LETTERS[1:5]) ; fun_comp_1d(obs1, obs2)
+ # obs1 = 1:5 ; obs2 = 1.1:6.1 ; fun_comp_1d(obs1, obs2)
+ # obs1 = as.table(1:5); obs2 = as.table(1:5) ; fun_comp_1d(obs1, obs2)
+ # obs1 = as.table(1:5); obs2 = 1:5 ; fun_comp_1d(obs1, obs2)
+ # DEBUGGING
+ # data1 = 1:5 ; data2 = 1:5 ; names(data1) <- LETTERS[1:5] ; names(data2) <- LETTERS[1:5] # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # argument checking
+ if( ! any(class(data1) %in% c("logical", "integer", "numeric", "character", "factor", "table"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A NON NULL VECTOR, FACTOR OR 1D TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(all(class(data1) %in% "table")){
+ if(length(dim(data1)) > 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A 1D TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! any(class(data2) %in% c("logical", "integer", "numeric", "character", "factor", "table"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A NON NULL VECTOR, FACTOR OR 1D TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(all(class(data2) %in% "table")){
+ if(length(dim(data2)) > 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A 1D TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
+ # end argument checking
+ # main code
+ same.class <- FALSE
+ class <- NULL
+ same.length <- FALSE
+ length <- NULL
+ same.levels <- NULL # not FALSE to deal with no factors
+ levels <- NULL
+ any.id.levels <- FALSE
+ same.levels.pos1 <- NULL
+ same.levels.pos2 <- NULL
+ same.levels.match1 <- NULL
+ same.levels.match2 <- NULL
+ common.levels <- NULL
+ same.names <- NULL # not FALSE to deal with absence of name
+ name <- NULL
+ any.id.name <- FALSE
+ same.names.pos1 <- NULL
+ same.names.pos2 <- NULL
+ same.names.match1 <- NULL
+ same.names.match2 <- NULL
+ common.names <- NULL
+ any.id.element <- FALSE
+ same.elements.pos1 <- NULL
+ same.elements.pos2 <- NULL
+ same.elements.match1 <- NULL
+ same.elements.match2 <- NULL
+ common.elements <- NULL
+ same.order <- NULL
+ order1 <- NULL
+ order2 <- NULL
+ identical.object <- FALSE
+ identical.content <- FALSE
+ if(identical(data1, data2)){
+ same.class <- TRUE
+ class <- class(data1)
+ same.length <- TRUE
+ length <- length(data1)
+ if(any(class(data1) %in% "factor")){
+ same.levels <- TRUE
+ levels <- levels(data1)
+ any.id.levels <- TRUE
+ same.levels.pos1 <- 1:length(levels(data1))
+ same.levels.pos2 <- 1:length(levels(data2))
+ same.levels.match1 <- 1:length(levels(data1))
+ same.levels.match2 <- 1:length(levels(data2))
+ common.levels <- levels(data1)
+ }
+ if( ! is.null(names(data1))){
+ same.names <- TRUE
+ name <- names(data1)
+ any.id.name <- TRUE
+ same.names.pos1 <- 1:length(data1)
+ same.names.pos2 <- 1:length(data2)
+ same.names.match1 <- 1:length(data1)
+ same.names.match2 <- 1:length(data2)
+ common.names <- names(data1)
+ }
+ any.id.element <- TRUE
+ same.elements.pos1 <- 1:length(data1)
+ same.elements.pos2 <- 1:length(data2)
+ same.elements.match1 <- 1:length(data1)
+ same.elements.match2 <- 1:length(data2)
+ common.elements <- data1
+ same.order <- TRUE
+ order1 <- order(data1)
+ order2 <- order(data2)
+ identical.object <- TRUE
+ identical.content <- TRUE
+ }else{
+ if(identical(class(data1), class(data2))){
+ same.class <- TRUE
+ class <- class(data1)
+ }
+ if(identical(length(data1), length(data2))){
+ same.length<- TRUE
+ length <- length(data1)
+ }
+ if(any(class(data1) %in% "factor") & any(class(data2) %in% "factor")){
+ if(identical(levels(data1), levels(data2))){
+ same.levels <- TRUE
+ levels <- levels(data1)
+ }else{
+ same.levels <- FALSE
+ }
+ if(any(levels(data1) %in% levels(data2))){
+ any.id.levels <- TRUE
+ same.levels.pos1 <- which(levels(data1) %in% levels(data2))
+ same.levels.match1 <- match(levels(data1), levels(data2))
+ }
+ if(any(levels(data2) %in% levels(data1))){
+ any.id.levels <- TRUE
+ same.levels.pos2 <- which(levels(data2) %in% levels(data1))
+ same.levels.match2 <- match(levels(data2), levels(data1))
+ }
+ if(any.id.levels == TRUE){
+ common.levels <- unique(c(levels(data1)[same.levels.pos1], levels(data2)[same.levels.pos2]))
+ }
+ }
+ if(any(class(data1) %in% "factor")){ # to compare content
+ data1 <- as.character(data1)
+ }
+ if(any(class(data2) %in% "factor")){ # to compare content
+ data2 <- as.character(data2)
+ }
+ if( ! (is.null(names(data1)) & is.null(names(data2)))){
+ if(identical(names(data1), names(data2))){
+ same.names <- TRUE
+ name <- names(data1)
+ }else{
+ same.names <- FALSE
+ }
+ if(any(names(data1) %in% names(data2))){
+ any.id.name <- TRUE
+ same.names.pos1 <- which(names(data1) %in% names(data2))
+ same.names.match1 <- match(names(data1), names(data2))
+ }
+ if(any(names(data2) %in% names(data1))){
+ any.id.name <- TRUE
+ same.names.pos2 <- which(names(data2) %in% names(data1))
+ same.names.match2 <- match(names(data2), names(data1))
+ }
+ if(any.id.name == TRUE){
+ common.names <- unique(c(names(data1)[same.names.pos1], names(data2)[same.names.pos2]))
+ }
+ }
+ names(data1) <- NULL # names solved -> to do not be disturbed by names
+ names(data2) <- NULL # names solved -> to do not be disturbed by names
+ if(any(data1 %in% data2)){
+ any.id.element <- TRUE
+ same.elements.pos1 <- which(data1 %in% data2)
+ same.elements.match1 <- match(data1, data2)
+ }
+ if(any(data2 %in% data1)){
+ any.id.element <- TRUE
+ same.elements.pos2 <- which(data2 %in% data1)
+ same.elements.match2 <- match(data2, data1)
+ }
+ if(any.id.element == TRUE){
+ common.elements <- unique(c(data1[same.elements.pos1], data2[same.elements.pos2]))
+ }
+ if(identical(data1, data2)){
+ identical.content <- TRUE
+ same.order <- TRUE
+ }else if(identical(sort(data1), sort(data2))){
+ same.order <- FALSE
+ order1 <- order(data1)
+ order2 <- order(data2)
+ }
+ }
+ output <- list(same.class = same.class, class = class, same.length = same.length, length = length, same.levels = same.levels, levels = levels, any.id.levels = any.id.levels, same.levels.pos1 = same.levels.pos1, same.levels.pos2 = same.levels.pos2, same.levels.match1 = same.levels.match1, same.levels.match2 = same.levels.match2, common.levels = common.levels, same.names = same.names, name = name, any.id.name = any.id.name, same.names.pos1 = same.names.pos1, same.names.pos2 = same.names.pos2, same.names.match1 = same.names.match1, same.names.match2 = same.names.match2, common.names = common.names, any.id.element = any.id.element, same.elements.pos1 = same.elements.pos1, same.elements.pos2 = same.elements.pos2, same.elements.match1 = same.elements.match1, same.elements.match2 = same.elements.match2, common.elements = common.elements, same.order = same.order, order1 = order1, order2 = order2, identical.object = identical.object, identical.content = identical.content)
+ return(output)
}
+
+
+######## fun_comp_2d() #### comparison of two 2D datasets (row & col names, dimensions, etc.)
+
+
+fun_comp_2d <- function(data1, data2){
+ # AIM
+ # compare two 2D datasets of the same class or not. Check and report in a list if the 2 datasets have:
+ # same class
+ # common row names
+ # common column names
+ # same row number
+ # same column number
+ # potential identical rows between the 2 datasets
+ # potential identical columns between the 2 datasets
+ # WARNINGS
+ # For data frames: content are compared after conversion of content into characters. This means that the comparison of the content of data frame, either row to row, or column to column, does not take into account the mode in the different columns. This concern the results in $any.id.row, $same.row.pos1, $same.row.pos2, $same.row.match1, $same.row.match2, $any.id.col, $same.row.col1, $same.row.col2, $same.col.match1, $same.col.match2 and $identical.content result
+ # "TOO BIG FOR EVALUATION" returned in $same.row.pos1, $same.row.pos2, $same.row.match1 and $same.row.match2 when nrow(data1) * nrow(data2) > 1e6 and $any.id.row remains NULL
+ # "TOO BIG FOR EVALUATION" returned in $same.row.col1, $same.row.col2, $same.col.match1 and $same.col.match2 when ncol(data1) * ncol(data2) > 1e6 and $any.id.col remains NULL
+ # ARGUMENTS
+ # data1: matrix, data frame or table
+ # data2: matrix, data frame or table
+ # RETURN
+ # a list containing:
+ # $same.class: logical. Are class identical ?
+ # $class: classes of the 2 datasets (NULL otherwise)
+ # $same.dim: logical. Are dimension identical ?
+ # $dim: dimension of the 2 datasets (NULL otherwise)
+ # $same.row.nb: logical. Are number of rows identical ?
+ # $row.nb: nb of rows of the 2 datasets if identical (NULL otherwise)
+ # $same.col.nb: logical. Are number of columns identical ?
+ # $col.nb: nb of columns of the 2 datasets if identical (NULL otherwise)
+ # $same.row.name: logical. Are row names identical ? NULL if no row names in the two 2D datasets
+ # $row.name: name of rows of the 2 datasets if identical (NULL otherwise)
+ # $any.id.row.name: logical. Is there any row names identical ? NULL if no row names in the two 2D datasets
+ # $same.row.names.pos1: positions, in data1, of the row names identical in data2
+ # $same.row.names.pos2: positions, in data2, of the row names identical in data1
+ # $same.row.names.match1: positions, in data2, of the row names that match the row names in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.row.names.match2: positions, in data1, of the row names that match the row names in data2, as given by match(data1, data2) (NULL otherwise)
+ # $common.row.names: common row names between data1 and data2 (can be a subset of $name or not). NULL if no common row names
+ # $same.col.name: logical. Are column names identical ? NULL if no col names in the two 2D datasets
+ # $col.name: name of columns of the 2 datasets if identical (NULL otherwise)
+ # $any.id.col.name: logical. Is there any column names identical ? NULL if no col names in the two 2D datasets
+ # $same.col.names.pos1: positions, in data1, of the column names identical in data2
+ # $same.col.names.pos2: positions, in data2, of the column names identical in data1
+ # $same.col.names.match1: positions, in data2, of the column names that match the column names in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.col.names.match2: positions, in data1, of the column names that match the column names in data2, as given by match(data1, data2) (NULL otherwise)
+ # $common.col.names: common column names between data1 and data2 (can be a subset of $name or not). NULL if no common column names
+ # $any.id.row: logical. is there identical rows (not considering row names)? NULL if nrow(data1) * nrow(data2) > 1e10
+ # $same.row.pos1: positions, in data1, of the rows identical in data2 (not considering row names). Return "TOO BIG FOR EVALUATION" if nrow(data1) * nrow(data2) > 1e10
+ # $same.row.pos2: positions, in data2, of the rows identical in data1 (not considering row names). Return "TOO BIG FOR EVALUATION" if nrow(data1) * nrow(data2) > 1e10
+ # $same.row.match1: positions, in data2, of the rows that match the rows in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.row.match2: positions, in data1, of the rows that match the rows in data2, as given by match(data1, data2) (NULL otherwise)
+ # $any.id.col: logical. is there identical columns (not considering column names)? NULL if ncol(data1) * ncol(data2) > 1e10
+ # $same.col.pos1: position in data1 of the cols identical in data2 (not considering column names). Return "TOO BIG FOR EVALUATION" if ncol(data1) * ncol(data2) > 1e10
+ # $same.col.pos2: position in data2 of the cols identical in data1 (not considering column names). Return "TOO BIG FOR EVALUATION" if ncol(data1) * ncol(data2) > 1e10
+ # $same.col.match1: positions, in data2, of the columns that match the columns in data1, as given by match(data1, data2) (NULL otherwise)
+ # $same.row.match2: positions, in data1, of the columns that match the columns in data2, as given by match(data1, data2) (NULL otherwise)
+ # $identical.object: logical. Are objects identical (including row & column names)?
+ # $identical.content: logical. Are content objects identical (identical excluding row & column names)?
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # none
+ # EXAMPLES
+ # obs1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = as.data.frame(matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])), stringsAsFactors = TRUE) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
+ # obs1 = matrix(101:110, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
+ # large matrices
+ # obs1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; obs2 = matrix(as.integer((1:1e6)+1e6/5), ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; head(obs1) ; head(obs2) ; fun_comp_2d(obs1, obs2)
+ # WARNING: when comparing content (rows, columns, or total), double and integer data are considered as different -> double(1) != integer(1)
+ # obs1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; obs2 = matrix((1:1e6)+1e6/5, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; head(obs1) ; head(obs2) ; fun_comp_2d(obs1, obs2)
+ # Matrices: same row conten tand same row names
+ # obs1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4]))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
+ # Matrices: same row content but not same row names -> works: same content is identified
+ # obs1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("x", "z", "y"), c(LETTERS[1:2], "k", LETTERS[5:4]))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
+ # obs1 = t(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = t(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
+ # Data frames: same row content and same row names, not same mode between columns
+ # obs1 = as.data.frame(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = as.data.frame(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1[, 5] <- as.character(obs1[, 5]) ; obs2[, 5] <- as.character(obs2[, 5]) ; obs1 ; obs2 ; str(obs1) ; str(obs2) ; fun_comp_2d(obs1, obs2)
+ # Data frames: same row content but not same row names -> works: same content is identified
+ # obs1 = as.data.frame(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = as.data.frame(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("x", "z", "y"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1[, 5] <- as.character(obs1[, 5]) ; obs2[, 5] <- as.character(obs2[, 5]) ; obs1 ; obs2 ; str(obs1) ; str(obs2) ; fun_comp_2d(obs1, obs2)
+ # DEBUGGING
+ # data1 = matrix(1:10, ncol = 5) ; data2 = matrix(1:10, ncol = 5) # for function debugging
+ # data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5) # for function debugging
+ # data1 = matrix(1:15, byrow = TRUE, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # data1 = matrix(1:15, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # data1 = matrix(1:15, ncol = 5, dimnames = list(paste0("A", letters[1:3]), LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # data1 = matrix(1:15, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:12, ncol = 4, dimnames = list(letters[1:3], LETTERS[1:4])) # for function debugging
+ # data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(101:110, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
+ # data1 = data.frame(a = 1:3, b= letters[1:3], row.names = LETTERS[1:3], stringsAsFactors = TRUE) ; data2 = data.frame(A = 1:3, B= letters[1:3], stringsAsFactors = TRUE) # for function debugging
+ # data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = as.data.frame(matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])), stringsAsFactors = TRUE) # for function debugging
+ # data1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4]))) # for function debugging
+ # data1 = table(Exp1 = c("A", "A", "A", "B", "B", "B"), Exp2 = c("A1", "B1", "A1", "C1", "C1", "B1")) ; data2 = data.frame(A = 1:3, B= letters[1:3], stringsAsFactors = TRUE) # for function debugging
+ # data1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; data2 = matrix((1:1e6)+1e6/5, ncol = 5, dimnames = list(NULL, LETTERS[1:5]))
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # argument checking
+ if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A MATRIX, DATA FRAME OR TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (any(class(data2) %in% c("data.frame", "table")) | all(class(data2) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data2) %in% c("matrix", "data.frame", "table"))
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A MATRIX, DATA FRAME OR TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
+ # end argument checking
+ # main code
+ same.class <- NULL
+ class <- NULL
+ same.dim <- NULL
+ dim <- NULL
+ same.row.nb <- NULL
+ row.nb <- NULL
+ same.col.nb <- NULL
+ col.nb <- NULL
+ same.row.name <- NULL
+ row.name <- NULL
+ any.id.row.name <- NULL
+ same.row.names.pos1 <- NULL
+ same.row.names.pos2 <- NULL
+ same.row.names.match1 <- NULL
+ same.row.names.match2 <- NULL
+ common.row.names <- NULL
+ same.col.name <- NULL
+ any.id.col.name <- NULL
+ same.col.names.pos1 <- NULL
+ same.col.names.pos2 <- NULL
+ same.col.names.match1 <- NULL
+ same.col.names.match2 <- NULL
+ common.col.names <- NULL
+ col.name <- NULL
+ any.id.row <- NULL
+ same.row.pos1 <- NULL
+ same.row.pos2 <- NULL
+ same.row.match1 <- NULL
+ same.row.match2 <- NULL
+ any.id.col <- NULL
+ same.col.pos1 <- NULL
+ same.col.pos2 <- NULL
+ same.col.match1 <- NULL
+ same.col.match2 <- NULL
+ identical.object <- NULL
+ identical.content <- NULL
+ if(identical(data1, data2) & (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
+ same.class <- TRUE
+ class <- class(data1)
+ same.dim <- TRUE
+ dim <- dim(data1)
+ same.row.nb <- TRUE
+ row.nb <- nrow(data1)
+ same.col.nb <- TRUE
+ col.nb <- ncol(data1)
+ same.row.name <- TRUE
+ row.name <- dimnames(data1)[[1]]
+ any.id.row.name <- TRUE
+ same.row.names.pos1 <- 1:row.nb
+ same.row.names.pos2 <- 1:row.nb
+ same.row.names.match1 <- 1:row.nb
+ same.row.names.match2 <- 1:row.nb
+ common.row.names <- dimnames(data1)[[1]]
+ same.col.name <- TRUE
+ col.name <- dimnames(data1)[[2]]
+ any.id.col.name <- TRUE
+ same.col.names.pos1 <- 1:col.nb
+ same.col.names.pos2 <- 1:col.nb
+ same.col.names.match1 <- 1:col.nb
+ same.col.names.match2 <- 1:col.nb
+ common.col.names <- dimnames(data1)[[2]]
+ any.id.row <- TRUE
+ same.row.pos1 <- 1:row.nb
+ same.row.pos2 <- 1:row.nb
+ same.row.match1 <- 1:row.nb
+ same.row.match2 <- 1:row.nb
+ any.id.col <- TRUE
+ same.col.pos1 <- 1:col.nb
+ same.col.pos2 <- 1:col.nb
+ same.col.match1 <- 1:col.nb
+ same.col.match2 <- 1:col.nb
+ identical.object <- TRUE
+ identical.content <- TRUE
+ }else{
+ identical.object <- FALSE
+ if(all(class(data1) == "table") & length(dim(data1)) == 1L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT IS A 1D TABLE. USE THE fun_comp_1d FUNCTION")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(all(class(data2) == "table") & length(dim(data2)) == 1L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT IS A 1D TABLE. USE THE fun_comp_1d FUNCTION")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! identical(class(data1), class(data2))){
+ same.class <- FALSE
+ }else if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 AND data2 ARGUMENTS MUST BE EITHER MATRIX, DATA FRAME OR TABLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ same.class <- TRUE
+ class <- class(data1)
+ }
+ if( ! identical(dim(data1), dim(data2))){
+ same.dim <- FALSE
+ }else{
+ same.dim <- TRUE
+ dim <- dim(data1)
+ }
+ if( ! identical(nrow(data1), nrow(data2))){
+ same.row.nb <- FALSE
+ }else{
+ same.row.nb <- TRUE
+ row.nb <- nrow(data1)
+ }
+ if( ! identical(ncol(data1), ncol(data2))){
+ same.col.nb <- FALSE
+ }else{
+ same.col.nb <- TRUE
+ col.nb <- ncol(data1)
+ }
+ # row and col names
+ if(is.null(dimnames(data1)) & is.null(dimnames(data2))){
+ same.row.name <- NULL # but already NULL
+ same.col.name <- NULL # but already NULL
+ # other row names param remain NULL
+ }else if((is.null(dimnames(data1)) & ! is.null(dimnames(data2))) | ( ! is.null(dimnames(data1)) & is.null(dimnames(data2)))){
+ same.row.name <- FALSE
+ same.col.name <- FALSE
+ any.id.row.name <- FALSE
+ any.id.col.name <- FALSE
+ # other row names param remain NULL
+ }else{
+ # row names
+ if(is.null(dimnames(data1)[[1]]) & is.null(dimnames(data2)[[1]])){
+ same.row.name <- NULL # but already NULL
+ # other row names param remain NULL
+ }else if((is.null(dimnames(data1)[[1]]) & ! is.null(dimnames(data2)[[1]])) | ( ! is.null(dimnames(data1)[[1]]) & is.null(dimnames(data2)[[1]]))){
+ same.row.name <- FALSE
+ any.id.row.name <- FALSE
+ # other row names param remain NULL
+ }else if(identical(dimnames(data1)[[1]], dimnames(data2)[[1]])){
+ same.row.name <- TRUE
+ row.name <- dimnames(data1)[[1]]
+ any.id.row.name <- TRUE
+ same.row.names.pos1 <- 1:nrow(data1)
+ same.row.names.pos2 <- 1:nrow(data1)
+ same.row.names.match1 <- 1:nrow(data1)
+ same.row.names.match2 <- 1:nrow(data1)
+ common.row.names <- dimnames(data1)[[1]]
+ }else{
+ same.row.name <- FALSE
+ any.id.row.name <- FALSE
+ if(any(dimnames(data1)[[1]] %in% dimnames(data2)[[1]])){
+ any.id.row.name <- TRUE
+ same.row.names.pos1 <- which(dimnames(data1)[[1]] %in% dimnames(data2)[[1]])
+ same.row.names.match1 <- match(dimnames(data1)[[1]], dimnames(data2)[[1]])
+ }
+ if(any(dimnames(data2)[[1]] %in% dimnames(data1)[[1]])){
+ any.id.row.name <- TRUE
+ same.row.names.pos2 <- which(dimnames(data2)[[1]] %in% dimnames(data1)[[1]])
+ same.row.names.match2 <- match(dimnames(data2)[[1]], dimnames(data1)[[1]])
+ }
+ if(any.id.row.name == TRUE){
+ common.row.names <- unique(c(dimnames(data1)[[1]][same.row.names.pos1], dimnames(data2)[[1]][same.row.names.pos2]))
+ }
+ }
+ # col names
+ if(is.null(dimnames(data1)[[2]]) & is.null(dimnames(data2)[[2]])){
+ same.col.name <- NULL # but already NULL
+ # other col names param remain NULL
+ }else if((is.null(dimnames(data1)[[2]]) & ! is.null(dimnames(data2)[[2]])) | ( ! is.null(dimnames(data1)[[2]]) & is.null(dimnames(data2)[[2]]))){
+ same.col.name <- FALSE
+ any.id.col.name <- FALSE
+ # other col names param remain NULL
+ }else if(identical(dimnames(data1)[[2]], dimnames(data2)[[2]])){
+ same.col.name <- TRUE
+ col.name <- dimnames(data1)[[2]]
+ any.id.col.name <- TRUE
+ same.col.names.pos1 <- 1:ncol(data1)
+ same.col.names.pos2 <- 1:ncol(data1)
+ same.col.names.match1 <- 1:ncol(data1)
+ same.col.names.match2 <- 1:ncol(data1)
+ common.col.names <- dimnames(data1)[[2]]
+ }else{
+ same.col.name <- FALSE
+ any.id.col.name <- FALSE
+ if(any(dimnames(data1)[[2]] %in% dimnames(data2)[[2]])){
+ any.id.col.name <- TRUE
+ same.col.names.pos1 <- which(dimnames(data1)[[2]] %in% dimnames(data2)[[2]])
+ same.col.names.match1 <- match(dimnames(data1)[[2]], dimnames(data2)[[2]])
+ }
+ if(any(dimnames(data2)[[2]] %in% dimnames(data1)[[2]])){
+ any.id.col.name <- TRUE
+ same.col.names.pos2 <- which(dimnames(data2)[[2]] %in% dimnames(data1)[[2]])
+ same.col.names.match2 <- match(dimnames(data2)[[2]], dimnames(data1)[[2]])
+ }
+ if(any.id.col.name == TRUE){
+ common.col.names <- unique(c(dimnames(data1)[[2]][same.col.names.pos1], dimnames(data2)[[2]][same.col.names.pos2]))
+ }
+ }
+ }
+ # identical row and col content
+ if(all(class(data1) == "table")){
+ data1 <- as.data.frame(matrix(data1, ncol = ncol(data1)), stringsAsFactors = FALSE) # conversion of table into data frame to facilitate inter class comparison
+ }else if(all(class(data1) %in% c("matrix", "array"))){
+ data1 <- as.data.frame(data1, stringsAsFactors = FALSE) # conversion of matrix into data frame to facilitate inter class comparison
+ }else if(all(class(data1) == "data.frame")){
+ # data1 <- data.frame(lapply(data1, as.character), stringsAsFactors = FALSE) # conversion of columns into characters
+ }
+ if(all(class(data2) == "table")){
+ data2 <- as.data.frame(matrix(data2, ncol = ncol(data2)), stringsAsFactors = FALSE) # conversion of table into data frame to facilitate inter class comparison
+ }else if(all(class(data2) %in% c("matrix", "array"))){
+ data2 <- as.data.frame(data2, stringsAsFactors = FALSE) # conversion of matrix into data frame to facilitate inter class comparison
+ }else if(all(class(data2) == "data.frame")){
+ # data2 <- data.frame(lapply(data2, as.character), stringsAsFactors = FALSE) # conversion of columns into characters
+ }
+ row.names(data1) <- paste0("A", 1:nrow(data1))
+ row.names(data2) <- paste0("A", 1:nrow(data2))
+ if(same.col.nb == TRUE){ # because if not the same col nb, the row cannot be identical
+ if(all(sapply(data1, FUN = typeof) == "integer") & all(sapply(data2, FUN = typeof) == "integer") & as.double(nrow(data1)) * nrow(data2) <= 1e10){ # fast method for integers (thus not data frames). as.double(nrow(data1)) to prevent integer overflow because R is 32 bits for integers
+ tempo1 <- c(as.data.frame(t(data1), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
+ tempo2 <- c(as.data.frame(t(data2), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
+ same.row.pos1 <- which(tempo1 %in% tempo2)
+ same.row.pos2 <- which(tempo2 %in% tempo1)
+ same.row.match1 <- match(tempo1, tempo2)
+ same.row.match2 <- match(tempo2, tempo1)
+ }else if(as.double(nrow(data1)) * nrow(data2) <= 1e6){ # as.double(nrow(data1)) to prevent integer overflow because R is 32 bits for integers
+ # inactivated because I would like to keep the mode during comparisons
+ # if(col.nb <= 10){ # if ncol is not to big, the t() should not be that long
+ # tempo1 <- c(as.data.frame(t(data1), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
+ # tempo2 <- c(as.data.frame(t(data2), stringsAsFactors = FALSE)) # conversion into list.
+ # same.row.pos1 <- which(tempo1 %in% tempo2)
+ # same.row.pos2 <- which(tempo2 %in% tempo1)
+ # same.row.match1 <- match(tempo1, tempo2)
+ # same.row.match2 <- match(tempo2, tempo1)
+ # }else{
+ # very long computation
+ same.row.pos1 <- logical(length = nrow(data1)) # FALSE by default
+ same.row.pos1[] <- FALSE # security
+ same.row.pos2 <- logical(length = nrow(data2)) # FALSE by default
+ same.row.pos2[] <- FALSE # security
+ same.row.match1 <- rep(NA, nrow(data1))
+ same.row.match2 <- rep(NA, nrow(data2))
+ for(i3 in 1:nrow(data1)){
+ for(i4 in 1:nrow(data2)){
+ tempo1 <- data1[i3, ]
+ tempo2 <- data2[i4, ]
+ rownames(tempo1) <- NULL # to have same row and column names
+ colnames(tempo1) <- NULL # to have same row and column names
+ rownames(tempo2) <- NULL # to have same row and column names
+ colnames(tempo2) <- NULL # to have same row and column names
+ if(identical(tempo1, tempo2)){
+ same.row.pos1[i3] <- TRUE
+ same.row.pos2[i4] <- TRUE
+ same.row.match1[i3] <- i4
+ same.row.match2[i4] <- i3
+ }
+ }
+ }
+ same.row.pos1 <- which(same.row.pos1)
+ same.row.pos2 <- which(same.row.pos2)
+ # }
+ }else{
+ same.row.pos1 <- "TOO BIG FOR EVALUATION"
+ same.row.pos2 <- "TOO BIG FOR EVALUATION"
+ same.row.match1 <- "TOO BIG FOR EVALUATION"
+ same.row.match2 <- "TOO BIG FOR EVALUATION"
+ }
+
+ names(same.row.pos1) <- NULL
+ names(same.row.pos2) <- NULL
+ if(all(is.na(same.row.pos1))){
+ same.row.pos1 <- NULL
+ }else{
+ same.row.pos1 <- same.row.pos1[ ! is.na(same.row.pos1)]
+ any.id.row <- TRUE
+ }
+ if(all(is.na(same.row.pos2))){
+ same.row.pos2 <- NULL
+ }else{
+ same.row.pos2 <- same.row.pos2[ ! is.na(same.row.pos2)]
+ any.id.row <- TRUE
+ }
+ if(is.null(same.row.pos1) & is.null(same.row.pos2)){
+ any.id.row <- FALSE
+ }else if(length(same.row.pos1) == 0L & length(same.row.pos2) == 0L){
+ any.id.row <- FALSE
+ }else if(all(same.row.pos1 == "TOO BIG FOR EVALUATION") & all(same.row.pos2 == "TOO BIG FOR EVALUATION")){
+ any.id.row <- NULL
+ }
+ }else{
+ any.id.row <- FALSE
+ # same.row.pos1 and 2 remain NULL
+ }
+ if(same.row.nb == TRUE){ # because if not the same row nb, the col cannot be identical
+ if(as.double(ncol(data1)) * ncol(data2) <= 1e10){ # comparison of data frame columns is much easier than rows because no need to use t() before converting to list for fast comparison. as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
+ # if(all(sapply(data1, FUN = typeof) == "integer") & all(sapply(data2, FUN = typeof) == "integer") & as.double(ncol(data1)) * ncol(data2) <= 1e10){ # fast method for integers (thus not data frames). as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
+ tempo1 <- c(data1)
+ tempo2 <- c(data2)
+ same.col.pos1 <- which(tempo1 %in% tempo2)
+ same.col.pos2 <- which(tempo2 %in% tempo1)
+ same.col.match1 <- match(tempo1, tempo2)
+ same.col.match2 <- match(tempo2, tempo1)
+ # }else if(as.double(ncol(data1)) * ncol(data2) <= 1e6){ # as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
+ # same.col.pos1 <- logical(length = ncol(data1)) # FALSE by default
+ # same.col.pos1[] <- FALSE # security
+ # same.col.pos2 <- logical(length = ncol(data2)) # FALSE by default
+ # same.col.pos2[] <- FALSE # security
+ # same.col.match1 <- rep(NA, ncol(data1))
+ # same.col.match2 <- rep(NA, ncol(data2))
+ # for(i3 in 1:ncol(data1)){
+ # for(i4 in 1:ncol(data2)){
+ # if(identical(data1[ , i3], data2[ , i4])){
+ # same.col.pos1[i3] <- TRUE
+ # same.col.pos2[i4] <- TRUE
+ # same.col.match1[i3] <- i4
+ # same.col.match2[i4] <- i3
+ # }
+ # }
+ # }
+ # same.col.pos1 <- which(same.col.pos1)
+ # same.col.pos2 <- which(same.col.pos2)
+ }else{
+ same.col.pos1 <- "TOO BIG FOR EVALUATION"
+ same.col.pos2 <- "TOO BIG FOR EVALUATION"
+ }
+ names(same.col.pos1) <- NULL
+ names(same.col.pos2) <- NULL
+ if(all(is.na(same.col.pos1))){
+ same.col.pos1 <- NULL
+ }else{
+ same.col.pos1 <- same.col.pos1[ ! is.na(same.col.pos1)]
+ any.id.col <- TRUE
+ }
+ if(all(is.na(same.col.pos2))){
+ same.col.pos2 <- NULL
+ }else{
+ same.col.pos2 <- same.col.pos2[ ! is.na(same.col.pos2)]
+ any.id.col <- TRUE
+ }
+ if(is.null(same.col.pos1) & is.null(same.col.pos2)){
+ any.id.col <- FALSE
+ }else if(length(same.col.pos1) == 0L & length(same.col.pos2) == 0L){
+ any.id.col <- FALSE
+ }else if(all(same.col.pos1 == "TOO BIG FOR EVALUATION") & all(same.col.pos2 == "TOO BIG FOR EVALUATION")){
+ any.id.col <- NULL
+ }
+ }else{
+ any.id.col <- FALSE
+ # same.col.pos1 and 2 remain NULL
+ }
+ if(same.dim == TRUE){
+ # names(data1) <- NULL
+ # row.names(data1) <- NULL
+ # names(data2) <- NULL
+ # row.names(data2) <- NULL
+ # if(identical(data1, data2)){
+ # identical.content <- TRUE
+ # }else{
+ # identical.content <- FALSE
+ # }
+ # code above inactivated because integer dataset are sometimes imported into R as integer or as double, which is different for identical(data1, data2)
+ if(all(data1 == data2)){
+ identical.content <- TRUE
+ }else{
+ identical.content <- FALSE
+ }
+ }else{
+ identical.content <- FALSE
+ }
+ }
+ output <- list(same.class = same.class, class = class, same.dim = same.dim, dim = dim, same.row.nb = same.row.nb, row.nb = row.nb, same.col.nb = same.col.nb , col.nb = col.nb, same.row.name = same.row.name, row.name = row.name, any.id.row.name = any.id.row.name, same.row.names.pos1 = same.row.names.pos1, same.row.names.pos2 = same.row.names.pos2, same.row.names.match1 = same.row.names.match1, same.row.names.match2 = same.row.names.match2, common.row.names = common.row.names, same.col.name = same.col.name, col.name = col.name,any.id.col.name = any.id.col.name, same.col.names.pos1 = same.col.names.pos1, same.col.names.pos2 = same.col.names.pos2, same.col.names.match1 = same.col.names.match1, same.col.names.match2 = same.col.names.match2, common.col.names = common.col.names, any.id.row = any.id.row, same.row.pos1 = same.row.pos1, same.row.pos2 = same.row.pos2, same.row.match1 = same.row.match1, same.row.match2 = same.row.match2, any.id.col = any.id.col, same.col.pos1 = same.col.pos1, same.col.pos2 = same.col.pos2, same.col.match1 = same.col.match1, same.col.match2 = same.col.match2, identical.object = identical.object, identical.content = identical.content)
+ return(output)
}
-if( ! any(class(data2) %in% c("logical", "integer", "numeric", "character", "factor", "table"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A NON NULL VECTOR, FACTOR OR 1D TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(all(class(data2) %in% "table")){
-if(length(dim(data2)) > 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A 1D TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+
+
+######## fun_comp_list() #### comparison of two lists
+
+
+fun_comp_list <- function(data1, data2){
+ # AIM
+ # compare two lists. Check and report in a list if the 2 datasets have:
+ # same length
+ # common names
+ # common compartments
+ # ARGUMENTS
+ # data1: list
+ # data2: list
+ # RETURN
+ # a list containing:
+ # $same.length: logical. Are number of elements identical?
+ # $length: number of elements in the 2 datasets (NULL otherwise)
+ # $same.names: logical. Are element names identical ?
+ # $name: name of elements of the 2 datasets if identical (NULL otherwise)
+ # $any.id.name: logical. Is there any element names identical ?
+ # $same.names.pos1: positions, in data1, of the element names identical in data2
+ # $same.names.pos2: positions, in data2, of the compartment names identical in data1
+ # $any.id.compartment: logical. is there any identical compartments ?
+ # $same.compartment.pos1: positions, in data1, of the compartments identical in data2
+ # $same.compartment.pos2: positions, in data2, of the compartments identical in data1
+ # $identical.object: logical. Are objects identical (kind of object, compartment names and content)?
+ # $identical.content: logical. Are content objects identical (identical compartments excluding compartment names)?
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # none
+ # EXAMPLES
+ # obs1 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; obs2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; fun_comp_list(obs1, obs2)
+ # obs1 = list(1:5, LETTERS[1:2]) ; obs2 = list(a = 1:5, b = LETTERS[1:2]) ; fun_comp_list(obs1, obs2)
+ # obs1 = list(b = 1:5, c = LETTERS[1:2]) ; obs2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; fun_comp_list(obs1, obs2)
+ # obs1 = list(b = 1:5, c = LETTERS[1:2]) ; obs2 = list(LETTERS[5:9], matrix(1:6), 1:5) ; fun_comp_list(obs1, obs2)
+ # DEBUGGING
+ # data1 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; data2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) # for function debugging
+ # data1 = list(a = 1:5, b = LETTERS[1:2]) ; data2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # argument checking
+ if( ! any(class(data1) %in% "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! any(class(data2) %in% "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
+ # end argument checking
+ # main code
+ same.length <- NULL
+ length <- NULL
+ same.names <- NULL
+ name <- NULL
+ any.id.name <- NULL
+ same.names.pos1 <- NULL
+ same.names.pos2 <- NULL
+ any.id.compartment <- NULL
+ same.compartment.pos1 <- NULL
+ same.compartment.pos2 <- NULL
+ identical.object <- NULL
+ identical.content <- NULL
+ if(identical(data1, data2)){
+ same.length <- TRUE
+ length <- length(data1)
+ if( ! is.null(names(data1))){
+ same.names <- TRUE
+ name <- names(data1)
+ any.id.name <- TRUE
+ same.names.pos1 <- 1:length(data1)
+ same.names.pos2 <- 1:length(data2)
+ }
+ any.id.compartment <- TRUE
+ same.compartment.pos1 <- 1:length(data1)
+ same.compartment.pos2 <- 1:length(data2)
+ identical.object <- TRUE
+ identical.content <- TRUE
+ }else{
+ identical.object <- FALSE
+ if( ! identical(length(data1), length(data2))){
+ same.length<- FALSE
+ }else{
+ same.length<- TRUE
+ length <- length(data1)
+ }
+ if( ! (is.null(names(data1)) & is.null(names(data2)))){
+ if( ! identical(names(data1), names(data2))){
+ same.names <- FALSE
+ }else{
+ same.names <- TRUE
+ name <- names(data1)
+ }
+ any.id.name <- FALSE
+ if(any(names(data1) %in% names(data2))){
+ any.id.name <- TRUE
+ same.names.pos1 <- which(names(data1) %in% names(data2))
+ }
+ if(any(names(data2) %in% names(data1))){
+ any.id.name <- TRUE
+ same.names.pos2 <- which(names(data2) %in% names(data1))
+ }
+ }
+ names(data1) <- NULL
+ names(data2) <- NULL
+ any.id.compartment <- FALSE
+ if(any(data1 %in% data2)){
+ any.id.compartment <- TRUE
+ same.compartment.pos1 <- which(data1 %in% data2)
+ }
+ if(any(data2 %in% data1)){
+ any.id.compartment <- TRUE
+ same.compartment.pos2 <- which(data2 %in% data1)
+ }
+ if(same.length == TRUE & ! all(is.null(same.compartment.pos1), is.null(same.compartment.pos2))){
+ if(identical(same.compartment.pos1, same.compartment.pos2)){
+ identical.content <- TRUE
+ }else{
+ identical.content <- FALSE
+ }
+ }else{
+ identical.content <- FALSE
+ }
+ }
+ output <- list(same.length = same.length, length = length, same.names = same.names, name = name, any.id.name = any.id.name, same.names.pos1 = same.names.pos1, same.names.pos2 = same.names.pos2, any.id.compartment = any.id.compartment, same.compartment.pos1 = same.compartment.pos1, same.compartment.pos2 = same.compartment.pos2, identical.object = identical.object, identical.content = identical.content)
+ return(output)
}
+
+
+######## fun_test() #### test combinations of argument values of a function and return errors (and graphs)
+
+
+# add traceback https://stackoverflow.com/questions/47414119/how-to-read-a-traceback-in-r
+
+fun_test <- function(
+ fun,
+ arg,
+ val,
+ expect.error = NULL,
+ parall = FALSE,
+ thread.nb = NULL,
+ print.count = 10,
+ plot.fun = FALSE,
+ export = FALSE,
+ res.path = NULL,
+ lib.path = NULL,
+ cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
+){
+ # AIM
+ # test combinations of argument values of a function
+ # WARNINGS
+ # Limited to 43 arguments with at least 2 values each. The total number of arguments tested can be more if the additional arguments have a single value. The limit is due to nested "for" loops (https://stat.ethz.ch/pipermail/r-help/2008-March/157341.html), but it should not be a problem since the number of tests would be 2^43 > 8e12
+ # ARGUMENTS
+ # fun: character string indicating the name of the function tested (without brackets)
+ # arg: vector of character strings of arguments of fun. At least arguments that do not have default values must be present in this vector
+ # val: list with number of compartments equal to length of arg, each compartment containing values of the corresponding argument in arg. Each different value must be in a list or in a vector. For instance, argument 3 in arg is a logical argument (values accepted TRUE, FALSE, NA). Thus, compartment 3 of val can be either list(TRUE, FALSE, NA), or c(TRUE, FALSE, NA). NULL value alone must be written list(NULL)
+ # expect.error: list of exactly the same structure as val argument, but containing FALSE or TRUE, depending on whether error is expected (TRUE) or not (FALSE) for each corresponding value of val. A message is returned depending on discrepancies between the expected and observed errors. BEWARE: not always possible to write the expected errors for all the combination of argument values. Ignored if NULL
+ # parall: logical. Force parallelization ?
+ # thread.nb: numeric value indicating the number of threads to use if ever parallelization is required. If NULL, all the available threads will be used. Ignored if parall is FALSE
+ # print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
+ # plot.fun: logical. Plot the plotting function tested for each test?
+ # export: logical. Export the results into a .RData file and into a .txt file? If FALSE, return a list into the console (see below). BEWARE: will be automatically set to TRUE if parall is TRUE. This means that when using parallelization, the results are systematically exported, not returned into the console
+ # res.path: character string indicating the absolute pathway of folder where the txt results and pdfs, containing all the plots, will be saved. Several txt and pdf, one per thread, if parallelization. Ignored if export is FALSE. Must be specified if parall is TRUE or if export is TRUE
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # cute.path: character string indicating the absolute path of the cute.R file. Will be remove when cute will be a package. Ignored if parall is FALSE
+ # REQUIRED PACKAGES
+ # lubridate
+ # parallel if parall arguemtn is TRUE (included in the R installation packages but not automatically loaded)
+ # pdftools if parall arguemtn is TRUE (included in the R installation packages but not automatically loaded)
+ # If the tested function is in a package, this package must be imported first (no parallelization) or must be in the classical R package folder indicated by the lib.path argument (parallelization)
+ # RETURN
+ # if export is FALSE a list containing:
+ # $fun: the tested function
+ # $instruction: the initial instruction
+ # $sys.info: system and packages info
+ # $data: a data frame of all the combination tested, containing the following columns:
+ # the different values tested, named by arguments
+ # $kind: a vector of character strings indicating the kind of test result: either "ERROR", or "WARNING", or "OK"
+ # $problem: a logical vector indicating if error or not
+ # $expected.error: optional logical vector indicating the expected error specified in the expect.error argument
+ # $message: either NULL if $kind is always "OK", or the messages
+ # if export is TRUE 1) the same list object into a .RData file, 2) also the $data data frame into a .txt file, and 3) if expect.error is non NULL and if any discrepancy, the $data data frame into a .txt file but containing only the rows with discrepancies between expected and observed errors
+ # one or several pdf if a plotting function is tested and if the plot.fun argument is TRUE
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_get_message()
+ # fun_pack()
+ # EXAMPLES
+ # fun_test(fun = "unique", arg = c("x", "incomparables"), val = list(x = list(1:10, c(1,1,2,8), NA), incomparable = c(TRUE, FALSE, NA)))
+ # fun_test(fun = "fun_round", arg = c("data", "dec.nb", "after.lead.zero"), val = list(L1 = list(c(1, 1.0002256, 1.23568), "a", NA), L2 = list(2, c(1,3), NA), L3 = c(TRUE, FALSE, NA)))
+ # fun_test(fun = "plot", arg = c("x", "y"), val = list(x = list(1:10, 12:13, NA, (1:10)^2), y = list(1:10, NA, NA)), expect.error = list(x = list(FALSE, TRUE, TRUE, FALSE), y = list(FALSE, TRUE, TRUE)), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = NULL)
+ # fun_test(fun = "plot", arg = c("x", "y"), val = list(x = list(1:10, 12:13, NA, (1:10)^2), y = list(1:10, NA, NA)), parall = FALSE, thread.nb = 4, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\")
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_test(fun = "fun_gg_boxplot", arg = c("data1", "y", "categ"), val = list(L1 = list(L1 = obs1), L2 = list(L1 = "Time"), L3 = list(L1 = "Group1")))
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_test(fun = "fun_gg_boxplot", arg = c("data1", "y", "categ"), val = list(L1 = list(obs1), L2 = "Time", L3 = "Group1"), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\")
+ # library(ggplot2) ; fun_test(fun = "geom_histogram", arg = c("data", "mapping"), val = list(x = list(data.frame(X = "a", stringsAsFactors = TRUE)), y = list(ggplot2::aes(x = X))), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\") # BEWARE: ggplot2::geom_histogram does not work
+ # DEBUGGING
+ # fun = "unique" ; arg = "x" ; val = list(x = list(1:10, c(1,1,2,8), NA)) ; expect.error = list(x = list(FALSE, FALSE, TRUE)) ; parall = FALSE ; thread.nb = NULL ; plot.fun = FALSE ; export = FALSE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 1 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
+ # fun = "unique" ; arg = c("x", "incomparables") ; val = list(x = list(1:10, c(1,1,2,8), NA), incomparable = c(TRUE, FALSE, NA)) ; expect.error = NULL ; parall = FALSE ; thread.nb = 2 ; plot.fun = FALSE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 10 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
+ # fun = "plot" ; arg = c("x", "y") ; val = list(x = list(1:10, 12:13, NA), y = list(1:10, NA, NA)) ; expect.error = list(x = list(FALSE, FALSE, TRUE, FALSE), y = list(FALSE, TRUE, TRUE)) ; print.count = 10 ; parall = FALSE ; thread.nb = NULL ; plot.fun = TRUE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL # for function debugging
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun = "fun_gg_boxplot" ; arg = c("data1", "y", "categ") ; val = list(L1 = list(L1 = obs1), L2 = list(L1 = "Time"), L3 = list(L1 = "Group1")) ; expect.error = NULL ; print.count = 10 ; parall = FALSE ; thread.nb = NULL ; plot.fun = TRUE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL # for function debugging
+ # fun = "unique" ; arg = "x" ; val = list(list(1:3, mean)) ; expect.error = list(TRUE, TRUE) ; parall = FALSE ; thread.nb = NULL ; plot.fun = FALSE ; export = FALSE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 1 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check",
+ "fun_get_message",
+ "fun_pack"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # arg with no default values
+ mandat.args <- c(
+ "fun",
+ "arg",
+ "val"
+ )
+ tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), "))")))
+ print(tempo)
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = fun, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = arg, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = val, class = "list", fun.name = function.name) ; eval(ee)
+ if( ! is.null(expect.error)){
+ tempo <- fun_check(data = expect.error, class = "list", fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = parall, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(parall == TRUE){
+ if( ! is.null(thread.nb)){
+ tempo <- fun_check(data = thread.nb, typeof = "integer", double.as.integer.allowed = TRUE, neg.values = FALSE, length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & thread.nb < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": thread.nb PARAMETER MUST EQUAL OR GREATER THAN 1: ", thread.nb)
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = plot.fun, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = export, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(res.path)){
+ tempo <- fun_check(data = res.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = cute.path, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # new environment
+ env.name <- paste0("env", as.numeric(Sys.time()))
+ if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ assign(env.name, new.env())
+ assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
+ }
+ # end new environment
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "fun",
+ "arg",
+ "val",
+ # "expect.erro", # because can be NULL
+ "parall",
+ # "thread.nb", # because can be NULL
+ "print.count",
+ "plot.fun",
+ "export",
+ # "res.path", # because can be NULL
+ # "lib.path", # because can be NULL
+ "cute.path"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ # end warning initiation
+ # other checkings
+ if(grepl(x = fun, pattern = "()$")){ # remove ()
+ fun <- sub(x = fun, pattern = "()$", replacement = "")
+ }
+ if( ! exists(fun)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CHARACTER STRING IN fun ARGUMENT DOES NOT EXIST IN THE R WORKING ENVIRONMENT: ", paste(fun, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }else if( ! all(base::class(get(fun)) == "function")){ # here no env = sys.nframe(), inherit = FALSE for get() because fun is a function in the classical scope
+ tempo.cat <- paste0("ERROR IN ", function.name, ": fun ARGUMENT IS NOT CLASS \"function\" BUT: ", paste(base::class(get(fun)), collapse = "\n"), "\nCHECK IF ANY CREATED OBJECT WOULD HAVE THE NAME OF THE TESTED FUNCTION")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(tempo$problem == FALSE & base::length(arg) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": arg ARGUMENT CANNOT BE LENGTH 0")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ for(i2 in 1:base::length(val)){
+ tempo1 <- fun_check(data = val[[i2]], class = "vector", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = val[[i2]], class = "list", na.contain = TRUE, fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": COMPARTMENT ", i2, " OF val ARGUMENT MUST BE A VECTOR OR A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }else if(tempo1$problem == FALSE){ # vector split into list compartments
+ val[[i2]] <- split(x = val[[i2]], f = 1:base::length(val[[i2]]))
+ }
+ }
+ if(base::length(arg) != base::length(val)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF arg ARGUMENT MUST BE IDENTICAL TO LENGTH OF val ARGUMENT:\nHERE IT IS: ", base::length(arg), " VERSUS ", base::length(val))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ args <- names(formals(get(fun))) # here no env = sys.nframe(), inherit = FALSE for get() because fun is a function in the classical scope
+ if( ! all(arg %in% args)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": SOME OF THE STRINGS IN arg ARE NOT ARGUMENTS OF fun\nfun ARGUMENTS: ", paste(args, collapse = " "),"\nPROBLEMATIC STRINGS IN arg: ", paste(arg[ ! arg %in% args], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(sum(sapply(val, FUN = length) > 1) > 43){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CANNOT TEST MORE THAN 43 ARGUMENTS IF THEY ALL HAVE AT LEAST 2 VALUES EACH\nHERE THE NUMBER IS: ", sum(sapply(val, FUN = length) > 1))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if( ! is.null(expect.error)){
+ if(base::length(val) != base::length(expect.error)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF val ARGUMENT MUST BE IDENTICAL TO LENGTH OF expect.error ARGUMENT:\nHERE IT IS: ", base::length(val), " VERSUS ", base::length(expect.error))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ for(i3 in 1:base::length(expect.error)){
+ tempo1 <- fun_check(data = expect.error[[i3]], class = "vector", mode = "logical", fun.name = function.name)
+ tempo2 <- fun_check(data = expect.error[[i3]], class = "list", fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": COMPARTMENT ", i3, " OF expect.error ARGUMENT MUST BE TRUE OR FALSE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }else if(tempo1$problem == FALSE){ # vector split into list compartments
+ expect.error[[i3]] <- split(x = expect.error[[i3]], f = 1:base::length(expect.error[[i3]]))
+ }
+ }
+ }
+ if( ! is.null(res.path)){
+ if( ! all(dir.exists(res.path))){ # separation to avoid the problem of tempo$problem == FALSE and res.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE res.path ARGUMENT DOES NOT EXISTS:\n", paste(res.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ }
+ if(parall == TRUE & is.null(res.path)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": res.path ARGUMENT MUST BE SPECIFIED IF parall ARGUMENT IS TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(is.null(res.path) & export == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": res.path ARGUMENT MUST BE SPECIFIED IF export ARGUMENT TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(parall == TRUE & export == FALSE){
+ export <- TRUE
+ tempo.cat <- paste0("WARNING FROM ", function.name, ": export ARGUMENT CONVERTED TO TRUE BECAUSE thread.nb ARGUMENT IS NOT NULL")
+ warning(paste0("\n", tempo.cat, "\n"), call. = FALSE)
+ }
+ if( ! is.null(lib.path)){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ }
+ if(parall == TRUE){
+ if(grepl(x = cute.path, pattern = "^http")){
+ tempo.error1 <- any(grepl(x = fun_get_message(data = "source(cute.path)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE)), pattern = "^[Ee]rror"))
+ tempo.error2 <- FALSE
+ }else{
+ tempo.error1 <- FALSE
+ tempo.error2 <- ! file.exists(cute.path)
+ }
+ if(tempo.error1 | tempo.error2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(grepl(x = cute.path, pattern = "^http"), "URL", "FILE"), " PATH INDICATED IN THE cute.path PARAMETER DOES NOT EXISTS:\n", cute.path)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ }
+ # end other checkings
+ # reserved word checking
+ # end reserved word checking
+ # end second round of checking and data preparation
+ # package checking
+ fun_pack(req.package = c("lubridate"), lib.path = lib.path)
+ if(parall == TRUE){
+ fun_pack(req.package = c("parallel", "pdftools"), lib.path = lib.path)
+ }
+ # end package checking
+ # declaration of special plot functions
+ sp.plot.fun <- c("fun_gg_scatter", "fun_gg_bar", "fun_gg_boxplot")
+ # end declaration of special plot functions
+ # main code
+ ini.warning.length <- base::options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ cat("\nfun_test JOB IGNITION\n")
+ ini.date <- Sys.time()
+ ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
+ if(export == TRUE){
+ res.path <- paste0(res.path, "/fun_test_res_", trunc(ini.time))
+ if(dir.exists(res.path)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": FOLDER ALREADY EXISTS\n", res.path, "\nPLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ dir.create(res.path)
+ }
+ }
+ total.comp.nb <- prod(sapply(val, FUN = "length"))
+ cat(paste0("\nTHE TOTAL NUMBER OF TESTS IS: ", total.comp.nb, "\n"))
+ # creation of the txt instruction that includes several loops
+ loop.string <- NULL
+ end.loop.string <- NULL
+ fun.args <- NULL
+ fun.args2 <- NULL
+ error.values <- NULL
+ arg.values <- "list("
+ for(i1 in 1:base::length(arg)){
+ if(parall == FALSE){
+ if(base::length(val[[i1]]) > 1){ # loop only if more than one value in base::length(val[[i1]])
+ loop.string <- paste0(loop.string, "for(i", i1, " in 1:", base::length(val[[i1]]), "){")
+ end.loop.string <- paste0(end.loop.string, "}")
+ }
+ }else{
+ loop.string <- "for(i in x){"
+ end.loop.string <- "}"
+ }
+ fun.args <- paste0(
+ fun.args,
+ ifelse(i1 == 1L, "", ", "),
+ arg[i1],
+ " = val[[",
+ i1,
+ "]][[",
+ if(parall == FALSE){
+ if(base::length(val[[i1]]) > 1){
+ paste0("i", i1)
+ }else{
+ "1" # a unique element in val[[i1]]
+ }
+ }else{
+ paste0("i.list[[", i1, "]][i]")
+ },
+ "]]"
+ )
+ fun.args2 <- paste0(
+ fun.args2,
+ ifelse(i1 == 1L, "", ", "),
+ arg[i1],
+ " = val[[",
+ i1,
+ "]][[', ",
+ if(parall == FALSE){
+ if(base::length(val[[i1]]) > 1){
+ paste0("i", i1)
+ }else{
+ "1" # a unique element in val[[i1]]
+ }
+ }else{
+ paste0("i.list[[", i1, "]][i]")
+ },
+ ", ']]"
+ )
+ arg.values <- paste0(
+ arg.values,
+ "val[[", i1, "]][[",
+ if(parall == FALSE){
+ if(base::length(val[[i1]]) > 1){
+ paste0("i", i1)
+ }else{
+ "1" # a unique element in val[[i1]]
+ }
+ }else{
+ paste0("i.list[[", i1, "]][i]")
+ },
+ "]]",
+ ifelse(i1 == base::length(arg), "", ", ")
+ )
+ error.values <- paste0(
+ error.values,
+ ifelse(i1 == 1L, "", " | "),
+ "expect.error[[", i1, "]][[",
+ if(parall == FALSE){
+ if(base::length(expect.error[[i1]]) > 1){
+ paste0("i", i1)
+ }else{
+ "1" # a unique element in expect.error[[i1]]
+ }
+ }else{
+ paste0("i.list[[", i1, "]][i]")
+ },
+ "]]"
+ )
+ }
+ arg.values <- paste0(arg.values, ")")
+ fun.test <- paste0(fun, "(", fun.args, ")")
+ fun.test2 <- paste0("paste0('", fun, "(", fun.args2, ")')")
+ # plot title for special plot functions
+ if(plot.fun == TRUE){
+ plot.kind <- "classic"
+ if(fun %in% sp.plot.fun){
+ plot.kind <- "special"
+ if(any(arg %in% "title")){ # this is for the special functions
+ tempo.match <- regmatches(x = fun.test, m = regexpr(text = fun.test, pattern = "title = .+[,)]"))
+ tempo.match <- substring(tempo.match , 1, nchar(tempo.match) - 1)
+ fun.test <- sub(x = fun.test, pattern = tempo.match, replacement = paste0(tempo.match, "\ntempo.title"))
+ }else{
+ fun.test <- sub(x = fun.test, pattern = ")$", replacement = ", title = tempo.title)")
+ }
+ }
+ }
+ # end plot title for special plot functions
+ kind <- character()
+ problem <- logical()
+ expected.error <- logical()
+ res <- character()
+ count <- 0
+ print.count.loop <- 0
+ plot.count <- 0
+ if(base::length(arg) == 1L){
+ data <- data.frame()
+ }else{ # base::length(arg) == 0L already tested above
+ data <- data.frame(t(vector("character", base::length(arg))), stringsAsFactors = FALSE)[-1, ] # -1 to remove the single row created and to have an empty data frame with base::length(arg) columns
+ }
+ code <- paste(
+ loop.string, '
+count <- count + 1
+print.count.loop <- print.count.loop + 1
+arg.values.print <- eval(parse(text = arg.values)) # recover the list of the i1 compartment
+for(j3 in 1:base::length(arg.values.print)){ # WARNING: do not use i1, i2 etc., here because already in loop.string
+tempo.capt <- capture.output(tempo.error <- fun_get_message(data = paste0("paste(arg.values.print[[", j3, "]])"), kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))) # collapsing arg.values sometimes does not work (with function for instance)
+if( ! is.null(tempo.error)){
+arg.values.print[[j3]] <- paste0("SPECIAL VALUE OF CLASS ", base::class(arg.values.print[[j3]]), " AND TYPE ", base::typeof(arg.values.print[[j3]]))
}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
-# end argument checking
-# main code
-same.class <- FALSE
-class <- NULL
-same.length <- FALSE
-length <- NULL
-same.levels <- NULL # not FALSE to deal with no factors
-levels <- NULL
-any.id.levels <- FALSE
-same.levels.pos1 <- NULL
-same.levels.pos2 <- NULL
-same.levels.match1 <- NULL
-same.levels.match2 <- NULL
-common.levels <- NULL
-same.names <- NULL # not FALSE to deal with absence of name
-name <- NULL
-any.id.name <- FALSE
-same.names.pos1 <- NULL
-same.names.pos2 <- NULL
-same.names.match1 <- NULL
-same.names.match2 <- NULL
-common.names <- NULL
-any.id.element <- FALSE
-same.elements.pos1 <- NULL
-same.elements.pos2 <- NULL
-same.elements.match1 <- NULL
-same.elements.match2 <- NULL
-common.elements <- NULL
-same.order <- NULL
-order1 <- NULL
-order2 <- NULL
-identical.object <- FALSE
-identical.content <- FALSE
-if(identical(data1, data2)){
-same.class <- TRUE
-class <- class(data1)
-same.length <- TRUE
-length <- length(data1)
-if(any(class(data1) %in% "factor")){
-same.levels <- TRUE
-levels <- levels(data1)
-any.id.levels <- TRUE
-same.levels.pos1 <- 1:length(levels(data1))
-same.levels.pos2 <- 1:length(levels(data2))
-same.levels.match1 <- 1:length(levels(data1))
-same.levels.match2 <- 1:length(levels(data2))
-common.levels <- levels(data1)
}
-if( ! is.null(names(data1))){
-same.names <- TRUE
-name <- names(data1)
-any.id.name <- TRUE
-same.names.pos1 <- 1:length(data1)
-same.names.pos2 <- 1:length(data2)
-same.names.match1 <- 1:length(data1)
-same.names.match2 <- 1:length(data2)
-common.names <- names(data1)
+data <- rbind(data, as.character(sapply(arg.values.print, FUN = "paste", collapse = " ")), stringsAsFactors = FALSE) # each colum is a test
+tempo.capt <- capture.output(tempo.try.error <- fun_get_message(data = eval(parse(text = fun.test2)), kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))) # data argument needs a character string but eval(parse(text = fun.test2)) provides it (eval parse replace the i1, i2, etc., by the correct values, meaning that only val is required in the env.name environment)
+tempo.capt <- capture.output(tempo.try.warning <- fun_get_message(data = eval(parse(text = fun.test2)), kind = "warning", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE), print.no = TRUE)) # data argument needs a character string but eval(parse(text = fun.test2)) provides it (eval parse replace the i1, i2, etc., by the correct values, meaning that only val is required in the env.name environment)
+if( ! is.null(expect.error)){
+expected.error <- c(expected.error, eval(parse(text = error.values)))
}
-any.id.element <- TRUE
-same.elements.pos1 <- 1:length(data1)
-same.elements.pos2 <- 1:length(data2)
-same.elements.match1 <- 1:length(data1)
-same.elements.match2 <- 1:length(data2)
-common.elements <- data1
-same.order <- TRUE
-order1 <- order(data1)
-order2 <- order(data2)
-identical.object <- TRUE
-identical.content <- TRUE
+if( ! is.null(tempo.try.error)){
+kind <- c(kind, "ERROR")
+problem <- c(problem, TRUE)
+res <- c(res, tempo.try.error)
}else{
-if(identical(class(data1), class(data2))){
-same.class <- TRUE
-class <- class(data1)
-}
-if(identical(length(data1), length(data2))){
-same.length<- TRUE
-length <- length(data1)
-}
-if(any(class(data1) %in% "factor") & any(class(data2) %in% "factor")){
-if(identical(levels(data1), levels(data2))){
-same.levels <- TRUE
-levels <- levels(data1)
+if( ! is.null(tempo.try.warning)){
+kind <- c(kind, "WARNING")
+problem <- c(problem, FALSE)
+res <- c(res, tempo.try.warning)
}else{
-same.levels <- FALSE
-}
-if(any(levels(data1) %in% levels(data2))){
-any.id.levels <- TRUE
-same.levels.pos1 <- which(levels(data1) %in% levels(data2))
-same.levels.match1 <- match(levels(data1), levels(data2))
-}
-if(any(levels(data2) %in% levels(data1))){
-any.id.levels <- TRUE
-same.levels.pos2 <- which(levels(data2) %in% levels(data1))
-same.levels.match2 <- match(levels(data2), levels(data1))
-}
-if(any.id.levels == TRUE){
-common.levels <- unique(c(levels(data1)[same.levels.pos1], levels(data2)[same.levels.pos2]))
-}
-}
-if(any(class(data1) %in% "factor")){ # to compare content
-data1 <- as.character(data1)
-}
-if(any(class(data2) %in% "factor")){ # to compare content
-data2 <- as.character(data2)
+kind <- c(kind, "OK")
+problem <- c(problem, FALSE)
+res <- c(res, "")
}
-if( ! (is.null(names(data1)) & is.null(names(data2)))){
-if(identical(names(data1), names(data2))){
-same.names <- TRUE
-name <- names(data1)
+if(plot.fun == TRUE){
+invisible(dev.set(window.nb))
+plot.count <- plot.count + 1
+tempo.title <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), ifelse(parall == FALSE, count, x[count])))
+if(plot.kind == "classic"){
+eval(parse(text = fun.test))
+tempo <- fun_post_plot(corner.text = tempo.title)
+}else if(plot.kind == "special"){
+eval(parse(text = fun.test))
}else{
-same.names <- FALSE
-}
-if(any(names(data1) %in% names(data2))){
-any.id.name <- TRUE
-same.names.pos1 <- which(names(data1) %in% names(data2))
-same.names.match1 <- match(names(data1), names(data2))
-}
-if(any(names(data2) %in% names(data1))){
-any.id.name <- TRUE
-same.names.pos2 <- which(names(data2) %in% names(data1))
-same.names.match2 <- match(names(data2), names(data1))
-}
-if(any.id.name == TRUE){
-common.names <- unique(c(names(data1)[same.names.pos1], names(data2)[same.names.pos2]))
+tempo.cat <- paste0("INTERNAL CODE ERROR 1 IN ", function.name, ": CODE HAS TO BE MODIFIED")
+stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
}
}
-names(data1) <- NULL # names solved -> to do not be disturbed by names
-names(data2) <- NULL # names solved -> to do not be disturbed by names
-if(any(data1 %in% data2)){
-any.id.element <- TRUE
-same.elements.pos1 <- which(data1 %in% data2)
-same.elements.match1 <- match(data1, data2)
}
-if(any(data2 %in% data1)){
-any.id.element <- TRUE
-same.elements.pos2 <- which(data2 %in% data1)
-same.elements.match2 <- match(data2, data1)
-}
-if(any.id.element == TRUE){
-common.elements <- unique(c(data1[same.elements.pos1], data2[same.elements.pos2]))
-}
-if(identical(data1, data2)){
-identical.content <- TRUE
-same.order <- TRUE
-}else if(identical(sort(data1), sort(data2))){
-same.order <- FALSE
-order1 <- order(data1)
-order2 <- order(data2)
+if(print.count.loop == print.count){
+print.count.loop <- 0
+tempo.time <- as.numeric(Sys.time())
+tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+final.loop <- (tempo.time - ini.time) / count * ifelse(parall == FALSE, total.comp.nb, base::length(x)) # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
+final.exp <- as.POSIXct(final.loop, origin = ini.date)
+cat(paste0(ifelse(parall == FALSE, "\n", paste0("\nIN PROCESS ", process.id, " | ")), "LOOP ", format(count, big.mark=","), " / ", format(ifelse(parall == FALSE, total.comp.nb, base::length(x)), big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
}
+if(count == ifelse(parall == FALSE, total.comp.nb, base::length(x))){
+tempo.time <- as.numeric(Sys.time())
+tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+cat(paste0(ifelse(parall == FALSE, "\nLOOP PROCESS ENDED | ", paste0("\nPROCESS ", process.id, " ENDED | ")), "LOOP ", format(count, big.mark=","), " / ", format(ifelse(parall == FALSE, total.comp.nb, base::length(x)), big.mark=","), " | TIME SPENT: ", tempo.lapse, "\n\n"))
}
-output <- list(same.class = same.class, class = class, same.length = same.length, length = length, same.levels = same.levels, levels = levels, any.id.levels = any.id.levels, same.levels.pos1 = same.levels.pos1, same.levels.pos2 = same.levels.pos2, same.levels.match1 = same.levels.match1, same.levels.match2 = same.levels.match2, common.levels = common.levels, same.names = same.names, name = name, any.id.name = any.id.name, same.names.pos1 = same.names.pos1, same.names.pos2 = same.names.pos2, same.names.match1 = same.names.match1, same.names.match2 = same.names.match2, common.names = common.names, any.id.element = any.id.element, same.elements.pos1 = same.elements.pos1, same.elements.pos2 = same.elements.pos2, same.elements.match1 = same.elements.match1, same.elements.match2 = same.elements.match2, common.elements = common.elements, same.order = same.order, order1 = order1, order2 = order2, identical.object = identical.object, identical.content = identical.content)
-return(output)
+',
+end.loop.string
+ )
+ # end creation of the txt instruction that includes several loops
+ if(parall == TRUE){
+ # list of i numbers that will be split
+ i.list <- vector("list", base::length(val)) # positions to split in parallel jobs
+ for(i2 in 1:base::length(arg)){
+ if(i2 == 1L){
+ tempo.divisor <- total.comp.nb / base::length(val[[i2]])
+ i.list[[i2]] <- rep(1:base::length(val[[i2]]), each = as.integer(tempo.divisor))
+ tempo.multi <- base::length(val[[i2]])
+ }else{
+ tempo.divisor <- tempo.divisor / base::length(val[[i2]])
+ i.list[[i2]] <- rep(rep(1:base::length(val[[i2]]), each = as.integer(tempo.divisor)), time = as.integer(tempo.multi))
+ tempo.multi <- tempo.multi * base::length(val[[i2]])
+ }
+ }
+ # end list of i numbers that will be split
+ tempo.cat <- paste0("PARALLELIZATION INITIATED AT: ", ini.date)
+ cat(paste0("\n", tempo.cat, "\n"))
+ tempo.thread.nb = parallel::detectCores(all.tests = FALSE, logical = TRUE) # detect the number of threads
+ if(tempo.thread.nb < thread.nb){
+ thread.nb <- tempo.thread.nb
+ }
+ tempo.cat <- paste0("NUMBER OF THREADS USED: ", thread.nb)
+ cat(paste0("\n ", tempo.cat, "\n"))
+ Clust <- parallel::makeCluster(thread.nb, outfile = paste0(res.path, "/fun_test_parall_log.txt")) # outfile to print or cat during parallelization (only possible in a file, outfile = "" do not work on windows)
+ tempo.cat <- paste0("SPLIT OF TEST NUMBERS IN PARALLELISATION:")
+ cat(paste0("\n ", tempo.cat, "\n"))
+ cluster.list <- parallel::clusterSplit(Clust, 1:total.comp.nb) # split according to the number of cluster
+ str(cluster.list) # using print(str()) add a NULL below the result
+ cat("\n")
+ paral.output.list <- parallel::clusterApply( # paral.output.list is a list made of thread.nb compartments, each made of n / thread.nb (mat theo column number) compartment. Each compartment receive the corresponding results of fun_permut(), i.e., data (permuted mat1.perm), warning message, cor (final correlation) and count (number of permutations)
+ cl = Clust,
+ x = cluster.list,
+ function.name = function.name,
+ instruction = instruction,
+ thread.nb = thread.nb,
+ print.count = print.count,
+ total.comp.nb = total.comp.nb,
+ sp.plot.fun = sp.plot.fun,
+ i.list = i.list,
+ fun.tested = fun,
+ arg.values = arg.values,
+ fun.test = fun.test,
+ fun.test2 = fun.test2,
+ kind = kind,
+ problem = problem,
+ res = res,
+ count = count,
+ plot.count = plot.count,
+ data = data,
+ code = code,
+ plot.fun = plot.fun,
+ res.path = res.path,
+ lib.path = lib.path,
+ cute.path = cute.path,
+ fun = function(
+ x,
+ function.name,
+ instruction,
+ thread.nb,
+ print.count,
+ total.comp.nb,
+ sp.plot.fun,
+ i.list,
+ fun.tested,
+ arg.values,
+ fun.test,
+ fun.test2,
+ kind,
+ problem,
+ res,
+ count,
+ plot.count,
+ data,
+ code,
+ plot.fun,
+ res.path,
+ lib.path,
+ cute.path
+ ){
+ # check again: very important because another R
+ process.id <- Sys.getpid()
+ cat(paste0("\nPROCESS ID ", process.id, " -> TESTS ", x[1], " TO ", x[base::length(x)], "\n"))
+ source(cute.path, local = .GlobalEnv)
+ fun_pack(req.package = "lubridate", lib.path = lib.path, load = TRUE) # load = TRUE to be sure that functions are present in the environment. And this prevent to use R.lib.path argument of fun_python_pack()
+ # end check again: very important because another R
+ # plot management
+ if(plot.fun == TRUE){
+ pdf(file = paste0(res.path, "/plots_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".pdf", paste0("-", x[base::length(x)], ".pdf"))))
+ }else{
+ pdf(file = NULL) # send plots into a NULL file, no pdf file created
+ }
+ window.nb <- dev.cur()
+ invisible(dev.set(window.nb))
+ # end plot management
+ # new environment
+ ini.date <- Sys.time()
+ ini.time <- as.numeric(ini.date) # time of process begin, converted into
+ env.name <- paste0("env", ini.time)
+ if(exists(env.name, where = -1)){ # verify if still ok when fun_test() is inside a function
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ assign(env.name, new.env())
+ assign("val", val, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # var replaced by val
+ }
+ # end new environment
+ print.count.loop <- 0
+ suppressMessages(suppressWarnings(eval(parse(text = code))))
+ colnames(data) <- arg
+ if( ! is.null(expect.error)){
+ data <- data.frame(data, kind = kind, problem = problem, expected.error = expected.error, message = res, stringsAsFactors = FALSE)
+ }else{
+ data <- data.frame(data, kind = kind, problem = problem, message = res, stringsAsFactors = FALSE)
+ }
+ row.names(data) <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), x))
+ sys.info <- sessionInfo()
+ sys.info$loadedOnly <- sys.info$loadedOnly[order(names(sys.info$loadedOnly))] # sort the packages
+ invisible(dev.off(window.nb))
+ rm(env.name) # optional, because should disappear at the end of the function execution
+ # output
+ output <- list(fun = fun, instruction = instruction, sys.info = sys.info) # data = data finally removed from the output list, because everything combined in a RData file at the end
+ save(output, file = paste0(res.path, "/fun_test_", x[1], ifelse(base::length(x) == 1L, ".RData", paste0("-", x[base::length(x)], ".RData"))))
+ if(plot.fun == TRUE & plot.count == 0L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN PROCESS ", process.id, ": NO PDF PLOT BECAUSE ONLY ERRORS REPORTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ file.remove(paste0(res.path, "/plots_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".pdf", paste0("-", x[base::length(x)], ".pdf"))))
+ }
+ table.out <- as.matrix(data)
+ # table.out[table.out == ""] <- " " # does not work # because otherwise read.table() converts "" into NA
+ table.out <- gsub(table.out, pattern = "\n", replacement = " ")
+ write.table(table.out, file = paste0(res.path, "/table_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".txt", paste0("-", x[base::length(x)], ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
+ }
+ )
+ parallel::stopCluster(Clust)
+ # files assembly
+ if(base::length(cluster.list) > 1){
+ for(i2 in 1:base::length(cluster.list)){
+ tempo.file <- paste0(res.path, "/table_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".txt", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".txt"))) # txt file
+ tempo <- read.table(file = tempo.file, header = TRUE, stringsAsFactors = FALSE, sep = "\t", row.names = 1, comment.char = "", colClasses = "character") # row.names = 1 (1st column) because now read.table() adds a NA in the header if the header starts by a tabulation, comment.char = "" because colors with #, colClasses = "character" otherwise convert "" (from NULL) into NA
+ if(file.exists(paste0(res.path, "/plots_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".pdf", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".pdf"))))){
+ tempo.pdf <- paste0(res.path, "/plots_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".pdf", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".pdf"))) # pdf file
+ }else{
+ tempo.pdf <- NULL
+ }
+ tempo.rdata <- paste0(res.path, "/fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".RData", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".RData"))) # RData file
+ if(i2 == 1L){
+ final.file <- tempo
+ final.pdf <- tempo.pdf
+ # new env for RData combining
+ env.name <- paste0("env", ini.time)
+ if(exists(env.name, where = -1)){ # verify if still ok when fun_test() is inside a function
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ # end new env for RData combining
+ }else{
+ assign(env.name, new.env())
+ load(tempo.rdata, envir = get(env.name))
+ tempo.rdata1 <- tempo.rdata
+ assign("final.output", get("output", envir = get(env.name)), envir = get(env.name))
+ }
+ }else{
+ final.file <- rbind(final.file, tempo, stringsAsFactors = TRUE)
+ final.pdf <- c(final.pdf, tempo.pdf)
+ load(tempo.rdata, envir = get(env.name))
+ if( ! identical(get("final.output", envir = get(env.name))[c("R.version", "locale", "platform")], get("output", envir = get(env.name))[c("R.version", "locale", "platform")])){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIFFERENCE BETWEEN OUTPUTS WHILE THEY SHOULD BE IDENTICAL\nPLEASE CHECK\n", tempo.rdata1, "\n", tempo.rdata)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ # add the differences in RData $sysinfo into final.output
+ tempo.base1 <- sort(get("final.output", envir = get(env.name))$sys.info$basePkgs)
+ tempo.base2 <- sort(get("output", envir = get(env.name))$sys.info$basePkgs)
+ tempo.other1 <- names(get("final.output", envir = get(env.name))$sys.info$otherPkgs)
+ tempo.other2 <- names(get("output", envir = get(env.name))$sys.info$otherPkgs)
+ tempo.loaded1 <- names(get("final.output", envir = get(env.name))$sys.info$loadedOnly)
+ tempo.loaded2 <- names(get("output", envir = get(env.name))$sys.info$loadedOnly)
+ assign("final.output", {
+ x <- get("final.output", envir = get(env.name))
+ y <- get("output", envir = get(env.name))
+ x$sys.info$basePkgs <- sort(unique(tempo.base1, tempo.base2))
+ if( ! all(tempo.other2 %in% tempo.other1)){
+ x$sys.info$otherPkgs <- c(x$sys.info$otherPkgs, y$sys.info$otherPkgs[ ! (tempo.other2 %in% tempo.other1)])
+ x$sys.info$otherPkgs <- x$sys.info$otherPkgs[order(names(x$sys.info$otherPkgs))]
+ }
+ if( ! all(tempo.loaded2 %in% tempo.loaded1)){
+ x$sys.info$loadedOnly <- c(x$sys.info$loadedOnly, y$sys.info$loadedOnly[ ! (tempo.loaded2 %in% tempo.loaded1)])
+ x$sys.info$loadedOnly <- x$sys.info$loadedOnly[order(names(x$sys.info$loadedOnly))]
+ }
+ x
+ }, envir = get(env.name))
+ # add the differences in RData $sysinfo into final.output
+ }
+ }
+ file.remove(c(tempo.file, tempo.rdata))
+ }
+ # combine pdf and save
+ if( ! is.null(final.pdf)){
+ pdftools::pdf_combine(
+ input = final.pdf,
+ output = paste0(res.path, "/plots_from_fun_test_1-", total.comp.nb, ".pdf")
+ )
+ file.remove(final.pdf)
+ }
+ # end combine pdf and save
+ # save RData
+ assign("output", c(get("final.output", envir = get(env.name)), data = list(final.file)), envir = get(env.name))
+ save(output, file = paste0(res.path, "/fun_test__1-", total.comp.nb, ".RData"), envir = get(env.name))
+ rm(env.name) # optional, because should disappear at the end of the function execution
+ # end save RData
+ # save txt
+ write.table(final.file, file = paste0(res.path, "/table_from_fun_test_1-", total.comp.nb, ".txt"), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
+ # end save txt
+ if( ! is.null(expect.error)){
+ final.file <- final.file[ ! final.file$problem == final.file$expected.error, ]
+ if(nrow(final.file) == 0L){
+ cat(paste0("NO DISCREPANCY BETWEEN EXPECTED AND OBSERVED ERRORS\n\n"))
+ }else{
+ cat(paste0("DISCREPANCIES BETWEEN EXPECTED AND OBSERVED ERRORS (SEE THE discrepancy_table_from_fun_test_1-", total.comp.nb, ".txt FILE)\n\n"))
+ write.table(final.file, file = paste0(res.path, "/discrepancy_table_from_fun_test_1-", total.comp.nb, ".txt"), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
+ }
+ }
+ }
+ # end files assembly
+ }else{
+ # plot management
+ if(plot.fun == TRUE){
+ pdf(file = paste0(res.path, "/plots_from_fun_test_1", ifelse(total.comp.nb == 1L, ".pdf", paste0("-", total.comp.nb, ".pdf"))))
+ }else{
+ pdf(file = NULL) # send plots into a NULL file, no pdf file created
+ }
+ window.nb <- dev.cur()
+ invisible(dev.set(window.nb))
+ # end plot management
+ # new environment
+ env.name <- paste0("env", ini.time)
+ if(exists(env.name, where = -1)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ assign(env.name, new.env())
+ assign("val", val, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # var replaced by val
+ }
+ # end new environment
+ suppressMessages(suppressWarnings(eval(parse(text = code))))
+ colnames(data) <- arg
+ expect.data <- data.frame()
+ if( ! is.null(expect.error)){
+ data <- data.frame(data, kind = kind, problem = problem, expected.error = expected.error, message = res, stringsAsFactors = FALSE)
+ }else{
+ data <- data.frame(data, kind = kind, problem = problem, message = res, stringsAsFactors = FALSE)
+ }
+ row.names(data) <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), 1:total.comp.nb))
+ sys.info <- sessionInfo()
+ sys.info$loadedOnly <- sys.info$loadedOnly[order(names(sys.info$loadedOnly))] # sort the packages
+ invisible(dev.off(window.nb))
+ rm(env.name) # optional, because should disappear at the end of the function execution
+ # output
+ output <- list(fun = fun, instruction = instruction, sys.info = sys.info, data = data)
+ if(plot.fun == TRUE & plot.count == 0L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO PDF PLOT BECAUSE ONLY ERRORS REPORTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ file.remove(paste0(res.path, "/plots_from_fun_test_1", ifelse(total.comp.nb == 1L, ".pdf", paste0("-", total.comp.nb, ".pdf"))))
+ }
+ if( ! is.null(expect.error)){
+ expect.data <- output$data[ ! output$data$problem == output$data$expected.error, ]
+ if(nrow(expect.data) == 0L){
+ cat(paste0("NO DISCREPANCY BETWEEN EXPECTED AND OBSERVED ERRORS\n\n"))
+ }else{
+ cat(paste0("DISCREPANCIES BETWEEN EXPECTED AND OBSERVED ERRORS (SEE THE ", if(export == TRUE){paste0("discrepancy_table_from_fun_test_1", ifelse(total.comp.nb == 1L, "", paste0("-", total.comp.nb)), ".txt FILE")}else{"$data RESULT"}, ")\n\n"))
+ if(export == TRUE){
+ expect.data <- as.matrix(expect.data)
+ expect.data <- gsub(expect.data, pattern = "\n", replacement = " ")
+ write.table(expect.data, file = paste0(res.path, "/discrepancy_table_from_fun_test_1", ifelse(total.comp.nb == 1L, ".txt", paste0("-", total.comp.nb, ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
+ }
+ }
+ }
+ if( ! is.null(warn)){
+ base::options(warning.length = 8170)
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = base::options(warning.length = ini.warning.length), add = TRUE)
+ if(export == TRUE){
+ save(output, file = paste0(res.path, "/fun_test_1", ifelse(total.comp.nb == 1L, ".RData", paste0("-", total.comp.nb, ".RData"))))
+ table.out <- as.matrix(output$data)
+ table.out <- gsub(table.out, pattern = "\n", replacement = " ")
+ write.table(table.out, file = paste0(res.path, "/table_from_fun_test_1", ifelse(total.comp.nb == 1L, ".txt", paste0("-", total.comp.nb, ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
+ }else{
+ return(output)
+ }
+ }
+ # after return() ?
+ end.date <- Sys.time()
+ end.time <- as.numeric(end.date)
+ total.lapse <- round(lubridate::seconds_to_period(end.time - ini.time))
+ cat(paste0("fun_test JOB END\n\nTIME: ", end.date, "\n\nTOTAL TIME LAPSE: ", total.lapse, "\n\n\n"))
}
-######## fun_comp_2d() #### comparison of two 2D datasets (row & col names, dimensions, etc.)
+################ Object modification
-fun_comp_2d <- function(data1, data2){
-# AIM
-# compare two 2D datasets of the same class or not. Check and report in a list if the 2 datasets have:
-# same class
-# common row names
-# common column names
-# same row number
-# same column number
-# potential identical rows between the 2 datasets
-# potential identical columns between the 2 datasets
-# WARNINGS
-# For data frames: content are compared after conversion of content into characters. This means that the comparison of the content of data frame, either row to row, or column to column, does not take into account the mode in the different columns. This concern the results in $any.id.row, $same.row.pos1, $same.row.pos2, $same.row.match1, $same.row.match2, $any.id.col, $same.row.col1, $same.row.col2, $same.col.match1, $same.col.match2 and $identical.content result
-# "TOO BIG FOR EVALUATION" returned in $same.row.pos1, $same.row.pos2, $same.row.match1 and $same.row.match2 when nrow(data1) * nrow(data2) > 1e6 and $any.id.row remains NULL
-# "TOO BIG FOR EVALUATION" returned in $same.row.col1, $same.row.col2, $same.col.match1 and $same.col.match2 when ncol(data1) * ncol(data2) > 1e6 and $any.id.col remains NULL
-# ARGUMENTS
-# data1: matrix, data frame or table
-# data2: matrix, data frame or table
-# RETURN
-# a list containing:
-# $same.class: logical. Are class identical ?
-# $class: classes of the 2 datasets (NULL otherwise)
-# $same.dim: logical. Are dimension identical ?
-# $dim: dimension of the 2 datasets (NULL otherwise)
-# $same.row.nb: logical. Are number of rows identical ?
-# $row.nb: nb of rows of the 2 datasets if identical (NULL otherwise)
-# $same.col.nb: logical. Are number of columns identical ?
-# $col.nb: nb of columns of the 2 datasets if identical (NULL otherwise)
-# $same.row.name: logical. Are row names identical ? NULL if no row names in the two 2D datasets
-# $row.name: name of rows of the 2 datasets if identical (NULL otherwise)
-# $any.id.row.name: logical. Is there any row names identical ? NULL if no row names in the two 2D datasets
-# $same.row.names.pos1: positions, in data1, of the row names identical in data2
-# $same.row.names.pos2: positions, in data2, of the row names identical in data1
-# $same.row.names.match1: positions, in data2, of the row names that match the row names in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.row.names.match2: positions, in data1, of the row names that match the row names in data2, as given by match(data1, data2) (NULL otherwise)
-# $common.row.names: common row names between data1 and data2 (can be a subset of $name or not). NULL if no common row names
-# $same.col.name: logical. Are column names identical ? NULL if no col names in the two 2D datasets
-# $col.name: name of columns of the 2 datasets if identical (NULL otherwise)
-# $any.id.col.name: logical. Is there any column names identical ? NULL if no col names in the two 2D datasets
-# $same.col.names.pos1: positions, in data1, of the column names identical in data2
-# $same.col.names.pos2: positions, in data2, of the column names identical in data1
-# $same.col.names.match1: positions, in data2, of the column names that match the column names in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.col.names.match2: positions, in data1, of the column names that match the column names in data2, as given by match(data1, data2) (NULL otherwise)
-# $common.col.names: common column names between data1 and data2 (can be a subset of $name or not). NULL if no common column names
-# $any.id.row: logical. is there identical rows (not considering row names)? NULL if nrow(data1) * nrow(data2) > 1e10
-# $same.row.pos1: positions, in data1, of the rows identical in data2 (not considering row names). Return "TOO BIG FOR EVALUATION" if nrow(data1) * nrow(data2) > 1e10
-# $same.row.pos2: positions, in data2, of the rows identical in data1 (not considering row names). Return "TOO BIG FOR EVALUATION" if nrow(data1) * nrow(data2) > 1e10
-# $same.row.match1: positions, in data2, of the rows that match the rows in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.row.match2: positions, in data1, of the rows that match the rows in data2, as given by match(data1, data2) (NULL otherwise)
-# $any.id.col: logical. is there identical columns (not considering column names)? NULL if ncol(data1) * ncol(data2) > 1e10
-# $same.col.pos1: position in data1 of the cols identical in data2 (not considering column names). Return "TOO BIG FOR EVALUATION" if ncol(data1) * ncol(data2) > 1e10
-# $same.col.pos2: position in data2 of the cols identical in data1 (not considering column names). Return "TOO BIG FOR EVALUATION" if ncol(data1) * ncol(data2) > 1e10
-# $same.col.match1: positions, in data2, of the columns that match the columns in data1, as given by match(data1, data2) (NULL otherwise)
-# $same.row.match2: positions, in data1, of the columns that match the columns in data2, as given by match(data1, data2) (NULL otherwise)
-# $identical.object: logical. Are objects identical (including row & column names)?
-# $identical.content: logical. Are content objects identical (identical excluding row & column names)?
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# none
-# EXAMPLES
-# obs1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = as.data.frame(matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])), stringsAsFactors = TRUE) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
-# obs1 = matrix(101:110, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
-# large matrices
-# obs1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; obs2 = matrix(as.integer((1:1e6)+1e6/5), ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; head(obs1) ; head(obs2) ; fun_comp_2d(obs1, obs2)
-# WARNING: when comparing content (rows, columns, or total), double and integer data are considered as different -> double(1) != integer(1)
-# obs1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; obs2 = matrix((1:1e6)+1e6/5, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; head(obs1) ; head(obs2) ; fun_comp_2d(obs1, obs2)
-# Matrices: same row conten tand same row names
-# obs1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4]))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
-# Matrices: same row content but not same row names -> works: same content is identified
-# obs1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; obs2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("x", "z", "y"), c(LETTERS[1:2], "k", LETTERS[5:4]))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
-# obs1 = t(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = t(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1 ; obs2 ; fun_comp_2d(obs1, obs2)
-# Data frames: same row content and same row names, not same mode between columns
-# obs1 = as.data.frame(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = as.data.frame(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1[, 5] <- as.character(obs1[, 5]) ; obs2[, 5] <- as.character(obs2[, 5]) ; obs1 ; obs2 ; str(obs1) ; str(obs2) ; fun_comp_2d(obs1, obs2)
-# Data frames: same row content but not same row names -> works: same content is identified
-# obs1 = as.data.frame(matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5]))) ; obs2 = as.data.frame(matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("x", "z", "y"), c(LETTERS[1:2], "k", LETTERS[5:4])))) ; obs1[, 5] <- as.character(obs1[, 5]) ; obs2[, 5] <- as.character(obs2[, 5]) ; obs1 ; obs2 ; str(obs1) ; str(obs2) ; fun_comp_2d(obs1, obs2)
-# DEBUGGING
-# data1 = matrix(1:10, ncol = 5) ; data2 = matrix(1:10, ncol = 5) # for function debugging
-# data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5) # for function debugging
-# data1 = matrix(1:15, byrow = TRUE, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# data1 = matrix(1:15, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# data1 = matrix(1:15, ncol = 5, dimnames = list(paste0("A", letters[1:3]), LETTERS[1:5])) ; data2 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# data1 = matrix(1:15, ncol = 5, dimnames = list(letters[1:3], LETTERS[1:5])) ; data2 = matrix(1:12, ncol = 4, dimnames = list(letters[1:3], LETTERS[1:4])) # for function debugging
-# data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(101:110, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) # for function debugging
-# data1 = data.frame(a = 1:3, b= letters[1:3], row.names = LETTERS[1:3], stringsAsFactors = TRUE) ; data2 = data.frame(A = 1:3, B= letters[1:3], stringsAsFactors = TRUE) # for function debugging
-# data1 = matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = as.data.frame(matrix(1:10, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])), stringsAsFactors = TRUE) # for function debugging
-# data1 = matrix(1:10, byrow = TRUE, ncol = 5, dimnames = list(letters[1:2], LETTERS[1:5])) ; data2 = matrix(c(1:5, 101:105, 6:10), byrow = TRUE, ncol = 5, dimnames = list(c("a", "z", "b"), c(LETTERS[1:2], "k", LETTERS[5:4]))) # for function debugging
-# data1 = table(Exp1 = c("A", "A", "A", "B", "B", "B"), Exp2 = c("A1", "B1", "A1", "C1", "C1", "B1")) ; data2 = data.frame(A = 1:3, B= letters[1:3], stringsAsFactors = TRUE) # for function debugging
-# data1 = matrix(1:1e6, ncol = 5, dimnames = list(NULL, LETTERS[1:5])) ; data2 = matrix((1:1e6)+1e6/5, ncol = 5, dimnames = list(NULL, LETTERS[1:5]))
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# argument checking
-if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A MATRIX, DATA FRAME OR TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (any(class(data2) %in% c("data.frame", "table")) | all(class(data2) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data2) %in% c("matrix", "data.frame", "table"))
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A MATRIX, DATA FRAME OR TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
-# end argument checking
-# main code
-same.class <- NULL
-class <- NULL
-same.dim <- NULL
-dim <- NULL
-same.row.nb <- NULL
-row.nb <- NULL
-same.col.nb <- NULL
-col.nb <- NULL
-same.row.name <- NULL
-row.name <- NULL
-any.id.row.name <- NULL
-same.row.names.pos1 <- NULL
-same.row.names.pos2 <- NULL
-same.row.names.match1 <- NULL
-same.row.names.match2 <- NULL
-common.row.names <- NULL
-same.col.name <- NULL
-any.id.col.name <- NULL
-same.col.names.pos1 <- NULL
-same.col.names.pos2 <- NULL
-same.col.names.match1 <- NULL
-same.col.names.match2 <- NULL
-common.col.names <- NULL
-col.name <- NULL
-any.id.row <- NULL
-same.row.pos1 <- NULL
-same.row.pos2 <- NULL
-same.row.match1 <- NULL
-same.row.match2 <- NULL
-any.id.col <- NULL
-same.col.pos1 <- NULL
-same.col.pos2 <- NULL
-same.col.match1 <- NULL
-same.col.match2 <- NULL
-identical.object <- NULL
-identical.content <- NULL
-if(identical(data1, data2) & (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
-same.class <- TRUE
-class <- class(data1)
-same.dim <- TRUE
-dim <- dim(data1)
-same.row.nb <- TRUE
-row.nb <- nrow(data1)
-same.col.nb <- TRUE
-col.nb <- ncol(data1)
-same.row.name <- TRUE
-row.name <- dimnames(data1)[[1]]
-any.id.row.name <- TRUE
-same.row.names.pos1 <- 1:row.nb
-same.row.names.pos2 <- 1:row.nb
-same.row.names.match1 <- 1:row.nb
-same.row.names.match2 <- 1:row.nb
-common.row.names <- dimnames(data1)[[1]]
-same.col.name <- TRUE
-col.name <- dimnames(data1)[[2]]
-any.id.col.name <- TRUE
-same.col.names.pos1 <- 1:col.nb
-same.col.names.pos2 <- 1:col.nb
-same.col.names.match1 <- 1:col.nb
-same.col.names.match2 <- 1:col.nb
-common.col.names <- dimnames(data1)[[2]]
-any.id.row <- TRUE
-same.row.pos1 <- 1:row.nb
-same.row.pos2 <- 1:row.nb
-same.row.match1 <- 1:row.nb
-same.row.match2 <- 1:row.nb
-any.id.col <- TRUE
-same.col.pos1 <- 1:col.nb
-same.col.pos2 <- 1:col.nb
-same.col.match1 <- 1:col.nb
-same.col.match2 <- 1:col.nb
-identical.object <- TRUE
-identical.content <- TRUE
-}else{
-identical.object <- FALSE
-if(all(class(data1) == "table") & length(dim(data1)) == 1L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT IS A 1D TABLE. USE THE fun_comp_1d FUNCTION")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(all(class(data2) == "table") & length(dim(data2)) == 1L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT IS A 1D TABLE. USE THE fun_comp_1d FUNCTION")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! identical(class(data1), class(data2))){
-same.class <- FALSE
-}else if( ! (any(class(data1) %in% c("data.frame", "table")) | all(class(data1) %in% c("matrix", "array")))){ # before R4.0.0, it was ! any(class(data1) %in% c("matrix", "data.frame", "table"))
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 AND data2 ARGUMENTS MUST BE EITHER MATRIX, DATA FRAME OR TABLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-same.class <- TRUE
-class <- class(data1)
-}
-if( ! identical(dim(data1), dim(data2))){
-same.dim <- FALSE
-}else{
-same.dim <- TRUE
-dim <- dim(data1)
-}
-if( ! identical(nrow(data1), nrow(data2))){
-same.row.nb <- FALSE
-}else{
-same.row.nb <- TRUE
-row.nb <- nrow(data1)
-}
-if( ! identical(ncol(data1), ncol(data2))){
-same.col.nb <- FALSE
-}else{
-same.col.nb <- TRUE
-col.nb <- ncol(data1)
-}
-# row and col names
-if(is.null(dimnames(data1)) & is.null(dimnames(data2))){
-same.row.name <- NULL # but already NULL
-same.col.name <- NULL # but already NULL
-# other row names param remain NULL
-}else if((is.null(dimnames(data1)) & ! is.null(dimnames(data2))) | ( ! is.null(dimnames(data1)) & is.null(dimnames(data2)))){
-same.row.name <- FALSE
-same.col.name <- FALSE
-any.id.row.name <- FALSE
-any.id.col.name <- FALSE
-# other row names param remain NULL
-}else{
-# row names
-if(is.null(dimnames(data1)[[1]]) & is.null(dimnames(data2)[[1]])){
-same.row.name <- NULL # but already NULL
-# other row names param remain NULL
-}else if((is.null(dimnames(data1)[[1]]) & ! is.null(dimnames(data2)[[1]])) | ( ! is.null(dimnames(data1)[[1]]) & is.null(dimnames(data2)[[1]]))){
-same.row.name <- FALSE
-any.id.row.name <- FALSE
-# other row names param remain NULL
-}else if(identical(dimnames(data1)[[1]], dimnames(data2)[[1]])){
-same.row.name <- TRUE
-row.name <- dimnames(data1)[[1]]
-any.id.row.name <- TRUE
-same.row.names.pos1 <- 1:nrow(data1)
-same.row.names.pos2 <- 1:nrow(data1)
-same.row.names.match1 <- 1:nrow(data1)
-same.row.names.match2 <- 1:nrow(data1)
-common.row.names <- dimnames(data1)[[1]]
-}else{
-same.row.name <- FALSE
-any.id.row.name <- FALSE
-if(any(dimnames(data1)[[1]] %in% dimnames(data2)[[1]])){
-any.id.row.name <- TRUE
-same.row.names.pos1 <- which(dimnames(data1)[[1]] %in% dimnames(data2)[[1]])
-same.row.names.match1 <- match(dimnames(data1)[[1]], dimnames(data2)[[1]])
-}
-if(any(dimnames(data2)[[1]] %in% dimnames(data1)[[1]])){
-any.id.row.name <- TRUE
-same.row.names.pos2 <- which(dimnames(data2)[[1]] %in% dimnames(data1)[[1]])
-same.row.names.match2 <- match(dimnames(data2)[[1]], dimnames(data1)[[1]])
-}
-if(any.id.row.name == TRUE){
-common.row.names <- unique(c(dimnames(data1)[[1]][same.row.names.pos1], dimnames(data2)[[1]][same.row.names.pos2]))
-}
-}
-# col names
-if(is.null(dimnames(data1)[[2]]) & is.null(dimnames(data2)[[2]])){
-same.col.name <- NULL # but already NULL
-# other col names param remain NULL
-}else if((is.null(dimnames(data1)[[2]]) & ! is.null(dimnames(data2)[[2]])) | ( ! is.null(dimnames(data1)[[2]]) & is.null(dimnames(data2)[[2]]))){
-same.col.name <- FALSE
-any.id.col.name <- FALSE
-# other col names param remain NULL
-}else if(identical(dimnames(data1)[[2]], dimnames(data2)[[2]])){
-same.col.name <- TRUE
-col.name <- dimnames(data1)[[2]]
-any.id.col.name <- TRUE
-same.col.names.pos1 <- 1:ncol(data1)
-same.col.names.pos2 <- 1:ncol(data1)
-same.col.names.match1 <- 1:ncol(data1)
-same.col.names.match2 <- 1:ncol(data1)
-common.col.names <- dimnames(data1)[[2]]
-}else{
-same.col.name <- FALSE
-any.id.col.name <- FALSE
-if(any(dimnames(data1)[[2]] %in% dimnames(data2)[[2]])){
-any.id.col.name <- TRUE
-same.col.names.pos1 <- which(dimnames(data1)[[2]] %in% dimnames(data2)[[2]])
-same.col.names.match1 <- match(dimnames(data1)[[2]], dimnames(data2)[[2]])
-}
-if(any(dimnames(data2)[[2]] %in% dimnames(data1)[[2]])){
-any.id.col.name <- TRUE
-same.col.names.pos2 <- which(dimnames(data2)[[2]] %in% dimnames(data1)[[2]])
-same.col.names.match2 <- match(dimnames(data2)[[2]], dimnames(data1)[[2]])
-}
-if(any.id.col.name == TRUE){
-common.col.names <- unique(c(dimnames(data1)[[2]][same.col.names.pos1], dimnames(data2)[[2]][same.col.names.pos2]))
-}
-}
-}
-# identical row and col content
-if(all(class(data1) == "table")){
-data1 <- as.data.frame(matrix(data1, ncol = ncol(data1)), stringsAsFactors = FALSE) # conversion of table into data frame to facilitate inter class comparison
-}else if(all(class(data1) %in% c("matrix", "array"))){
-data1 <- as.data.frame(data1, stringsAsFactors = FALSE) # conversion of matrix into data frame to facilitate inter class comparison
-}else if(all(class(data1) == "data.frame")){
-# data1 <- data.frame(lapply(data1, as.character), stringsAsFactors = FALSE) # conversion of columns into characters
-}
-if(all(class(data2) == "table")){
-data2 <- as.data.frame(matrix(data2, ncol = ncol(data2)), stringsAsFactors = FALSE) # conversion of table into data frame to facilitate inter class comparison
-}else if(all(class(data2) %in% c("matrix", "array"))){
-data2 <- as.data.frame(data2, stringsAsFactors = FALSE) # conversion of matrix into data frame to facilitate inter class comparison
-}else if(all(class(data2) == "data.frame")){
-# data2 <- data.frame(lapply(data2, as.character), stringsAsFactors = FALSE) # conversion of columns into characters
-}
-row.names(data1) <- paste0("A", 1:nrow(data1))
-row.names(data2) <- paste0("A", 1:nrow(data2))
-if(same.col.nb == TRUE){ # because if not the same col nb, the row cannot be identical
-if(all(sapply(data1, FUN = typeof) == "integer") & all(sapply(data2, FUN = typeof) == "integer") & as.double(nrow(data1)) * nrow(data2) <= 1e10){ # fast method for integers (thus not data frames). as.double(nrow(data1)) to prevent integer overflow because R is 32 bits for integers
-tempo1 <- c(as.data.frame(t(data1), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
-tempo2 <- c(as.data.frame(t(data2), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
-same.row.pos1 <- which(tempo1 %in% tempo2)
-same.row.pos2 <- which(tempo2 %in% tempo1)
-same.row.match1 <- match(tempo1, tempo2)
-same.row.match2 <- match(tempo2, tempo1)
-}else if(as.double(nrow(data1)) * nrow(data2) <= 1e6){ # as.double(nrow(data1)) to prevent integer overflow because R is 32 bits for integers
-# inactivated because I would like to keep the mode during comparisons
-# if(col.nb <= 10){ # if ncol is not to big, the t() should not be that long
-# tempo1 <- c(as.data.frame(t(data1), stringsAsFactors = FALSE)) # conversion into list. This work fast with only integers (because 32 bits)
-# tempo2 <- c(as.data.frame(t(data2), stringsAsFactors = FALSE)) # conversion into list.
-# same.row.pos1 <- which(tempo1 %in% tempo2)
-# same.row.pos2 <- which(tempo2 %in% tempo1)
-# same.row.match1 <- match(tempo1, tempo2)
-# same.row.match2 <- match(tempo2, tempo1)
-# }else{
-# very long computation
-same.row.pos1 <- logical(length = nrow(data1)) # FALSE by default
-same.row.pos1[] <- FALSE # security
-same.row.pos2 <- logical(length = nrow(data2)) # FALSE by default
-same.row.pos2[] <- FALSE # security
-same.row.match1 <- rep(NA, nrow(data1))
-same.row.match2 <- rep(NA, nrow(data2))
-for(i3 in 1:nrow(data1)){
-for(i4 in 1:nrow(data2)){
-tempo1 <- data1[i3, ]
-tempo2 <- data2[i4, ]
-rownames(tempo1) <- NULL # to have same row and column names
-colnames(tempo1) <- NULL # to have same row and column names
-rownames(tempo2) <- NULL # to have same row and column names
-colnames(tempo2) <- NULL # to have same row and column names
-if(identical(tempo1, tempo2)){
-same.row.pos1[i3] <- TRUE
-same.row.pos2[i4] <- TRUE
-same.row.match1[i3] <- i4
-same.row.match2[i4] <- i3
-}
-}
-}
-same.row.pos1 <- which(same.row.pos1)
-same.row.pos2 <- which(same.row.pos2)
-# }
-}else{
-same.row.pos1 <- "TOO BIG FOR EVALUATION"
-same.row.pos2 <- "TOO BIG FOR EVALUATION"
-same.row.match1 <- "TOO BIG FOR EVALUATION"
-same.row.match2 <- "TOO BIG FOR EVALUATION"
-}
+######## fun_name_change() #### check a vector of character strings and modify any string if present in another vector
-names(same.row.pos1) <- NULL
-names(same.row.pos2) <- NULL
-if(all(is.na(same.row.pos1))){
-same.row.pos1 <- NULL
-}else{
-same.row.pos1 <- same.row.pos1[ ! is.na(same.row.pos1)]
-any.id.row <- TRUE
-}
-if(all(is.na(same.row.pos2))){
-same.row.pos2 <- NULL
-}else{
-same.row.pos2 <- same.row.pos2[ ! is.na(same.row.pos2)]
-any.id.row <- TRUE
-}
-if(is.null(same.row.pos1) & is.null(same.row.pos2)){
-any.id.row <- FALSE
-}else if(length(same.row.pos1) == 0L & length(same.row.pos2) == 0L){
-any.id.row <- FALSE
-}else if(all(same.row.pos1 == "TOO BIG FOR EVALUATION") & all(same.row.pos2 == "TOO BIG FOR EVALUATION")){
-any.id.row <- NULL
-}
-}else{
-any.id.row <- FALSE
-# same.row.pos1 and 2 remain NULL
-}
-if(same.row.nb == TRUE){ # because if not the same row nb, the col cannot be identical
-if(as.double(ncol(data1)) * ncol(data2) <= 1e10){ # comparison of data frame columns is much easier than rows because no need to use t() before converting to list for fast comparison. as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
-# if(all(sapply(data1, FUN = typeof) == "integer") & all(sapply(data2, FUN = typeof) == "integer") & as.double(ncol(data1)) * ncol(data2) <= 1e10){ # fast method for integers (thus not data frames). as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
-tempo1 <- c(data1)
-tempo2 <- c(data2)
-same.col.pos1 <- which(tempo1 %in% tempo2)
-same.col.pos2 <- which(tempo2 %in% tempo1)
-same.col.match1 <- match(tempo1, tempo2)
-same.col.match2 <- match(tempo2, tempo1)
-# }else if(as.double(ncol(data1)) * ncol(data2) <= 1e6){ # as.double(ncol(data1)) to prevent integer overflow because R is 32 bits for integers
-# same.col.pos1 <- logical(length = ncol(data1)) # FALSE by default
-# same.col.pos1[] <- FALSE # security
-# same.col.pos2 <- logical(length = ncol(data2)) # FALSE by default
-# same.col.pos2[] <- FALSE # security
-# same.col.match1 <- rep(NA, ncol(data1))
-# same.col.match2 <- rep(NA, ncol(data2))
-# for(i3 in 1:ncol(data1)){
-# for(i4 in 1:ncol(data2)){
-# if(identical(data1[ , i3], data2[ , i4])){
-# same.col.pos1[i3] <- TRUE
-# same.col.pos2[i4] <- TRUE
-# same.col.match1[i3] <- i4
-# same.col.match2[i4] <- i3
-# }
-# }
-# }
-# same.col.pos1 <- which(same.col.pos1)
-# same.col.pos2 <- which(same.col.pos2)
-}else{
-same.col.pos1 <- "TOO BIG FOR EVALUATION"
-same.col.pos2 <- "TOO BIG FOR EVALUATION"
-}
-names(same.col.pos1) <- NULL
-names(same.col.pos2) <- NULL
-if(all(is.na(same.col.pos1))){
-same.col.pos1 <- NULL
-}else{
-same.col.pos1 <- same.col.pos1[ ! is.na(same.col.pos1)]
-any.id.col <- TRUE
-}
-if(all(is.na(same.col.pos2))){
-same.col.pos2 <- NULL
-}else{
-same.col.pos2 <- same.col.pos2[ ! is.na(same.col.pos2)]
-any.id.col <- TRUE
-}
-if(is.null(same.col.pos1) & is.null(same.col.pos2)){
-any.id.col <- FALSE
-}else if(length(same.col.pos1) == 0L & length(same.col.pos2) == 0L){
-any.id.col <- FALSE
-}else if(all(same.col.pos1 == "TOO BIG FOR EVALUATION") & all(same.col.pos2 == "TOO BIG FOR EVALUATION")){
-any.id.col <- NULL
-}
-}else{
-any.id.col <- FALSE
-# same.col.pos1 and 2 remain NULL
-}
-if(same.dim == TRUE){
-names(data1) <- NULL
-row.names(data1) <- NULL
-names(data2) <- NULL
-row.names(data2) <- NULL
-if(identical(data1, data2)){
-identical.content <- TRUE
-}else{
-identical.content <- FALSE
-}
-}else{
-identical.content <- FALSE
-}
-}
-output <- list(same.class = same.class, class = class, same.dim = same.dim, dim = dim, same.row.nb = same.row.nb, row.nb = row.nb, same.col.nb = same.col.nb , col.nb = col.nb, same.row.name = same.row.name, row.name = row.name, any.id.row.name = any.id.row.name, same.row.names.pos1 = same.row.names.pos1, same.row.names.pos2 = same.row.names.pos2, same.row.names.match1 = same.row.names.match1, same.row.names.match2 = same.row.names.match2, common.row.names = common.row.names, same.col.name = same.col.name, col.name = col.name,any.id.col.name = any.id.col.name, same.col.names.pos1 = same.col.names.pos1, same.col.names.pos2 = same.col.names.pos2, same.col.names.match1 = same.col.names.match1, same.col.names.match2 = same.col.names.match2, common.col.names = common.col.names, any.id.row = any.id.row, same.row.pos1 = same.row.pos1, same.row.pos2 = same.row.pos2, same.row.match1 = same.row.match1, same.row.match2 = same.row.match2, any.id.col = any.id.col, same.col.pos1 = same.col.pos1, same.col.pos2 = same.col.pos2, same.col.match1 = same.col.match1, same.col.match2 = same.col.match2, identical.object = identical.object, identical.content = identical.content)
-return(output)
+
+fun_name_change <- function(data1, data2, added.string = "_modif"){
+ # AIM
+ # this function allow to check if a vector of character strings, like column names of a data frame, has elements present in another vector (vector of reserved words or column names of another data frame before merging)
+ # ARGUMENTS
+ # data1: vector of character strings to check and modify
+ # data2: reference vector of character strings
+ # added.string: string added at the end of the modified string in data1 if present in data2
+ # RETURN
+ # a list containing
+ # $data: the modified data1 (in the same order as in the initial data1)
+ # $ini: the initial elements before modification. NULL if no modification
+ # $post: the modified elements in the same order as in ini. NULL if no modification
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # obs1 <- c("A", "B", "C", "D") ; obs2 <- c("A", "C") ; fun_name_change(obs1, obs2)
+ # obs1 <- c("A", "B", "C", "C_modif1", "D") ; obs2 <- c("A", "A_modif1", "C") ; fun_name_change(obs1, obs2) # the function checks that the new names are neither in obs1 nor in obs2 (increment the number after the added string)
+ # DEBUGGING
+ # data1 = c("A", "B", "C", "D") ; data2 <- c("A", "C") ; added.string = "_modif" # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data1, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = data2, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = added.string, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ ini <- NULL
+ post <- NULL
+ if(any(data1 %in% data2)){
+ tempo.names <- data1[data1 %in% data2]
+ ini <- NULL
+ post <- NULL
+ for(i2 in 1:length(tempo.names)){
+ count <- 0
+ tempo <- tempo.names[i2]
+ while(any(tempo %in% data2) | any(tempo %in% data1)){
+ count <- count + 1
+ tempo <- paste0(tempo.names[i2], "_modif", count)
+ }
+ data1[data1 %in% tempo.names[i2]] <- paste0(tempo.names[i2], "_modif", count)
+ if(count != 0){
+ ini <- c(ini, tempo.names[i2])
+ post <- c(post, paste0(tempo.names[i2], "_modif", count))
+ }
+ }
+ data <- data1
+ }else{
+ data <- data1
+ }
+ output <- list(data = data, ini = ini, post = post)
+ return(output)
}
-######## fun_comp_list() #### comparison of two lists
+######## fun_df_remod() #### remodeling a data frame to have column name as a qualitative values and vice-versa
-fun_comp_list <- function(data1, data2){
-# AIM
-# compare two lists. Check and report in a list if the 2 datasets have:
-# same length
-# common names
-# common compartments
-# ARGUMENTS
-# data1: list
-# data2: list
-# RETURN
-# a list containing:
-# $same.length: logical. Are number of elements identical?
-# $length: number of elements in the 2 datasets (NULL otherwise)
-# $same.names: logical. Are element names identical ?
-# $name: name of elements of the 2 datasets if identical (NULL otherwise)
-# $any.id.name: logical. Is there any element names identical ?
-# $same.names.pos1: positions, in data1, of the element names identical in data2
-# $same.names.pos2: positions, in data2, of the compartment names identical in data1
-# $any.id.compartment: logical. is there any identical compartments ?
-# $same.compartment.pos1: positions, in data1, of the compartments identical in data2
-# $same.compartment.pos2: positions, in data2, of the compartments identical in data1
-# $identical.object: logical. Are objects identical (kind of object, compartment names and content)?
-# $identical.content: logical. Are content objects identical (identical compartments excluding compartment names)?
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# none
-# EXAMPLES
-# obs1 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; obs2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; fun_comp_list(obs1, obs2)
-# obs1 = list(1:5, LETTERS[1:2]) ; obs2 = list(a = 1:5, b = LETTERS[1:2]) ; fun_comp_list(obs1, obs2)
-# obs1 = list(b = 1:5, c = LETTERS[1:2]) ; obs2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; fun_comp_list(obs1, obs2)
-# obs1 = list(b = 1:5, c = LETTERS[1:2]) ; obs2 = list(LETTERS[5:9], matrix(1:6), 1:5) ; fun_comp_list(obs1, obs2)
-# DEBUGGING
-# data1 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) ; data2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) # for function debugging
-# data1 = list(a = 1:5, b = LETTERS[1:2]) ; data2 = list(a = 1:5, b = LETTERS[1:2], d = matrix(1:6)) # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# argument checking
-if( ! any(class(data1) %in% "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! any(class(data2) %in% "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) # activate this line and use the function to check arguments status
-# end argument checking
-# main code
-same.length <- NULL
-length <- NULL
-same.names <- NULL
-name <- NULL
-any.id.name <- NULL
-same.names.pos1 <- NULL
-same.names.pos2 <- NULL
-any.id.compartment <- NULL
-same.compartment.pos1 <- NULL
-same.compartment.pos2 <- NULL
-identical.object <- NULL
-identical.content <- NULL
-if(identical(data1, data2)){
-same.length <- TRUE
-length <- length(data1)
-if( ! is.null(names(data1))){
-same.names <- TRUE
-name <- names(data1)
-any.id.name <- TRUE
-same.names.pos1 <- 1:length(data1)
-same.names.pos2 <- 1:length(data2)
-}
-any.id.compartment <- TRUE
-same.compartment.pos1 <- 1:length(data1)
-same.compartment.pos2 <- 1:length(data2)
-identical.object <- TRUE
-identical.content <- TRUE
-}else{
-identical.object <- FALSE
-if( ! identical(length(data1), length(data2))){
-same.length<- FALSE
-}else{
-same.length<- TRUE
-length <- length(data1)
-}
-if( ! (is.null(names(data1)) & is.null(names(data2)))){
-if( ! identical(names(data1), names(data2))){
-same.names <- FALSE
-}else{
-same.names <- TRUE
-name <- names(data1)
-}
-any.id.name <- FALSE
-if(any(names(data1) %in% names(data2))){
-any.id.name <- TRUE
-same.names.pos1 <- which(names(data1) %in% names(data2))
+fun_df_remod <- function(
+ data,
+ quanti.col.name = "quanti",
+ quali.col.name = "quali"
+){
+ # AIM
+ # if the data frame is made of n numeric columns, a new data frame is created, with the 1st column gathering all the numeric values, and the 2nd column being the name of the columns of the initial data frame. If row names were present in the initial data frame, then a new ini_rowname column is added with the names of the rows
+
+
+ # If the data frame is made of one numeric column and one character or factor column, a new data frame is created, with the new columns corresponding to the split numeric values (according to the character column). NA are added a the end of each column to have the same number of rows. BEWARE: in such data frame, rows are not individuals. This means that in the example below, values 10 and 20 are associated on the same row but that means nothing in term of association
+
+
+
+ # ARGUMENTS
+ # data: data frame to convert
+ # quanti.col.name: optional name for the quanti column of the new data frame
+ # quali.col.name: optional name for the quali column of the new data frame
+ # RETURN
+ # the modified data frame
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # obs <- data.frame(col1 = (1:4)*10, col2 = c("A", "B", "A", "A"), stringsAsFactors = TRUE) ; obs ; fun_df_remod(obs)
+ # obs <- data.frame(col1 = (1:4)*10, col2 = 5:8, stringsAsFactors = TRUE) ; obs ; fun_df_remod(obs, quanti.col.name = "quanti", quali.col.name = "quali")
+ # obs <- data.frame(col1 = (1:4)*10, col2 = 5:8, stringsAsFactors = TRUE) ; rownames(obs) <- paste0("row", 1:4) ; obs ; fun_df_remod(obs, quanti.col.name = "quanti", quali.col.name = "quali")
+ # DEBUGGING
+ # data = data.frame(a = 1:3, b = 4:6, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
+ # data = data.frame(a = 1:3, b = 4:6, c = 11:13, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
+ # data = data.frame(a = 1:3, b = letters[1:3], stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
+ # data = data.frame(a = 1:3, b = letters[1:3], stringsAsFactors = TRUE) ; quanti.col.name = "TEST" ; quali.col.name = "quali" # for function debugging
+ # data = data.frame(b = letters[1:3], a = 1:3, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
+ # data = data.frame(b = c("e", "e", "h"), a = 1:3, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking without fun_check()
+ if( ! any(class(data) %in% "data.frame")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data ARGUMENT MUST BE A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking without fun_check()
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = quanti.col.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = quali.col.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ tempo.factor <- unlist(lapply(data, class))
+ for(i in 1:length(tempo.factor)){ # convert factor columns as character
+ if(all(tempo.factor[i] == "factor")){
+ data[, i] <- as.character(data[, i])
+ }
+ }
+ tempo.factor <- unlist(lapply(data, mode))
+ if(length(data) == 2L){
+ if( ! ((base::mode(data[, 1]) == "character" & base::mode(data[, 2]) == "numeric") | base::mode(data[, 2]) == "character" & base::mode(data[, 1]) == "numeric" | base::mode(data[, 2]) == "numeric" & base::mode(data[, 1]) == "numeric") ){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": IF data ARGUMENT IS A DATA FRAME MADE OF 2 COLUMNS, EITHER A COLUMN MUST BE NUMERIC AND THE OTHER CHARACTER, OR THE TWO COLUMNS MUST BE NUMERIC")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if((base::mode(data[, 1]) == "character" | base::mode(data[, 2]) == "character") & (quanti.col.name != "quanti" | quali.col.name != "quali")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": IMPROPER quanti.col.name OR quali.col.name RESETTINGS. THESE ARGUMENTS ARE RESERVED FOR DATA FRAMES MADE OF n NUMERIC COLUMNS ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ if( ! all(tempo.factor %in% "numeric")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": IF data ARGUMENT IS A DATA FRAME MADE OF ONE COLUMN, OR MORE THAN 2 COLUMNS, THESE COLUMNS MUST BE NUMERIC")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(( ! any(tempo.factor %in% "character")) & is.null(names(data))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": NUMERIC DATA FRAME in the data ARGUMENT MUST HAVE COLUMN NAMES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(all(tempo.factor %in% "numeric")){ # transfo 1
+ quanti <- NULL
+ for(i in 1:length(data)){
+ quanti <-c(quanti, data[, i])
+ }
+ quali <- rep(names(data), each = nrow(data))
+ output.data <- data.frame(quanti, quali, stringsAsFactors = TRUE, check.names = FALSE)
+ names(output.data) <- c(quanti.col.name, quali.col.name)
+ # add the ini_rowname column
+ ini.rownames <- rownames(data)
+ tempo.data <- data
+ rownames(tempo.data) <- NULL
+ null.rownames <- (tempo.data)
+ if( ! identical(ini.rownames, null.rownames)){
+ ini_rowname <- rep(ini.rownames, times = ncol(data))
+ output.data <- cbind(output.data, ini_rowname, stringsAsFactors = TRUE)
+ }
+ }else{ # transfo 2
+ if(class(data[, 1]) == "character"){
+ data <- cbind(data[2], data[1], stringsAsFactors = TRUE)
+ }
+ nc.max <- max(table(data[, 2])) # effectif maximum des classes
+ nb.na <- nc.max - table(data[,2]) # nombre de NA à ajouter pour réaliser la data frame
+ tempo<-split(data[, 1], data[, 2])
+ for(i in 1:length(tempo)){tempo[[i]] <- append(tempo[[i]], rep(NA, nb.na[i]))} # des NA doivent être ajoutés lorsque les effectifs sont différents entre les classes. C'est uniquement pour que chaque colonne ait le même nombre de lignes
+ output.data<-data.frame(tempo, stringsAsFactors = TRUE, check.names = FALSE)
+ }
+ return(output.data)
}
-if(any(names(data2) %in% names(data1))){
-any.id.name <- TRUE
-same.names.pos2 <- which(names(data2) %in% names(data1))
+
+
+
+
+######## fun_round() #### rounding number if decimal present
+
+
+fun_round <- function(data, dec.nb = 2, after.lead.zero = TRUE){
+ # AIM
+ # round a vector of values, if decimal, with the desired number of decimal digits after the decimal leading zeros
+ # WARNINGS
+ # Work well with numbers as character strings, but not always with numerical numbers because of the floating point
+ # Numeric values are really truncated from a part of their decimal digits, whatever options(digits) settings
+ # See ?.Machine or https://stackoverflow.com/questions/5173692/how-to-return-number-of-decimal-places-in-r, with the interexting formula: abs(x - round(x)) > .Machine$double.eps^0.5
+ # ARGUMENTS
+ # data: a vector of numbers (numeric or character mode)
+ # dec.nb: number of required decimal digits
+ # after.lead.zero: logical. If FALSE, rounding is performed for all the decimal numbers, whatever the leading zeros (e.g., 0.123 -> 0.12 and 0.00128 -> 0.00). If TRUE, dec.nb are taken after the leading zeros (e.g., 0.123 -> 0.12 and 0.00128 -> 0.0013)
+ # RETURN
+ # the modified vector
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, 10, 100.001, 333.0001254, 12312.1235), dec.nb = 2, after.lead.zero = FALSE), "\n\n") ; options(digits = ini.options)
+ # ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, 10, 100.001, 333.0001254, 12312.1235), dec.nb = 2, after.lead.zero = TRUE), "\n\n") ; options(digits = ini.options)
+ # ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, "10", "100.001", "333.0001254", "12312.1235"), dec.nb = 2, after.lead.zero = FALSE), "\n\n") ; options(digits = ini.options)
+ # ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, "10", "100.001", "333.0001254", "12312.1235"), dec.nb = 2, after.lead.zero = TRUE), "\n\n") ; options(digits = ini.options)
+ # DEBUGGING
+ # data = data = c(10, 100.001, 333.0001254, 12312.1235) ; dec.nb = 2 ; after.lead.zero = FALSE # # for function debugging
+ # data = data = c("10", "100.001", "333.0001254", "12312.1235") ; dec.nb = 2 ; after.lead.zero = TRUE # # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking without fun_check()
+ if( ! (all(typeof(data) == "character") | all(typeof(data) == "double") | all(typeof(data) == "integer"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A VECTOR OF NUMBERS (IN NUMERIC OR CHARACTER MODE)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking without fun_check()
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data, class = "vector", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dec.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = after.lead.zero, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ tempo <- grepl(x = data, pattern = "\\.") # detection of decimal numbers
+ ini.mode <- base::mode(data)
+ data <- as.character(data) # to really truncate decimal digits
+ for(i in 1:length(data)){ # scan all the numbers of the vector
+ if(tempo[i] == TRUE){ # means decimal number
+ if(after.lead.zero == TRUE){
+ zero.pos <- unlist(gregexpr(text=data[i], pattern = 0)) # recover all the position of the zeros in the number. -1 if no zeros (do not record the leading and trailing zeros)
+ }else{
+ zero.pos <- -1 # -1 as if no zero
+ }
+ dot.pos <- unlist(gregexpr(text=data[i], pattern = "\\.")) # recover all the position of the zeros in the number
+ digit.pos <- unlist(gregexpr(text=data[i], pattern = "[[:digit:]]")) # recover all the position of the digits in the number
+ dec.pos <- digit.pos[digit.pos > dot.pos]
+ count <- 0
+ while((dot.pos + count + 1) %in% zero.pos & (dot.pos + count + 1) <= max(dec.pos) & (count + dec.nb) < length(dec.pos)){ # count the number of leading zeros in the decimal part
+ count <- count + 1
+ }
+ data[i] <- formatC(as.numeric(data[i]), digits = (count + dec.nb), format = "f")
+ }
+ }
+ if(ini.mode != "character"){
+ data <- as.numeric(data)
+ }
+ return(data)
}
+
+
+######## fun_mat_rotate() #### 90° clockwise matrix rotation
+
+
+fun_mat_rotate <- function(data){
+ # AIM
+ # 90° clockwise matrix rotation
+ # applied twice, the function provide the mirror matrix, according to vertical and horizontal symmetry
+ # ARGUMENTS
+ # data: matrix (matrix class)
+ # RETURN
+ # the modified matrix
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # obs <- matrix(1:10, ncol = 1) ; obs ; fun_mat_rotate(obs)
+ # obs <- matrix(LETTERS[1:10], ncol = 5) ; obs ; fun_mat_rotate(obs)
+ # DEBUGGING
+ # data = matrix(1:10, ncol = 1)
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data, class = "matrix", fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ for (i in 1:ncol(data)){data[,i] <- rev(data[,i])}
+ data <- t(data)
+ return(data)
}
-names(data1) <- NULL
-names(data2) <- NULL
-any.id.compartment <- FALSE
-if(any(data1 %in% data2)){
-any.id.compartment <- TRUE
-same.compartment.pos1 <- which(data1 %in% data2)
+
+
+######## fun_mat_num2color() #### convert a numeric matrix into hexadecimal color matrix
+
+
+fun_mat_num2color <- function(
+ mat1,
+ mat.hsv.h = TRUE,
+ notch = 1,
+ s = 1,
+ v = 1,
+ forced.color = NULL
+){
+ # AIM
+ # convert a matrix made of numbers into a hexadecimal matrix for rgb colorization
+ # ARGUMENTS:
+ # mat1: matrix 1 of non negative numerical values that has to be colored (matrix class). NA allowed
+ # mat.hsv.h: logical. Is mat1 the h of hsv colors ? (if TRUE, mat1 must be between zero and 1). If FALSE, mat1 must be made of positive integer values without 0
+ # notch: single value between 0 and 1 to shift the successive colors on the hsv circle by + notch
+ # s: s argument of hsv(). Must be between 0 and 1
+ # v: v argument of hsv(). Must be between 0 and 1
+ # forced.color: Must be NULL or hexadecimal color code or name given by colors(). The first minimal values of mat1 will be these colors. All the color of mat1 can be forced using this argument
+ # RETURN
+ # a list containing:
+ # $mat1.name: name of mat1
+ # $colored.mat: colors of mat1 in hexa
+ # $problem: logical. Is any colors of forced.color overlap the colors designed by the function. NULL if forced.color = NULL
+ # $text.problem: text when overlapping colors. NULL if forced.color = NULL or problem == FALSE
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # mat1 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; dimnames(mat1) <- list(LETTERS[1:4], letters[1:2]) ; fun_mat_num2color(mat1, mat.hsv.h = FALSE, notch = 1, s = 1, v = 1, forced.color = NULL)
+ # DEBUGGING
+ # mat1 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; dimnames(mat1) <- list(LETTERS[1:4], letters[1:2]); mat.hsv.h = FALSE ; notch = 1 ; s = 1 ; v = 1 ; forced.color = c(hsv(1,1,1), hsv(0,0,0)) # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = mat1, mode = "numeric", class = "matrix", na.contain = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = mat.hsv.h, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = notch, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = s, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = v, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # argument checking without fun_check()
+ if(mat.hsv.h == TRUE & fun_check(data = mat1, mode = "numeric", prop = TRUE)$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 ARGUMENT MUST BE A MATRIX OF PROPORTIONS SINCE THE mat.hsv.h ARGUMENT IS SET TO TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(forced.color)){
+ tempo <- fun_check(data = forced.color, class = "character")
+ if(any(tempo$problem == TRUE)){
+ paste0("\n\n================\n\n", paste(tempo$text[tempo$problem], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! all(forced.color %in% colors() | grepl(pattern = "^#", forced.color))){ # check that all strings of forced.color start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": forced.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end argument checking without fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ problem <- NULL
+ text.problem <- NULL
+ mat1.name <- deparse(substitute(mat1))
+ # change the scale of the plotted matrix
+ if(mat.hsv.h == TRUE){
+ if(any(min(mat1, na.rm = TRUE) < 0 | max(mat1, na.rm = TRUE) > 1, na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 MUST BE MADE OF VALUES BETWEEN 0 AND 1 BECAUSE mat.hsv.h ARGUMENT SET TO TRUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ if(any(mat1 - floor(mat1) > 0, na.rm = TRUE) | any(mat1 == 0L, na.rm = TRUE)){ # no need of isTRUE(all.equal()) because we do not require approx here but strictly 0, thus == is ok
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 MUST BE MADE OF INTEGER VALUES WITHOUT 0 BECAUSE mat.hsv.h ARGUMENT SET TO FALSE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ mat1 <- mat1 / max(mat1, na.rm = TRUE)
+ }
+ }
+ if(notch != 1){
+ different.color <- unique(as.vector(mat1))
+ different.color <- different.color[ ! is.na(different.color)]
+ tempo.different.color <- different.color + c(0, cumsum(rep(notch, length(different.color) - 1)))
+ tempo.different.color <- tempo.different.color - floor(tempo.different.color)
+ if(any(duplicated(tempo.different.color) == TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DUPLICATED VALUES AFTER USING notch (", paste(tempo.different.color[duplicated(tempo.different.color)], collapse = " "), "). TRY ANOTHER notch VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(length(different.color) != length(tempo.different.color)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF different.color (", paste(different.color, collapse = " "), ") DIFFERENT FROM LENGTH OF tempo.different.color (", paste(tempo.different.color, collapse = " "), ")")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ for(i in 1:length(different.color)){
+ mat1[mat1 == different.color[i]] <- tempo.different.color[i] # no need of isTRUE(all.equal()) because different.color comes from mat1
+ }
+ }
+ }
+ if( ! is.null(forced.color)){
+ hexa.values.to.change <- hsv(unique(sort(mat1))[1:length(forced.color)], s, v)
+ }
+ mat1[ ! is.na(mat1)] <- hsv(mat1[ ! is.na(mat1)], s, v)
+ if( ! is.null(forced.color)){
+ if(any(forced.color %in% mat1, na.rm = TRUE)){
+ problem <- TRUE
+ text.problem <- paste0("THE FOLLOWING COLORS WHERE INTRODUCED USING forced.color BUT WHERE ALREADY PRESENT IN THE COLORED MATRIX :", paste(forced.color[forced.color %in% mat1], collapse = " "))
+ }else{
+ problem <- FALSE
+ }
+ for(i in 1:length(hexa.values.to.change)){
+ if( ! any(mat1 == hexa.values.to.change[i], na.rm = TRUE)){# no need of isTRUE(all.equal()) because character
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE ", hexa.values.to.change[i], " VALUE FROM hexa.values.to.change IS NOT REPRESENTED IN mat1 : ", paste(unique(as.vector(mat1)), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ mat1[which(mat1 == hexa.values.to.change[i])] <- forced.color[i] # no need of isTRUE(all.equal()) because character
+ }
+ }
+ }
+ output <- list(mat1.name = mat1.name, colored.mat = mat1, problem = problem, text.problem = text.problem)
+ return(output)
}
-if(any(data2 %in% data1)){
-any.id.compartment <- TRUE
-same.compartment.pos2 <- which(data2 %in% data1)
+
+
+######## fun_mat_op() #### assemble several matrices with operation
+
+
+fun_mat_op <- function(mat.list, kind.of.operation = "+"){
+ # AIM
+ # assemble several matrices of same dimensions by performing by case operation. For instance add the value of all the case 1 (row1 & column1) of the matrices and put it in the case 1 of a new matrix M, add the value of all the case 2 (row2 & column1) of the matrices and put it in the case 2 of a new matrix M, etc.
+
+ # c: case
+ # i: row number
+ # j: column number
+ # k: matrix number
+ # z: number of matrices
+ # ARGUMENTS:
+ # mat.list: list of matrices
+ # kind.of.operation: either "+" (by case addition), "-" (by case subtraction) or "*" (by case multiplication)
+ # RETURN
+ # the assembled matrix, with row and/or column names only if all the matrices have identical row/column names
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_comp_2d()
+ # EXAMPLES
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "+")
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "*")
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], c(NA, NA))) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "-")
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(c("A1", "A2", "A3", "A4"), letters[1:2])) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat3 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2, mat3), kind.of.operation = "+")
+ # DEBUGGING
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; mat.list = list(mat1, mat2) ; kind.of.operation = "+" # for function debugging
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], c(NA, NA))) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat.list = list(mat1, mat2) ; kind.of.operation = "*" # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_comp_2d() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = mat.list, class = "list", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = kind.of.operation, options = c("+", "-", "*"), length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # argument checking without fun_check()
+ if(length(mat.list) < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat.list ARGUMENT MUST BE A LIST CONTAINING AT LEAST 2 MATRICES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ for(i1 in 1:length(mat.list)){
+ tempo <- fun_check(data = mat.list[[i1]], class = "matrix", mode = "numeric", na.contain = TRUE)
+ if(tempo$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ELEMENT ", i1, " OF mat.list ARGUMENT MUST BE A NUMERIC MATRIX")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ ident.row.names <- TRUE
+ ident.col.names <- TRUE
+ for(i1 in 2:length(mat.list)){
+ tempo <- fun_comp_2d(data1 = mat.list[[1]], data2 = mat.list[[i1]])
+ if(tempo$same.dim == FALSE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX ", i1, " OF mat.list ARGUMENT MUST HAVE THE SAME DIMENSION (", paste(dim(mat.list[[i1]]), collapse = " "), ") THAN THE MATRIX 1 IN mat.list (", paste(dim(mat.list[[1]]), collapse = " "), ")")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(tempo$same.row.name)){
+ if(tempo$same.row.name != TRUE){ # != TRUE to deal with NA
+ ident.row.names <- FALSE
+ }
+ }
+ if( ! is.null(tempo$same.col.name)){
+ if(tempo$same.col.name != TRUE){ # != TRUE to deal with NA
+ ident.col.names <- FALSE
+ }
+ }
+ }
+ # end argument checking without fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ output <- mat.list[[1]]
+ for(i1 in 2:length(mat.list)){
+ output <- get(kind.of.operation)(output, mat.list[[i1]]) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ }
+ dimnames(output) <- NULL
+ if(ident.row.names == TRUE){
+ rownames(output) <- rownames(mat.list[[1]])
+ }
+ if(ident.col.names == TRUE){
+ colnames(output) <- colnames(mat.list[[1]])
+ }
+ return(output)
}
-if(same.length == TRUE & ! all(is.null(same.compartment.pos1), is.null(same.compartment.pos2))){
-if(identical(same.compartment.pos1, same.compartment.pos2)){
-identical.content <- TRUE
-}else{
-identical.content <- FALSE
+
+
+######## fun_mat_inv() #### return the inverse of a square matrix
+
+
+fun_mat_inv <- function(mat){
+ # AIM
+ # return the inverse of a square matrix when solve() cannot
+ # ARGUMENTS:
+ # mat: a square numeric matrix without NULL, NA, Inf or single case (dimension 1, 1) of 0
+ # RETURN
+ # the inversed matrix
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # mat1 = matrix(c(1,1,1,2,1,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1) # use solve()
+ # mat1 = matrix(c(0,0,0,0,0,0,0,0,0), ncol = 3) ; fun_mat_inv(mat = mat1) # use the trick
+ # mat1 = matrix(c(1,1,1,2,Inf,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1)
+ # mat1 = matrix(c(1,1,1,2,NA,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1)
+ # mat1 = matrix(c(1,2), ncol = 1) ; fun_mat_inv(mat = mat1)
+ # mat1 = matrix(0, ncol = 1) ; fun_mat_inv(mat = mat1)
+ # mat1 = matrix(2, ncol = 1) ; fun_mat_inv(mat = mat1)
+ # DEBUGGING
+ # mat = matrix(c(1,1,1,2,1,5,9,8,9), ncol = 3) # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = mat, class = "matrix", mode = "numeric", fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # argument checking without fun_check()
+ if(ncol(mat) != nrow(mat)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A SQUARE MATRIX")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(mat %in% c(Inf, -Inf, NA))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A MATRIX WITHOUT Inf, -Inf OR NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(all(mat == 0L) & ncol(mat) == 1L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT CANNOT BE A SQUARE MATRIX MADE OF A SINGLE CASE OF 0")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking without fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if(any(grepl(x = try(solve(mat), silent = TRUE)[], pattern = "[Ee]rror"))){
+ tempo <- svd(mat)
+ val.critique <- which(tempo$d < 10^-8)
+ Diag.mod <- diag(1 / tempo$d)
+ for(i in val.critique){
+ Diag.mod[i, i] <- 0
+ }
+ return(tempo$v %*% Diag.mod %*% t(tempo$u))
+ }else{
+ return(solve(mat))
+ }
}
-}else{
-identical.content <- FALSE
+
+
+######## fun_mat_fill() #### fill the empty half part of a symmetric square matrix
+
+
+fun_mat_fill <- function(mat, empty.cell.string = 0, warn.print = FALSE){
+ # AIM
+ # detect the empty half part of a symmetric square matrix (either topleft, topright, bottomleft or bottomright)
+ # fill this empty half part using the other symmetric half part of the matrix
+ # WARNINGS
+ # a plot verification using fun_gg_heatmap() is recommanded
+ # ARGUMENTS:
+ # mat: a numeric or character square matrix with the half part (according to the grand diagonal) filled with NA (any kind of matrix), "0" (character matrix) or 0 (numeric matrix) exclusively (not a mix of 0 and NA in the empty part)
+ # empty.cell.string: a numeric, character or NA (no quotes) indicating what empty cells are filled with
+ # warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
+ # RETURN
+ # a list containing:
+ # $mat: the filled matrix
+ # $warn: the warning messages. Use cat() for proper display. NULL if no warning
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # mat1 = matrix(c(1,NA,NA,NA, 0,2,NA,NA, NA,3,4,NA, 5,6,7,8), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = NA, warn.print = TRUE) # bottomleft example
+ # mat1 = matrix(c(1,1,1,2, 0,2,3,0, NA,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = NA, warn.print = TRUE) # error example
+ # mat1 = matrix(c(1,1,1,2, 0,2,3,0, NA,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # bottomright example
+ # mat1 = matrix(c(1,1,1,2, "a",2,3,NA, "a","a",0,0, "a","a","a",0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = "a", warn.print = TRUE) # topright example
+ # mat1 = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # topleft example
+ # mat1 = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # error example
+ # DEBUGGING
+ # mat = matrix(c(1,NA,NA,NA, 0,2,NA,NA, NA,3,4,NA, 5,6,7,8), ncol = 4) ; empty.cell.string = NA ; warn.print = TRUE # for function debugging
+ # mat = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; empty.cell.string = 0 ; warn.print = TRUE # for function debugging # topleft example
+ # mat = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; empty.cell.string = NA ; warn.print = TRUE # for function debugging # topleft example
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = mat, class = "matrix", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = empty.cell.string, class = "vector", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # argument checking without fun_check()
+ if(ncol(mat) != nrow(mat)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A SQUARE MATRIX")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (base::mode(mat) %in% c("numeric", "character"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A NUMERIC OR CHARACTER MATRIX")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(nrow(mat) == 1L & ncol(mat) == 1L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT CANNOT BE A SQUARE MATRIX MADE OF A SINGLE CASE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(ifelse(is.na(empty.cell.string), ! any(is.na(mat)), ! any(mat == empty.cell.string, na.rm = TRUE))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MATRIX MUST HAVE CELLS WITH THE EMPTY STRING SPECIFIED IN empty.cell.string ARGUMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking without fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ list.diag <- vector("list", length = nrow(mat) - 1)
+ for(i1 in 1:(nrow(mat) - 1)){
+ list.diag[[i1]] <- numeric(length = nrow(mat) - i1) # list made of zero
+ }
+ sector <- c("topleft", "topright", "bottomright", "bottomleft")
+ diag.scan <-c( # same order as sector. Recover each diag from center to corner
+ "mat[as.matrix(as.data.frame(list(1:(nrow(mat) - i2), (ncol(mat) -i2):1), stringsAsFactors = TRUE))]", # topleft part
+ "mat[as.matrix(as.data.frame(list(1:(nrow(mat) - i2), (1:ncol(mat))[-(1:i2)]), stringsAsFactors = TRUE))]", # topright part
+ "mat[as.matrix(as.data.frame(list((1 + i2):nrow(mat), ncol(mat):(1 + i2)), stringsAsFactors = TRUE))]", # bottomright part
+ "mat[as.matrix(as.data.frame(list((1 + i2):nrow(mat), 1:(ncol(mat) -i2)), stringsAsFactors = TRUE))]" # bottomleft part
+ )
+ # empty part detection
+ empty.sector <- NULL
+ full.sector <- NULL
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ for(i1 in 1:length(sector)){
+ tempo.list.diag <- list.diag
+ for(i2 in 1:(nrow(mat) - 1)){
+ tempo.list.diag[[i2]] <- eval(parse(text = diag.scan[i1]))
+ if(ifelse(is.na(empty.cell.string), ! all(is.na(tempo.list.diag[[i2]])), ! (all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = TRUE) & ! (is.na(all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE)))))){ # I had to add this ! (is.na(all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE))) because all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE) gives NA and not FALSE if one NA in tempo.list.diag[[i2]] -> not good for if()
+ full.sector <- c(full.sector, sector[i1])
+ break
+ }
+ }
+ if(i2 == nrow(mat) - 1){
+ if(all(unlist(lapply(tempo.list.diag, FUN = function(x){if(is.na(empty.cell.string)){is.na(x)}else{x == empty.cell.string}})), na.rm = TRUE)){
+ empty.sector <- c(empty.sector, sector[i1])
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") EMPTY SECTOR DETECTED ON THE ", toupper(sector[i1]), " CORNER, FULL OF ", empty.cell.string)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE ", toupper(sector[i1]), " SECTOR, DETECTED AS EMPTY, IS NOT? DIFFERENT VALUES IN THIS SECTOR:\n", paste(names(table(unlist(tempo.list.diag), useNA = "ifany")), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ }
+ # end empty part detection
+ if(length(empty.sector) == 0L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") ACCORDING TO empty.cell.string ARGUMENT (", empty.cell.string, "), mat ARGUMENT MATRIX HAS ZERO EMPTY HALF PART")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ if(length(empty.sector) > 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ACCORDING TO empty.cell.string ARGUMENT (", empty.cell.string, "), mat ARGUMENT MATRIX HAS MORE THAN ONE EMPTY HALF PART (ACCORDING TO THE GRAND DIAGONAL): ", paste(empty.sector, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(any(full.sector %in% empty.sector, na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE FUNCTION HAS DETECTED EMPTY AND NON EMPTY HALF PART IN THE SAME SECTOR: ", paste(full.sector[full.sector %in% empty.sector], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(length(empty.sector) + length(full.sector)!= 4){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE FUNCTION HAS DETECTED MORE OR LESS SECTORS THAN 4:\nHALF SECTORS:", paste(empty.sector, collapse = " "), "\nFULL SECTORS:", paste(full.sector, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") ", toupper(empty.sector), " SECTOR HAS BEEN COMPLETED TO BECOME SYMMETRICAL")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # matrix filling
+ for(i2 in 1:(nrow(mat) - 1)){
+ if(empty.sector == "topleft"){
+ eval(parse(text = paste0(diag.scan[1], " <- ", diag.scan[3])))
+ }else if(empty.sector == "topright"){
+ eval(parse(text = paste0(diag.scan[2], " <- ", diag.scan[4])))
+ }else if(empty.sector == "bottomright"){
+ eval(parse(text = paste0(diag.scan[3], " <- ", diag.scan[1])))
+ }else if(empty.sector == "bottomleft"){
+ eval(parse(text = paste0(diag.scan[4], " <- ", diag.scan[2])))
+ }
+ }
+ # end matrix filling
+ }
+ if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ return(list(mat = mat, warn = warn))
}
+
+
+######## fun_permut() #### progressively breaks a vector order
+
+
+fun_permut <- function(
+ data1,
+ data2 = NULL,
+ n = NULL,
+ seed = NULL,
+ print.count = 10,
+ text.print = "",
+ cor.method = "spearman",
+ cor.limit = 0.2,
+ warn.print = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # reorder the elements of the data1 vector by flipping 2 randomly selected consecutive positions either:
+ # 1) n times (when n is precised) or
+ # 2) until the correlation between data1 and data2 decreases down to the cor.limit (0.2 by default). See cor.limit below to deal with negative correlations
+ # Example of consecutive position flipping: ABCD -> BACD -> BADC, etc.
+ # designed for discrete values, but worls also for continuous values
+ # WARNINGS
+ # see # https://www.r-bloggers.com/strategies-to-speedup-r-code/ for code speedup
+ # the random switch of non consecutive positions (ABCD -> DBCA for instance) does not work very well as the correlation is quickly obtained but the initial vector structure is mainly kept (no much order). Ths code would be: pos <- ini.pos[1:2] ; pos <- sample.int(n = n , size = 2, replace = FALSE) ; tempo.pos[pos] <- tempo.pos[rev(pos)]
+ # ARGUMENTS
+ # data1: a vector of at least 2 elements. Must be numeric if data2 is specified
+ # data2: a numeric vector of same length as data1
+ # n: number of times "flipping 2 randomly selected consecutive positions". Ignored if data2 is specified
+ # seed: integer number used by set.seed(). Write NULL if random result is required, an integer otherwise. BEWARE: if not NULL, fun_permut() will systematically return the same result when the other parameters keep the same settings
+ # print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
+ # text.print: optional message to add to the working progress message every print.count loop
+ # cor.method: correlation method. Either "pearson", "kendall" or "spearman". Ignored if data2 is not specified
+ # cor.limit: a correlation limit (between 0 and 1). Ignored if data2 is not specified. Compute the correlation between data1 and data2, permute the data1 values, and stop the permutation process when the correlation between data1 and data2 decreases down below the cor limit value (0.2 by default). If cor(data1, data2) is negative, then -cor.limit is used and the process stops until the correlation between data1 and data2 increases up over cor.limit (-0.2 by default). BEWARE: write a positive cor.limit even if cor(data1, data2) is known to be negative. The function will automatically uses -cor.limit. If the initial correlation is already below cor.limit (positive correlation) or over -cor.limit (negative correlation), then the data1 value positions are completely randomized (correlation between data1 and data2 is expected to be 0)
+ # warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # RETURN
+ # a list containing:
+ # $data: the modified vector
+ # $warn: potential warning messages (in case of negative correlation when data2 is specified). NULL if non warning message
+ # $cor: a spearman correlation between the initial positions (1:length(data1) and the final positions if data2 is not specified and the final correlation between data1 and data2 otherwise, according to cor.method
+ # $count: the number of loops used
+ # REQUIRED PACKAGES
+ # lubridate
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # fun_round()
+ # EXAMPLES
+ # example (1) showing that for loop, used in fun_permut(), is faster than while loop
+ # ini.time <- as.numeric(Sys.time()) ; count <- 0 ; for(i0 in 1:1e9){count <- count + 1} ; tempo.time <- as.numeric(Sys.time()) ; tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time)) ; tempo.lapse
+ # example (2) showing that for loop, used in fun_permut(), is faster than while loop
+ # ini.time <- as.numeric(Sys.time()) ; count <- 0 ; while(count < 1e9){count <- count + 1} ; tempo.time <- as.numeric(Sys.time()) ; tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time)) ; tempo.lapse
+ # fun_permut(data1 = LETTERS[1:5], data2 = NULL, n = 100, seed = 1, print.count = 10, text.print = "CPU NB 4")
+ # fun_permut(data1 = 101:110, data2 = 21:30, seed = 1, print.count = 1e4, text.print = "", cor.method = "spearman", cor.limit = 0.2)
+ # a way to use the cor.limit argument just considering data1
+ # obs1 <- 101:110 ; fun_permut(data1 = obs1, data2 = obs1, seed = 1, print.count = 10, cor.method = "spearman", cor.limit = 0.2)
+ # fun_permut(data1 = 1:1e3, data2 = 1e3:1, seed = 1, print.count = 1e6, text.print = "", cor.method = "spearman", cor.limit = 0.7)
+ # fun_permut(data1 = 1:1e2, data2 = 1e2:1, seed = 1, print.count = 1e3, cor.limit = 0.5)
+ # fun_permut(data1 = c(0,0,0,0,0), n = 5, data2 = NULL, seed = 1, print.count = 1e3, cor.limit = 0.5)
+ # DEBUGGING
+ # data1 = LETTERS[1:5] ; data2 = NULL ; n = 1e6 ; seed = NULL ; print.count = 1e3 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
+ # data1 = LETTERS[1:5] ; data2 = NULL ; n = 10 ; seed = 22 ; print.count = 10 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
+ # data1 = 101:110 ; data2 = 21:30 ; n = 10 ; seed = 22 ; print.count = 10 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
+ # data1 = 1:1e3 ; data2 = 1e3:1 ; n = 20 ; seed = 22 ; print.count = 1e6 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.5 ; warn.print = TRUE ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_round", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data1, class = "vector", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & length(data1) < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A VECTOR OF MINIMUM LENGTH 2. HERE IT IS: ", length(data1))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(data2)){
+ tempo <- fun_check(data = data1, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 MUST BE A NUMERIC VECTOR IF data2 ARGUMENT IS SPECIFIED")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = data2, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
+ if(length(data1) != length(data2)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 AND data2 MUST BE VECTOR OF SAME LENGTH. HERE IT IS ", length(data1)," AND ", length(data2))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else if(is.null(n)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT CANNOT BE NULL IF data2 ARGUMENT IS NULL")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(n)){
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(seed)){
+ tempo <- fun_check(data = seed, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = text.print, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = cor.method, options = c("pearson", "kendall", "spearman"), length =1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = cor.limit, class = "vector", mode = "numeric", prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # package checking
+ fun_pack(req.package = "lubridate", lib.path = lib.path)
+ # end package checking
+ # main code
+ # code that protects set.seed() in the global environment
+ # see also Protocol 100-rev0 Parallelization in R.docx
+ if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
+ tempo.random.seed <- .Random.seed
+ on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
+ }else{
+ on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
+ }
+ set.seed(seed)
+ # end code that protects set.seed() in the global environment
+ ini.date <- Sys.time() # time of process begin, converted into seconds
+ ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
+ ini.pos <- 1:length(data1) # positions of data1 before permutation loops
+ tempo.pos <- ini.pos # positions of data1 that will be modified during loops
+ # pos.selec.seq <- ini.pos[-length(data1)] # selection of 1 position in initial position, without the last because always up permutation (pos -> pos+1 & pos+1 -> pos)
+ pos.selec.seq.max <- length(ini.pos) - 1 # max position (used by sample.int() function). See below for - 1
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ count <- 0
+ round <- 0
+ BREAK <- FALSE
+ tempo.cor <- 0
+ if(is.null(data2)){
+ if(length(table(data1)) == 1L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data1 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data1)))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
+ }else{
+ if(print.count > n){
+ print.count <- n
+ }
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP OF ", n, " LOOPS INITIATED | LOOP COUNT: ", format(count, big.mark=",")))
+ print.count.loop <- logical(length = print.count)
+ print.count.loop[length(print.count.loop)] <- TRUE # not this to avoid long vector, but not forget to reset during printing: print.count.loop[(1:trunc(n / print.count) * print.count)] <- TRUE # counter to speedup
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.date.loop <- Sys.time()
+ tempo.time.loop <- as.numeric(tempo.date.loop)
+ for(i3 in 1:n){
+ count.loop <- count.loop + 1
+ pos2 <- pos[count.loop] # selection of 1 position
+ tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ if(print.count.loop[count.loop]){
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
+ final.loop <- (tempo.time - tempo.time.loop) / i3 * n # expected duration in seconds
+ final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP ", i3, " / ", n, " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
+ }
+ }
+ count <- count + n # out of the loop to speedup
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP ENDED | LOOP COUNT: ", format(count, big.mark=",")))
+ cat("\n\n")
+ }
+ }else{
+ if(length(table(data1)) == 1L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data1 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data1)))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
+ tempo.cor <- 1
+ }else if(length(table(data2)) == 1L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data2 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data2)))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
+ tempo.cor <- 1
+ }else{
+ cor.ini <- cor(x = data1, y = data2, use = "pairwise.complete.obs", method = cor.method)
+ tempo.cor <- cor.ini # correlation that will be modified during loops
+ neg.cor <- FALSE
+ if(tempo.cor < 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") INITIAL ", toupper(cor.method), " CORRELATION BETWEEN data1 AND data2 HAS BEEN DETECTED AS NEGATIVE: ", tempo.cor, ". THE LOOP STEPS WILL BE PERFORMED USING POSITIVE CORRELATIONS BUT THE FINAL CORRELATION WILL BE NEGATIVE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
+ neg.cor <- TRUE
+ tempo.cor <- abs(tempo.cor)
+ cor.ini <- abs(cor.ini)
+ }
+ if(tempo.cor < cor.limit){ # randomize directly all the position to be close to correlation zero
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") INITIAL ABSOLUTE VALUE OF THE ", toupper(cor.method), " CORRELATION ", fun_round(tempo.cor), " BETWEEN data1 AND data2 HAS BEEN DETECTED AS BELOW THE CORRELATION LIMIT PARAMETER ", cor.limit, "\nTHE data1 SEQUENCE HAS BEEN COMPLETELY RANDOMIZED TO CORRESPOND TO CORRELATION ZERO")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
+ for(i4 in 1:5){ # done 5 times to be sure of the complete randomness
+ tempo.pos <- sample(x = tempo.pos, size = length(tempo.pos), replace = FALSE)
+ }
+ count <- count + 5 # out of the loop to speedup
+ }else{
+ # smallest correlation decrease
+ count <- count + 1 # 1 and not 0 because already 1 performed just below
+ pos <- sample.int(n = pos.selec.seq.max , size = 1, replace = TRUE) # selection of 1 position # pos.selec.seq.max because selection of 1 position in initial position, without the last because always up permutation (pos -> pos+1 & pos+1 -> pos)
+ tempo.pos[c(pos + 1, pos)] <- tempo.pos[c(pos, pos + 1)]
+ tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
+ smallest.cor.dec <- cor.ini - tempo.cor
+ # end smallest correlation decrease
+ # going out of tempo.cor == cor.ini
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "CORRELATION DECREASE AFTER A SINGLE PERMUTATION: ", fun_round(smallest.cor.dec, 4)))
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP -> GOING OUT FROM EQUALITY | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
+ print.count.loop <- logical(length = print.count)
+ print.count.loop[length(print.count.loop)] <- TRUE # counter to speedup
+ count.loop <- 0 #
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.date.loop <- Sys.time()
+ tempo.time.loop <- as.numeric(tempo.date.loop)
+ while(tempo.cor == cor.ini){ # to be out of equality between tempo.cor and cor.ini at the beginning (only valid for very long vector)
+ count <- count + 1
+ count.loop <- count.loop + 1
+ pos2 <- pos[count.loop]
+ tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
+ if(print.count.loop[count.loop]){
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP", format(count.loop, big.mark=","), " / ? | COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TIME SPENT: ", tempo.lapse))
+ }
+ }
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP END | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TOTAL SPENT TIME: ", tempo.lapse))
+ if(tempo.cor < cor.limit){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE FIRST FOR & WHILE LOOP STEPS HAVE BEEN TOO FAR AND SUBSEQUENT LOOP STEPS WILL NOT RUN")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end going out of tempo.cor == cor.ini
+ # estimation of the average correlation decrease per loop on x loops and for loop execution
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE/FOR LOOPS INITIATION | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
+ count.est <- 1e5
+ first.round <- TRUE
+ GOBACK <- FALSE
+ while(tempo.cor > cor.limit){
+ round <- round + 1
+ # estimation step
+ if(first.round == TRUE){
+ first.round <- FALSE
+ cor.dec.per.loop <- numeric(length = 5)
+ loop.nb.est <- Inf
+ cor.est.ini <- tempo.cor
+ cor.est <- numeric(length = 5)
+ for(i6 in 1:5){ # connected to cor.dec.per.loop
+ tempo.pos.est <- tempo.pos
+ pos <- sample.int(n = pos.selec.seq.max , size = count.est, replace = TRUE) # selection of n position
+ for(i7 in 1:count.est){
+ pos2 <- pos[i7] # selection of 1 position
+ tempo.pos.est[c(pos2 + 1, pos2)] <- tempo.pos.est[c(pos2, pos2 + 1)]
+ }
+ tempo.cor.est <- abs(cor(x = data1[tempo.pos.est], y = data2, use = "pairwise.complete.obs", method = cor.method))
+ cor.est[i6] <- tempo.cor.est
+ tempo.cor.dec.per.loop <- (cor.est.ini - tempo.cor.est) / count.est # correlation decrease per loop
+ if(is.na(tempo.cor.dec.per.loop) | ! is.finite(tempo.cor.dec.per.loop)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2\ncor.est.ini: ", cor.est.ini, "\ntempo.cor.est: ", tempo.cor.est)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ cor.dec.per.loop[i6] <- tempo.cor.dec.per.loop
+ }
+ cor.est <- cor.est[which.max(cor.dec.per.loop)] # max to avoid to go to far with for loop (tempo.cor below tempo.limit)
+ cor.dec.per.loop <- max(cor.dec.per.loop, na.rm = TRUE) # max to avoid to go to far with for loop (tempo.cor below tempo.limit)
+ loop.nb.est <- round((tempo.cor - cor.limit) / cor.dec.per.loop)
+ }else{
+ if(GOBACK == TRUE){
+ loop.nb.est <- round(loop.nb.est / 2)
+ }else{
+ cor.dec.per.loop <- (cor.ini - tempo.cor) / count
+ loop.nb.est <- round((tempo.cor - cor.limit) / cor.dec.per.loop)
+ }
+ }
+ # end estimation step
+ # loop step
+ if(is.na(loop.nb.est) | ! is.finite(loop.nb.est)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1\nloop.nb.est: ", loop.nb.est, "\ncor.ini: ", cor.ini, "\ntempo.cor: ", tempo.cor, "\ncor.limit: ", cor.limit, "\ncor.dec.per.loop: ", cor.dec.per.loop)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(loop.nb.est > 1e4){ # below -> leave the while loop
+ tempo.pos.secu <- tempo.pos
+ count.secu <- count
+ tempo.cor.secu <- tempo.cor
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "INITIAL SETTINGS BEFORE ROUND: ", round, " | LOOP COUNT: ", format(count, big.mark=","), " | GO BACK: ", GOBACK, " | LOOP NUMBER ESTIMATION: ", format(loop.nb.est, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
+ print.count.loop <- logical(length = print.count)
+ print.count.loop[length(print.count.loop)] <- TRUE # not this to avoid long vector, but not forget to reset during printing: print.count.loop[(1:trunc(n / print.count) * print.count)] <- TRUE # counter to speedup
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.date.loop <- Sys.time()
+ tempo.time.loop <- as.numeric(tempo.date.loop)
+ for(i6 in 1:loop.nb.est){
+ count.loop <- count.loop + 1
+ pos2 <- pos[count.loop] # selection of 1 position
+ tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ if(print.count.loop[count.loop]){
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
+ final.loop <- (tempo.time - tempo.time.loop) / i6 * loop.nb.est # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
+ final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP | ROUND ", round, " | LOOP: ", format(i6, big.mark=","), " / ", format(loop.nb.est, big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
+ }
+ }
+ count <- count + loop.nb.est # out of the loop to speedup
+ tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
+ if(tempo.cor > tempo.cor.secu | ((tempo.cor - cor.limit) < 0 & abs(tempo.cor - cor.limit) > smallest.cor.dec * round(log10(max(ini.pos, na.rm = TRUE))))){
+ GOBACK <- TRUE
+ tempo.pos <- tempo.pos.secu
+ count <- count.secu
+ tempo.cor <- tempo.cor.secu
+ }else{
+ GOBACK <- FALSE
+ }
+ }else{
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FINAL WHILE LOOP | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
+ print.count.loop <- logical(length = print.count)
+ print.count.loop[length(print.count.loop)] <- TRUE # counter to speedup
+ count.loop <- 0 #
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.cor.loop <- tempo.cor
+ tempo.date.loop <- Sys.time()
+ tempo.time.loop <- as.numeric(tempo.date.loop)
+ while(tempo.cor > cor.limit){
+ count <- count + 1
+ count.loop <- count.loop + 1
+ pos2 <- pos[count.loop]
+ tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
+ tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
+ if(print.count.loop[count.loop]){
+ count.loop <- 0
+ pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
+ final.loop <- (tempo.time - tempo.time.loop) / (tempo.cor.loop - tempo.cor) * (tempo.cor - cor.limit) # expected duration in seconds # tempo.cor.loop - tempo.cor always positive and tempo.cor decreases progressively starting from tempo.cor.loop
+ final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE LOOP | LOOP NB: ", format(count.loop, big.mark=","), " | COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
+ }
+ }
+ }
+ }
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+ cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE/FOR LOOPS END | LOOP COUNT: ", format(count, big.mark=","), " | NB OF ROUNDS: ", round, " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TOTAL SPENT TIME: ", tempo.lapse))
+ }
+ tempo.cor <- ifelse(neg.cor == TRUE, -tempo.cor, tempo.cor)
+ }
+ }
+ cat("\n\n")
+ if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE), add = TRUE)
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ output <- list(data = data1[tempo.pos], warn = warn, cor = if(is.null(data2)){cor(ini.pos, tempo.pos, method = "spearman")}else{tempo.cor}, count = count)
+ return(output)
}
-output <- list(same.length = same.length, length = length, same.names = same.names, name = name, any.id.name = any.id.name, same.names.pos1 = same.names.pos1, same.names.pos2 = same.names.pos2, any.id.compartment = any.id.compartment, same.compartment.pos1 = same.compartment.pos1, same.compartment.pos2 = same.compartment.pos2, identical.object = identical.object, identical.content = identical.content)
-return(output)
+
+
+######## fun_slide() #### return a computation made on a vector using a sliding window
+
+
+fun_slide <- function(
+ data,
+ window.size,
+ step,
+ from = NULL,
+ to = NULL,
+ fun,
+ args = NULL,
+ boundary = "left",
+ parall = FALSE,
+ thread.nb = NULL,
+ print.count = 100,
+ res.path = NULL,
+ lib.path = NULL,
+ verbose = TRUE,
+ cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
+){
+ # AIM
+ # return a computation made on a vector using a sliding window
+ # WARNINGS
+ # The function uses two strategies, depending on the amout of memory required which depends on the data, window.size and step arguments. The first one uses lapply(), is generally fast but requires lots of memory. The second one uses a parallelized loop. The choice between the two strategies is automatic if parall argument is FALSE, and is forced toward parallelization if parall argument is TRUE
+ # The parall argument forces the parallelization, which is convenient when the data argument is big, because the lapply() function is sometimes slower than the parallelization
+ # Always use the env argument when fun_slide() is used inside functions
+ # ARGUMENTS
+ # data: vector, matrix, table or array of numeric values (mode must be numeric). Inf not allowed. NA will be removed before computation
+ # window.size: single numeric value indicating the width of the window sliding across data (in the same unit as data value)
+ # step: single numeric value indicating the step between each window (in the same unit as data value). Cannot be larger than window.size
+ # from: value of the left boundary of the first sliding window. If NULL, min(data) is used. The first window will strictly have from or min(data) as left boundary
+ # to: value of the right boundary of the last sliding window. If NULL, max(data) is used. Warning: (1) the final last window will not necessary have to|max(data) as right boundary. In fact the last window will be the one that contains to|max(data) for the first time, i.e., min[from|min(data) + window.size + n * step >= to|max(data)]; (2) In fact, the >= in min[from|min(data) + window.size + n * step >= to|max(data)] depends on the boundary argument (>= for "right" and > for "left"); (3) to have the rule (1) but for the center of the last window, use to argument as to = to|max(data) + window.size / 2
+ # fun: function or character string (without brackets) indicating the name of the function to apply in each window. Example: fun = "mean", or fun = mean
+ # args: character string of additional arguments of fun (separated by a comma between the quotes). Example args = "na.rm = TRUE" for fun = mean. Ignored if NULL
+ # boundary: either "left" or "right". Indicates if the sliding window includes values equal to left boundary and exclude values equal to right boundary ("left") or the opposite ("right")
+ # parall: logical. Force parallelization ?
+ # thread.nb: numeric value indicating the number of threads to use if ever parallelization is required. If NULL, all the available threads will be used. Ignored if parall is FALSE
+ # print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
+ # res.path: character string indicating the absolute pathway where the parallelization log file will be created if parallelization is used. If NULL, will be created in the R current directory
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # verbose: logical. Display messages?
+ # cute.path: character string indicating the absolute path of the cute.R file. Will be remove when cute will be a package. Ignored if parall is FALSE
+ # RETURN
+ # a data frame containing
+ #$left : the left boundary of each window (in the unit of the data argument)
+ #$right : the right boundary of each window (in the unit of data argument)
+ #$center : the center of each window (in the unit of data argument)
+ #$value : the computed value by the fun argument in each window)
+ # REQUIRED PACKAGES
+ # lubridate
+ # parallel if parall argument is TRUE (included in the R installation packages but not automatically loaded)
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_get_message
+ # fun_pack()
+ # EXAMPLES
+ # fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "left")
+ # fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "right") # effect of boundary argument
+ # fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "left", parall = TRUE) # effect of parall argument
+ # DEBUGGING
+ # data = c(1:10, 100:110, 500) ; window.size = 5 ; step = 2 ; from = NULL ; to = NULL ; fun = length ; args = NULL ; boundary = "left" ; parall = FALSE ; thread.nb = NULL ; print.count = 100 ; res.path = NULL ; lib.path = NULL ; verbose = TRUE ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
+ # data = lag.pos; window.size = window.size; step = step; fun = length; from = min(a$pos); to = max(a$pos)
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check",
+ "fun_get_message",
+ "fun_pack"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # arg with no default values
+ mandat.args <- c(
+ "data",
+ "window.size",
+ "step",
+ "fun"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data, mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = window.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = step, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(from)){
+ tempo <- fun_check(data = from, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(to)){
+ tempo <- fun_check(data = to, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo1 <- fun_check(data = fun, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = fun, class = "function", length = 1, fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": fun ARGUMENT MUST BE A FUNCTION OR A CHARACTER STRING OF THE NAME OF A FUNCTION")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(args)){
+ tempo <- fun_check(data = args, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = boundary, options = c("left", "right"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = parall, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(parall == TRUE){
+ if( ! is.null(thread.nb)){
+ tempo <- fun_check(data = thread.nb, typeof = "integer", double.as.integer.allowed = TRUE, neg.values = FALSE, length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & thread.nb < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": thread.nb PARAMETER MUST EQUAL OR GREATER THAN 1: ", thread.nb)
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(res.path)){
+ tempo <- fun_check(data = res.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = verbose, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = cute.path, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # new environment
+ env.name <- paste0("env", as.numeric(Sys.time()))
+ if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ assign(env.name, new.env())
+ }
+ # end new environment
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "data",
+ "window.size",
+ "step",
+ # "from", # because can be NULL
+ # "to", # because can be NULL
+ "fun",
+ # "args", # because can be NULL
+ "boundary",
+ "parall",
+ # "thread.nb", # because can be NULL
+ "print.count",
+ # "res.path", # because can be NULL
+ # "lib.path", # because can be NULL
+ "verbose",
+ "cute.path"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ # end warning initiation
+ # other checkings
+ if(length(data) == 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT CANNOT BE LENGTH 0")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(any( ! is.finite(data))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT CANNOT CONTAIN Inf VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if(step > window.size){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": step ARGUMENT MUST BE LOWER THAN window.size ARGUMENT\nstep: ", paste(step, collapse = " "), "\nwindow.size: ", paste(window.size, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ if( ! is.null(res.path)){
+ if( ! all(dir.exists(res.path))){ # separation to avoid the problem of tempo$problem == FALSE and res.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE res.path ARGUMENT DOES NOT EXISTS:\n", paste(res.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ }else{
+ res.path <- getwd() # working directory
+ }
+ if( ! is.null(lib.path)){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
+ }
+ }
+ if(grepl(x = cute.path, pattern = "^http")){
+ tempo.error1 <- any(grepl(x = fun_get_message(data = "source(cute.path)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE)), pattern = "^[Ee]rror"))
+ tempo.error2 <- FALSE
+ }else{
+ tempo.error1 <- FALSE
+ tempo.error2 <- ! file.exists(cute.path)
+ }
+ if(tempo.error1 | tempo.error2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(grepl(x = cute.path, pattern = "^http"), "URL", "FILE"), " PATH INDICATED IN THE cute.path PARAMETER DOES NOT EXISTS:\n", cute.path)
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ # end other checkings
+ # reserved word checking
+ # end reserved word checking
+ # end second round of checking and data preparation
+ # package checking
+ fun_pack(req.package = c("lubridate"), lib.path = lib.path)
+ fun_pack(req.package = c("parallel"), lib.path = lib.path)
+ # end package checking
+ # main code
+ if(verbose == TRUE){
+ cat("\nfun_slide JOB IGNITION\n")
+ }
+ ini.date <- Sys.time()
+ ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
+ fun <- match.fun(fun) # make fun <- get(fun) is fun is a function name written as character string of length 1
+ if(boundary == "left"){
+ left <- ">="
+ right <- "<"
+ right.last.wind <- ">"
+ }else if(boundary == "right"){
+ left <- ">"
+ right <- "<="
+ right.last.wind <- ">="
+ }else{
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ data <- as.vector(data)
+ data <- sort(data, na.last = NA) # NA removed
+ wind <- data.frame(left = seq(from = if(is.null(from)){min(data, na.rm = TRUE)}else{from}, to = if(is.null(to)){max(data, na.rm = TRUE)}else{to}, by = step), stringsAsFactors = TRUE)
+ wind <- data.frame(wind, right = wind$left + window.size, stringsAsFactors = TRUE)
+ wind <- data.frame(wind, center = (wind$left + wind$right) / 2, stringsAsFactors = TRUE)
+ if(all(wind$right < if(is.null(to)){max(data, na.rm = TRUE)}else{to})){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 2")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # The 3 next lines is for the rule of to argument with center (see to argument description)
+ # if(any(wind$center > max(data, na.rm = TRUE))){
+ # wind <- wind[ ! wind$center > max(data, na.rm = TRUE),]
+ # }
+ if(sum(get(right.last.wind)(wind$right, if(is.null(to)){max(data, na.rm = TRUE)}else{to}), na.rm = TRUE) > 1){ # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ tempo.log <- get(right.last.wind)(wind$right, if(is.null(to)){max(data, na.rm = TRUE)}else{to}) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ tempo.log[min(which(tempo.log), na.rm = TRUE)] <- FALSE # convert the first left boundary that goes above max(data, na.rm = TRUE) to FALSE to keep it (the next ones will be removed)
+ wind <- wind[ ! tempo.log,]
+ }
+
+ # test if lapply can be used
+ if(parall == FALSE){
+ assign("wind", wind, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # wind assigned in a new envir for test
+ assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
+ tempo.message <- fun_get_message(data="lapply(X = wind$left, Y = data, FUN = function(X, Y){res <- get(left)(Y, X) ; return(res)})", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE), print.no = FALSE) # no env = sys.nframe(), inherit = FALSE in get(left) because look for function in the classical scope
+ # rm(env.name) # optional, because should disappear at the end of the function execution
+ }else{
+ tempo.message <- "ERROR" # with this, force the parallelization by default
+ }
+ # end test if lapply can be used
+ if( ! any(grepl(x = tempo.message, pattern = "ERROR.*"))){
+ left.log <- lapply(X = wind$left, Y = data, FUN = function(X, Y){
+ res <- get(left)(Y, X) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ return(res)
+ })
+ right.log <- lapply(X = wind$right, Y = data, FUN = function(X, Y){
+ res <- get(right)(Y, X) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ return(res)
+ })
+ log <- mapply(FUN = "&", left.log, right.log, SIMPLIFY = FALSE)
+ output <- eval(parse(text = paste0("sapply(lapply(log, FUN = function(X){(data[X])}), FUN = fun", if( ! is.null(args)){paste0(", ", args)}, ")"))) # take the values of the data vector according to log (list of logical, each compartment of length(data)) and apply fun with args of fun
+ if(length(output) != nrow(wind)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ output <- data.frame(wind, value = output, stringsAsFactors = TRUE)
+ }
+ }else{
+ if(verbose == TRUE){
+ tempo.cat <- paste0("PARALLELIZATION INITIATED AT: ", ini.date)
+ cat(paste0("\n", tempo.cat, "\n"))
+ }
+ tempo.thread.nb = parallel::detectCores(all.tests = FALSE, logical = TRUE) # detect the number of threads
+ if( ! is.null(thread.nb)){
+ if(tempo.thread.nb < thread.nb){
+ thread.nb <- tempo.thread.nb
+ if(verbose == TRUE){
+ tempo.cat <- paste0("ONLY: ", tempo.thread.nb, " THREADS AVAILABLE")
+ cat(paste0("\n", tempo.cat, "\n"))
+ }
+ }
+ }else{
+ thread.nb <- tempo.thread.nb
+ }
+ if(verbose == TRUE){
+ tempo.cat <- paste0("NUMBER OF THREADS USED: ", thread.nb)
+ cat(paste0("\n ", tempo.cat, "\n"))
+ }
+ Clust <- parallel::makeCluster(thread.nb, outfile = paste0(res.path, "/fun_slide_parall_log.txt")) # outfile to print or cat during parallelization (only possible in a file, outfile = "" do not work on windows)
+ cluster.list <- parallel::clusterSplit(Clust, 1:nrow(wind)) # split according to the number of cluster
+ if(verbose == TRUE){
+ tempo.cat <- paste0("SPLIT OF TEST NUMBERS IN PARALLELISATION:")
+ cat(paste0("\n ", tempo.cat, "\n"))
+ str(cluster.list) # using print(str()) add a NULL below the result
+ cat("\n")
+ }
+ paral.output.list <- parallel::clusterApply( #
+ cl = Clust,
+ x = cluster.list,
+ function.name = function.name,
+ data = data,
+ FUN = fun, # because fun argument of clusterApply
+ args = args,
+ thread.nb = thread.nb,
+ print.count = print.count,
+ wind = wind,
+ left = left,
+ right = right,
+ res.path = res.path,
+ lib.path = lib.path,
+ verbose = verbose,
+ cute.path = cute.path,
+ fun = function(
+ x,
+ function.name,
+ data,
+ FUN,
+ args,
+ thread.nb,
+ print.count,
+ wind,
+ left,
+ right,
+ res.path,
+ lib.path,
+ verbose,
+ cute.path
+ ){
+ # check again: very important because another R
+ process.id <- Sys.getpid()
+ cat(paste0("\nPROCESS ID ", process.id, " -> TESTS ", x[1], " TO ", x[length(x)], "\n"))
+ source(cute.path, local = .GlobalEnv)
+ fun_pack(req.package = "lubridate", lib.path = lib.path, load = TRUE) # load = TRUE to be sure that functions are present in the environment. And this prevent to use R.lib.path argument of fun_python_pack()
+ # end check again: very important because another R
+ ini.date <- Sys.time()
+ ini.time <- as.numeric(ini.date) # time of process begin, converted into
+ output <- NULL
+ print.count.loop <- 0
+ for(i4 in 1:length(x)){
+ print.count.loop <- print.count.loop + 1
+ log <- get(left)(data, wind$left[x[i4]]) & get(right)(data, wind$right[x[i4]]) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ output <- c(output, eval(parse(text = paste0("FUN(data[log]", if( ! is.null(args)){paste0(", ", args)}, ")"))))
+ if(verbose == TRUE){
+ if(print.count.loop == print.count){
+ print.count.loop <- 0
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+ final.loop <- (tempo.time - ini.time) / i4 * length(x) # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
+ final.exp <- as.POSIXct(final.loop, origin = ini.date)
+ cat(paste0("\nIN PROCESS ", process.id, " | LOOP ", format(i4, big.mark=","), " / ", format(length(x), big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
+ }
+ if(i4 == length(x)){
+ tempo.time <- as.numeric(Sys.time())
+ tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
+ cat(paste0("\nPROCESS ", process.id, " ENDED | LOOP ", format(i4, big.mark=","), " / ", format(length(x), big.mark=","), " | TIME SPENT: ", tempo.lapse, "\n\n"))
+ }
+ }
+ }
+ wind <- wind[x, ]
+ if(length(output) != nrow(wind)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ output <- data.frame(wind, value = output, stringsAsFactors = TRUE)
+ return(output)
+ }
+ }
+ )
+ parallel::stopCluster(Clust)
+ # result assembly
+ output <- data.frame()
+ for(i2 in 1:length(paral.output.list)){ # compartment relatives to each parallelization
+ output <- rbind(output, paral.output.list[[i2]], stringsAsFactors = TRUE)
+ }
+ # end result assembly
+ if(nrow(output) != nrow(wind)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 5\nlength(output): ", length(output), "\nnrow(wind): ", nrow(wind))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ output <- output[order(output$left), ]
+ }
+ }
+ if(verbose == TRUE){
+ end.date <- Sys.time()
+ end.time <- as.numeric(end.date)
+ total.lapse <- round(lubridate::seconds_to_period(end.time - ini.time))
+ cat(paste0("\nfun_slide JOB END\n\nTIME: ", end.date, "\n\nTOTAL TIME LAPSE: ", total.lapse, "\n\n\n"))
+ }
+ return(output)
}
-######## fun_test() #### test combinations of argument values of a function and return errors (and graphs)
-# add traceback https://stackoverflow.com/questions/47414119/how-to-read-a-traceback-in-r
+######## fun_codon2aa() #### convert codon to amino acid using standard genetic code
-fun_test <- function(
-fun,
-arg,
-val,
-expect.error = NULL,
-parall = FALSE,
-thread.nb = NULL,
-print.count = 10,
-plot.fun = FALSE,
-export = FALSE,
-res.path = NULL,
-lib.path = NULL,
-cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
+
+fun_codon2aa <- function(
+ data,
+ display = FALSE
){
-# AIM
-# test combinations of argument values of a function
-# WARNINGS
-# Limited to 43 arguments with at least 2 values each. The total number of arguments tested can be more if the additional arguments have a single value. The limit is due to nested "for" loops (https://stat.ethz.ch/pipermail/r-help/2008-March/157341.html), but it should not be a problem since the number of tests would be 2^43 > 8e12
-# ARGUMENTS
-# fun: character string indicating the name of the function tested (without brackets)
-# arg: vector of character strings of arguments of fun. At least arguments that do not have default values must be present in this vector
-# val: list with number of compartments equal to length of arg, each compartment containing values of the corresponding argument in arg. Each different value must be in a list or in a vector. For instance, argument 3 in arg is a logical argument (values accepted TRUE, FALSE, NA). Thus, compartment 3 of val can be either list(TRUE, FALSE, NA), or c(TRUE, FALSE, NA). NULL value alone must be written list(NULL)
-# expect.error: list of exactly the same structure as val argument, but containing FALSE or TRUE, depending on whether error is expected (TRUE) or not (FALSE) for each corresponding value of val. A message is returned depending on discrepancies between the expected and observed errors. BEWARE: not always possible to write the expected errors for all the combination of argument values. Ignored if NULL
-# parall: logical. Force parallelization ?
-# thread.nb: numeric value indicating the number of threads to use if ever parallelization is required. If NULL, all the available threads will be used. Ignored if parall is FALSE
-# print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
-# plot.fun: logical. Plot the plotting function tested for each test?
-# export: logical. Export the results into a .RData file and into a .txt file? If FALSE, return a list into the console (see below). BEWARE: will be automatically set to TRUE if parall is TRUE. This means that when using parallelization, the results are systematically exported, not returned into the console
-# res.path: character string indicating the absolute pathway of folder where the txt results and pdfs, containing all the plots, will be saved. Several txt and pdf, one per thread, if parallelization. Ignored if export is FALSE. Must be specified if parall is TRUE or if export is TRUE
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# cute.path: character string indicating the absolute path of the cute.R file. Will be remove when cute will be a package. Ignored if parall is FALSE
-# REQUIRED PACKAGES
-# lubridate
-# parallel if parall arguemtn is TRUE (included in the R installation packages but not automatically loaded)
-# pdftools if parall arguemtn is TRUE (included in the R installation packages but not automatically loaded)
-# If the tested function is in a package, this package must be imported first (no parallelization) or must be in the classical R package folder indicated by the lib.path argument (parallelization)
-# RETURN
-# if export is FALSE a list containing:
-# $fun: the tested function
-# $instruction: the initial instruction
-# $sys.info: system and packages info
-# $data: a data frame of all the combination tested, containing the following columns:
-# the different values tested, named by arguments
-# $kind: a vector of character strings indicating the kind of test result: either "ERROR", or "WARNING", or "OK"
-# $problem: a logical vector indicating if error or not
-# $expected.error: optional logical vector indicating the expected error specified in the expect.error argument
-# $message: either NULL if $kind is always "OK", or the messages
-# if export is TRUE 1) the same list object into a .RData file, 2) also the $data data frame into a .txt file, and 3) if expect.error is non NULL and if any discrepancy, the $data data frame into a .txt file but containing only the rows with discrepancies between expected and observed errors
-# one or several pdf if a plotting function is tested and if the plot.fun argument is TRUE
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_get_message()
-# fun_pack()
-# EXAMPLES
-# fun_test(fun = "unique", arg = c("x", "incomparables"), val = list(x = list(1:10, c(1,1,2,8), NA), incomparable = c(TRUE, FALSE, NA)))
-# fun_test(fun = "fun_round", arg = c("data", "dec.nb", "after.lead.zero"), val = list(L1 = list(c(1, 1.0002256, 1.23568), "a", NA), L2 = list(2, c(1,3), NA), L3 = c(TRUE, FALSE, NA)))
-# fun_test(fun = "plot", arg = c("x", "y"), val = list(x = list(1:10, 12:13, NA, (1:10)^2), y = list(1:10, NA, NA)), expect.error = list(x = list(FALSE, TRUE, TRUE, FALSE), y = list(FALSE, TRUE, TRUE)), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = NULL)
-# fun_test(fun = "plot", arg = c("x", "y"), val = list(x = list(1:10, 12:13, NA, (1:10)^2), y = list(1:10, NA, NA)), parall = FALSE, thread.nb = 4, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\")
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_test(fun = "fun_gg_boxplot", arg = c("data1", "y", "categ"), val = list(L1 = list(L1 = obs1), L2 = list(L1 = "Time"), L3 = list(L1 = "Group1")))
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_test(fun = "fun_gg_boxplot", arg = c("data1", "y", "categ"), val = list(L1 = list(obs1), L2 = "Time", L3 = "Group1"), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\")
-# library(ggplot2) ; fun_test(fun = "geom_histogram", arg = c("data", "mapping"), val = list(x = list(data.frame(X = "a", stringsAsFactors = TRUE)), y = list(ggplot2::aes(x = X))), parall = FALSE, thread.nb = NULL, plot.fun = TRUE, res.path = "C:\\Users\\Gael\\Desktop\\", lib.path = "C:\\Program Files\\R\\R-4.0.2\\library\\") # BEWARE: ggplot2::geom_histogram does not work
-# DEBUGGING
-# fun = "unique" ; arg = "x" ; val = list(x = list(1:10, c(1,1,2,8), NA)) ; expect.error = list(x = list(FALSE, FALSE, TRUE)) ; parall = FALSE ; thread.nb = NULL ; plot.fun = FALSE ; export = FALSE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 1 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
-# fun = "unique" ; arg = c("x", "incomparables") ; val = list(x = list(1:10, c(1,1,2,8), NA), incomparable = c(TRUE, FALSE, NA)) ; expect.error = NULL ; parall = FALSE ; thread.nb = 2 ; plot.fun = FALSE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 10 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
-# fun = "plot" ; arg = c("x", "y") ; val = list(x = list(1:10, 12:13, NA), y = list(1:10, NA, NA)) ; expect.error = list(x = list(FALSE, FALSE, TRUE, FALSE), y = list(FALSE, TRUE, TRUE)) ; print.count = 10 ; parall = FALSE ; thread.nb = NULL ; plot.fun = TRUE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL # for function debugging
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun = "fun_gg_boxplot" ; arg = c("data1", "y", "categ") ; val = list(L1 = list(L1 = obs1), L2 = list(L1 = "Time"), L3 = list(L1 = "Group1")) ; expect.error = NULL ; print.count = 10 ; parall = FALSE ; thread.nb = NULL ; plot.fun = TRUE ; export = TRUE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL # for function debugging
-# fun = "unique" ; arg = "x" ; val = list(list(1:3, mean)) ; expect.error = list(TRUE, TRUE) ; parall = FALSE ; thread.nb = NULL ; plot.fun = FALSE ; export = FALSE ; res.path = "C:\\Users\\Gael\\Desktop\\" ; lib.path = NULL ; print.count = 1 ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R" # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check",
-"fun_get_message",
-"fun_pack"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words
-# end reserved words
-# arg with no default values
-mandat.args <- c(
-"fun",
-"arg",
-"val"
-)
-tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), "))")))
-print(tempo)
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = fun, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = arg, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = val, class = "list", fun.name = function.name) ; eval(ee)
-if( ! is.null(expect.error)){
-tempo <- fun_check(data = expect.error, class = "list", fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = parall, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(parall == TRUE){
-if( ! is.null(thread.nb)){
-tempo <- fun_check(data = thread.nb, typeof = "integer", double.as.integer.allowed = TRUE, neg.values = FALSE, length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & thread.nb < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": thread.nb PARAMETER MUST EQUAL OR GREATER THAN 1: ", thread.nb)
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = plot.fun, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = export, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(res.path)){
-tempo <- fun_check(data = res.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = cute.path, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# new environment
-env.name <- paste0("env", as.numeric(Sys.time()))
-if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-assign(env.name, new.env())
-assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
-}
-# end new environment
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"fun",
-"arg",
-"val",
-# "expect.erro", # because can be NULL
-"parall",
-# "thread.nb", # because can be NULL
-"print.count",
-"plot.fun",
-"export",
-# "res.path", # because can be NULL
-# "lib.path", # because can be NULL
-"cute.path"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-# end warning initiation
-# other checkings
-if(grepl(x = fun, pattern = "()$")){ # remove ()
-fun <- sub(x = fun, pattern = "()$", replacement = "")
-}
-if( ! exists(fun)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CHARACTER STRING IN fun ARGUMENT DOES NOT EXIST IN THE R WORKING ENVIRONMENT: ", paste(fun, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}else if( ! all(base::class(get(fun)) == "function")){ # here no env = sys.nframe(), inherit = FALSE for get() because fun is a function in the classical scope
-tempo.cat <- paste0("ERROR IN ", function.name, ": fun ARGUMENT IS NOT CLASS \"function\" BUT: ", paste(base::class(get(fun)), collapse = "\n"), "\nCHECK IF ANY CREATED OBJECT WOULD HAVE THE NAME OF THE TESTED FUNCTION")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(tempo$problem == FALSE & base::length(arg) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": arg ARGUMENT CANNOT BE LENGTH 0")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-for(i2 in 1:base::length(val)){
-tempo1 <- fun_check(data = val[[i2]], class = "vector", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = val[[i2]], class = "list", na.contain = TRUE, fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": COMPARTMENT ", i2, " OF val ARGUMENT MUST BE A VECTOR OR A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}else if(tempo1$problem == FALSE){ # vector split into list compartments
-val[[i2]] <- split(x = val[[i2]], f = 1:base::length(val[[i2]]))
-}
-}
-if(base::length(arg) != base::length(val)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF arg ARGUMENT MUST BE IDENTICAL TO LENGTH OF val ARGUMENT:\nHERE IT IS: ", base::length(arg), " VERSUS ", base::length(val))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-args <- names(formals(get(fun))) # here no env = sys.nframe(), inherit = FALSE for get() because fun is a function in the classical scope
-if( ! all(arg %in% args)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": SOME OF THE STRINGS IN arg ARE NOT ARGUMENTS OF fun\nfun ARGUMENTS: ", paste(args, collapse = " "),"\nPROBLEMATIC STRINGS IN arg: ", paste(arg[ ! arg %in% args], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(sum(sapply(val, FUN = length) > 1) > 43){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CANNOT TEST MORE THAN 43 ARGUMENTS IF THEY ALL HAVE AT LEAST 2 VALUES EACH\nHERE THE NUMBER IS: ", sum(sapply(val, FUN = length) > 1))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if( ! is.null(expect.error)){
-if(base::length(val) != base::length(expect.error)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF val ARGUMENT MUST BE IDENTICAL TO LENGTH OF expect.error ARGUMENT:\nHERE IT IS: ", base::length(val), " VERSUS ", base::length(expect.error))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-for(i3 in 1:base::length(expect.error)){
-tempo1 <- fun_check(data = expect.error[[i3]], class = "vector", mode = "logical", fun.name = function.name)
-tempo2 <- fun_check(data = expect.error[[i3]], class = "list", fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": COMPARTMENT ", i3, " OF expect.error ARGUMENT MUST BE TRUE OR FALSE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}else if(tempo1$problem == FALSE){ # vector split into list compartments
-expect.error[[i3]] <- split(x = expect.error[[i3]], f = 1:base::length(expect.error[[i3]]))
+ # AIM
+ # Convert codon to amino acid using standard genetic code indicated in https://en.wikipedia.org/wiki/DNA_and_RNA_codon_tables
+ # WARNINGS
+ # None
+ # ARGUMENTS
+ # data: single caracter string of three characters, or vector of three caracters, indicating the DNA codon (only "A", "T", "G" and "C" allowed). Case insensitive. Omitted if display argument is TRUE
+ # display: logical. Display the whole genetic table? if TRUE, override data
+ # RETURN
+ # The 1 letter uppercase amino acid of the submitted codon or the whole table if display argument is TRUE
+ # REQUIRED PACKAGES
+ # None
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # fun_check()
+ # EXAMPLE
+ # fun_codon2aa(data = "ATC", display = TRUE)
+ # see http
+ # DEBUGGING
+ # data = "atg" ; display = FALSE
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # arg with no default values
+ mandat.args <- c(
+ "data"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data, class = "vector", typeof = "character", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = display, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){ # normally no NA
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "data",
+ "display"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ # end warning initiation
+ # other checkings
+ if(length(data) == 1L){
+ data <- unlist(strsplit(data, split = ""))
+ }else if(length(data) != 3L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A STRING OF THREE CHARACTERS OR A VECTOR OF THREE CHARACTERS, MADE OF \"A\", \"C\", \"G\", \"T\" ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! all(toupper(data) %in% c("A", "C", "G","T"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A STRING OF THREE CHARACTERS OR A VECTOR OF THREE CHARACTERS, MADE OF \"A\", \"C\", \"G\", \"T\" ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end other checkings
+ # reserved word checking
+ # end reserved word checking
+ # end second round of checking and data preparation
+ # package checking
+ # end package checking
+ # main code
+ # standard genetic code
+ sgc <- array(
+ c(
+ "F", "L", "I", "V",
+ "S", "P", "T", "A",
+ "Y", "H", "N", "D",
+ "C", "R", "S", "G",
+
+ "F", "L", "I", "V",
+ "S", "P", "T", "A",
+ "Y", "H", "N", "D",
+ "C", "R", "S", "G",
+
+ "L", "L", "I", "V",
+ "S", "P", "T", "A",
+ "stop", "Q", "K", "E",
+ "stop", "R", "R", "G",
+
+ "L", "L", "M", "V",
+ "S", "P", "T", "A",
+ "stop", "Q", "K", "E",
+ "W", "R", "R", "G"
+ ),
+ dim = c(4, 4, 4),
+ dimnames = list(
+ first = c("T", "C", "A", "G"),
+ second = c("T", "C", "A", "G"),
+ third = c("T", "C", "A", "G")
+ )
+ )
+ # end standard genetic code
+ if(display == TRUE){
+ output <- sgc
+ }else{
+ data <- toupper(data)
+ output <- eval(parse(text = paste0("sgc['", paste0(data, collapse = "','"), "']")))
+ }
+ return(output)
}
+
+
+######## fun_codon_finder() #### gives the codon number and position in the codon of nucleotid positions
+
+
+fun_codon_finder <- function(
+ pos,
+ begin,
+ end
+){
+ # AIM
+ # gives the codon number and position in the codon of nucleotid positions
+ # WARNINGS
+ # Only for coding sequences (no introns): ((end - begin) + 1) / 3 must be an integer (i.e., modulo zero)
+ # Negatives positions allowed but this implies that one base has the position 0 in the sequence
+ # ARGUMENTS
+ # pos: vector of integers indicating the positions of nucleotids in a sequence. Must be between begin and end arguments
+ # begin: single integer indicating the position of the first base of the coding sequence
+ # end: single indicating the position of the last base of the coding sequence
+ # RETURN
+ # a data frame with column names:
+ # pos: values of the pos argument
+ # codon_nb: the codon number in the CDS encompassing the pos value
+ # codon_pos: the position of pos in the codon (either 1, 2 or 3)
+ # codon_begin: the first base position of the codon
+ # codon_end: the last base position of the codon
+ # REQUIRED PACKAGES
+ # None
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # fun_check()
+ # EXAMPLE
+ # fun_codon_finder(c(5, 6, 8, 10), begin = 5, end = 10)
+ # fun_codon_finder(c(0, 5, 6, 8, 10), begin = -2, end = 12)
+ # see http
+ # DEBUGGING
+ # pos = c(5, 6, 8, 10) ; begin = 5 ; end = 10
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words
+ # end reserved words
+ # arg with no default values
+ mandat.args <- c(
+ "pos",
+ "begin",
+ "end"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = pos, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = begin, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = end, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){ # normally no NA
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "pos",
+ "begin",
+ "end"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ # end warning initiation
+ # other checkings
+ if(begin >= end){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": end ARGUMENT MUST BE STRICTLY GREATER THAN begin ARGUMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if((end - begin + 1) %% 3 != 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ((end - begin) + 1) / 3 MUST BE AN INTEGER (I.E., MODULO ZERO)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(pos < begin | pos > end)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": pos ARGUMENT VALUES MUST BE BETWEEN begin AND end ARGUMENT VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end other checkings
+ # reserved word checking
+ # end reserved word checking
+ # end second round of checking and data preparation
+ # package checking
+ # end package checking
+ # main code
+ first <- seq.int(from = begin, to = end, by = 3)
+ last <- seq.int(from = begin + 2, to = end, by = 3)
+ tempo <- lapply(X = pos, FUN = function(x = X){
+ tempo.log <- x >= first & x <= last
+ if(sum(tempo.log, na.rm = TRUE) != 1){ # check that 1 possible TRUE
+ tempo.cat <- paste0("ERROR IN ", function.name, ": INTERNAL ERROR. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ codon_nb <- which(tempo.log)
+ codon_pos <- as.integer((x - (begin + (codon_nb - 1) * 3) + 1))
+ codon_begin <- as.integer(first[tempo.log])
+ codon_end <- as.integer(last[tempo.log])
+ }
+ return(data.frame(codon_nb = codon_nb, codon_pos = codon_pos, codon_begin = codon_begin, codon_end = codon_end))
+ })
+ tempo <- do.call("rbind", tempo)
+ output <- data.frame(pos = as.integer(pos), tempo)
+ return(output)
}
+
+
+################ Graphics management
+
+
+# this order can be used:
+# fun_width()
+# fun_open()
+# fun_prior_plot() # not for ggplot2
+# plot() or any other plotting
+# fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
+# fun_close()
+
+
+######## fun_width() #### window width depending on classes to plot
+
+
+fun_width <- function(
+ class.nb,
+ inches.per.class.nb = 1,
+ ini.window.width = 7,
+ inch.left.space,
+ inch.right.space,
+ boundarie.space = 0.5
+){
+ # AIM
+ # rescale the width of a window to open depending on the number of classes to plot
+ # can be used for height, considering that it is as if it was a width
+ # this order can be used:
+ # fun_width()
+ # fun_open()
+ # fun_prior_plot() # not for ggplot2
+ # plot() or any other plotting
+ # fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
+ # fun_close()
+ # ARGUMENTS
+ # class.nb: number of class to plot
+ # inches.per.class.nb: number of inches per unit of class.nb. 2 means 2 inches for each boxplot for instance
+ # ini.window.width:initial window width in inches
+ # inch.left.space: left horizontal margin of the figure region (in inches)
+ # inch.right.space: right horizontal margin of the figure region (in inches)
+ # boundarie.space: space between the right and left limits of the plotting region and the plot (0.5 means half a class width)
+ # RETURN
+ # the new window width in inches
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_width(class.nb = 10, inches.per.class.nb = 0.2, ini.window.width = 7, inch.left.space = 1, inch.right.space = 1, boundarie.space = 0.5)
+ # DEBUGGING
+ # class.nb = 10 ; inches.per.class.nb = 0.2 ; ini.window.width = 7 ; inch.left.space = 1 ; inch.right.space = 1 ; boundarie.space = 0.5 # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = class.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = inches.per.class.nb, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = ini.window.width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = inch.left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = inch.right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = boundarie.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ range.max <- class.nb + boundarie.space # the max range of the future plot
+ range.min <- boundarie.space # the min range of the future plot
+ window.width <- inch.left.space + inch.right.space + inches.per.class.nb * (range.max - range.min)
+ return(window.width)
}
-if( ! is.null(res.path)){
-if( ! all(dir.exists(res.path))){ # separation to avoid the problem of tempo$problem == FALSE and res.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE res.path ARGUMENT DOES NOT EXISTS:\n", paste(res.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-}
-if(parall == TRUE & is.null(res.path)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": res.path ARGUMENT MUST BE SPECIFIED IF parall ARGUMENT IS TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(is.null(res.path) & export == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": res.path ARGUMENT MUST BE SPECIFIED IF export ARGUMENT TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(parall == TRUE & export == FALSE){
-export <- TRUE
-tempo.cat <- paste0("WARNING FROM ", function.name, ": export ARGUMENT CONVERTED TO TRUE BECAUSE thread.nb ARGUMENT IS NOT NULL")
-warning(paste0("\n", tempo.cat, "\n"), call. = FALSE)
-}
-if( ! is.null(lib.path)){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-}
-if(parall == TRUE){
-if(grepl(x = cute.path, pattern = "^http")){
-tempo.error1 <- any(grepl(x = fun_get_message(data = "source(cute.path)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE)), pattern = "^[Ee]rror"))
-tempo.error2 <- FALSE
-}else{
-tempo.error1 <- FALSE
-tempo.error2 <- ! file.exists(cute.path)
-}
-if(tempo.error1 | tempo.error2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(grepl(x = cute.path, pattern = "^http"), "URL", "FILE"), " PATH INDICATED IN THE cute.path PARAMETER DOES NOT EXISTS:\n", cute.path)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-}
-# end other checkings
-# reserved word checking
-# end reserved word checking
-# end second round of checking and data preparation
-# package checking
-fun_pack(req.package = c("lubridate"), lib.path = lib.path)
-if(parall == TRUE){
-fun_pack(req.package = c("parallel", "pdftools"), lib.path = lib.path)
-}
-# end package checking
-# declaration of special plot functions
-sp.plot.fun <- c("fun_gg_scatter", "fun_gg_bar", "fun_gg_boxplot")
-# end declaration of special plot functions
-# main code
-ini.warning.length <- base::options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-cat("\nfun_test JOB IGNITION\n")
-ini.date <- Sys.time()
-ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
-if(export == TRUE){
-res.path <- paste0(res.path, "/fun_test_res_", trunc(ini.time))
-if(dir.exists(res.path)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": FOLDER ALREADY EXISTS\n", res.path, "\nPLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-dir.create(res.path)
-}
-}
-total.comp.nb <- prod(sapply(val, FUN = "length"))
-cat(paste0("\nTHE TOTAL NUMBER OF TESTS IS: ", total.comp.nb, "\n"))
-# creation of the txt instruction that includes several loops
-loop.string <- NULL
-end.loop.string <- NULL
-fun.args <- NULL
-fun.args2 <- NULL
-error.values <- NULL
-arg.values <- "list("
-for(i1 in 1:base::length(arg)){
-if(parall == FALSE){
-if(base::length(val[[i1]]) > 1){ # loop only if more than one value in base::length(val[[i1]])
-loop.string <- paste0(loop.string, "for(i", i1, " in 1:", base::length(val[[i1]]), "){")
-end.loop.string <- paste0(end.loop.string, "}")
-}
-}else{
-loop.string <- "for(i in x){"
-end.loop.string <- "}"
-}
-fun.args <- paste0(
-fun.args,
-ifelse(i1 == 1L, "", ", "),
-arg[i1],
-" = val[[",
-i1,
-"]][[",
-if(parall == FALSE){
-if(base::length(val[[i1]]) > 1){
-paste0("i", i1)
-}else{
-"1" # a unique element in val[[i1]]
-}
-}else{
-paste0("i.list[[", i1, "]][i]")
-},
-"]]"
-)
-fun.args2 <- paste0(
-fun.args2,
-ifelse(i1 == 1L, "", ", "),
-arg[i1],
-" = val[[",
-i1,
-"]][[', ",
-if(parall == FALSE){
-if(base::length(val[[i1]]) > 1){
-paste0("i", i1)
-}else{
-"1" # a unique element in val[[i1]]
-}
-}else{
-paste0("i.list[[", i1, "]][i]")
-},
-", ']]"
-)
-arg.values <- paste0(
-arg.values,
-"val[[", i1, "]][[",
-if(parall == FALSE){
-if(base::length(val[[i1]]) > 1){
-paste0("i", i1)
-}else{
-"1" # a unique element in val[[i1]]
-}
-}else{
-paste0("i.list[[", i1, "]][i]")
-},
-"]]",
-ifelse(i1 == base::length(arg), "", ", ")
-)
-error.values <- paste0(
-error.values,
-ifelse(i1 == 1L, "", " | "),
-"expect.error[[", i1, "]][[",
-if(parall == FALSE){
-if(base::length(expect.error[[i1]]) > 1){
-paste0("i", i1)
-}else{
-"1" # a unique element in expect.error[[i1]]
-}
-}else{
-paste0("i.list[[", i1, "]][i]")
-},
-"]]"
-)
-}
-arg.values <- paste0(arg.values, ")")
-fun.test <- paste0(fun, "(", fun.args, ")")
-fun.test2 <- paste0("paste0('", fun, "(", fun.args2, ")')")
-# plot title for special plot functions
-if(plot.fun == TRUE){
-plot.kind <- "classic"
-if(fun %in% sp.plot.fun){
-plot.kind <- "special"
-if(any(arg %in% "title")){ # this is for the special functions
-tempo.match <- regmatches(x = fun.test, m = regexpr(text = fun.test, pattern = "title = .+[,)]"))
-tempo.match <- substring(tempo.match , 1, nchar(tempo.match) - 1)
-fun.test <- sub(x = fun.test, pattern = tempo.match, replacement = paste0(tempo.match, "\ntempo.title"))
-}else{
-fun.test <- sub(x = fun.test, pattern = ")$", replacement = ", title = tempo.title)")
-}
-}
-}
-# end plot title for special plot functions
-kind <- character()
-problem <- logical()
-expected.error <- logical()
-res <- character()
-count <- 0
-print.count.loop <- 0
-plot.count <- 0
-if(base::length(arg) == 1L){
-data <- data.frame()
-}else{ # base::length(arg) == 0L already tested above
-data <- data.frame(t(vector("character", base::length(arg))), stringsAsFactors = FALSE)[-1, ] # -1 to remove the single row created and to have an empty data frame with base::length(arg) columns
-}
-code <- paste(
-loop.string, '
-count <- count + 1
-print.count.loop <- print.count.loop + 1
-arg.values.print <- eval(parse(text = arg.values)) # recover the list of the i1 compartment
-for(j3 in 1:base::length(arg.values.print)){ # WARNING: do not use i1, i2 etc., here because already in loop.string
-tempo.capt <- capture.output(tempo.error <- fun_get_message(data = paste0("paste(arg.values.print[[", j3, "]])"), kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))) # collapsing arg.values sometimes does not work (with function for instance)
-if( ! is.null(tempo.error)){
-arg.values.print[[j3]] <- paste0("SPECIAL VALUE OF CLASS ", base::class(arg.values.print[[j3]]), " AND TYPE ", base::typeof(arg.values.print[[j3]]))
-}
-}
-data <- rbind(data, as.character(sapply(arg.values.print, FUN = "paste", collapse = " ")), stringsAsFactors = FALSE) # each colum is a test
-tempo.capt <- capture.output(tempo.try.error <- fun_get_message(data = eval(parse(text = fun.test2)), kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE))) # data argument needs a character string but eval(parse(text = fun.test2)) provides it (eval parse replace the i1, i2, etc., by the correct values, meaning that only val is required in the env.name environment)
-tempo.capt <- capture.output(tempo.try.warning <- fun_get_message(data = eval(parse(text = fun.test2)), kind = "warning", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE), print.no = TRUE)) # data argument needs a character string but eval(parse(text = fun.test2)) provides it (eval parse replace the i1, i2, etc., by the correct values, meaning that only val is required in the env.name environment)
-if( ! is.null(expect.error)){
-expected.error <- c(expected.error, eval(parse(text = error.values)))
-}
-if( ! is.null(tempo.try.error)){
-kind <- c(kind, "ERROR")
-problem <- c(problem, TRUE)
-res <- c(res, tempo.try.error)
-}else{
-if( ! is.null(tempo.try.warning)){
-kind <- c(kind, "WARNING")
-problem <- c(problem, FALSE)
-res <- c(res, tempo.try.warning)
-}else{
-kind <- c(kind, "OK")
-problem <- c(problem, FALSE)
-res <- c(res, "")
-}
-if(plot.fun == TRUE){
-invisible(dev.set(window.nb))
-plot.count <- plot.count + 1
-tempo.title <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), ifelse(parall == FALSE, count, x[count])))
-if(plot.kind == "classic"){
-eval(parse(text = fun.test))
-tempo <- fun_post_plot(corner.text = tempo.title)
-}else if(plot.kind == "special"){
-eval(parse(text = fun.test))
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR 1 IN ", function.name, ": CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}
-if(print.count.loop == print.count){
-print.count.loop <- 0
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-final.loop <- (tempo.time - ini.time) / count * ifelse(parall == FALSE, total.comp.nb, base::length(x)) # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
-final.exp <- as.POSIXct(final.loop, origin = ini.date)
-cat(paste0(ifelse(parall == FALSE, "\n", paste0("\nIN PROCESS ", process.id, " | ")), "LOOP ", format(count, big.mark=","), " / ", format(ifelse(parall == FALSE, total.comp.nb, base::length(x)), big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
-}
-if(count == ifelse(parall == FALSE, total.comp.nb, base::length(x))){
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-cat(paste0(ifelse(parall == FALSE, "\nLOOP PROCESS ENDED | ", paste0("\nPROCESS ", process.id, " ENDED | ")), "LOOP ", format(count, big.mark=","), " / ", format(ifelse(parall == FALSE, total.comp.nb, base::length(x)), big.mark=","), " | TIME SPENT: ", tempo.lapse, "\n\n"))
-}
-',
-end.loop.string
-)
-# end creation of the txt instruction that includes several loops
-if(parall == TRUE){
-# list of i numbers that will be split
-i.list <- vector("list", base::length(val)) # positions to split in parallel jobs
-for(i2 in 1:base::length(arg)){
-if(i2 == 1L){
-tempo.divisor <- total.comp.nb / base::length(val[[i2]])
-i.list[[i2]] <- rep(1:base::length(val[[i2]]), each = as.integer(tempo.divisor))
-tempo.multi <- base::length(val[[i2]])
-}else{
-tempo.divisor <- tempo.divisor / base::length(val[[i2]])
-i.list[[i2]] <- rep(rep(1:base::length(val[[i2]]), each = as.integer(tempo.divisor)), time = as.integer(tempo.multi))
-tempo.multi <- tempo.multi * base::length(val[[i2]])
-}
-}
-# end list of i numbers that will be split
-tempo.cat <- paste0("PARALLELIZATION INITIATED AT: ", ini.date)
-cat(paste0("\n", tempo.cat, "\n"))
-tempo.thread.nb = parallel::detectCores(all.tests = FALSE, logical = TRUE) # detect the number of threads
-if(tempo.thread.nb < thread.nb){
-thread.nb <- tempo.thread.nb
-}
-tempo.cat <- paste0("NUMBER OF THREADS USED: ", thread.nb)
-cat(paste0("\n ", tempo.cat, "\n"))
-Clust <- parallel::makeCluster(thread.nb, outfile = paste0(res.path, "/fun_test_parall_log.txt")) # outfile to print or cat during parallelization (only possible in a file, outfile = "" do not work on windows)
-tempo.cat <- paste0("SPLIT OF TEST NUMBERS IN PARALLELISATION:")
-cat(paste0("\n ", tempo.cat, "\n"))
-cluster.list <- parallel::clusterSplit(Clust, 1:total.comp.nb) # split according to the number of cluster
-str(cluster.list) # using print(str()) add a NULL below the result
-cat("\n")
-paral.output.list <- parallel::clusterApply( # paral.output.list is a list made of thread.nb compartments, each made of n / thread.nb (mat theo column number) compartment. Each compartment receive the corresponding results of fun_permut(), i.e., data (permuted mat1.perm), warning message, cor (final correlation) and count (number of permutations)
-cl = Clust,
-x = cluster.list,
-function.name = function.name,
-instruction = instruction,
-thread.nb = thread.nb,
-print.count = print.count,
-total.comp.nb = total.comp.nb,
-sp.plot.fun = sp.plot.fun,
-i.list = i.list,
-fun.tested = fun,
-arg.values = arg.values,
-fun.test = fun.test,
-fun.test2 = fun.test2,
-kind = kind,
-problem = problem,
-res = res,
-count = count,
-plot.count = plot.count,
-data = data,
-code = code,
-plot.fun = plot.fun,
-res.path = res.path,
-lib.path = lib.path,
-cute.path = cute.path,
-fun = function(
-x,
-function.name,
-instruction,
-thread.nb,
-print.count,
-total.comp.nb,
-sp.plot.fun,
-i.list,
-fun.tested,
-arg.values,
-fun.test,
-fun.test2,
-kind,
-problem,
-res,
-count,
-plot.count,
-data,
-code,
-plot.fun,
-res.path,
-lib.path,
-cute.path
-){
-# check again: very important because another R
-process.id <- Sys.getpid()
-cat(paste0("\nPROCESS ID ", process.id, " -> TESTS ", x[1], " TO ", x[base::length(x)], "\n"))
-source(cute.path, local = .GlobalEnv)
-fun_pack(req.package = "lubridate", lib.path = lib.path, load = TRUE) # load = TRUE to be sure that functions are present in the environment. And this prevent to use R.lib.path argument of fun_python_pack()
-# end check again: very important because another R
-# plot management
-if(plot.fun == TRUE){
-pdf(file = paste0(res.path, "/plots_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".pdf", paste0("-", x[base::length(x)], ".pdf"))))
-}else{
-pdf(file = NULL) # send plots into a NULL file, no pdf file created
-}
-window.nb <- dev.cur()
-invisible(dev.set(window.nb))
-# end plot management
-# new environment
-ini.date <- Sys.time()
-ini.time <- as.numeric(ini.date) # time of process begin, converted into
-env.name <- paste0("env", ini.time)
-if(exists(env.name, where = -1)){ # verify if still ok when fun_test() is inside a function
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-assign(env.name, new.env())
-assign("val", val, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # var replaced by val
-}
-# end new environment
-print.count.loop <- 0
-suppressMessages(suppressWarnings(eval(parse(text = code))))
-colnames(data) <- arg
-if( ! is.null(expect.error)){
-data <- data.frame(data, kind = kind, problem = problem, expected.error = expected.error, message = res, stringsAsFactors = FALSE)
-}else{
-data <- data.frame(data, kind = kind, problem = problem, message = res, stringsAsFactors = FALSE)
-}
-row.names(data) <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), x))
-sys.info <- sessionInfo()
-sys.info$loadedOnly <- sys.info$loadedOnly[order(names(sys.info$loadedOnly))] # sort the packages
-invisible(dev.off(window.nb))
-rm(env.name) # optional, because should disappear at the end of the function execution
-# output
-output <- list(fun = fun, instruction = instruction, sys.info = sys.info) # data = data finally removed from the output list, because everything combined in a RData file at the end
-save(output, file = paste0(res.path, "/fun_test_", x[1], ifelse(base::length(x) == 1L, ".RData", paste0("-", x[base::length(x)], ".RData"))))
-if(plot.fun == TRUE & plot.count == 0L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN PROCESS ", process.id, ": NO PDF PLOT BECAUSE ONLY ERRORS REPORTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-file.remove(paste0(res.path, "/plots_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".pdf", paste0("-", x[base::length(x)], ".pdf"))))
-}
-table.out <- as.matrix(data)
-# table.out[table.out == ""] <- " " # does not work # because otherwise read.table() converts "" into NA
-table.out <- gsub(table.out, pattern = "\n", replacement = " ")
-write.table(table.out, file = paste0(res.path, "/table_from_fun_test_", x[1], ifelse(base::length(x) == 1L, ".txt", paste0("-", x[base::length(x)], ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
-}
-)
-parallel::stopCluster(Clust)
-# files assembly
-if(base::length(cluster.list) > 1){
-for(i2 in 1:base::length(cluster.list)){
-tempo.file <- paste0(res.path, "/table_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".txt", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".txt"))) # txt file
-tempo <- read.table(file = tempo.file, header = TRUE, stringsAsFactors = FALSE, sep = "\t", row.names = 1, comment.char = "", colClasses = "character") # row.names = 1 (1st column) because now read.table() adds a NA in the header if the header starts by a tabulation, comment.char = "" because colors with #, colClasses = "character" otherwise convert "" (from NULL) into NA
-if(file.exists(paste0(res.path, "/plots_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".pdf", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".pdf"))))){
-tempo.pdf <- paste0(res.path, "/plots_from_fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".pdf", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".pdf"))) # pdf file
-}else{
-tempo.pdf <- NULL
-}
-tempo.rdata <- paste0(res.path, "/fun_test_", min(cluster.list[[i2]], na.rm = TRUE), ifelse(base::length(cluster.list[[i2]]) == 1L, ".RData", paste0("-", max(cluster.list[[i2]], na.rm = TRUE), ".RData"))) # RData file
-if(i2 == 1L){
-final.file <- tempo
-final.pdf <- tempo.pdf
-# new env for RData combining
-env.name <- paste0("env", ini.time)
-if(exists(env.name, where = -1)){ # verify if still ok when fun_test() is inside a function
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-# end new env for RData combining
-}else{
-assign(env.name, new.env())
-load(tempo.rdata, envir = get(env.name))
-tempo.rdata1 <- tempo.rdata
-assign("final.output", get("output", envir = get(env.name)), envir = get(env.name))
-}
-}else{
-final.file <- rbind(final.file, tempo, stringsAsFactors = TRUE)
-final.pdf <- c(final.pdf, tempo.pdf)
-load(tempo.rdata, envir = get(env.name))
-if( ! identical(get("final.output", envir = get(env.name))[c("R.version", "locale", "platform")], get("output", envir = get(env.name))[c("R.version", "locale", "platform")])){
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIFFERENCE BETWEEN OUTPUTS WHILE THEY SHOULD BE IDENTICAL\nPLEASE CHECK\n", tempo.rdata1, "\n", tempo.rdata)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-# add the differences in RData $sysinfo into final.output
-tempo.base1 <- sort(get("final.output", envir = get(env.name))$sys.info$basePkgs)
-tempo.base2 <- sort(get("output", envir = get(env.name))$sys.info$basePkgs)
-tempo.other1 <- names(get("final.output", envir = get(env.name))$sys.info$otherPkgs)
-tempo.other2 <- names(get("output", envir = get(env.name))$sys.info$otherPkgs)
-tempo.loaded1 <- names(get("final.output", envir = get(env.name))$sys.info$loadedOnly)
-tempo.loaded2 <- names(get("output", envir = get(env.name))$sys.info$loadedOnly)
-assign("final.output", {
-x <- get("final.output", envir = get(env.name))
-y <- get("output", envir = get(env.name))
-x$sys.info$basePkgs <- sort(unique(tempo.base1, tempo.base2))
-if( ! all(tempo.other2 %in% tempo.other1)){
-x$sys.info$otherPkgs <- c(x$sys.info$otherPkgs, y$sys.info$otherPkgs[ ! (tempo.other2 %in% tempo.other1)])
-x$sys.info$otherPkgs <- x$sys.info$otherPkgs[order(names(x$sys.info$otherPkgs))]
-}
-if( ! all(tempo.loaded2 %in% tempo.loaded1)){
-x$sys.info$loadedOnly <- c(x$sys.info$loadedOnly, y$sys.info$loadedOnly[ ! (tempo.loaded2 %in% tempo.loaded1)])
-x$sys.info$loadedOnly <- x$sys.info$loadedOnly[order(names(x$sys.info$loadedOnly))]
-}
-x
-}, envir = get(env.name))
-# add the differences in RData $sysinfo into final.output
-}
-}
-file.remove(c(tempo.file, tempo.rdata))
-}
-# combine pdf and save
-if( ! is.null(final.pdf)){
-pdftools::pdf_combine(
-input = final.pdf,
-output = paste0(res.path, "/plots_from_fun_test_1-", total.comp.nb, ".pdf")
-)
-file.remove(final.pdf)
-}
-# end combine pdf and save
-# save RData
-assign("output", c(get("final.output", envir = get(env.name)), data = list(final.file)), envir = get(env.name))
-save(output, file = paste0(res.path, "/fun_test__1-", total.comp.nb, ".RData"), envir = get(env.name))
-rm(env.name) # optional, because should disappear at the end of the function execution
-# end save RData
-# save txt
-write.table(final.file, file = paste0(res.path, "/table_from_fun_test_1-", total.comp.nb, ".txt"), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
-# end save txt
-if( ! is.null(expect.error)){
-final.file <- final.file[ ! final.file$problem == final.file$expected.error, ]
-if(nrow(final.file) == 0L){
-cat(paste0("NO DISCREPANCY BETWEEN EXPECTED AND OBSERVED ERRORS\n\n"))
-}else{
-cat(paste0("DISCREPANCIES BETWEEN EXPECTED AND OBSERVED ERRORS (SEE THE discrepancy_table_from_fun_test_1-", total.comp.nb, ".txt FILE)\n\n"))
-write.table(final.file, file = paste0(res.path, "/discrepancy_table_from_fun_test_1-", total.comp.nb, ".txt"), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
-}
-}
-}
-# end files assembly
-}else{
-# plot management
-if(plot.fun == TRUE){
-pdf(file = paste0(res.path, "/plots_from_fun_test_1", ifelse(total.comp.nb == 1L, ".pdf", paste0("-", total.comp.nb, ".pdf"))))
-}else{
-pdf(file = NULL) # send plots into a NULL file, no pdf file created
-}
-window.nb <- dev.cur()
-invisible(dev.set(window.nb))
-# end plot management
-# new environment
-env.name <- paste0("env", ini.time)
-if(exists(env.name, where = -1)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-assign(env.name, new.env())
-assign("val", val, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # var replaced by val
-}
-# end new environment
-suppressMessages(suppressWarnings(eval(parse(text = code))))
-colnames(data) <- arg
-expect.data <- data.frame()
-if( ! is.null(expect.error)){
-data <- data.frame(data, kind = kind, problem = problem, expected.error = expected.error, message = res, stringsAsFactors = FALSE)
-}else{
-data <- data.frame(data, kind = kind, problem = problem, message = res, stringsAsFactors = FALSE)
-}
-row.names(data) <- paste0("test_", sprintf(paste0("%0", nchar(total.comp.nb), "d"), 1:total.comp.nb))
-sys.info <- sessionInfo()
-sys.info$loadedOnly <- sys.info$loadedOnly[order(names(sys.info$loadedOnly))] # sort the packages
-invisible(dev.off(window.nb))
-rm(env.name) # optional, because should disappear at the end of the function execution
-# output
-output <- list(fun = fun, instruction = instruction, sys.info = sys.info, data = data)
-if(plot.fun == TRUE & plot.count == 0L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO PDF PLOT BECAUSE ONLY ERRORS REPORTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-file.remove(paste0(res.path, "/plots_from_fun_test_1", ifelse(total.comp.nb == 1L, ".pdf", paste0("-", total.comp.nb, ".pdf"))))
-}
-if( ! is.null(expect.error)){
-expect.data <- output$data[ ! output$data$problem == output$data$expected.error, ]
-if(nrow(expect.data) == 0L){
-cat(paste0("NO DISCREPANCY BETWEEN EXPECTED AND OBSERVED ERRORS\n\n"))
-}else{
-cat(paste0("DISCREPANCIES BETWEEN EXPECTED AND OBSERVED ERRORS (SEE THE ", if(export == TRUE){paste0("discrepancy_table_from_fun_test_1", ifelse(total.comp.nb == 1L, "", paste0("-", total.comp.nb)), ".txt FILE")}else{"$data RESULT"}, ")\n\n"))
-if(export == TRUE){
-expect.data <- as.matrix(expect.data)
-expect.data <- gsub(expect.data, pattern = "\n", replacement = " ")
-write.table(expect.data, file = paste0(res.path, "/discrepancy_table_from_fun_test_1", ifelse(total.comp.nb == 1L, ".txt", paste0("-", total.comp.nb, ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
-}
-}
-}
-if( ! is.null(warn)){
-base::options(warning.length = 8170)
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = base::options(warning.length = ini.warning.length), add = TRUE)
-if(export == TRUE){
-save(output, file = paste0(res.path, "/fun_test_1", ifelse(total.comp.nb == 1L, ".RData", paste0("-", total.comp.nb, ".RData"))))
-table.out <- as.matrix(output$data)
-table.out <- gsub(table.out, pattern = "\n", replacement = " ")
-write.table(table.out, file = paste0(res.path, "/table_from_fun_test_1", ifelse(total.comp.nb == 1L, ".txt", paste0("-", total.comp.nb, ".txt"))), row.names = TRUE, col.names = NA, append = FALSE, quote = FALSE, sep = "\t", eol = "\n", na = "")
-}else{
-return(output)
-}
-}
-# after return() ?
-end.date <- Sys.time()
-end.time <- as.numeric(end.date)
-total.lapse <- round(lubridate::seconds_to_period(end.time - ini.time))
-cat(paste0("fun_test JOB END\n\nTIME: ", end.date, "\n\nTOTAL TIME LAPSE: ", total.lapse, "\n\n\n"))
-}
-
-
-################ Object modification
-
-
-######## fun_name_change() #### check a vector of character strings and modify any string if present in another vector
-
-
-fun_name_change <- function(data1, data2, added.string = "_modif"){
-# AIM
-# this function allow to check if a vector of character strings, like column names of a data frame, has elements present in another vector (vector of reserved words or column names of another data frame before merging)
-# ARGUMENTS
-# data1: vector of character strings to check and modify
-# data2: reference vector of character strings
-# added.string: string added at the end of the modified string in data1 if present in data2
-# RETURN
-# a list containing
-# $data: the modified data1 (in the same order as in the initial data1)
-# $ini: the initial elements before modification. NULL if no modification
-# $post: the modified elements in the same order as in ini. NULL if no modification
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# obs1 <- c("A", "B", "C", "D") ; obs2 <- c("A", "C") ; fun_name_change(obs1, obs2)
-# obs1 <- c("A", "B", "C", "C_modif1", "D") ; obs2 <- c("A", "A_modif1", "C") ; fun_name_change(obs1, obs2) # the function checks that the new names are neither in obs1 nor in obs2 (increment the number after the added string)
-# DEBUGGING
-# data1 = c("A", "B", "C", "D") ; data2 <- c("A", "C") ; added.string = "_modif" # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data1, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = data2, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = added.string, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-ini <- NULL
-post <- NULL
-if(any(data1 %in% data2)){
-tempo.names <- data1[data1 %in% data2]
-ini <- NULL
-post <- NULL
-for(i2 in 1:length(tempo.names)){
-count <- 0
-tempo <- tempo.names[i2]
-while(any(tempo %in% data2) | any(tempo %in% data1)){
-count <- count + 1
-tempo <- paste0(tempo.names[i2], "_modif", count)
-}
-data1[data1 %in% tempo.names[i2]] <- paste0(tempo.names[i2], "_modif", count)
-if(count != 0){
-ini <- c(ini, tempo.names[i2])
-post <- c(post, paste0(tempo.names[i2], "_modif", count))
-}
-}
-data <- data1
-}else{
-data <- data1
-}
-output <- list(data = data, ini = ini, post = post)
-return(output)
-}
-
-
-######## fun_df_remod() #### remodeling a data frame to have column name as a qualitative values and vice-versa
-
-
-fun_df_remod <- function(
-data,
-quanti.col.name = "quanti",
-quali.col.name = "quali"
-){
-# AIM
-# if the data frame is made of n numeric columns, a new data frame is created, with the 1st column gathering all the numeric values, and the 2nd column being the name of the columns of the initial data frame. If row names were present in the initial data frame, then a new ini_rowname column is added with the names of the rows
-
-
-# If the data frame is made of one numeric column and one character or factor column, a new data frame is created, with the new columns corresponding to the split numeric values (according to the character column). NA are added a the end of each column to have the same number of rows. BEWARE: in such data frame, rows are not individuals. This means that in the example below, values 10 and 20 are associated on the same row but that means nothing in term of association
-
-
-
-# ARGUMENTS
-# data: data frame to convert
-# quanti.col.name: optional name for the quanti column of the new data frame
-# quali.col.name: optional name for the quali column of the new data frame
-# RETURN
-# the modified data frame
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# obs <- data.frame(col1 = (1:4)*10, col2 = c("A", "B", "A", "A"), stringsAsFactors = TRUE) ; obs ; fun_df_remod(obs)
-# obs <- data.frame(col1 = (1:4)*10, col2 = 5:8, stringsAsFactors = TRUE) ; obs ; fun_df_remod(obs, quanti.col.name = "quanti", quali.col.name = "quali")
-# obs <- data.frame(col1 = (1:4)*10, col2 = 5:8, stringsAsFactors = TRUE) ; rownames(obs) <- paste0("row", 1:4) ; obs ; fun_df_remod(obs, quanti.col.name = "quanti", quali.col.name = "quali")
-# DEBUGGING
-# data = data.frame(a = 1:3, b = 4:6, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
-# data = data.frame(a = 1:3, b = 4:6, c = 11:13, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
-# data = data.frame(a = 1:3, b = letters[1:3], stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
-# data = data.frame(a = 1:3, b = letters[1:3], stringsAsFactors = TRUE) ; quanti.col.name = "TEST" ; quali.col.name = "quali" # for function debugging
-# data = data.frame(b = letters[1:3], a = 1:3, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
-# data = data.frame(b = c("e", "e", "h"), a = 1:3, stringsAsFactors = TRUE) ; quanti.col.name = "quanti" ; quali.col.name = "quali" # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking without fun_check()
-if( ! any(class(data) %in% "data.frame")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data ARGUMENT MUST BE A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking without fun_check()
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = quanti.col.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = quali.col.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-tempo.factor <- unlist(lapply(data, class))
-for(i in 1:length(tempo.factor)){ # convert factor columns as character
-if(all(tempo.factor[i] == "factor")){
-data[, i] <- as.character(data[, i])
-}
-}
-tempo.factor <- unlist(lapply(data, mode))
-if(length(data) == 2L){
-if( ! ((base::mode(data[, 1]) == "character" & base::mode(data[, 2]) == "numeric") | base::mode(data[, 2]) == "character" & base::mode(data[, 1]) == "numeric" | base::mode(data[, 2]) == "numeric" & base::mode(data[, 1]) == "numeric") ){
-tempo.cat <- paste0("ERROR IN ", function.name, ": IF data ARGUMENT IS A DATA FRAME MADE OF 2 COLUMNS, EITHER A COLUMN MUST BE NUMERIC AND THE OTHER CHARACTER, OR THE TWO COLUMNS MUST BE NUMERIC")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if((base::mode(data[, 1]) == "character" | base::mode(data[, 2]) == "character") & (quanti.col.name != "quanti" | quali.col.name != "quali")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": IMPROPER quanti.col.name OR quali.col.name RESETTINGS. THESE ARGUMENTS ARE RESERVED FOR DATA FRAMES MADE OF n NUMERIC COLUMNS ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-if( ! all(tempo.factor %in% "numeric")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": IF data ARGUMENT IS A DATA FRAME MADE OF ONE COLUMN, OR MORE THAN 2 COLUMNS, THESE COLUMNS MUST BE NUMERIC")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(( ! any(tempo.factor %in% "character")) & is.null(names(data))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": NUMERIC DATA FRAME in the data ARGUMENT MUST HAVE COLUMN NAMES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(all(tempo.factor %in% "numeric")){ # transfo 1
-quanti <- NULL
-for(i in 1:length(data)){
-quanti <-c(quanti, data[, i])
-}
-quali <- rep(names(data), each = nrow(data))
-output.data <- data.frame(quanti, quali, stringsAsFactors = TRUE, check.names = FALSE)
-names(output.data) <- c(quanti.col.name, quali.col.name)
-# add the ini_rowname column
-ini.rownames <- rownames(data)
-tempo.data <- data
-rownames(tempo.data) <- NULL
-null.rownames <- (tempo.data)
-if( ! identical(ini.rownames, null.rownames)){
-ini_rowname <- rep(ini.rownames, times = ncol(data))
-output.data <- cbind(output.data, ini_rowname, stringsAsFactors = TRUE)
-}
-}else{ # transfo 2
-if(class(data[, 1]) == "character"){
-data <- cbind(data[2], data[1], stringsAsFactors = TRUE)
-}
-nc.max <- max(table(data[, 2])) # effectif maximum des classes
-nb.na <- nc.max - table(data[,2]) # nombre de NA à ajouter pour réaliser la data frame
-tempo<-split(data[, 1], data[, 2])
-for(i in 1:length(tempo)){tempo[[i]] <- append(tempo[[i]], rep(NA, nb.na[i]))} # des NA doivent être ajoutés lorsque les effectifs sont différents entre les classes. C'est uniquement pour que chaque colonne ait le même nombre de lignes
-output.data<-data.frame(tempo, stringsAsFactors = TRUE, check.names = FALSE)
-}
-return(output.data)
-}
-
-
-
-
-######## fun_round() #### rounding number if decimal present
-
-
-fun_round <- function(data, dec.nb = 2, after.lead.zero = TRUE){
-# AIM
-# round a vector of values, if decimal, with the desired number of decimal digits after the decimal leading zeros
-# WARNINGS
-# Work well with numbers as character strings, but not always with numerical numbers because of the floating point
-# Numeric values are really truncated from a part of their decimal digits, whatever options(digits) settings
-# See ?.Machine or https://stackoverflow.com/questions/5173692/how-to-return-number-of-decimal-places-in-r, with the interexting formula: abs(x - round(x)) > .Machine$double.eps^0.5
-# ARGUMENTS
-# data: a vector of numbers (numeric or character mode)
-# dec.nb: number of required decimal digits
-# after.lead.zero: logical. If FALSE, rounding is performed for all the decimal numbers, whatever the leading zeros (e.g., 0.123 -> 0.12 and 0.00128 -> 0.00). If TRUE, dec.nb are taken after the leading zeros (e.g., 0.123 -> 0.12 and 0.00128 -> 0.0013)
-# RETURN
-# the modified vector
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, 10, 100.001, 333.0001254, 12312.1235), dec.nb = 2, after.lead.zero = FALSE), "\n\n") ; options(digits = ini.options)
-# ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, 10, 100.001, 333.0001254, 12312.1235), dec.nb = 2, after.lead.zero = TRUE), "\n\n") ; options(digits = ini.options)
-# ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, "10", "100.001", "333.0001254", "12312.1235"), dec.nb = 2, after.lead.zero = FALSE), "\n\n") ; options(digits = ini.options)
-# ini.options <- options()$digits ; options(digits = 8) ; cat(fun_round(data = c(NA, "10", "100.001", "333.0001254", "12312.1235"), dec.nb = 2, after.lead.zero = TRUE), "\n\n") ; options(digits = ini.options)
-# DEBUGGING
-# data = data = c(10, 100.001, 333.0001254, 12312.1235) ; dec.nb = 2 ; after.lead.zero = FALSE # # for function debugging
-# data = data = c("10", "100.001", "333.0001254", "12312.1235") ; dec.nb = 2 ; after.lead.zero = TRUE # # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking without fun_check()
-if( ! (all(typeof(data) == "character") | all(typeof(data) == "double") | all(typeof(data) == "integer"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A VECTOR OF NUMBERS (IN NUMERIC OR CHARACTER MODE)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking without fun_check()
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data, class = "vector", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dec.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = after.lead.zero, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-tempo <- grepl(x = data, pattern = "\\.") # detection of decimal numbers
-ini.mode <- base::mode(data)
-data <- as.character(data) # to really truncate decimal digits
-for(i in 1:length(data)){ # scan all the numbers of the vector
-if(tempo[i] == TRUE){ # means decimal number
-if(after.lead.zero == TRUE){
-zero.pos <- unlist(gregexpr(text=data[i], pattern = 0)) # recover all the position of the zeros in the number. -1 if no zeros (do not record the leading and trailing zeros)
-}else{
-zero.pos <- -1 # -1 as if no zero
-}
-dot.pos <- unlist(gregexpr(text=data[i], pattern = "\\.")) # recover all the position of the zeros in the number
-digit.pos <- unlist(gregexpr(text=data[i], pattern = "[[:digit:]]")) # recover all the position of the digits in the number
-dec.pos <- digit.pos[digit.pos > dot.pos]
-count <- 0
-while((dot.pos + count + 1) %in% zero.pos & (dot.pos + count + 1) <= max(dec.pos) & (count + dec.nb) < length(dec.pos)){ # count the number of leading zeros in the decimal part
-count <- count + 1
-}
-data[i] <- formatC(as.numeric(data[i]), digits = (count + dec.nb), format = "f")
-}
-}
-if(ini.mode != "character"){
-data <- as.numeric(data)
-}
-return(data)
-}
-
-
-######## fun_mat_rotate() #### 90° clockwise matrix rotation
-
-
-fun_mat_rotate <- function(data){
-# AIM
-# 90° clockwise matrix rotation
-# applied twice, the function provide the mirror matrix, according to vertical and horizontal symmetry
-# ARGUMENTS
-# data: matrix (matrix class)
-# RETURN
-# the modified matrix
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# obs <- matrix(1:10, ncol = 1) ; obs ; fun_mat_rotate(obs)
-# obs <- matrix(LETTERS[1:10], ncol = 5) ; obs ; fun_mat_rotate(obs)
-# DEBUGGING
-# data = matrix(1:10, ncol = 1)
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data, class = "matrix", fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-for (i in 1:ncol(data)){data[,i] <- rev(data[,i])}
-data <- t(data)
-return(data)
-}
-
-
-######## fun_mat_num2color() #### convert a numeric matrix into hexadecimal color matrix
-
-
-fun_mat_num2color <- function(
-mat1,
-mat.hsv.h = TRUE,
-notch = 1,
-s = 1,
-v = 1,
-forced.color = NULL
-){
-# AIM
-# convert a matrix made of numbers into a hexadecimal matrix for rgb colorization
-# ARGUMENTS:
-# mat1: matrix 1 of non negative numerical values that has to be colored (matrix class). NA allowed
-# mat.hsv.h: logical. Is mat1 the h of hsv colors ? (if TRUE, mat1 must be between zero and 1). If FALSE, mat1 must be made of positive integer values without 0
-# notch: single value between 0 and 1 to shift the successive colors on the hsv circle by + notch
-# s: s argument of hsv(). Must be between 0 and 1
-# v: v argument of hsv(). Must be between 0 and 1
-# forced.color: Must be NULL or hexadecimal color code or name given by colors(). The first minimal values of mat1 will be these colors. All the color of mat1 can be forced using this argument
-# RETURN
-# a list containing:
-# $mat1.name: name of mat1
-# $colored.mat: colors of mat1 in hexa
-# $problem: logical. Is any colors of forced.color overlap the colors designed by the function. NULL if forced.color = NULL
-# $text.problem: text when overlapping colors. NULL if forced.color = NULL or problem == FALSE
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# mat1 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; dimnames(mat1) <- list(LETTERS[1:4], letters[1:2]) ; fun_mat_num2color(mat1, mat.hsv.h = FALSE, notch = 1, s = 1, v = 1, forced.color = NULL)
-# DEBUGGING
-# mat1 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; dimnames(mat1) <- list(LETTERS[1:4], letters[1:2]); mat.hsv.h = FALSE ; notch = 1 ; s = 1 ; v = 1 ; forced.color = c(hsv(1,1,1), hsv(0,0,0)) # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = mat1, mode = "numeric", class = "matrix", na.contain = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = mat.hsv.h, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = notch, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = s, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = v, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# argument checking without fun_check()
-if(mat.hsv.h == TRUE & fun_check(data = mat1, mode = "numeric", prop = TRUE)$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 ARGUMENT MUST BE A MATRIX OF PROPORTIONS SINCE THE mat.hsv.h ARGUMENT IS SET TO TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(forced.color)){
-tempo <- fun_check(data = forced.color, class = "character")
-if(any(tempo$problem == TRUE)){
-paste0("\n\n================\n\n", paste(tempo$text[tempo$problem], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! all(forced.color %in% colors() | grepl(pattern = "^#", forced.color))){ # check that all strings of forced.color start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": forced.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end argument checking without fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-problem <- NULL
-text.problem <- NULL
-mat1.name <- deparse(substitute(mat1))
-# change the scale of the plotted matrix
-if(mat.hsv.h == TRUE){
-if(any(min(mat1, na.rm = TRUE) < 0 | max(mat1, na.rm = TRUE) > 1, na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 MUST BE MADE OF VALUES BETWEEN 0 AND 1 BECAUSE mat.hsv.h ARGUMENT SET TO TRUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-if(any(mat1 - floor(mat1) > 0, na.rm = TRUE) | any(mat1 == 0L, na.rm = TRUE)){ # no need of isTRUE(all.equal()) because we do not require approx here but strictly 0, thus == is ok
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat1 MUST BE MADE OF INTEGER VALUES WITHOUT 0 BECAUSE mat.hsv.h ARGUMENT SET TO FALSE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-mat1 <- mat1 / max(mat1, na.rm = TRUE)
-}
-}
-if(notch != 1){
-different.color <- unique(as.vector(mat1))
-different.color <- different.color[ ! is.na(different.color)]
-tempo.different.color <- different.color + c(0, cumsum(rep(notch, length(different.color) - 1)))
-tempo.different.color <- tempo.different.color - floor(tempo.different.color)
-if(any(duplicated(tempo.different.color) == TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": DUPLICATED VALUES AFTER USING notch (", paste(tempo.different.color[duplicated(tempo.different.color)], collapse = " "), "). TRY ANOTHER notch VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(length(different.color) != length(tempo.different.color)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": LENGTH OF different.color (", paste(different.color, collapse = " "), ") DIFFERENT FROM LENGTH OF tempo.different.color (", paste(tempo.different.color, collapse = " "), ")")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-for(i in 1:length(different.color)){
-mat1[mat1 == different.color[i]] <- tempo.different.color[i] # no need of isTRUE(all.equal()) because different.color comes from mat1
-}
-}
-}
-if( ! is.null(forced.color)){
-hexa.values.to.change <- hsv(unique(sort(mat1))[1:length(forced.color)], s, v)
-}
-mat1[ ! is.na(mat1)] <- hsv(mat1[ ! is.na(mat1)], s, v)
-if( ! is.null(forced.color)){
-if(any(forced.color %in% mat1, na.rm = TRUE)){
-problem <- TRUE
-text.problem <- paste0("THE FOLLOWING COLORS WHERE INTRODUCED USING forced.color BUT WHERE ALREADY PRESENT IN THE COLORED MATRIX :", paste(forced.color[forced.color %in% mat1], collapse = " "))
-}else{
-problem <- FALSE
-}
-for(i in 1:length(hexa.values.to.change)){
-if( ! any(mat1 == hexa.values.to.change[i], na.rm = TRUE)){# no need of isTRUE(all.equal()) because character
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE ", hexa.values.to.change[i], " VALUE FROM hexa.values.to.change IS NOT REPRESENTED IN mat1 : ", paste(unique(as.vector(mat1)), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-mat1[which(mat1 == hexa.values.to.change[i])] <- forced.color[i] # no need of isTRUE(all.equal()) because character
-}
-}
-}
-output <- list(mat1.name = mat1.name, colored.mat = mat1, problem = problem, text.problem = text.problem)
-return(output)
-}
-
-
-######## fun_mat_op() #### assemble several matrices with operation
-
-
-fun_mat_op <- function(mat.list, kind.of.operation = "+"){
-# AIM
-# assemble several matrices of same dimensions by performing by case operation. For instance add the value of all the case 1 (row1 & column1) of the matrices and put it in the case 1 of a new matrix M, add the value of all the case 2 (row2 & column1) of the matrices and put it in the case 2 of a new matrix M, etc.
-
-# c: case
-# i: row number
-# j: column number
-# k: matrix number
-# z: number of matrices
-# ARGUMENTS:
-# mat.list: list of matrices
-# kind.of.operation: either "+" (by case addition), "-" (by case subtraction) or "*" (by case multiplication)
-# RETURN
-# the assembled matrix, with row and/or column names only if all the matrices have identical row/column names
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_comp_2d()
-# EXAMPLES
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "+")
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "*")
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], c(NA, NA))) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2), kind.of.operation = "-")
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(c("A1", "A2", "A3", "A4"), letters[1:2])) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat3 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; fun_mat_op(mat.list = list(mat1, mat2, mat3), kind.of.operation = "+")
-# DEBUGGING
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2) ; mat.list = list(mat1, mat2) ; kind.of.operation = "+" # for function debugging
-# mat1 = matrix(c(1,1,1,2,1,5,9,8), ncol = 2, dimnames = list(LETTERS[1:4], c(NA, NA))) ; mat2 = matrix(c(1,1,1,2,1,5,9,NA), ncol = 2, dimnames = list(LETTERS[1:4], letters[1:2])) ; mat.list = list(mat1, mat2) ; kind.of.operation = "*" # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_comp_2d() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = mat.list, class = "list", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = kind.of.operation, options = c("+", "-", "*"), length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# argument checking without fun_check()
-if(length(mat.list) < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat.list ARGUMENT MUST BE A LIST CONTAINING AT LEAST 2 MATRICES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-for(i1 in 1:length(mat.list)){
-tempo <- fun_check(data = mat.list[[i1]], class = "matrix", mode = "numeric", na.contain = TRUE)
-if(tempo$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ELEMENT ", i1, " OF mat.list ARGUMENT MUST BE A NUMERIC MATRIX")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-ident.row.names <- TRUE
-ident.col.names <- TRUE
-for(i1 in 2:length(mat.list)){
-tempo <- fun_comp_2d(data1 = mat.list[[1]], data2 = mat.list[[i1]])
-if(tempo$same.dim == FALSE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX ", i1, " OF mat.list ARGUMENT MUST HAVE THE SAME DIMENSION (", paste(dim(mat.list[[i1]]), collapse = " "), ") THAN THE MATRIX 1 IN mat.list (", paste(dim(mat.list[[1]]), collapse = " "), ")")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(tempo$same.row.name)){
-if(tempo$same.row.name != TRUE){ # != TRUE to deal with NA
-ident.row.names <- FALSE
-}
-}
-if( ! is.null(tempo$same.col.name)){
-if(tempo$same.col.name != TRUE){ # != TRUE to deal with NA
-ident.col.names <- FALSE
-}
-}
-}
-# end argument checking without fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-output <- mat.list[[1]]
-for(i1 in 2:length(mat.list)){
-output <- get(kind.of.operation)(output, mat.list[[i1]]) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-}
-dimnames(output) <- NULL
-if(ident.row.names == TRUE){
-rownames(output) <- rownames(mat.list[[1]])
-}
-if(ident.col.names == TRUE){
-colnames(output) <- colnames(mat.list[[1]])
-}
-return(output)
-}
-
-
-######## fun_mat_inv() #### return the inverse of a square matrix
-
-
-fun_mat_inv <- function(mat){
-# AIM
-# return the inverse of a square matrix when solve() cannot
-# ARGUMENTS:
-# mat: a square numeric matrix without NULL, NA, Inf or single case (dimension 1, 1) of 0
-# RETURN
-# the inversed matrix
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# mat1 = matrix(c(1,1,1,2,1,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1) # use solve()
-# mat1 = matrix(c(0,0,0,0,0,0,0,0,0), ncol = 3) ; fun_mat_inv(mat = mat1) # use the trick
-# mat1 = matrix(c(1,1,1,2,Inf,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1)
-# mat1 = matrix(c(1,1,1,2,NA,5,9,8,9), ncol = 3) ; fun_mat_inv(mat = mat1)
-# mat1 = matrix(c(1,2), ncol = 1) ; fun_mat_inv(mat = mat1)
-# mat1 = matrix(0, ncol = 1) ; fun_mat_inv(mat = mat1)
-# mat1 = matrix(2, ncol = 1) ; fun_mat_inv(mat = mat1)
-# DEBUGGING
-# mat = matrix(c(1,1,1,2,1,5,9,8,9), ncol = 3) # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = mat, class = "matrix", mode = "numeric", fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# argument checking without fun_check()
-if(ncol(mat) != nrow(mat)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A SQUARE MATRIX")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(mat %in% c(Inf, -Inf, NA))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A MATRIX WITHOUT Inf, -Inf OR NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(all(mat == 0L) & ncol(mat) == 1L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT CANNOT BE A SQUARE MATRIX MADE OF A SINGLE CASE OF 0")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking without fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if(any(grepl(x = try(solve(mat), silent = TRUE)[], pattern = "[Ee]rror"))){
-tempo <- svd(mat)
-val.critique <- which(tempo$d < 10^-8)
-Diag.mod <- diag(1 / tempo$d)
-for(i in val.critique){
-Diag.mod[i, i] <- 0
-}
-return(tempo$v %*% Diag.mod %*% t(tempo$u))
-}else{
-return(solve(mat))
-}
-}
-
-
-######## fun_mat_fill() #### fill the empty half part of a symmetric square matrix
-
-
-fun_mat_fill <- function(mat, empty.cell.string = 0, warn.print = FALSE){
-# AIM
-# detect the empty half part of a symmetric square matrix (either topleft, topright, bottomleft or bottomright)
-# fill this empty half part using the other symmetric half part of the matrix
-# WARNINGS
-# a plot verification using fun_gg_heatmap() is recommanded
-# ARGUMENTS:
-# mat: a numeric or character square matrix with the half part (according to the grand diagonal) filled with NA (any kind of matrix), "0" (character matrix) or 0 (numeric matrix) exclusively (not a mix of 0 and NA in the empty part)
-# empty.cell.string: a numeric, character or NA (no quotes) indicating what empty cells are filled with
-# warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
-# RETURN
-# a list containing:
-# $mat: the filled matrix
-# $warn: the warning messages. Use cat() for proper display. NULL if no warning
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# mat1 = matrix(c(1,NA,NA,NA, 0,2,NA,NA, NA,3,4,NA, 5,6,7,8), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = NA, warn.print = TRUE) # bottomleft example
-# mat1 = matrix(c(1,1,1,2, 0,2,3,0, NA,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = NA, warn.print = TRUE) # error example
-# mat1 = matrix(c(1,1,1,2, 0,2,3,0, NA,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # bottomright example
-# mat1 = matrix(c(1,1,1,2, "a",2,3,NA, "a","a",0,0, "a","a","a",0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = "a", warn.print = TRUE) # topright example
-# mat1 = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # topleft example
-# mat1 = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,0, 5,0,0,0), ncol = 4) ; mat1 ; fun_mat_fill(mat = mat1, empty.cell.string = 0, warn.print = TRUE) # error example
-# DEBUGGING
-# mat = matrix(c(1,NA,NA,NA, 0,2,NA,NA, NA,3,4,NA, 5,6,7,8), ncol = 4) ; empty.cell.string = NA ; warn.print = TRUE # for function debugging
-# mat = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; empty.cell.string = 0 ; warn.print = TRUE # for function debugging # topleft example
-# mat = matrix(c(0,0,0,2, 0,0,3,0, 0,3,0,NA, 5,0,0,0), ncol = 4) ; empty.cell.string = NA ; warn.print = TRUE # for function debugging # topleft example
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = mat, class = "matrix", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = empty.cell.string, class = "vector", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# argument checking without fun_check()
-if(ncol(mat) != nrow(mat)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A SQUARE MATRIX")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (base::mode(mat) %in% c("numeric", "character"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MUST BE A NUMERIC OR CHARACTER MATRIX")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(nrow(mat) == 1L & ncol(mat) == 1L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT CANNOT BE A SQUARE MATRIX MADE OF A SINGLE CASE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(ifelse(is.na(empty.cell.string), ! any(is.na(mat)), ! any(mat == empty.cell.string, na.rm = TRUE))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": mat ARGUMENT MATRIX MUST HAVE CELLS WITH THE EMPTY STRING SPECIFIED IN empty.cell.string ARGUMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking without fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-list.diag <- vector("list", length = nrow(mat) - 1)
-for(i1 in 1:(nrow(mat) - 1)){
-list.diag[[i1]] <- numeric(length = nrow(mat) - i1) # list made of zero
-}
-sector <- c("topleft", "topright", "bottomright", "bottomleft")
-diag.scan <-c( # same order as sector. Recover each diag from center to corner
-"mat[as.matrix(as.data.frame(list(1:(nrow(mat) - i2), (ncol(mat) -i2):1), stringsAsFactors = TRUE))]", # topleft part
-"mat[as.matrix(as.data.frame(list(1:(nrow(mat) - i2), (1:ncol(mat))[-(1:i2)]), stringsAsFactors = TRUE))]", # topright part
-"mat[as.matrix(as.data.frame(list((1 + i2):nrow(mat), ncol(mat):(1 + i2)), stringsAsFactors = TRUE))]", # bottomright part
-"mat[as.matrix(as.data.frame(list((1 + i2):nrow(mat), 1:(ncol(mat) -i2)), stringsAsFactors = TRUE))]" # bottomleft part
-)
-# empty part detection
-empty.sector <- NULL
-full.sector <- NULL
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-for(i1 in 1:length(sector)){
-tempo.list.diag <- list.diag
-for(i2 in 1:(nrow(mat) - 1)){
-tempo.list.diag[[i2]] <- eval(parse(text = diag.scan[i1]))
-if(ifelse(is.na(empty.cell.string), ! all(is.na(tempo.list.diag[[i2]])), ! (all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = TRUE) & ! (is.na(all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE)))))){ # I had to add this ! (is.na(all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE))) because all(tempo.list.diag[[i2]] == empty.cell.string, na.rm = FALSE) gives NA and not FALSE if one NA in tempo.list.diag[[i2]] -> not good for if()
-full.sector <- c(full.sector, sector[i1])
-break
-}
-}
-if(i2 == nrow(mat) - 1){
-if(all(unlist(lapply(tempo.list.diag, FUN = function(x){if(is.na(empty.cell.string)){is.na(x)}else{x == empty.cell.string}})), na.rm = TRUE)){
-empty.sector <- c(empty.sector, sector[i1])
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") EMPTY SECTOR DETECTED ON THE ", toupper(sector[i1]), " CORNER, FULL OF ", empty.cell.string)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE ", toupper(sector[i1]), " SECTOR, DETECTED AS EMPTY, IS NOT? DIFFERENT VALUES IN THIS SECTOR:\n", paste(names(table(unlist(tempo.list.diag), useNA = "ifany")), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}
-# end empty part detection
-if(length(empty.sector) == 0L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") ACCORDING TO empty.cell.string ARGUMENT (", empty.cell.string, "), mat ARGUMENT MATRIX HAS ZERO EMPTY HALF PART")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-if(length(empty.sector) > 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ACCORDING TO empty.cell.string ARGUMENT (", empty.cell.string, "), mat ARGUMENT MATRIX HAS MORE THAN ONE EMPTY HALF PART (ACCORDING TO THE GRAND DIAGONAL): ", paste(empty.sector, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(any(full.sector %in% empty.sector, na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE FUNCTION HAS DETECTED EMPTY AND NON EMPTY HALF PART IN THE SAME SECTOR: ", paste(full.sector[full.sector %in% empty.sector], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(length(empty.sector) + length(full.sector)!= 4){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE FUNCTION HAS DETECTED MORE OR LESS SECTORS THAN 4:\nHALF SECTORS:", paste(empty.sector, collapse = " "), "\nFULL SECTORS:", paste(full.sector, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") ", toupper(empty.sector), " SECTOR HAS BEEN COMPLETED TO BECOME SYMMETRICAL")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# matrix filling
-for(i2 in 1:(nrow(mat) - 1)){
-if(empty.sector == "topleft"){
-eval(parse(text = paste0(diag.scan[1], " <- ", diag.scan[3])))
-}else if(empty.sector == "topright"){
-eval(parse(text = paste0(diag.scan[2], " <- ", diag.scan[4])))
-}else if(empty.sector == "bottomright"){
-eval(parse(text = paste0(diag.scan[3], " <- ", diag.scan[1])))
-}else if(empty.sector == "bottomleft"){
-eval(parse(text = paste0(diag.scan[4], " <- ", diag.scan[2])))
-}
-}
-# end matrix filling
-}
-if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-return(list(mat = mat, warn = warn))
-}
-
-
-######## fun_permut() #### progressively breaks a vector order
-
-
-fun_permut <- function(
-data1,
-data2 = NULL,
-n = NULL,
-seed = NULL,
-print.count = 10,
-text.print = "",
-cor.method = "spearman",
-cor.limit = 0.2,
-warn.print = FALSE,
-lib.path = NULL
-){
-# AIM
-# reorder the elements of the data1 vector by flipping 2 randomly selected consecutive positions either:
-# 1) n times (when n is precised) or
-# 2) until the correlation between data1 and data2 decreases down to the cor.limit (0.2 by default). See cor.limit below to deal with negative correlations
-# Example of consecutive position flipping: ABCD -> BACD -> BADC, etc.
-# designed for discrete values, but worls also for continuous values
-# WARNINGS
-# see # https://www.r-bloggers.com/strategies-to-speedup-r-code/ for code speedup
-# the random switch of non consecutive positions (ABCD -> DBCA for instance) does not work very well as the correlation is quickly obtained but the initial vector structure is mainly kept (no much order). Ths code would be: pos <- ini.pos[1:2] ; pos <- sample.int(n = n , size = 2, replace = FALSE) ; tempo.pos[pos] <- tempo.pos[rev(pos)]
-# ARGUMENTS
-# data1: a vector of at least 2 elements. Must be numeric if data2 is specified
-# data2: a numeric vector of same length as data1
-# n: number of times "flipping 2 randomly selected consecutive positions". Ignored if data2 is specified
-# seed: integer number used by set.seed(). Write NULL if random result is required, an integer otherwise. BEWARE: if not NULL, fun_permut() will systematically return the same result when the other parameters keep the same settings
-# print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
-# text.print: optional message to add to the working progress message every print.count loop
-# cor.method: correlation method. Either "pearson", "kendall" or "spearman". Ignored if data2 is not specified
-# cor.limit: a correlation limit (between 0 and 1). Ignored if data2 is not specified. Compute the correlation between data1 and data2, permute the data1 values, and stop the permutation process when the correlation between data1 and data2 decreases down below the cor limit value (0.2 by default). If cor(data1, data2) is negative, then -cor.limit is used and the process stops until the correlation between data1 and data2 increases up over cor.limit (-0.2 by default). BEWARE: write a positive cor.limit even if cor(data1, data2) is known to be negative. The function will automatically uses -cor.limit. If the initial correlation is already below cor.limit (positive correlation) or over -cor.limit (negative correlation), then the data1 value positions are completely randomized (correlation between data1 and data2 is expected to be 0)
-# warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# RETURN
-# a list containing:
-# $data: the modified vector
-# $warn: potential warning messages (in case of negative correlation when data2 is specified). NULL if non warning message
-# $cor: a spearman correlation between the initial positions (1:length(data1) and the final positions if data2 is not specified and the final correlation between data1 and data2 otherwise, according to cor.method
-# $count: the number of loops used
-# REQUIRED PACKAGES
-# lubridate
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# fun_round()
-# EXAMPLES
-# example (1) showing that for loop, used in fun_permut(), is faster than while loop
-# ini.time <- as.numeric(Sys.time()) ; count <- 0 ; for(i0 in 1:1e9){count <- count + 1} ; tempo.time <- as.numeric(Sys.time()) ; tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time)) ; tempo.lapse
-# example (2) showing that for loop, used in fun_permut(), is faster than while loop
-# ini.time <- as.numeric(Sys.time()) ; count <- 0 ; while(count < 1e9){count <- count + 1} ; tempo.time <- as.numeric(Sys.time()) ; tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time)) ; tempo.lapse
-# fun_permut(data1 = LETTERS[1:5], data2 = NULL, n = 100, seed = 1, print.count = 10, text.print = "CPU NB 4")
-# fun_permut(data1 = 101:110, data2 = 21:30, seed = 1, print.count = 1e4, text.print = "", cor.method = "spearman", cor.limit = 0.2)
-# a way to use the cor.limit argument just considering data1
-# obs1 <- 101:110 ; fun_permut(data1 = obs1, data2 = obs1, seed = 1, print.count = 10, cor.method = "spearman", cor.limit = 0.2)
-# fun_permut(data1 = 1:1e3, data2 = 1e3:1, seed = 1, print.count = 1e6, text.print = "", cor.method = "spearman", cor.limit = 0.7)
-# fun_permut(data1 = 1:1e2, data2 = 1e2:1, seed = 1, print.count = 1e3, cor.limit = 0.5)
-# fun_permut(data1 = c(0,0,0,0,0), n = 5, data2 = NULL, seed = 1, print.count = 1e3, cor.limit = 0.5)
-# DEBUGGING
-# data1 = LETTERS[1:5] ; data2 = NULL ; n = 1e6 ; seed = NULL ; print.count = 1e3 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
-# data1 = LETTERS[1:5] ; data2 = NULL ; n = 10 ; seed = 22 ; print.count = 10 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
-# data1 = 101:110 ; data2 = 21:30 ; n = 10 ; seed = 22 ; print.count = 10 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.2 ; warn.print = TRUE ; lib.path = NULL
-# data1 = 1:1e3 ; data2 = 1e3:1 ; n = 20 ; seed = 22 ; print.count = 1e6 ; text.print = "" ; cor.method = "spearman" ; cor.limit = 0.5 ; warn.print = TRUE ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_round", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data1, class = "vector", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & length(data1) < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A VECTOR OF MINIMUM LENGTH 2. HERE IT IS: ", length(data1))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(data2)){
-tempo <- fun_check(data = data1, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
-if(tempo$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 MUST BE A NUMERIC VECTOR IF data2 ARGUMENT IS SPECIFIED")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = data2, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
-if(length(data1) != length(data2)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 AND data2 MUST BE VECTOR OF SAME LENGTH. HERE IT IS ", length(data1)," AND ", length(data2))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else if(is.null(n)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT CANNOT BE NULL IF data2 ARGUMENT IS NULL")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(n)){
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(seed)){
-tempo <- fun_check(data = seed, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = text.print, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = cor.method, options = c("pearson", "kendall", "spearman"), length =1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = cor.limit, class = "vector", mode = "numeric", prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# package checking
-fun_pack(req.package = "lubridate", lib.path = lib.path)
-# end package checking
-# main code
-# code that protects set.seed() in the global environment
-# see also Protocol 100-rev0 Parallelization in R.docx
-if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
-tempo.random.seed <- .Random.seed
-on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
-}else{
-on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
-}
-set.seed(seed)
-# end code that protects set.seed() in the global environment
-ini.date <- Sys.time() # time of process begin, converted into seconds
-ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
-ini.pos <- 1:length(data1) # positions of data1 before permutation loops
-tempo.pos <- ini.pos # positions of data1 that will be modified during loops
-# pos.selec.seq <- ini.pos[-length(data1)] # selection of 1 position in initial position, without the last because always up permutation (pos -> pos+1 & pos+1 -> pos)
-pos.selec.seq.max <- length(ini.pos) - 1 # max position (used by sample.int() function). See below for - 1
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-count <- 0
-round <- 0
-BREAK <- FALSE
-tempo.cor <- 0
-if(is.null(data2)){
-if(length(table(data1)) == 1L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data1 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data1)))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
-}else{
-if(print.count > n){
-print.count <- n
-}
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP OF ", n, " LOOPS INITIATED | LOOP COUNT: ", format(count, big.mark=",")))
-print.count.loop <- logical(length = print.count)
-print.count.loop[length(print.count.loop)] <- TRUE # not this to avoid long vector, but not forget to reset during printing: print.count.loop[(1:trunc(n / print.count) * print.count)] <- TRUE # counter to speedup
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.date.loop <- Sys.time()
-tempo.time.loop <- as.numeric(tempo.date.loop)
-for(i3 in 1:n){
-count.loop <- count.loop + 1
-pos2 <- pos[count.loop] # selection of 1 position
-tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-if(print.count.loop[count.loop]){
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
-final.loop <- (tempo.time - tempo.time.loop) / i3 * n # expected duration in seconds
-final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP ", i3, " / ", n, " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
-}
-}
-count <- count + n # out of the loop to speedup
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP ENDED | LOOP COUNT: ", format(count, big.mark=",")))
-cat("\n\n")
-}
-}else{
-if(length(table(data1)) == 1L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data1 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data1)))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
-tempo.cor <- 1
-}else if(length(table(data2)) == 1L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO PERMUTATION PERFORMED BECAUSE data2 ARGUMENT SEEMS TO BE MADE OF IDENTICAL ELEMENTS: ", names(table(data2)))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
-tempo.cor <- 1
-}else{
-cor.ini <- cor(x = data1, y = data2, use = "pairwise.complete.obs", method = cor.method)
-tempo.cor <- cor.ini # correlation that will be modified during loops
-neg.cor <- FALSE
-if(tempo.cor < 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") INITIAL ", toupper(cor.method), " CORRELATION BETWEEN data1 AND data2 HAS BEEN DETECTED AS NEGATIVE: ", tempo.cor, ". THE LOOP STEPS WILL BE PERFORMED USING POSITIVE CORRELATIONS BUT THE FINAL CORRELATION WILL BE NEGATIVE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
-neg.cor <- TRUE
-tempo.cor <- abs(tempo.cor)
-cor.ini <- abs(cor.ini)
-}
-if(tempo.cor < cor.limit){ # randomize directly all the position to be close to correlation zero
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") INITIAL ABSOLUTE VALUE OF THE ", toupper(cor.method), " CORRELATION ", fun_round(tempo.cor), " BETWEEN data1 AND data2 HAS BEEN DETECTED AS BELOW THE CORRELATION LIMIT PARAMETER ", cor.limit, "\nTHE data1 SEQUENCE HAS BEEN COMPLETELY RANDOMIZED TO CORRESPOND TO CORRELATION ZERO")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn))) #
-for(i4 in 1:5){ # done 5 times to be sure of the complete randomness
-tempo.pos <- sample(x = tempo.pos, size = length(tempo.pos), replace = FALSE)
-}
-count <- count + 5 # out of the loop to speedup
-}else{
-# smallest correlation decrease
-count <- count + 1 # 1 and not 0 because already 1 performed just below
-pos <- sample.int(n = pos.selec.seq.max , size = 1, replace = TRUE) # selection of 1 position # pos.selec.seq.max because selection of 1 position in initial position, without the last because always up permutation (pos -> pos+1 & pos+1 -> pos)
-tempo.pos[c(pos + 1, pos)] <- tempo.pos[c(pos, pos + 1)]
-tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
-smallest.cor.dec <- cor.ini - tempo.cor
-# end smallest correlation decrease
-# going out of tempo.cor == cor.ini
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "CORRELATION DECREASE AFTER A SINGLE PERMUTATION: ", fun_round(smallest.cor.dec, 4)))
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP -> GOING OUT FROM EQUALITY | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
-print.count.loop <- logical(length = print.count)
-print.count.loop[length(print.count.loop)] <- TRUE # counter to speedup
-count.loop <- 0 #
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.date.loop <- Sys.time()
-tempo.time.loop <- as.numeric(tempo.date.loop)
-while(tempo.cor == cor.ini){ # to be out of equality between tempo.cor and cor.ini at the beginning (only valid for very long vector)
-count <- count + 1
-count.loop <- count.loop + 1
-pos2 <- pos[count.loop]
-tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
-if(print.count.loop[count.loop]){
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP", format(count.loop, big.mark=","), " / ? | COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TIME SPENT: ", tempo.lapse))
-}
-}
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FIRST WHILE LOOP STEP END | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TOTAL SPENT TIME: ", tempo.lapse))
-if(tempo.cor < cor.limit){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE FIRST FOR & WHILE LOOP STEPS HAVE BEEN TOO FAR AND SUBSEQUENT LOOP STEPS WILL NOT RUN")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end going out of tempo.cor == cor.ini
-# estimation of the average correlation decrease per loop on x loops and for loop execution
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE/FOR LOOPS INITIATION | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
-count.est <- 1e5
-first.round <- TRUE
-GOBACK <- FALSE
-while(tempo.cor > cor.limit){
-round <- round + 1
-# estimation step
-if(first.round == TRUE){
-first.round <- FALSE
-cor.dec.per.loop <- numeric(length = 5)
-loop.nb.est <- Inf
-cor.est.ini <- tempo.cor
-cor.est <- numeric(length = 5)
-for(i6 in 1:5){ # connected to cor.dec.per.loop
-tempo.pos.est <- tempo.pos
-pos <- sample.int(n = pos.selec.seq.max , size = count.est, replace = TRUE) # selection of n position
-for(i7 in 1:count.est){
-pos2 <- pos[i7] # selection of 1 position
-tempo.pos.est[c(pos2 + 1, pos2)] <- tempo.pos.est[c(pos2, pos2 + 1)]
-}
-tempo.cor.est <- abs(cor(x = data1[tempo.pos.est], y = data2, use = "pairwise.complete.obs", method = cor.method))
-cor.est[i6] <- tempo.cor.est
-tempo.cor.dec.per.loop <- (cor.est.ini - tempo.cor.est) / count.est # correlation decrease per loop
-if(is.na(tempo.cor.dec.per.loop) | ! is.finite(tempo.cor.dec.per.loop)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2\ncor.est.ini: ", cor.est.ini, "\ntempo.cor.est: ", tempo.cor.est)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-cor.dec.per.loop[i6] <- tempo.cor.dec.per.loop
-}
-cor.est <- cor.est[which.max(cor.dec.per.loop)] # max to avoid to go to far with for loop (tempo.cor below tempo.limit)
-cor.dec.per.loop <- max(cor.dec.per.loop, na.rm = TRUE) # max to avoid to go to far with for loop (tempo.cor below tempo.limit)
-loop.nb.est <- round((tempo.cor - cor.limit) / cor.dec.per.loop)
-}else{
-if(GOBACK == TRUE){
-loop.nb.est <- round(loop.nb.est / 2)
-}else{
-cor.dec.per.loop <- (cor.ini - tempo.cor) / count
-loop.nb.est <- round((tempo.cor - cor.limit) / cor.dec.per.loop)
-}
-}
-# end estimation step
-# loop step
-if(is.na(loop.nb.est) | ! is.finite(loop.nb.est)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1\nloop.nb.est: ", loop.nb.est, "\ncor.ini: ", cor.ini, "\ntempo.cor: ", tempo.cor, "\ncor.limit: ", cor.limit, "\ncor.dec.per.loop: ", cor.dec.per.loop)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(loop.nb.est > 1e4){ # below -> leave the while loop
-tempo.pos.secu <- tempo.pos
-count.secu <- count
-tempo.cor.secu <- tempo.cor
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "INITIAL SETTINGS BEFORE ROUND: ", round, " | LOOP COUNT: ", format(count, big.mark=","), " | GO BACK: ", GOBACK, " | LOOP NUMBER ESTIMATION: ", format(loop.nb.est, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
-print.count.loop <- logical(length = print.count)
-print.count.loop[length(print.count.loop)] <- TRUE # not this to avoid long vector, but not forget to reset during printing: print.count.loop[(1:trunc(n / print.count) * print.count)] <- TRUE # counter to speedup
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.date.loop <- Sys.time()
-tempo.time.loop <- as.numeric(tempo.date.loop)
-for(i6 in 1:loop.nb.est){
-count.loop <- count.loop + 1
-pos2 <- pos[count.loop] # selection of 1 position
-tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-if(print.count.loop[count.loop]){
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
-final.loop <- (tempo.time - tempo.time.loop) / i6 * loop.nb.est # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
-final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FOR LOOP | ROUND ", round, " | LOOP: ", format(i6, big.mark=","), " / ", format(loop.nb.est, big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
-}
-}
-count <- count + loop.nb.est # out of the loop to speedup
-tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
-if(tempo.cor > tempo.cor.secu | ((tempo.cor - cor.limit) < 0 & abs(tempo.cor - cor.limit) > smallest.cor.dec * round(log10(max(ini.pos, na.rm = TRUE))))){
-GOBACK <- TRUE
-tempo.pos <- tempo.pos.secu
-count <- count.secu
-tempo.cor <- tempo.cor.secu
-}else{
-GOBACK <- FALSE
-}
-}else{
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "FINAL WHILE LOOP | LOOP COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4)))
-print.count.loop <- logical(length = print.count)
-print.count.loop[length(print.count.loop)] <- TRUE # counter to speedup
-count.loop <- 0 #
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # selection of random positions. BEWARE: n = pos.selec.seq.max because already - 1 (see above) but is connected to tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.cor.loop <- tempo.cor
-tempo.date.loop <- Sys.time()
-tempo.time.loop <- as.numeric(tempo.date.loop)
-while(tempo.cor > cor.limit){
-count <- count + 1
-count.loop <- count.loop + 1
-pos2 <- pos[count.loop]
-tempo.pos[c(pos2 + 1, pos2)] <- tempo.pos[c(pos2, pos2 + 1)]
-tempo.cor <- abs(cor(x = data1[tempo.pos], y = data2, use = "pairwise.complete.obs", method = cor.method))
-if(print.count.loop[count.loop]){
-count.loop <- 0
-pos <- sample.int(n = pos.selec.seq.max , size = print.count, replace = TRUE) # BEWARE: never forget to resample here
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - tempo.time.loop))
-final.loop <- (tempo.time - tempo.time.loop) / (tempo.cor.loop - tempo.cor) * (tempo.cor - cor.limit) # expected duration in seconds # tempo.cor.loop - tempo.cor always positive and tempo.cor decreases progressively starting from tempo.cor.loop
-final.exp <- as.POSIXct(final.loop, origin = tempo.date.loop)
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE LOOP | LOOP NB: ", format(count.loop, big.mark=","), " | COUNT: ", format(count, big.mark=","), " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
-}
-}
-}
-}
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-cat(paste0("\n", ifelse(text.print == "", "", paste0(text.print, " | ")), "WHILE/FOR LOOPS END | LOOP COUNT: ", format(count, big.mark=","), " | NB OF ROUNDS: ", round, " | CORRELATION LIMIT: ", fun_round(cor.limit, 4), " | ABS TEMPO CORRELATION: ", fun_round(tempo.cor, 4), " | TOTAL SPENT TIME: ", tempo.lapse))
-}
-tempo.cor <- ifelse(neg.cor == TRUE, -tempo.cor, tempo.cor)
-}
-}
-cat("\n\n")
-if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE), add = TRUE)
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-output <- list(data = data1[tempo.pos], warn = warn, cor = if(is.null(data2)){cor(ini.pos, tempo.pos, method = "spearman")}else{tempo.cor}, count = count)
-return(output)
-}
-
-
-######## fun_slide() #### return a computation made on a vector using a sliding window
-
-
-fun_slide <- function(
-data,
-window.size,
-step,
-from = NULL,
-to = NULL,
-fun,
-args = NULL,
-boundary = "left",
-parall = FALSE,
-thread.nb = NULL,
-print.count = 100,
-res.path = NULL,
-lib.path = NULL,
-verbose = TRUE,
-cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
-){
-# AIM
-# return a computation made on a vector using a sliding window
-# WARNINGS
-# The function uses two strategies, depending on the amout of memory required which depends on the data, window.size and step arguments. The first one uses lapply(), is generally fast but requires lots of memory. The second one uses a parallelized loop. The choice between the two strategies is automatic if parall argument is FALSE, and is forced toward parallelization if parall argument is TRUE
-# The parall argument forces the parallelization, which is convenient when the data argument is big, because the lapply() function is sometimes slower than the parallelization
-# Always use the env argument when fun_slide() is used inside functions
-# ARGUMENTS
-# data: vector, matrix, table or array of numeric values (mode must be numeric). Inf not allowed. NA will be removed before computation
-# window.size: single numeric value indicating the width of the window sliding across data (in the same unit as data value)
-# step: single numeric value indicating the step between each window (in the same unit as data value). Cannot be larger than window.size
-# from: value of the left boundary of the first sliding window. If NULL, min(data) is used. The first window will strictly have from or min(data) as left boundary
-# to: value of the right boundary of the last sliding window. If NULL, max(data) is used. Warning: (1) the final last window will not necessary have to|max(data) as right boundary. In fact the last window will be the one that contains to|max(data) for the first time, i.e., min[from|min(data) + window.size + n * step >= to|max(data)]; (2) In fact, the >= in min[from|min(data) + window.size + n * step >= to|max(data)] depends on the boundary argument (>= for "right" and > for "left"); (3) to have the rule (1) but for the center of the last window, use to argument as to = to|max(data) + window.size / 2
-# fun: function or character string (without brackets) indicating the name of the function to apply in each window. Example: fun = "mean", or fun = mean
-# args: character string of additional arguments of fun (separated by a comma between the quotes). Example args = "na.rm = TRUE" for fun = mean. Ignored if NULL
-# boundary: either "left" or "right". Indicates if the sliding window includes values equal to left boundary and exclude values equal to right boundary ("left") or the opposite ("right")
-# parall: logical. Force parallelization ?
-# thread.nb: numeric value indicating the number of threads to use if ever parallelization is required. If NULL, all the available threads will be used. Ignored if parall is FALSE
-# print.count: interger value. Print a working progress message every print.count during loops. BEWARE: can increase substentially the time to complete the process using a small value, like 10 for instance. Use Inf is no loop message desired
-# res.path: character string indicating the absolute pathway where the parallelization log file will be created if parallelization is used. If NULL, will be created in the R current directory
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# verbose: logical. Display messages?
-# cute.path: character string indicating the absolute path of the cute.R file. Will be remove when cute will be a package. Ignored if parall is FALSE
-# RETURN
-# a data frame containing
-#$left : the left boundary of each window (in the unit of the data argument)
-#$right : the right boundary of each window (in the unit of data argument)
-#$center : the center of each window (in the unit of data argument)
-#$value : the computed value by the fun argument in each window)
-# REQUIRED PACKAGES
-# lubridate
-# parallel if parall argument is TRUE (included in the R installation packages but not automatically loaded)
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_get_message
-# fun_pack()
-# EXAMPLES
-# fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "left")
-# fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "right") # effect of boundary argument
-# fun_slide(data = c(1:10, 100:110, 500), window.size = 5, step = 2, fun = length, boundary = "left", parall = TRUE) # effect of parall argument
-# DEBUGGING
-# data = c(1:10, 100:110, 500) ; window.size = 5 ; step = 2 ; from = NULL ; to = NULL ; fun = length ; args = NULL ; boundary = "left" ; parall = FALSE ; thread.nb = NULL ; print.count = 100 ; res.path = NULL ; lib.path = NULL ; verbose = TRUE ; cute.path = "C:\\Users\\Gael\\Documents\\Git_projects\\cute_little_R_functions\\cute_little_R_functions.R"
-# data = lag.pos; window.size = window.size; step = step; fun = length; from = min(a$pos); to = max(a$pos)
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check",
-"fun_get_message",
-"fun_pack"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words
-# end reserved words
-# arg with no default values
-mandat.args <- c(
-"data",
-"window.size",
-"step",
-"fun"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data, mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = window.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = step, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(from)){
-tempo <- fun_check(data = from, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(to)){
-tempo <- fun_check(data = to, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo1 <- fun_check(data = fun, class = "vector", mode = "character", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = fun, class = "function", length = 1, fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": fun ARGUMENT MUST BE A FUNCTION OR A CHARACTER STRING OF THE NAME OF A FUNCTION")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(args)){
-tempo <- fun_check(data = args, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = boundary, options = c("left", "right"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = parall, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(parall == TRUE){
-if( ! is.null(thread.nb)){
-tempo <- fun_check(data = thread.nb, typeof = "integer", double.as.integer.allowed = TRUE, neg.values = FALSE, length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & thread.nb < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": thread.nb PARAMETER MUST EQUAL OR GREATER THAN 1: ", thread.nb)
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-tempo <- fun_check(data = print.count, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if( ! is.null(res.path)){
-tempo <- fun_check(data = res.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = verbose, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = cute.path, class = "vector", typeof = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# new environment
-env.name <- paste0("env", as.numeric(Sys.time()))
-if(exists(env.name, where = -1)){ # verify if still ok when fun_info() is inside a function
-tempo.cat <- paste0("ERROR IN ", function.name, ": ENVIRONMENT env.name ALREADY EXISTS. PLEASE RERUN ONCE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-assign(env.name, new.env())
-}
-# end new environment
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS", "THIS ARGUMENT"), " CANNOT JUST BE NA:", paste0(tempo.arg[tempo.log], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"data",
-"window.size",
-"step",
-# "from", # because can be NULL
-# "to", # because can be NULL
-"fun",
-# "args", # because can be NULL
-"boundary",
-"parall",
-# "thread.nb", # because can be NULL
-"print.count",
-# "res.path", # because can be NULL
-# "lib.path", # because can be NULL
-"verbose",
-"cute.path"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-# end warning initiation
-# other checkings
-if(length(data) == 0){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT CANNOT BE LENGTH 0")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(any( ! is.finite(data))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT CANNOT CONTAIN Inf VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if(step > window.size){
-tempo.cat <- paste0("ERROR IN ", function.name, ": step ARGUMENT MUST BE LOWER THAN window.size ARGUMENT\nstep: ", paste(step, collapse = " "), "\nwindow.size: ", paste(window.size, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-if( ! is.null(res.path)){
-if( ! all(dir.exists(res.path))){ # separation to avoid the problem of tempo$problem == FALSE and res.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE res.path ARGUMENT DOES NOT EXISTS:\n", paste(res.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-}else{
-res.path <- getwd() # working directory
-}
-if( ! is.null(lib.path)){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE)
-}
-}
-if(grepl(x = cute.path, pattern = "^http")){
-tempo.error1 <- any(grepl(x = fun_get_message(data = "source(cute.path)", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE)), pattern = "^[Ee]rror"))
-tempo.error2 <- FALSE
-}else{
-tempo.error1 <- FALSE
-tempo.error2 <- ! file.exists(cute.path)
-}
-if(tempo.error1 | tempo.error2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(grepl(x = cute.path, pattern = "^http"), "URL", "FILE"), " PATH INDICATED IN THE cute.path PARAMETER DOES NOT EXISTS:\n", cute.path)
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-# end other checkings
-# reserved word checking
-# end reserved word checking
-# end second round of checking and data preparation
-# package checking
-fun_pack(req.package = c("lubridate"), lib.path = lib.path)
-fun_pack(req.package = c("parallel"), lib.path = lib.path)
-# end package checking
-# main code
-if(verbose == TRUE){
-cat("\nfun_slide JOB IGNITION\n")
-}
-ini.date <- Sys.time()
-ini.time <- as.numeric(ini.date) # time of process begin, converted into seconds
-fun <- match.fun(fun) # make fun <- get(fun) is fun is a function name written as character string of length 1
-if(boundary == "left"){
-left <- ">="
-right <- "<"
-right.last.wind <- ">"
-}else if(boundary == "right"){
-left <- ">"
-right <- "<="
-right.last.wind <- ">="
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-data <- as.vector(data)
-data <- sort(data, na.last = NA) # NA removed
-wind <- data.frame(left = seq(from = if(is.null(from)){min(data, na.rm = TRUE)}else{from}, to = if(is.null(to)){max(data, na.rm = TRUE)}else{to}, by = step), stringsAsFactors = TRUE)
-wind <- data.frame(wind, right = wind$left + window.size, stringsAsFactors = TRUE)
-wind <- data.frame(wind, center = (wind$left + wind$right) / 2, stringsAsFactors = TRUE)
-if(all(wind$right < if(is.null(to)){max(data, na.rm = TRUE)}else{to})){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 2")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# The 3 next lines is for the rule of to argument with center (see to argument description)
-# if(any(wind$center > max(data, na.rm = TRUE))){
-# wind <- wind[ ! wind$center > max(data, na.rm = TRUE),]
-# }
-if(sum(get(right.last.wind)(wind$right, if(is.null(to)){max(data, na.rm = TRUE)}else{to}), na.rm = TRUE) > 1){ # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-tempo.log <- get(right.last.wind)(wind$right, if(is.null(to)){max(data, na.rm = TRUE)}else{to}) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-tempo.log[min(which(tempo.log), na.rm = TRUE)] <- FALSE # convert the first left boundary that goes above max(data, na.rm = TRUE) to FALSE to keep it (the next ones will be removed)
-wind <- wind[ ! tempo.log,]
-}
-
-# test if lapply can be used
-if(parall == FALSE){
-assign("wind", wind, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # wind assigned in a new envir for test
-assign("data", data, envir = get(env.name, env = sys.nframe(), inherit = FALSE)) # data assigned in a new envir for test
-tempo.message <- fun_get_message(data="lapply(X = wind$left, Y = data, FUN = function(X, Y){res <- get(left)(Y, X) ; return(res)})", kind = "error", header = FALSE, env = get(env.name, env = sys.nframe(), inherit = FALSE), print.no = FALSE) # no env = sys.nframe(), inherit = FALSE in get(left) because look for function in the classical scope
-# rm(env.name) # optional, because should disappear at the end of the function execution
-}else{
-tempo.message <- "ERROR" # with this, force the parallelization by default
-}
-# end test if lapply can be used
-if( ! any(grepl(x = tempo.message, pattern = "ERROR.*"))){
-left.log <- lapply(X = wind$left, Y = data, FUN = function(X, Y){
-res <- get(left)(Y, X) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-return(res)
-})
-right.log <- lapply(X = wind$right, Y = data, FUN = function(X, Y){
-res <- get(right)(Y, X) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-return(res)
-})
-log <- mapply(FUN = "&", left.log, right.log, SIMPLIFY = FALSE)
-output <- eval(parse(text = paste0("sapply(lapply(log, FUN = function(X){(data[X])}), FUN = fun", if( ! is.null(args)){paste0(", ", args)}, ")"))) # take the values of the data vector according to log (list of logical, each compartment of length(data)) and apply fun with args of fun
-if(length(output) != nrow(wind)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-output <- data.frame(wind, value = output, stringsAsFactors = TRUE)
-}
-}else{
-if(verbose == TRUE){
-tempo.cat <- paste0("PARALLELIZATION INITIATED AT: ", ini.date)
-cat(paste0("\n", tempo.cat, "\n"))
-}
-tempo.thread.nb = parallel::detectCores(all.tests = FALSE, logical = TRUE) # detect the number of threads
-if( ! is.null(thread.nb)){
-if(tempo.thread.nb < thread.nb){
-thread.nb <- tempo.thread.nb
-if(verbose == TRUE){
-tempo.cat <- paste0("ONLY: ", tempo.thread.nb, " THREADS AVAILABLE")
-cat(paste0("\n", tempo.cat, "\n"))
-}
-}
-}else{
-thread.nb <- tempo.thread.nb
-}
-if(verbose == TRUE){
-tempo.cat <- paste0("NUMBER OF THREADS USED: ", thread.nb)
-cat(paste0("\n ", tempo.cat, "\n"))
-}
-Clust <- parallel::makeCluster(thread.nb, outfile = paste0(res.path, "/fun_slide_parall_log.txt")) # outfile to print or cat during parallelization (only possible in a file, outfile = "" do not work on windows)
-cluster.list <- parallel::clusterSplit(Clust, 1:nrow(wind)) # split according to the number of cluster
-if(verbose == TRUE){
-tempo.cat <- paste0("SPLIT OF TEST NUMBERS IN PARALLELISATION:")
-cat(paste0("\n ", tempo.cat, "\n"))
-str(cluster.list) # using print(str()) add a NULL below the result
-cat("\n")
-}
-paral.output.list <- parallel::clusterApply( #
-cl = Clust,
-x = cluster.list,
-function.name = function.name,
-data = data,
-FUN = fun, # because fun argument of clusterApply
-args = args,
-thread.nb = thread.nb,
-print.count = print.count,
-wind = wind,
-left = left,
-right = right,
-res.path = res.path,
-lib.path = lib.path,
-verbose = verbose,
-cute.path = cute.path,
-fun = function(
-x,
-function.name,
-data,
-FUN,
-args,
-thread.nb,
-print.count,
-wind,
-left,
-right,
-res.path,
-lib.path,
-verbose,
-cute.path
-){
-# check again: very important because another R
-process.id <- Sys.getpid()
-cat(paste0("\nPROCESS ID ", process.id, " -> TESTS ", x[1], " TO ", x[length(x)], "\n"))
-source(cute.path, local = .GlobalEnv)
-fun_pack(req.package = "lubridate", lib.path = lib.path, load = TRUE) # load = TRUE to be sure that functions are present in the environment. And this prevent to use R.lib.path argument of fun_python_pack()
-# end check again: very important because another R
-ini.date <- Sys.time()
-ini.time <- as.numeric(ini.date) # time of process begin, converted into
-output <- NULL
-print.count.loop <- 0
-for(i4 in 1:length(x)){
-print.count.loop <- print.count.loop + 1
-log <- get(left)(data, wind$left[x[i4]]) & get(right)(data, wind$right[x[i4]]) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-output <- c(output, eval(parse(text = paste0("FUN(data[log]", if( ! is.null(args)){paste0(", ", args)}, ")"))))
-if(verbose == TRUE){
-if(print.count.loop == print.count){
-print.count.loop <- 0
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-final.loop <- (tempo.time - ini.time) / i4 * length(x) # expected duration in seconds # intra nb.compar loop lapse: time lapse / cycles done * cycles remaining
-final.exp <- as.POSIXct(final.loop, origin = ini.date)
-cat(paste0("\nIN PROCESS ", process.id, " | LOOP ", format(i4, big.mark=","), " / ", format(length(x), big.mark=","), " | TIME SPENT: ", tempo.lapse, " | EXPECTED END: ", final.exp))
-}
-if(i4 == length(x)){
-tempo.time <- as.numeric(Sys.time())
-tempo.lapse <- round(lubridate::seconds_to_period(tempo.time - ini.time))
-cat(paste0("\nPROCESS ", process.id, " ENDED | LOOP ", format(i4, big.mark=","), " / ", format(length(x), big.mark=","), " | TIME SPENT: ", tempo.lapse, "\n\n"))
-}
-}
-}
-wind <- wind[x, ]
-if(length(output) != nrow(wind)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-output <- data.frame(wind, value = output, stringsAsFactors = TRUE)
-return(output)
-}
-}
-)
-parallel::stopCluster(Clust)
-# result assembly
-output <- data.frame()
-for(i2 in 1:length(paral.output.list)){ # compartment relatives to each parallelization
-output <- rbind(output, paral.output.list[[i2]], stringsAsFactors = TRUE)
-}
-# end result assembly
-if(nrow(output) != nrow(wind)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 5\nlength(output): ", length(output), "\nnrow(wind): ", nrow(wind))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-output <- output[order(output$left), ]
-}
-}
-if(verbose == TRUE){
-end.date <- Sys.time()
-end.time <- as.numeric(end.date)
-total.lapse <- round(lubridate::seconds_to_period(end.time - ini.time))
-cat(paste0("\nfun_slide JOB END\n\nTIME: ", end.date, "\n\nTOTAL TIME LAPSE: ", total.lapse, "\n\n\n"))
-}
-return(output)
-}
-
-
-
-
-######## fun_codon2aa() #### convert codon to amino acid using standard genetic code
-
-
-fun_codon2aa <- function(
-data,
-display = FALSE
-){
-# AIM
-# Convert codon to amino acid using standard genetic code indicated in https://en.wikipedia.org/wiki/DNA_and_RNA_codon_tables
-# WARNINGS
-# None
-# ARGUMENTS
-# data: single caracter string of three characters, or vector of three caracters, indicating the DNA codon (only "A", "T", "G" and "C" allowed). Case insensitive. Omitted if display argument is TRUE
-# display: logical. Display the whole genetic table? if TRUE, override data
-# RETURN
-# The 1 letter uppercase amino acid of the submitted codon or the whole table if display argument is TRUE
-# REQUIRED PACKAGES
-# None
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# fun_check()
-# EXAMPLE
-# fun_codon2aa(data = "ATC", display = TRUE)
-# see http
-# DEBUGGING
-# data = "atg" ; display = FALSE
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words
-# end reserved words
-# arg with no default values
-mandat.args <- c(
-"data"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data, class = "vector", typeof = "character", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = display, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){ # normally no NA
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"data",
-"display"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-# end warning initiation
-# other checkings
-if(length(data) == 1L){
-data <- unlist(strsplit(data, split = ""))
-}else if(length(data) != 3L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A STRING OF THREE CHARACTERS OR A VECTOR OF THREE CHARACTERS, MADE OF \"A\", \"C\", \"G\", \"T\" ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! all(toupper(data) %in% c("A", "C", "G","T"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A STRING OF THREE CHARACTERS OR A VECTOR OF THREE CHARACTERS, MADE OF \"A\", \"C\", \"G\", \"T\" ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end other checkings
-# reserved word checking
-# end reserved word checking
-# end second round of checking and data preparation
-# package checking
-# end package checking
-# main code
-# standard genetic code
-sgc <- array(
-c(
-"F", "L", "I", "V",
-"S", "P", "T", "A",
-"Y", "H", "N", "D",
-"C", "R", "S", "G",
-
-"F", "L", "I", "V",
-"S", "P", "T", "A",
-"Y", "H", "N", "D",
-"C", "R", "S", "G",
-
-"L", "L", "I", "V",
-"S", "P", "T", "A",
-"stop", "Q", "K", "E",
-"stop", "R", "R", "G",
-
-"L", "L", "M", "V",
-"S", "P", "T", "A",
-"stop", "Q", "K", "E",
-"W", "R", "R", "G"
-),
-dim = c(4, 4, 4),
-dimnames = list(
-first = c("T", "C", "A", "G"),
-second = c("T", "C", "A", "G"),
-third = c("T", "C", "A", "G")
-)
-)
-# end standard genetic code
-if(display == TRUE){
-output <- sgc
-}else{
-data <- toupper(data)
-output <- eval(parse(text = paste0("sgc['", paste0(data, collapse = "','"), "']")))
-}
-return(output)
-}
-
-
-######## fun_codon_finder() #### gives the codon number and position in the codon of nucleotid positions
-
-
-fun_codon_finder <- function(
-pos,
-begin,
-end
-){
-# AIM
-# gives the codon number and position in the codon of nucleotid positions
-# WARNINGS
-# Only for coding sequences (no introns): ((end - begin) + 1) / 3 must be an integer (i.e., modulo zero)
-# Negatives positions allowed but this implies that one base has the position 0 in the sequence
-# ARGUMENTS
-# pos: vector of integers indicating the positions of nucleotids in a sequence. Must be between begin and end arguments
-# begin: single integer indicating the position of the first base of the coding sequence
-# end: single indicating the position of the last base of the coding sequence
-# RETURN
-# a data frame with column names:
-# pos: values of the pos argument
-# codon_nb: the codon number in the CDS encompassing the pos value
-# codon_pos: the position of pos in the codon (either 1, 2 or 3)
-# codon_begin: the first base position of the codon
-# codon_end: the last base position of the codon
-# REQUIRED PACKAGES
-# None
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# fun_check()
-# EXAMPLE
-# fun_codon_finder(c(5, 6, 8, 10), begin = 5, end = 10)
-# fun_codon_finder(c(0, 5, 6, 8, 10), begin = -2, end = 12)
-# see http
-# DEBUGGING
-# pos = c(5, 6, 8, 10) ; begin = 5 ; end = 10
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words
-# end reserved words
-# arg with no default values
-mandat.args <- c(
-"pos",
-"begin",
-"end"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = pos, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = begin, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = end, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){ # normally no NA
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"pos",
-"begin",
-"end"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-# end warning initiation
-# other checkings
-if(begin >= end){
-tempo.cat <- paste0("ERROR IN ", function.name, ": end ARGUMENT MUST BE STRICTLY GREATER THAN begin ARGUMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if((end - begin + 1) %% 3 != 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ((end - begin) + 1) / 3 MUST BE AN INTEGER (I.E., MODULO ZERO)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(pos < begin | pos > end)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": pos ARGUMENT VALUES MUST BE BETWEEN begin AND end ARGUMENT VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end other checkings
-# reserved word checking
-# end reserved word checking
-# end second round of checking and data preparation
-# package checking
-# end package checking
-# main code
-first <- seq.int(from = begin, to = end, by = 3)
-last <- seq.int(from = begin + 2, to = end, by = 3)
-tempo <- lapply(X = pos, FUN = function(x = X){
-tempo.log <- x >= first & x <= last
-if(sum(tempo.log, na.rm = TRUE) != 1){ # check that 1 possible TRUE
-tempo.cat <- paste0("ERROR IN ", function.name, ": INTERNAL ERROR. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-codon_nb <- which(tempo.log)
-codon_pos <- as.integer((x - (begin + (codon_nb - 1) * 3) + 1))
-codon_begin <- as.integer(first[tempo.log])
-codon_end <- as.integer(last[tempo.log])
-}
-return(data.frame(codon_nb = codon_nb, codon_pos = codon_pos, codon_begin = codon_begin, codon_end = codon_end))
-})
-tempo <- do.call("rbind", tempo)
-output <- data.frame(pos = as.integer(pos), tempo)
-return(output)
-}
-
-
-################ Graphics management
-
-
-# this order can be used:
-# fun_width()
-# fun_open()
-# fun_prior_plot() # not for ggplot2
-# plot() or any other plotting
-# fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
-# fun_close()
-
-
-######## fun_width() #### window width depending on classes to plot
-
-
-fun_width <- function(
-class.nb,
-inches.per.class.nb = 1,
-ini.window.width = 7,
-inch.left.space,
-inch.right.space,
-boundarie.space = 0.5
-){
-# AIM
-# rescale the width of a window to open depending on the number of classes to plot
-# can be used for height, considering that it is as if it was a width
-# this order can be used:
-# fun_width()
-# fun_open()
-# fun_prior_plot() # not for ggplot2
-# plot() or any other plotting
-# fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
-# fun_close()
-# ARGUMENTS
-# class.nb: number of class to plot
-# inches.per.class.nb: number of inches per unit of class.nb. 2 means 2 inches for each boxplot for instance
-# ini.window.width:initial window width in inches
-# inch.left.space: left horizontal margin of the figure region (in inches)
-# inch.right.space: right horizontal margin of the figure region (in inches)
-# boundarie.space: space between the right and left limits of the plotting region and the plot (0.5 means half a class width)
-# RETURN
-# the new window width in inches
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_width(class.nb = 10, inches.per.class.nb = 0.2, ini.window.width = 7, inch.left.space = 1, inch.right.space = 1, boundarie.space = 0.5)
-# DEBUGGING
-# class.nb = 10 ; inches.per.class.nb = 0.2 ; ini.window.width = 7 ; inch.left.space = 1 ; inch.right.space = 1 ; boundarie.space = 0.5 # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = class.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = inches.per.class.nb, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = ini.window.width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = inch.left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = inch.right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = boundarie.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-range.max <- class.nb + boundarie.space # the max range of the future plot
-range.min <- boundarie.space # the min range of the future plot
-window.width <- inch.left.space + inch.right.space + inches.per.class.nb * (range.max - range.min)
-return(window.width)
-}
-
-
-######## fun_open() #### open a GUI or pdf graphic window
-
-
-fun_open <- function(
-pdf = TRUE,
-pdf.path = "working.dir",
-pdf.name = "graph",
-width = 7,
-height = 7,
-paper = "special",
-pdf.overwrite = FALSE,
-rescale = "fixed",
-remove.read.only = TRUE,
-return.output = FALSE
-){
-# AIM
-# open a pdf or screen (GUI) graphic window and return initial graphic parameters
-# this order can be used:
-# fun_width()
-# fun_open()
-# fun_prior_plot() # not for ggplot2
-# plot() or any other plotting
-# fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
-# fun_close()
-# WARNINGS
-# On Linux, use pdf = TRUE, if (GUI) graphic window is not always available, meaning that X is not installed (clusters for instance). Use X11() in R to test if available
-# ARGUMENTS:
-# pdf: logical. Use pdf display? If FALSE, a GUI is opened
-# pdf.path: where the pdf is saved (do not terminate by / or \\). Write "working.dir" if working directory is required (default). Ignored if pdf == FALSE
-# pdf.name: name of the pdf file containing the graphs (the .pdf extension is added by the function, if not detected in the name end). Ignored if pdf == FALSE
-# width: width of the window (in inches)
-# height: height of the window (in inches)
-# paper: paper argument of the pdf function (paper format). Only used for pdf(). Either "a4", "letter", "legal", "us", "executive", "a4r", "USr" or "special". If "special", means that the paper dimension will be width and height. With another paper format, if width or height is over the size of the paper, width or height will be modified such that the plot is adjusted to the paper dimension (see $dim in the returned list below to see the modified dimensions). Ignored if pdf == FALSE
-# pdf.overwrite: logical. Existing pdf can be overwritten? . Ignored if pdf == FALSE
-# rescale: kind of GUI. Either "R", "fit", or "fixed". Ignored on Mac and Linux OS. See ?windows for details
-# remove.read.only: logical. remove the read only (R.O.) graphical parameters? If TRUE, the graphical parameters are returned without the R.O. parameters. The returned $ini.par list can be used to set the par() of a new graphical device. If FALSE, graphical parameters are returned with the R.O. parameters, which provides information like text dimension (see ?par() ). The returned $ini.par list can be used to set the par() of a new graphical device, but generate a warning message. Ignored if return.output == FALSE.
-# return.output: logical. Return output ? If TRUE the output list is displayed
-# RETURN
-# a list containing:
-# $pdf.loc: path of the pdf created
-# $ini.par: initial par() parameters
-# $zone.ini: initial window spliting
-# $dim: dimension of the graphical device (in inches)
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_open(pdf = FALSE, pdf.path = "C:/Users/Gael/Desktop", pdf.name = "graph", width = 7, height = 7, paper = "special", pdf.overwrite = FALSE, return.output = TRUE)
-# DEBUGGING
-# pdf = TRUE ; pdf.path = "C:/Users/Gael/Desktop" ; pdf.name = "graphs" ; width = 7 ; height = 7 ; paper = "special" ; pdf.overwrite = FALSE ; rescale = "fixed" ; remove.read.only = TRUE ; return.output = TRUE # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = pdf, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = pdf.path, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = pdf.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = height, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = paper, options = c("a4", "letter", "legal", "us", "executive", "a4r", "USr", "special", "A4", "LETTER", "LEGAL", "US"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data =pdf.overwrite, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = rescale, options = c("R", "fit", "fixed"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = remove.read.only, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.output, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if(pdf.path == "working.dir"){
-pdf.path <- getwd()
-}else{
-if(grepl(x = pdf.path, pattern = ".+/$")){
-pdf.path <- sub(x = pdf.path, pattern = "/$", replacement = "") # remove the last /
-}else if(grepl(x = pdf.path, pattern = ".+[\\]$")){ # or ".+\\\\$" # cannot be ".+\$" because \$ does not exist contrary to \n
-pdf.path <- sub(x = pdf.path, pattern = "[\\]$", replacement = "") # remove the last /
-}
-if(dir.exists(pdf.path) == FALSE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\npdf.path ARGUMENT DOES NOT CORRESPOND TO EXISTING DIRECTORY\n", pdf.path)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# par.ini recovery
-# cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
-if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
-open.fail <- NULL
-grDevices::windows()
-ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-}else if(Sys.info()["sysname"] == "Linux"){
-if(pdf == TRUE){# cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
-if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
-tempo.random.seed <- .Random.seed
-on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
-}else{
-on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
-}
-set.seed(NULL)
-tempo.code <- sample(x = 1:1e7, size = 1)
-while(file.exists(paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf")) == TRUE){
-tempo.code <- tempo.code + 1
-}
-grDevices::pdf(width = width, height = height, file=paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf"), paper = paper)
-ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the pdf window
-file.remove(paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf")) # remove the pdf file
-}else{
-# test if X11 can be opened
-if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHIS FUNCTION CANNOT BE USED ON LINUX IF A Rplots.pdf FILE ALREADY EXISTS HERE\n", getwd())
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-open.fail <- suppressWarnings(try(grDevices::X11(), silent = TRUE))[] # try to open a X11 window. If open.fail == NULL, no problem, meaning that the X11 window is opened. If open.fail != NULL, a pdf can be opened here paste0(getwd(), "/Rplots.pdf")
-if(is.null(open.fail)){
-ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-}else if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
-file.remove(paste0(getwd(), "/Rplots.pdf")) # remove the pdf file
-tempo.cat <- ("ERROR IN fun_open()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, EITHER SET THE X GRAPHIC INTERFACE OF THE SYSTEM OR SET THE pdf ARGUMENT OF THE fun_open() FUNCTION TO TRUE AND RERUN")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}
-}else{
-open.fail <- NULL
-grDevices::quartz()
-ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-}
-# end par.ini recovery
-zone.ini <- matrix(1, ncol=1) # to recover the initial parameters for next figure region when device region split into several figure regions
-if(pdf == TRUE){
-if(grepl(x = pdf.name, pattern = "\\.pdf$")){
-pdf.name <- sub(x = pdf.name, pattern = "\\.pdf$", replacement = "") # remove the last .pdf
-}
-pdf.loc <- paste0(pdf.path, "/", pdf.name, ".pdf")
-if(file.exists(pdf.loc) == TRUE & pdf.overwrite == FALSE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\npdf.loc FILE ALREADY EXISTS AND CANNOT BE OVERWRITTEN DUE TO pdf.overwrite ARGUMENT SET TO TRUE\n", pdf.loc)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-grDevices::pdf(width = width, height = height, file=pdf.loc, paper = paper)
-}
-}else if(pdf == FALSE){
-pdf.loc <- NULL
-if(Sys.info()["sysname"] == "Windows"){ # .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
-grDevices::windows(width = width, height = height, rescale = rescale)
-}else if(Sys.info()["sysname"] == "Linux"){
-if( ! is.null(open.fail)){
-tempo.cat <- "ERROR IN fun_open()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, EITHER SET THE X GRAPHIC INTERFACE OF THE SYSTEM OR SET THE pdf ARGUMENT OF THE fun_open() FUNCTION TO TRUE AND RERUN"
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-grDevices::X11(width = width, height = height)
-}
-}else{
-grDevices::quartz(width = width, height = height)
-}
-}
-if(return.output == TRUE){
-output <- list(pdf.loc = pdf.loc, ini.par = ini.par, zone.ini = zone.ini, dim = dev.size())
-return(output)
-}
-}
-
-
-######## fun_prior_plot() #### set graph param before plotting (erase axes for instance)
-
-
-fun_prior_plot <- function(
-param.reinitial = FALSE,
-xlog.scale = FALSE,
-ylog.scale = FALSE,
-remove.label = TRUE,
-remove.x.axis = TRUE,
-remove.y.axis = TRUE,
-std.x.range = TRUE,
-std.y.range = TRUE,
-down.space = 1,
-left.space = 1,
-up.space = 1,
-right.space = 1,
-orient = 1,
-dist.legend = 3.5,
-tick.length = 0.5,
-box.type = "n",
-amplif.label = 1,
-amplif.axis = 1,
-display.extend = FALSE,
-return.par = FALSE
-){
-# AIM
-# very convenient to erase the axes for post plot axis redrawing using fun_post_plot()
-# reinitialize and set the graphic parameters before plotting
-# CANNOT be used if no graphic device already opened
-# ARGUMENTS
-# param.reinitial: reinitialize graphic parameters before applying the new ones, as defined by the other arguments? Either TRUE or FALSE
-# xlog.scale: Log scale for the x-axis? Either TRUE or FALSE. If TRUE, erases the x-axis, except legend, for further drawing by fun_post_plot()(xlog argument of par())
-# ylog.scale: Log scale for the y-axis? Either TRUE or FALSE. If TRUE, erases the y-axis, except legend, for further drawing by fun_post_plot()(ylog argument of par())
-# remove.label: remove labels (axis legend) of the two axes? Either TRUE or FALSE (ann argument of par())
-# remove.x.axis: remove x-axis except legend? Either TRUE or FALSE (control the xaxt argument of par()). Automately set to TRUE if xlog.scale == TRUE
-# remove.y.axis: remove y-axis except legend? Either TRUE or FALSE (control the yaxt argument of par()). Automately set to TRUE if ylog.scale == TRUE
-# std.x.range: standard range on the x-axis? TRUE (no range extend) or FALSE (4% range extend). Controls xaxs argument of par() (TRUE is xaxs = "i", FALSE is xaxs = "r")
-# std.y.range: standard range on the y-axis? TRUE (no range extend) or FALSE (4% range extend). Controls yaxs argument of par() (TRUE is yaxs = "i", FALSE is yaxs = "r")
-# down.space: lower vertical margin (in inches, mai argument of par())
-# left.space: left horizontal margin (in inches, mai argument of par())
-# up.space: upper vertical margin between plot region and grapical window (in inches, mai argument of par())
-# right.space: right horizontal margin (in inches, mai argument of par())
-# orient: scale number orientation (las argument of par()). 0, always parallel to the axis; 1, always horizontal; 2, always perpendicular to the axis; 3, always vertical
-# dist.legend: numeric value that moves axis legends away in inches (first number of mgp argument of par() but in inches thus / 0.2)
-# tick.length: length of the ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc. 0 means no tick
-# box.type: bty argument of par(). Either "o", "l", "7", "c", "u", "]", the resulting box resembles the corresponding upper case letter. A value of "n" suppresses the box
-# amplif.label: increase or decrease the size of the text in legends
-# amplif.axis: increase or decrease the size of the scale numbers in axis
-# display.extend: extend display beyond plotting region? Either TRUE or FALSE (xpd argument of par() without NA)
-# return.par: return graphic parameter modification?
-# RETURN
-# return graphic parameter modification
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_prior_plot(param.reinitial = FALSE, xlog.scale = FALSE, ylog.scale = FALSE, remove.label = TRUE, remove.x.axis = TRUE, remove.y.axis = TRUE, std.x.range = TRUE, std.y.range = TRUE, down.space = 1, left.space = 1, up.space = 1, right.space = 1, orient = 1, dist.legend = 4.5, tick.length = 0.5, box.type = "n", amplif.label = 1, amplif.axis = 1, display.extend = FALSE, return.par = FALSE)
-# DEBUGGING
-# param.reinitial = FALSE ; xlog.scale = FALSE ; ylog.scale = FALSE ; remove.label = TRUE ; remove.x.axis = TRUE ; remove.y.axis = TRUE ; std.x.range = TRUE ; std.y.range = TRUE ; down.space = 1 ; left.space = 1 ; up.space = 1 ; right.space = 1 ; orient = 1 ; dist.legend = 4.5 ; tick.length = 0.5 ; box.type = "n" ; amplif.label = 1 ; amplif.axis = 1 ; display.extend = FALSE ; return.par = FALSE # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = param.reinitial, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = xlog.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = ylog.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = remove.label, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = remove.x.axis, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = remove.y.axis, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = std.x.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = std.y.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = down.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = up.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = orient, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.type, options = c("o", "l", "7", "c", "u", "]", "n"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = amplif.label, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = amplif.axis, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = display.extend, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.par, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if(is.null(dev.list())){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THIS FUNCTION CANNOT BE USED IF NO GRAPHIC DEVICE ALREADY OPENED (dev.list() IS CURRENTLY NULL)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# par.ini recovery
-# cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
-if(param.reinitial == TRUE){
-if( ! all(names(dev.cur()) == "null device")){
-active.wind.nb <- dev.cur()
-}else{
-active.wind.nb <- 0
-}
-if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
-grDevices::windows()
-ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-}else if(Sys.info()["sysname"] == "Linux"){
-if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THIS FUNCTION CANNOT BE USED ON LINUX WITH param.reinitial SET TO TRUE IF A Rplots.pdf FILE ALREADY EXISTS HERE: ", getwd())
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-open.fail <- suppressWarnings(try(grDevices::X11(), silent = TRUE))[] # try to open a X11 window. If open.fail == NULL, no problem, meaning that the X11 window is opened. If open.fail != NULL, a pdf can be opened here paste0(getwd(), "/Rplots.pdf")
-if(is.null(open.fail)){
-ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-}else if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
-ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-invisible(dev.off()) # close the new window
-file.remove(paste0(getwd(), "/Rplots.pdf")) # remove the pdf file
-}else{
-tempo.cat <- ("ERROR IN fun_prior_plot()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, PLEASE USE A PDF GRAPHIC INTERFACE AND RERUN")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}else{ # macOS
-grDevices::quartz()
-ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened)
-invisible(dev.off()) # close the new window
-}
-if( ! all(names(dev.cur()) == "null device")){
-invisible(dev.set(active.wind.nb)) # go back to the active window if exists
-par(ini.par) # apply the initial par to current window
-}
-}
-# end par.ini recovery
-if(remove.x.axis == TRUE){
-par(xaxt = "n") # suppress the y-axis label
-}else{
-par(xaxt = "s")
-}
-if(remove.y.axis == TRUE){
-par(yaxt = "n") # suppress the y-axis label
-}else{
-par(yaxt = "s")
-}
-if(std.x.range == TRUE){
-par(xaxs = "i")
-}else{
-par(xaxs = "r")
-}
-if(std.y.range == TRUE){
-par(yaxs = "i")
-}else{
-par(yaxs = "r")
-}
-par(mai = c(down.space, left.space, up.space, right.space), ann = ! remove.label, las = orient, mgp = c(dist.legend/0.2, 1, 0), xpd = display.extend, bty= box.type, cex.lab = amplif.label, cex.axis = amplif.axis)
-par(tcl = -par()$mgp[2] * tick.length) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
-if(xlog.scale == TRUE){
-par(xaxt = "n", xlog = TRUE) # suppress the x-axis label
-}else{
-par(xlog = FALSE)
-}
-if(ylog.scale == TRUE){
-par(yaxt = "n", ylog = TRUE) # suppress the y-axis label
-}else{
-par(ylog = FALSE)
-}
-if(return.par == TRUE){
-tempo.par <- par()
-return(tempo.par)
-}
-}
-
-
-######## fun_scale() #### select nice label numbers when setting number of ticks on an axis
-
-
-
-
-
-fun_scale <- function(n, lim, kind = "approx", lib.path = NULL){
-# AIM
-# attempt to select nice scale numbers when setting n ticks on a lim axis range
-# ARGUMENTS
-# n: desired number of main ticks on the axis (integer above 0)
-# lim: vector of 2 numbers indicating the limit range of the axis. Order of the 2 values matters (for inverted axis). Can be log transformed values
-# kind: either "approx" (approximative), "strict" (strict) or "strict.cl" (strict clean). If "approx", use the scales::trans_breaks() function to provide an easy to read scale of approximately n ticks spanning the range of the lim argument. If "strict", cut the range of the lim argument into n + 1 equidistant part and return the n numbers at each boundary. This often generates numbers uneasy to read. If "strict.cl", provide an easy to read scale of exactly n ticks, but sometimes not completely spanning the range of the lim argument
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# RETURN
-# a vector of numbers
-# REQUIRED PACKAGES
-# if kind = "approx":
-# ggplot2
-# scales
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_round()
-# EXAMPLES
-# approximate number of main ticks
-# ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "approx") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
-# strict number of main ticks
-# ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "strict") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
-# strict "clean" number of main ticks
-# ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "strict.cl") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
-# approximate number of main ticks, scale inversion
-# ymin = 3.101 ; ymax = 2 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "approx") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
-# DEBUGGING
-# n = 9 ; lim = c(2, 3.101) ; kind = "approx" ; lib.path = NULL # for function debugging
-# n = 10 ; lim = c(1e-4, 1e6) ; kind = "approx" ; lib.path = NULL # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# end initial argument checking
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_round", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_round() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & isTRUE(all.equal(n, 0))){ # isTRUE(all.equal(n, 0)) equivalent to n == 0 but deals with floats (approx ok)
-tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A NON NULL AND POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE) #
-}
-tempo <- fun_check(data = lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & all(diff(lim) == 0L)){ # isTRUE(all.equal(diff(lim), rep(0, length(diff(lim))))) not used because we strictly need zero as a result
-tempo.cat <- paste0("ERROR IN ", function.name, ": lim ARGUMENT HAS A NULL RANGE (2 IDENTICAL VALUES)")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & any(lim %in% c(Inf, -Inf))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = kind, options = c("approx", "strict", "strict.cl"), length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-lim.rank <- rank(lim) # to deal with inverted axis
-lim <- sort(lim)
-if(kind == "approx"){
-# package checking
-fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
-fun_pack(req.package = c("scales"), lib.path = lib.path)
-# end package checking
-output <- ggplot2::ggplot_build(ggplot2::ggplot() + ggplot2::scale_y_continuous(
-breaks = scales::trans_breaks(
-trans = "identity",
-inv = "identity",
-n = n
-),
-limits = lim
-))$layout$panel_params[[1]]$y$breaks # pretty() alone is not appropriate: tempo.pret <- pretty(seq(lim[1] ,lim[2], length.out = n)) ; tempo.pret[tempo.pret > = lim[1] & tempo.pret < = lim[2]]. # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks
-if( ! is.null(attributes(output))){ # layout$panel_params[[1]]$y$breaks can be characters (labels of the axis). In that case, it has attributes that corresponds to positions
-output <- unlist(attributes(output))
-}
-output <- output[ ! is.na(output)]
-}else if(kind == "strict"){
-output <- fun_round(seq(lim[1] ,lim[2], length.out = n), 2)
-}else if(kind == "strict.cl"){
-tempo.range <- diff(sort(lim))
-tempo.max <- max(lim)
-tempo.min <- min(lim)
-mid <- tempo.min + (tempo.range/2) # middle of axis
-tempo.inter <- tempo.range / (n + 1) # current interval between two ticks, between 0 and Inf
-if(tempo.inter == 0L){ # isTRUE(all.equal(tempo.inter, rep(0, length(tempo.inter)))) not used because we strictly need zero as a result
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE INTERVAL BETWEEN TWO TICKS OF THE SCALE IS NULL. MODIFY THE lim OR n ARGUMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-log10.abs.lim <- 200
-log10.range <- (-log10.abs.lim):log10.abs.lim
-log10.vec <- 10^log10.range
-round.vec <- c(5, 4, 3, 2.5, 2, 1.25, 1)
-dec.table <- outer(log10.vec, round.vec) # table containing the scale units (row: power of ten from -201 to +199, column: the 5, 2.5, 2, 1.25, 1 notches
-
-
-
-# recover the number of leading zeros in tempo.inter
-ini.scipen <- options()$scipen
-options(scipen = -1000) # force scientific format
-if(any(grepl(pattern = "\\+", x = tempo.inter))){ # tempo.inter > 1
-power10.exp <- as.integer(substring(text = tempo.inter, first = (regexpr(pattern = "\\+", text = tempo.inter) + 1))) # recover the power of 10. Example recover 08 from 1e+08
-mantisse <- as.numeric(substr(x = tempo.inter, start = 1, stop = (regexpr(pattern = "\\+", text = tempo.inter) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
-}else if(any(grepl(pattern = "\\-", x = tempo.inter))){ # tempo.inter < 1
-power10.exp <- as.integer(substring(text = tempo.inter, first = (regexpr(pattern = "\\-", text = tempo.inter)))) # recover the power of 10. Example recover 08 from 1e+08
-mantisse <- as.numeric(substr(x = tempo.inter, start = 1, stop = (regexpr(pattern = "\\-", text = tempo.inter) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-tempo.scale <- dec.table[log10.range == power10.exp, ]
-# new interval
-inter.select <- NULL
-for(i1 in 1:length(tempo.scale)){
-tempo.first.tick <- trunc((tempo.min + tempo.scale[i1]) / tempo.scale[i1]) * (tempo.scale[i1]) # this would be use to have a number not multiple of tempo.scale[i1]: ceiling(tempo.min) + tempo.scale[i1] * 10^power10.exp
-tempo.last.tick <- tempo.first.tick + tempo.scale[i1] * (n - 1)
-if((tempo.first.tick >= tempo.min) & (tempo.last.tick <= tempo.max)){
-inter.select <- tempo.scale[i1]
-break()
-}
-}
-if(is.null(inter.select)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-options(scipen = ini.scipen) # restore the initial scientific penalty
-# end new interval
-# centering the new scale
-tempo.mid <- trunc((mid + (-1:1) * inter.select) / inter.select) * inter.select # tempo middle tick closest to the middle axis
-mid.tick <- tempo.mid[which.min(abs(tempo.mid - mid))]
-if(isTRUE(all.equal(n, rep(1, length(n))))){ # isTRUE(all.equal(n, rep(1, length(n)))) is similar to n == 1L but deals with float
-output <- mid.tick
-}else if(isTRUE(all.equal(n, rep(2, length(n))))){ # isTRUE(all.equal(n, rep(0, length(n)))) is similar to n == 2L but deals with float
-output <- mid.tick
-tempo.min.dist <- mid.tick - inter.select - tempo.min
-tempo.max.dist <- tempo.max - mid.tick + inter.select
-if(tempo.min.dist <= tempo.max.dist){ # distance between lowest tick and bottom axis <= distance between highest tick and top axis. If yes, extra tick but at the top, otherwise at the bottom
-output <- c(mid.tick, mid.tick + inter.select)
-}else{
-output <- c(mid.tick - inter.select, mid.tick)
-}
-}else if((n / 2 - trunc(n / 2)) > 0.1){ # > 0.1 to avoid floating point. Because result can only be 0 or 0.5. Thus, > 0.1 means odd number
-output <- c(mid.tick - (trunc(n / 2):1) * inter.select, mid.tick, mid.tick + (1:trunc(n / 2)) * inter.select)
-}else if((n / 2 - trunc(n / 2)) < 0.1){ # < 0.1 to avoid floating point. Because result can only be 0 or 0.5. Thus, < 0.1 means even number
-tempo.min.dist <- mid.tick - trunc(n / 2) * inter.select - tempo.min
-tempo.max.dist <- tempo.max - mid.tick + trunc(n / 2) * inter.select
-if(tempo.min.dist <= tempo.max.dist){ # distance between lowest tick and bottom axis <= distance between highest tick and top axis. If yes, extra tick but at the bottom, otherwise at the top
-output <- c(mid.tick - ((trunc(n / 2) - 1):1) * inter.select, mid.tick, mid.tick + (1:trunc(n / 2)) * inter.select)
-}else{
-output <- c(mid.tick - (trunc(n / 2):1) * inter.select, mid.tick, mid.tick + (1:(trunc(n / 2) - 1)) * inter.select)
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end centering the new scale
-# last check
-if(min(output) < tempo.min){
-output <- c(output[-1], max(output) + inter.select) # remove the lowest tick and add a tick at the top
-}else if( max(output) > tempo.max){
-output <- c(min(output) - inter.select, output[-length(output)])
-}
-if(min(output) < tempo.min | max(output) > tempo.max){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(is.na(output))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5 (NA GENERATION)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end last check
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 6")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(diff(lim.rank) < 0){
-output <- rev(output)
-}
-return(output)
-}
-
-
-######## fun_inter_ticks() #### define coordinates of secondary ticks
-
-
-fun_inter_ticks <- function(
-lim,
-log = "log10",
-breaks = NULL,
-n = NULL,
-warn.print = TRUE
-){
-# AIM
-# define coordinates and values of secondary ticks
-# ARGUMENTS
-# lim: vector of 2 numbers indicating the limit range of the axis. Order of the 2 values matters (for inverted axis). If log argument is "log2" or "log10", values in lim must be already log transformed. Thus, negative or zero values are allowed
-# log: either "log2" (values in the lim argument are log2 transformed) or "log10" (values in the lim argument are log10 transformed), or "no"
-# breaks: mandatory vector of numbers indicating the main ticks values/positions when log argument is "no". Ignored when log argument is "log2" or "log10"
-# n: number of secondary ticks between each main tick when log argument is "no". Ignored when log argument is "log2" or "log10"
-# warn.print: logical. Print potential warning messages at the end of the execution? If FALSE, warning messages are never printed, but can still be recovered in the returned list
-# RETURN
-# a list containing
-# $log: value of the log argument used
-# $coordinates: the coordinates of the secondary ticks on the axis, between the lim values
-# $values: the corresponding values associated to each coordinate (with log scale, 2^$values or 10^$values is equivalent to the labels of the axis)
-# $warn: the potential warning messages. Use cat() for proper display. NULL if no warning
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# no log scale
-# fun_inter_ticks(lim = c(-4,4), log = "no", breaks = c(-2, 0, 2), n = 3)
-# fun_inter_ticks(lim = c(10, 0), log = "no", breaks = c(10, 8, 6, 4, 2, 0), n = 4)
-# log2
-# fun_inter_ticks(lim = c(-4,4), log = "log2")
-# log10
-# fun_inter_ticks(lim = c(-2,3), log = "log10")
-# DEBUGGING
-# lim = c(2, 3.101) ; log = "no" ; breaks = NULL ; n = NULL ; warn.print = TRUE # for function debugging
-# lim = c(0, 26.5) ; log = "no" ; breaks = c(0, 10, 20) ; n = 3 # for function debugging
-# lim = c(10, 0); log = "no"; breaks = c(10, 8, 6, 4, 2, 0); n = 4 # for function debugging
-# lim = c(-10, -20); log = "no"; breaks = c(-20, -15, -10); n = 4 # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-req.function <- c(
-"fun_check"
-)
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED ", i1, "() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end required function checking
-# argument primary checking
-# arg with no default values
-mandat.args <- c(
-"lim"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# using fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(breaks)){
-tempo <- fun_check(data = breaks, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(n)){
-tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = warn.print, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA
-if(any(is.na(lim)) | any(is.na(log)) | any(is.na(breaks)) | any(is.na(n)) | any(is.na(warn.print))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nNO ARGUMENT CAN HAVE NA VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA
-# management of NULL
-if(is.null(lim) | is.null(log)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHESE ARGUMENTS\nlim\nlog\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL
-if(all(diff(lim) == 0L)){ # isTRUE(all.equal(diff(lim), rep(0, length(diff(lim))))) not used because we strictly need zero as a result
-tempo.cat <- paste0("ERROR IN ", function.name, "\nlim ARGUMENT HAS A NULL RANGE (2 IDENTICAL VALUES): ", paste(lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(any(lim %in% c(Inf, -Inf))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nlim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(log == "no" & is.null(breaks)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT CANNOT BE NULL IF log ARGUMENT IS \"no\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(breaks)){
-if(length(breaks) < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT MUST HAVE 2 VALUES AT LEAST (OTHERWISE, INTER TICK POSITIONS CANNOT BE COMPUTED): ", paste(breaks, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! isTRUE(all.equal(diff(sort(breaks)), rep(diff(sort(breaks))[1], length(diff(sort(breaks))))))){ # isTRUE(all.equal(n, 0)) equivalent to n == 0 but deals with floats (approx ok)
-tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT MUST HAVE EQUIDISTANT VALUES (OTHERWISE, EQUAL NUMBER OF INTER TICK BETWEEN MAIN TICKS CANNOT BE COMPUTED): ", paste(breaks, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if( ! is.null(n)){
-if(n <= 0){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nn ARGUMENT MUST BE A POSITIVE AND NON NULL INTEGER: ", paste(n, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end second round of checking and data preparation
-# main code
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-lim.rank <- rank(lim) # to deal with inverse axis
-if(log != "no"){
-ini.scipen <- options()$scipen
-options(scipen = -1000) # force scientific format
-power10.exp <- as.integer(substring(text = 10^lim, first = (regexpr(pattern = "\\+|\\-", text = 10^lim)))) # recover the power of 10, i.e., integer part of lim. Example recover 08 from 1e+08. Works for log2
-# mantisse <- as.numeric(substr(x = 10^lim, start = 1, stop = (regexpr(pattern = "\\+|\\-", text = 10^lim) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
-options(scipen = ini.scipen) # restore the initial scientific penalty
-tick.pos <- unique(as.vector(outer(2:10, ifelse(log == "log2", 2, 10)^((power10.exp[1] - ifelse(diff(lim.rank) > 0, 1, -1)):(power10.exp[2] + ifelse(diff(lim.rank) > 0, 1, -1)))))) # use log10(2:10) even if log2: it is to get log values between 0 and 1
-tick.pos <- sort(tick.pos, decreasing = ifelse(diff(lim.rank) > 0, FALSE, TRUE))
-if(log == "log2"){
-tick.values <- tick.pos[tick.pos >= min(2^lim) & tick.pos <= max(2^lim)]
-tick.pos <- log2(tick.values)
-}else if(log == "log10"){
-tick.values <- tick.pos[tick.pos >= min(10^lim) & tick.pos <= max(10^lim)]
-tick.pos <- log10(tick.values)
-}
-}else{
-# if(length(breaks) > 1){ # not required because already checked above
-breaks.rank <- rank(c(breaks[1], breaks[length(breaks)]))
-if(diff(breaks.rank) != diff(lim.rank)){
-breaks <- sort(breaks, decreasing = ifelse(diff(lim.rank) < 0, TRUE, FALSE))
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") VALUES IN breaks ARGUMENT NOT IN THE SAME ORDER AS IN lim ARGUMENT -> VALUES REORDERED AS IN lim: ", paste(breaks, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-breaks.rank <- rank(c(breaks[1], breaks[length(breaks)]))
-}
-# }
-main.tick.dist <- mean(diff(breaks), na.rm = TRUE)
-tick.dist <- main.tick.dist / (n + 1)
-tempo.extra.margin <- max(abs(diff(breaks)), na.rm = TRUE)
-tick.pos <- seq(
-if(diff(breaks.rank) > 0){breaks[1] - tempo.extra.margin}else{breaks[1] + tempo.extra.margin},
-if(diff(breaks.rank) > 0){breaks[length(breaks)] + tempo.extra.margin}else{breaks[length(breaks)] - tempo.extra.margin},
-by = tick.dist
-)
-tick.pos <- tick.pos[tick.pos >= min(lim) & tick.pos <= max(lim)]
-tick.values <- tick.pos
-}
-if(any(is.na(tick.pos) | ! is.finite(tick.pos))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, ": NA or Inf GENERATED FOR THE INTER TICK POSITIONS: ", paste(tick.pos, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(tick.pos) == 0L){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO INTER TICKS COMPUTED BETWEEN THE LIMITS INDICATED: ", paste(lim, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-output <- list(log = log, coordinates = tick.pos, values = tick.values, warn = warn)
-if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE)) # to recover the warning messages, see $warn
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-return(output)
-}
-
-
-######## fun_post_plot() #### set graph param after plotting (axes redesign for instance)
-
-
-
-
-
-fun_post_plot <- function(
-x.side = 0,
-x.log.scale = FALSE,
-x.categ = NULL,
-x.categ.pos = NULL,
-x.lab = "",
-x.axis.size = 1.5,
-x.label.size = 1.5,
-x.dist.legend = 0.5,
-x.nb.inter.tick = 1,
-y.side = 0,
-y.log.scale = FALSE,
-y.categ = NULL,
-y.categ.pos = NULL,
-y.lab = "",
-y.axis.size = 1.5,
-y.label.size = 1.5,
-y.dist.legend = 0.5,
-y.nb.inter.tick = 1,
-text.angle = 90,
-tick.length = 0.5,
-sec.tick.length = 0.3,
-bg.color = NULL,
-grid.lwd = NULL,
-grid.col = "white",
-corner.text = "",
-corner.text.size = 1,
-just.label.add = FALSE,
-par.reset = FALSE,
-custom.par = NULL
-){
-# AIM
-# redesign axis. If x.side = 0, y.side = 0, the function just adds text at topright of the graph and reset par() for next graphics and provides outputs (see below)
-# provide also positions for legend or additional text on the graph
-# use fun_prior_plot() before this function for initial inactivation of the axis drawings
-# ARGUMENTS
-# x.side: axis at the bottom (1) or top (3) of the region figure. Write 0 for no change
-# x.log.scale: Log scale for the x-axis? Either TRUE or FALSE
-# x.categ: character vector representing the classes (levels()) to specify when the x-axis is qualititative(stripchart, boxplot)
-# x.categ.pos: position of the classes names (numeric vector of identical length than x.categ). If left NULL, this will be 1:length(levels())
-# x.lab: label of the x-axis. If x.side == 0 and x.lab != "", then x.lab is printed
-# x.axis.size: positive numeric. Increase or decrease the size of the x axis numbers. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2. Also control the size of displayed categories
-# x.label.size: positive numeric. Increase or decrease the size of the x axis legend text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
-# x.dist.legend: increase the number to move x-axis legends away in inches (first number of mgp argument of par() but in inches)
-# x.nb.inter.tick: number of secondary ticks between main ticks on x-axis (only if not log scale). 0 means no secondary ticks
-# y.side: axis at the left (2) or right (4) of the region figure. Write 0 for no change
-# y.log.scale: Log scale for the y-axis? Either TRUE or FALSE
-# y.categ: classes (levels()) to specify when the y-axis is qualititative(stripchart, boxplot)
-# y.categ.pos: position of the classes names (numeric vector of identical length than y.categ). If left NULL, this will be 1:length(levels())
-# y.lab: label of the y-axis. If y.side == 0 and y.lab != "", then y.lab is printed
-# y.axis.size: positive numeric. Increase or decrease the size of the y axis numbers. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2. Also control the size of displayed categories
-# y.label.size: positive numeric. Increase or decrease the size of the y axis legend text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
-# y.dist.legend: increase the number to move y-axis legends away in inches (first number of mgp argument of par() but in inches)
-# y.nb.inter.tick: number of secondary ticks between main ticks on y-axis (only if not log scale). 0 means non secondary ticks
-# text.angle: angle of the text when axis is qualitative
-# tick.length: length of the main ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
-# sec.tick.length: length of the secondary ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
-# bg.color: background color of the plot region. NULL for no color. BEWARE: cover/hide an existing plot !
-# grid.lwd: if non NULL, activate the grid line (specify the line width)
-# grid.col: grid line color (only if grid.lwd non NULL)
-# corner.text: text to add at the top right corner of the window
-# corner.text.size: positive numeric. Increase or decrease the size of the text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
-# par.reset: to reset all the graphics parameters. BEWARE: TRUE can generate display problems, mainly in graphic devices with multiple figure regions
-# just.label.add: just add axis labels (legend)? Either TRUE or FALSE. If TRUE, at least (x.side == 0 & x.lab != "") or (y.side == 0 & y.lab != "") must be set to display the corresponding x.lab or y.lab
-# custom.par: list that provides the parameters that reset all the graphics parameters. BEWARE: if NULL and par.reset == TRUE, the default par() parameters are used
-# RETURN
-# a list containing:
-# $x.mid.left.dev.region: middle of the left margin of the device region, in coordinates of the x-axis
-# $x.left.dev.region: left side of the left margin (including the potential margin of the device region), in coordinates of the x-axis
-# $x.mid.right.dev.region: middle of the right margin of the device region, in coordinates of the x-axis
-# $x.right.dev.region: right side of the right margin (including the potential margin of the device region), in coordinates of the x-axis
-# $x.mid.left.fig.region: middle of the left margin of the figure region, in coordinates of the x-axis
-# $x.left.fig.region: left side of the left margin, in coordinates of the x-axis
-# $x.mid.right.fig.region: middle of the right margin of the figure region, in coordinates of the x-axis
-# $x.right.fig.region: right side of the right margin, in coordinates of the x-axis
-# $x.left.plot.region: left side of the plot region, in coordinates of the x-axis
-# $x.right.plot.region: right side of the plot region, in coordinates of the x-axis
-# $x.mid.plot.region: middle of the plot region, in coordinates of the x-axis
-# $y.mid.bottom.dev.region: middle of the bottom margin of the device region, in coordinates of the y-axis
-# $y.bottom.dev.region: bottom side of the bottom margin (including the potential margin of the device region), in coordinates of the y-axis
-# $y.mid.top.dev.region: middle of the top margin of the device region, in coordinates of the y-axis
-# $y.top.dev.region: top side of the top margin (including the potential margin of the device region), in coordinates of the y-axis
-# $y.mid.bottom.fig.region: middle of the bottom margin of the figure region, in coordinates of the y-axis
-# $y.bottom.fig.region: bottom of the bottom margin of the figure region, in coordinates of the y-axis
-# $y.mid.top.fig.region: middle of the top margin of the figure region, in coordinates of the y-axis
-# $y.top.fig.region: top of the top margin of the figure region, in coordinates of the y-axis
-# $y.top.plot.region: top of the plot region, in coordinates of the y-axis
-# $y.bottom.plot.region: bottom of the plot region, in coordinates of the y-axis
-# $y.mid.plot.region: middle of the plot region, in coordinates of the y-axis
-# $text: warning text
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_open() to reinitialize graph parameters if par.reset = TRUE and custom.par = NULL
-# EXAMPLES
-# Example of log axis with log y-axis and unmodified x-axis:
-# prior.par <- fun_prior_plot(param.reinitial = TRUE, xlog.scale = FALSE, ylog.scale = TRUE, remove.label = TRUE, remove.x.axis = FALSE, remove.y.axis = TRUE, down.space = 1, left.space = 1, up.space = 1, right.space = 1, orient = 1, dist.legend = 0.5, tick.length = 0.5, box.type = "n", amplif.label = 1, amplif.axis = 1, display.extend = FALSE, return.par = TRUE) ; plot(1:100, log = "y") ; fun_post_plot(y.side = 2, y.log.scale = prior.par$ylog, x.lab = "Values", y.lab = "TEST", y.axis.size = 1.25, y.label.size = 1.5, y.dist.legend = 0.7, just.label.add = ! prior.par$ann)
-# Example of log axis with redrawn x-axis and y-axis:
-# prior.par <- fun_prior_plot(param.reinitial = TRUE) ; plot(1:100) ; fun_post_plot(x.side = 1, x.lab = "Values", y.side = 2, y.lab = "TEST", y.axis.size = 1, y.label.size = 2, y.dist.legend = 0.6)
-# Example of title easily added to a plot:
-# plot(1:100) ; para <- fun_post_plot(corner.text = "TITLE ADDED") # try also: par(xpd = TRUE) ; text(x = para$x.mid.left.fig.region, y = para$y.mid.top.fig.region, labels = "TITLE ADDED", cex = 0.5)
-# example with margins in the device region:
-# windows(5,5) ; fun_prior_plot(box.type = "o") ; par(mai=c(0.5,0.5,0.5,0.5), omi = c(0.25,0.25,1,0.25), xaxs = "i", yaxs = "i") ; plot(0:10) ; a <- fun_post_plot(x.side = 0, y.side = 0) ; x <- c(a$x.mid.left.dev.region, a$x.left.dev.region, a$x.mid.right.dev.region, a$x.right.dev.region, a$x.mid.left.fig.region, a$x.left.fig.region, a$x.mid.right.fig.region, a$x.right.fig.region, a$x.right.plot.region, a$x.left.plot.region, a$x.mid.plot.region) ; y <- c(a$y.mid.bottom.dev.region, a$y.bottom.dev.region, a$y.mid.top.dev.region, a$y.top.dev.region, a$y.mid.bottom.fig.region, a$y.bottom.fig.region, a$y.mid.top.fig.region, a$y.top.fig.region, a$y.top.plot.region, a$y.bottom.plot.region, a$y.mid.plot.region) ; par(xpd = NA) ; points(x = rep(5, length(y)), y = y, pch = 16, col = "red") ; text(x = rep(5, length(y)), y = y, c("y.mid.bottom.dev.region", "y.bottom.dev.region", "y.mid.top.dev.region", "y.top.dev.region", "y.mid.bottom.fig.region", "y.bottom.fig.region", "y.mid.top.fig.region", "y.top.fig.region", "y.top.plot.region", "y.bottom.plot.region", "y.mid.plot.region"), cex = 0.65, col = grey(0.25)) ; points(y = rep(5, length(x)), x = x, pch = 16, col = "blue") ; text(y = rep(5, length(x)), x = x, c("x.mid.left.dev.region", "x.left.dev.region", "x.mid.right.dev.region", "x.right.dev.region", "x.mid.left.fig.region", "x.left.fig.region", "x.mid.right.fig.region", "x.right.fig.region", "x.right.plot.region", "x.left.plot.region", "x.mid.plot.region"), cex = 0.65, srt = 90, col = grey(0.25))
-# DEBUGGING
-# x.side = 0 ; x.log.scale = FALSE ; x.categ = NULL ; x.categ.pos = NULL ; x.lab = "" ; x.axis.size = 1.5 ; x.label.size = 1.5 ; x.dist.legend = 1 ; x.nb.inter.tick = 1 ; y.side = 0 ; y.log.scale = FALSE ; y.categ = NULL ; y.categ.pos = NULL ; y.lab = "" ; y.axis.size = 1.5 ; y.label.size = 1.5 ; y.dist.legend = 0.7 ; y.nb.inter.tick = 1 ; text.angle = 90 ; tick.length = 0.5 ; sec.tick.length = 0.3 ; bg.color = NULL ; grid.lwd = NULL ; grid.col = "white" ; corner.text = "" ; corner.text.size = 1 ; just.label.add = FALSE ; par.reset = FALSE ; custom.par = NULL # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_open", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_open() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = x.side, options = c(0, 1, 3), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.log.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(x.categ)){
-tempo <- fun_check(data = x.categ, class = "character", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(x.categ.pos)){
-tempo <- fun_check(data = x.categ.pos, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = x.lab, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.axis.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.label.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.nb.inter.tick, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.side, options = c(0, 2, 4), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.log.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(y.categ)){
-tempo <- fun_check(data = y.categ, class = "character", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(y.categ.pos)){
-tempo <- fun_check(data = y.categ.pos, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = y.lab, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.axis.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.label.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.nb.inter.tick, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = text.angle, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = sec.tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-if( ! is.null(bg.color)){
-tempo <- fun_check(data = bg.color, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if( ! (bg.color %in% colors() | grepl(pattern = "^#", bg.color))){ # check color
-tempo.cat <- paste0("ERROR IN ", function.name, ": bg.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # OR A COLOR NAME GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-if( ! is.null(grid.lwd)){
-tempo <- fun_check(data = grid.lwd, class = "vector", mode = "numeric", neg.values = FALSE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(grid.col)){
-tempo <- fun_check(data = grid.col, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if( ! (grid.col %in% colors() | grepl(pattern = "^#", grid.col))){ # check color
-tempo.cat <- paste0("ERROR IN ", function.name, ": grid.col ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # OR A COLOR NAME GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = corner.text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = corner.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = just.label.add, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = par.reset, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(custom.par)){
-tempo <- fun_check(data = custom.par, typeof = "list", length = 1, fun.name = function.name) ; eval(ee)
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-text <- NULL
-par(tcl = -par()$mgp[2] * tick.length)
-if(x.log.scale == TRUE){
-grid.coord.x <- c(10^par("usr")[1], 10^par("usr")[2])
-}else{
-grid.coord.x <- c(par("usr")[1], par("usr")[2])
-}
-if(y.log.scale == TRUE){
-grid.coord.y <- c(10^par("usr")[3], 10^par("usr")[4])
-}else{
-grid.coord.y <- c(par("usr")[3], par("usr")[4])
-}
-if( ! is.null(bg.color)){
-rect(grid.coord.x[1], grid.coord.y[1], grid.coord.x[2], grid.coord.y[2], col = bg.color, border = NA)
-}
-if( ! is.null(grid.lwd)){
-grid(nx = NA, ny = NULL, col = grid.col, lty = 1, lwd = grid.lwd)
-}
-if(x.log.scale == TRUE){
-x.mid.left.dev.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
-x.left.dev.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1]) # in x coordinates
-x.mid.right.dev.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
-x.right.dev.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2])) # in x coordinates
-x.mid.left.fig.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
-x.left.fig.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1]) # in x coordinates
-x.mid.right.fig.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
-x.right.fig.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2])) # in x coordinates
-x.left.plot.region <- 10^par("usr")[1] # in x coordinates, left of the plot region (according to x scale)
-x.right.plot.region <- 10^par("usr")[2] # in x coordinates, right of the plot region (according to x scale)
-x.mid.plot.region <- 10^((par("usr")[2] + par("usr")[1]) / 2) # in x coordinates, right of the plot region (according to x scale)
-}else{
-x.mid.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
-x.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1]) # in x coordinates
-x.mid.right.dev.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
-x.right.dev.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2])) # in x coordinates
-x.mid.left.fig.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
-x.left.fig.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1]) # in x coordinates
-x.mid.right.fig.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
-x.right.fig.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2])) # in x coordinates
-x.left.plot.region <- par("usr")[1] # in x coordinates, left of the plot region (according to x scale)
-x.right.plot.region <- par("usr")[2] # in x coordinates, right of the plot region (according to x scale)
-x.mid.plot.region <- (par("usr")[2] + par("usr")[1]) / 2 # in x coordinates, right of the plot region (according to x scale)
-}
-if(y.log.scale == TRUE){
-y.mid.bottom.dev.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (par("omd")[3] / 2)) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
-y.bottom.dev.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * par("omd")[3]) # in y coordinates
-y.mid.top.dev.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
-y.top.dev.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4])) # in y coordinates
-y.mid.bottom.fig.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] / 2) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
-y.bottom.fig.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3]) # in y coordinates
-y.mid.top.fig.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
-y.top.fig.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4])) # in y coordinates
-y.top.plot.region <- 10^par("usr")[4] # in y coordinates, top of the plot region (according to y scale)
-y.bottom.plot.region <- 10^par("usr")[3] # in y coordinates, bottom of the plot region (according to y scale)
-y.mid.plot.region <- (par("usr")[3] + par("usr")[4]) / 2 # in x coordinates, right of the plot region (according to x scale)
-}else{
-y.mid.bottom.dev.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (par("omd")[3] / 2)) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
-y.bottom.dev.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * par("omd")[3]) # in y coordinates
-y.mid.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
-y.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4])) # in y coordinates
-y.mid.bottom.fig.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] / 2) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
-y.bottom.fig.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3]) # in y coordinates
-y.mid.top.fig.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
-y.top.fig.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4])) # in y coordinates
-y.top.plot.region <- par("usr")[4] # in y coordinates, top of the plot region (according to y scale)
-y.bottom.plot.region <- par("usr")[3] # in y coordinates, bottom of the plot region (according to y scale)
-y.mid.plot.region <- ((par("usr")[3] + par("usr")[4]) / 2) # in x coordinates, right of the plot region (according to x scale)
-}
-if(any(sapply(FUN = all.equal, c(1, 3), x.side) == TRUE)){
-par(xpd=FALSE, xaxt="s")
-if(is.null(x.categ) & x.log.scale == TRUE){
-if(any(par()$xaxp[1:2] == 0L)){ # any(sapply(FUN = all.equal, par()$xaxp[1:2], 0) == TRUE) not used because we strictly need zero as a result. Beware: write "== TRUE", because the result is otherwise character and a warning message appears using any()
-if(par()$xaxp[1] == 0L){ # isTRUE(all.equal(par()$xaxp[1], 0)) not used because we strictly need zero as a result
-par(xaxp = c(10^-30, par()$xaxp[2:3])) # because log10(par()$xaxp[1] == 0) == -Inf
-}
-if(par()$xaxp[2] == 0L){ # isTRUE(all.equal(par()$xaxp[1], 0)) not used because we strictly need zero as a result
-par(xaxp = c(par()$xaxp[1], 10^-30, par()$xaxp[3])) # because log10(par()$xaxp[2] == 0) == -Inf
-}
-}
-axis(side = x.side, at = c(10^par()$usr[1], 10^par()$usr[2]), labels=rep("", 2), lwd=1, lwd.ticks = 0) # draw the axis line
-mtext(side = x.side, text = x.lab, line = x.dist.legend / 0.2, las = 0, cex = x.label.size)
-par(tcl = -par()$mgp[2] * sec.tick.length) # length of the secondary ticks are reduced
-suppressWarnings(rug(10^outer(c((log10(par("xaxp")[1]) -1):log10(par("xaxp")[2])), log10(1:10), "+"), ticksize = NA, side = x.side)) # ticksize = NA to allow the use of par()$tcl value
-par(tcl = -par()$mgp[2] * tick.length) # back to main ticks
-axis(side = x.side, at = c(1e-15, 1e-14, 1e-13, 1e-12, 1e-11, 1e-10, 1e-9, 1e-8, 1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 1e-1, 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10), labels = expression(10^-15, 10^-14, 10^-13, 10^-12, 10^-11, 10^-10, 10^-9, 10^-8, 10^-7, 10^-6, 10^-5, 10^-4, 10^-3, 10^-2, 10^-1, 10^0, 10^1, 10^2, 10^3, 10^4, 10^5, 10^6, 10^7, 10^8, 10^9, 10^10), lwd = 0, lwd.ticks = 1, cex.axis = x.axis.size)
-x.text <- 10^par("usr")[2]
-}else if(is.null(x.categ) & x.log.scale == FALSE){
-axis(side=x.side, at=c(par()$usr[1], par()$usr[2]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
-axis(side=x.side, at=round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), cex.axis = x.axis.size) # axis(side=x.side, at=round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), labels = format(round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), big.mark=','), cex.axis = x.axis.size) # to get the 1000 comma separator
-mtext(side = x.side, text = x.lab, line = x.dist.legend / 0.2, las = 0, cex = x.label.size)
-if(x.nb.inter.tick > 0){
-inter.tick.unit <- (par("xaxp")[2] - par("xaxp")[1]) / par("xaxp")[3]
-par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
-suppressWarnings(rug(seq(par("xaxp")[1] - 10 * inter.tick.unit, par("xaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + x.nb.inter.tick)), ticksize = NA, x.side)) # ticksize = NA to allow the use of par()$tcl value
-par(tcl = -par()$mgp[2] * tick.length) # back to main ticks
-}
-x.text <- par("usr")[2]
-}else if(( ! is.null(x.categ)) & x.log.scale == FALSE){
-if(is.null(x.categ.pos)){
-x.categ.pos <- 1:length(x.categ)
-}else if(length(x.categ.pos) != length(x.categ)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.categ.pos MUST BE THE SAME LENGTH AS x.categ")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-par(xpd = TRUE)
-if(isTRUE(all.equal(x.side, 1))){ #isTRUE(all.equal(x.side, 1)) is similar to x.side == 1L but deals with float
-segments(x0 = x.left.plot.region, x1 = x.right.plot.region, y0 = y.bottom.plot.region, y1 = y.bottom.plot.region) # draw the line of the axis
-text(x = x.categ.pos, y = y.mid.bottom.fig.region, labels = x.categ, srt = text.angle, cex = x.axis.size)
-}else if(isTRUE(all.equal(x.side, 3))){ #isTRUE(all.equal(x.side, 1)) is similar to x.side == 3L but deals with float
-segments(x0 = x.left.plot.region, x1 = x.right.plot.region, y0 = y.top.plot.region, y1 = y.top.plot.region) # draw the line of the axis
-text(x = x.categ.pos, y = y.mid.top.fig.region, labels = x.categ, srt = text.angle, cex = x.axis.size)
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": ARGUMENT x.side CAN ONLY BE 1 OR 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-par(xpd = FALSE)
-x.text <- par("usr")[2]
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": PROBLEM WITH THE x.side (", x.side ,") OR x.log.scale (", x.log.scale,") ARGUMENTS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-x.text <- par("usr")[2]
-}
-if(any(sapply(FUN = all.equal, c(2, 4), y.side) == TRUE)){
-par(xpd=FALSE, yaxt="s")
-if(is.null(y.categ) & y.log.scale == TRUE){
-if(any(par()$yaxp[1:2] == 0L)){ # any(sapply(FUN = all.equal, par()$yaxp[1:2], 0) == TRUE) not used because we strictly need zero as a result. Beware: write "== TRUE", because the result is otherwise character and a warning message appears using any()
-if(par()$yaxp[1] == 0L){ # strict zero needed
-par(yaxp = c(10^-30, par()$yaxp[2:3])) # because log10(par()$yaxp[1] == 0) == -Inf
-}
-if(par()$yaxp[2] == 0L){ # strict zero needed
-par(yaxp = c(par()$yaxp[1], 10^-30, par()$yaxp[3])) # because log10(par()$yaxp[2] == 0) == -Inf
-}
-}
-axis(side=y.side, at=c(10^par()$usr[3], 10^par()$usr[4]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
-par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
-suppressWarnings(rug(10^outer(c((log10(par("yaxp")[1])-1):log10(par("yaxp")[2])), log10(1:10), "+"), ticksize = NA, side = y.side)) # ticksize = NA to allow the use of par()$tcl value
-par(tcl = -par()$mgp[2] * tick.length) # back to main tick length
-axis(side = y.side, at = c(1e-15, 1e-14, 1e-13, 1e-12, 1e-11, 1e-10, 1e-9, 1e-8, 1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 1e-1, 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10), labels = expression(10^-15, 10^-14, 10^-13, 10^-12, 10^-11, 10^-10, 10^-9, 10^-8, 10^-7, 10^-6, 10^-5, 10^-4, 10^-3, 10^-2, 10^-1, 10^0, 10^1, 10^2, 10^3, 10^4, 10^5, 10^6, 10^7, 10^8, 10^9, 10^10), lwd = 0, lwd.ticks = 1, cex.axis = y.axis.size)
-y.text <- 10^(par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
-mtext(side = y.side, text = y.lab, line = y.dist.legend / 0.2, las = 0, cex = y.label.size)
-}else if(is.null(y.categ) & y.log.scale == FALSE){
-axis(side=y.side, at=c(par()$usr[3], par()$usr[4]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
-axis(side=y.side, at=round(seq(par()$yaxp[1], par()$yaxp[2], length.out=par()$yaxp[3]+1), 2), cex.axis = y.axis.size)
-mtext(side = y.side, text = y.lab, line = y.dist.legend / 0.2, las = 0, cex = y.label.size)
-if(y.nb.inter.tick > 0){
-inter.tick.unit <- (par("yaxp")[2] - par("yaxp")[1]) / par("yaxp")[3]
-par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
-suppressWarnings(rug(seq(par("yaxp")[1] - 10 * inter.tick.unit, par("yaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + y.nb.inter.tick)), ticksize = NA, side=y.side)) # ticksize = NA to allow the use of par()$tcl value
-par(tcl = -par()$mgp[2] * tick.length) # back to main tick length
-}
-y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
-}else if(( ! is.null(y.categ)) & y.log.scale == FALSE){
-if(is.null(y.categ.pos)){
-y.categ.pos <- 1:length(y.categ)
-}else if(length(y.categ.pos) != length(y.categ)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.categ.pos MUST BE THE SAME LENGTH AS y.categ")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-axis(side = y.side, at = y.categ.pos, labels = rep("", length(y.categ)), lwd=0, lwd.ticks=1) # draw the line of the axis
-par(xpd = TRUE)
-if(isTRUE(all.equal(y.side, 2))){ #isTRUE(all.equal(y.side, 2)) is similar to y.side == 2L but deals with float
-text(x = x.mid.left.fig.region, y = y.categ.pos, labels = y.categ, srt = text.angle, cex = y.axis.size)
-}else if(isTRUE(all.equal(y.side, 4))){ # idem
-text(x = x.mid.right.fig.region, y = y.categ.pos, labels = y.categ, srt = text.angle, cex = y.axis.size)
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": ARGUMENT y.side CAN ONLY BE 2 OR 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-par(xpd = FALSE)
-y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": PROBLEM WITH THE y.side (", y.side ,") OR y.log.scale (", y.log.scale,") ARGUMENTS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
-}
-par(xpd=NA)
-text(x = x.mid.right.fig.region, y = y.text, corner.text, adj=c(1, 1.1), cex = corner.text.size) # text at the topright corner. Replace x.right.fig.region by x.text if text at the right edge of the plot region
-if(just.label.add == TRUE & isTRUE(all.equal(x.side, 0)) & x.lab != ""){
-text(x = x.mid.plot.region, y = y.mid.bottom.fig.region, x.lab, adj=c(0.5, 0.5), cex = x.label.size) # x label
-}
-if(just.label.add == TRUE & isTRUE(all.equal(y.side, 0)) & y.lab != ""){
-text(x = y.mid.plot.region, y = x.mid.left.fig.region, y.lab, adj=c(0.5, 0.5), cex = y.label.size) # x label
-}
-par(xpd=FALSE)
-if(par.reset == TRUE){
-tempo.par <- fun_open(pdf = FALSE, return.output = TRUE)
-invisible(dev.off()) # close the new window
-if( ! is.null(custom.par)){
-if( ! names(custom.par) %in% names(tempo.par$ini.par)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": custom.par ARGUMENT SHOULD HAVE THE NAMES OF THE COMPARTMENT LIST COMING FROM THE par() LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-par(custom.par)
-text <- c(text, "\nGRAPH PARAMETERS SET TO VALUES DEFINED BY custom.par ARGUMENT\n")
-}else{
-par(tempo.par$ini.par)
-text <- c(text, "\nGRAPH PARAMETERS RESET TO par() DEFAULT VALUES\n")
-}
-}
-output <- list(x.mid.left.dev.region = x.mid.left.dev.region, x.left.dev.region = x.left.dev.region, x.mid.right.dev.region = x.mid.right.dev.region, x.right.dev.region = x.right.dev.region, x.mid.left.fig.region = x.mid.left.fig.region, x.left.fig.region = x.left.fig.region, x.mid.right.fig.region = x.mid.right.fig.region, x.right.fig.region = x.right.fig.region, x.left.plot.region = x.left.plot.region, x.right.plot.region = x.right.plot.region, x.mid.plot.region = x.mid.plot.region, y.mid.bottom.dev.region = y.mid.bottom.dev.region, y.bottom.dev.region = y.bottom.dev.region, y.mid.top.dev.region = y.mid.top.dev.region, y.top.dev.region = y.top.dev.region, y.mid.bottom.fig.region = y.mid.bottom.fig.region, y.bottom.fig.region = y.bottom.fig.region, y.mid.top.fig.region = y.mid.top.fig.region, y.top.fig.region = y.top.fig.region, y.top.plot.region = y.top.plot.region, y.bottom.plot.region = y.bottom.plot.region, y.mid.plot.region = y.mid.plot.region, text = text)
-return(output)
-}
-
-
-######## fun_close() #### close specific graphic windows
-
-
-fun_close <- function(kind = "pdf", return.text = FALSE){
-# AIM
-# close only specific graphic windows (devices)
-# ARGUMENTS:
-# kind: vector, among c("windows", "quartz", "x11", "X11", "pdf", "bmp", "png", "tiff"), indicating the kind of graphic windows (devices) to close. BEWARE: either "windows", "quartz", "x11" or "X11" means that all the X11 GUI graphics devices will be closed, whatever the OS used
-# return.text: print text regarding the kind parameter and the devices that were finally closed?
-# RETURN
-# text regarding the kind parameter and the devices that were finally closed
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# windows() ; windows() ; pdf() ; dev.list() ; fun_close(kind = c("pdf", "x11"), return.text = TRUE) ; dev.list()
-# DEBUGGING
-# kind = c("windows", "pdf") ; return.text = FALSE # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = kind, options = c("windows", "quartz", "x11", "X11", "pdf", "bmp", "png", "tiff"), fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.text, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-text <- paste0("THE REQUIRED KIND OF GRAPHIC DEVICES TO CLOSE ARE ", paste(kind, collapse = " "))
-if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
-if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
-tempo <- kind %in% c("windows", "quartz", "x11", "X11")
-kind[tempo] <- "windows" # term are replaced by what is displayed when using a <- dev.list() ; names(a)
-}
-}else if(Sys.info()["sysname"] == "Linux"){
-if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
-tempo.device <- suppressWarnings(try(X11(), silent = TRUE))[] # open a X11 window to try to recover the X11 system used
-if( ! is.null(tempo.device)){
-text <- paste0(text, "\nCANNOT CLOSE GUI GRAPHIC DEVICES AS REQUIRED BECAUSE THIS LINUX SYSTEM DOES NOT HAVE IT")
-}else{
-tempo <- kind %in% c("windows", "quartz", "x11", "X11")
-kind[tempo] <- names(dev.list()[length(dev.list())]) # term are replaced by what is displayed when using a <- dev.list() ; names(a)
-invisible(dev.off()) # close the X11 opened by tempo
-}
-}
-}else{ # for macOS
-if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
-tempo <- kind %in% c("windows", "quartz", "x11", "X11")
-kind[tempo] <- "quartz" # term are replaced by what is displayed when using a <- dev.list() ; names(a)
-}
-}
-kind <- unique(kind)
-if(length(dev.list()) != 0){
-for(i in length(names(dev.list())):1){
-if(names(dev.list())[i] %in% kind){
-text <- paste0(text, "\n", names(dev.list())[i], " DEVICE NUMBER ", dev.list()[i], " HAS BEEN CLOSED")
-invisible(dev.off(dev.list()[i]))
-}
-}
-}
-if(return.text == TRUE){
-return(text)
-}
-}
-
-
-################ Standard graphics
-
-
-######## fun_empty_graph() #### text to display for empty graphs
-
-
-
-
-
-fun_empty_graph <- function(
-text = NULL,
-text.size = 1,
-title = NULL,
-title.size = 1.5
-){
-# AIM
-# display an empty plot with a text in the middle of the window (for instance to specify that no plot can be drawn)
-# ARGUMENTS
-# text: character string of the message to display
-# text.size: numeric value of the text size
-# title: character string of the graph title
-# title.size: numeric value of the title size (in points)
-# RETURN
-# an empty plot
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# simple example
-# fun_empty_graph(text = "NO GRAPH")
-# white page
-# fun_empty_graph() # white page
-# all the arguments
-# fun_empty_graph(text = "NO GRAPH", text.size = 2, title = "GRAPH1", title.size = 1)
-# DEBUGGING
-# text = "NO GRAPH" ; title = "GRAPH1" ; text.size = 1
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if( ! is.null(text)){
-tempo <- fun_check(data = text, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(title)){
-tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = title.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-ini.par <- par(no.readonly = TRUE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
-par(ann=FALSE, xaxt="n", yaxt="n", mar = rep(1, 4), bty = "n", xpd = NA)
-plot(1, 1, type = "n") # no display with type = "n"
-x.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1])
-y.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]))
-if( ! is.null(text)){
-text(x = 1, y = 1, labels = text, cex = text.size)
-}
-if( ! is.null(title)){
-text(x = x.left.dev.region, y = y.top.dev.region, labels = title, adj=c(0, 1), cex = title.size)
-}
-par(ini.par)
-}
-
-
-################ gg graphics
-
-
-######## fun_gg_palette() #### ggplot2 default color palette
-
-
-
-
-
-fun_gg_palette <- function(n, kind = "std"){
-# AIM
-# provide colors used by ggplot2
-# the interest is to use another single color that is not the red one used by default
-# for ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
-# ARGUMENTS
-# n: number of groups on the graph
-# kind: either "std" for standard gg colors, "dark" for darkened gg colors, or "light" for pastel gg colors
-# RETURN
-# the vector of hexadecimal colors
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# output of the function
-# fun_gg_palette(n = 2)
-# the ggplot2 palette when asking for 7 different colors
-# plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7))
-# selection of the 5th color of the ggplot2 palette made of 7 different colors
-# plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7)[5])
-# the ggplot2 palette made of 7 darkened colors
-# plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7, kind = "dark"))
-# the ggplot2 palette made of 7 lighten colors
-# plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7, kind = "light"))
-# DEBUGGING
-# n = 0
-# kind = "std"
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = n, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & isTRUE(all.equal(n, 0))){ # isTRUE(all.equal(n, 0))) is similar to n == 0 but deals with float
-tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A NON ZERO INTEGER. HERE IT IS: ", paste(n, collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-tempo <- fun_check(data = kind, options = c("std", "dark", "light"), length = 1, fun.name = function.name) ; eval(ee)
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-hues = seq(15, 375, length = n + 1)
-hcl(h = hues, l = if(kind == "std"){65}else if(kind == "dark"){35}else if(kind == "light"){85}, c = 100)[1:n]
-}
-
-
-######## fun_gg_just() #### ggplot2 justification of the axis labeling, depending on angle
-
-
-
-
-
-fun_gg_just <- function(angle, pos, kind = "axis"){
-# AIM
-# provide correct justification for text labeling, depending on the chosen angle
-# WARNINGS
-# justification behave differently on plot, depending whether it is used for annotayed text or for axis labelling. Indeed the latter has labelling constrained
-# Of note, a bug in ggplot2: vjust sometimes does not work, i.e., the same justification result is obtained whatever the value used. This is the case with angle = 90, pos = "top", kind = "axis". While everything is fine with angle = 90, pos = "bottom", kind = "axis". At least, everything seems fine for kind = "axis" and pos = c("left", "bottom")
-# ARGUMENTS
-# angle: integer value of the text angle, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
-# pos: where text is? Either "top", "right", "bottom" or "left" of the elements to justify from
-# kind: kind of text? Either "axis" or "text". In the first case, the pos argument refers to the axis position, and in the second to annotated text (using ggplot2::annotate() or ggplot2::geom_text())
-# RETURN
-# a list containing:
-# $angle: the submitted angle (value potentially reduced to fit the [-360 ; 360] interval, e.g., 460 -> 100, without impact on the final angle displayed)
-# $pos: the selected position (argument pos)
-# $kind: the selected kind of text (argument kind)
-# $hjust: the horizontal justification
-# $vjust: the vertical justification
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_gg_just(angle = 45, pos = "bottom")
-# fun_gg_just(angle = (360*2 + 45), pos = "left")
-# output <- fun_gg_just(angle = 45, pos = "bottom") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.x = ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust))
-# output <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.y = ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)) + ggplot2::coord_flip()
-# output1 <- fun_gg_just(angle = 90, pos = "bottom") ; output2 <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.x = ggplot2::element_text(angle = output1$angle, hjust = output1$hjust, vjust = output1$vjust), axis.text.y = ggplot2::element_text(angle = output2$angle, hjust = output2$hjust, vjust = output2$vjust))
-# output <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1, km = 1, bird = "pigeon", stringsAsFactors = FALSE) ; ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point() + ggplot2::geom_text(mapping = ggplot2::aes(label = bird), angle = output$angle, hjust = output$hjust, vjust = output$vjust)
-# obs1 <- data.frame(time = 1:10, km = 1:10, bird = c(NA, NA, NA, "pigeon", NA, "cat", NA, NA, NA, NA), stringsAsFactors = FALSE) ; fun_open(width = 4, height = 4) ; for(i0 in c("text", "axis")){for(i1 in c("top", "right", "bottom", "left")){for(i2 in c(0, 45, 90, 135, 180, 225, 270, 315, 360)){output <- fun_gg_just(angle = i2, pos = i1, kind = i0) ; title <- paste0("kind: ", i0, " | pos: ", i1, " | angle = ", i2, " | hjust: ", output$hjust, " | vjust: ", output$vjust) ; if(i0 == "text"){print(ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point(color = fun_gg_palette(1), alpha = 0.5) + ggplot2::ggtitle(title) + ggplot2::geom_text(mapping = ggplot2::aes(label = bird), angle = output$angle, hjust = output$hjust, vjust = output$vjust) + ggplot2::theme(title = ggplot2::element_text(size = 5)))}else{print(ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point(color = fun_gg_palette(1), alpha = 0.5) + ggplot2::ggtitle(title) + ggplot2::geom_text(mapping = ggplot2::aes(label = bird)) + ggplot2::scale_x_continuous(position = ifelse(i1 == "top", "top", "bottom")) + ggplot2::scale_y_continuous(position = ifelse(i1 == "right", "right", "left")) + ggplot2::theme(title = ggplot2::element_text(size = 5), axis.text.x = if(i1 %in% c("top", "bottom")){ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)}, axis.text.y = if(i1 %in% c("right", "left")){ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)}))}}}} ; dev.off()
-# DEBUGGING
-# angle = 45 ; pos = "left" ; kind = "axis"
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument primary checking
-# arg with no default values
-mandat.args <- c(
-"angle",
-"pos"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# using fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = angle, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = pos, options = c("left", "top", "right", "bottom"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = kind, options = c("axis", "text"), length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <- c(
-"angle",
-"pos",
-"kind"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# end second round of checking and data preparation
-# main code
-# to get angle between -360 and 360
-while(angle > 360){
-angle <- angle - 360
-}
-while(angle < -360){
-angle <- angle + 360
-}
-# end to get angle between -360 and 360
-# justifications
-if(pos %in% c("bottom", "top")){
-# code below is for if(pos == "bottom"){
-if(any(sapply(FUN = all.equal, c(-360, -180, 0, 180, 360), angle) == TRUE)){ # equivalent of angle == -360 | angle == -180 | angle == 0 | angle == 180 | angle == 360 but deals with floats
-hjust <- 0.5
-if(kind == "text"){
-if(any(sapply(FUN = all.equal, c(-360, 0, 360), angle) == TRUE)){
-vjust <- 1
-}else if(any(sapply(FUN = all.equal, c(-180, 180), angle) == TRUE)){
-vjust <- 0
-}
-}else{
-vjust <- 0.5
-}
-}else if(any(sapply(FUN = all.equal, c(-270, 90), angle) == TRUE)){
-hjust <- 1
-vjust <- 0.5
-}else if(any(sapply(FUN = all.equal, c(-90, 270), angle) == TRUE)){
-hjust <- 0
-vjust <- 0.5
-}else if((angle > -360 & angle < -270) | (angle > 0 & angle < 90)){
-hjust <- 1
-vjust <- 1
-}else if((angle > -270 & angle < -180) | (angle > 90 & angle < 180)){
-hjust <- 1
-vjust <- 0
-}else if((angle > -180 & angle < -90) | (angle > 180 & angle < 270)){
-hjust <- 0
-vjust <- 0
-if(kind == "text" & pos == "top"){
-hjust <- 1
-}
-}else if((angle > -90 & angle < 0) | (angle > 270 & angle < 360)){
-hjust <- 0
-vjust <- 1
-}
-if(pos == "top"){
-if( ! ((angle > -180 & angle < -90) | (angle > 180 & angle < 270))){
-hjust <- 1 - hjust
-}
-vjust <- 1 - vjust
-}
-}else if(pos %in% c("left", "right")){
-# code below is for if(pos == "left"){
-if(any(sapply(FUN = all.equal, c(-270, -90, 90, 270), angle) == TRUE)){ # equivalent of angle == -270 | angle == -90 | angle == 90 | angle == 270 but deals with floats
-hjust <- 0.5
-if(kind == "text"){
-if(any(sapply(FUN = all.equal, c(-90, 90), angle) == TRUE)){
-vjust <- 0
-}else if(any(sapply(FUN = all.equal, c(-270, 270), angle) == TRUE)){
-vjust <- 1
-}
-}else{
-vjust <- 0.5
-}
-}else if(any(sapply(FUN = all.equal, c(-360, 0, 360), angle) == TRUE)){
-hjust <- 1
-vjust <- 0.5
-}else if(any(sapply(FUN = all.equal, c(-180, 180), angle) == TRUE)){
-hjust <- 0
-vjust <- 0.5
-}else if((angle > -360 & angle < -270) | (angle > 0 & angle < 90)){
-hjust <- 1
-vjust <- 0
-}else if((angle > -270 & angle < -180) | (angle > 90 & angle < 180)){
-hjust <- 0
-vjust <- 0
-}else if((angle > -180 & angle < -90) | (angle > 180 & angle < 270)){
-hjust <- 0
-vjust <- 1
-}else if((angle > -90 & angle < 0) | (angle > 270 & angle < 360)){
-hjust <- 1
-vjust <- 1
-}
-if(pos == "right"){
-hjust <- 1 - hjust
-if( ! (((angle > -270 & angle < -180) | (angle > 90 & angle < 180)) | ((angle > -180 & angle < -90) | (angle > 180 & angle < 270)))){
-vjust <- 1 - vjust
-}
-}
-}
-# end justifications
-output <- list(angle = angle, pos = pos, kind = kind, hjust = hjust, vjust = vjust)
-return(output)
-}
-
-
-######## fun_gg_get_legend() #### get the legend of ggplot objects
-
-
-
-
-
-fun_gg_get_legend <- function(ggplot_built, fun.name = NULL, lib.path = NULL){
-# AIM
-# get legend of ggplot objects
-# # from https://stackoverflow.com/questions/12539348/ggplot-separate-legend-and-plot
-# ARGUMENTS
-# ggplot_built: a ggplot build object
-# fun.name: single character string indicating the name of the function using fun_gg_get_legend() for warning and error messages. Ignored if NULL
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# RETURN
-# a list of class c("gtable", "gTree", "grob", "gDesc"), providing legend information of ggplot_built objet, or NULL if the ggplot_built object has no legend
-# REQUIRED PACKAGES
-# ggplot2
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# EXAMPLES
-# Simple example
-# obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time, fill = group)) ; fun_gg_get_legend(ggplot_built = ggplot2::ggplot_build(p))
-# Error message because no legend in the ggplot
-# obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time)) ; fun_gg_get_legend(ggplot_built = ggplot2::ggplot_build(p))
-# DEBUGGING
-# obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time)) ; ggplot_built = ggplot2::ggplot_build(p) ; fun.name = NULL ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-req.function <- c(
-"fun_check",
-"fun_pack"
-)
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED ", i1, "() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end required function checking
-# argument primary checking
-# arg with no default values
-mandat.args <- c(
-"ggplot_built"
-)
-tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# using fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = ggplot_built, class = "ggplot_built", mode = "list", fun.name = function.name) ; eval(ee)
-if( ! is.null(fun.name)){
-tempo <- fun_check(data = fun.name, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(arg.check)){
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-}
-# end using fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking
-# management of NA
-if(any(is.na(ggplot_built)) | any(is.na(fun.name)) | any(is.na(lib.path))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": NO ARGUMENT CAN HAVE NA VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA
-# management of NULL
-if(is.null(ggplot_built)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nggplot_built ARGUMENT CANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL
-if( ! is.null(lib.path)){
-if( ! all(dir.exists(lib.path))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end second round of checking
-# package checking
-fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
-# end package checking
-# main code
-win.nb <- dev.cur()
-pdf(file = NULL)
-tmp <- ggplot2::ggplot_gtable(ggplot_built)
-# BEWARE with ggplot_gtable : open a blanck device https://stackoverflow.com/questions/17012518/why-does-this-r-ggplot2-code-bring-up-a-blank-display-device
-invisible(dev.off())
-if(win.nb > 1){ # to go back to the previous active device, if == 1 means no opened device
-dev.set(win.nb)
-}
-leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
-if(length(leg) == 0L){
-legend <- NULL
-}else{
-legend <- tmp$grobs[[leg]]
-}
-return(legend)
-}
-
-
-######## fun_gg_point_rast() #### ggplot2 raster scatterplot layer
-
-
-
-
-
-fun_gg_point_rast <- function(
-data = NULL,
-mapping = NULL,
-stat = "identity",
-position = "identity",
-...,
-na.rm = FALSE,
-show.legend = NA,
-inherit.aes = TRUE,
-raster.width = NULL,
-raster.height = NULL,
-raster.dpi = 300,
-inactivate = TRUE,
-lib.path = NULL
-){
-# AIM
-# equivalent to ggplot2::geom_point() but in raster mode
-# use it like ggplot2::geom_point() with the main raster.dpi additional argument
-# WARNINGS
-# can be long to generate the plot
-# use a square plot region. Otherwise, the dots will have ellipsoid shape
-# solve the transparency problems with some GUI
-# this function is derived from the geom_point_rast() function, created by Viktor Petukhov , and present in the ggrastr package (https://rdrr.io/github/VPetukhov/ggrastr/src/R/geom-point-rast.R, MIT License, Copyright (c) 2017 Viktor Petukhov). Has been placed here to minimize package dependencies
-# ARGUMENTS
-# classical arguments of geom_point(), shown here https://rdrr.io/github/VPetukhov/ggrastr/man/geom_point_rast.html
-# raster.width : width of the result image (in inches). Default: deterined by the current device parameters
-# raster.height: height of the result image (in inches). Default: deterined by the current device parameters
-# raster.dpi: resolution of the result image
-# inactivate: logical. Inactivate the fun.name argument of the fun_check() function? If TRUE, the name of the fun_check() function in error messages coming from this function. Use TRUE if fun_gg_point_rast() is used like this: eval(parse(text = "fun_gg_point_rast"))
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# RETURN
-# a raster scatter plot
-# REQUIRED PACKAGES
-# ggplot2
-# grid (included in the R installation packages but not automatically loaded)
-# Cairo
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# EXAMPLES
-# Two pdf in the current directory
-# set.seed(1) ; data1 = data.frame(x = rnorm(100000), y = rnorm(10000), stringsAsFactors = TRUE) ; fun_open(pdf.name = "Raster") ; ggplot2::ggplot() + fun_gg_point_rast(data = data1, mapping = ggplot2::aes(x = x, y = y)) ; fun_open(pdf.name = "Vectorial") ; ggplot2::ggplot() + ggplot2::geom_point(data = data1, mapping = ggplot2::aes(x = x, y = y)) ; dev.off() ; dev.off()
-# DEBUGGING
-#
-# function name
-if(all(inactivate == FALSE)){ # inactivate has to be used here but will be fully checked below
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-}else if(all(inactivate == TRUE)){
-function.name <- NULL
-}else{
-tempo.cat <- paste0("ERROR IN fun_gg_point_rast(): CODE INCONSISTENCY 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if( ! is.null(data)){
-tempo <- fun_check(data = data, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(mapping)){
-tempo <- fun_check(data = mapping, class = "uneval", typeof = "list", fun.name = function.name) ; eval(ee) # aes() is tested
-}
-# stat and position not tested because too complicate
-tempo <- fun_check(data = na.rm, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = show.legend, class = "vector", mode = "logical", length = 1, na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = inherit.aes, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(raster.width)){
-tempo <- fun_check(data = raster.width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(raster.height)){
-tempo <- fun_check(data = raster.height, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = raster.dpi, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = inactivate, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# package checking
-fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
-fun_pack(req.package = c("grid"), lib.path = lib.path)
-fun_pack(req.package = c("Cairo"), lib.path = lib.path)
-# end package checking
-# additional functions
-DrawGeomPointRast <- function(data, panel_params, coord, na.rm = FALSE, raster.width = NULL, raster.height= NULL, raster.dpi = raster.dpi){
-if (is.null(raster.width)){
-raster.width <- par('fin')[1]
-}
-if (is.null(raster.height)){
- raster.height <- par('fin')[2]
-}
-prev_dev_id <- dev.cur()
-p <- ggplot2::GeomPoint$draw_panel(data, panel_params, coord)
-dev_id <- Cairo::Cairo(type='raster', width = raster.width*raster.dpi, height = raster.height*raster.dpi, dpi = raster.dpi, units = 'px', bg = "transparent")[1]
-grid::pushViewport(grid::viewport(width = 1, height = 1))
-grid::grid.points(x = p$x, y = p$y, pch = p$pch, size = p$size,
-name = p$name, gp = p$gp, vp = p$vp, draw = T)
-grid::popViewport()
-cap <- grid::grid.cap()
-invisible(dev.off(dev_id))
-invisible(dev.set(prev_dev_id))
-grid::rasterGrob(cap, x = 0, y = 0, width = 1, height = 1, default.units = "native", just = c("left","bottom"))
-}
-# end additional functions
-# main code
-GeomPointRast <- ggplot2::ggproto("GeomPointRast", ggplot2::GeomPoint, draw_panel = DrawGeomPointRast)
-ggplot2::layer(
-data = data,
-mapping = mapping,
-stat = stat,
-geom = GeomPointRast,
-position = position,
-show.legend = show.legend,
-inherit.aes = inherit.aes,
-params = list(
-na.rm = na.rm,
-raster.width = raster.width,
-raster.height = raster.height,
-raster.dpi = raster.dpi,
-...
-)
-)
-# end main code
-}
-
-
-######## fun_gg_boxplot() #### ggplot2 boxplot + background dots if required
-
-
-
-
-######## fun_gg_scatter() #### ggplot2 scatterplot + lines (up to 6 overlays totally)
-
-
-
-
-######## fun_gg_heatmap() #### ggplot2 heatmap + overlaid mask if required
-
-
-#test plot.margin = margin(up.space.mds, right.space.mds, down.space.mds, left.space.mds, "inches") to set the dim of the region plot ?
-# if matrix is full of zero (or same value I guess), heatmap is complicate. Test it and error message
-
-fun_gg_heatmap <- function(
-data1,
-legend.name1 = "",
-low.color1 = "blue",
-mid.color1 = "white",
-high.color1 = "red",
-limit1 = NULL,
-midpoint1 = NULL,
-data2 = NULL,
-color2 = "black",
-alpha2 = 0.5,
-invert2 = FALSE,
-text.size = 12,
-title = "",
-title.text.size = 12,
-show.scale = TRUE,
-rotate = FALSE,
-return = FALSE,
-plot = TRUE,
-add = NULL,
-warn.print = FALSE,
-lib.path = NULL
-){
-# AIM
-# ggplot2 heatmap with the possibility to overlay a mask
-# see also:
-# draw : http://www.sthda.com/english/wiki/ggplot2-quick-correlation-matrix-heatmap-r-software-and-data-visualization
-# same range scale : https://stackoverflow.com/questions/44655723/r-ggplot2-heatmap-fixed-scale-color-between-graphs
-# for ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
-# ARGUMENTS
-# data1: numeric matrix or data frame resulting from the conversion of the numeric matrix by reshape2::melt()
-# legend.name1: character string of the data1 heatmap scale legend
-# low.color1: character string of the color (i.e., "blue" or "#0000FF") of the lowest scale value
-# mid.color1: same as low.color1 but for the middle scale value. If NULL, the middle color is the default color between low.color1 and high.color1. BEWARE: argument midpoint1 is not ignored, even if mid.color1 is NULL, meaning that the default mid color can still be controled
-# high.color1: same as low.color1 but for the highest scale value
-# limit1: 2 numeric values defining the lowest and higest color scale values. If NULL, take the range of data1 values
-# midpoint1: single numeric value defining the value corresponding to the mid.color1 argument. A warning message is returned if midpoint1 does not correspond to the mean of limit1 values, because the color scale is not linear anymore. If NULL, takes the mean of limit1 values. Mean of data1, instead of mean of limit1, can be used here if required
-# data2: binary mask matrix (made of 0 and 1) of same dimension as data1 or a data frame resulting from the conversion of the binary mask matrix by reshape2::melt(). Value 1 of data2 will correspond to color2 argument (value 0 will be NA color), and the opposite if invert2 argument is TRUE (inverted mask)
-# color2: color of the 1 values of the binary mask matrix. The 0 values will be color NA
-# alpha2: numeric value (from 0 to 1) of the mask transparency
-# invert2: logical. Invert the mask (1 -> 0 and 0 -> 1)?
-# text.size: numeric value of the size of the texts in scale
-# title: character string of the graph title
-# title.text.size: numeric value of the title size (in points)
-# show.scale: logical. Show color scale?
-# rotate: logical. Rotate the heatmap 90° clockwise?
-# return: logical. Return the graph parameters?
-# plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
-# add: character string allowing to add more ggplot2 features (dots, lines, themes, etc.). BEWARE: (1) must start with "+" just after the simple or double opening quote (no space, end of line, carriage return, etc., allowed), (2) must finish with ")" just before the simple or double closing quote (no space, end of line, carriage return, etc., allowed) and (3) each function must be preceded by "ggplot2::" (for instance: "ggplot2::coord_flip()). If the character string contains the "ggplot2::theme" string, then internal ggplot2 theme() and theme_classic() functions will be inactivated to be reused by add. BEWARE: handle this argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
-# warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
-# lib.path: absolute path of the required packages, if not in the default folders
-# RETURN
-# a heatmap if plot argument is TRUE
-# a list of the graph info if return argument is TRUE:
-# $data: a list of the graphic info
-# $axes: a list of the axes info
-# $scale: the scale info (lowest, mid and highest values)
-# $warn: the warning messages. Use cat() for proper display. NULL if no warning
-# REQUIRED PACKAGES
-# ggplot2
-# reshape2
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# fun_round()
-# EXAMPLES
-# fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), title = "GRAPH 1")
-# fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), return = TRUE)
-# fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), legend.name1 = "VALUE", title = "GRAPH 1", text.size = 5, data2 = matrix(rep(c(1,0,0,0), 4), ncol = 4), invert2 = FALSE, return = TRUE)
-# diagonal matrix
-# fun_gg_heatmap(data1 = matrix(c(1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1), ncol = 4))
-# fun_gg_heatmap(data1 = reshape2::melt(matrix(c(1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1), ncol = 4)))
-# error message
-# fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), data2 = matrix(rep(c(1,0,0,0), 5), ncol = 5))
-# fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), data2 = reshape2::melt(matrix(rep(c(1,0,0,0), 4), ncol = 4)))
-# fun_gg_heatmap(data1 = reshape2::melt(matrix(1:16, ncol = 4)), data2 = reshape2::melt(matrix(rep(c(1,0,0,0), 4), ncol = 4)))
-# DEBUGGING
-# data1 = matrix(1:16, ncol = 4) ; legend.name1 = "" ; low.color1 = "blue" ; mid.color1 = "white" ; high.color1 = "red" ; limit1 = NULL ; midpoint1 = NULL ; data2 = matrix(rep(c(1,0,0,0), 4), ncol = 4) ; color2 = "black" ; alpha2 = 0.5 ; invert2 = FALSE ; text.size = 12 ; title = "" ; title.text.size = 12 ; show.scale = TRUE ; rotate = FALSE ; return = FALSE ; plot = TRUE ; add = NULL ; warn.print = TRUE ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_round", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_round() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# no reserved words required for this function
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if(all(is.matrix(data1))){
-tempo <- fun_check(data = data1, class = "matrix", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}else if(all(is.data.frame(data1))){
-tempo <- fun_check(data = data1, class = "data.frame", length = 3, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-# structure of reshape2::melt() data frame
-tempo <- fun_check(data = data1[, 1], data.name = "COLUMN 1 OF data1 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = data1[, 2], data.name = "COLUMN 2 OF data1 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = data1[, 3], data.name = "COLUMN 3 OF data1 (reshape2::melt() DATA FRAME)", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A NUMERIC MATRIX OR A DATA FRAME OUTPUT OF THE reshape::melt() FUNCTION")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = legend.name1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = low.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (all(low.color1 %in% colors() | grepl(pattern = "^#", low.color1)))){ # check that all strings of low.color1 start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": low.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(mid.color1)){
-tempo <- fun_check(data = mid.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (all(mid.color1 %in% colors() | grepl(pattern = "^#", mid.color1)))){ # check that all strings of mid.color1 start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": mid.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = high.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (all(high.color1 %in% colors() | grepl(pattern = "^#", high.color1)))){ # check that all strings of high.color1 start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": high.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(limit1)){
-tempo <- fun_check(data = limit1, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & any(limit1 %in% c(Inf, -Inf))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": limit1 ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-if( ! is.null(midpoint1)){
-tempo <- fun_check(data = midpoint1, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(data2)){
-if(all(is.matrix(data2))){
-tempo <- fun_check(data = data2, class = "matrix", mode = "numeric", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! all(unique(data2) %in% c(0,1))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX IN data2 MUST BE MADE OF 0 AND 1 ONLY (MASK MATRIX)")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & all(is.matrix(data1)) & ! identical(dim(data1), dim(data2))){ # matrix and matrix
-tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX DIMENSION IN data2 MUST BE IDENTICAL AS MATRIX DIMENSION IN data1. HERE IT IS RESPECTIVELY:\n", paste(dim(data2), collapse = " "), "\n", paste(dim(data1), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & all(is.data.frame(data1)) & nrow(data1) != prod(dim(data2))){ # reshape2 and matrix
-tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME IN data2 MUST HAVE ROW NUMBER EQUAL TO PRODUCT OF DIMENSIONS OF data1 MATRIX. HERE IT IS RESPECTIVELY:\n", paste(nrow(data1), collapse = " "), "\n", paste(prod(dim(data2)), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else if(all(is.data.frame(data2))){
-tempo <- fun_check(data = data2, class = "data.frame", length = 3, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-# structure of reshape2::melt() data frame
-tempo <- fun_check(data = data2[, 1], data.name = "COLUMN 1 OF data2 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = data2[, 2], data.name = "COLUMN 2 OF data2 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = data2[, 3], data.name = "COLUMN 3 OF data2 (reshape2::melt() DATA FRAME)", mode = "numeric", fun.name = function.name) ; eval(ee)
-}
-if(tempo$problem == FALSE & ! all(unique(data2[, 3]) %in% c(0,1))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THIRD COLUMN OF DATA FRAME IN data2 MUST BE MADE OF 0 AND 1 ONLY (MASK DATA FRAME)")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & all(is.data.frame(data1)) & ! identical(dim(data1), dim(data2))){ # data frame and data frame
-tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME DIMENSION IN data2 MUST BE IDENTICAL TO DATA FRAME DIMENSION IN data1. HERE IT IS RESPECTIVELY:\n", paste(dim(data2), collapse = " "), "\n", paste(dim(data1), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & all(is.matrix(data1)) & nrow(data2) != prod(dim(data1))){ # reshape2 and matrix
-tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME IN data2 MUST HAVE ROW NUMBER EQUAL TO PRODUCT OF DIMENSION OF data1 MATRIX. HERE IT IS RESPECTIVELY:\n", paste(nrow(data2), collapse = " "), "\n", paste(prod(dim(data1)), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A NUMERIC MATRIX OR A DATA FRAME OUTPUT OF THE reshape::melt() FUNCTION")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = color2, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (all(color2 %in% colors() | grepl(pattern = "^#", color2)))){ # check that all strings of color2 start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": color2 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = alpha2, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = invert2, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = show.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(add)){
-tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! grepl(pattern = "^\\+", add)){ # check that the add string start by +
-tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & ! grepl(pattern = "ggplot2::", add)){ #
-tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST CONTAIN \"ggplot2::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & ! grepl(pattern = ")$", add)){ # check that the add string finished by )
-tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# package checking
-fun_pack(req.package = c("reshape2", "ggplot2"), lib.path = lib.path)
-# end package checking
-# main code
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-if(all(is.matrix(data1))){
-data1 <- reshape2::melt(data1) # transform a matrix into a data frame with 2 coordinates columns and the third intensity column
-}
-if(rotate == TRUE){
-data1[, 1] <- rev(data1[, 1])
-}
-if(is.null(limit1)){
-if(any(data1[, 3] %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE THIRD COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-limit1 <- range(data1[, 3], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE limit1 ARGUMENT IS NULL -> RANGE OF data1 ARGUMENT HAS BEEN TAKEN: ", paste(fun_round(limit1), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-if(suppressWarnings(any(limit1 %in% c(Inf, -Inf)))){
-tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED LIMIT CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(is.null(midpoint1)){
-midpoint1 <- mean(limit1, na.rm = TRUE)
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE midpoint1 ARGUMENT IS NULL -> MEAN OF limit1 ARGUMENT HAS BEEN TAKEN: ", paste(fun_round(midpoint1), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else if(fun_round(midpoint1, 9) != fun_round(mean(limit1), 9)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE midpoint1 ARGUMENT (", fun_round(mean(midpoint1), 9), ") DOES NOT CORRESPOND TO THE MEAN OF THE limit1 ARGUMENT (", fun_round(mean(limit1), 9), "). COLOR SCALE IS NOT LINEAR")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data2)){
-if(all(is.matrix(data2))){
-data2 <- reshape2::melt(data2) # transform a matrix into a data frame with 2 coordinates columns and the third intensity column
-}
-if(rotate == TRUE){
-data2[, 1] <- rev(data2[, 1])
-}
-data2[, 3] <- factor(data2[, 3]) # to converte continuous scale into discrete scale
-}
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0 # to facilitate debugging
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggplot())
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data1, mapping = ggplot2::aes_string(x = names(data1)[ifelse(rotate == FALSE, 2, 1)], y = names(data1)[ifelse(rotate == FALSE, 1, 2)], fill = names(data1)[3]), show.legend = show.scale)) # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_fill_gradient2(low = low.color1, high = high.color1, mid = mid.color1, midpoint = midpoint1, limit = limit1, breaks = c(limit1[1], midpoint1, limit1[2]), labels = fun_round(c(limit1[1], midpoint1, limit1[2])), name = legend.name1))
-if( ! is.null(data2)){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data2, mapping = ggplot2::aes_string(x = names(data2)[ifelse(rotate == FALSE, 2, 1)], y = names(data2)[ifelse(rotate == FALSE, 1, 2)], alpha = names(data2)[3]), fill = color2, show.legend = FALSE))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", values = if(invert2 == FALSE){c(0, alpha2)}else{c(alpha2, 0)}, guide = FALSE))
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data2, mapping = ggplot2::aes_string(x = names(data2)[ifelse(rotate == FALSE, 2, 1)], y = names(data2)[ifelse(rotate == FALSE, 1, 2)], group = names(data2)[3]), fill = data2[, 3], alpha = alpha2, show.legend = FALSE)) # BEWARE: this does not work if NA present, because geom_raster() has a tendency to complete empty spaces, and thus, behave differently than geom_tile(). See https://github.com/tidyverse/ggplot2/issues/3025
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_fixed()) # x = y
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_y_reverse())
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
-add.check <- TRUE
-if( ! is.null(add)){ # if add is NULL, then = 0
-if(grepl(pattern = "ggplot2::theme", add) == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT -> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-add.check <- FALSE
-}
-}
-if(add.check == TRUE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_blank(),
-axis.title = ggplot2::element_blank(),
-axis.text = ggplot2::element_blank(),
-axis.ticks = ggplot2::element_blank(),
-panel.background = ggplot2::element_blank()
-))
-}
-if(plot == TRUE){
-# suppressWarnings(
-print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))
-# )
-}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-if(return == TRUE){
-output <- ggplot2::ggplot_build(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "))))
-output <- output$data
-names(output)[1] <- "heatmap"
-if( ! is.null(data2)){
-names(output)[2] <- "mask"
-}
-return(list(data = output, axes = output$layout$panel_params[[1]], scale = c(limit1[1], midpoint1, limit1[2]), warn = warn))
-}
-}
-
-
-######## fun_gg_empty_graph() #### text to display for empty graphs
-
-
-
-
-
-fun_gg_empty_graph <- function(
-text = NULL,
-text.size = 12,
-title = NULL,
-title.size = 8,
-lib.path = NULL
-){
-# AIM
-# display an empty ggplot2 plot with a text in the middle of the window (for instance to specify that no plot can be drawn)
-# ARGUMENTS
-# text: character string of the message to display
-# text.size: numeric value of the text size (in points)
-# title: character string of the graph title
-# title.size: numeric value of the title size (in points)
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
-# RETURN
-# an empty plot
-# REQUIRED PACKAGES
-# ggplot2
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# EXAMPLES
-### simple example
-# fun_gg_empty_graph(text = "NO GRAPH")
-### white page
-# fun_gg_empty_graph()
-### all the arguments
-# fun_gg_empty_graph(text = "NO GRAPH", text.size = 8, title = "GRAPH1", title.size = 10, lib.path = NULL)
-# DEBUGGING
-# text = "NO GRAPH" ; text.size = 12 ; title = "GRAPH1" ; title.size = 8 ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if( ! is.null(text)){
-tempo <- fun_check(data = text, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(title)){
-tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = title.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# package checking
-fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
-# end package checking
-# main code
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0
-# no need loop part
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggplot())
-if( ! is.null(text)){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(data = data.frame(x = 1, y = 1, stringsAsFactors = TRUE), ggplot2::aes(x = x, y = y, label = text), size = text.size))
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_void())
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-plot.title = ggplot2::element_text(size = title.size) # stronger than text
-))
-suppressWarnings(print(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ")))))
-}
-
-
-################ Graphic extraction
-
-
-######## fun_trim() #### display values from a quantitative variable and trim according to defined cut-offs
-
-# Add name of the variable in the graph
-# not max and min for boxplot but 1.5IQR
-fun_trim <- function(
-data,
-displayed.nb = NULL,
-single.value.display = FALSE,
-trim.method = "",
-trim.cutoffs = c(0.05,
-0.975),
-interval.scale.disp = TRUE,
-down.space = 0.75,
-left.space = 0.75,
-up.space = 0.3,
-right.space = 0.25,
-orient = 1,
-dist.legend = 0.37,
-box.type = "l",
-amplif.label = 1.25,
-amplif.axis = 1.25,
-std.x.range = TRUE,
-std.y.range = TRUE,
-cex.pt = 0.2,
-col.box = hsv(0.55,
-0.8,
-0.8),
-x.nb.inter.tick = 4,
-y.nb.inter.tick = 0,
-tick.length = 1,
-sec.tick.length = 0.75,
-corner.text = "",
-amplif.legend = 1,
-corner.text.size = 0.75,
-trim.return = FALSE
-){
-# AIM
-# trim and display values from a numeric vector or matrix
-# plot 4 graphs: stripchart of values, stripchart of rank of values, histogram and normal QQPlot
-# different kinds of intervals are displayed on the top of graphes to facilitate the analysis of the variable and a trimming setting
-# the trimming interval chosen is displayed on top of graphs
-# both trimmed and not trimmed values are returned in a list
-# ARGUMENTS
-# data: values to plot (either a numeric vector or a numeric matrix)
-# displayed.nb: number of values displayed. If NULL, all the values are displayed. Otherwise, if the number of values is over displayed.nb, then displayed.nb values are displayed after random selection
-# single.value.display: provide the 4 graphs if data is made of a single (potentially repeated value)? If FALSE, an empty graph is displayed if data is made of a single (potentially repeated value). And the return list is made of NULL compartments
-# trim.method: Write "" if not required. write "mean.sd" if mean +/- sd has to be displayed as a trimming interval (only recommanded for normal distribution). Write "quantile" to display a trimming interval based on quantile cut-offs. No other possibility allowed. See trim.cutoffs below
-# trim.cutoffs: 2 values cutoff for the trimming interval displayed, each value between 0 and 1. Not used if trim.method == "".The couple of values c(lower, upper) represents the lower and upper boundaries of the trimming interval (in proportion), which represent the interval of distribution kept (between 0 and 1). Example: trim.cutoffs = c(0.05, 0.975). What is strictly kept for the display is ]lower , upper[, boundaries excluded. Using the "mean.sd" method, 0.025 and 0.975 represent 95% CI which is mean +/- 1.96 * sd
-# interval.scale.disp: display sd and quantiles intervals on top of graphs ?
-# down.space: lower vertical margin (in inches, mai argument of par())
-# left.space: left horizontal margin (in inches, mai argument of par())
-# up.space: upper vertical margin between plot region and grapical window (in inches, mai argument of par())
-# right.space: right horizontal margin (in inches, mai argument of par())
-# orient: scale number orientation (las argument of par()). 0, always parallel to the axis; 1, always horizontal; 2, always perpendicular to the axis; 3, always vertical
-# dist.legend: numeric value that moves axis legends away in inches (first number of mgp argument of par() but in inches thus / 0.2)
-# box.type: bty argument of par(). Either "o", "l", "7", "c", "u", "]", the resulting box resembles the corresponding upper case letter. A value of "n" suppresses the box
-# amplif.label: increase or decrease the size of the text in legends
-# amplif.axis: increase or decrease the size of the scale numbers in axis
-# std.x.range: standard range on the x-axis? TRUE (no range extend) or FALSE (4% range extend). Controls xaxs argument of par() (TRUE is xaxs = "i", FALSE is xaxs = "r")
-# std.y.range: standard range on the y-axis? TRUE (no range extend) or FALSE (4% range extend). Controls yaxs argument of par() (TRUE is yaxs = "i", FALSE is yaxs = "r")
-# cex.pt: size of points in stripcharts (in inches, thus cex.pt will be thereafter / 0.2)
-# col.box: color of boxplot
-# x.nb.inter.tick: number of secondary ticks between main ticks on x-axis (only if not log scale). Zero means non secondary ticks
-# y.nb.inter.tick: number of secondary ticks between main ticks on y-axis (only if not log scale). Zero means non secondary ticks
-# tick.length: length of the ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc. 0 means no tick
-# sec.tick.length: length of the secondary ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
-# corner.text: text to add at the top right corner of the window
-# amplif.legend: increase or decrease the size of the text of legend
-# corner.text.size: positive numeric. Increase or decrease the size of the text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
-# trim.return: return the trimmed and non trimmed values? NULL returned for trimmed and non trimmed values if trim.method == ""
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# RETURN
-# a list containing:
-# $trim.method: correspond to trim.method above
-# $trim.cutoffs: correspond to trim.cutoffs above
-# $real.trim.cutoffs: the two boundary values (in the unit of the numeric vector or numeric matrix analyzed). NULL
-# $trimmed.values: the values outside of the trimming interval as defined in trim.cutoffs above
-# $kept.values: the values inside the trimming interval as defined in trim.cutoffs above
-# EXAMPLES
-# fun_trim(data = c(1:100, 1:10), displayed.nb = NULL, single.value.display = FALSE, trim.method = "mean.sd", trim.cutoffs = c(0.05, 0.975), interval.scale.disp = TRUE, down.space = 0.75, left.space = 0.75, up.space = 0.3, right.space = 0.25, orient = 1, dist.legend = 0.37, box.type = "l", amplif.label = 1.25, amplif.axis = 1.25, std.x.range = TRUE, std.y.range = TRUE, cex.pt = 0.2, col.box = hsv(0.55, 0.8, 0.8), x.nb.inter.tick = 4, y.nb.inter.tick = 0, tick.length = 0.5, sec.tick.length = 0.3, corner.text = "", amplif.legend = 1, corner.text.size = 0.75, trim.return = TRUE)
-# DEBUGGING
-# data = c(1:100, 1:10) ; displayed.nb = NULL ; single.value.display = FALSE ; trim.method = "quantile" ; trim.cutoffs = c(0.05, 0.975) ; interval.scale.disp = TRUE ; down.space = 1 ; left.space = 1 ; up.space = 0.5 ; right.space = 0.25 ; orient = 1 ; dist.legend = 0.5 ; box.type = "l" ; amplif.label = 1 ; amplif.axis = 1 ; std.x.range = TRUE ; std.y.range = TRUE ; cex.pt = 0.1 ; col.box = hsv(0.55, 0.8, 0.8) ; x.nb.inter.tick = 4 ; y.nb.inter.tick = 0 ; tick.length = 0.5 ; sec.tick.length = 0.3 ; corner.text = "" ; amplif.legend = 1 ; corner.text.size = 0.75 ; trim.return = TRUE # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-# argument checking without fun_check()
-if( ! (all(class(data) == "numeric") | all(class(data) == "integer") | (all(class(data) %in% c("matrix", "array")) & base::mode(data) == "numeric"))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A NUMERIC VECTOR OR NUMERIC MATRIX")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking without fun_check()
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-if( ! is.null(displayed.nb)){
-tempo <- fun_check(data = displayed.nb, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(displayed.nb < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": displayed.nb ARGUMENT MUST BE A SINGLE INTEGER VALUE GREATER THAN 1 AND NOT: ", paste(displayed.nb, collapse = " "))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = single.value.display, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = trim.method, options = c("", "mean.sd", "quantile"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = trim.cutoffs, class = "vector", mode = "numeric", length = 2, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = interval.scale.disp, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = down.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = up.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = orient, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.type, options = c("o", "l", "7", "c", "u", "]", "n"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = amplif.label, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = amplif.axis, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = std.x.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = std.y.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = cex.pt, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = col.box, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.nb.inter.tick, class = "integer", length = 1, neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.nb.inter.tick, class = "integer", length = 1, neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = sec.tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = corner.text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = amplif.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = corner.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = trim.return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-if(all(is.na(data) | ! is.finite(data))){
-tempo.cat <- paste0("ERROR IN fun_trim FUNCTION\ndata ARGUMENT CONTAINS ONLY NA OR Inf")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end argument checking
-# main code
-if(all(class(data)%in% c("matrix", "array"))){
-data <- as.vector(data)
-}
-na.nb <- NULL
-if(any(is.na(data))){
-na.nb <- sum(c(is.na(data)))
-data <- data[ ! is.na(data)]
-}
-color.cut <- hsv(0.75, 1, 1) # color of interval selected
-col.mean <- hsv(0.25, 1, 0.8) # color of interval using mean+/-sd
-col.quantile <- "orange" # color of interval using quantiles
-quantiles.selection <- c(0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 0.75, 0.9, 0.95, 0.975, 0.99) # quantiles used in axis to help for choosing trimming cutoffs
-if(single.value.display == FALSE & length(unique(data)) == 1L){
-par(bty = "n", xaxt = "n", yaxt = "n", xpd = TRUE)
-plot(1, pch = 16, col = "white", xlab = "", ylab = "")
-text(x = 1, y = 1, paste0("No graphic displayed\nBecause data made of a single different value (", formatC(as.double(table(data))), ")"), cex = 2)
-output <- list(trim.method = NULL, trim.cutoffs = NULL, real.trim.cutoffs = NULL, trimmed.values = NULL, kept.values = NULL)
-}else{
-output <- list(trim.method = trim.method, trim.cutoffs = trim.cutoffs, real.trim.cutoffs = NULL, trimmed.values = NULL, kept.values = NULL)
-fun.rug <- function(sec.tick.length.f = sec.tick.length, x.nb.inter.tick.f = x.nb.inter.tick, y.nb.inter.tick.f = y.nb.inter.tick){
-if(x.nb.inter.tick.f > 0){
-inter.tick.unit <- (par("xaxp")[2] - par("xaxp")[1]) / par("xaxp")[3]
-par.ini <- par()[c("xpd", "tcl")]
-par(xpd = FALSE)
-par(tcl = -par()$mgp[2] * sec.tick.length.f) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
-suppressWarnings(rug(seq(par("xaxp")[1] - 10 * inter.tick.unit, par("xaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + x.nb.inter.tick.f)), ticksize = NA, side = 1)) # ticksize = NA to allow the use of par()$tcl value
-par(par.ini)
-rm(par.ini)
-}
-if(y.nb.inter.tick.f > 0){
-inter.tick.unit <- (par("yaxp")[2] - par("yaxp")[1]) / par("yaxp")[3]
-par.ini <- par()[c("xpd", "tcl")]
-par(xpd = FALSE)
-par(tcl = -par()$mgp[2] * sec.tick.length.f) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
-suppressWarnings(rug(seq(par("yaxp")[1] - 10 * inter.tick.unit, par("yaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + y.nb.inter.tick.f)), ticksize = NA, side = 2)) # ticksize = NA to allow the use of par()$tcl value
-par(par.ini)
-rm(par.ini)
-}
-}
-fun.add.cut <- function(data.f, trim.method.f = trim.method, trim.cutoffs.f = trim.cutoffs, color.cut.f = color.cut, return.f = FALSE){
-# DEBUGGING
-# data.f = data ; trim.method.f = "mean.sd"; trim.cutoffs.f = trim.cutoffs ; color.cut.f = color.cut ; return.f = TRUE
-real.trim.cutoffs.f <- NULL
-if(trim.method.f != ""){
-data.f <- sort(data.f)
-par.ini <- par()$xpd
-par(xpd = FALSE)
-if(trim.method.f == "mean.sd"){
-real.trim.cutoffs.f <- qnorm(trim.cutoffs.f, mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE))
-abline(v = qnorm(trim.cutoffs.f, mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), col = color.cut.f)
-segments(qnorm(trim.cutoffs.f[1], mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), par()$usr[4] * 0.75, qnorm(trim.cutoffs.f[2], mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), par()$usr[4] * 0.75, col = color.cut.f)
-}
-if(trim.method.f == "quantile"){
-real.trim.cutoffs.f <- quantile(data.f, probs = trim.cutoffs.f, type = 7, na.rm = TRUE)
-abline(v = quantile(data.f, probs = trim.cutoffs.f, type = 7, na.rm = TRUE), col = color.cut.f)
-segments(quantile(data.f, probs = trim.cutoffs.f[1], type = 7, na.rm = TRUE), par()$usr[4] * 0.75, quantile(data.f, probs = trim.cutoffs.f[2], type = 7, na.rm = TRUE), par()$usr[4] * 0.75, col = color.cut.f)
-}
-par(par.ini)
-if(return.f == TRUE){
-trimmed.values.f <- data.f[data.f <= real.trim.cutoffs.f[1] | data.f >= real.trim.cutoffs.f[2]]
-kept.values.f <- data.f[data.f > real.trim.cutoffs.f[1] & data.f < real.trim.cutoffs.f[2]]
-}
-}else{
-real.trim.cutoffs.f <- NULL
-trimmed.values.f <- NULL
-kept.values.f <- NULL
-}
-if(return.f == TRUE){
-output <- list(trim.method = trim.method.f, trim.cutoffs = trim.cutoffs.f, real.trim.cutoffs = real.trim.cutoffs.f, trimmed.values = trimmed.values.f, kept.values = kept.values.f)
-return(output)
-}
-}
-fun.interval.scale.display <- function(data.f, col.quantile.f = col.quantile, quantiles.selection.f = quantiles.selection, col.mean.f = col.mean){ # intervals on top of graphs
-par.ini <- par()[c("mgp", "xpd")]
-par(mgp = c(0.25, 0.25, 0), xpd = NA)
-axis(side = 3, at = c(par()$usr[1], par()$usr[2]), labels = rep("", 2), col = col.quantile.f, lwd.ticks = 0)
-par(xpd = FALSE)
-axis(side = 3, at = quantile(as.vector(data.f), probs = quantiles.selection.f, type = 7, na.rm = TRUE), labels = quantiles.selection.f, col.axis = col.quantile.f, col = col.quantile.f)
-par(mgp = c(1.75, 1.75, 1.5), xpd = NA)
-axis(side = 3, at = c(par()$usr[1], par()$usr[2]), labels = rep("", 2), col = col.mean.f, lwd.ticks = 0)
-par(xpd = FALSE)
-axis(side = 3, at = m + s * qnorm(quantiles.selection.f), labels = formatC(round(qnorm(quantiles.selection.f), 2)), col.axis = col.mean.f, col = col.mean.f, lwd.ticks = 1)
-par(par.ini)
-}
-zone<-matrix(1:4, ncol=2)
-layout(zone)
-par(omi = c(0, 0, 1.5, 0), mai = c(down.space, left.space, up.space, right.space), las = orient, mgp = c(dist.legend / 0.2, 0.5, 0), xpd = FALSE, bty= box.type, cex.lab = amplif.label, cex.axis = amplif.axis, xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
-par(tcl = -par()$mgp[2] * tick.length) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
-if(is.null(displayed.nb)){
-sampled.data <- as.vector(data)
-if(corner.text == ""){
-corner.text <- paste0("ALL VALUES OF THE DATASET DISPLAYED")
-}else{
-corner.text <- paste0(corner.text, "\nALL VALUES OF THE DATASET DISPLAYED")
-}
-}else{
-if(length(as.vector(data)) > displayed.nb){
-sampled.data <- sample(as.vector(data), displayed.nb, replace = FALSE)
-if(corner.text == ""){
-corner.text <- paste0("WARNING: ONLY ", displayed.nb, " VALUES ARE DISPLAYED AMONG THE ", length(as.vector(data)), " VALUES OF THE DATASET ANALYZED")
-}else{
-corner.text <- paste0(corner.text, "\nWARNING: ONLY ", displayed.nb, " VALUES ARE DISPLAYED AMONG THE ", length(as.vector(data)), " VALUES OF THE DATASET ANALYZED")
-}
-}else{
-sampled.data <- as.vector(data)
-if(corner.text == ""){
-corner.text <- paste0("WARNING: THE DISPLAYED NUMBER OF VALUES PARAMETER ", deparse(substitute(displayed.nb)), " HAS BEEN SET TO ", displayed.nb, " WHICH IS ABOVE THE NUMBER OF VALUES OF THE DATASET ANALYZED -> ALL VALUES DISPLAYED")
-}else{
-corner.text <- paste0(corner.text, "\nWARNING: THE DISPLAYED NUMBER OF VALUES PARAMETER ", deparse(substitute(displayed.nb)), " HAS BEEN SET TO ", displayed.nb, " WHICH IS ABOVE THE NUMBER OF VALUES OF THE DATASET ANALYZED -> ALL VALUES DISPLAYED")
-}
-}
-}
-if( ! is.null(na.nb)){
-if(corner.text == ""){
-corner.text <- paste0("WARNING: NUMBER OF NA REMOVED IS ", na.nb)
-}else{
-corner.text <- paste0("WARNING: NUMBER OF NA REMOVED IS ", na.nb)
-}
-}
-stripchart(sampled.data, method="jitter", jitter=0.4, vertical=FALSE, ylim=c(0.5, 1.5), group.names = "", xlab = "Value", ylab="", pch=1, cex = cex.pt / 0.2)
-fun.rug(y.nb.inter.tick.f = 0)
-boxplot(as.vector(data), horizontal=TRUE, add=TRUE, boxwex = 0.4, staplecol = col.box, whiskcol = col.box, medcol = col.box, boxcol = col.box, range = 0, whisklty = 1)
-m <- mean(as.vector(data), na.rm = TRUE)
-s <- sd(as.vector(data), na.rm = TRUE)
-segments(m, 0.8, m, 1, lwd=2, col="red") # mean
-segments(m -1.96 * s, 0.9, m + 1.96 * s, 0.9, lwd=1, col="red") # mean
-graph.xlim <- par()$usr[1:2] # for hist() and qqnorm() below
-if(interval.scale.disp == TRUE){
-fun.interval.scale.display(data.f = data)
-if(corner.text == ""){
-corner.text <- paste0("MULTIPLYING FACTOR DISPLAYED (MEAN +/- SD) ON SCALES: ", paste(formatC(round(qnorm(quantiles.selection), 2))[-(1:(length(quantiles.selection) - 1) / 2)], collapse = ", "), "\nQUANTILES DISPLAYED ON SCALES: ", paste(quantiles.selection, collapse = ", "))
-}else{
-corner.text <- paste0(corner.text, "\nMULTIPLYING FACTOR DISPLAYED (MEAN +/- SD) ON SCALES: ", paste(formatC(round(qnorm(quantiles.selection), 2))[-(1:(length(quantiles.selection) - 1) / 2)], collapse = ", "), "\nQUANTILES DISPLAYED ON SCALES: ", paste(quantiles.selection, collapse = ", "))
-}
-}
-output.tempo <- fun.add.cut(data.f = data, return.f = TRUE) # to recover real.trim.cutoffs
-if(trim.return == TRUE){
-output <- output.tempo
-}
-par(xpd = NA)
-if(trim.method != ""){
-if(corner.text == ""){
-corner.text <- paste0("SELECTED CUT-OFFS (PROPORTION): ", paste(trim.cutoffs, collapse = ", "), "\nSELECTED CUT-OFFS: ", paste(output.tempo$real.trim.cutoffs, collapse = ", "))
-}else{
-corner.text <- paste0(corner.text, "\nSELECTED CUT-OFFS (PROPORTION): ", paste(trim.cutoffs, collapse = ", "), "\nSELECTED CUT-OFFS: ", paste(output.tempo$real.trim.cutoffs, collapse = ", "))
-}
-if(interval.scale.disp == TRUE){
-legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- 1.96sd", paste0("Trimming interval: ", paste0(trim.cutoffs, collapse = " , ")), "Mean +/- sd multiplying factor", "Quantile"), yjust = 0, lty=1, col=c(col.box, "red", color.cut, col.mean, col.quantile), bty="n", cex = amplif.legend)
-}else{
-legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- 1.96sd", paste0("Trimming interval: ", paste0(trim.cutoffs, collapse = " , "))), yjust = 0, lty=1, col=c(col.box, "red", color.cut), bty="n", cex = amplif.legend, y.intersp=1.25)
-}
-}else{
-if(interval.scale.disp == TRUE){
-legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- sd", "Mean +/- sd multiplying factor", "Quantile"), yjust = 0, lty=1, col=c(col.box, "red", col.mean, col.quantile), bty="n", cex = amplif.legend)
-}else{
-legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- sd"), yjust = 0, lty=1, col=c(col.box, "red"), bty="n", cex = amplif.legend, y.intersp=1.25)
-}
-}
-par(xpd = FALSE, xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
-hist(as.vector(data), main = "", xlim = graph.xlim, xlab = "Value", ylab="Density", col = grey(0.25)) # removed: breaks = seq(min(as.vector(data), na.rm = TRUE), max(as.vector(data), na.rm = TRUE), length.out = length(as.vector(data)) / 10)
-abline(h = par()$usr[3])
-fun.rug()
-if(interval.scale.disp == TRUE){
-fun.interval.scale.display(data.f = data)
-}
-fun.add.cut(data.f = data)
-par(xaxs = ifelse(std.x.range, "i", "r"))
-stripchart(rank(sampled.data), method="stack", vertical=FALSE, ylim=c(0.99, 1.3), group.names = "", xlab = "Rank of values", ylab="", pch=1, cex = cex.pt / 0.2)
-fun.rug(y.nb.inter.tick.f = 0)
-x.text <- par("usr")[2] + (par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1]) * (1 - par("plt")[2]) / 2
-y.text <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par()$omd[4] / 2) * ((par("plt")[4] - par("plt")[3])))) * (1 - par("omd")[4])) # BEWARE. Here in "(par()$omd[4] / 2", division by two because there are 2 graphs staked on the y axis, and not one
-par(xpd=NA)
-text(x = x.text, y = y.text, paste0(corner.text), adj=c(1, 1.1), cex = corner.text.size) # text at the topright corner
-par(xpd=FALSE)
-par(xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
-qqnorm(as.vector(sampled.data), main = "", datax = TRUE, ylab = "Value", pch = 1, col = "red", cex = cex.pt / 0.2)
-fun.rug()
-if(diff(quantile(as.vector(data), probs = c(0.25, 0.75), na.rm = TRUE)) != 0){ # otherwise, error generated
-qqline(as.vector(data), datax = TRUE)
-}
-if(interval.scale.disp == TRUE){
-fun.interval.scale.display(data.f = data)
-}
-fun.add.cut(data.f = data)
-}
-if(trim.return == TRUE){
-return(output)
-}
-}
-
-
-######## fun_segmentation() #### segment a dot cloud on a scatterplot and define the dots from another cloud outside the segmentation
-
-
-fun_segmentation <- function(
-data1,
-x1,
-y1,
-x.range.split = NULL,
-x.step.factor = 10,
-y.range.split = NULL,
-y.step.factor = 10,
-error = 0,
-data2 = NULL,
-x2,
-y2,
-data2.pb.dot = "unknown",
-xy.cross.kind = "&",
-plot = FALSE,
-graph.in.file = FALSE,
-raster = TRUE,
-warn.print = FALSE,
-lib.path = NULL
-){
-# AIM
-# if data1 is a data frame corresponding to the data set of a scatterplot (with a x column for x-axis values and a y column for the y-axis column), then fun_segmentation() delimits a frame around the dots cloud using a sliding window set by x.range.split and x.step.factor to frame the top and bottom part of the cloud, and set by y.range.split and y.step.factor to frame the left and right part of the cloud
-# if a second data frame is provided, corresponding to the data set of a scatterplot (with a x column for x-axis values and a y column for the y-axis column), then fun_segmentation() defines the dots of this data frame, outside of the frame of the first data frame
-# WARNINGS
-# if dots from data2 look significant on the graph (outside the frame) but are not (not black on the last figure), this is probably because the frame is flat on the zero coordinate (no volume inside the frame at this position). Thus, no way to conclude that data2 dots here are significant. These dots are refered to as "unknown". The pb.dot argument deals with such dots
-# dots that are sometimes inside and outside the frame, depending on the sliding window, are treated differently: they are removed. Such dots are neither classified as "signif", "non signif" or "unknown", but as "inconsistent"
-# unknown dots are treated as finally significant, not significant, or unknown (data2.pb.dot argument) for each x-axis and y-axis separately. Then, the union or intersection of significant dots is performed (argument xy.cross.kind). See the example section
-# ARGUMENTS
-# data1: a data frame containing a column of x-axis values and a column of y-axis values
-# x1: character string of the data1 column name for x-axis (first column of data1 by default)
-# y1: character string of the data1 column name for y-axis (second column of data1 by default)
-# x.range.split: positive non null numeric value giving the number of interval on the x value range. if x.range is the range of the dots on the x-axis, then abs(diff(x.range) / x.range.split) gives the window size. Window size decreases when range.split increases. In unit of x-axis. Write NULL if not required. At least one of the x.range.split and y.range.split must be non NULL
-# x.step.factor: positive non null numeric value giving the shift step of the window. If x.step.factor = 1, no overlap during the sliding (when the window slides from position n to position n+1, no overlap between the two positions). If x.step.factor = 2, 50% of overlap (when the window slides from position n to position n+1, the window on position n+1 overlap 50% of the window when it was on position n)
-# y.range.split: same as x.range.split for the y-axis. At least one of the x.range.split and y.range.split must be non NULL
-# y.step.factor: same as x.step.factor for the y-axis
-# error: proportion (from 0 to 1) of false positives (i.e., proportion of dots from data1 outside of the frame). 0.05 means 5% of the dots from data1 outside of the frame
-# data2: a data frame containing a column of x-axis values and a column of y-axis values, for which outside dots of the data1 cloud has to be determined. Write NULL if not required
-# x2: character string of the data1 column name for x-axis (first column of data1 by default)
-# y2: character string of the data1 column name for y-axis (second column of data1 by default)
-# data2.pb.dot: unknown dots are explain in the warning section above. If "signif", then the unknown dots are finally considered as significant (outside the frame). If "not.signif", then the unknown dots are finally considered as non significant (inside the frame). If "unknown", no conclusion are drawn from these dots. See the examples below
-# xy.cross.kind: if data2 is non null and if both x.range.split and y.range.split are non null, which dots are finally significants? Write "&" for intersection of outside dots on x and on y. Write "|" for union of outside dots on x and on y. See the examples below
-# plot: logical. Print graphs that check the frame?
-# graph.in.file: logical. Graphs sent into a graphic device already opened? If FALSE, GUI are opened for each graph. If TRUE, no GUI are opended. The graphs are displayed on the current active graphic device. Ignored if plot is FALSE
-# raster: logical. Dots in raster mode? If FALSE, dots from each geom_point from geom argument are in vectorial mode (bigger pdf and long to display if millions of dots). If TRUE, dots from each geom_point from geom argument are in matricial mode (smaller pdf and easy display if millions of dots, but long to generate the layer). If TRUE, the region plot will be square to avoid a bug in fun_gg_point_rast(). If TRUE, solve the transparency problem with some GUI. Not considered if plot is FALSE
-# warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
-# lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL. Ignored if plot is FALSE
-# RETURN
-# several graphs if plot is TRUE
-# a list containing:
-# $data1.removed.row.nb: which rows have been removed due to NA; NaN, -Inf or Inf detection in x1 or y1 columns (NULL if no row removed)
-# $data1.removed.rows: removed rows (NULL if no row removed)
-# $data2.removed.row.nb: which rows have been removed due to NA; NaN, -Inf or Inf detection in x2 or y2 columns (NULL if no row removed)
-# $data2.removed.rows: removed rows (NULL if no row removed)
-# $hframe: x and y coordinates of the bottom and top frames for frame plotting (frame1 for the left step and frame2 for the right step)
-# $vframe: x and y coordinates of the left and right frames for frame plotting (frame1 for the down step and frame2 for the top step)
-# $data1.signif.dot: the significant dots of data1 (i.e., dots outside the frame). A good segmentation should not have any data1.signif.dot
-# $data1.non.signif.dot: the non significant dots of data1 (i.e., dots inside the frame)
-# $data1.inconsistent.dot: see the warning section above
-# $data2.signif.dot: the significant dots of data2 if non NULL (i.e., dots outside the frame)
-# $data2.non.signif.dot: the non significant dots of data2 (i.e., dots inside the frame)
-# $data2.unknown.dot: the problematic dots of data2 (i.e., data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots). Is systematically NULL except if argument data2.pb.dot = "unknown" and some data2 dots are in such situation. Modifying the segmentation x.range.split, x.step.factor, y.range.split, y.step.factor arguments can solve this problem
-# $data2.inconsistent.dot: see the warning section above
-# $axes: the x-axis and y-axis info
-# $warn: the warning messages. Use cat() for proper display. NULL if no warning
-# REQUIRED PACKAGES
-# ggplot2 if plot is TRUE
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# if plot is TRUE:
-# fun_pack()
-# fun_open()
-# fun_gg_palette()
-# fun_gg_scatter()
-# fun_gg_empty_graph()
-# fun_close()
-# EXAMPLES
-# example explaining the unknown and inconsistent dots, and the cross
-
-# set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data1[5:7, 2] <- NA ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; data2[11:13, 1] <- Inf ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = 20, x.step.factor = 10, y.range.split = 23, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "not.signif", xy.cross.kind = "|", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
-# set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = NULL, x.step.factor = 10, y.range.split = 23, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "unknown", xy.cross.kind = "|", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
-# set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = 20, x.step.factor = 10, y.range.split = NULL, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "unknown", xy.cross.kind = "&", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
-# DEBUGGING
-# set.seed(1) ; data1 = data.frame(x = rnorm(50), y = rnorm(50), stringsAsFactors = TRUE) ; data1[5:7, 2] <- NA ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 5 ; x.step.factor = 10 ; y.range.split = 5 ; y.step.factor = 10 ; error = 0 ; data2 = data.frame(x = rnorm(50, 0, 2), y = rnorm(50, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "unknown" ; xy.cross.kind = "|" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
-# set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 20 ; x.step.factor = 10 ; y.range.split = 23 ; y.step.factor = 10 ; error = 0 ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "not.signif" ; xy.cross.kind = "|" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
-# set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 20 ; x.step.factor = 10 ; y.range.split = NULL ; y.step.factor = 10 ; error = 0 ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "unknown" ; xy.cross.kind = "&" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-ini.warning.length <- options()$warning.length
-warn <- NULL
-warn.count <- 0
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & length(data1) < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OF AT LEAST 2 COLUMNS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = x1, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (x1 %in% names(data1))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x1 ARGUMENT MUST BE A COLUMN NAME OF data1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & x1 %in% names(data1)){
-tempo <- fun_check(data = data1[, x1], data.name = "x1 COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = y1, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (y1 %in% names(data1))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y1 ARGUMENT MUST BE A COLUMN NAME OF data1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & y1 %in% names(data1)){
-tempo <- fun_check(data = data1[, y1], data.name = "y1 COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-if(is.null(x.range.split) & is.null(y.range.split)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE OF THE x.range.split AND y.range.split ARGUMENTS MUST BE NON NULL")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(x.range.split)){
-tempo <- fun_check(data = x.range.split, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & x.range.split < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.range.split ARGUMENT CANNOT BE LOWER THAN 1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-if( ! is.null(y.range.split)){
-tempo <- fun_check(data = y.range.split, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & y.range.split < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.range.split ARGUMENT CANNOT BE LOWER THAN 1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = x.step.factor, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & x.step.factor < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.step.factor ARGUMENT CANNOT BE LOWER THAN 1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = y.step.factor, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & y.step.factor < 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.step.factor ARGUMENT CANNOT BE LOWER THAN 1")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = error, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(data2)){
-if(is.null(x2) | is.null(y2)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x2 AND y2 ARGUMENTS CANNOT BE NULL IF data2 ARGUMENT IS NON NULL")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = data2, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & length(data2) < 2){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data2 ARGUMENT MUST BE A DATA FRAME OF AT LEAST 2 COLUMNS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(x2)){
-tempo <- fun_check(data = x2, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (x2 %in% names(data2))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x2 ARGUMENT MUST BE A COLUMN NAME OF data2")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & x2 %in% names(data2)){
-tempo <- fun_check(data = data2[, x2], data.name = "x2 COLUMN OF data2", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-}
-if( ! is.null(y2)){
-tempo <- fun_check(data = y2, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & ! (y2 %in% names(data2))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y2 ARGUMENT MUST BE A COLUMN NAME OF data2")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & y2 %in% names(data2)){
-tempo <- fun_check(data = data2[, y2], data.name = "y2 COLUMN OF data2", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-}
-}
-}
-if( ! is.null(data2)){
-tempo <- fun_check(data = data2.pb.dot, options = c("signif", "not.signif", "unknown"), length = 1, fun.name = function.name) ; eval(ee)
-}
-if( ! (is.null(x.range.split)) & ! (is.null(y.range.split))){
-tempo <- fun_check(data = xy.cross.kind, options = c("&", "|"), length = 1, fun.name = function.name) ; eval(ee)
-}
-tempo <- fun_check(data = plot, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & plot == TRUE){
-tempo <- fun_check(data = raster, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = graph.in.file, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & graph.in.file == TRUE & is.null(dev.list())){
-tempo.cat <- paste0("ERROR IN ", function.name, ": \ngraph.in.file PARAMETER SET TO TRUE BUT NO ACTIVE GRAPHIC DEVICE DETECTED")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo$problem == FALSE & graph.in.file == TRUE & ! is.null(dev.list())){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHS PRINTED IN THE CURRENT DEVICE (TYPE ", toupper(names(dev.cur())), ")")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# other required function checking
-if(plot == TRUE){
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_open", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_open() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_gg_palette", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_palette() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_gg_empty_graph", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_empty_graph() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_gg_scatter", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_scatter() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_close", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_close() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end other required function checking
-# package checking
-if(plot == TRUE){
-fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
-}
-# end package checking
-# main code
-# na and Inf detection and removal (done now to be sure of the correct length of categ)
-data1.removed.row.nb <- NULL
-data1.removed.rows <- NULL
-data2.removed.row.nb <- NULL
-data2.removed.rows <- NULL
-if(any(is.na(data1[, c(x1, y1)])) | any(is.infinite(data1[, x1])) | any(is.infinite(data1[, y1]))){
-tempo.na <- unlist(lapply(lapply(c(data1[c(x1, y1)]), FUN = is.na), FUN = which))
-tempo.inf <- unlist(lapply(lapply(c(data1[c(x1, y1)]), FUN = is.infinite), FUN = which))
-data1.removed.row.nb <- sort(unique(c(tempo.na, tempo.inf)))
-if(length(data1.removed.row.nb) > 0){
-data1.removed.rows <- data1[data1.removed.row.nb, ]
-}
-if(length(data1.removed.row.nb) == nrow(data1)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE NA, NaN, -Inf OR Inf DETECTED IN EACH ROW OF data1. FUNCTION CANNOT BE USED ON EMPTY DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(data1.removed.row.nb) > 0){
-data1 <- data1[-data1.removed.row.nb, ]
-}
-if(nrow(data1) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x1, y1), collapse = " "), " OF data1 AND CORRESPONDING ROWS REMOVED (SEE $data1.removed.row.nb AND $data1.removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x1, y1), collapse = " "), " OF data1. NO ROW REMOVED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data2)){
-if(any(is.na(data2[, c(x2, y2)])) | any(is.infinite(data2[, x2])) | any(is.infinite(data2[, y2]))){
-tempo.na <- unlist(lapply(lapply(c(data2[c(x2, y2)]), FUN = is.na), FUN = which))
-tempo.inf <- unlist(lapply(lapply(c(data2[c(x2, y2)]), FUN = is.infinite), FUN = which))
-data2.removed.row.nb <- sort(unique(c(tempo.na, tempo.inf)))
-if(length(data2.removed.row.nb) > 0){
-data2.removed.rows <- data2[data2.removed.row.nb, ]
-}
-if(length(data2.removed.row.nb) == nrow(data2)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE NA, NaN, -Inf OR Inf DETECTED IN EACH ROW OF data2. FUNCTION CANNOT BE USED ON EMPTY DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(data2.removed.row.nb) > 0){
-data2 <- data2[-data2.removed.row.nb, ]
-}
-if(nrow(data2) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x2, y2), collapse = " "), " OF data2 AND CORRESPONDING ROWS REMOVED (SEE $data2.removed.row.nb AND $data2.removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NO NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x2, y2), collapse = " "), " OF data2. NO ROW REMOVED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end na and Inf detection and removal (done now to be sure of the correct length of categ)
-# row annotation (dot number)
-# data1 <- data1[ ! duplicated(data1[, c(x1, y1)]), ] # do not remove the dots that have same x and y values, because they will have different dot number -> not the same position on the matrices (so true for symmetric matrices)
-data1 <- cbind(data1, DOT_NB = 1:nrow(data1), stringsAsFactors = TRUE)
-if( ! is.null(data2)){
-# data2 <- data2[ ! duplicated(data2[, c(x2, y2)]), ] # do not remove the dots that have same x and y values, because they will have different dot number -> not the same position on the matrices (so true for symmetric matrices)
-data2 <- cbind(data2, DOT_NB = 1:nrow(data2), stringsAsFactors = TRUE)
-}
-# end row annotation (dot number)
-
-
-
-
-# Method using x unit interval
-# may be create vector of each column to increase speed
-x.data1.l <- NULL # x coord of the y upper and lower limits defined on the data1 cloud for left step line
-x.data1.r <- NULL # x coord of the y upper and lower limits defined on the data1 cloud for right step line
-y.data1.down.limit.l <- NULL # lower limit of the data1 cloud for left step line
-y.data1.top.limit.l <- NULL # upper limit of the data1 cloud for left step line
-y.data1.down.limit.r <- NULL # lower limit of the data1 cloud for right step line
-y.data1.top.limit.r <- NULL # upper limit of the data1 cloud for left step line
-if(any(data1[, x1] %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE x1 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-x.range <- range(data1[, x1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-if(suppressWarnings(any(x.range %in% c(Inf, -Inf)))){
-tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED x.range CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1[, y1] %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE y1 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-y.range <- range(data1[, y1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-if(suppressWarnings(any(x.range %in% c(Inf, -Inf)))){
-tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED y.range CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-x.range.plot <- range(data1[, x1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-y.range.plot <- range(data1[, y1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-if( ! is.null(data2)){
-if(any(data2[, x2] %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE data2 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE x2 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-x.range.plot <- range(data1[, x1], data2[, x2], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-if(any(data2[, y2] %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE data2 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE y2 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-y.range.plot <- range(data1[, y1], data2[, y2], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-}
-if(suppressWarnings(any(x.range.plot %in% c(Inf, -Inf)))){
-tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED x.range.plot CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 (AND data2?) ARGUMENTS ARE NA OR Inf ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(suppressWarnings(any(y.range.plot %in% c(Inf, -Inf)))){
-tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED y.range.plot CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 (AND data2?) ARGUMENTS ARE NA OR Inf ONLY")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(x.range.split)){
-# data.frame ordering to slide the window from small to big values + sliding window definition
-data1 <- data1[order(data1[, x1], na.last = TRUE), ]
-if( ! is.null(data2)){
-data2 <- data2[order(data2[, x2], na.last = TRUE), ]
-}
-x.win.size <- abs(diff(x.range) / x.range.split) # in unit of x-axis
-step <- x.win.size / x.step.factor
-# end data.frame ordering to slide the window from small to big values + sliding window definition
-# x-axis sliding and y-axis limits of the data1 cloud -> y significant data2
-loop.nb <- ceiling((diff(x.range) - x.win.size) / step) # x.win.size + n * step covers the x range if x.win.size + n * step >= diff(x.range), thus if n >= (diff(x.range) - x.win.size) / step
-y.outside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are upper or lower than the frame
-y.inside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are not upper or lower than the frame
-y.data1.median <- median(data1[, y1], na.rm = TRUE) # will be used for sliding windows without data1 in it
-if( ! is.null(data2)){
-y.outside.data2.dot.nb <- integer() # vector that will contain the selected 1D coordinates (i.e., dots) of data2 that are upper or lower than the data1 frame
-y.inside.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are not upper or lower than the data1 frame
-y.unknown.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are problematic: data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots
-# recover data2 dots outside the range of data1
-if(any(data2[, x2] < x.range[1])){
-y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[data2[, x2] < x.range[1]])
-#tempo.warn & indicate the interval
-}
-if(any(data2[, x2] > x.range[2])){
-y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[data2[, x2] > x.range[2]])
-#tempo.warn & indicate the interval
-}
-# end recover data2 dots outside the range of data1
-}
-# loop.ini.time <- as.numeric(Sys.time())
-for(i1 in 0:(loop.nb + 1)){
-min.pos <- x.range[1] + step * i1 # lower position of the sliding window in data1
-max.pos <- min.pos + x.win.size # upper position of the sliding window in data1
-x.data1.l <- c(x.data1.l, min.pos, min.pos + step) # min.pos + step to make the steps
-x.data1.r <- c(x.data1.r, max.pos, max.pos + step) # max.pos + step to make the steps
-x.data1.dot.here <- data1[, x1] >= min.pos & data1[, x1] < max.pos # is there data1 dot present in the sliding window, considering the x axis?
-if( ! is.null(data2)){
-x.data2.dot.here <- data2[, x2] >= min.pos & data2[, x2] < max.pos # is there data2 dot present in the sliding window, considering the x axis?
-}
-# recover the data1 dots outside the frame
-if(any(x.data1.dot.here == TRUE)){
-tempo.y.data1.top.limit <- quantile(data1[x.data1.dot.here, y1], probs = 1 - error, na.rm = TRUE)
-tempo.y.data1.down.limit <- quantile(data1[x.data1.dot.here, y1], probs = 0 + error, na.rm = TRUE)
-y.data1.top.limit.l <- c(y.data1.top.limit.l, tempo.y.data1.top.limit, tempo.y.data1.top.limit)
-y.data1.down.limit.l <- c(y.data1.down.limit.l, tempo.y.data1.down.limit, tempo.y.data1.down.limit)
-y.data1.top.limit.r <- c(y.data1.top.limit.r, tempo.y.data1.top.limit, tempo.y.data1.top.limit)
-y.data1.down.limit.r <- c(y.data1.down.limit.r, tempo.y.data1.down.limit, tempo.y.data1.down.limit)
-y.data1.dot.signif <- ( ! ((data1[, y1] <= tempo.y.data1.top.limit) & (data1[, y1] >= tempo.y.data1.down.limit))) & x.data1.dot.here # is there data1 dot present in the sliding window, above or below the data1 limits, considering the y axis?
-y.data1.dot.not.signif <- x.data1.dot.here & ! y.data1.dot.signif
-y.outside.data1.dot.nb <- c(y.outside.data1.dot.nb, data1$DOT_NB[y.data1.dot.signif]) # recover the row number of data1
-y.outside.data1.dot.nb <- unique(y.outside.data1.dot.nb)
-y.inside.data1.dot.nb <- c(y.inside.data1.dot.nb, data1$DOT_NB[y.data1.dot.not.signif])
-y.inside.data1.dot.nb <- unique(y.inside.data1.dot.nb)
-}else{
-y.data1.top.limit.l <- c(y.data1.top.limit.l, y.data1.median, y.data1.median)
-y.data1.down.limit.l <- c(y.data1.down.limit.l, y.data1.median, y.data1.median)
-y.data1.top.limit.r <- c(y.data1.top.limit.r, y.data1.median, y.data1.median)
-y.data1.down.limit.r <- c(y.data1.down.limit.r, y.data1.median, y.data1.median)
-}
-# end recover the data1 dots outside the frame
-# recover the data2 dots outside the frame
-if( ! is.null(data2)){
-if(any(x.data1.dot.here == TRUE) & any(x.data2.dot.here == TRUE)){
-y.data2.dot.signif <- ( ! ((data2[, y2] <= tempo.y.data1.top.limit) & (data2[, y2] >= tempo.y.data1.down.limit))) & x.data2.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the y axis?
-y.data2.dot.not.signif <- x.data2.dot.here & ! y.data2.dot.signif
-y.outside.data2.dot.nb <- c(y.outside.data2.dot.nb, data2$DOT_NB[y.data2.dot.signif])
-y.outside.data2.dot.nb <- unique(y.outside.data2.dot.nb)
-y.inside.data2.dot.nb <- c(y.inside.data2.dot.nb, data2$DOT_NB[y.data2.dot.not.signif])
-y.inside.data2.dot.nb <- unique(y.inside.data2.dot.nb)
-}else if(any(x.data1.dot.here == FALSE) & any(x.data2.dot.here == TRUE)){ # problem: data2 dots in the the window but no data1 dots to generates the quantiles
-y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[x.data2.dot.here])
-y.unknown.data2.dot.nb <- unique(y.unknown.data2.dot.nb)
-#tempo.warn & indicate the interval
-
-
-
-
-# tempo.warn <- paste0("FROM FUNCTION ", function.name, ": THE [", round(min.pos, 3), " ; ", round(max.pos, 3), "] INTERVAL DOES NOT CONTAIN data1 X VALUES BUT CONTAINS data2 X VALUES WHICH CANNOT BE EVALUATED.\nTHE CONCERNED data2 ROW NUMBERS ARE:\n", paste(which(x.data1.dot.here == FALSE & x.data2.dot.here == TRUE), collapse = "\n"))
-# warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end recover the data2 dots outside the frame
-# if(any(i1 == seq(1, loop.nb, 500))){
-# loop.fin.time <- as.numeric(Sys.time()) # time of process end
-# cat(paste0("COMPUTATION TIME OF LOOP ", i1, " / ", loop.nb, ": ", as.character(lubridate::seconds_to_period(round(loop.fin.time - loop.ini.time))), "\n"))
-# }
-}
-if(max.pos < x.range[2]){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE SLIDING WINDOW HAS NOT REACHED THE MAX VALUE OF data1 ON THE X-AXIS: ", max.pos, " VERSUS ", x.range[2])
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-y.incon.data1.dot.nb.final <- unique(c(y.outside.data1.dot.nb[y.outside.data1.dot.nb %in% y.inside.data1.dot.nb], y.inside.data1.dot.nb[y.inside.data1.dot.nb %in% y.outside.data1.dot.nb])) # inconsistent dots: if a row number of y.inside.data1.dot.nb is present in y.outside.data1.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
-y.outside.data1.dot.nb.final <- y.outside.data1.dot.nb[ ! (y.outside.data1.dot.nb %in% y.incon.data1.dot.nb.final)] # inconsistent dots removed from the outside list
-y.inside.data1.dot.nb.final <- y.inside.data1.dot.nb[ ! (y.inside.data1.dot.nb %in% y.incon.data1.dot.nb.final)] # inconsistent dots removed from the inside list
-if( ! is.null(data2)){
-# if some unknown dots are also inside, and/or outside, they are put in the inside and/or outside. Ok, because then the intersection between inside and outside is treated -> inconsistent dots
-tempo.unknown.out <- y.unknown.data2.dot.nb[y.unknown.data2.dot.nb %in% y.outside.data2.dot.nb]
-y.outside.data2.dot.nb <- unique(c(y.outside.data2.dot.nb, tempo.unknown.out)) # if a row number of y.unknown.data2.dot.nb is present in y.outside.data2.dot.nb, it is put into outside
-tempo.unknown.in <- y.unknown.data2.dot.nb[y.unknown.data2.dot.nb %in% y.inside.data2.dot.nb]
-y.inside.data2.dot.nb <- unique(c(y.inside.data2.dot.nb, tempo.unknown.in)) # if a row number of y.unknown.data2.dot.nb is present in y.inside.data2.dot.nb, it is put into inside
-y.unknown.data2.dot.nb.final <- y.unknown.data2.dot.nb[ ! (y.unknown.data2.dot.nb %in% c(y.outside.data2.dot.nb, y.inside.data2.dot.nb))] # then dots also in inside and outside are remove from unknown
-y.incon.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb[y.outside.data2.dot.nb %in% y.inside.data2.dot.nb], y.inside.data2.dot.nb[y.inside.data2.dot.nb %in% y.outside.data2.dot.nb])) # inconsistent dots: if a row number of y.inside.data2.dot.nb is present in y.outside.data2.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
-y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb[ ! (y.outside.data2.dot.nb %in% y.incon.data2.dot.nb.final)] # inconsistent dots removed from the outside list
-y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb[ ! (y.inside.data2.dot.nb %in% y.incon.data2.dot.nb.final)] # inconsistent dots removed from the inside list
-}
-# end x-axis sliding and y-axis limits of the data1 cloud -> y significant data2
-}
-# end Method using x unit interval
-
-
-
-
-# Method using y unit interval
-y.data1.d <- NULL # y coord of the x upper and lower limits defined on the data1 cloud for down step line
-y.data1.t <- NULL # y coord of the x upper and lower limits defined on the data1 cloud for top step line
-x.data1.left.limit.d <- NULL # left limit of the data1 cloud for down step line
-x.data1.right.limit.d <- NULL # right limit of the data1 cloud for down step line
-x.data1.left.limit.t <- NULL # left limit of the data1 cloud for top step line
-x.data1.right.limit.t <- NULL # right limit of the data1 cloud for top step line
-if( ! is.null(y.range.split)){
-# data.frame ordering to slide the window from small to big values + sliding window definition
-data1 <- data1[order(data1[, y1], na.last = TRUE), ]
-if( ! is.null(data2)){
-data2 <- data2[order(data2[, y2], na.last = TRUE), ]
-}
-y.win.size <- abs(diff(y.range) / y.range.split) # in unit of y-axis
-step <- y.win.size / y.step.factor
-# end data.frame ordering to slide the window from small to big values + sliding window definition
-# y-axis sliding and x-axis limits of the data1 cloud -> x significant data2
-loop.nb <- ceiling((diff(y.range) - y.win.size) / step) # y.win.size + n * step covers the y range if y.win.size + n * step >= diff(y.range), thus if n >= (diff(y.range) - y.win.size) / step
-x.outside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are upper or lower than the frame
-x.inside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are not upper or lower than the frame
-x.data1.median <- median(data1[, x1], na.rm = TRUE) # will be used for sliding window without data1 in it
-if( ! is.null(data2)){
-x.outside.data2.dot.nb <- integer() # vector that will contain the selected 1D coordinates (i.e., dots) of data2 that are upper or lower than the data1 frame
-x.inside.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are not upper or lower than the data1 frame
-x.unknown.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are problematic: data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots
-# recover data2 dots outside the range of data1
-if(any(data2[, y2] < y.range[1])){
-x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[data2[, y2] < y.range[1]])
-}
-if(any(data2[, y2] > y.range[2])){
-x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[data2[, y2] > y.range[2]])
-}
-# end recover data2 dots outside the range of data1
-}
-# loop.ini.time <- as.numeric(Sys.time())
-for(i1 in 0:(loop.nb + 1)){
-min.pos <- y.range[1] + step * i1 # lower position of the sliding window in data1
-max.pos <- min.pos + y.win.size # upper position of the sliding window in data1
-y.data1.d <- c(y.data1.d, min.pos, min.pos + step) # min.pos + step to make the steps
-y.data1.t <- c(y.data1.t, max.pos, max.pos + step) # max.pos + step to make the steps
-y.data1.dot.here <- data1[, y1] >= min.pos & data1[, y1] < max.pos # is there data1 dot present in the sliding window, considering the y axis?
-if( ! is.null(data2)){
-y.data2.dot.here <- data2[, y2] >= min.pos & data2[, y2] < max.pos # is there data2 dot present in the sliding window, considering the y axis?
-}
-# recover the data1 dots outside the frame
-if(any(y.data1.dot.here == TRUE)){
-tempo.x.data1.right.limit <- quantile(data1[y.data1.dot.here, x1], probs = 1 - error, na.rm = TRUE)
-tempo.x.data1.left.limit <- quantile(data1[y.data1.dot.here, x1], probs = 0 + error, na.rm = TRUE)
-x.data1.right.limit.d <- c(x.data1.right.limit.d, tempo.x.data1.right.limit, tempo.x.data1.right.limit)
-x.data1.left.limit.d <- c(x.data1.left.limit.d, tempo.x.data1.left.limit, tempo.x.data1.left.limit)
-x.data1.right.limit.t <- c(x.data1.right.limit.t, tempo.x.data1.right.limit, tempo.x.data1.right.limit)
-x.data1.left.limit.t <- c(x.data1.left.limit.t, tempo.x.data1.left.limit, tempo.x.data1.left.limit)
-x.data1.dot.signif <- ( ! ((data1[, x1] <= tempo.x.data1.right.limit) & (data1[, x1] >= tempo.x.data1.left.limit))) & y.data1.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the x axis?
-x.data1.dot.not.signif <- y.data1.dot.here & ! x.data1.dot.signif
-x.outside.data1.dot.nb <- c(x.outside.data1.dot.nb, data1$DOT_NB[x.data1.dot.signif]) # recover the row number of data1
-x.outside.data1.dot.nb <- unique(x.outside.data1.dot.nb)
-x.inside.data1.dot.nb <- c(x.inside.data1.dot.nb, data1$DOT_NB[x.data1.dot.not.signif])
-x.inside.data1.dot.nb <- unique(x.inside.data1.dot.nb)
-}else{
-x.data1.right.limit.d <- c(x.data1.right.limit.d, x.data1.median, x.data1.median)
-x.data1.left.limit.d <- c(x.data1.left.limit.d, x.data1.median, x.data1.median)
-x.data1.right.limit.t <- c(x.data1.right.limit.t, x.data1.median, x.data1.median)
-x.data1.left.limit.t <- c(x.data1.left.limit.t, x.data1.median, x.data1.median)
-}
-# end recover the data1 dots outside the frame
-# recover the data2 dots outside the frame
-if( ! is.null(data2)){
-if(any(y.data1.dot.here == TRUE) & any(y.data2.dot.here == TRUE)){
-x.data2.dot.signif <- ( ! ((data2[, x2] <= tempo.x.data1.right.limit) & (data2[, x2] >= tempo.x.data1.left.limit))) & y.data2.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the x axis?
-x.data2.dot.not.signif <- y.data2.dot.here & ! x.data2.dot.signif
-x.outside.data2.dot.nb <- c(x.outside.data2.dot.nb, data2$DOT_NB[x.data2.dot.signif])
-x.outside.data2.dot.nb <- unique(x.outside.data2.dot.nb)
-x.inside.data2.dot.nb <- c(x.inside.data2.dot.nb, data2$DOT_NB[x.data2.dot.not.signif])
-x.inside.data2.dot.nb <- unique(x.inside.data2.dot.nb)
-}else if(any(y.data1.dot.here == FALSE) & any(y.data2.dot.here == TRUE)){ # recover the data2 dots outside the range of the data1 cloud
-x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[y.data2.dot.here])
-x.unknown.data2.dot.nb <- unique(x.unknown.data2.dot.nb)
-
-
-
-# tempo.warn <- paste0("FROM FUNCTION ", function.name, ": THE [", round(min.pos, 3), " ; ", round(max.pos, 3), "] INTERVAL DOES NOT CONTAIN data1 Y VALUES BUT CONTAINS data2 Y VALUES WHICH CANNOT BE EVALUATED.\nTHE CONCERNED data2 ROW NUMBERS ARE:\n", paste(which(y.data1.dot.here == FALSE & y.data2.dot.here == TRUE), collapse = "\n"))
-# warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end recover the data2 dots outside the frame
-# if(any(i1 == seq(1, loop.nb, 500))){
-# loop.fin.time <- as.numeric(Sys.time()) # time of process end
-# cat(paste0("COMPUTATION TIME OF LOOP ", i1, " / ", loop.nb, ": ", as.character(lubridate::seconds_to_period(round(loop.fin.time - loop.ini.time))), "\n"))
-# }
-}
-if(max.pos < y.range[2]){
-tempo.cat <- paste0("ERROR IN ", function.name, ": THE SLIDING WINDOW HAS NOT REACHED THE MAX VALUE OF data1 ON THE Y-AXIS: ", max.pos, " VERSUS ", y.range[2])
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-x.incon.data1.dot.nb.final <- unique(c(x.outside.data1.dot.nb[x.outside.data1.dot.nb %in% x.inside.data1.dot.nb], x.inside.data1.dot.nb[x.inside.data1.dot.nb %in% x.outside.data1.dot.nb])) # inconsistent dots: if a row number of x.inside.data1.dot.nb is present in x.outside.data1.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
-x.outside.data1.dot.nb.final <- x.outside.data1.dot.nb[ ! (x.outside.data1.dot.nb %in% x.incon.data1.dot.nb.final)] # inconsistent dots removed from the outside list
-x.inside.data1.dot.nb.final <- x.inside.data1.dot.nb[ ! (x.inside.data1.dot.nb %in% x.incon.data1.dot.nb.final)] # inconsistent dots removed from the inside list
-if( ! is.null(data2)){
-# if some unknown dots are also inside, and/or outside, they are put in the inside and/or outside. Ok, because then the intersection between inside and outside is treated -> inconsistent dots
-tempo.unknown.out <- x.unknown.data2.dot.nb[x.unknown.data2.dot.nb %in% x.outside.data2.dot.nb]
-x.outside.data2.dot.nb <- unique(c(x.outside.data2.dot.nb, tempo.unknown.out)) # if a row number of x.unknown.data2.dot.nb is present in x.outside.data2.dot.nb, it is put into outside
-tempo.unknown.in <- x.unknown.data2.dot.nb[x.unknown.data2.dot.nb %in% x.inside.data2.dot.nb]
-x.inside.data2.dot.nb <- unique(c(x.inside.data2.dot.nb, tempo.unknown.in)) # if a row number of x.unknown.data2.dot.nb is present in x.inside.data2.dot.nb, it is put into inside
-x.unknown.data2.dot.nb.final <- x.unknown.data2.dot.nb[ ! (x.unknown.data2.dot.nb %in% c(x.outside.data2.dot.nb, x.inside.data2.dot.nb))] # then dots also in inside and outside are remove from unknown
-x.incon.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb[x.outside.data2.dot.nb %in% x.inside.data2.dot.nb], x.inside.data2.dot.nb[x.inside.data2.dot.nb %in% x.outside.data2.dot.nb])) # inconsistent dots: if a row number of x.inside.data2.dot.nb is present in x.outside.data2.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
-x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb[ ! (x.outside.data2.dot.nb %in% x.incon.data2.dot.nb.final)] # inconsistent dots removed from the outside list
-x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb[ ! (x.inside.data2.dot.nb %in% x.incon.data2.dot.nb.final)] # inconsistent dots removed from the inside list
-}
-# end y-axis sliding and x-axis limits of the data1 cloud -> x significant data2
-}
-# end Method using y unit interval
-
-
-
-# recovering the frame coordinates
-hframe = rbind(
-data.frame(
-x = if(is.null(x.data1.l)){NULL}else{x.data1.l},
-y = if(is.null(x.data1.l)){NULL}else{y.data1.down.limit.l},
-kind = if(is.null(x.data1.l)){NULL}else{"down.frame1"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(x.data1.r)){NULL}else{x.data1.r},
-y = if(is.null(x.data1.r)){NULL}else{y.data1.down.limit.r},
-kind = if(is.null(x.data1.r)){NULL}else{"down.frame2"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(x.data1.l)){NULL}else{x.data1.l},
-y = if(is.null(x.data1.l)){NULL}else{y.data1.top.limit.l},
-kind = if(is.null(x.data1.l)){NULL}else{"top.frame1"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(x.data1.r)){NULL}else{x.data1.r},
-y = if(is.null(x.data1.r)){NULL}else{y.data1.top.limit.r},
-kind = if(is.null(x.data1.r)){NULL}else{"top.frame2"},
-stringsAsFactors = TRUE
-),
-stringsAsFactors = TRUE
-)
-vframe = rbind(
-data.frame(
-x = if(is.null(y.data1.d)){NULL}else{x.data1.left.limit.d},
-y = if(is.null(y.data1.d)){NULL}else{y.data1.d},
-kind = if(is.null(y.data1.d)){NULL}else{"left.frame1"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(y.data1.t)){NULL}else{x.data1.left.limit.t},
-y = if(is.null(y.data1.t)){NULL}else{y.data1.t},
-kind = if(is.null(y.data1.t)){NULL}else{"left.frame2"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(y.data1.d)){NULL}else{x.data1.right.limit.d},
-y = if(is.null(y.data1.d)){NULL}else{y.data1.d},
-kind = if(is.null(y.data1.d)){NULL}else{"right.frame1"},
-stringsAsFactors = TRUE
-),
-data.frame(
-x = if(is.null(y.data1.t)){NULL}else{x.data1.right.limit.t},
-y = if(is.null(y.data1.t)){NULL}else{y.data1.t},
-kind = if(is.null(y.data1.t)){NULL}else{"right.frame2"},
-stringsAsFactors = TRUE
-),
-stringsAsFactors = TRUE
-)
-# end recovering the frame coordinates
-# recovering the dot coordinates
-data1.signif.dot <- NULL
-data1.non.signif.dot <- NULL
-data1.incon.dot <- NULL
-data2.signif.dot <- NULL
-data2.non.signif.dot <- NULL
-data2.unknown.dot <- NULL
-data2.incon.dot <- NULL
-if(( ! is.null(x.range.split)) & ( ! is.null(y.range.split))){
-# inconsistent dots recovery
-if(length(unique(c(x.incon.data1.dot.nb.final, y.incon.data1.dot.nb.final))) > 0){
-data1.incon.dot <- data1[data1$DOT_NB %in% unique(c(x.incon.data1.dot.nb.final, y.incon.data1.dot.nb.final)), ] # if a dot in inconsistent in x or y -> classified as inconsistent (so unique() used)
-# removal of the inconsistent dot in the other classifications
-x.inside.data1.dot.nb.final <- x.inside.data1.dot.nb.final[ ! x.inside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-y.inside.data1.dot.nb.final <- y.inside.data1.dot.nb.final[ ! y.inside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-x.outside.data1.dot.nb.final <- x.outside.data1.dot.nb.final[ ! x.outside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-y.outside.data1.dot.nb.final <- y.outside.data1.dot.nb.final[ ! y.outside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-x.unknown.data1.dot.nb.final <- x.unknown.data1.dot.nb.final[ ! x.unknown.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-y.unknown.data1.dot.nb.final <- y.unknown.data1.dot.nb.final[ ! y.unknown.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
-# end removal of the inconsistent dot in the other classifications
-}
-if( ! is.null(data2)){
-if(length(unique(c(x.incon.data2.dot.nb.final, y.incon.data2.dot.nb.final))) > 0){
-data2.incon.dot <- data2[data2$DOT_NB %in% unique(c(x.incon.data2.dot.nb.final, y.incon.data2.dot.nb.final)), ]
-# removal of the inconsistent dot in the other classifications
-x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-x.unknown.data2.dot.nb.final <- x.unknown.data2.dot.nb.final[ ! x.unknown.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-y.unknown.data2.dot.nb.final <- y.unknown.data2.dot.nb.final[ ! y.unknown.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
-# end removal of the inconsistent dot in the other classifications
-}
-}
-# end inconsistent dots recovery
-# unknown dots recovery
-if( ! is.null(data2)){
-if(data2.pb.dot == "signif"){
-x.outside.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
-x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% x.unknown.data2.dot.nb.final] # remove x.unknown.data2.dot.nb.final from x.inside.data2.dot.nb.final
-y.outside.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
-y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% y.unknown.data2.dot.nb.final] # remove y.unknown.data2.dot.nb.final from y.inside.data2.dot.nb.final
-x.unknown.data2.dot.nb.final <- NULL
-y.unknown.data2.dot.nb.final <- NULL
-data2.unknown.dot <- NULL
-}else if(data2.pb.dot == "not.signif"){
-x.inside.data2.dot.nb.final <- unique(c(x.inside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
-x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% x.unknown.data2.dot.nb.final] # remove x.unknown.data2.dot.nb.final from x.outside.data2.dot.nb.final
-y.inside.data2.dot.nb.final <- unique(c(y.inside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
-y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% y.unknown.data2.dot.nb.final] # remove y.unknown.data2.dot.nb.final from y.outside.data2.dot.nb.final
-x.unknown.data2.dot.nb.final <- NULL
-y.unknown.data2.dot.nb.final <- NULL
-data2.unknown.dot <- NULL
-}else if(data2.pb.dot == "unknown"){
-if(length(unique(c(x.unknown.data2.dot.nb.final, y.unknown.data2.dot.nb.final))) > 0){
-data2.unknown.dot <- data2[data2$DOT_NB %in% unique(c(x.unknown.data2.dot.nb.final, y.unknown.data2.dot.nb.final)), ] # if a dot in unknown in x or y -> classified as unknown (so unique() used)
-x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove x.unknown.data2.dot.nb.final from x.outside.data2.dot.nb.final
-x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove x.unknown.data2.dot.nb.final from x.inside.data2.dot.nb.final
-y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove y.unknown.data2.dot.nb.final from y.outside.data2.dot.nb.final
-y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove y.unknown.data2.dot.nb.final from y.inside.data2.dot.nb.final
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end unknown dots recovery
-# sign and non sign dot recovery
-if(xy.cross.kind == "|"){ # here the problem is to deal with significant dots depending on x and y. Thus I start with that, recover dots finally non significant in outside and put them in inside (when &), and remove from inside the dots in outside
-if(length(unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final))) > 0){
-tempo.outside <- unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final)) # union so unique() used
-tempo.inside <- unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final))
-tempo.inside <- tempo.inside[ ! tempo.inside %in% tempo.outside]
-data1.signif.dot <- data1[data1$DOT_NB %in% tempo.outside, ]
-data1.non.signif.dot <- data1[data1$DOT_NB %in% tempo.inside, ]
-}else{
-data1.non.signif.dot <- data1[unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
-}
-}else if(xy.cross.kind == "&"){
-if(sum(x.outside.data1.dot.nb.final %in% y.outside.data1.dot.nb.final) > 0){ # that is intersection
-tempo.outside <- unique(x.outside.data1.dot.nb.final[x.outside.data1.dot.nb.final %in% y.outside.data1.dot.nb.final]) # intersection
-tempo.outside.removed <- unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final))[ ! unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final)) %in% tempo.outside]
-tempo.inside <- unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final))
-data1.signif.dot <- data1[data1$DOT_NB %in% tempo.outside, ]
-data1.non.signif.dot <- data1[data1$DOT_NB %in% tempo.inside, ]
-}else{
-data1.non.signif.dot <- data1[unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(data2)){
-if(xy.cross.kind == "|"){ # here the problem is to deal with significant dots depending on x and y. Thus I start with that, recover dots finally non significant in outside and put them in inside (when &), and remove from inside the dots in outside
-if(length(unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final))) > 0){
-tempo.outside <- unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final)) # union so unique() used
-tempo.inside <- unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final))
-tempo.inside <- tempo.inside[ ! tempo.inside %in% tempo.outside]
-data2.signif.dot <- data2[data2$DOT_NB %in% tempo.outside, ]
-data2.non.signif.dot <- data2[data2$DOT_NB %in% tempo.inside, ]
-}else{
-data2.non.signif.dot <- data2[unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
-}
-}else if(xy.cross.kind == "&"){
-if(sum(x.outside.data2.dot.nb.final %in% y.outside.data2.dot.nb.final) > 0){ # that is intersection
-tempo.outside <- unique(x.outside.data2.dot.nb.final[x.outside.data2.dot.nb.final %in% y.outside.data2.dot.nb.final]) # intersection
-tempo.outside.removed <- unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final))[ ! unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final)) %in% tempo.outside]
-tempo.inside <- unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final))
-data2.signif.dot <- data2[data2$DOT_NB %in% tempo.outside, ]
-data2.non.signif.dot <- data2[data2$DOT_NB %in% tempo.inside, ]
-}else{
-data2.non.signif.dot <- data2[unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end sign and non sign dot recovery
-}else if(( ! is.null(x.range.split)) & is.null(y.range.split)){
-# inconsistent dots recovery
-if(length(y.incon.data1.dot.nb.final) > 0){
-data1.incon.dot <- data1[data1$DOT_NB %in% y.incon.data1.dot.nb.final, ]
-}
-if( ! is.null(data2)){
-if(length(y.incon.data2.dot.nb.final) > 0){
-data2.incon.dot <- data2[data2$DOT_NB %in% y.incon.data2.dot.nb.final, ]
-}
-}# end inconsistent dots recovery
-# unknown dots recovery
-if( ! is.null(data2)){
-if(data2.pb.dot == "signif"){
-y.outside.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
-}else if(data2.pb.dot == "not.signif"){
-y.inside.data2.dot.nb.final <- unique(c(y.inside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
-}else if(data2.pb.dot == "unknown"){
-if(length(y.unknown.data2.dot.nb.final) > 0){
-data2.unknown.dot <- data2[data2$DOT_NB %in% y.unknown.data2.dot.nb.final, ]
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 6")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end unknown dots recovery
-# sign and non sign dot recovery
-if(length(y.outside.data1.dot.nb.final) > 0){
-data1.signif.dot <- data1[data1$DOT_NB %in% y.outside.data1.dot.nb.final, ]
-}
-if(length(y.inside.data1.dot.nb.final) > 0){
-data1.non.signif.dot <- data1[data1$DOT_NB %in% y.inside.data1.dot.nb.final, ]
-}
-if( ! is.null(data2)){
-if(length(y.outside.data2.dot.nb.final) > 0){
-data2.signif.dot <- data2[data2$DOT_NB %in% y.outside.data2.dot.nb.final, ]
-}
-if(length(y.inside.data2.dot.nb.final) > 0){
-data2.non.signif.dot <- data2[data2$DOT_NB %in% y.inside.data2.dot.nb.final, ]
-}
-}
-# end sign and non sign dot recovery
-}else if(is.null(x.range.split) & ( ! is.null(y.range.split))){
-# inconsistent dots recovery
-if(length(x.incon.data1.dot.nb.final) > 0){
-data1.incon.dot <- data1[data1$DOT_NB %in% x.incon.data1.dot.nb.final, ]
-}
-if( ! is.null(data2)){
-if(length(x.incon.data2.dot.nb.final) > 0){
-data2.incon.dot <- data2[data2$DOT_NB %in% x.incon.data2.dot.nb.final, ]
-}
-}# end inconsistent dots recovery
-# unknown dots recovery
-if( ! is.null(data2)){
-if(data2.pb.dot == "signif"){
-x.outside.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
-}else if(data2.pb.dot == "not.signif"){
-x.inside.data2.dot.nb.final <- unique(c(x.inside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
-}else if(data2.pb.dot == "unknown"){
-if(length(x.unknown.data2.dot.nb.final) > 0){
-data2.unknown.dot <- data2[data2$DOT_NB %in% x.unknown.data2.dot.nb.final, ]
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 7")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end unknown dots recovery
-# sign and non sign dot recovery
-if(length(x.outside.data1.dot.nb.final) > 0){
-data1.signif.dot <- data1[data1$DOT_NB %in% x.outside.data1.dot.nb.final, ]
-}
-if(length(x.inside.data1.dot.nb.final) > 0){
-data1.non.signif.dot <- data1[data1$DOT_NB %in% x.inside.data1.dot.nb.final, ]
-}
-if( ! is.null(data2)){
-if(length(x.outside.data2.dot.nb.final) > 0){
-data2.signif.dot <- data2[data2$DOT_NB %in% x.outside.data2.dot.nb.final, ]
-}
-if(length(x.inside.data2.dot.nb.final) > 0){
-data2.non.signif.dot <- data2[data2$DOT_NB %in% x.inside.data2.dot.nb.final, ]
-}
-}
-# end sign and non sign dot recovery
-}
-# end recovering the dot coordinates
-# verif
-if(any(data1.signif.dot$DOT_NB %in% data1.non.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", FUNCTION.NAME, ": CODE INCONSISTENCY 8")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1.non.signif.dot$DOT_NB %in% data1.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", FUNCTION.NAME, ": CODE INCONSISTENCY 9")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1.signif.dot$DOT_NB %in% data1.incon.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 10")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1.incon.dot$DOT_NB %in% data1.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 11")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1.non.signif.dot$DOT_NB %in% data1.incon.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 12")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data1.incon.dot$DOT_NB %in% data1.non.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 13")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(data2)){
-if(any(data2.signif.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 14")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.non.signif.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 15")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.signif.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 16")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.unknown.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 17")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.signif.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 18")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.incon.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 19")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.non.signif.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 20")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.unknown.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 21")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.non.signif.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 22")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.incon.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 23")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.unknown.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 24")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(data2.incon.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 25")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end verif
-# plot
-# recovering the axes data whatever plot or not
-if(is.null(data2)){
-axes <- fun_gg_scatter(data1 = list(data1), x = list(x1), y = list(y1), categ = list(NULL), color = list(fun_gg_palette(2)[2]), geom = list("geom_point"), alpha = list(0.5), x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, plot = FALSE, return = TRUE)$axes
-}else{
-axes <- fun_gg_scatter(data1 = list(data1, data2), x = list(x1, x2), y = list(y1, y2), categ = list(NULL, NULL), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1]), geom = list("geom_point", "geom_point"), alpha = list(0.5, 0.5), x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, plot = FALSE, return = TRUE)$axes
-}
-# end recovering the axes data whatever plot or not
-if(plot == TRUE){
-# add a categ for plot legend
-tempo.df.name <- c("data1", "data1.signif.dot", "data1.incon.dot", "data2", "data2.signif.dot", "data2.unknown.dot", "data2.incon.dot")
-tempo.class.name <- c("data1", "data1", "data1", "data2", "data2", "data2", "data2")
-for(i2 in 1:length(tempo.df.name)){
-if( ! is.null(get(tempo.df.name[i2], env = sys.nframe(), inherit = FALSE))){
-assign(tempo.df.name[i2], data.frame(get(tempo.df.name[i2], env = sys.nframe(), inherit = FALSE), kind = tempo.class.name[i2]),
-stringsAsFactors = TRUE)
-}
-}
-# end add a categ for plot legend
-if(( ! is.null(x.range.split)) & ( ! is.null(y.range.split))){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe), x = list(x1, "x", "x"), y = list(y1, "y", "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data1.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe, data1.signif.dot), x = list(x1, "x", "x", x1), y = list(y1, "y", "y", y1), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
-}
-if( ! is.null(data1.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe, data1.incon.dot), x = list(x1, "x", "x", x1), y = list(y1, "y", "y", y1), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
-}
-if( ! is.null(data2)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, hframe , vframe), x = list(x1, x2, "x", "x"), y = list(y1, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data2.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
-}
-if( ! is.null(data2.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
-}
-if( ! is.null(data2.unknown.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 12, title = "DATA2 + DATA2 UNKNOWN DOTS")
-}
-}
-}else if(( ! is.null(x.range.split)) & is.null(y.range.split)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe), x = list(x1, "x"), y = list(y1, "y"), categ = list("kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_path"), alpha = list(0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data1.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, data1.signif.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
-}
-if( ! is.null(data1.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, data1.incon.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
-}
-if( ! is.null(data2)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, hframe), x = list(x1, x2, "x"), y = list(y1, y2, "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data2.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
-}
-if( ! is.null(data2.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
-}
-if( ! is.null(data2.unknown.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 8, title = "DATA2 + DATA2 UNKNOWN DOTS")
-}
-}
-}else if(is.null(x.range.split) & ( ! is.null(y.range.split))){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe), x = list(x1, "x"), y = list(y1, "y"), categ = list("kind", "kind"), legend.name = list("DATASET", "VERT FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_path"), alpha = list(0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data1.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe, data1.signif.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "VERT FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
-}
-if( ! is.null(data1.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe, data1.incon.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "VERT FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
-}
-if( ! is.null(data2)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, vframe), x = list(x1, x2, "x"), y = list(y1, y2, "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(data2.signif.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
-}
-if( ! is.null(data2.incon.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
-}
-if( ! is.null(data2.unknown.dot)){
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
-if( ! is.null(tempo.graph$warn)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(graph.in.file == FALSE){
-fun_open(pdf = FALSE)
-}
-fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 8, title = "DATA2 + DATA2 UNKNOWN DOTS")
-}
-}
-}
-}
-# end plot
-if(warn.print == TRUE & ! is.null(warn)){
-options(warning.length = 8170)
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-tempo.list <- list(data1.removed.row.nb = data1.removed.row.nb, data1.removed.rows = data1.removed.rows, data2.removed.row.nb = data2.removed.row.nb, data2.removed.rows = data2.removed.rows, hframe = hframe, vframe = vframe, data1.signif.dot = data1.signif.dot, data1.non.signif.dot = data1.non.signif.dot, data1.inconsistent.dot = data1.incon.dot, data2.signif.dot = data2.signif.dot, data2.non.signif.dot = data2.non.signif.dot, data2.unknown.dot = data2.unknown.dot, data2.inconsistent.dot = data2.incon.dot, axes = axes, warn = warn)
-return(tempo.list)
-}
-
-
-################ Import
-
-
-######## fun_pack() #### check if R packages are present and import into the working environment
-
-
-fun_pack <- function(
-req.package,
-load = FALSE,
-lib.path = NULL
-){
-# AIM
-# check if the specified R packages are present in the computer and import them into the working environment
-# ARGUMENTS
-# req.package: character vector of package names to import
-# load: logical. Load the package into the environement (using library())? Interesting if packages are not in default folders or for checking the functions names of packages using search()
-# lib.path: optional character vector specifying the absolute pathways of the directories containing some of the listed packages in the req.package argument, if not in the default directories. Ignored if NULL
-# RETURN
-# nothing
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_pack(req.package = "nopackage")
-# fun_pack(req.package = "ggplot2")
-# fun_pack(req.package = "ggplot2", lib.path = "blablabla")
-# DEBUGGING
-# req.package = "ggplot2" ; lib.path = "C:/Program Files/R/R-3.5.1/library"
-# req.package = "serpentine" ; lib.path = "C:/users/gael/appdata/roaming/python/python36/site-packages"
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = req.package, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = load, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-if(is.null(lib.path)){
-lib.path <- .libPaths() # .libPaths(new = lib.path) # or .libPaths(new = c(.libPaths(), lib.path))
-}else{
-.libPaths(new = sub(x = lib.path, pattern = "/$|\\\\$", replacement = "")) # .libPaths(new = ) add path to default path. BEWARE: .libPaths() does not support / at the end of a submitted path. Thus check and replace last / or \\ in path
-lib.path <- .libPaths()
-}
-tempo <- NULL
-for(i1 in 1:length(req.package)){
-if( ! req.package[i1] %in% rownames(utils::installed.packages(lib.loc = lib.path))){
-tempo <- c(tempo, req.package[i1])
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0(
-"ERROR IN ",
-function.name,
-": PACKAGE",
-ifelse(length(tempo) == 1L, paste0("\n\n", tempo, "\n\n"), paste0("S\n", paste(tempo, collapse = "\n"), "\n")),
-"MUST BE INSTALLED IN",
-ifelse(length(lib.path) == 1L, "", " ONE OF THESE FOLDERS"),
-":\n",
-paste(lib.path, collapse = "\n")
-)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(load == TRUE){
-for(i2 in 1:length(req.package)){
-suppressMessages(suppressWarnings(suppressPackageStartupMessages(library(req.package[i2], lib.loc = lib.path, quietly = TRUE, character.only = TRUE))))
-}
-}
-}
-
-
-######## fun_python_pack() #### check if python packages are present
-
-
-fun_python_pack <- function(
-req.package,
-python.exec.path = NULL,
-lib.path = NULL,
-R.lib.path = NULL
-){
-# AIM
-# check if the specified python packages are present in the computer (no import)
-# WARNINGS
-# for python 3.7. Previous versions return an error "Error in sys$stdout$flush() : attempt to apply non-function"
-# ARGUMENTS
-# req.package: character vector of package names to import
-# python.exec.path: optional character vector specifying the absolute pathways of the executable python file to use (associated to the packages to use). If NULL, the reticulate::import_from_path() function used in fun_python_pack() seeks for an available version of python.exe, and then uses python_config(python_version, required_module, python_versions). But might not be the correct one for the lib.path parameter specified. Thus, it is recommanded to do not leave NULL, notably when using computing clusters
-# lib.path: optional character vector specifying the absolute pathways of the directories containing some of the listed packages in the req.package argument, if not in the default directories
-# R.lib.path: absolute path of the reticulate packages, if not in the default folders
-# RETURN
-# nothing
-# REQUIRED PACKAGES
-# reticulate
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# fun_pack()
-# EXAMPLES
-# example of error message
-# fun_python_pack(req.package = "nopackage")
-# example without error message (require the installation of the python serpentine package from https://github.com/koszullab/serpentine
-# fun_python_pack(req.package = "serpentine", python.exec.path = "C:/ProgramData/Anaconda3/python.exe", lib.path = "c:/programdata/anaconda3/lib/site-packages/")
-# another example of error message
-# fun_python_pack(req.package = "serpentine", lib.path = "blablabla")
-# DEBUGGING
-# req.package = "serpentine" ; python.exec.path = "C:/ProgramData/Anaconda3/python.exe" ; lib.path = "c:/programdata/anaconda3/lib/site-packages/" ; R.lib.path = NULL
-# req.package = "bad" ; lib.path = NULL ; R.lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(utils::find("fun_pack", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = req.package, class = "character", fun.name = function.name) ; eval(ee)
-if( ! is.null(python.exec.path)){
-tempo <- fun_check(data = python.exec.path, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(file.exists(python.exec.path))){ # separation to avoid the problem of tempo$problem == FALSE and python.exec.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": FILE PATH INDICATED IN THE python.exec.path ARGUMENT DOES NOT EXISTS:\n", paste(python.exec.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if( ! is.null(R.lib.path)){
-tempo <- fun_check(data = R.lib.path, class = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(R.lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and R.lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE R.lib.path ARGUMENT DOES NOT EXISTS:\n", paste(R.lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# package checking
-fun_pack(req.package = "reticulate", lib.path = R.lib.path)
-# end package checking
-# main code
-if(is.null(python.exec.path)){
-python.exec.path <- reticulate::py_run_string("
-import sys ;
-path_lib = sys.path
-") # python string
-python.exec.path <- python.exec.path$path_lib
-}
-if(is.null(lib.path)){
-lib.path <- reticulate::py_run_string("
-import sys ;
-path_lib = sys.path
-") # python string
-lib.path <- lib.path$path_lib
-}
-reticulate::use_python(Sys.which(python.exec.path), required = TRUE) # required to avoid the use of erratic python exec by reticulate::import_from_path()
-for(i1 in 1:length(req.package)){
-tempo.try <- vector("list", length = length(lib.path))
-for(i2 in 1:length(lib.path)){
-tempo.try[[i2]] <- suppressWarnings(try(reticulate::import_from_path(req.package[i1], path = lib.path[i2]), silent = TRUE))
-tempo.try[[i2]] <- suppressWarnings(try(reticulate::import_from_path(req.package[i1], path = lib.path[i2]), silent = TRUE)) # done twice to avoid the error message about flushing present the first time but not the second time. see https://stackoverflow.com/questions/57357001/reticulate-1-13-error-in-sysstdoutflush-attempt-to-apply-non-function
-}
-if(all(sapply(tempo.try, FUN = grepl, pattern = "[Ee]rror"))){
-print(tempo.try)
-tempo.cat <- paste0("ERROR IN ", function.name, ": PACKAGE ", req.package[i1], " MUST BE INSTALLED IN THE MENTIONNED DIRECTORY:\n", paste(lib.path, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-} # else{
-# suppressMessages(suppressWarnings(suppressPackageStartupMessages(assign(req.package[i1], reticulate::import(req.package[i1]))))) # not required because try() already evaluates
-# }
-}
-}
-
-
-################ Print / Exporting results (text & tables)
-
-
-######## fun_report() #### print string or data object into output file
-
-
-# Problem with 1D tables : names over the table not printed. In addition, see how the 2D tables are printed.
-
-fun_report <- function(
-data,
-output = "results.txt",
-path = "C:/Users/Gael/Desktop/",
-overwrite = FALSE,
-rownames.kept = FALSE,
-vector.cat = FALSE,
-noquote = TRUE,
-sep = 2
-){
-# AIM
-# log file function: print a character string or a data object into a same output file
-# ARGUMENTS
-# data: object to print in the output file. If NULL, nothing is done, with no warning
-# output: name of the output file
-# path: location of the output file
-# overwrite: (logical) if output file already exists, defines if the printing is appended (default FALSE) or if the output file content is erased before printing (TRUE)
-# rownames.kept: (logical) defines whether row names have to be removed or not in small tables (less than length.rows rows)
-# vector.cat (logical). If TRUE print a vector of length > 1 using cat() instead of capture.output(). Otherwise (default FALSE) the opposite
-# noquote: (logical). If TRUE no quote are present for the characters
-# sep: number of separating lines after printed data (must be integer)
-# RETURN
-# nothing
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_report()
-# fun_report(data = 1:3, output = "results.txt", path = "C:/Users/Gael/Desktop", overwrite = TRUE, rownames.kept = FALSE, vector.cat = FALSE, noquote = FALSE, sep = 2)
-# DEBUGGING
-# data = 1:3 ; output = "results.txt" ; path = "C:/Users/Gael/Desktop" ; overwrite = TRUE ; rownames.kept = FALSE ; vector.cat = FALSE ; noquote = FALSE ; sep = 2 # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# argument checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = output, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & output == ""){
-tempo.cat <- paste0("ERROR IN ", function.name, ": output ARGUMENT AS \"\" DOES NOT CORRESPOND TO A VALID FILE NAME")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo <- fun_check(data = path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": path ARGUMENT DOES NOT CORRESPOND TO EXISTING DIRECTORY\n", paste(path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = overwrite, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = rownames.kept, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = vector.cat, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = noquote, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = sep, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# the 4 next lines are inactivated but kept because at a time, I might have a problem with data (solved with data = NULL). These 4 lines are just to know how to detect a missing argument. Important here because if data is not provided, print the code of the data function
-# arg.user.list <- as.list(match.call(expand.dots = FALSE))[-1] # recover all the arguments provided by the function user (excluding the argument with defaults values not provided by the user. Thus, it is really the list indicated by the user)
-# default.arg.list <- formals(fun = sys.function(sys.parent())) # list of all the arguments of the function with their default values (not the values of the user !). It seems that ls() as first line of the function provide the names of the arguments (empty, called, etc., or not)
-# arg.without.default.value <- sapply(default.arg.list, is.symbol) & sapply(sapply(default.arg.list, as.character), identical, "") # logical to detect argument without default values (these are typeof "symbol" and class "name" and empty character
-# if( ! all(names(default.arg.list)[arg.without.default.value] %in% names(arg.user.list))){ # test that the arguments with no null values are provided by the user
-# tempo.cat <- paste0("ERROR IN ", function.name, ": VALUE REQUIRED FOR THESE ARGUMENTS WITH NO DEFAULTS VALUES: ", paste(names(default.arg.list)[arg.without.default.value][ ! names(default.arg.list)[arg.without.default.value] %in% names(arg.user.list)], collapse = " "))
-# stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-# }
-# end argument checking
-# main code
-if( ! is.null(data)){
-if(all(class(data) == "data.frame") | all(class(data) == "table") | all(class(data) %in% c("matrix", "array"))){ # before R4.0.0, it was all(class(data) %in% c("matrix", "data.frame", "table"))
-if(rownames.kept == FALSE & all(class(data) == "data.frame") & nrow(data) != 0 & nrow(data) <= 4){ # for data frames with nrows <= 4
-rownames.output.tables <- ""
-length.rows <- nrow(data)
-for(i in 1:length.rows){ # replace the rownames of the first 4 rows by increasing number of spaces (because identical row names not allowed in data frames). This method cannot be extended to more rows as the printed data frame is shifted on the right because of "big empty rownames"
-rownames.output.tables <- c(rownames.output.tables, paste0(rownames.output.tables[i]," ", collapse=""))
-}
-row.names(data) <- rownames.output.tables[1:length.rows]
-}else if(rownames.kept == FALSE & (all(class(data) == "table") | all(class(data) %in% c("matrix", "array")))){ # before R4.0.0, it was & all(class(data) %in% c("matrix", "table"))
-rownames(data) <- rep("", nrow(data)) # identical row names allowed in matrices and tables
-}
-if(noquote == TRUE){
-utils::capture.output(noquote(data), file=paste0(path, "/", output), append = ! overwrite)
-}else{
-utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
-}
-}else if(is.vector(data) & all(class(data) != "list") & (length(data) == 1L | vector.cat == TRUE)){
-if(noquote == TRUE){
-cat(noquote(data), file= paste0(path, "/", output), append = ! overwrite)
-}else{
-cat(data, file= paste0(path, "/", output), append = ! overwrite)
-}
-}else if(all(base::mode(data) == "character")){ # characters (array, list, factor or vector with vector.cat = FALSE)
-if(noquote == TRUE){
-utils::capture.output(noquote(data), file=paste0(path, "/", output), append = ! overwrite)
-}else{
-utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
-}
-}else{ # other object (S4 for instance, which do not like noquote()
-utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
-}
-sep.final <- paste0(rep("\n", sep), collapse = "")
-write(sep.final, file= paste0(path, "/", output), append = TRUE) # add a sep
-}
-}
-
-
-######## fun_get_message() #### return error/warning/other messages of an expression (that can be exported)
-
-
-fun_get_message <- function(
-data,
-kind = "error",
-header = TRUE,
-print.no = FALSE,
-text = NULL,
-env = NULL
-){
-# AIM
-# evaluate an instruction written between "" and return the first of the error, or warning or standard (non error non warning) messages if ever exist
-# using argument print.no = FALSE, return NULL if no message, which is convenient in some cases
-# WARNINGS
-# Only the first message is returned
-# Always use the env argument when fun_get_message() is used inside functions
-# The function does not prevent printing if print() is used inside the instruction tested. To prevent that, use tempo <- capture.output(error <- fun_get_message(data = "fun_check(data = 'a', class = mean, neg.values = FALSE, print = TRUE)")). The return of fun_get_message() is assigned into error and the printed messages are captured by capture.output() and assigned into tempo. See the examples
-# ARGUMENTS
-# data: character string to evaluate
-# kind: character string. Either "error" to get error messages, or "warning" to get warning messages, or "message" to get non error and non warning messages
-# header: logical. Add a header in the returned message?
-# print.no: logical. Print a message saying that no message reported?
-# text: character string added to the output message (even if no message exists and print.no is TRUE). Inactivated if header is FALSE
-# env: the name of an existing environment. NULL if not required
-# RETURN
-# the message or NULL if no message and print.no is FALSE
-# REQUIRED PACKAGES
-# none
-# REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
-# fun_check()
-# EXAMPLES
-# fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "error", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "warning", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "message", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "wilcox.test()", kind = "error", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "sum(1)", kind = "error", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "message('ahah')", kind = "error", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "message('ahah')", kind = "message", print.no = TRUE, text = "IN A")
-# fun_get_message(data = "ggplot2::ggplot(data = data.frame(X = 1:10, stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()", kind = "message", print.no = TRUE, text = "IN FUNCTION 1")
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_get_message(data = 'fun_gg_boxplot(data = obs1, y = "Time", categ = "Group1")', kind = "message", print.no = TRUE, text = "IN FUNCTION 1")
-# DEBUGGING
-# data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)" ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
-# data = "sum(1)" ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; data = 'fun_gg_boxplot(data1 = obs1, y = "Time", categ = "Group1")' ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
-# data = "message('ahah')" ; kind = "error" ; header = TRUE ; print.no = TRUE ; text = "IN A" ; env = NULL
-# data = 'ggplot2::ggplot(data = data.frame(X = "a", stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()' ; kind = "message" ; header = TRUE ; print.no = FALSE ; text = NULL # for function debugging
-# data = 'ggplot2::ggplot(data = data.frame(X = "a", stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()' ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL # for function debugging
-# data = "emmeans::emmeans(object = emm.rg, specs = contrast.var)" ; kind = "message" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-# end function name
-# required function checking
-if(length(utils::find("fun_check", mode = "function")) == 0L){
-tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# no need to use reserved words to avoid bugs, because it is local, and exists("tempo.warning", inherit = FALSE), never use the scope
-# argument checking
-# argument checking with fun_check()
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = kind, options = c("error", "warning", "message"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = print.no, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = header, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(text)){
-tempo <- fun_check(data = text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-if( ! is.null(env)){
-tempo <- fun_check(data = env, class = "environment", fun.name = function.name) ; eval(ee) #
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# end argument checking with fun_check()
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument checking
-# main code
-pdf(file = NULL) # send plots into a NULL file, no pdf file created
-window.nb <- dev.cur()
-invisible(dev.set(window.nb))
-# last warning cannot be used because suppressWarnings() does not modify last.warning present in the base evironment (created at first warning in a new R session), or warnings() # to reset the warning history : unlockBinding("last.warning", baseenv()) ; assign("last.warning", NULL, envir = baseenv())
-output <- NULL
-tempo.error <- try(suppressMessages(suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))), silent = TRUE) # get error message, not warning or messages
-if(any(class(tempo.error) %in% c("gg", "ggplot"))){
-tempo.error <- try(suppressMessages(suppressWarnings(ggplot2::ggplot_build(tempo.error))), silent = TRUE)[1]
-}
-if(exists("tempo.error", inherit = FALSE) == TRUE){ # inherit = FALSE avoid the portee lexical and thus the declared word
-if( ! all(class(tempo.error) == "try-error")){ # deal with NULL and S4 objects. Old code: ! (all(class(tempo.error) == "try-error") & any(grepl(x = tempo.error, pattern = "^Error|^error|^ERROR"))) but problem with S4 objects. Old code : if((length(tempo.error) > 0 & ! any(grepl(x = tempo.error, pattern = "^Error|^error|^ERROR"))) | (length(tempo.error) == 0) ){ but problem when tempo.error is a list but added this did not work: | ! all(class(tempo.error) == "character")
-tempo.error <- NULL
-}
-}else{
-tempo.error <- NULL
-}
-if(kind == "error" & ! is.null(tempo.error)){ #
-if(header == TRUE){
-tempo.error[1] <- gsub(x = tempo.error[1], pattern = "^Error i|^error i|^ERROR I", replacement = "I")
-output <- paste0("ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.error[1]) #
-}else{
-output <- tempo.error[1] #
-}
-}else if(kind == "error" & is.null(tempo.error) & print.no == TRUE){
-output <- paste0("NO ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-}else if(kind != "error" & ( ! is.null(tempo.error)) & print.no == TRUE){
-output <- paste0("NO ", ifelse(kind == "warning", "WARNING", "STANDARD (NON ERROR AND NON WARNING)"), " MESSAGE BECAUSE OF ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-}else if(is.null(tempo.error)){
-fun.warning.capture <- function(expr){
-# from demo(error.catching) typed in the R console, coming from ?tryCatch
-# see also http://mazamascience.com/WorkingWithData/?p=912
-# return a character string or NULL
-# expr <- wilcox.test.default(c(1, 1, 3), c(1, 2, 4), paired = TRUE)
-W <- NULL
-w.handler <- function(w){ # warning handler
-W <<- w # send to the above env, i.e., the inside of the fun.warning.capture function
-invokeRestart("muffleWarning") # here w.handler() muffles all the warnings. See http://romainfrancois.blog.free.fr/index.php?post/2009/05/20/Disable-specific-warnings to muffle specific warnings and print others
-}
-output <- list(
-value = suppressMessages(withCallingHandlers(tryCatch(expr, error = function(e){e}), warning = w.handler)), # BEWARE: w.handler is a function written without (), like in other functions with FUN argument
-warning = W # processed by w.handler()
-)
-return(if(is.null(output$warning)){NULL}else{as.character(output$warning)})
-}
-tempo.warn <- fun.warning.capture(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))
-# warn.options.ini <- options()$warn ; options(warn = 1) ; tempo.warn <- utils::capture.output({tempo <- suppressMessages(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))}, type = "message") ; options(warn = warn.options.ini) # this recover warnings not messages and not errors but does not work in all enviroments
-tempo.message <- utils::capture.output({
-tempo <- suppressMessages(suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env})))
-if(any(class(tempo) %in% c("gg", "ggplot"))){
-tempo <- ggplot2::ggplot_build(tempo)
-}else{
-tempo <- suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))
-}
-}, type = "message") # recover messages not warnings and not errors
-if(kind == "warning" & ! is.null(tempo.warn)){
-if(length(tempo.warn) > 0){ # to avoid character(0)
-if( ! any(sapply(tempo.warn, FUN = "grepl", pattern = "() FUNCTION:$"))){
-tempo.warn <- paste(unique(tempo.warn), collapse = "\n") # if FALSE, means that the tested data is a special function. If TRUE, means that the data is a standard function. In that case, the output of capture.output() is two strings per warning messages: if several warning messages -> identical first string, which is removed in next messages by unique()
-}else{
-tempo.warn <- paste(tempo.warn, collapse = "\n")
-}
-if(header == TRUE){
-if(any(grepl(x = tempo.warn[[1]], pattern = "^simpleWarning i"))){
-tempo.warn[[1]] <- gsub(x = tempo.warn[[1]], pattern = "^Warning i", replacement = "I")
-}
-if(any(grepl(x = tempo.warn[[1]], pattern = "^Warning i"))){
-tempo.warn[[1]] <- gsub(x = tempo.warn[[1]], pattern = "^Warning i", replacement = "I")
-}
-output <- paste0("WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.warn) #
-}else{
-output <- tempo.warn #
-}
-}else{
-if(print.no == TRUE){
-output <- paste0("NO WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-} # no need else{} here because output is already NULL at first
-}
-}else if(kind == "warning" & is.null(tempo.warn) & print.no == TRUE){
-output <- paste0("NO WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-}else if(kind == "message" & exists("tempo.message", inherit = FALSE) == TRUE){ # inherit = FALSE avoid the portee lexical and thus the declared word
-if(length(tempo.message) > 0){ # if something is returned by capture.ouptput() (only in this env) with a length more than 1
-if(header == TRUE){
-output <- paste0("STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.message) #
-}else{
-output <- tempo.message #
-}
-}else{
-if(print.no == TRUE){
-output <- paste0("NO STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-} # no need else{} here because output is already NULL at first
-}
-}else if(kind == "message" & exists("tempo.message", inherit = FALSE) == FALSE & print.no == TRUE){
-output <- paste0("NO STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
-} # no need else{} here because output is already NULL at first
-} # no need else{} here because output is already NULL at first
-invisible(dev.off(window.nb)) # end send plots into a NULL file
-return(output) # do not use cat() because the idea is to reuse the message
-}
-
-
-
-
-
-# Error: class order not good when a class is removed due to NA
-# Error: line 136 in check 20201126 with add argument
-# Solve this: sometimes error messages can be more than the max display (8170). Thus, check every paste0("ERROR IN ", function.name, and trunck the message if to big. In addition, add at the begining of the warning message that it is too long and see the $warn output for complete message. Add also this into fun_scatter
-# add dot.shape ? See with available aesthetic layers
-# rasterise: https://cran.r-project.org/web/packages/ggrastr/vignettes/Raster_geoms.html
-# add horizontal argument and deal any conflict with vertical argument. Start with horizontal = NULL as default. If ! is.null() -> convert vertical if required
-
-fun_gg_boxplot <- function(
-data1,
-y,
-categ,
-categ.class.order = NULL,
-categ.color = NULL,
-box.legend.name = NULL,
-box.fill = FALSE,
-box.width = 0.5,
-box.space = 0.1,
-box.line.size = 0.75,
-box.notch = FALSE,
-box.alpha = 1,
-box.mean = TRUE,
-box.whisker.kind = "std",
-box.whisker.width = 0,
-dot.color = grey(0.25),
-dot.categ = NULL,
-dot.categ.class.order = NULL,
-dot.legend.name = NULL,
-dot.tidy = FALSE,
-dot.tidy.bin.nb = 50,
-dot.jitter = 0.5,
-dot.seed = 2,
-dot.size = 3,
-dot.alpha = 0.5,
-dot.border.size = 0.5,
-dot.border.color = NULL,
-x.lab = NULL,
-x.angle = 0,
-y.lab = NULL,
-y.lim = NULL,
-y.log = "no",
-y.tick.nb = NULL,
-y.second.tick.nb = 1,
-y.include.zero = FALSE,
-y.top.extra.margin = 0.05,
-y.bottom.extra.margin = 0.05,
-stat.pos = "top",
-stat.mean = FALSE,
-stat.size = 4,
-stat.dist = 5,
-stat.angle = 0,
-vertical = TRUE,
-text.size = 12,
-title = "",
-title.text.size = 8,
-legend.show = TRUE,
-legend.width = 0.5,
-article = TRUE,
-grid = FALSE,
-add = NULL,
-return = FALSE,
-return.ggplot = FALSE,
-return.gtable = TRUE,
-plot = TRUE,
-warn.print = FALSE,
-lib.path = NULL
-){
-# AIM
-# Plot ggplot2 boxplots + dots + means
-# For ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
-# WARNINGS
-# Rows containing NA in data1[, c(y, categ)] will be removed before processing, with a warning (see below)
-# Hinges are not computed like in the classical boxplot() function of R. See https://ggplot2.tidyverse.org/reference/geom_boxplot.html
-# To have a single box, please create a factor column with a single class and specify the name of this column in the categ argument. For a single set of grouped boxes, create a factor column with a single class and specify this column in categ argument as first element (i.e., as categ1, knowing that categ2 must also be specified in this situation). See categ argument below
-# The dot.alpha argument can alter the display of the color boxes when using pdf output
-# Size arguments (box.line.size, dot.size, dot.border.size, stat.size, text.size and title.text.size) are in mm. See Hadley comment in https://stackoverflow.com/questions/17311917/ggplot2-the-unit-of-size. See also http://sape.inf.usi.ch/quick-reference/ggplot2/size). Unit object are not accepted, but conversion can be used (e.g., grid::convertUnit(grid::unit(0.2, "inches"), "mm", valueOnly = TRUE))
-# Display seems to be done twice on Windows devices (like a blink). However, no double plots on pdf devices. Thus, the blink remains mysterious
-# To remove boxes and have only dots, use box.alpha = 0
-# ARGUMENTS
-# data1: data frame containing one column of quantitative values (see the y argument below) and one or two columns of categories (see the categ argument below). Duplicated column names are not allowed
-# y: character string of the data1 column name for y-axis (column containing numeric values). Numeric values will be split according to the classes of the column names indicated in the categ argument to generate the boxes and will also be used to plot the dots
-# categ: vector of character strings of the data1 column name for categories (column of characters or factors). Must be either one or two column names. If a single column name (further referred to as categ1), then one box per class of categ1. If two column names (further referred to as categ1 and categ2), then one box per class of categ2, which form a group of boxes in each class of categ1. WARNING: no empty classes allowed. To have a single box, create a factor column with a single class and specify the name of this column in the categ argument (here, no categ2 in categ argument). For a single set of grouped boxes, create a factor column with a single class and specify this column in categ argument as first element (i.e., as categ1), in addition to the already used category (as categ2 in this situation)
-# categ.class.order: list indicating the order of the classes of categ1 and categ2 represented on the boxplot (the first compartment for categ1 and and the second for categ2). If categ.class.order == NULL, classes are represented according to the alphabetical order. Some compartments can be NULL and others not. See the categ argument for categ1 and categ2 description
-# categ.color: vector of color character string for box frames (see the categ argument for categ1 and categ2 description)
-# If categ.color == NULL, default colors of ggplot2, whatever categ1 and categ2
-# If categ.color is non-null and only categ1 in categ argument, categ.color can be either:
-# (1) a single color string. All the boxes will have this color, whatever the number of classes of categ1
-# (2) a vector of string colors, one for each class of categ1. Each color will be associated according to categ.class.order of categ1
-# (3) a vector or factor of string colors, like if it was one of the column of data1 data frame. WARNING: a single color per class of categ1 and a single class of categ1 per color must be respected
-# Color functions, like grey(), hsv(), etc., are also accepted
-# Positive integers are also accepted instead of character strings, as long as above rules about length are respected. Integers will be processed by fun_gg_palette() using the maximal integer value among all the integers in categ.color (see fun_gg_palette())
-# If categ.color is non-null and categ1 and categ2 are specified, all the rules described above will apply to categ2 instead of categ1 (colors will be determined for boxes inside a group of boxes)
-# box.legend.name: character string of the legend title. If box.legend.name is NULL, then box.legend.name <- categ1 if only categ1 is present, and box.legend.name <- categ2 if categ1 and categ2 are present in the categ argument. Write "" if no legend required. See the categ argument for categ1 and categ2 description
-# box.fill: logical. Fill the box? If TRUE, the categ.color argument will be used to generate filled boxplots (the box frames being black) as well as filled outlier dots (the dot border being controlled by the dot.border.color argument). If all the dots are plotted (argument dot.color other than NULL), they will be over the boxes. If FALSE, the categ.color argument will be used to color the box frames and the outlier dot borders. If all the dots are plotted, they will be beneath the boxes
-# box.width: single numeric value (from 0 to 1) of width of either boxes or group of boxes
-# When categ argument has a single categ1 element (i.e., separate boxes. See the categ argument for categ1 and categ2 description), then each class of categ1 is represented by a single box. In that case, box.width argument defines each box width, from 0 (no box width) to 1 (max box width), but also the space between boxes (the code uses 1 - box.width for the box spaces). Of note, xmin and xmax of the fun_gg_boxplot() output report the box boundaries (around x-axis unit 1, 2, 3, etc., for each box)
-# When categ argument has a two categ1 and categ2 elements (i.e., grouped boxes), box.width argument defines the width allocated for each set of grouped boxes, from 0 (no group width) to 1 (max group width), but also the space between grouped boxes (the code uses 1 - box.width for the spaces). Of note, xmin and xmax of the fun_gg_boxplot() output report the box boundaries (around x-axis unit 1, 2, 3, etc., for each set of grouped box)
-# box.space: single numeric value (from 0 to 1) indicating the box separation inside grouped boxes, when categ argument has a two categ1 and categ2 elements. 0 means no space and 1 means boxes shrunk to a vertical line. Ignored if categ argument has a single categ1 element
-# box.line.size: single numeric value of line width of boxes and whiskers in mm
-# box.notch: logical. Notched boxplot? It TRUE, display notched boxplot, notches corresponding approximately to the 95% confidence interval of the median (the notch interval is exactly 1.58 x Inter Quartile Range (IQR) / sqrt(n), with n the number of values that made the box). If notch intervals between two boxes do not overlap, it can be interpreted as significant median differences
-# box.alpha: single numeric value (from 0 to 1) of box transparency (full transparent to full opaque, respectively). To remove boxplots, use box.alpha = 0
-# box.mean: logical. Add mean value? If TRUE, a diamond-shaped dot, with the horizontal diagonal corresponding to the mean value, is displayed over each boxplot
-# box.whisker.kind: range of the whiskers. Either "no" (no whiskers), or "std" (length of each whisker equal to 1.5 x Inter Quartile Range (IQR)), or "max" (length of the whiskers up or down to the most distant dot)
-# box.whisker.width: single numeric value (from 0 to 1) of the whisker width, with 0 meaning no whiskers and 1 meaning a width equal to the box width
-# dot.color: vector of color character string ruling the dot colors and the dot display. See the example section below for easier understanding of the rules described here
-# If NULL, no dots plotted
-# If "same", the dots will have the same colors as the respective boxplots
-# Otherwise, as in the rule (1), (2) or (3) described in the categ.color argument, except that in the possibility (3), the rule "a single color per class of categ and a single class of categ per color", does not have to be respected (for instance, each dot can have a different color). Colors will also depend on the dot.categ argument. If dot.categ is NULL, then colors will be applied to each class of the last column name specified in categ. If dot.categ is non-NULL, colors will be applied to each class of the column name specified in dot.categ. See examples
-# dot.categ: optional single character string of a column name (further referred to as categ3) of the data1 argument. This column of data1 will be used to generate a legend for dots, in addition to the legend for boxes. See the dot.color argument for details about the way the legend is built using the two dot.categ and dot.color arguments. If NULL, no legend created and the colors of dots will depend on dot.color and categ arguments (as explained in the dot.color argument)
-# dot.categ.class.order: optional vector of character strings indicating the order of the classes of categ3 (see the dot.categ argument). If dot.categ is non-NULL and dot.categ.class.order is NULL, classes are displayed in the legend according to the alphabetical order. Ignored if dot.categ is NULL
-# dot.legend.name: optional character string of the legend title for categ3 (see the dot.categ argument). If dot.legend.name == NULL, dot.categ value is used (name of the column in data1). Write "" if no legend required. Ignored if dot.categ is NULL
-# dot.tidy: logical. Nice dot spreading? If TRUE, use the geom_dotplot() function for a nice representation. WARNING: change the true quantitative coordinates of dots (i.e., y-axis values for vertical display) because of binning. Thus, the gain in aestheticism is associated with a loss in precision that can be very important. If FALSE, dots are randomly spread on the qualitative axis, using the dot.jitter argument (see below) keeping the true quantitative coordinates
-# dot.tidy.bin.nb: positive integer indicating the number of bins (i.e., nb of separations) of the y.lim range. Each dot will then be put in one of the bin, with a diameter of the width of the bin. In other words, increase the number of bins to have smaller dots. Not considered if dot.tidy is FALSE
-# dot.jitter: numeric value (from 0 to 1) of random dot horizontal dispersion (for vertical display), with 0 meaning no dispersion and 1 meaning dispersion in the corresponding box width interval. Not considered if dot.tidy is TRUE
-# dot.seed: integer value that set the random seed. Using the same number will generate the same dot jittering. Write NULL to have different jittering each time the same instruction is run. Ignored if dot.tidy is TRUE
-# dot.size: numeric value of dot diameter in mm. Not considered if dot.tidy is TRUE
-# dot.alpha: numeric value (from 0 to 1) of dot transparency (full transparent to full opaque, respectively)
-# dot.border.size: numeric value of border dot width in mm. Write zero for no dot border. If dot.tidy is TRUE, value 0 remove the border and other values leave the border without size control (geom_doplot() feature)
-# dot.border.color: single character color string defining the color of the dot border (same color for all the dots, whatever their categories). If dot.border.color == NULL, the border color will be the same as the dot color. A single integer is also accepted instead of a character string, that will be processed by fun_gg_palette()
-# x.lab: a character string or expression for x-axis legend. If NULL, character string of categ1 (see the categ argument for categ1 and categ2 description)
-# x.angle: integer value of the text angle for the x-axis numbers, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
-# y.lab: a character string or expression for y-axis legend. If NULL, character string of the y argument
-# y.lim: 2 numeric values indicating the range of the y-axis. Order matters (for inverted axis). If NULL, the range of the x column name of data1 will be used.
-# y.log: either "no", "log2" (values in the y argument column of the data1 data frame will be log2 transformed and y-axis will be log2 scaled) or "log10" (values in the y argument column of the data1 data frame will be log10 transformed and y-axis will be log10 scaled). WARNING: not possible to have horizontal boxes with a log axis, due to a bug in ggplot2 (see https://github.com/tidyverse/ggplot2/issues/881)
-# y.tick.nb: approximate number of desired values labeling the y-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if y.log is "no", then the number of labeling values is set by ggplot2. If NULL and if y.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the y.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for y.lim = c(9, 1200)). WARNING: if non-NULL and if y.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
-# y.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if y.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$y.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
-# y.include.zero: logical. Does y.lim range include 0? Ignored if y.log is "log2" or "log10"
-# y.top.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to y.lim. If different from 0, add the range of the axis multiplied by y.top.extra.margin (e.g., abs(y.lim[2] - y.lim[1]) * y.top.extra.margin) to the top of y-axis
-# y.bottom.extra.margin: idem as y.top.extra.margin but to the bottom of y-axis
-# stat.pos: add the median number above the corresponding box. Either NULL (no number shown), "top" (at the top of the plot region) or "above" (above each box)
-# stat.mean: logical. Display mean numbers instead of median numbers? Ignored if stat.pos is NULL
-# stat.size: numeric value of the stat font size in mm. Ignored if stat.pos is NULL
-# stat.dist: numeric value of the stat distance in percentage of the y-axis range (stat.dist = 5 means move the number displayed at 5% of the y-axis range). Ignored if stat.pos is NULL or "top"
-# stat.angle: integer value of the angle of stat, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
-# vertical: logical. Vertical boxes? WARNING: will be automatically set to TRUE if y.log argument is other than "no". Indeed, not possible to have horizontal boxes with a log axis, due to a bug in ggplot2 (see https://github.com/tidyverse/ggplot2/issues/881)
-# text.size: numeric value of the font size of the (1) axis numbers, (2) axis labels and (3) texts in the graphic legend (in mm)
-# title: character string of the graph title
-# title.text.size: numeric value of the title font size in mm
-# legend.show: logical. Show legend? Not considered if categ argument is NULL, because this already generate no legend, excepted if legend.width argument is non-NULL. In that specific case (categ is NULL, legend.show is TRUE and legend.width is non-NULL), an empty legend space is created. This can be useful when desiring graphs of exactly the same width, whatever they have legends or not
-# legend.width: single proportion (between 0 and 1) indicating the relative width of the legend sector (on the right of the plot) relative to the width of the plot. Value 1 means that the window device width is split in 2, half for the plot and half for the legend. Value 0 means no room for the legend, which will overlay the plot region. Write NULL to inactivate the legend sector. In such case, ggplot2 will manage the room required for the legend display, meaning that the width of the plotting region can vary between graphs, depending on the text in the legend
-# article: logical. If TRUE, use an article theme (article like). If FALSE, use a classic related ggplot theme. Use the add argument (e.g., add = "+ggplot2::theme_classic()" for the exact classic ggplot theme
-# grid: logical. Draw lines in the background to better read the box values? Not considered if article == FALSE (grid systematically present)
-# add: character string allowing to add more ggplot2 features (dots, lines, themes, facet, etc.). Ignored if NULL
-# WARNING: (1) the string must start with "+", (2) the string must finish with ")" and (3) each function must be preceded by "ggplot2::". Example: "+ ggplot2::coord_flip() + ggplot2::theme_bw()"
-# If the character string contains the "ggplot2::theme" string, then the article argument of fun_gg_boxplot() (see above) is ignored with a warning. In addition, some arguments can be overwritten, like x.angle (check all the arguments)
-# Handle the add argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
-# WARNING: the call of objects inside the quotes of add can lead to an error if the name of these objects are some of the fun_gg_boxplot() arguments. Indeed, the function will use the internal argument instead of the global environment object. Example article <- "a" in the working environment and add = '+ ggplot2::ggtitle(article)'. The risk here is to have TRUE as title. To solve this, use add = '+ ggplot2::ggtitle(get("article", envir = .GlobalEnv))'
-# return: logical. Return the graph parameters?
-# return.ggplot: logical. Return the ggplot object in the output list? Ignored if return argument is FALSE. WARNING: always assign the fun_gg_boxplot() function (e.g., a <- fun_gg_boxplot()) if return.ggplot argument is TRUE, otherwise, double plotting is performed. See $ggplot in the RETURN section below for more details
-# return.gtable: logical. Return the ggplot object as gtable of grobs in the output list? Ignored if plot argument is FALSE. Indeed, the graph must be plotted to get the grobs dispositions. See $gtable in the RETURN section below for more details
-# plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
-# warn.print: logical. Print warnings at the end of the execution? ? If FALSE, warning messages are never printed, but can still be recovered in the returned list. Some of the warning messages (those delivered by the internal ggplot2 functions) are not apparent when using the argument plot = FALSE
-# lib.path: character string indicating the absolute path of the required packages (see below). if NULL, the function will use the R library default folders
-# RETURN
-# A boxplot if plot argument is TRUE
-# A list of the graph info if return argument is TRUE:
-# $data: the initial data with graphic information added
-# $stat: the graphic statistics (mostly equivalent to ggplot_build()$data[[2]])
-# $removed.row.nb: which rows have been removed due to NA/Inf detection in y and categ columns (NULL if no row removed)
-# $removed.rows: removed rows (NULL if no row removed)
-# $plot: the graphic box and dot coordinates
-# $dots: dot coordinates
-# $main.box: coordinates of boxes
-# $median: median coordinates
-# $sup.whisker: coordinates of top whiskers (y for base and y.end for extremities)
-# $inf.whisker: coordinates of bottom whiskers (y for base and y.end for extremities)
-# $sup.whisker.edge: coordinates of top whisker edges (x and xend)
-# $inf.whisker.edge: coordinates of bottom whisker edges(x and xend)
-# $mean: diamond mean coordinates (only if box.mean argument is TRUE)
-# $stat.pos: coordinates of stat numbers (only if stat.pos argument is not NULL)
-# y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non-NULL or if y.log argument is different from "no")
-# y.second.tick.values: values of secondary ticks. NULL except if y.second.tick.nb argument is non-NULL or if y.log argument is different from "no")
-# $panel: the variable names used for the panels (NULL if no panels). WARNING: NA can be present according to ggplot2 upgrade to v3.3.0
-# $axes: the x-axis and y-axis info
-# $warn: the warning messages. Use cat() for proper display. NULL if no warning. WARNING: warning messages delivered by the internal ggplot2 functions are not apparent when using the argument plot = FALSE
-# $ggplot: ggplot object that can be used for reprint (use print(...$ggplot) or update (use ...$ggplot + ggplot2::...). NULL if return.ggplot argument is FALSE. Of note, a non-NULL $ggplot in the output list is sometimes annoying as the manipulation of this list prints the plot
-# $gtable: gtable object that can be used for reprint (use gridExtra::grid.arrange(...$ggplot) or with additionnal grobs (see the grob decomposition in the examples). NULL if return.ggplot argument is FALSE. Contrary to $ggplot, a non-NULL $gtable in the output list is not annoying as the manipulation of this list does not print the plot
-# REQUIRED PACKAGES
-# ggplot2
-# gridExtra
-# lemon (in case of use in the add argument)
-# scales
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# fun_check()
-# fun_comp_1d()
-# fun_comp_2d()
-# fun_gg_just()
-# fun_gg_palette()
-# fun_inter_ticks()
-# fun_name_change()
-# fun_pack()
-# fun_round()
-# fun_scale()
-# EXAMPLE
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(20, 100, 10), rnorm(20, 200, 50), rnorm(20, 500, 60), rnorm(20, 100, 50)), Categ1 = rep(c("CAT", "DOG"), times = 40), Categ2 = rep(c("A", "B", "C", "D"), each = 20), Color1 = rep(c("coral", "lightblue"), times = 40), Color2 = rep(c("#9F2108", "#306100", "#007479", "#8500C0"), each = 20), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_gg_boxplot(data1 = obs1, y = "Time", categ = "Categ1")
-# see http
-# DEBUGGING
-# set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Categ1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; set.seed(NULL) ; obs1$Time[1:10] <- NA ; data1 = obs1 ; y = "Time" ; categ = c("Categ1") ; categ.class.order = NULL ; categ.color = NULL ; box.legend.name = NULL ; box.fill = FALSE ; box.width = 0.5 ; box.space = 0.1 ; box.line.size = 0.75 ; box.notch = FALSE ; box.alpha = 1 ; box.mean = TRUE ; box.whisker.kind = "std" ; box.whisker.width = 0 ; dot.color = grey(0.25) ; dot.categ = NULL ; dot.categ.class.order = NULL ; dot.legend.name = NULL ; dot.tidy = FALSE ; dot.tidy.bin.nb = 50 ; dot.jitter = 0.5 ; dot.seed = 2 ; dot.size = 3 ; dot.alpha = 0.5 ; dot.border.size = 0.5 ; dot.border.color = NULL ; x.lab = NULL ; x.angle = 0 ; y.lab = NULL ; y.lim = NULL ; y.log = "no" ; y.tick.nb = NULL ; y.second.tick.nb = 1 ; y.include.zero = FALSE ; y.top.extra.margin = 0.05 ; y.bottom.extra.margin = 0.05 ; stat.pos = "top" ; stat.mean = FALSE ; stat.size = 4 ; stat.dist = 5 ; stat.angle = 0 ; vertical = TRUE ; text.size = 12 ; title = "" ; title.text.size = 8 ; legend.show = TRUE ; legend.width = 0.5 ; article = TRUE ; grid = FALSE ; add = NULL ; return = FALSE ; return.ggplot = FALSE ; return.gtable = TRUE ; plot = TRUE ; warn.print = FALSE ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_comp_2d",
-"fun_gg_just",
-"fun_gg_palette",
-"fun_name_change",
-"fun_pack",
-"fun_check",
-"fun_round",
-"fun_scale",
-"fun_inter_ticks"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function")) == 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words to avoid bugs (names of dataframe columns used in this function)
-reserved.words <- c("categ.check", "categ.color", "dot.color", "dot.categ", "dot.max", "dot.min", "group", "PANEL", "group.check", "MEAN", "tempo.categ1", "tempo.categ2", "text.max.pos", "text.min.pos", "x", "x.y", "y", "y.check", "y_from.dot.max", "ymax", "tidy_group", "binwidth")
-# end reserved words to avoid bugs (used in this function)
-# arg with no default values
-mandat.args <- c(
-"data1",
-"y",
-"categ"
-)
-tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), ")")))
-if(any(tempo)){ # normally no NA for missing() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", "HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = categ, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if( ! is.null(categ.class.order)){
-tempo <- fun_check(data = categ.class.order, class = "list", fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = categ.class.order, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(box.legend.name)){
-tempo <- fun_check(data = box.legend.name, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = box.legend.name, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(categ.color)){
-tempo1 <- fun_check(data = categ.color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = categ.color, class = "factor", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.check.color <- fun_check(data = categ.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, neg.values = FALSE, fun.name = function.name)$problem
-if(tempo.check.color == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(any(categ.color == 0L, na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = categ.color, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = box.fill, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.space, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.line.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.notch, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.alpha, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.mean, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.whisker.kind, options = c("no", "std", "max"), length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = box.whisker.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(dot.color)){
-tempo1 <- fun_check(data = dot.color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = dot.color, class = "factor", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.check.color <- fun_check(data = dot.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, neg.values = FALSE, fun.name = function.name)$problem
-if(tempo.check.color == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(any(dot.color == 0L, na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.color, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(dot.categ)){
-tempo <- fun_check(data = dot.categ, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.categ, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(dot.categ.class.order)){
-tempo <- fun_check(data = dot.categ.class.order, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.categ.class.order, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(dot.legend.name)){
-tempo <- fun_check(data = dot.legend.name, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.legend.name, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = dot.tidy, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dot.tidy.bin.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if(dot.tidy.bin.nb == 0L){ # length and NA checked above
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.tidy.bin.nb ARGUMENT MUST BE A NON-NULL AND POSITVE INTEGER VALUE") # integer possible because dealt above
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-tempo <- fun_check(data = dot.jitter, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(dot.seed)){
-tempo <- fun_check(data = dot.seed, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.seed, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = dot.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dot.alpha, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = dot.border.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if( ! is.null(dot.border.color)){
-tempo1 <- fun_check(data = dot.border.color, class = "vector", mode = "character", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.border.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR STRING STARTING BY #, OR (2) A COLOR NAME GIVEN BY colors(), OR (3) AN INTEGER VALUE")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}else if(tempo1$problem == FALSE & tempo2$problem == TRUE){
-if( ! all(dot.border.color %in% colors() | grepl(pattern = "^#", dot.border.color), na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.border.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR STRING STARTING BY #, OR (2) A COLOR NAME GIVEN BY colors(), OR (3) AN INTEGER VALUE")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.border.color, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(x.lab)){
-tempo1 <- fun_check(data = x.lab, class = "expression", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = x.lab, class = "vector", mode = "character", length = 1, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lab ARGUMENT MUST BE A SINGLE CHARACTER STRING OR EXPRESSION")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x.lab, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = x.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-if( ! is.null(y.lab)){
-tempo1 <- fun_check(data = y.lab, class = "expression", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = y.lab, class = "vector", mode = "character", length = 1, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lab ARGUMENT MUST BE A SINGLE CHARACTER STRING OR EXPRESSION")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.lab, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(y.lim)){
-tempo <- fun_check(data = y.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if(any(is.infinite(y.lim))){ # normally no NA for is.infinite() output
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.lim, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = y.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(y.tick.nb)){
-tempo <- fun_check(data = y.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if(y.tick.nb < 0){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.tick.nb ARGUMENT MUST BE A NON NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(y.second.tick.nb)){
-tempo <- fun_check(data = y.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if(y.second.tick.nb <= 0){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.second.tick.nb ARGUMENT MUST BE A NON NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.second.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = y.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.top.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.bottom.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(stat.pos)){
-tempo <- fun_check(data = stat.pos, options = c("top", "above"), length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = stat.pos, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = stat.mean, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = stat.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = stat.dist, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = stat.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = vertical, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = legend.show, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(legend.width)){
-tempo <- fun_check(data = legend.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = legend.width, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = article, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = grid, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(add)){
-tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = add, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.ggplot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.gtable, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path), na.rm = TRUE)){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = lib.path, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if(any(arg.check) == TRUE){ # normally no NA
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"data1",
-"y",
-"categ",
-"box.fill",
-"box.width",
-"box.space",
-"box.line.size",
-"box.notch",
-"box.alpha",
-"box.mean",
-"box.whisker.kind",
-"box.whisker.width",
-# "dot.color", # inactivated because can be null
-"dot.tidy",
-"dot.tidy.bin.nb",
-"dot.jitter",
-# "dot.seed", # inactivated because can be null
-"dot.size",
-"dot.alpha",
-"dot.border.size",
-"x.angle",
-"y.log",
-# "y.second.tick.nb", # inactivated because can be null
-"y.include.zero",
-"y.top.extra.margin",
-"y.bottom.extra.margin",
-# "stat.pos", # inactivated because can be null
-"stat.mean",
-"stat.size",
-"stat.dist",
-"stat.angle",
-"vertical",
-"text.size",
-"title",
-"title.text.size",
-"legend.show",
-# "legend.width", # inactivated because can be null
-"article",
-"grid",
-"return",
-"return.ggplot",
-"return.gtable",
-"plot",
-"warn.print"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){# normally no NA with is.null()
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# see also Protocol 100-rev0 Parallelization in R.docx
-if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
-tempo.random.seed <- .Random.seed
-on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
-}else{
-on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
-}
-set.seed(dot.seed)
-# end code that protects set.seed() in the global environment
-# warning initiation
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-# end warning initiation
-# other checkings
-if(any(duplicated(names(data1)), na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDUPLICATED COLUMN NAMES OF data1 ARGUMENT NOT ALLOWED:\n", paste(names(data1)[duplicated(names(data1))], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! (y %in% names(data1))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny ARGUMENT MUST BE A COLUMN NAME OF data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-tempo <- fun_check(data = data1[, y], data.name = "y COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
-if(tempo$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny ARGUMENT MUST BE NUMERIC COLUMN IN data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(length(categ) > 2){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg ARGUMENT CANNOT HAVE MORE THAN 2 COLUMN NAMES OF data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! all(categ %in% names(data1))){ # all() without na.rm -> ok because categ cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg ARGUMENT MUST BE COLUMN NAMES OF data1. HERE IT IS:\n", paste(categ, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(length(dot.categ) > 1){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT CANNOT HAVE MORE THAN 1 COLUMN NAMES OF data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! all(dot.categ %in% names(data1))){ # all() without na.rm -> ok because dot.categ cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT MUST BE COLUMN NAMES OF data1. HERE IT IS:\n", paste(dot.categ, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# reserved word checking
-if(any(names(data1) %in% reserved.words, na.rm = TRUE)){
-if(any(duplicated(names(data1)), na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDUPLICATED COLUMN NAMES OF data1 ARGUMENT NOT ALLOWED:\n", paste(names(data1)[duplicated(names(data1))], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(dot.categ)){
-if(dot.categ %in% categ){
-reserved.words <- c(reserved.words, paste0(dot.categ, "_DOT")) # paste0(dot.categ, "_DOT") is added to the reserved words because in such situation, a new column will be added to data1 that is named paste0(dot.categ, "_DOT")
-}
-}
-tempo.output <- fun_name_change(names(data1), reserved.words)
-for(i2 in 1:length(tempo.output$ini)){ # a loop to be sure to take the good ones
-names(data1)[names(data1) == tempo.output$ini[i2]] <- tempo.output$post[i2]
-if(any(y == tempo.output$ini[i2])){ # any() without na.rm -> ok because y cannot be NA (tested above)
-y[y == tempo.output$ini[i2]] <- tempo.output$post[i2]
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN y ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN y ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# WARNING: names of y argument potentially replaced
-if(any(categ == tempo.output$ini[i2])){ # any() without na.rm -> ok because categ cannot be NA (tested above)
-categ[categ == tempo.output$ini[i2]] <- tempo.output$post[i2]
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN categ ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN categ ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# WARNING: names of categ argument potentially replaced
-if( ! is.null(dot.categ)){
-if(any(dot.categ == tempo.output$ini[i2])){ # any() without na.rm -> ok because dot.categ cannot be NA (tested above)
-dot.categ[dot.categ == tempo.output$ini[i2]] <- tempo.output$post[i2]
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN dot.categ ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN dot.categ ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# WARNING: names of dot.categ argument potentially replaced
-}
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") REGARDING COLUMN NAMES REPLACEMENT, THE NAMES\n", paste(tempo.output$ini, collapse = " "), "\nHAVE BEEN REPLACED BY\n", paste(tempo.output$post, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-if( ! (is.null(add) | is.null(tempo.output$ini))){
-if(grepl(x = add, pattern = paste(tempo.output$ini, collapse = "|"))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF COLUMN NAMES OF data1 IN THE add ARGUMENT STRING, THAT CORRESPOND TO RESERVED STRINGS FOR ", function.name, "\nCOLUMN NAMES HAVE TO BE CHANGED\nTHE PROBLEMATIC COLUMN NAMES ARE SOME OF THESE NAMES:\n", paste(tempo.output$ini, collapse = " "), "\nIN THE DATA FRAME OF data1 AND IN THE STRING OF add ARGUMENT, TRY TO REPLACE NAMES BY:\n", paste(tempo.output$post, collapse = " "), "\n\nFOR INFORMATION, THE RESERVED WORDS ARE:\n", paste(reserved.words, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-}
-if( ! (is.null(add))){
-if(any(sapply(X = arg.names, FUN = grepl, x = add), na.rm = TRUE)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NAMES OF ", function.name, " ARGUMENTS DETECTED IN THE add STRING:\n", paste(arg.names[sapply(X = arg.names, FUN = grepl, x = add)], collapse = "\n"), "\nRISK OF WRONG OBJECT USAGE INSIDE ", function.name)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end reserved word checking
-# verif of add
-if( ! is.null(add)){
-if( ! grepl(pattern = "^\\s*\\+", add)){ # check that the add string start by +
-tempo.cat <- paste0("ERROR IN ", function.name, "\nadd ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! grepl(pattern = "(ggplot2|lemon)\\s*::", add)){ #
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOR EASIER FUNCTION DETECTION, add ARGUMENT MUST CONTAIN \"ggplot2::\" OR \"lemon::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! grepl(pattern = ")\\s*$", add)){ # check that the add string finished by )
-tempo.cat <- paste0("ERROR IN ", function.name, "\nadd ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end verif of add
-# management of add containing facet
-facet.categ <- NULL
-if( ! is.null(add)){
-facet.check <- TRUE
-tempo <- unlist(strsplit(x = add, split = "\\s*\\+\\s*(ggplot2|lemon)\\s*::\\s*")) #
-tempo <- sub(x = tempo, pattern = "^facet_wrap", replacement = "ggplot2::facet_wrap")
-tempo <- sub(x = tempo, pattern = "^facet_grid", replacement = "ggplot2::facet_grid")
-tempo <- sub(x = tempo, pattern = "^facet_rep", replacement = "lemon::facet_rep")
-if(any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap"), na.rm = TRUE)){
-tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")])))
-facet.categ <- names(tempo1$params$facets)
-tempo.text <- "facet_wrap OR facet_rep_wrap"
-facet.check <- FALSE
-}else if(grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")){
-tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")])))
-facet.categ <- c(names(tempo1$params$rows), names(tempo1$params$cols))
-tempo.text <- "facet_grid OR facet_rep_grid"
-facet.check <- FALSE
-}
-if(facet.check == FALSE & ! all(facet.categ %in% names(data1))){ # WARNING: all(facet.categ %in% names(data1)) is TRUE when facet.categ is NULL # all() without na.rm -> ok because facet.categ cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF \"", tempo.text, "\" STRING IN THE add ARGUMENT BUT PROBLEM OF VARIABLE DETECTION (COLUMN NAMES OF data1)\nTHE DETECTED VARIABLES ARE:\n", paste(facet.categ, collapse = " "), "\nTHE data1 COLUMN NAMES ARE:\n", paste(names(data1), collapse = " "), "\nPLEASE REWRITE THE add STRING AND RERUN")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end management of add containing facet
-# conversion of categ columns in data1 into factors
-for(i1 in 1:length(categ)){
-tempo1 <- fun_check(data = data1[, categ[i1]], data.name = paste0("categ NUMBER ", i1, " OF data1"), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = data1[, categ[i1]], data.name = paste0("categ NUMBER ", i1, " OF data1"), class = "factor", na.contain = TRUE, fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\n", paste0("categ NUMBER ", i1, " OF data1"), " MUST BE A FACTOR OR CHARACTER VECTOR")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(tempo1$problem == FALSE){ # character vector
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN categ NUMBER ", i1, " IN data1, THE CHARACTER COLUMN HAS BEEN CONVERTED TO FACTOR, WITH LEVELS ACCORDING TO THE ALPHABETICAL ORDER")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-data1[, categ[i1]] <- factor(data1[, categ[i1]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-}
-# OK: all the categ columns of data1 are factors from here
-# end conversion of categ columns in data1 into factors
-
-
-
-# management of log scale and Inf removal
-if(any(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf # normally no NA with is.finite0() and is.na()
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN THE ", y, " COLUMN OF THE data1 ARGUMENT AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-data1.ini <- data1 # strictly identical to data1 except that in data1 y is log converted if and only if y.log != "no"
-if(y.log != "no"){
-tempo1 <- ! is.finite(data1[, y]) # where are initial NA and Inf
-data1[, y] <- suppressWarnings(get(y.log)(data1[, y]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-if(any( ! (tempo1 | is.finite(data1[, y])))){ # normally no NA with is.finite
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN THE ", y, " COLUMN OF THE data1 ARGUMENT AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# Inf removal
-if(any(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf # normally no NA with is.finite
-removed.row.nb <- which(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))
-removed.rows <- data1.ini[removed.row.nb, ] # here data1.ini used to have the y = O rows that will be removed because of Inf creation after log transformation
-data1 <- data1[-removed.row.nb, ] #
-data1.ini <- data1.ini[-removed.row.nb, ] #
-}else{
-removed.row.nb <- NULL
-removed.rows <- data.frame(stringsAsFactors = FALSE)
-}
-# From here, data1 and data.ini have no more Inf
-# end Inf removal
-if(y.log != "no" & ! is.null(y.lim)){
-if(any(y.lim <= 0)){ # any() without na.rm -> ok because y.lim cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(any( ! is.finite(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}))){ # normally no NA with is.finite
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT RETURNS INF/NA WITH THE y.log ARGUMENT SET TO ", y.log, "\nAS SCALE COMPUTATION IS ", ifelse(y.log == "log10", "log10", "log2"), ":\n", paste(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(y.log != "no" & y.include.zero == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") y.log ARGUMENT SET TO ", y.log, " AND y.include.zero ARGUMENT SET TO TRUE -> y.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-y.include.zero <- FALSE
-}
-if(y.log != "no" & vertical == FALSE){
-vertical <- TRUE
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") BECAUSE OF A BUG IN ggplot2, CANNOT FLIP BOXES HORIZONTALLY WITH A Y.LOG SCALE -> vertical ARGUMENT RESET TO TRUE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end management of log scale and Inf removal
-# na detection and removal (done now to be sure of the correct length of categ)
-column.check <- unique(c(y, categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){dot.categ}, if( ! is.null(facet.categ)){facet.categ})) # dot.categ because can be a 3rd column of data1, categ.color and dot.color will be tested later
-if(any(is.na(data1[, column.check]))){ # data1 used here instead of data1.ini in case of new NaN created by log conversion (neg values) # normally no NA with is.na
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS OF data1 AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-for(i2 in 1:length(column.check)){
-if(any(is.na(data1[, column.check[i2]]))){ # normally no NA with is.na
-tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i2], " OF data1")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n", tempo.warn)))
-}
-}
-tempo <- unique(unlist(lapply(lapply(c(data1[column.check]), FUN = is.na), FUN = which)))
-removed.row.nb <- c(removed.row.nb, tempo) # removed.row.nb created to remove Inf
-removed.rows <- rbind(removed.rows, data1.ini[tempo, ], stringsAsFactors = FALSE) # here data1.ini used to have the non NA rows that will be removed because of NAN creation after log transformation (neg values for instance)
-column.check <- column.check[ ! column.check == y] # remove y to keep quali columns
-if(length(tempo) != 0){
-data1 <- data1[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
-data1.ini <- data1.ini[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers than the former
-for(i3 in 1:length(column.check)){
-if(any( ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]]), na.rm = TRUE)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i3], " OF data1, THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA/Inf REMOVAL (IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[, column.check[i3]])[ ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]])], collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-}
-count.categ <- 0
-for(i2 in 1:length(column.check)){
-if(column.check[i2] %in% categ){
-count.categ <- count.categ + 1
-}
-if(column.check[i2] == categ[count.categ]){
-categ.class.order[count.categ] <- list(levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])]) # remove the absent color in the character vector
-data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(categ.class.order[[count.categ]]))
-}
-if( ! is.null(dot.color) & ! is.null(dot.categ)){ # reminder : dot.categ cannot be a column name of categ anymore (because in that case dot.categ name is changed into "..._DOT"
-if(column.check[i2] == dot.categ){
-dot.categ.class.order <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
-data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(dot.categ.class.order))
-}
-}
-if(column.check[i2] %in% facet.categ){ # works if facet.categ == NULL this method should keep the order of levels when removing some levels
-tempo.levels <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(as.character(data1[, column.check[i2]]))]
-data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = tempo.levels)
-}
-}
-}
-# end na detection and removal (done now to be sure of the correct length of categ)
-# From here, data1 and data.ini have no more NA or NaN in y, categ, dot.categ (if dot.color != NULL) and facet.categ
-
-
-
-if( ! is.null(categ.class.order)){
-if(length(categ.class.order) != length(categ)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.class.order ARGUMENT MUST BE A LIST OF LENGTH EQUAL TO LENGTH OF categ\nHERE IT IS LENGTH: ", length(categ.class.order), " VERSUS ", length(categ))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-for(i3 in 1:length(categ.class.order)){
-if(is.null(categ.class.order[[i3]])){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE categ.class.order COMPARTMENT ", i3, " IS NULL. ALPHABETICAL ORDER WILL BE APPLIED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-data1[, categ[i3]] <- factor(as.character(data1[, categ[i3]])) # if already a factor, change nothing, if characters, levels according to alphabetical order
-categ.class.order[[i3]] <- levels(data1[, categ[i3]]) # character vector that will be used later
-}else{
-tempo <- fun_check(data = categ.class.order[[i3]], data.name = paste0("COMPARTMENT ", i3 , " OF categ.class.order ARGUMENT"), class = "vector", mode = "character", length = length(levels(data1[, categ[i3]])), fun.name = function.name) # length(data1[, categ[i1]) -> if data1[, categ[i1] was initially character vector, then conversion as factor after the NA removal, thus class number ok. If data1[, categ[i1] was initially factor, no modification after the NA removal, thus class number ok
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(any(duplicated(categ.class.order[[i3]]), na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i3, " OF categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(categ.class.order[[i3]], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! (all(categ.class.order[[i3]] %in% unique(data1[, categ[i3]]), na.rm = TRUE) & all(unique(data1[, categ[i3]]) %in% categ.class.order[[i3]], na.rm = TRUE))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i3, " OF categ.class.order ARGUMENT MUST BE CLASSES OF ELEMENT ", i3, " OF categ ARGUMENT\nHERE IT IS:\n", paste(categ.class.order[[i3]], collapse = " "), "\nFOR COMPARTMENT ", i3, " OF categ.class.order AND IT IS:\n", paste(unique(data1[, categ[i3]]), collapse = " "), "\nFOR COLUMN ", categ[i3], " OF data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-data1[, categ[i3]] <- factor(data1[, categ[i3]], levels = categ.class.order[[i3]]) # reorder the factor
-
-}
-names(categ.class.order)[i3] <- categ[i3]
-}
-}
-}else{
-categ.class.order <- vector("list", length = length(categ))
-tempo.categ.class.order <- NULL
-for(i2 in 1:length(categ.class.order)){
-categ.class.order[[i2]] <- levels(data1[, categ[i2]])
-names(categ.class.order)[i2] <- categ[i2]
-tempo.categ.class.order <- c(tempo.categ.class.order, ifelse(i2 != 1, "\n", ""), categ.class.order[[i2]])
-}
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR BOX ORDERING:\n", paste(tempo.categ.class.order, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# categ.class.order not NULL anymore (list)
-if(is.null(box.legend.name) & box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE box.legend.name SETTING IS NULL. NAMES OF categ WILL BE USED: ", paste(categ, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-box.legend.name <- categ[length(categ)] # if only categ1, then legend name of categ1, if length(categ) == 2L, then legend name of categ2
-}
-# box.legend.name not NULL anymore (character string)
-# management of categ.color
-if( ! is.null(categ.color)){
-# check the nature of color
-# integer colors into gg_palette
-tempo.check.color <- fun_check(data = categ.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem
-if(tempo.check.color == FALSE){
-# convert integers into colors
-categ.color <- fun_gg_palette(max(categ.color, na.rm = TRUE))[categ.color]
-}
-# end integer colors into gg_palette
-if( ! (all(categ.color %in% colors() | grepl(pattern = "^#", categ.color)))){ # check that all strings of low.color start by #, # all() without na.rm -> ok because categ.color cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors(): ", paste(unique(categ.color), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(is.na(categ.color)) & box.alpha != 0){ # normally no NA with is.na
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT CONTAINS NA")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end check the nature of color
-# check the length of color
-categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
-if(length(data1[, categ[categ.len]]) == length(levels(data1[, categ[categ.len]])) & length(categ.color) == length(data1[, categ[categ.len]])){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE NUMBER OF CLASSES OF THE COLUMN ", categ[categ.len], " THE NUMBER OF ROWS OF THIS COLUMN AND THE NUMBER OF COLORS OF THE categ.color ARGUMENT ARE ALL EQUAL. BOX COLORS WILL BE ATTRIBUTED ACCORDING THE LEVELS OF ", categ[categ.len], ", NOT ACCORDING TO THE ROWS OF ", categ[categ.len])
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if(length(categ.color) == length(levels(data1[, categ[categ.len]]))){ # here length(categ.color) is equal to the different number of categ
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-data1 <- data.frame(data1, categ.color = data1[, categ[categ.len]], stringsAsFactors = TRUE) # no need stringsAsFactors here for stat.nolog as factors remain factors
-data1$categ.color <- factor(data1$categ.color, labels = categ.color) # replace the characters of data1[, categ[categ.len]] put in the categ.color column by the categ.color (can be write like this because categ.color is length of levels of data1[, categ[categ.len]])
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", categ[categ.len], " OF categ ARGUMENT, THE FOLLOWING COLORS:\n", paste(categ.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else if(length(categ.color) == length(data1[, categ[categ.len]])){# here length(categ.color) is equal to nrow(data1) -> Modif to have length(categ.color) equal to the different number of categ (length(categ.color) == length(levels(data1[, categ[categ.len]])))
-data1 <- data.frame(data1, categ.color = categ.color, stringsAsFactors = TRUE)
-tempo.check <- unique(data1[ , c(categ[categ.len], "categ.color")])
-if( ! (nrow(tempo.check) == length(unique(categ.color)) & nrow(tempo.check) == length(unique(data1[ , categ[categ.len]])))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT HAS THE LENGTH OF data1 ROW NUMBER\nBUT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF categ ", categ[categ.len], ":\n", paste(unique(mapply(FUN = "paste", data1[ ,categ[categ.len]], data1[ ,"categ.color"])), collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-categ.color <- unique(data1$categ.color[order(data1[, categ[categ.len]])]) # Modif to have length(categ.color) equal to the different number of categ (length(categ.color) == length(levels(data1[, categ[categ.len]])))
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT HAS THE LENGTH OF data1 ROW NUMBER\nCOLORS HAVE BEEN RESPECTIVELY ASSOCIATED TO EACH CLASS OF categ ", categ[categ.len], " AS:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\n", paste(categ.color, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-}else if(length(categ.color) == 1L){
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-data1 <- data.frame(data1, categ.color = categ.color, stringsAsFactors = TRUE)
-categ.color <- rep(categ.color, length(levels(data1[, categ[categ.len]])))
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT HAS LENGTH 1, MEANING THAT ALL THE DIFFERENT CLASSES OF ", categ[categ.len], "\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(categ.color, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE (1) LENGTH 1, OR (2) THE LENGTH OF data1 NROWS AFTER NA/Inf REMOVAL, OR (3) THE LENGTH OF THE CLASSES IN THE categ ", categ[categ.len], " COLUMN. HERE IT IS COLOR LENGTH ", length(categ.color), " VERSUS CATEG LENGTH ", length(data1[, categ[categ.len]]), " AND CATEG CLASS LENGTH ", length(unique(data1[, categ[categ.len]])), "\nPRESENCE OF NA/Inf COULD BE THE PROBLEM")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-categ.color <- fun_gg_palette(length(levels(data1[, categ[categ.len]])))
-data1 <- data.frame(data1, categ.color = data1[, categ[categ.len]], stringsAsFactors = TRUE)
-data1$categ.color <- factor(data1$categ.color, labels = categ.color) # replace the characters of data1[, categ[categ.len]] put in the categ.color column by the categ.color (can be write like this because categ.color is length of levels of data1[, categ[categ.len]])
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL categ.color ARGUMENT -> COLORS RESPECTIVELY ATTRIBUTED TO EACH CLASS OF ", categ[categ.len], " IN data1:\n", paste(categ.color, collapse = " "), "\n", paste(levels(data1[, categ[categ.len]]), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# categ.color not NULL anymore
-categ.color <- as.character(categ.color)
-# categ.color is a character string representing the diff classes
-data1$categ.color <- factor(data1$categ.color, levels = unique(categ.color)) # ok because if categ.color is a character string, the order make class 1, class 2, etc. unique() because no duplicates allowed
-# data1$categ.color is a factor with order of levels -> categ.color
-# end management of categ.color
-# management of dot.color
-if( ! is.null(dot.color)){
-# optional legend of dot colors
-if( ! is.null(dot.categ)){
-ini.dot.categ <- dot.categ
-if( ! dot.categ %in% names(data1)){ # no need to use all() because length(dot.categ) = 1
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT MUST BE A COLUMN NAME OF data1. HERE IT IS:\n", dot.categ)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(dot.categ %in% categ){ # no need to use all() because length(dot.categ) = 1. Do not use dot.categ %in% categ[length(categ)] -> error
-# management of dot legend if dot.categ %in% categ (because legends with the same name are joined in ggplot2)
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE COLUMN NAME OF data1 INDICATED IN THE dot.categ ARGUMENT (", dot.categ, ") HAS BEEN REPLACED BY ", paste0(dot.categ, "_DOT"), " TO AVOID MERGED LEGEND BY GGPLOT2")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-data1 <- data.frame(data1, dot.categ = data1[, dot.categ], stringsAsFactors = TRUE) # dot.categ is not a column name of data1 (checked above with reserved words)
-dot.categ <- paste0(dot.categ, "_DOT")
-names(data1)[names(data1) == "dot.categ"] <- dot.categ # paste0(dot.categ, "_DOT") is not a column name of data1 (checked above with reserved words)
-# tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT CANNOT BE A COLUMN NAME OF data1 ALREADY SPECIFIED IN THE categ ARGUMENT:\n", dot.categ, "\nINDEED, dot.categ ARGUMENT IS MADE TO HAVE MULTIPLE DOT COLORS NOT RELATED TO THE BOXPLOT CATEGORIES")
-# stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-tempo1 <- fun_check(data = data1[, dot.categ], data.name = paste0(dot.categ, " COLUMN OF data1"), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = data1[, dot.categ], data.name = paste0(dot.categ, " COLUMN OF data1"), class = "factor", na.contain = TRUE, fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ COLUMN MUST BE A FACTOR OR CHARACTER VECTOR") #
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-data1[, dot.categ] <- factor(data1[, dot.categ]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-# dot.categ column of data1 is factor from here
-if( ! is.null(dot.categ.class.order)){
-if(any(duplicated(dot.categ.class.order), na.rm = TRUE)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(dot.categ.class.order, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! (all(dot.categ.class.order %in% levels(data1[, dot.categ])) & all(levels(data1[, dot.categ]) %in% dot.categ.class.order, na.rm = TRUE))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ.class.order ARGUMENT MUST BE CLASSES OF dot.categ ARGUMENT\nHERE IT IS:\n", paste(dot.categ.class.order, collapse = " "), "\nFOR dot.categ.class.order AND IT IS:\n", paste(levels(data1[, dot.categ]), collapse = " "), "\nFOR dot.categ COLUMN (", ini.dot.categ, ") OF data1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
-}
-}else{
-if(all(dot.color == "same") & length(dot.color)== 1L){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
-dot.categ.class.order <- unlist(categ.class.order[length(categ)])
-data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order SETTING IS NULL AND dot.color IS \"same\". ORDER OF categ.class.order WILL BE APPLIED FOR LEGEND DISPLAY: ", paste(dot.categ.class.order, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-dot.categ.class.order <- sort(levels(data1[, dot.categ]))
-data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR LEGEND DISPLAY: ", paste(dot.categ.class.order, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# dot.categ.class.order not NULL anymore (character string) if dot.categ is not NULL
-if(all(dot.color == "same") & length(dot.color)== 1L){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
-if( ! identical(ini.dot.categ, categ[length(categ)])){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN dot.color ARGUMENT IS \"same\", THE COLUMN NAME IN dot.categ ARGUMENT MUST BE IDENTICAL TO THE LAST COLUMN NAME IN categ ARGUMENT. HERE IT IS:\ndot.categ: ", paste(ini.dot.categ, collapse = " "), "\ncateg: ", paste(categ, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if( ! fun_comp_1d(unlist(categ.class.order[length(categ)]), dot.categ.class.order)$identical.content){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN dot.color ARGUMENT IS \"same\",\nLAST COMPARTMENT OF categ.class.order ARGUMENT AND dot.categ.class.order ARGUMENT CANNOT BE DIFFERENT:\nLAST COMPARTMENT OF categ.class.order: ", paste(unlist(categ.class.order[length(categ)]), collapse = " "), "\ndot.categ.class.order: ", paste(dot.categ.class.order, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-for(i3 in 1:length(categ)){
-if(identical(categ[i3], ini.dot.categ) & ! identical(unlist(categ.class.order[i3]), dot.categ.class.order) & identical(sort(unlist(categ.class.order[i3])), sort(dot.categ.class.order))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE dot.categ ARGUMENT SETTING IS PRESENT IN THE categ ARGUMENT SETTING, BUT ORDER OF THE CLASSES IS NOT THE SAME:\ncateg.class.order: ", paste(unlist(categ.class.order[i3]), collapse = " "), "\ndot.categ.class.order: ", paste(dot.categ.class.order, collapse = " "), "\nNOTE THAT ORDER OF categ.class.order IS THE ONE USED FOR THE AXIS REPRESENTATION")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(is.null(dot.legend.name)){
-dot.legend.name <- if(ini.dot.categ %in% categ[length(categ)]){dot.categ}else{ini.dot.categ} #
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE dot.legend.name SETTING IS NULL -> ", dot.legend.name, " WILL BE USED AS LEGEND TITLE OF DOTS")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# dot.legend.name not NULL anymore (character string)
-}else{
-if( ! is.null(dot.categ.class.order)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order ARGUMENT IS NOT NULL, BUT IS THE dot.categ ARGUMENT\n-> dot.categ.class.order NOT CONSIDERED AS NO LEGEND WILL BE DRAWN")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# But dot.categ.class.order will be converted to NULL below (not now)
-}
-# end optional legend of dot colors
-# check the nature of color
-# integer colors into gg_palette
-tempo.check.color <- fun_check(data = dot.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem
-if(tempo.check.color == FALSE){
-# convert integers into colors
-dot.color <- fun_gg_palette(max(dot.color, na.rm = TRUE))[dot.color]
-}
-# end integer colors into gg_palette
-if(all(dot.color == "same") & length(dot.color)== 1L){# all() without na.rm -> ok because dot.color cannot be NA (tested above)
-dot.color <- categ.color # same color of the dots as the corresponding box color
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT HAS BEEN SET TO \"same\"\nTHUS, DOTS WILL HAVE THE SAME COLORS AS THE CORRESPONDING BOXPLOT")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else if( ! (all(dot.color %in% colors() | grepl(pattern = "^#", dot.color)))){ # check that all strings of low.color start by #, # all() without na.rm -> ok because dot.color cannot be NA (tested above)
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR VECTOR STARTING BY #, OR (2) COLOR NAMES GIVEN BY colors(), OR (3) INTEGERS, OR THE STRING \"same\"\nHERE IT IS: ", paste(unique(dot.color), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(any(is.na(dot.color))){ # normally no NA with is.finite
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT CONTAINS NA")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end check the nature of color
-# check the length of color
-if( ! is.null(dot.categ)){
-# optional legend of dot colors
-if(length(data1[, dot.categ]) == length(levels(data1[, dot.categ])) & length(dot.color) == length(data1[, dot.categ])){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE NUMBER OF CLASSES OF THE COLUMN ", dot.categ, " THE NUMBER OF ROWS OF THIS COLUMN AND THE NUMBER OF COLORS OF THE dot.color ARGUMENT ARE ALL EQUAL. DOT COLORS WILL BE ATTRIBUTED ACCORDING THE LEVELS OF ", dot.categ, ", NOT ACCORDING TO THE ROWS OF ", dot.categ)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if(length(dot.color) > 1 & ! (length(dot.color) == length(unique(data1[, dot.categ])) | length(dot.color) == length(data1[, dot.categ]))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN LENGTH OF THE dot.color ARGUMENT IS MORE THAN 1, IT MUST BE EQUAL TO THE NUMBER OF 1) ROWS OR 2) LEVELS OF dot.categ COLUMN (", dot.categ, "):\ndot.color: ", paste(dot.color, collapse = " "), "\ndot.categ LEVELS: ", paste(levels(data1[, dot.categ]), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else if(length(dot.color) > 1 & length(dot.color) == length(unique(data1[, dot.categ]))){
-data1 <- data.frame(data1, dot.color = data1[, dot.categ], stringsAsFactors = TRUE)
-data1$dot.color <- factor(data1$dot.color, labels = dot.color) # do not use labels = unique(dot.color). Otherwise, we can have green1 green2 when dot.color is c("green", "green")
-}else if(length(dot.color) > 1 & length(dot.color) == length(data1[, dot.categ])){
-data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
-}else if(length(dot.color)== 1L){ # to deal with single color. Warning: & length(dot.categ.class.order) > 1 removed because otherwise, the data1 is not with dot.color column when length(dot.categ.class.order) == 1
-data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
-}
-dot.color <- as.character(unique(data1$dot.color[order(data1[, dot.categ])])) # reorder the dot.color character vector
-if(length(dot.color)== 1L & length(dot.categ.class.order) > 1){ # to deal with single color
-dot.color <- rep(dot.color, length(dot.categ.class.order))
-}
-tempo.check <- unique(data1[ , c(dot.categ, "dot.color")])
-if(length(unique(data1[ , "dot.color"])) > 1 & ( ! (nrow(tempo.check) == length(unique(data1[ , "dot.color"])) & nrow(tempo.check) == length(unique(data1[ , dot.categ]))))){ # length(unique(data1[ , "dot.color"])) > 1 because if only one color, can be attributed to each class of dot.categ
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF dot.categ (", dot.categ, ") COLUMN:\n", paste(unique(mapply(FUN = "paste", data1[ , dot.categ], data1[ ,"dot.color"])), collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN dot.categ ARGUMENT (", ini.dot.categ, "), THE FOLLOWING COLORS OF DOTS:\n", paste(dot.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(data1[, dot.categ]), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# dot.color is a character string representing the diff classes of dot.categ
-# data1$dot.color is a factor with order of levels -> dot.categ
-# end optional legend of dot colors
-}else{
-categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
-if(length(dot.color) == length(levels(data1[, categ[categ.len]]))){ # here length(dot.color) is equal to the different number of categ
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-data1 <- data.frame(data1, dot.color = data1[, categ[categ.len]], stringsAsFactors = TRUE)
-data1$dot.color <- factor(data1$dot.color, labels = dot.color)
-if(box.alpha != 0){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", categ[categ.len], " OF categ ARGUMENT, THE FOLLOWING COLORS:\n", paste(dot.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else if(length(dot.color) == length(data1[, categ[categ.len]])){# here length(dot.color) is equal to nrow(data1) -> Modif to have length(dot.color) equal to the different number of categ (length(dot.color) == length(levels(data1[, categ[categ.len]])))
-data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
-}else if(length(dot.color)== 1L & ! all(dot.color == "same")){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
-# data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
-data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
-dot.color <- rep(dot.color, length(levels(data1[, categ[categ.len]])))
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT HAS LENGTH 1, MEANING THAT ALL THE DIFFERENT CLASSES OF ", categ[categ.len], "\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(dot.color, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE (1) LENGTH 1, OR (2) THE LENGTH OF data1 NROWS AFTER NA/Inf REMOVAL, OR (3) THE LENGTH OF THE CLASSES IN THE categ ", categ[categ.len], " COLUMN. HERE IT IS COLOR LENGTH ", length(dot.color), " VERSUS CATEG LENGTH ", length(data1[, categ[categ.len]]), " AND CATEG CLASS LENGTH ", length(unique(data1[, categ[categ.len]])), "\nPRESENCE OF NA/Inf COULD BE THE PROBLEM")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end check the length of color
-dot.color <- as.character(dot.color)
-# dot.color is a character string representing the diff classes
-data1$dot.color <- factor(data1$dot.color, levels = unique(dot.color)) # ok because if dot.color is a character string, the order make class 1, class 2, etc. If dot.color is a column of data1, then levels will be created, without incidence, except if dot.categ specified (see below). unique() because no duplicates allowed
-# data1$dot.color is a factor with order of levels -> dot.color
-}
-# end optional legend of dot colors
-}else if(is.null(dot.color) & ! (is.null(dot.categ) & is.null(dot.categ.class.order) & is.null(dot.legend.name))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") dot.categ OR dot.categ.class.order OR dot.legend.name ARGUMENT HAS BEEN SPECIFIED BUT dot.color ARGUMENT IS NULL (NO DOT PLOTTED)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# dot.color either NULL (no dot plotted) or character string (potentially representing the diff classes of dot.categ)
-# data1$dot.color is either NA or a factor (with order of levels -> depending on dot.categ or categ[length(categ)], or other
-if(is.null(dot.categ)){
-dot.categ.class.order <- NULL # because not used anyway
-}
-# dot.categ.class.order either NULL if dot.categ is NULL (no legend displayed) or character string (potentially representing the diff classes of dot.categ)
-# end management of dot.color
-if(is.null(dot.color) & box.fill == FALSE & dot.alpha <= 0.025){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE FOLLOWING ARGUMENTS WERE SET AS:\ndot.color = NULL (NOT ALL DOTS BUT ONLY POTENTIAL OUTLIER DOTS DISPLAYED)\nbox.fill = FALSE (NO FILLING COLOR FOR BOTH BOXES AND POTENTIAL OUTLIER DOTS)\ndot.alpha = ", fun_round(dot.alpha, 4), "\n-> POTENTIAL OUTLIER DOTS MIGHT NOT BE VISIBLE BECAUSE ALMOST TRANSPARENT")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if(is.null(dot.color) & box.fill == FALSE & dot.border.size == 0){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE FOLLOWING ARGUMENTS WERE SET AS:\ndot.color = NULL (NOT ALL DOTS BUT ONLY POTENTIAL OUTLIER DOTS DISPLAYED)\nbox.fill = FALSE (NO FILLING COLOR FOR BOTH BOXES AND POTENTIAL OUTLIER DOTS)\ndot.border.size = 0 (NO BORDER FOR POTENTIAL OUTLIER DOTS)\n-> THESE SETTINGS ARE NOT ALLOWED BECAUSE THE POTENTIAL OUTLIER DOTS WILL NOT BE VISIBLE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# integer dot.border.color into gg_palette
-if( ! is.null(dot.border.color)){
-tempo <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
-if(tempo$problem == FALSE){ # convert integers into colors
-dot.border.color <- fun_gg_palette(max(dot.border.color, na.rm = TRUE))[dot.border.color]
-}
-}
-# end integer dot.border.color into gg_palette
-# na detection and removal (done now to be sure of the correct length of categ)
-column.check <- c("categ.color", if( ! is.null(dot.color)){"dot.color"}) #
-if(any(is.na(data1[, column.check]))){ # data1 used here instead of data1.ini in case of new NaN created by log conversion (neg values) # normally no NA with is.na
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS ", paste(column.check, collapse = " "), " OF data1 AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-for(i2 in 1:length(column.check)){
-if(any(is.na(data1[, column.check[i2]]))){ # normally no NA with is.na
-tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i2], " OF data1")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n", tempo.warn)))
-}
-}
-tempo <- unique(unlist(lapply(lapply(c(data1[column.check]), FUN = is.na), FUN = which)))
-removed.row.nb <- c(removed.row.nb, tempo)
-removed.rows <- rbind(removed.rows, data1[tempo, ], stringsAsFactors = FALSE) # here data1 used because categorical columns tested
-if(length(tempo) != 0){
-data1 <- data1[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
-data1.ini <- data1.ini[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
-for(i3 in 1:length(column.check)){
-if(any( ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]]), na.rm = TRUE)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i3], " OF data1, THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA/Inf REMOVAL (IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[, column.check[i3]])[ ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]])], collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-}
-for(i2 in 1:length(column.check)){
-if(column.check[i2] == "categ.color"){
-categ.color <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
-if(length(categ.color)== 1L & length(unlist(categ.class.order[length(categ)])) > 1){ # to deal with single color
-categ.color <- rep(categ.color, length(unlist(categ.class.order[length(categ)])))
-}
-data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(categ.color))
-}
-if(column.check[i2] == "dot.color"){
-dot.color <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
-if(length(dot.color)== 1L & length(dot.categ.class.order) > 1){ # to deal with single color. If dot.categ.class.order == NULL (which is systematically the case if dot.categ == NULL), no rep(dot.color, length(dot.categ.class.order)
-dot.color <- rep(dot.color, length(dot.categ.class.order))
-}
-data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(dot.color))
-}
-}
-}
-# end na detection and removal (done now to be sure of the correct length of categ)
-# From here, data1 and data.ini have no more NA or NaN
-# end other checkings
-# reserved word checking
-#already done above
-# end reserved word checking
-# end second round of checking and data preparation
-
-
-# package checking
-fun_pack(req.package = c(
-"ggplot2",
-"gridExtra",
-"lemon",
-"scales"
-), lib.path = lib.path)
-# end package checking
-
-
-
-
-
-# main code
-# y coordinates recovery (create ini.box.coord, dot.coord and modify data1)
-if(length(categ)== 1L){
-# width commputations
-box.width2 <- box.width
-box.space <- 0 # to inactivate the shrink that add space between grouped boxes, because no grouped boxes here
-# end width commputations
-# data1 check categ order for dots coordinates recovery
-data1 <- data.frame(data1, categ.check = data1[, categ[1]], stringsAsFactors = TRUE)
-data1$categ.check <- as.integer(data1$categ.check) # to check that data1[, categ[1]] and dot.coord$group are similar, during merging
-# end data1 check categ order for dots coordinates recovery
-# per box dots coordinates recovery
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[1]), stroke = dot.border.size, size = dot.size, alpha = dot.alpha, shape = 21))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(is.null(categ.color)){rep(NA, length(unique(data1[, categ[1]])))}else if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[1]])))}else{categ.color})) # categ.color used for dot colors because at that stage, we do not care about colors
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[1]), coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf})) # fill because this is what is used with geom_box # to easily have the equivalent of the grouped boxes
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[1]])))}else{categ.color}))
-# end per box dots coordinates recovery
-}else if(length(categ) == 2L){
-# width commputations
-box.width2 <- box.width / length(unique(data1[, categ[length(categ)]])) # real width of each box in x-axis unit, among the set of grouped box. Not relevant if no grouped boxes length(categ)== 1L
-# end width commputations
-# data1 check categ order for dots coordinates recovery
-tempo.factor <- paste0(data1[order(data1[, categ[2]], data1[, categ[1]]), categ[2]], "_", data1[order(data1[, categ[2]], data1[, categ[1]]), categ[1]])
-data1 <- data.frame(data1[order(data1[, categ[2]], data1[, categ[1]]), ], categ.check = factor(tempo.factor, levels = unique(tempo.factor)), stringsAsFactors = TRUE)
-data1$categ.check <- as.integer(data1$categ.check)
-# end data1 check categ order for dots coordinates recovery
-# per box dots coordinates recovery
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[2]), stroke = dot.border.size, size = dot.size, alpha = dot.alpha, shape = 21))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(is.null(categ.color)){rep(NA, length(unique(data1[, categ[2]])))}else if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[2]])))}else{categ.color})) # categ.color used for dot colors because at that stage, we do not care about colors
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[2]), coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf})) # fill because this is what is used with geom_box # to easily have the equivalent of the grouped boxes
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[2]])))}else{categ.color}))
-# end per box dots coordinates recovery
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if( ! is.null(stat.pos)){
-stat.just <- fun_gg_just(
-angle = stat.angle,
-pos = ifelse(
-vertical == TRUE,
-ifelse(stat.pos == "top", "bottom", "top"), # "bottom" because we want justification for text that are below the ref point which is the top of the graph. The opposite for "above"
-ifelse(stat.pos == "top", "left", "right") # "left" because we want justification for text that are on the left of the ref point which is the right border of the graph. The opposite for "above"
-),
-kind = "text"
-)
-}
-# has in fact no interest because ggplot2 does not create room for geom_text()
-tempo.data.max <- data1[which.max(data1[, y]), ]
-tempo.data.max <- data.frame(tempo.data.max, label = formatC(tempo.data.max[, y], digit = 2, drop0trailing = TRUE, format = "f"), stringsAsFactors = TRUE)
-# end has in fact no interest because ggplot2 does not create room for geom_text()
-tempo.graph.info.ini <- ggplot2::ggplot_build(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if( ! is.null(stat.pos)){' + ggplot2::geom_text(data = tempo.data.max, mapping = ggplot2::aes_string(x = 1, y = y, label = "label"), size = stat.size, color = "black", angle = stat.angle, hjust = stat.just$hjust, vjust = stat.just$vjust)'})))) # added here to have room for annotation
-dot.coord <- tempo.graph.info.ini$data[[1]]
-dot.coord$x <- as.numeric(dot.coord$x) # because weird class
-dot.coord$PANEL <- as.numeric(dot.coord$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
-tempo.mean <- aggregate(x = dot.coord$y, by = list(dot.coord$group, dot.coord$PANEL), FUN = mean, na.rm = TRUE)
-names(tempo.mean)[names(tempo.mean) == "x"] <- "MEAN"
-names(tempo.mean)[names(tempo.mean) == "Group.1"] <- "BOX"
-names(tempo.mean)[names(tempo.mean) == "Group.2"] <- "PANEL"
-dot.coord <- data.frame(
-dot.coord[order(dot.coord$group, dot.coord$y), ], # dot.coord$PANEL deals below
-y.check = as.double(data1[order(data1$categ.check, data1[, y]), y]),
-categ.check = data1[order(data1$categ.check, data1[, y]), "categ.check"],
-dot.color = if(is.null(dot.color)){NA}else{data1[order(data1$categ.check, data1[, y]), "dot.color"]},
-data1[order(data1$categ.check, data1[, y]), ][categ], # avoid the renaming below
-stringsAsFactors = TRUE
-) # y.check to be sure that the order is the same between the y of data1 and the y of dot.coord
-# names(dot.coord)[names(dot.coord) == "tempo.categ1"] <- categ[1]
-if( ! is.null(dot.categ)){
-dot.coord <- data.frame(dot.coord, data1[order(data1$categ.check, data1[, y]), ][dot.categ], stringsAsFactors = TRUE) # avoid the renaming
-}
-if( ! is.null(facet.categ)){
-dot.coord <- data.frame(dot.coord, data1[order(data1$categ.check, data1[, y]), ][facet.categ], stringsAsFactors = TRUE) # for facet panels
-tempo.test <- NULL
-for(i2 in 1:length(facet.categ)){
-tempo.test <- paste0(tempo.test, ".", formatC(as.numeric(dot.coord[, facet.categ[i2]]), width = nchar(max(as.numeric(dot.coord[, facet.categ[i2]]), na.rm = TRUE)), flag = "0")) # convert factor into numeric with leading zero for proper ranking # merge the formatC() to create a new factor. The convertion to integer should recreate the correct group number. Here as.numeric is used and not as.integer in case of numeric in facet.categ (because comes from add and not checked by fun_check, contrary to categ)
-}
-tempo.test <- as.integer(factor(tempo.test))
-if( ! identical(as.integer(dot.coord$PANEL), tempo.test)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nas.integer(dot.coord$PANEL) AND tempo.test MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(dot.tidy == TRUE){
-if( ! is.null(dot.categ)){
-dot.coord <- data.frame(dot.coord, tidy_group = data1[order(data1$categ.check, data1[, y]), ][, dot.categ], stringsAsFactors = TRUE) # avoid the renaming
-# tidy_group_coord is to be able to fuse table when creating the table for dot coordinates
-if(dot.categ %in% categ){
-dot.coord <- data.frame(dot.coord, tidy_group_coord = dot.coord$group, stringsAsFactors = TRUE)
-}else{
-dot.coord <- data.frame(dot.coord, tidy_group_coord = as.integer(factor(paste0(
-formatC(as.integer(dot.coord[, categ[1]]), width = nchar(max(as.integer(dot.coord[, categ[1]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
-".",
-if(length(categ) == 2L){formatC(as.integer(dot.coord[, categ[2]]), width = nchar(max(as.integer(dot.coord[, categ[2]]), na.rm = TRUE)), flag = "0")}, # convert factor into numeric with leading zero for proper ranking
-if(length(categ) == 2L){"."},
-formatC(as.integer(dot.coord[, dot.categ]), width = nchar(max(as.integer(dot.coord[, dot.categ]), na.rm = TRUE)), flag = "0") # convert factor into numeric with leading zero for proper ranking
-)), stringsAsFactors = TRUE) # merge the 2 or 3 formatC() to create a new factor. The convertion to integer should recreate the correct group number
-) # for tidy dot plots
-}
-}else{
-dot.coord <- data.frame(dot.coord, tidy_group = if(length(categ)== 1L){
-dot.coord[, categ]}else{as.integer(factor(paste0(
-formatC(as.integer(dot.coord[, categ[1]]), width = nchar(max(as.integer(dot.coord[, categ[1]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
-".",
-formatC(as.integer(dot.coord[, categ[2]]), width = nchar(max(as.integer(dot.coord[, categ[2]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
-)), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
-}) # for tidy dot plots
-# tidy_group_coord is to be able to fuse table when creating the table for dot coordinates
-dot.coord <- data.frame(dot.coord, tidy_group_coord = dot.coord$group, stringsAsFactors = TRUE)
-}
-}
-if( ! (identical(dot.coord$y, dot.coord$y.check) & identical(dot.coord$group, dot.coord$categ.check))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\n(dot.coord$y AND dot.coord$y.check) AS WELL AS (dot.coord$group AND dot.coord$categ.check) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-if( ! identical(tempo.mean[order(tempo.mean$BOX, tempo.mean$PANEL), ]$BOX, unique(dot.coord[order(dot.coord$group, dot.coord$PANEL), c("group", "PANEL")])$group)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\n(tempo.mean$BOX, tempo.mean$PANEL) AND (dot.coord$group, dot.coord$PANEL) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-tempo <- unique(dot.coord[order(dot.coord$group, dot.coord$PANEL), c(categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){if(dot.categ != ini.dot.categ){dot.categ}}, if( ! is.null(facet.categ)){facet.categ}), drop = FALSE])
-# names(tempo) <- paste0(names(tempo), ".mean")
-tempo.mean <- data.frame(tempo.mean[order(tempo.mean$BOX, tempo.mean$PANEL), ], tempo, stringsAsFactors = TRUE)
-}
-}
-# at that stage, categ color and dot color are correctly attributed in data1, box.coord and dot.coord
-# end y dot coordinates recovery (create ini.box.coord, dot.coord and modify data1)
-# ylim range
-if(is.null(y.lim)){
-y.lim <- tempo.graph.info.ini$layout$panel_params[[1]]$y.range # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-if(any(( ! is.finite(y.lim)) | is.na(y.lim)) | length(y.lim) != 2){ # kept but normally no more Inf in data1 # normally no NA with is.finite, etc.
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\ntempo.graph.info.ini$layout$panel_params[[1]]$y.range[1] CONTAINS NA OR Inf OR HAS LENGTH 1")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else if(y.log != "no"){
-y.lim <- get(y.log)(y.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-}
-if(y.log != "no"){
-# normally this control is not necessary anymore
-if(any( ! is.finite(y.lim))){ # normally no NA with is.finite
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-if(suppressWarnings(all(y.lim %in% c(Inf, -Inf)))){ # all() without na.rm -> ok because y.lim cannot be NA (tested above)
-# normally this control is not necessary anymore
-tempo.cat <- paste0("ERROR IN ", function.name, " y.lim CONTAINS Inf VALUES, MAYBE BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY OR BECAUSE OF LOG SCALE REQUIREMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-if(suppressWarnings(any(is.na(y.lim)))){ # normally no NA with is.na
-# normally this control is not necessary anymore
-tempo.cat <- paste0("ERROR IN ", function.name, " y.lim CONTAINS NA OR NaN VALUES, MAYBE BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY OR BECAUSE OF LOG SCALE REQUIREMENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-y.lim.order <- order(y.lim) # to deal with inverse axis
-y.lim <- sort(y.lim)
-y.lim[1] <- y.lim[1] - abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.bottom.extra.margin, y.top.extra.margin) # diff(y.lim.order) > 0 medians not inversed axis
-y.lim[2] <- y.lim[2] + abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.top.extra.margin, y.bottom.extra.margin) # diff(y.lim.order) > 0 medians not inversed axis
-if(y.include.zero == TRUE){ # no need to check y.log != "no" because done before
-y.lim <- range(c(y.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-}
-y.lim <- y.lim[y.lim.order]
-if(any(is.na(y.lim))){ # normally no NA with is.na
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 2")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end ylim range
-
-
-
-
-
-
-# drawing
-# constant part
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add is directly put here to deal with additional variable of data, like when using facet_grid. No problem if add is a theme, will be dealt below
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::xlab(if(is.null(x.lab)){categ[1]}else{x.lab}))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ylab(if(is.null(y.lab)){y}else{y.lab}))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
-# text angle management
-axis.just <- fun_gg_just(angle = x.angle, pos = ifelse(vertical == TRUE, "bottom", "left"), kind = "axis")
-# end text angle management
-add.check <- TRUE
-if( ! is.null(add)){ # if add is NULL, then = 0
-if(grepl(pattern = "ggplot2\\s*::\\s*theme", add) == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT\n-> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER\n-> article ARGUMENT WILL BE IGNORED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-add.check <- FALSE
-}
-}
-if(add.check == TRUE & article == TRUE){
-# WARNING: not possible to add theme()several times. NO message but the last one overwrites the others
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
-if(grid == TRUE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_line(size = 0.5),
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-axis.line.y.left = ggplot2::element_line(colour = "black"), # draw lines for the y axis
-axis.line.x.bottom = ggplot2::element_line(colour = "black"), # draw lines for the x axis
-panel.grid.major.x = if(vertical == TRUE){NULL}else{ggplot2::element_line(colour = "grey85", size = 0.75)},
-panel.grid.major.y = if(vertical == TRUE){ggplot2::element_line(colour = "grey85", size = 0.75)}else{NULL},
-panel.grid.minor.y = if(vertical == TRUE){ggplot2::element_line(colour = "grey90", size = 0.25)}else{NULL},
-axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
-axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)},
-strip.background = ggplot2::element_rect(fill = NA, colour = NA) # for facet background
-))
-}else{
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_line(size = 0.5),
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-axis.line.y.left = ggplot2::element_line(colour = "black"),
-axis.line.x.bottom = ggplot2::element_line(colour = "black"),
-axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
-axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)},
-strip.background = ggplot2::element_rect(fill = NA, colour = NA)
-))
-}
-}else if(add.check == TRUE & article == FALSE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_line(size = 0.5),
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-panel.background = ggplot2::element_rect(fill = "grey95"),
-axis.line.y.left = ggplot2::element_line(colour = "black"),
-axis.line.x.bottom = ggplot2::element_line(colour = "black"),
-panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.minor.x = ggplot2::element_blank(),
-panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
-strip.background = ggplot2::element_rect(fill = NA, colour = NA),
-axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
-axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}
-))
-}
-# Contrary to fun_gg_bar(), cannot plot the boxplot right now, because I need the dots plotted first
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, group = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), color = NA, width = box.width, fill = NA)) # this is to set the graph (i.e., a blanck boxplot to be able to use x coordinates to plot dots before boxes)
-# end constant part
-
-
-
-
-# graphic info recovery (including means)
-tempo.graph.info <- ggplot2::ggplot_build(eval(parse(text = paste0(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, notch = box.notch, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}) + ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color})')))) # will be recovered later again, when ylim will be considered
-tempo.yx.ratio <- (tempo.graph.info$layout$panel_params[[1]]$y.range[2] - tempo.graph.info$layout$panel_params[[1]]$y.range[1]) / (tempo.graph.info$layout$panel_params[[1]]$x.range[2] - tempo.graph.info$layout$panel_params[[1]]$x.range[1])
-box.coord <- tempo.graph.info$data[[2]] # to have the summary statistics of the plot. Contrary to ini.box.plot, now integrates ylim Here because can be required for stat.pos when just box are plotted
-box.coord$x <- as.numeric(box.coord$x) # because x is of special class that block comparison of values using identical
-box.coord$PANEL <- as.numeric(box.coord$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
-box.coord <- box.coord[order(box.coord$group, box.coord$PANEL), ]
-if( ! (identical(tempo.mean$BOX, box.coord$group) & identical(tempo.mean$PANEL, box.coord$PANEL))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nidentical(tempo.mean$BOX, box.coord$group) & identical(tempo.mean$PANEL, box.coord$PANEL) DO NOT HAVE THE SAME VALUE ORDER")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-# tempo <- c(categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){if(dot.categ != ini.dot.categ){dot.categ}}, if( ! is.null(facet.categ)){facet.categ})
-if(any(names(tempo.mean) %in% names(box.coord), na.rm = TRUE)){
-names(tempo.mean)[names(tempo.mean) %in% names(box.coord)] <- paste0(names(tempo.mean)[names(tempo.mean) %in% names(box.coord)], ".mean")
-}
-box.coord <- data.frame(box.coord, tempo.mean, stringsAsFactors = TRUE)
-}
-# end graphic info recovery (including means)
-
-
-
-# stat output (will also serve for boxplot and mean display)
-# x not added now (to do not have them in stat.nolog)
-stat <- data.frame(
-MIN = box.coord$ymin_final,
-QUART1 = box.coord$lower,
-MEDIAN = box.coord$middle,
-MEAN = box.coord$MEAN,
-QUART3 = box.coord$upper,
-MAX = box.coord$ymax_final,
-WHISK_INF = box.coord$ymin,
-BOX_INF = box.coord$lower,
-NOTCH_INF = box.coord$notchlower,
-NOTCH_SUP = box.coord$notchupper,
-BOX_SUP = box.coord$upper,
-WHISK_SUP = box.coord$ymax,
-OUTLIERS = box.coord["outliers"],
-tempo.mean[colnames(tempo.mean) != "MEAN"],
-COLOR = box.coord$fill,
-stringsAsFactors = TRUE
-) # box.coord["outliers"] written like this because it is a list. X coordinates not put now because several features to set
-names(stat)[names(stat) == "outliers"] <- "OUTLIERS"
-stat.nolog <- stat # stat.nolog ini will serve for outputs
-if(y.log != "no"){
-stat.nolog[c("MIN", "QUART1", "MEDIAN", "MEAN", "QUART3", "MAX", "WHISK_INF", "BOX_INF", "NOTCH_INF", "NOTCH_SUP", "BOX_SUP", "WHISK_SUP")] <- ifelse(y.log == "log2", 2, 10)^(stat.nolog[c("MIN", "QUART1", "MEDIAN", "MEAN", "QUART3", "MAX", "WHISK_INF", "BOX_INF", "NOTCH_INF", "NOTCH_SUP", "BOX_SUP", "WHISK_SUP")])
-stat.nolog$OUTLIERS <- lapply(stat.nolog$OUTLIERS, FUN = function(X){ifelse(y.log == "log2", 2, 10)^X})
-}
-# end stat output (will also serve for boxplot and mean display)
-
-
-
-
-
-
-# x coordinates management (for random plotting and for stat display)
-# width commputations
-width.ini <- c(box.coord$xmax - box.coord$xmin)[1] # all the box widths are equal here. Only the first one taken
-width.correct <- width.ini * box.space / 2
-if( ! (identical(stat$BOX, box.coord$group) & identical(stat$PANEL, box.coord$PANEL))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nidentical(stat$BOX, box.coord$group) & identical(stat$PANEL, box.coord$PANEL) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-stat <- data.frame(
-stat,
-X = box.coord$x,
-X_BOX_INF = box.coord$xmin + width.correct,
-X_BOX_SUP = box.coord$xmax - width.correct,
-X_NOTCH_INF = box.coord$x - (box.coord$x - (box.coord$xmin + width.correct)) / 2,
-X_NOTCH_SUP = box.coord$x + (box.coord$x - (box.coord$xmin + width.correct)) / 2,
-X_WHISK_INF = box.coord$x - (box.coord$x - (box.coord$xmin + width.correct)) * box.whisker.width,
-X_WHISK_SUP = box.coord$x + (box.coord$x - (box.coord$xmin + width.correct)) * box.whisker.width,
-# tempo.mean[colnames(tempo.mean) != "MEAN"], # already added above
-stringsAsFactors = TRUE
-)
-stat$COLOR <- factor(stat$COLOR, levels = unique(categ.color))
-if( ! all(stat$NOTCH_SUP < stat$BOX_SUP & stat$NOTCH_INF > stat$BOX_INF, na.rm = TRUE) & box.notch == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") SOME NOTCHES ARE BEYOND BOX HINGES. TRY ARGUMENT box.notch = FALSE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-dot.jitter <- c((box.coord$xmax - width.correct) - (box.coord$xmin + width.correct))[1] * dot.jitter # real dot.jitter. (box.coord$xmin + width.correct) - (box.coord$xmax - width.correct))[1] is the width of the box. Is equivalent to (box.coord$x - (box.coord$xmin + width.correct))[1] * 2
-# end width commputations
-if( ! is.null(dot.color)){
-# random dots
-if(dot.tidy == FALSE){
-dot.coord.rd1 <- merge(dot.coord, box.coord[c("fill", "PANEL", "group", "x")], by = c("PANEL", "group"), sort = FALSE) # rd for random. Send the coord of the boxes into the coord data.frame of the dots (in the column x.y). WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in box.coord. Thus, no need to consider fill column
-if(nrow(dot.coord.rd1) != nrow(dot.coord)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.rd1 DATA FRAME. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-sampled.dot.jitter <- if(nrow(dot.coord.rd1)== 1L){runif(n = nrow(dot.coord.rd1), min = - dot.jitter / 2, max = dot.jitter / 2)}else{sample(x = runif(n = nrow(dot.coord.rd1), min = - dot.jitter / 2, max = dot.jitter / 2), size = nrow(dot.coord.rd1), replace = FALSE)}
-dot.coord.rd2 <- data.frame(dot.coord.rd1, dot.x = dot.coord.rd1$x.y + sampled.dot.jitter, stringsAsFactors = TRUE) # set the dot.jitter thanks to runif and dot.jitter range. Then, send the coord of the boxes into the coord data.frame of the dots (in the column x.y)
-if(length(categ)== 1L){
-tempo.data1 <- unique(data.frame(data1[categ[1]], group = as.integer(data1[, categ[1]]), stringsAsFactors = TRUE)) # categ[1] is factor
-names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
-verif <- paste0(categ[1], ".check")
-}else if(length(categ) == 2L){
-tempo.data1 <- unique(
-data.frame(
-data1[c(categ[1], categ[2])],
-group = as.integer(factor(paste0(
-formatC(as.integer(data1[, categ[2]]), width = nchar(max(as.integer(data1[, categ[2]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
-".",
-formatC(as.integer(data1[, categ[1]]), width = nchar(max(as.integer(data1[, categ[1]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
-)), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
-)
-) # categ[2] first if categ[2] is used to make the categories in ggplot and categ[1] is used to make the x-axis
-names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
-names(tempo.data1)[names(tempo.data1) == categ[2]] <- paste0(categ[2], ".check")
-verif <- c(paste0(categ[1], ".check"), paste0(categ[2], ".check"))
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-dot.coord.rd3 <- merge(dot.coord.rd2, tempo.data1, by = intersect("group", "group"), sort = FALSE) # send the factors of data1 into coord. WARNING: I have replaced by = "group" by intersect("group", "group") because of an error due to wrong group group merging in dot.coord.rd3
-if(nrow(dot.coord.rd3) != nrow(dot.coord) | ( ! fun_comp_2d(dot.coord.rd3[categ], dot.coord.rd3[verif])$identical.content)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.rd3 DATA FRAME. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end random dots
-}
-# tidy dots
-# coordinates are recovered during plotting (see dot.coord.tidy1 below)
-# end tidy dots
-}
-# end x coordinates management (for random plotting and for stat display)
-
-
-
-
-
-# boxplot display before dot display if box.fill = TRUE
-coord.names <- NULL
-# creation of the data frame for (main box + legend) and data frame for means
-if(box.notch == FALSE){
-for(i3 in 1:length(categ)){
-if(i3== 1L){
-tempo.polygon <- data.frame(GROUPX = c(t(stat[, rep(categ[i3], 5)])), stringsAsFactors = TRUE)
-}else{
-tempo.polygon <- cbind(tempo.polygon, c(t(stat[, rep(categ[i3], 5)])), stringsAsFactors = TRUE)
-}
-}
-names(tempo.polygon) <- categ
-tempo.polygon <- data.frame(X = c(t(stat[, c("X_BOX_INF", "X_BOX_SUP", "X_BOX_SUP", "X_BOX_INF", "X_BOX_INF")])), Y = c(t(stat[, c("BOX_INF", "BOX_INF", "BOX_SUP", "BOX_SUP", "BOX_INF")])), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), BOX = as.character(c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX")]))), tempo.polygon, stringsAsFactors = TRUE)
-if( ! is.null(facet.categ)){
-for(i4 in 1:length(facet.categ)){
-tempo.polygon <- data.frame(tempo.polygon, c(t(stat[, c(facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4])])), stringsAsFactors = TRUE)
-names(tempo.polygon)[length(names(tempo.polygon))] <- facet.categ[i4]
-}
-}
-}else{
-for(i3 in 1:length(categ)){
-if(i3== 1L){
-tempo.polygon <- data.frame(GROUPX = c(t(stat[, rep(categ[i3], 11)])), stringsAsFactors = TRUE)
-}else{
-tempo.polygon <- cbind(tempo.polygon, c(t(stat[, rep(categ[i3], 11)])), stringsAsFactors = TRUE)
-}
-}
-names(tempo.polygon) <- categ
-tempo.polygon <- data.frame(X = c(t(stat[, c("X_BOX_INF", "X_BOX_SUP", "X_BOX_SUP", "X_NOTCH_SUP", "X_BOX_SUP", "X_BOX_SUP", "X_BOX_INF", "X_BOX_INF", "X_NOTCH_INF", "X_BOX_INF", "X_BOX_INF")])), Y = c(t(stat[, c("BOX_INF", "BOX_INF", "NOTCH_INF", "MEDIAN", "NOTCH_SUP", "BOX_SUP", "BOX_SUP", "NOTCH_SUP", "MEDIAN", "NOTCH_INF", "BOX_INF")])), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), BOX = as.character(c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX")]))), tempo.polygon, stringsAsFactors = TRUE)
-if( ! is.null(facet.categ)){
-for(i4 in 1:length(facet.categ)){
-tempo.polygon <- data.frame(tempo.polygon, c(t(stat[, c(facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4])])), stringsAsFactors = TRUE)
-names(tempo.polygon)[length(names(tempo.polygon))] <- facet.categ[i4]
-}
-}
-}
-tempo.polygon$COLOR <- factor(tempo.polygon$COLOR, levels = unique(categ.color))
-if( ! is.null(categ.class.order)){
-for(i3 in 1:length(categ)){
-tempo.polygon[, categ[i3]] <- factor(tempo.polygon[, categ[i3]], levels = categ.class.order[[i3]])
-}
-}
-# modified name of dot.categ column (e.g., "Categ1_DOT") must be included for boxplot using ridy dots
-if( ! is.null(dot.color) & ! is.null(dot.categ)){
-if(dot.categ != ini.dot.categ){
-tempo.polygon <- data.frame(tempo.polygon, GROUPX = tempo.polygon[, ini.dot.categ], stringsAsFactors = TRUE)
-names(tempo.polygon)[names(tempo.polygon) == "GROUPX"] <- dot.categ
-
-}
-}
-tempo.diamon.mean <- data.frame(X = c(t(stat[, c("X", "X_NOTCH_INF", "X", "X_NOTCH_SUP", "X")])), Y = c(t(cbind(stat["MEAN"] - (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stat["MEAN"], stat["MEAN"] + (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stat["MEAN"], stat["MEAN"] - (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stringsAsFactors = TRUE))), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), GROUP = c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX")])), stringsAsFactors = TRUE) # stringsAsFactors = TRUE for cbind() because stat["MEAN"] is a data frame. Otherwise, stringsAsFactors is not an argument for cbind() on vectors
-if( ! is.null(facet.categ)){
-for(i3 in 1:length(facet.categ)){
-tempo.diamon.mean <- data.frame(tempo.diamon.mean, c(t(stat[, c(facet.categ[i3], facet.categ[i3], facet.categ[i3], facet.categ[i3], facet.categ[i3])])), stringsAsFactors = TRUE)
-names(tempo.diamon.mean)[length(names(tempo.diamon.mean))] <- facet.categ[i3]
-}
-}
-tempo.diamon.mean$COLOR <- factor(tempo.diamon.mean$COLOR, levels = unique(categ.color))
-# end creation of the data frame for (main box + legend) and data frame for means
-if(box.fill == TRUE){
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[length(categ)], fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, size = box.line.size, notch = box.notch, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}, alpha = box.alpha, outlier.shape = if( ! is.null(dot.color)){NA}else{21}, outlier.color = if( ! is.null(dot.color)){NA}else{dot.border.color}, outlier.fill = if( ! is.null(dot.color)){NA}else{NULL}, outlier.size = if( ! is.null(dot.color)){NA}else{dot.size}, outlier.stroke = if( ! is.null(dot.color)){NA}else{dot.border.size}, outlier.alpha = if( ! is.null(dot.color)){NA}else{dot.alpha})) # the color, size, etc. of the outliers are dealt here. outlier.color = NA to do not plot outliers when dots are already plotted. Finally, boxplot redrawn (see below)
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_polygon(
-data = tempo.polygon,
-mapping = ggplot2::aes_string(x = "X", y = "Y", group = "BOX", fill = categ[length(categ)], color = categ[length(categ)]),
-size = box.line.size,
-alpha = box.alpha # works only for fill, not for color
-))
-coord.names <- c(coord.names, "main.box")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha)) #
-coord.names <- c(coord.names, "sup.whisker")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_INF, yend = WHISK_INF, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha)) #
-coord.names <- c(coord.names, "inf.whisker")
-if(box.whisker.width > 0){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha, lineend = "round")) #
-coord.names <- c(coord.names, "sup.whisker.edge")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_INF, yend = WHISK_INF, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha, lineend = "round")) #
-coord.names <- c(coord.names, "inf.whisker.edge")
-}
-if(box.mean == TRUE){
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = stat, mapping = ggplot2::aes_string(x = "X", y = "MEAN", group = categ[length(categ)]), shape = 23, stroke = box.line.size * 2, fill = stat$COLOR, size = box.mean.size, color = "black", alpha = box.alpha)) # group used in aesthetic to do not have it in the legend
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_polygon(
-data = tempo.diamon.mean,
-mapping = ggplot2::aes(x = X, y = Y, group = GROUP),
-fill = tempo.diamon.mean[, "COLOR"],
-color = hsv(0, 0, 0, alpha = box.alpha), # outline of the polygon in black but with alpha
-size = box.line.size,
-alpha = box.alpha
-))
-coord.names <- c(coord.names, "mean")
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = if(box.notch == FALSE){X_BOX_INF}else{X_NOTCH_INF}, xend = if(box.notch == FALSE){X_BOX_SUP}else{X_NOTCH_SUP}, y = MEDIAN, yend = MEDIAN, group = categ[length(categ)]), color = "black", size = box.line.size * 2, alpha = box.alpha)) #
-coord.names <- c(coord.names, "median")
-}
-# end boxplot display before dot display if box.fill = TRUE
-
-
-
-
-
-
-# dot display
-if( ! is.null(dot.color)){
-if(dot.tidy == FALSE){
-if(is.null(dot.categ)){
-if(dot.border.size == 0){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
-data = dot.coord.rd3,
-mapping = ggplot2::aes_string(x = "dot.x", y = "y", group = categ[length(categ)]),
-size = dot.size,
-shape = 19,
-color = dot.coord.rd3$dot.color,
-alpha = dot.alpha
-)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
-}else{
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
-data = dot.coord.rd3,
-mapping = ggplot2::aes_string(x = "dot.x", y = "y", group = categ[length(categ)]),
-shape = 21,
-stroke = dot.border.size,
-color = if(is.null(dot.border.color)){dot.coord.rd3$dot.color}else{rep(dot.border.color, nrow(dot.coord.rd3))},
-size = dot.size,
-fill = dot.coord.rd3$dot.color,
-alpha = dot.alpha
-)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
-}
-}else{
-if(dot.border.size == 0){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
-data = dot.coord.rd3,
-mapping = ggplot2::aes_string(x = "dot.x", y = "y", alpha = dot.categ),
-size = dot.size,
-shape = 19,
-color = dot.coord.rd3$dot.color
-)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
-}else{
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
-data = dot.coord.rd3,
-mapping = ggplot2::aes_string(x = "dot.x", y = "y", alpha = dot.categ),
-size = dot.size,
-shape = 21,
-stroke = dot.border.size,
-color = if(is.null(dot.border.color)){dot.coord.rd3$dot.color}else{rep(dot.border.color, nrow(dot.coord.rd3))},
-fill = dot.coord.rd3$dot.color
-)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = dot.legend.name, values = rep(dot.alpha, length(dot.categ.class.order)), guide = ggplot2::guide_legend(override.aes = list(fill = dot.color, color = if(is.null(dot.border.color)){dot.color}else{dot.border.color}, stroke = dot.border.size, alpha = dot.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-}
-coord.names <- c(coord.names, "dots")
-}else if(dot.tidy == TRUE){
-# here plot using group -> no scale
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_dotplot(
-data = dot.coord,
-mapping = ggplot2::aes_string(x = categ[1], y = "y", group = "group"), # not dot.categ here because the classes of dot.categ create new separations
-position = ggplot2::position_dodge(width = box.width),
-binpositions = "all",
-binaxis = "y",
-stackdir = "center",
-alpha = dot.alpha,
-fill = dot.coord$dot.color,
-stroke = dot.border.size,
-color = if(is.null(dot.border.color)){dot.coord$dot.color}else{rep(dot.border.color, nrow(dot.coord))},
-show.legend = FALSE, # WARNING: do not use show.legend = TRUE because it uses the arguments outside aes() as aesthetics (here color and fill). Thus I must find a way using ggplot2::scale_discrete_manual()
-binwidth = (y.lim[2] - y.lim[1]) / dot.tidy.bin.nb
-)) # geom_dotplot ggplot2 v3.3.0: I had to remove rev() in fill and color # very weird behavior of geom_dotplot ggplot2 v3.2.1, (1) because with aes group = (to avoid legend), the dot plotting is not good in term of coordinates, and (2) because data1 seems reorderer according to x = categ[1] before plotting. Thus, I have to use fill = dot.coord[rev(order(dot.coord[, categ[1]], decreasing = TRUE)), "dot.color"] to have the good corresponding colors # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
-coord.names <- c(coord.names, "dots")
-if( ! is.null(dot.categ)){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_dotplot(
-data = dot.coord,
-mapping = ggplot2::aes_string(x = categ[1], y = "y", alpha = dot.categ), # not dot.categ here because the classes of dot.categ create new separations
-position = ggplot2::position_dodge(width = box.width),
-binpositions = "all",
-binaxis = "y",
-stackdir = "center",
-fill = NA,
-stroke = NA,
-color = NA,
-# WARNING: do not use show.legend = TRUE because it uses the arguments outside aes() as aesthetics (here color and fill). Thus I must find a way using ggplot2::scale_discrete_manual()
-binwidth = (y.lim[2] - y.lim[1]) / dot.tidy.bin.nb
-)) # geom_dotplot ggplot2 v3.3.0: I had to remove rev() in fill and color # very weird behavior of geom_dotplot ggplot2 v3.2.1, (1) because with aes group = (to avoid legend), the dot plotting is not good in term of coordinates, and (2) because data1 seems reorderer according to x = categ[1] before plotting. Thus, I have to use fill = dot.coord[rev(order(dot.coord[, categ[1]], decreasing = TRUE)), "dot.color"] to have the good corresponding colors # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "linetype", name = dot.legend.name, values = rep(1, length(categ.color)))) # values = rep("black", length(categ.color)) are the values of color (which is the border color of dots), and this modify the border color on the plot. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-coord.names <- c(coord.names, "bad_remove")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = dot.legend.name, values = rep(dot.alpha, length(dot.categ.class.order)), labels = dot.categ.class.order, guide = ggplot2::guide_legend(title = if(ini.dot.categ == categ[length(categ)]){dot.categ}else{ini.dot.categ}, override.aes = list(fill = levels(dot.coord$dot.color), color = if(is.null(dot.border.color)){levels(dot.coord$dot.color)}else{dot.border.color}, stroke = dot.border.size, alpha = dot.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-}
-# coordinates of tidy dots
-tempo.coord <- ggplot2::ggplot_build(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "))))$data # to have the tidy dot coordinates
-if(length(which(sapply(X = tempo.coord, FUN = function(X){any(names(X) == "binwidth", na.rm = TRUE)}))) != 1){ # detect the compartment of tempo.coord which is the binned data frame
-# if(length(which(sapply(tempo.coord, FUN = nrow) == nrow(data1))) > if(is.null(dot.categ)){1}else{2}){ # this does not work if only one dot per class, thus replaced by above # if(is.null(dot.categ)){1}else{2} because 1 dotplot if dot.categ is NULL and 2 dotplots if not, with the second being a blank dotplot with wrong coordinates. Thus take the first in that situation
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nEITHER MORE THAN 1 OR NO COMPARTMENT HAVING A DATA FRAME WITH binwidth AS COLUMN NAME IN THE tempo.coord LIST (FOR TIDY DOT COORDINATES). CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-# dot.coord.tidy1 <- tempo.coord[[which(sapply(tempo.coord, FUN = nrow) == nrow(data1))[1]]] # this does not work if only one dot per class, thus replaced by above # the second being a blank dotplot with wrong coordinates. Thus take the first whatever situation
-dot.coord.tidy1 <- tempo.coord[[which(sapply(X = tempo.coord, FUN = function(X){any(names(X) == "binwidth", na.rm = TRUE)}))]] # detect the compartment of tempo.coord which is the binned data frame
-dot.coord.tidy1$x <- as.numeric(dot.coord.tidy1$x) # because weird class
-dot.coord.tidy1$PANEL <- as.numeric(dot.coord.tidy1$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
-}
-# tempo.box.coord <- merge(box.coord, unique(dot.coord[, c("PANEL", "group", categ)]), by = c("PANEL", "group"), sort = FALSE) # not required anymore because box.coord already contains categ do not add dot.categ and tidy_group_coord here because the coordinates are for stats. Add the categ in box.coord. WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in box.coord. Thus, no need to consider fill column
-# below inactivated because not true when dealing with dot.categ different from categ
-# if(nrow(tempo.box.coord) != nrow(box.coord)){
-# tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT tempo.box.coord DATA FRAME. CODE HAS TO BE MODIFIED")
-# stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-# }
-dot.coord.tidy2 <- merge(dot.coord.tidy1, box.coord[c("fill", "PANEL", "group", "x", categ)], by = c("PANEL", "group"), sort = FALSE) # send the coord of the boxes into the coord data.frame of the dots (in the column x.y).WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in tempo.box.coord. Thus, no need to consider fill colum # DANGER: from here the fill.y and x.y (from tempo.box.coord) are not good in dot.coord.tidy2. It is ok because Categ1 Categ2 from tempo.box.coord are ok with the group column from dot.coord.tidy1. This is due to the fact that dot.coord.tidy resulting from geom_dotplot does not make the same groups as the other functions
-if(nrow(dot.coord.tidy2) != nrow(dot.coord)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.tidy2 DATA FRAME. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# From here, check for dot.coord.tidy3 which wil be important for stat over the plot. WARNING: dot.categ has nothing to do here for stat coordinates. Thus, not in tempo.data1
-if(length(categ)== 1L){
-tempo.data1 <- unique(data.frame(data1[categ[1]], group = as.integer(data1[, categ[1]]), stringsAsFactors = TRUE)) # categ[1] is factor
-names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
-verif <- paste0(categ[1], ".check")
-}else if(length(categ) == 2L){
-tempo.data1 <- unique(
-data.frame(
-data1[c(categ[1], categ[2])],
-group = as.integer(factor(paste0(
-formatC(as.integer(data1[, categ[2]]), width = nchar(max(as.integer(data1[, categ[2]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
-".",
-formatC(as.integer(data1[, categ[1]]), width = nchar(max(as.integer(data1[, categ[1]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
-)), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
-)
-) # categ[2] first if categ[2] is used to make the categories in ggplot and categ[1] is used to make the x-axis
-names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
-names(tempo.data1)[names(tempo.data1) == categ[2]] <- paste0(categ[2], ".check")
-verif <- c(paste0(categ[1], ".check"), paste0(categ[2], ".check"))
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-dot.coord.tidy3 <- merge(dot.coord.tidy2, tempo.data1, by = intersect("group", "group"), sort = FALSE) # send the factors of data1 into coord. WARNING: I have tested intersect("group", "group") instead of by = "group". May be come back to by = "group" in case of error. But I did this because of an error in dot.coord.rd3 above
-if(nrow(dot.coord.tidy3) != nrow(dot.coord) | ( ! fun_comp_2d(dot.coord.tidy3[categ], dot.coord.tidy3[verif])$identical.content)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.tidy3 DATA FRAME. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end coordinates of tidy dots
-}
-}
-# end dot display
-
-
-
-# boxplot display (if box.fill = FALSE, otherwise, already plotted above)
-if(box.fill == TRUE){
-# overcome "work only for the filling of boxes, not for the frame. See https://github.com/tidyverse/ggplot2/issues/252"
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color}, guide = ggplot2::guide_legend(order = 1))) #, guide = ggplot2::guide_legend(override.aes = list(fill = levels(tempo.polygon$COLOR), color = "black")))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = rep(hsv(0, 0, 0, alpha = box.alpha), length(unique(data1[, categ[length(categ)]]))), guide = ggplot2::guide_legend(order = 1))) # , guide = ggplot2::guide_legend(override.aes = list(color = "black", alpha = box.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor # outline of the polygon in black but with alpha
-}else{
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[length(categ)], fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, size = box.line.size, notch = box.notch, alpha = box.alpha, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}, outlier.shape = if( ! is.null(dot.color)){NA}else{21}, outlier.color = if( ! is.null(dot.color)){NA}else{if(dot.border.size == 0){NA}else{dot.border.color}}, outlier.fill = if( ! is.null(dot.color)){NA}else{NULL}, outlier.size = if( ! is.null(dot.color)){NA}else{dot.size}, outlier.stroke = if( ! is.null(dot.color)){NA}else{dot.border.size}, outlier.alpha = if( ! is.null(dot.color)){NA}else{dot.alpha})) # the color, size, etc. of the outliers are dealt here. outlier.color = NA to do not plot outliers when dots are already plotted
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_path(
-data = tempo.polygon,
-mapping = ggplot2::aes_string(x = "X", y = "Y", group = "BOX", color = categ[length(categ)]),
-size = box.line.size,
-alpha = box.alpha,
-lineend = "round",
-linejoin = "round"
-))
-coord.names <- c(coord.names, "main.box")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = if(box.notch == FALSE){X_BOX_INF}else{X_NOTCH_INF}, xend = if(box.notch == FALSE){X_BOX_SUP}else{X_NOTCH_SUP}, y = MEDIAN, yend = MEDIAN, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size * 2, alpha = box.alpha)) #
-coord.names <- c(coord.names, "median")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha)) #
-coord.names <- c(coord.names, "sup.whisker")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_INF, yend = WHISK_INF, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha)) #
-coord.names <- c(coord.names, "inf.whisker")
-if(box.whisker.width > 0){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha, lineend = "round")) #
-coord.names <- c(coord.names, "sup.whisker.edge")
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_INF, yend = WHISK_INF, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha, lineend = "round")) #
-coord.names <- c(coord.names, "inf.whisker.edge")
-}
-if(box.mean == TRUE){
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = stat, mapping = ggplot2::aes_string(x = "X", y = "MEAN", group = categ[length(categ)]), shape = 23, stroke = box.line.size * 2, color = stat$COLOR, size = box.mean.size, fill = NA, alpha = box.alpha)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_path(
-data = tempo.diamon.mean,
-mapping = ggplot2::aes(x = X, y = Y, group = GROUP),
-color = tempo.diamon.mean[, "COLOR"],
-size = box.line.size,
-alpha = box.alpha,
-lineend = "round",
-linejoin = "round"
-))
-coord.names <- c(coord.names, "mean")
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = rep(NA, length(unique(data1[, categ[length(categ)]]))))) #, guide = ggplot2::guide_legend(override.aes = list(color = categ.color)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color}, guide = ggplot2::guide_legend(override.aes = list(alpha = if(plot == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list()) == 0L & Sys.info()["sysname"] == "Windows"))){1}else{box.alpha})))) # , guide = ggplot2::guide_legend(override.aes = list(color = as.character(categ.color))))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
-if(plot == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list()) == 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-# to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(box.alpha == 0){ # remove box legend because no boxes drawn
-# add this after the scale_xxx_manual() for boxplots
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none")) # inactivate the legend
-}
-# end boxplot display (if box.fill = FALSE, otherwise, already plotted above)
-
-
-
-
-# stat display
-# layer after dots but ok, behind dots on the plot
-if( ! is.null(stat.pos)){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NUMBERS DISPLAYED ARE ", ifelse(stat.mean == FALSE, "MEDIANS", "MEANS"))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-if(stat.pos == "top"){
-tempo.stat <- data.frame(stat, Y = y.lim[2], stringsAsFactors = TRUE) # I had to create a data frame for geom_tex() so that facet is taken into account, (ggplot2::annotate() does not deal with facet because no data and mapping arguments). Of note, facet.categ is in tempo.stat, via tempo.mean, via dot.coord
-if(stat.mean == FALSE){tempo.stat$MEDIAN <- formatC(stat.nolog$MEDIAN, digit = 2, drop0trailing = TRUE, format = "f")}else{tempo.stat$MEAN <- formatC(stat.nolog$MEAN, digit = 2, drop0trailing = TRUE, format = "f")}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
-data = tempo.stat,
-mapping = ggplot2::aes_string(x = "X", y = "Y", label = ifelse(stat.mean == FALSE, "MEDIAN", "MEAN")),
-size = stat.size,
-color = "black",
-angle = stat.angle,
-hjust = stat.just$hjust,
-vjust = stat.just$vjust
-)) # stat$X used here because identical to stat.nolog but has the X. WARNING: no need of order() for labels because box.coord$x set the order. For justification, see https://stackoverflow.com/questions/7263849/what-do-hjust-and-vjust-do-when-making-a-plot-using-ggplot
-coord.names <- c(coord.names, "stat.pos")
-}else if(stat.pos == "above"){
-# stat coordinates
-if( ! is.null(dot.color)){ # for text just above max dot
-if(dot.tidy == FALSE){
-tempo.stat.ini <- dot.coord.rd3
-}else if(dot.tidy == TRUE){
-tempo.stat.ini <- dot.coord.tidy3
-tempo.stat.ini$x.y <- tempo.stat.ini$x.x # this is just to be able to use tempo.stat.ini$x.y for untidy or tidy dots (remember that dot.coord.tidy3$x.y is not good, see above)
-}
-stat.coord1 <- aggregate(x = tempo.stat.ini["y"], by = {x.env <- if(length(categ)== 1L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]])}else if(length(categ) == 2L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]], tempo.stat.ini[, categ[2]])} ; names(x.env) <- if(length(categ)== 1L){c("group", "PANEL", "x.y", categ[1])}else if(length(categ) == 2L){c("group", "PANEL", "x.y", categ[1], categ[2])} ; x.env}, FUN = min, na.rm = TRUE)
-names(stat.coord1)[names(stat.coord1) == "y"] <- "dot.min"
-stat.coord2 <- aggregate(x = tempo.stat.ini["y"], by = {x.env <- if(length(categ)== 1L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]])}else if(length(categ) == 2L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]], tempo.stat.ini[, categ[2]])} ; names(x.env) <- if(length(categ)== 1L){c("group", "PANEL", "x.y", categ[1])}else if(length(categ) == 2L){c("group", "PANEL", "x.y", categ[1], categ[2])} ; x.env}, FUN = max, na.rm = TRUE)
-names(stat.coord2) <- paste0(names(stat.coord2), "_from.dot.max")
-names(stat.coord2)[names(stat.coord2) == "y_from.dot.max"] <- "dot.max"
-stat.coord3 <- cbind(box.coord[order(box.coord$group, box.coord$PANEL), ], stat.coord1[order(stat.coord1$group, stat.coord1$x.y), ], stat.coord2[order(stat.coord2$group, stat.coord2$x.y), ], stringsAsFactors = TRUE) #
-if( ! all(identical(round(stat.coord3$x, 9), round(as.numeric(stat.coord3$x.y), 9)), na.rm = TRUE)){ # as.numeric() because stat.coord3$x is class "mapped_discrete" "numeric"
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nFUSION OF box.coord, stat.coord1 AND stat.coord2 ACCORDING TO box.coord$x, stat.coord1$x.y AND stat.coord2$x.y IS NOT CORRECT. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# text.coord <- stat.coord3[, c("x", "group", "dot.min", "dot.max")]
-# names(text.coord)[names(text.coord) == "dot.min"] <- "text.min.pos"
-#names(text.coord)[names(text.coord) == "dot.max"] <- "text.max.pos"
-box.coord <- box.coord[order(box.coord$x, box.coord$group, box.coord$PANEL), ]
-# text.coord <- text.coord[order(text.coord$x), ] # to be sure to have the two objects in the same order for x. WARNING: cannot add identical(as.integer(text.coord$group), as.integer(box.coord$group)) because with error, the correspondence between x and group is not the same
-stat.coord3 <- stat.coord3[order(stat.coord3$x, stat.coord3$group, stat.coord3$PANEL), ] # to be sure to have the two objects in the same order for x. WARNING: cannot add identical(as.integer(text.coord$group), as.integer(box.coord$group)) because with error, the correspondence between x and group is not the same
-if( ! (identical(box.coord$x, stat.coord3$x) & identical(box.coord$group, stat.coord3$group) & identical(box.coord$PANEL, stat.coord3$PANEL))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\ntext.coord AND box.coord DO NOT HAVE THE SAME x, group AND PANEL COLUMN CONTENT")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}else{
-stat.coord3 <- box.coord
-}
-stat.coord3 <- data.frame(
-stat.coord3,
-Y = stat.coord3[, ifelse(
-is.null(dot.color),
-ifelse(diff(y.lim) > 0, "ymax", "ymin"),
-ifelse(diff(y.lim) > 0, "ymax_final", "ymin_final")
-)],
-stringsAsFactors = TRUE
-) # ymax is top whisker, ymax_final is top dot
-# stat.coord3 <- data.frame(stat.coord3, Y = vector("numeric", length = nrow(stat.coord3)), stringsAsFactors = TRUE)
-# check.Y <- as.logical(stat.coord3$Y) # convert everything in Y into FALSE (because Y is full of zero)
-# end stat coordinates
-# stat display
-# performed twice: first for y values >=0, then y values < 0, because only a single value allowed for hjust anf vjust
-if(stat.mean == FALSE){
-tempo.center.ref <- "middle"
-}else{
-tempo.center.ref <- "MEAN"
-}
-# if(is.null(dot.color)){
-# tempo.low.ref <- "ymin"
-# tempo.high.ref <- "ymax"
-# }else{
-# tempo.low.ref <- "ymin_final"
-# tempo.high.ref <- "ymax_final"
-# }
-# tempo.log.high <- if(diff(y.lim) > 0){stat.coord3[, tempo.center.ref] >= 0}else{stat.coord3[, tempo.center.ref] < 0}
-# tempo.log.low <- if(diff(y.lim) > 0){stat.coord3[, tempo.center.ref] < 0}else{stat.coord3[, tempo.center.ref] >= 0}
-# stat.coord3$Y[tempo.log.high] <- stat.coord3[tempo.log.high, tempo.high.ref]
-# stat.coord3$Y[tempo.log.low] <- stat.coord3[tempo.log.low, tempo.low.ref]
-# add distance
-stat.coord3$Y <- stat.coord3$Y + diff(y.lim) * stat.dist / 100
-# end add distance
-# correct median or mean text format
-if(y.log != "no"){
-stat.coord3[, tempo.center.ref] <- ifelse(y.log == "log2", 2, 10)^(stat.coord3[, tempo.center.ref])
-}
-stat.coord3[, tempo.center.ref] <- formatC(stat.coord3[, tempo.center.ref], digit = 2, drop0trailing = TRUE, format = "f")
-# end correct median or mean text format
-# if(any(tempo.log.high) == TRUE){
-# tempo.stat <- stat.coord3[tempo.log.high,]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
-data = stat.coord3,
-mapping = ggplot2::aes_string(x = "x", y = "Y", label = tempo.center.ref),
-size = stat.size,
-color = "black",
-angle = stat.angle,
-hjust = stat.just$hjust,
-vjust = stat.just$vjust
-)) # WARNING: no need of order() for labels because box.coord$x set the order
-coord.names <- c(coord.names, "stat.pos")
-# }
-# if(any(tempo.log.low) == TRUE){
-# tempo.stat <- stat.coord3[tempo.log.low,]
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
-# data = tempo.stat,
-# mapping = ggplot2::aes_string(x = "x", y = "Y", label = tempo.center.ref),
-# size = stat.size,
-# color = "black",
-# hjust = ifelse(vertical == TRUE, 0.5, 0.5 + stat.dist),
-# vjust = ifelse(vertical == TRUE, 0.5 + stat.dist, 0.5)
-# )) # WARNING: no need of order() for labels because box.coord$x set the order
-# coord.names <- c(coord.names, "stat.pos.negative")
-# }
-# end stat display
-}else{
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 5")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-}
-# end stat display
-# legend management
-if(legend.show == FALSE){ # must be here because must be before bef.final.plot <-
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none", alpha = "none")) # inactivate the initial legend
-}
-# end legend management
-
-
-
-# y scale management (cannot be before dot plot management)
-# the rescaling aspect is complicated and not intuitive. See:
-# explaination: https://github.com/tidyverse/ggplot2/issues/3948
-# the oob argument of scale_y_continuous() https://ggplot2.tidyverse.org/reference/scale_continuous.html
-# see also https://github.com/rstudio/cheatsheets/blob/master/data-visualization-2.1.pdf
-# secondary ticks
-bef.final.plot <- ggplot2::ggplot_build(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + if(vertical == TRUE){ggplot2::scale_y_continuous(expand = c(0, 0), limits = sort(y.lim), oob = scales::rescale_none)}else{ggplot2::coord_flip(ylim = y.lim)}')))) # here I do not need the x-axis and y-axis orientation, I just need the number of main ticks and the legend. I DI NOT UNDERSTAND THE COMMENT HERE BECAUSE WE NEED COORD_FLiP
-tempo.coord <- bef.final.plot$layout$panel_params[[1]]
-# y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non NULL or if y.log argument is different from "no")
-if(y.log != "no"){ # integer main ticks for log2 and log10
-tempo.scale <- (as.integer(min(y.lim, na.rm = TRUE)) - 1):(as.integer(max(y.lim, na.rm = TRUE)) + 1)
-}else{
-tempo <- if(is.null(attributes(tempo.coord$y$breaks))){tempo.coord$y$breaks}else{unlist(attributes(tempo.coord$y$breaks))}
-if(all(is.na(tempo))){# all() without na.rm -> ok because is.na() cannot be NA
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$y$breaks")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-tempo.scale <- fun_scale(lim = y.lim, n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks. If fact: n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) replaced by n = ifelse(is.null(y.tick.nb), 4, y.tick.nb))
-}
-y.second.tick.values <- NULL
-y.second.tick.pos <- NULL
-if(y.log != "no"){
-tempo <- fun_inter_ticks(lim = y.lim, log = y.log)
-y.second.tick.values <- tempo$values
-y.second.tick.pos <- tempo$coordinates
-# if(vertical == TRUE){ # do not remove in case the bug is fixed
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", y = y.second.tick.pos, yend = y.second.tick.pos, x = tempo.coord$x.range[1], xend = tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80))
-# }else{ # not working because of the ggplot2 bug
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", x = y.second.tick.pos, xend = y.second.tick.pos, y = tempo.coord$y.range[1], yend = tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80))
-# }
-coord.names <- c(coord.names, "y.second.tick.positions")
-}else if(( ! is.null(y.second.tick.nb)) & y.log == "no"){
-# if(y.second.tick.nb > 0){ #inactivated because already checked before
-if(length(tempo.scale) < 2){
-tempo.cat1 <- c("y.tick.nb", "y.second.tick.nb")
-tempo.cat2 <- sapply(list(y.tick.nb, y.second.tick.nb), FUN = paste0, collapse = " ")
-tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE Y-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-tempo <- fun_inter_ticks(lim = y.lim, log = y.log, breaks = tempo.scale, n = y.second.tick.nb)
-}
-y.second.tick.values <- tempo$values
-y.second.tick.pos <- tempo$coordinates
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
-geom = "segment",
-y = y.second.tick.pos,
-yend = y.second.tick.pos,
-x = if(vertical == TRUE){tempo.coord$x.range[1]}else{tempo.coord$y.range[1]},
-xend = if(vertical == TRUE){tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80}else{tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80}
-))
-coord.names <- c(coord.names, "y.second.tick.positions")
-}
-# end y.second.tick.positions
-# for the ggplot2 bug with y.log, this does not work: eval(parse(text = ifelse(vertical == FALSE & y.log == "log10", "ggplot2::scale_x_continuous", "ggplot2::scale_y_continuous")))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_y_continuous(
-breaks = tempo.scale,
-minor_breaks = y.second.tick.pos,
-labels = if(y.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(y.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(y.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 6") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)}, # == in stop() to be able to add several messages between ==
-expand = c(0, 0), # remove space after after axis limits
-limits = sort(y.lim), # NA indicate that limits must correspond to data limits but ylim() already used
-oob = scales::rescale_none,
-trans = ifelse(diff(y.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_y_reverse() but create the problem of y-axis label disappearance with y.lim decreasing. Thus, do not use. Use ylim() below and after this
-))
-if(vertical == TRUE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_cartesian(ylim = y.lim)) # problem of ggplot2::ylim() is that it redraws new breaks # coord_cartesian(ylim = y.lim)) not used because bug -> y-axis label disappearance with y.lim decreasing I DO NOT UNDERSTAND THIS MESSAGE WHILE I USE COORD_CARTESIAN # clip = "off" to have secondary ticks outside plot region does not work
-}else{
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_flip(ylim = y.lim)) # clip = "off" to have secondary ticks outside plot region does not work # create the problem of y-axis label disappearance with y.lim decreasing. IDEM ABOVE
-
-}
-# end y scale management (cannot be before dot plot management)
-
-
-# legend management
-if( ! is.null(legend.width)){
-legend.final <- fun_gg_get_legend(ggplot_built = bef.final.plot, fun.name = function.name, lib.path = lib.path) # get legend
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none", alpha = "none")) # inactivate the initial legend
-if(is.null(legend.final) & plot == TRUE){ # even if any(unlist(legend.disp)) is TRUE
-legend.final <- ggplot2::ggplot()+ggplot2::theme_void() # empty graph instead of legend
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON NULL legend.width ARGUMENT\n")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end legend management
-
-
-# drawing
-fin.plot <- suppressMessages(suppressWarnings(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ")))))
-grob.save <- NULL
-if(plot == TRUE){
-# following lines inactivated because of problem in warn.recov and message.recov
-# assign("env_fun_get_message", new.env())
-# assign("tempo.gg.name", tempo.gg.name, envir = env_fun_get_message)
-# assign("tempo.gg.count", tempo.gg.count, envir = env_fun_get_message)
-# assign("add", add, envir = env_fun_get_message)
-# two next line: for the moment, I cannot prevent the warning printing
-# warn.recov <- fun_get_message(paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}), kind = "warning", header = FALSE, print.no = FALSE, env = env_fun_get_message) # for recovering warnings printed by ggplot() functions
-# message.recov <- fun_get_message('print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))', kind = "message", header = FALSE, print.no = FALSE, env = env_fun_get_message) # for recovering messages printed by ggplot() functions
-# if( ! (return == TRUE & return.ggplot == TRUE)){ # because return() plots when return.ggplot is TRUE # finally not used -> see return.ggplot description
-if(is.null(legend.width)){
-grob.save <- suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot)))
-}else{
-grob.save <-suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot, legend.final, ncol=2, widths=c(1, legend.width))))
-}
-# }
-# suppressMessages(suppressWarnings(print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))))
-}else{
-# following lines inactivated because of problem in warn.recov and message.recov
-# message.recov <- NULL
-# warn.recov <- NULL
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end drawing
-
-
-
-# output
-# following lines inactivated because of problem in warn.recov and message.recov
-# if( ! (is.null(warn) & is.null(warn.recov) & is.null(message.recov))){
-# warn <- paste0(warn, "\n\n", if(length(warn.recov) > 0 | length(message.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", ifelse( ! is.null(warn.recov), unique(message.recov), ""), ifelse( ! is.null(message.recov), unique(message.recov), ""), collapse = "\n\n"), "\n\n")})
-# }else if( ! (is.null(warn) & is.null(warn.recov)) & is.null(message.recov)){
-# warn <- paste0(warn, "\n\n", if(length(warn.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", unique(warn.recov), collapse = "\n\n"), "\n\n")})
-# }else if( ! (is.null(warn) & is.null(message.recov)) & is.null(warn.recov)){
-# warn <- paste0(warn, "\n\n", if(length(message.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", unique(message.recov), collapse = "\n\n"), "\n\n")})
-# }
-if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
-}
-on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
-if(return == TRUE){
-tempo.output <- ggplot2::ggplot_build(fin.plot)
-tempo.output$data <- tempo.output$data[-1] # remove the first data because corresponds to the initial empty boxplot
-if(length(tempo.output$data) != length(coord.names)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nlength(tempo.output$data) AND length(coord.names) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-names(tempo.output$data) <- coord.names
-tempo.output$data <- tempo.output$data[coord.names != "bad_remove"]
-}
-tempo <- tempo.output$layout$panel_params[[1]]
-output <- list(
-data = data1.ini,
-stat = stat.nolog,
-removed.row.nb = removed.row.nb,
-removed.rows = removed.rows,
-plot = c(tempo.output$data, y.second.tick.values = list(y.second.tick.values)),
-panel = facet.categ,
-axes = list(
-x.range = tempo$x.range,
-x.labels = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{tempo$x$scale$get_labels()}, # is.null(attributes(tempo$x$breaks)) test if it is number (TRUE) or character (FALSE)
-x.positions = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{unlist(attributes(tempo$x$breaks))},
-y.range = tempo$y.range,
-y.labels = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{tempo$y$scale$get_labels()},
-y.positions = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{unlist(attributes(tempo$y$breaks))}
-),
-warn = paste0("\n", warn, "\n\n"),
-ggplot = if(return.ggplot == TRUE){fin.plot}else{NULL}, # fin.plot plots the graph if return == TRUE
-gtable = if(return.gtable == TRUE){grob.save}else{NULL}
-)
-return(output) # this plots the graph if return.ggplot is TRUE and if no assignment
-}
-# end output
-# end main code
-}
-
-
-
-
-
-
-# add density
-# rasterise all kind: https://cran.r-project.org/web/packages/ggrastr/vignettes/Raster_geoms.html
-
-
-fun_gg_scatter <- function(
-data1,
-x,
-y,
-categ = NULL,
-categ.class.order = NULL,
-color = NULL,
-geom = "geom_point",
-geom.step.dir = "hv",
-geom.stick.base = NULL,
-alpha = 0.5,
-dot.size = 2,
-dot.shape = 21,
-dot.border.size = 0.5,
-dot.border.color = NULL,
-line.size = 0.5,
-line.type = "solid",
-x.lim = NULL,
-x.lab = NULL,
-x.log = "no",
-x.tick.nb = NULL,
-x.second.tick.nb = NULL,
-x.include.zero = FALSE,
-x.left.extra.margin = 0.05,
-x.right.extra.margin = 0.05,
-x.text.angle = 0,
-y.lim = NULL,
-y.lab = NULL,
-y.log = "no",
-y.tick.nb = NULL,
-y.second.tick.nb = NULL,
-y.include.zero = FALSE,
-y.top.extra.margin = 0.05,
-y.bottom.extra.margin = 0.05,
-y.text.angle = 0,
-raster = FALSE,
-raster.ratio = 1,
-raster.threshold = NULL,
-text.size = 12,
-title = "",
-title.text.size = 12,
-legend.show = TRUE,
-legend.width = 0.5,
-legend.name = NULL,
-article = TRUE,
-grid = FALSE,
-add = NULL,
-return = FALSE,
-return.ggplot = FALSE,
-return.gtable = TRUE,
-plot = TRUE,
-warn.print = FALSE,
-lib.path = NULL
-){
-# AIM
-# Plot ggplot2 scatterplot with the possibility to overlay dots from up to 3 different data frames (-> three different legends) and lines from up to 3 different data frames (-> three different legends) -> up to 6 overlays totally
-# For ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
-# WARNINGS
-# Rows containing NA in data1[, c(x, y, categ)] will be removed before processing, with a warning (see below)
-# Size arguments (dot.size, dot.border.size, line.size, text.size and title.text.size) are in mm. See Hadley comment in https://stackoverflow.com/questions/17311917/ggplot2-the-unit-of-size. See also http://sape.inf.usi.ch/quick-reference/ggplot2/size). Unit object are not accepted, but conversion can be used (e.g., grid::convertUnit(grid::unit(0.2, "inches"), "mm", valueOnly = TRUE))
-# ARGUMENTS
-# data1: a dataframe compatible with ggplot2, or a list of data frames. Order matters for the order of the legend and for the layer staking (starting from below to top)
-# x: single character string of the data1 column name for x-axis coordinates. If data1 is a list, then x must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Write NULL for each "geom_hline" in geom argument
-# y: single character string of the data1 column name for y-axis coordinates. If data1 is a list, then y must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Write NULL for each "geom_vline" in geom argument
-# categ: either NULL or a single character string or a list of single character strings, indicating the data1 column names to use for categories which creates legend display
-# If categ == NULL, no categories -> no legend displayed
-# If data1 is a data frame, categ must be a single character string of the data1 column name for categories
-# If data1 is a list, then categ must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL (no legend display for these compartments), and other not
-# categ.class.order: either (1) NULL or (2) a vector of character strings or (3) a list of these vectors, setting the order of the classes of categ in the legend display
-# If categ.class.order is NULL, classes are represented according to the alphabetical order
-# If data1 is a data frame, categ.class.order must be a vector of character strings specifying the different classes in the categ column name of data1
-# If data1 is a list, then categ.class.order must be a list of vector of character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL (alphabetical order for these compartments), and other not
-# color: either (1) NULL, or (2) a vector of character strings or integers, or (3) a list of vectors of character strings or integers
-# If color is NULL, default colors of ggplot2
-# If data1 is a data frame, color argument can be either:
-# (1) a single color string. All the dots of the corresponding data1 will have this color, whatever the categ value (NULL or not)
-# (2) if categ is non-null, a vector of string colors, one for each class of categ. Each color will be associated according to the categ.class.order argument if specified, or to the alphabetical order of categ classes otherwise
-# (3) if categ is non-null, a vector or factor of string colors, like if it was one of the column of data1 data frame. WARNING: a single color per class of categ and a single class of categ per color must be respected
-# Positive integers are also accepted instead of character strings, as long as above rules about length are respected. Integers will be processed by fun_gg_palette() using the max integer value among all the integers in color (see fun_gg_palette())
-# If data1 is a list, then color argument must be either:
-# (1) a list of character strings or integers, of same size as data1, with compartment 1 related to compartment 1 of data1, etc.
-# (2) a single character string or a single integer
-# With a list (first possibility), the rules described for when data1 is a data frame apply to each compartment of the list. Some of the compartments can be NULL. In that case, a different grey color will be used for each NULL compartment. With a single value (second possibility), the same color will be used for all the dots and lines, whatever the data1 list
-# geom: single character string of the kind of plot, or a list of single character strings
-# Either:
-# "geom_point" (scatterplot)
-# "geom_line" (coordinates plotted then line connection, from the lowest to highest x coordinates first and from the lowest to highest y coordinates thenafter)
-# "geom_path" (coordinates plotted then line connection respecting the row order in data1)
-# "geom_step" coordinates plotted then line connection respecting the row order in data1 but drawn in steps). See the geom.step.dir argument
-# "geom_hline" (horizontal line, no x value provided)
-# "geom_vline" (vertical line, no y value provided)
-# "geom_stick" (dots as vertical bars)
-# If data1 is a list, then geom must be either:
-# (1) a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc.
-# (2) a single character string. In that case the same kind of plot will apply for the different compartments of the data1 list
-# WARNING concerning "geom_hline" or "geom_vline":
-# (1) x or y argument must be NULL, respectively
-# (2) x.lim or y.lim argument must NOT be NULL, respectively, if only these kind of lines are drawn (if other geom present, then x.lim = NULL and y.lim = NULL will generate x.lim and y.lim defined by these other geom, which is not possible with "geom_hline" or "geom_vline" alone)
-# (3) the function will draw n lines for n values in the x argument column name of the data1 data frame. If several colors required, the categ argument must be specified and the corresponding categ column name must exist in the data1 data frame with a different class name for each row
-# geom.step.dir: single character string indicating the direction when using "geom_step" of the geom argument, or a list of single character strings
-# Either:
-# "vh" (vertical then horizontal)
-# "hv" (horizontal then vertical)
-# "mid" (step half-way between adjacent x-values)
-# See https://ggplot2.tidyverse.org/reference/geom_path.html
-# If data1 is a list, then geom.step.dir must be either:
-# (1) a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. The value in compartments related to other geom values than "geom_step" will be ignored
-# (2) a single character string, which will be used for all the "geom_step" values of the geom argument, whatever the data1 list
-# geom.stick.base: either (1) NULL or (2) a single numeric value or (3) a list of single numeric values, setting the base of the sticks when using "geom_stick" of the geom argument
-# If geom.stick.base is NULL, the bottom limit of the y-axis is taken as the base
-# If data1 is a list, then geom.stick.base must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the values in compartments related to other geom values than "geom_stick" will be ignored. With a single value (latter possibility), the same base will be used for all the sticks, whatever the data1 list
-# Warning: the y-axis limits are not modified by the value of geom.stick.base, meaning that this value can be outside of the range of y.lim. Add the value of geom.stick.base also in the y.lim argument if required
-# Warning: if geom.stick.base is NULL, the bottom limit of the y-axis is taken as the base. Thus, be careful with inverted y-axis
-# alpha: single numeric value (from 0 to 1) of transparency. If data1 is a list, then alpha must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. In that case the same transparency will apply for the different compartments of the data1 list
-# dot.size: single numeric value of dot shape radius? in mm. If data1 is a list, then dot.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.size will be used for all the dots, whatever the data1 list
-# dot.shape: value indicating the shape of the dots (see https://ggplot2.tidyverse.org/articles/ggplot2-specs.html) If data1 is a list, then dot.shape must be either (1) a list of single shape values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single shape value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.shape will be used for all the dots, whatever the data1 list
-# dot.border.size: single numeric value of border dot width in mm. Write zero for no dot border. If data1 is a list, then dot.border.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.border.size will be used for all the dots, whatever the data1 list
-# dot.border.color: single character color string defining the color of the dot border (same border color for all the dots, whatever their categories). If dot.border.color == NULL, the border color will be the same as the dot color. A single integer is also accepted instead of a character string, that will be processed by fun_gg_palette()
-# line.size: single numeric value of line width in mm. If data1 is a list, then line.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to dots will be ignored. With a single value (latter possibility), the same line.size will be used for all the lines, whatever the data1 list
-# line.type: value indicating the kind of lines (see https://ggplot2.tidyverse.org/articles/ggplot2-specs.html) If data1 is a list, then line.type must be either (1) a list of single line kind values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single line kind value. With a list (former possibility), the value in compartments related to dots will be ignored. With a single value (latter possibility), the same line.type will be used for all the lines, whatever the data1 list
-# x.lim: 2 numeric values setting the x-axis range. Order of the 2 values matters (for inverted axis). If NULL, the range of the x column name of data1 will be used
-# x.lab: a character string or expression for x-axis label. If NULL, will use the first value of x (x column name of the first data frame in data1). Warning message if the elements in x are different between data frames in data1
-# x.log: either "no", "log2" (values in the x column name of the data1 data frame will be log2 transformed and x-axis will be log2 scaled) or "log10" (values in the x column name of the data1 data frame will be log10 transformed and x-axis will be log10 scaled)
-# x.tick.nb: approximate number of desired values labeling the x-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if x.log is "no", then the number of labeling values is set by ggplot2. If NULL and if x.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the x.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for x.lim = c(9, 1200)). WARNING: if non-NULL and if x.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
-# x.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if x.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$x.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
-# x.include.zero: logical. Does x.lim range include 0? Ignored if x.log is "log2" or "log10"
-# x.left.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to x.lim. If different from 0, add the range of the axis multiplied by x.left.extra.margin (e.g., abs(x.lim[2] - x.lim[1]) * x.left.extra.margin) to the left of x-axis
-# x.right.extra.margin: idem as x.left.extra.margin but to the right of x-axis
-# x.text.angle: integer value of the text angle for the x-axis labeling values, using the same rules as in ggplot2. Use positive value for clockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Use negative values for counterclockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
-# y.lim: 2 numeric values setting the y-axis range. Order of the 2 values matters (for inverted axis). If NULL, the range of the y column name of data1 will be used
-# y.lab: a character string or expression for y-axis label. If NULL, will use the first value of y (y column name of the first data frame in data1). Warning message if the elements in y are different between data frames in data1
-# y.log: either "no", "log2" (values in the y column name of the data1 data frame will be log2 transformed and y-axis will be log2 scaled) or "log10" (values in the y column name of the data1 data frame will be log10 transformed and y-axis will be log10 scaled)
-# y.tick.nb: approximate number of desired values labeling the y-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if y.log is "no", then the number of labeling values is set by ggplot2. If NULL and if y.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the y.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for y.lim = c(9, 1200)). WARNING: if non-NULL and if y.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
-# y.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if y.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$y.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
-# y.include.zero: logical. Does y.lim range include 0? Ignored if y.log is "log2" or "log10"
-# y.top.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to y.lim. If different from 0, add the range of the axis multiplied by y.top.extra.margin (e.g., abs(y.lim[2] - y.lim[1]) * y.top.extra.margin) to the top of y-axis
-# y.bottom.extra.margin: idem as y.top.extra.margin but to the bottom of y-axis
-# y.text.angle: integer value of the text angle for the y-axis labeling values, using the same rules as in ggplot2. Use positive value for clockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Use negative values for counterclockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
-# raster: logical. Dots in raster mode? If FALSE, dots from each "geom_point" from geom argument are plotted in vectorial mode (bigger pdf and long to display if lots of dots). If TRUE, dots from each "geom_point" from geom argument are plotted in matricial mode (smaller pdf and easy display if lots of dots, but it takes time to generate the layer). If TRUE, the raster.ratio argument is used to avoid an ellipsoid representation of the dots. If TRUE, solve the transparency problem with some GUI. Overriden by the non-NULL raster.threshold argument
-# raster.ratio: single numeric value indicating the height / width ratio of the graphic device used (for instance provided by the $dim compartment in the output of the fun_open() function). The default value is 1 because by default R opens a square graphic device. But this argument has to be set when using other device dimensions. Ignored if raster == FALSE
-# raster.threshold: positive integer value indicating the limit of the dot number above which "geom_point" layers from the geom argument switch from vectorial mode to matricial mode (see the raster argument). If any layer is matricial, then the raster.ratio argument is used to avoid an ellipsoid representation of the dots. If non-NULL, it overrides the raster argument
-# text.size: numeric value of the font size of the (1) axis numbers and axis legends and (2) texts in the graphic legend (in mm)
-# title: character string of the graph title
-# title.text.size: numeric value of the title font size in mm
-# legend.show: logical. Show legend? Not considered if categ argument is NULL, because this already generate no legend, excepted if legend.width argument is non-NULL. In that specific case (categ is NULL, legend.show is TRUE and legend.width is non-NULL), an empty legend space is created. This can be useful when desiring graphs of exactly the same width, whatever they have legends or not
-# legend.width: single proportion (between 0 and 1) indicating the relative width of the legend sector (on the right of the plot) relative to the width of the plot. Value 1 means that the window device width is split in 2, half for the plot and half for the legend. Value 0 means no room for the legend, which will overlay the plot region. Write NULL to inactivate the legend sector. In such case, ggplot2 will manage the room required for the legend display, meaning that the width of the plotting region can vary between graphs, depending on the text in the legend
-# legend.name: character string of the legend title. If legend.name is NULL and categ argument is not NULL, then legend.name <- categ. If data1 is a list, then legend.name must be a list of character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL, and other not
-# article: logical. If TRUE, use an article theme (article like). If FALSE, use a classic related ggplot theme. Use the add argument (e.g., add = "+ggplot2::theme_classic()" for the exact classic ggplot theme
-# grid: logical. Draw lines in the background to better read the box values? Not considered if article == FALSE (grid systematically present)
-# add: character string allowing to add more ggplot2 features (dots, lines, themes, facet, etc.). Ignored if NULL
-# WARNING: (1) the string must start with "+", (2) the string must finish with ")" and (3) each function must be preceded by "ggplot2::". Example: "+ ggplot2::coord_flip() + ggplot2::theme_bw()"
-# If the character string contains the "ggplot2::theme" string, then the article argument of fun_gg_scatter() (see above) is ignored with a warning. In addition, some arguments can be overwritten, like x.angle (check all the arguments)
-# Handle the add argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
-# WARNING: the call of objects inside the quotes of add can lead to an error if the name of these objects are some of the fun_gg_scatter() arguments. Indeed, the function will use the internal argument instead of the global environment object. Example article <- "a" in the working environment and add = '+ ggplot2::ggtitle(article)'. The risk here is to have TRUE as title. To solve this, use add = '+ ggplot2::ggtitle(get("article", envir = .GlobalEnv))'
-# return: logical. Return the graph parameters?
-# return.ggplot: logical. Return the ggplot object in the output list? Ignored if return argument is FALSE. WARNING: always assign the fun_gg_scatter() function (e.g., a <- fun_gg_scatter()) if return.ggplot argument is TRUE, otherwise, double plotting is performed. See $ggplot in the RETURN section below for more details
-# return.gtable: logical. Return the ggplot object as gtable of grobs in the output list? Ignored if plot argument is FALSE. Indeed, the graph must be plotted to get the grobs dispositions. See $gtable in the RETURN section below for more details
-# plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
-# warn.print: logical. Print warnings at the end of the execution? ? If FALSE, warning messages are never printed, but can still be recovered in the returned list. Some of the warning messages (those delivered by the internal ggplot2 functions) are not apparent when using the argument plot = FALSE
-# lib.path: character string indicating the absolute path of the required packages (see below). if NULL, the function will use the R library default folders
-# RETURN
-# a scatter plot if plot argument is TRUE
-# a list of the graph info if return argument is TRUE:
-# $data: the initial data with graphic information added. WARNING: if the x.log or y.log argument is not "no", x or y argument column of the data1 data frame are log2 or log10 converted in $data, respectively. Use 2^values or 10^$values to recover the initial values
-# $removed.row.nb: a list of the removed rows numbers in data frames (because of NA). NULL if no row removed
-# $removed.rows: a list of the removed rows in data frames (because of NA). NULL if no row removed
-# $plot: the graphic box and dot coordinates
-# $dots: dot coordinates
-# y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non-null or if y.log argument is different from "no")
-# y.second.tick.values: values of secondary ticks. NULL except if y.second.tick.nb argument is non-null or if y.log argument is different from "no")
-# $panel: the variable names used for the panels (NULL if no panels). WARNING: NA can be present according to ggplot2 upgrade to v3.3.0
-# $axes: the x-axis and y-axis info
-# $warn: the warning messages. Use cat() for proper display. NULL if no warning. WARNING: warning messages delivered by the internal ggplot2 functions are not apparent when using the argument plot = FALSE
-# $ggplot: ggplot object that can be used for reprint (use print($ggplot) or update (use $ggplot + ggplot2::...). NULL if return.ggplot argument is FALSE. Of note, a non-null $ggplot in the output list is sometimes annoying as the manipulation of this list prints the plot
-# $gtable: gtable object that can be used for reprint (use gridExtra::grid.arrange(...$ggplot) or with additionnal grobs (see the grob decomposition in the examples). NULL if return.ggplot argument is FALSE. Contrary to $ggplot, a non-NULL $gtable in the output list is not annoying as the manipulation of this list does not print the plot
-# REQUIRED PACKAGES
-# ggplot2
-# gridExtra
-# lemon (in case of use in the add argument)
-# scales
-# if raster plots are drawn (see the raster and raster.threshold arguments):
-# Cairo
-# grid
-# REQUIRED FUNCTIONS FROM THE cute PACKAGE
-# fun_gg_empty_graph()
-# fun_gg_palette()
-# fun_gg_point_rast()
-# fun_pack()
-# fun_check()
-# fun_round()
-# fun_scale()
-# fun_inter_ticks()
-# EXAMPLES
-# set.seed(1) ; obs1 <- data.frame(Km = c(2, 1, 6, 5, 4, 7), Time = c(2, 1, 6, 5, 4, 7)^2, Car = c("TUUT", "TUUT", "TUUT", "WIIM", "WIIM", "WIIM"), Color1 = rep(c("coral", "lightblue"), each = 3), stringsAsFactors = TRUE) ; fun_gg_scatter(data1 = obs1, x = "Km", y = "Time")
-# DEBUGGING
-# set.seed(1) ; obs1 <- data.frame(km = rnorm(1000, 10, 3), time = rnorm(1000, 10, 3), group1 = rep(c("A1", "A2"), 500), stringsAsFactors = TRUE) ; obs2 <-data.frame(km = rnorm(1000, 15, 3), time = rnorm(1000, 15, 3), group2 = rep(c("G1", "G2"), 500), stringsAsFactors = TRUE) ; set.seed(NULL) ; obs1$km[2:3] <- NA ; data1 = list(L1 = obs1, L2 = obs2) ; x = list(L1 = "km", L2 = "km") ; y = list(L1 = "time", L2 = "time") ; categ = list(L1 = "group1", L2 = "group2") ; categ = NULL ; categ.class.order = NULL ; color = NULL ; geom = "geom_point" ; geom.step.dir = "hv" ; geom.stick.base = NULL ; alpha = 0.5 ; dot.size = 2 ; dot.shape = 21 ; dot.border.size = 0.5 ; dot.border.color = NULL ; line.size = 0.5 ; line.type = "solid" ; x.lim = NULL ; x.lab = NULL ; x.log = "no" ; x.tick.nb = NULL ; x.second.tick.nb = NULL ; x.include.zero = FALSE ; x.left.extra.margin = 0.05 ; x.right.extra.margin = 0.05 ; x.text.angle = 0 ; y.lim = NULL ; y.lab = NULL ; y.log = "no" ; y.tick.nb = NULL ; y.second.tick.nb = NULL ; y.include.zero = FALSE ; y.top.extra.margin = 0.05 ; y.bottom.extra.margin = 0.05 ; y.text.angle = 0 ; raster = FALSE ; raster.ratio = 1 ; raster.threshold = NULL ; text.size = 12 ; title = "" ; title.text.size = 12 ; legend.show = TRUE ; legend.width = 0.5 ; legend.name = NULL ; article = TRUE ; grid = FALSE ; add = NULL ; return = FALSE ; return.ggplot = FALSE ; return.gtable = TRUE ; plot = TRUE ; warn.print = FALSE ; lib.path = NULL
-# function name
-function.name <- paste0(as.list(match.call(expand.dots=FALSE))[[1]], "()")
-arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
-arg.user.setting <- as.list(match.call(expand.dots=FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
-# end function name
-# required function checking
-req.function <- c(
-"fun_check",
-"fun_gg_just",
-"fun_gg_empty_graph",
-"fun_gg_palette",
-"fun_gg_point_rast",
-"fun_round",
-"fun_pack",
-"fun_scale",
-"fun_inter_ticks"
-)
-tempo <- NULL
-for(i1 in req.function){
-if(length(find(i1, mode = "function"))== 0L){
-tempo <- c(tempo, i1)
-}
-}
-if( ! is.null(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end required function checking
-# reserved words to avoid bugs (used in this function)
-reserved.words <- c("fake_x", "fake_y", "fake_categ")
-# end reserved words to avoid bugs (used in this function)
-# arg with no default values
-mandat.args <- c(
-"data1",
-"x",
-"y"
-)
-tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), ")")))
-if(any(tempo)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", "HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end arg with no default values
-# argument primary checking
-arg.check <- NULL #
-text.check <- NULL #
-checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
-ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
-tempo1 <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = data1, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OR A LIST OF DATA FRAMES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(x)){
-tempo1 <- fun_check(data = x, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = x, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(y)){
-tempo1 <- fun_check(data = y, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = y, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(categ)){
-tempo1 <- fun_check(data = categ, class = "vector", mode = "character", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = categ, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = categ, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(categ.class.order)){
-if(is.null(categ)){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT IS NOT NULL, BUT categ IS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = categ.class.order, class = "vector", mode = "character", fun.name = function.name)
-tempo2 <- fun_check(data = categ.class.order, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT MUST BE A VECTOR OF CHARACTER STRINGS OR A LIST OF VECTOR OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = categ.class.order, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(legend.name)){
-tempo1 <- fun_check(data = legend.name, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = legend.name, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = legend.name, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(color)){
-tempo1 <- fun_check(data = color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = color, class = "factor", na.contain = TRUE, fun.name = function.name)
-tempo3 <- fun_check(data = color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)
-tempo4 <- fun_check(data = color, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo4$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE & tempo3$problem == TRUE & tempo4$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT MUST BE A VECTOR (OF CHARACTER STRINGS OR INTEGERS) OR A FACTOR OR A LIST OF THESE POSSIBILITIES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = color, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo1 <- fun_check(data = geom, class = "vector", mode = "character", na.contain = FALSE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = geom, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = geom.step.dir, options = c("vh", "hv", "mid"), na.contain = FALSE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = geom.step.dir, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT MUST BE A SINGLE CHARACTER STRING (\"vh\" OR \"hv\" OR \"mid\") OR A LIST OF THESE CHARACTER STRINGS")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(geom.stick.base)){
-tempo1 <- fun_check(data = geom.stick.base, class = "vector", mode = "numeric", na.contain = FALSE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = color, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = geom.stick.base, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo1 <- fun_check(data = alpha, prop = TRUE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = alpha, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT MUST BE A SINGLE NUMERIC VALUE BETWEEN 0 AND 1 OR A LIST OF SUCH VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = dot.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
-tempo2 <- fun_check(data = dot.size, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = dot.shape, class = "vector", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = dot.shape, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT MUST BE A SINGLE SHAPE VALUE OR A LIST OF SINGLE SHAPE VALUES (SEE https://ggplot2.tidyverse.org/articles/ggplot2-specs.html)")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = dot.border.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
-tempo2 <- fun_check(data = dot.border.size, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(dot.border.color)){
-tempo1 <- fun_check(data = dot.border.color, class = "vector", mode = "character", length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-# integer colors -> gg_palette
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color MUST BE A SINGLE CHARACTER STRING OF COLOR OR A SINGLE INTEGER VALUE")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = dot.border.color, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo1 <- fun_check(data = line.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
-tempo2 <- fun_check(data = line.size, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo2$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-tempo1 <- fun_check(data = line.type, class = "vector", typeof = "integer", double.as.integer.allowed = FALSE, length = 1, fun.name = function.name)
-tempo2 <- fun_check(data = line.type, class = "vector", mode = "character", length = 1, fun.name = function.name)
-tempo3 <- fun_check(data = line.type, class = "list", na.contain = TRUE, fun.name = function.name)
-checked.arg.names <- c(checked.arg.names, tempo3$object.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE & tempo3$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT MUST BE A SINGLE LINE KIND VALUE OR A LIST OF SINGLE LINE KIND VALUES (SEE https://ggplot2.tidyverse.org/articles/ggplot2-specs.html)")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-if( ! is.null(x.lim)){
-tempo <- fun_check(data = x.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & any(x.lim %in% c(Inf, -Inf))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x.lim, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(x.lab)){
-if(all(class(x.lab) %in% "expression")){ # to deal with math symbols
-tempo <- fun_check(data = x.lab, class = "expression", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-tempo <- fun_check(data = x.lab, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x.lab, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = x.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(x.tick.nb)){
-tempo <- fun_check(data = x.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & x.tick.nb < 0){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(x.second.tick.nb)){
-tempo <- fun_check(data = x.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & x.second.tick.nb <= 0){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x.second.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = x.second.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = x.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.left.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.right.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = x.text.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-if( ! is.null(y.lim)){
-tempo <- fun_check(data = y.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & any(y.lim %in% c(Inf, -Inf))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.lim, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(y.lab)){
-if(all(class(y.lab) %in% "expression")){ # to deal with math symbols
-tempo <- fun_check(data = y.lab, class = "expression", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-tempo <- fun_check(data = y.lab, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.lab, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = y.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(y.tick.nb)){
-tempo <- fun_check(data = y.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & y.tick.nb < 0){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if( ! is.null(y.second.tick.nb)){
-tempo <- fun_check(data = y.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE & y.second.tick.nb <= 0){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y.second.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = y.second.tick.nb, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = y.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.top.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.bottom.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = y.text.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = raster, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = raster.ratio, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-if( ! is.null(raster.threshold)){
-tempo <- fun_check(data = raster.threshold, class = "vector", typeof = "integer", neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = raster.threshold, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = legend.show, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(legend.width)){
-tempo <- fun_check(data = legend.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = legend.width, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = article, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = grid, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(add)){
-tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = add, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.ggplot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = return.gtable, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
-if( ! is.null(lib.path)){
-tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
-if(tempo$problem == FALSE){
-if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
-tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
-text.check <- c(text.check, tempo.cat)
-arg.check <- c(arg.check, TRUE)
-}
-}
-}else{
-# no fun_check test here, it is just for checked.arg.names
-tempo <- fun_check(data = lib.path, class = "vector")
-checked.arg.names <- c(checked.arg.names, tempo$object.name)
-}
-if(any(arg.check) == TRUE){
-stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
-}
-# source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
-# end argument primary checking
-
-# second round of checking and data preparation
-# management of NA arguments
-tempo.arg <- names(arg.user.setting) # values provided by the user
-tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length)== 1L # no argument provided by the user can be just NA
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NA arguments
-# management of NULL arguments
-tempo.arg <-c(
-"data1",
-# "x", # inactivated because of hline or vline
-# "y", # inactivated because of hline or vline
-"geom",
-"geom.step.dir",
-# "geom.stick.base", # inactivated because can be null
-"alpha",
-"dot.size",
-"dot.shape",
-"dot.border.size",
-"line.size",
-"line.type",
-"x.log",
-"x.include.zero",
-"x.left.extra.margin",
-"x.right.extra.margin",
-"x.text.angle",
-"y.log",
-"y.include.zero",
-"y.top.extra.margin",
-"y.bottom.extra.margin",
-"y.text.angle",
-"raster",
-"raster.ratio",
-"text.size",
-"title",
-"title.text.size",
-"legend.show",
-# "legend.width", # inactivated because can be null
-"article",
-"grid",
-"return",
-"return.ggplot",
-"return.gtable",
-"plot",
-"warn.print"
-)
-tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
-if(any(tempo.log) == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
-}
-# end management of NULL arguments
-# code that protects set.seed() in the global environment
-# end code that protects set.seed() in the global environment
-# warning initiation
-ini.warning.length <- options()$warning.length
-options(warning.length = 8170)
-warn <- NULL
-warn.count <- 0
-# end warning initiation
-# other checkings
-# check list lengths (and names of data1 compartments if present)
-list.color <- NULL
-list.geom <- NULL
-list.geom.step.dir <- NULL
-list.geom.stick.base <- NULL
-list.alpha <- NULL
-list.dot.size <- NULL
-list.dot.shape <- NULL
-list.dot.border.size <- NULL
-list.dot.border.color <- NULL
-list.line.size <- NULL
-list.line.type <- NULL
-if(all(class(data1) == "list")){
-if(length(data1) > 6){
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A LIST OF 6 DATA FRAMES MAXIMUM (6 OVERLAYS MAX)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(is.null(names(data1))){
-names(data1) <- paste0("L", 1:length(data1))
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL NAME COMPARTMENT OF data1 LIST -> NAMES RESPECTIVELY ATTRIBUTED TO EACH COMPARTMENT:\n", paste(names(data1), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-if( ! is.null(x)){
-if( ! (all(class(x) == "list") & length(data1) == length(x))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}else{
-x <- vector("list", length(data1))
-}
-if( ! is.null(y)){
-if( ! (all(class(y) == "list") & length(data1) == length(y))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}else{
-y <- vector("list", length(data1))
-}
-if( ! is.null(categ)){
-if( ! (all(class(categ) == "list") & length(data1) == length(categ))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if( ! is.null(categ.class.order)){
-if( ! (all(class(categ.class.order) == "list") & length(data1) == length(categ.class.order))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if( ! is.null(color)){
-if( ! ((all(class(color) == "list") & length(data1) == length(color)) | ((all(mode(color) == "character") | all(mode(color) == "numeric")) & length(color)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER STRING OR INTEGER")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if((all(mode(color) == "character") | all(mode(color) == "numeric")) & length(color)== 1L){ # convert the single value into a list of single value
-list.color <- vector(mode = "list", length = length(data1))
-list.color[] <- color
-}
-}
-if( ! ((all(class(geom) == "list") & length(data1) == length(geom)) | (all(mode(geom) == "character") & length(geom)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(geom) == "character") & length(geom)== 1L){ # convert the single value into a list of single value
-list.geom <- vector(mode = "list", length = length(data1))
-list.geom[] <- geom
-}
-if( ! ((all(class(geom.step.dir) == "list") & length(data1) == length(geom.step.dir)) | (all(mode(geom.step.dir) == "character") & length(geom.step.dir)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(geom.step.dir) == "character") & length(geom.step.dir)== 1L){ # convert the single value into a list of single value
-list.geom.step.dir <- vector(mode = "list", length = length(data1))
-list.geom.step.dir[] <- geom.step.dir
-}
-if( ! is.null(geom.stick.base)){
-if( ! ((all(class(geom.stick.base) == "list") & length(data1) == length(geom.stick.base)) | (all(mode(geom.stick.base) == "numeric") & length(geom.stick.base)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(geom.stick.base) == "numeric") & length(geom.stick.base)== 1L){ # convert the single value into a list of single value
-list.geom.stick.base <- vector(mode = "list", length = length(data1))
-list.geom.stick.base[] <- geom.stick.base
-}
-}
-if( ! ((all(class(alpha) == "list") & length(data1) == length(alpha)) | (all(mode(alpha) == "numeric") & length(alpha)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(alpha) == "numeric") & length(alpha)== 1L){ # convert the single value into a list of single value
-list.alpha <- vector(mode = "list", length = length(data1))
-list.alpha[] <- alpha
-}
-if( ! ((all(class(dot.size) == "list") & length(data1) == length(dot.size)) | (all(mode(dot.size) == "numeric") & length(dot.size)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(dot.size) == "numeric") & length(dot.size)== 1L){ # convert the single value into a list of single value
-list.dot.size <- vector(mode = "list", length = length(data1))
-list.dot.size[] <- dot.size
-}
-if( ! ((all(class(dot.shape) == "list") & length(data1) == length(dot.shape)) | (all(mode(dot.shape) != "list") & length(dot.shape)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE SHAPE VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(dot.shape) != "list") & length(dot.shape)== 1L){ # convert the single value into a list of single value
-list.dot.shape <- vector(mode = "list", length = length(data1))
-list.dot.shape[] <- dot.shape
-}
-if( ! ((all(class(dot.border.size) == "list") & length(data1) == length(dot.border.size)) | (all(mode(dot.border.size) == "numeric") & length(dot.border.size)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(dot.border.size) == "numeric") & length(dot.border.size)== 1L){ # convert the single value into a list of single value
-list.dot.border.size <- vector(mode = "list", length = length(data1))
-list.dot.border.size[] <- dot.border.size
-}
-if( ! is.null(dot.border.color)){
-if( ! ((all(class(dot.border.color) == "list") & length(data1) == length(dot.border.color)) | ((all(mode(dot.border.color) == "character") | all(mode(dot.border.color) == "numeric")) & length(dot.border.color)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER STRING OR INTEGER")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if((all(mode(dot.border.color) == "character") | all(mode(dot.border.color) == "numeric")) & length(dot.border.color)== 1L){ # convert the single value into a list of single value
-list.dot.border.color <- vector(mode = "list", length = length(data1))
-list.dot.border.color[] <- dot.border.color
-}
-}
-if( ! ((all(class(line.size) == "list") & length(data1) == length(line.size)) | (all(mode(line.size) == "numeric") & length(line.size)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(line.size) == "numeric") & length(line.size)== 1L){ # convert the single value into a list of single value
-list.line.size <- vector(mode = "list", length = length(data1))
-list.line.size[] <- line.size
-}
-if( ! ((all(class(line.type) == "list") & length(data1) == length(line.type)) | (all(mode(line.type) != "list") & length(line.type)== 1L))){ # list of same length as data1 or single value
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE LINE KIND VALUE")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(all(mode(line.type) != "list") & length(line.type)== 1L){ # convert the single value into a list of single value
-list.line.type <- vector(mode = "list", length = length(data1))
-list.line.type[] <- line.type
-}
-if( ! is.null(legend.name)){
-if( ! (all(class(legend.name) == "list") & length(data1) == length(legend.name))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-}
-# end check list lengths (and names of data1 compartments if present)
-# conversion into lists
-if(all(is.data.frame(data1))){
-data1 <- list(L1 = data1)
-if(all(class(x) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-x <- list(L1 = x)
-}
-if(all(class(y) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-y <- list(L1 = y)
-}
-if( ! is.null(categ)){
-if(all(class(categ) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-categ <- list(L1 = categ)
-}
-}
-if( ! is.null(categ.class.order)){
-if(all(class(categ.class.order) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-categ.class.order <- list(L1 = categ.class.order)
-}
-}
-if( ! is.null(color)){
-if(all(class(color) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-color <- list(L1 = color)
-}
-}
-if(all(class(geom) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-geom <- list(L1 = geom)
-}
-if(all(class(geom.step.dir) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-geom.step.dir <- list(L1 = geom.step.dir)
-}
-if( ! is.null(geom.stick.base)){
-if(all(class(geom.stick.base) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-geom.stick.base <- list(L1 = geom.stick.base)
-}
-}
-if(all(class(alpha) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-alpha <- list(L1 = alpha)
-}
-if(all(class(dot.size) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-dot.size <- list(L1 = dot.size)
-}
-if(all(class(dot.shape) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-dot.shape <- list(L1 = dot.shape)
-}
-if(all(class(dot.border.size) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-dot.border.size <- list(L1 = dot.border.size)
-}
-if( ! is.null(dot.border.color)){
-if(all(class(dot.border.color) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-dot.border.color <- list(L1 = dot.border.color)
-}
-}
-if(all(class(line.size) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-line.size <- list(L1 = line.size)
-}
-if(all(class(line.type) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-line.type <- list(L1 = line.type)
-}
-if( ! is.null(legend.name)){
-if(all(class(legend.name) == "list")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-legend.name <- list(L1 = legend.name)
-}
-}
-}else if( ! all(sapply(data1, FUN = "class") == "data.frame")){ # if not a data frame, data1 can only be a list, as tested above
-tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OR A LIST OF DATA FRAMES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-# single value converted into list now reattributed to the argument name
-if( ! is.null(color)){
-if( ! is.null(list.color)){
-color <- list.color
-}
-}
-if( ! is.null(list.geom)){
-geom <- list.geom
-}
-if( ! is.null(list.geom.step.dir)){
-geom.step.dir <- list.geom.step.dir
-}
-if( ! is.null(geom.stick.base)){
-if( ! is.null(list.geom.stick.base)){
-geom.stick.base <- list.geom.stick.base
-}
-}
-if( ! is.null(list.alpha)){
-alpha <- list.alpha
-}
-if( ! is.null(list.dot.size)){
-dot.size <- list.dot.size
-}
-if( ! is.null(list.dot.shape)){
-dot.shape <- list.dot.shape
-}
-if( ! is.null(list.dot.border.size)){
-dot.border.size <- list.dot.border.size
-}
-if( ! is.null(dot.border.color)){
-if( ! is.null(list.dot.border.color)){
-dot.border.color <- list.dot.border.color
-}
-}
-if( ! is.null(list.line.size)){
-line.size <- list.line.size
-}
-if( ! is.null(list.line.type)){
-line.type <- list.line.type
-}
-# end single value converted into list now reattributed to the argument name
-# data, x, y, geom, alpha, dot.size, shape, dot.border.size, line.size, line.type, legend.name are list now
-# if non-null, categ, categ.class.order, legend.name, color, dot.border.color are list now
-# end conversion into lists
-# verif of add
-if( ! is.null(add)){
-if( ! grepl(pattern = "^\\s*\\+", add)){ # check that the add string start by +
-tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-
-}else if( ! grepl(pattern = "(ggplot2|lemon)\\s*::", add)){ #
-tempo.cat <- paste0("ERROR IN ", function.name, ": FOR EASIER FUNCTION DETECTION, add ARGUMENT MUST CONTAIN \"ggplot2::\" OR \"lemon::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if( ! grepl(pattern = ")\\s*$", add)){ # check that the add string finished by )
-tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-# end verif of add
-# management of add containing facet
-facet.categ <- NULL
-if( ! is.null(add)){
-facet.check <- TRUE
-tempo <- unlist(strsplit(x = add, split = "\\s*\\+\\s*(ggplot2|lemon)\\s*::\\s*")) #
-tempo <- sub(x = tempo, pattern = "^facet_wrap", replacement = "ggplot2::facet_wrap")
-tempo <- sub(x = tempo, pattern = "^facet_grid", replacement = "ggplot2::facet_grid")
-tempo <- sub(x = tempo, pattern = "^facet_rep", replacement = "lemon::facet_rep")
-if(length(data1) > 1 & (any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")) | grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid"))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nfacet PANELS CANNOT BE USED IF MORE THAN ONE DATA FRAME IN THE data1 ARGUMENT\nPLEASE REWRITE THE add STRING AND RERUN")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-if(any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap"))){
-tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")])))
-facet.categ <- list(names(tempo1$params$facets)) # list of length 1
-tempo.text <- "facet_wrap OR facet_rep_wrap"
-facet.check <- FALSE
-}else if(grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")){
-tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")])))
-facet.categ <- list(c(names(tempo1$params$rows), names(tempo1$params$cols))) # list of length 1
-tempo.text <- "facet_grid OR facet_rep_grid"
-facet.check <- FALSE
-}
-if(facet.check == FALSE & ! all(facet.categ %in% names(data1[[1]]))){ # WARNING: all(facet.categ %in% names(data1)) is TRUE when facet.categ is NULL
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF \"", tempo.text, "\" STRING IN THE add ARGUMENT BUT PROBLEM OF VARIABLE DETECTION (COLUMN NAMES OF data1)\nTHE DETECTED VARIABLES ARE:\n", paste(facet.categ, collapse = " "), "\nTHE data1 COLUMN NAMES ARE:\n", paste(names(data1[[1]]), collapse = " "), "\nPLEASE REWRITE THE add STRING AND RERUN")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-}
-# if facet.categ is not NULL, it is a list of length 1 now
-# end management of add containing facet
-# legend name filling
-if(is.null(legend.name) & ! is.null(categ)){
-legend.name <- categ
-}else if(is.null(legend.name) & is.null(categ)){
-legend.name <- vector("list", length(data1)) # null list
-}
-# legend.name not NULL anymore (list)
-# end legend name filling
-# ini categ for legend display
-fin.lg.disp <- vector("list", 6) # will be used at the end to display or not legends
-fin.lg.disp[] <- FALSE
-legend.disp <- vector("list", length(data1))
-if(is.null(categ) | legend.show == FALSE){
-legend.disp[] <- FALSE
-}else{
-for(i2 in 1:length(data1)){
-if(is.null(categ[[i2]])){
-legend.disp[[i2]] <- FALSE
-}else{
-legend.disp[[i2]] <- TRUE
-}
-}
-}
-# end ini categ for legend display
-# integer colors into gg_palette
-tempo.check.color <- NULL
-for(i1 in 1:length(data1)){
-if(any(is.na(color[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), ": color ARGUMENT CANNOT CONTAIN NA")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo.check.color <- c(tempo.check.color, fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem)
-}
-tempo.check.color <- ! tempo.check.color # invert TRUE and FALSE because if integer, then problem = FALSE
-if(any(tempo.check.color == TRUE)){ # convert integers into colors
-tempo.integer <- unlist(color[tempo.check.color])
-tempo.color <- fun_gg_palette(max(tempo.integer, na.rm = TRUE))
-for(i1 in 1:length(data1)){
-if(tempo.check.color[i1] == TRUE){
-color[[i1]] <-tempo.color[color[[i1]]]
-}
-}
-}
-# end integer colors into gg_palette
-# loop (checking inside list compartment)
-compart.null.color <- 0 # will be used to attribute a color when color is non-null but a compartment of color is NULL
-data1.ini <- data1 # to report NA removal
-removed.row.nb <- vector("list", length = length(data1)) # to report NA removal. Contains NULL
-removed.rows <- vector("list", length = length(data1)) # to report NA removal. Contains NULL
-for(i1 in 1:length(data1)){
-tempo <- fun_check(data = data1[[i1]], data.name = ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "data.frame", na.contain = TRUE, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-# reserved word checking
-if(any(names(data1[[i1]]) %in% reserved.words)){ # I do not use fun_name_change() because cannot control y before creating "fake_y". But ok because reserved are not that common
-tempo.cat <- paste0("ERROR IN ", function.name, ": COLUMN NAMES OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " ARGUMENT CANNOT BE ONE OF THESE WORDS\n", paste(reserved.words, collapse = " "), "\nTHESE ARE RESERVED FOR THE ", function.name, " FUNCTION")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if( ! (is.null(add))){
-if(any(sapply(X = reserved.words, FUN = grepl, x = add))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF COLUMN NAMES OF data1 IN THE add ARGUMENT STRING, THAT CORRESPOND TO RESERVED STRINGS FOR ", function.name, "\nFOLLOWING COLUMN NAMES HAVE TO BE CHANGED:\n", paste(arg.names[sapply(X = reserved.words, FUN = grepl, x = add)], collapse = "\n"), "\nFOR INFORMATION, THE RESERVED WORDS ARE:\n", paste(reserved.words, collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(any(sapply(X = arg.names, FUN = grepl, x = add))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NAMES OF ", function.name, " ARGUMENTS DETECTED IN THE add STRING:\n", paste(arg.names[sapply(X = arg.names, FUN = grepl, x = add)], collapse = "\n"), "\nRISK OF WRONG OBJECT USAGE INSIDE ", function.name)
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end reserved word checking
-# check of geom now because required for y argument
-tempo <- fun_check(data = geom[[i1]], data.name = ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), options = c("geom_point", "geom_line", "geom_path", "geom_step", "geom_hline", "geom_vline", "geom_stick"), length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(geom[[i1]] == "geom_step" & is.null(geom.step.dir[[i1]])){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(geom.step.dir)== 1L, "geom.step.dir", paste0("ELEMENT ", i1, " OF geom.step.dir ARGUMENT")), ": geom.step.dir ARGUMENT CANNOT BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("ELEMENT ", i1, " OF geom")), " ARGUMENT IS \"geom_step\"\nHERE geom.step.dir ARGUMENT IS: ", paste(geom.step.dir[[i1]], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(geom[[i1]] == "geom_step" & ! is.null(geom.step.dir[[i1]])){
-tempo <- fun_check(data = geom.step.dir[[i1]], data.name = ifelse(length(geom.step.dir)== 1L, "geom.step.dir", paste0("geom.step.dir NUMBER ", i1)), options = c("vh", "hv", "mid"), length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if( ! (is.null(geom.stick.base))){
-if(geom[[i1]] == "geom_stick" & ! is.null(geom.stick.base[[i1]])){
-tempo <- fun_check(data = geom.stick.base[[i1]], data.name = ifelse(length(geom.stick.base)== 1L, "geom.stick.base", paste0("geom.stick.base NUMBER ", i1)), mode = "numeric", length = 1, na.contain = FALSE, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-}
-# end check of geom now because required for y argument
-if(is.null(x[[i1]])){
-if(all(geom[[i1]] != "geom_hline")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": x ARGUMENT CANNOT BE NULL EXCEPT IF ", ifelse(length(geom)== 1L, "x", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_hline\"\nHERE geom ARGUMENT IS: ", paste(geom[[i1]], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-x[[i1]] <- "fake_x"
-data1[[i1]] <- cbind(data1[[i1]], fake_x = NA, stringsAsFactors = TRUE)
-data1[[i1]][, "fake_x"] <- as.numeric(data1[[i1]][, "fake_x"])
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT ASSOCIATED TO ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT ", geom[[i1]], " -> FAKE COLUMN ADDED TO DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", NAMED \"fake_x\" FOR FINAL DRAWING")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(all(geom[[i1]] == "geom_hline")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": x ARGUMENT MUST BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_hline\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo <- fun_check(data = x[[i1]], data.name = ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), class = "vector", mode = "character", length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(is.null(y[[i1]])){
-if(all(geom[[i1]] != "geom_vline")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": y ARGUMENT CANNOT BE NULL EXCEPT IF ", ifelse(length(geom)== 1L, "y", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_vline\"\nHERE geom ARGUMENT IS: ", paste(geom[[i1]], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-y[[i1]] <- "fake_y"
-data1[[i1]] <- cbind(data1[[i1]], fake_y = NA, stringsAsFactors = TRUE)
-data1[[i1]][, "fake_y"] <- as.numeric(data1[[i1]][, "fake_y"])
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT ASSOCIATED TO ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT ", geom[[i1]], " -> FAKE COLUMN ADDED TO DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", NAMED \"fake_y\" FOR FINAL DRAWING")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else{
-if(all(geom[[i1]] == "geom_vline")){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": y ARGUMENT MUST BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_vline\"")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo <- fun_check(data = y[[i1]], data.name = ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), class = "vector", mode = "character", length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-# x[[i1]] and y[[i1]] not NULL anymore
-if( ! (x[[i1]] %in% names(data1[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(x[[i1]], collapse = " "))))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if( ! (y[[i1]] %in% names(data1[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(y[[i1]], collapse = " "))))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo <- fun_check(data = data1[[i1]][, x[[i1]]], data.name = ifelse(length(x)== 1L, "x ARGUMENT (AS COLUMN NAME OF data1 DATA FRAME)", paste0("ELEMENT ", i1, " OF x ARGUMENT", " (AS COLUMN NAME OF data1 DATA FRAME NUMBER ", i1, ")")), class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo <- fun_check(data = data1[[i1]][, y[[i1]]], data.name = ifelse(length(y)== 1L, "y ARGUMENT (AS COLUMN NAME OF data1 DATA FRAME)", paste0("ELEMENT ", i1, " OF y ARGUMENT", " (AS COLUMN NAME OF data1 DATA FRAME NUMBER ", i1, ")")), class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(x[[i1]] == "fake_x" & y[[i1]] == "fake_y"){ # because the code cannot accept to be both "fake_x" and "fake_y" at the same time
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2\nTHE CODE CANNOT ACCEPT x AND y TO BE \"fake_x\" AND \"fake_y\" IN THE SAME DATA FRAME ", i1, " ")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(( ! is.null(categ)) & ( ! is.null(categ[[i1]]))){ # is.null(categ[[i1]]) works even if categ is NULL # is.null(categ[[i1]]) works even if categ is NULL # if categ[[i1]] = NULL, fake_categ will be created later on
-tempo <- fun_check(data = categ[[i1]], data.name = ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")),, class = "vector", mode = "character", length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if( ! (categ[[i1]] %in% names(data1[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(categ[[i1]], collapse = " "))))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-tempo1 <- fun_check(data = data1[[i1]][, categ[[i1]]], data.name = ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
-tempo2 <- fun_check(data = data1[[i1]][, categ[[i1]]], data.name = ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "factor", na.contain = TRUE, fun.name = function.name)
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " MUST BE A FACTOR OR CHARACTER VECTOR")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(tempo1$problem == FALSE){
-data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE CHARACTER COLUMN HAS BEEN CONVERTED TO FACTOR")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+######## fun_open() #### open a GUI or pdf graphic window
-}
-if(geom[[i1]] == "geom_vline" | geom[[i1]] == "geom_hline"){
-if(length(unique(data1[[i1]][, categ[[i1]]])) != nrow(data1[[i1]])){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(geom)== 1L, "geom OF data1 ARGUMENT", paste0("geom NUMBER ", i1, " OF DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " ARGUMENT IS ", geom[[i1]], ", MEANING THAT ", ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " MUST HAVE A DIFFERENT CLASS PER LINE OF data1 (ONE x VALUE PER CLASS)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-}else if(( ! is.null(categ)) & is.null(categ[[i1]])){ # is.null(categ[[i1]]) works even if categ is NULL # if categ[[i1]] = NULL, fake_categ will be created. WARNING: is.null(categ[[i1]]) means no legend display (see above), because categ has not been precised. This also means a single color for data1[[i1]]
-if(length(color[[i1]]) > 1){ # 0 means is.null(color[[i1]]) or is.null(color) and 1 is ok -> single color for data1[[i1]]
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT BUT CORRESPONDING COLORS IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " HAS LENGTH OVER 1\n", paste(color[[i1]], collapse = " "), "\nWHICH IS NOT COMPATIBLE WITH NULL CATEG -> COLOR RESET TO A SINGLE COLOR")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-color[i1] <- list(NULL) # will provide a single color below # Warning color[[i1]] <- NULL removes the compartment
-}
-categ[[i1]] <- "fake_categ"
-data1[[i1]] <- cbind(data1[[i1]], fake_categ = "", stringsAsFactors = TRUE)
-# inactivated because give a different color to different "Line_" categ while a single color for all the data1[[i1]] required. Thus, put back after the color management
-# if(geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline"){
-# data1[[i1]][, "fake_categ"] <- paste0("Line_", 1:nrow(data1[[i1]]))
-# }else{
-data1[[i1]][, "fake_categ"] <- data1[[i1]][, "fake_categ"] # as.numeric("") create a vector of NA but class numeric
-# }
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT -> FOR DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT:", paste0("NUMBER ", i1, " OF data1 ARGUMENT:")), "\n- FAKE \"fake_categ\" COLUMN ADDED FILLED WITH \"\"(OR WITH \"Line_...\" FOR LINES)\n- SINGLE COLOR USED FOR PLOTTING\n- NO LEGEND DISPLAYED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# OK: if categ is not NULL, all the non-null categ columns of data1 are factors from here
-
-# management of log scale and Inf removal
-if(x[[i1]] != "fake_x"){
-if(any(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN ", ifelse(length(categ)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(y[[i1]] != "fake_y"){
-if(any(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN ", ifelse(length(categ)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# log conversion
-if(x.log != "no"){
-tempo1 <- ! is.finite(data1[[i1]][, x[[i1]]]) # where are initial NA and Inf
-data1[[i1]][, x[[i1]]] <- suppressWarnings(get(x.log)(data1[[i1]][, x[[i1]]]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-if(any( ! (tempo1 | is.finite(data1[[i1]][, x[[i1]]])))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN ", ifelse(length(categ)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(y.log != "no"){
-tempo1 <- ! is.finite(data1[[i1]][, y[[i1]]]) # where are initial NA and Inf
-data1[[i1]][, y[[i1]]] <- suppressWarnings(get(y.log)(data1[[i1]][, y[[i1]]]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-if(any( ! (tempo1 | is.finite(data1[[i1]][, y[[i1]]])))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN ", ifelse(length(categ)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# Inf removal
-# removed.row.nb[[i1]] <- NULL # already NULL and Warning this removes the compartment
-removed.rows[[i1]] <- data.frame(stringsAsFactors = FALSE)
-if(any(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])) detects only Inf
-removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], which(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]]))))
-}
-if(any(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])) detects only Inf
-removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], which(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]]))))
-}
-if( ! is.null(removed.row.nb[[i1]])){
-removed.row.nb[[i1]] <- unique(removed.row.nb[[i1]]) # to remove the duplicated positions (NA in both x and y)
-removed.rows[[i1]] <- rbind(removed.rows[[i1]], data1.ini[[i1]][removed.row.nb[[i1]], ]) # here data1.ini used to have the y = O rows that will be removed because of Inf creation after log transformation
-data1[[i1]] <- data1[[i1]][-removed.row.nb[[i1]], ]
-data1.ini[[i1]] <- data1.ini[[i1]][-removed.row.nb[[i1]], ] #
-}
-# From here, data1 and data.ini have no more Inf
-# end Inf removal
-# x.lim and y.lim dealt later on, after the end f the loop
-# end management of log scale and Inf removal
-# na detection and removal
-column.check <- unique(unlist(c( # unlist because creates a list
-if(x[[i1]] == "fake_x"){NULL}else{x[[i1]]},
-if(y[[i1]] == "fake_y"){NULL}else{y[[i1]]},
-if( ! is.null(categ)){if(is.null(categ[[i1]])){NULL}else{categ[[i1]]}},
-if( ! is.null(facet.categ)){if(is.null(facet.categ[[i1]])){NULL}else{facet.categ[[i1]]}}
-))) # dot.categ because can be a 3rd column of data1
-if(any(is.na(data1[[i1]][, column.check]))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS ", paste(column.check, collapse = " "), " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-for(i3 in 1:length(column.check)){
-if(any(is.na(data1[[i1]][, column.check[i3]]))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i3], " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-tempo <- unique(unlist(lapply(lapply(c(data1[[i1]][column.check]), FUN = is.na), FUN = which)))
-removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], tempo)
-removed.rows[[i1]] <- rbind(removed.rows[[i1]], data1.ini[[i1]][tempo, ]) # # tempo used because removed.row.nb is not empty. Here data1.ini used to have the non NA rows that will be removed because of NAN creation after log transformation (neg values for instance)
-column.check <- column.check[ ! (column.check == x[[i1]] | column.check == y[[i1]])] # remove x and y to keep quali columns
-if(length(tempo) != 0){
-data1[[i1]] <- data1[[i1]][-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
-data1.ini[[i1]] <- data1.ini[[i1]][-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers than the former
-for(i4 in 1:length(column.check)){
-if(any( ! unique(removed.rows[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]]))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i4], " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA REMOVAL\n(IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[[i1]][, column.check[i4]])[ ! unique(removed.rows[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]])], collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-tempo.levels <- levels(data1[[i1]][, column.check[i4]])[levels(data1[[i1]][, column.check[i4]]) %in% unique(as.character(data1[[i1]][, column.check[i4]]))]
-data1[[i1]][, column.check[i4]] <- factor(as.character(data1[[i1]][, column.check[i4]]), levels = tempo.levels)
-if(column.check[i4] %in% categ[[i1]] & ! is.null(categ.class.order)){
-categ.class.order[[i1]] <- levels(data1[[i1]][, column.check[i4]])[levels(data1[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]])] # remove the absent class in the categ.class.order vector
-data1[[i1]][, column.check[i4]] <- factor(as.character(data1[[i1]][, column.check[i4]]), levels = unique(categ.class.order[[i1]]))
-}
-}
-}
-}
-}
-# end na detection and removal
-# From here, data1 and data.ini have no more NA or NaN in x, y, categ (if categ != NULL) and facet.categ (if categ != NULL)
-if( ! is.null(categ.class.order)){
-# the following check will be done several times but I prefer to keep it here, after the creation of categ
-if(is.null(categ[[i1]]) & ! is.null(categ.class.order[[i1]])){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ ARGUMENT CANNOT BE NULL IF COMPARTMENT ", i1, " OF categ.class.order ARGUMENT IS NOT NULL: ", paste(categ.class.order, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-if(is.null(categ.class.order[[i1]])){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE categ.class.order COMPARTMENT ", i1, " IS NULL. ALPHABETICAL ORDER WILL BE APPLIED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-data1[[i1]][, categ[[i1]]] <- factor(as.character(data1[[i1]][, categ[[i1]]])) # if already a factor, change nothing, if characters, levels according to alphabetical order
-categ.class.order[[i1]] <- levels(data1[[i1]][, categ[[i1]]]) # character vector that will be used later
-}else{
-tempo <- fun_check(data = categ.class.order[[i1]], data.name = paste0("COMPARTMENT ", i1 , " OF categ.class.order ARGUMENT"), class = "vector", mode = "character", length = length(levels(data1[[i1]][, categ[[i1]]])), fun.name = function.name) # length(data1[, categ[i1]) -> if data1[, categ[i1] was initially character vector, then conversion as factor after the NA removal, thus class number ok. If data1[, categ[i1] was initially factor, no modification after the NA removal, thus class number ok
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(any(duplicated(categ.class.order[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(categ.class.order[[i1]], collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if( ! (all(categ.class.order[[i1]] %in% unique(data1[[i1]][, categ[[i1]]])) & all(unique(data1[[i1]][, categ[[i1]]]) %in% categ.class.order[[i1]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ.class.order ARGUMENT MUST BE CLASSES OF COMPARTMENT ", i1, " OF categ ARGUMENT\nHERE IT IS:\n", paste(categ.class.order[[i1]], collapse = " "), "\nFOR COMPARTMENT ", i1, " OF categ.class.order AND IT IS:\n", paste(unique(data1[[i1]][, categ[[i1]]]), collapse = " "), "\nFOR COLUMN ", categ[[i1]], " OF data1 NUMBER ", i1)
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]], levels = categ.class.order[[i1]]) # reorder the factor
-}
-names(categ.class.order)[i1] <- categ[[i1]]
-}
-}
-# OK: if categ.class.order is not NULL, all the NULL categ.class.order columns of data1 are character from here
-if( ! is.null(legend.name[[i1]])){
-tempo <- fun_check(data = legend.name[[i1]], data.name = ifelse(length(legend.name)== 1L, "legend.name", paste0("legend.name NUMBER ", i1)),, class = "vector", mode = "character", length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if( ! is.null(color)){ # if color is NULL, will be filled later on
-# check the nature of color
-if(is.null(color[[i1]])){
-compart.null.color <- compart.null.color + 1
-color[[i1]] <- grey(compart.null.color / 8) # cannot be more than 7 overlays. Thus 7 different greys. 8/8 is excluded because white dots
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL COLOR IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " ASSOCIATED TO ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", SINGLE COLOR ", paste(color[[i1]], collapse = " "), " HAS BEEN ATTRIBUTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-tempo1 <- fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name) # na.contain = TRUE in case of colum of data1
-tempo2 <- fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "factor", na.contain = TRUE, fun.name = function.name) # idem
-if(tempo1$problem == TRUE & tempo2$problem == TRUE){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A FACTOR OR CHARACTER VECTOR OR INTEGER VECTOR") # integer possible because dealt above
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if( ! (all(color[[i1]] %in% colors() | grepl(pattern = "^#", color[[i1]])))){ # check that all strings of low.color start by #
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors(): ", paste(unique(color[[i1]]), collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(any(is.na(color[[i1]]))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), ", THE COLORS:\n", paste(unique(color[[i1]]), collapse = " "), "\nCONTAINS NA")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# end check the nature of color
-# check the length of color
-if(is.null(categ) & length(color[[i1]]) != 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A SINGLE COLOR IF categ IS NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if( ! is.null(categ)){
-# No problem of NA management by ggplot2 because already removed
-if(categ[[i1]] == "fake_categ" & length(color[[i1]]) != 1){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A SINGLE COLOR IF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IS NULL")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(length(color[[i1]]) == length(unique(data1[[i1]][, categ[[i1]]]))){ # here length(color) is equal to the different number of categ
-data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE FOLLOWING COLORS:\n", paste(color[[i1]], collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else if(length(color[[i1]]) == length(data1[[i1]][, categ[[i1]]])){# here length(color) is equal to nrow(data1[[i1]]) -> Modif to have length(color) equal to the different number of categ (length(color) == length(levels(data1[[i1]][, categ[[i1]]])))
-data1[[i1]] <- cbind(data1[[i1]], color = color[[i1]], stringsAsFactors = TRUE)
-tempo.check <- unique(data1[[i1]][ , c(categ[[i1]], "color")])
-if( ! (nrow(data1[[i1]]) == length(color[[i1]]) & nrow(tempo.check) == length(unique(data1[[i1]][ , categ[[i1]]])))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color")), " ARGUMENT HAS THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nBUT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF THIS categ:\n", paste(unique(mapply(FUN = "paste", data1[[i1]][ ,categ[[i1]]], data1[[i1]][ ,"color"])), collapse = "\n"))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-color[[i1]] <- unique(color[[i1]][order(data1[[i1]][, categ[[i1]]])]) # Modif to have length(color) equal to the different number of categ (length(color) == length(levels(data1[[i1]][, categ[[i1]]])))
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count, ") FROM FUNCTION ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " HAS THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " COLUMN VALUES\nCOLORS HAVE BEEN RESPECTIVELY ASSOCIATED TO EACH CLASS OF categ AS:\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "), "\n", paste(color[[i1]], collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else if(length(color[[i1]])== 1L){
-data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
-color[[i1]] <- rep(color[[i1]], length(levels(data1[[i1]][, categ[[i1]]])))
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", COLOR HAS LENGTH 1 MEANING THAT ALL THE DIFFERENT CLASSES OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), "\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(color[[i1]], collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE\n(1) LENGTH 1\nOR (2) THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " COLUMN VALUES\nOR (3) THE LENGTH OF THE CLASSES IN THIS COLUMN\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LENGTH ", length(data1[[i1]][, categ[[i1]]]), " AND CATEG CLASS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])), "\nPRESENCE OF NA IN THE COLUMN x, y OR categ OF data1 COULD BE THE PROBLEME")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-}
-if((geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline") & ! is.null(categ[[i1]])){ # add here after the color management, to deal with the different lines to plot inside any data[[i1]]
-if(categ[[i1]] == "fake_categ"){
-data1[[i1]][, "fake_categ"] <- factor(paste0("Line_", formatC(1:nrow(data1[[i2]]), width = nchar(nrow(data1[[i2]])), flag = "0")))
-}
-}
-tempo <- fun_check(data = alpha[[i1]], data.name = ifelse(length(alpha)== 1L, "alpha", paste0("alpha NUMBER ", i1)), prop = TRUE, length = 1, fun.name = function.name)
-if(tempo$problem == TRUE){
-stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-# end loop (checking inside list compartment)
-if(length(data1) > 1){
-if(length(unique(unlist(x)[ ! x == "fake_x"])) > 1){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE x ARGUMENT DOES NOT CONTAIN IDENTICAL COLUMN NAMES:\n", paste(unlist(x), collapse = " "), "\nX-AXIS OVERLAYING DIFFERENT VARIABLES?")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(length(data1) > 1){
-if(length(unique(unlist(y)[ ! y == "fake_y"])) > 1){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE y ARGUMENT DOES NOT CONTAIN IDENTICAL COLUMN NAMES:\n", paste(unlist(y), collapse = " "), "\nY-AXIS OVERLAYING DIFFERENT VARIABLES?")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-if(sum(geom %in% "geom_point") > 3){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT HAVE MORE THAN THREE \"geom_point\" ELEMENTS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(length(geom) - sum(geom %in% "geom_point") > 3){
-tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT HAVE MORE THAN THREE LINE ELEMENTS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-# x.lim management before transfo by x.log
-if(x.log != "no" & ! is.null(x.lim)){
-if(any(x.lim <= 0)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE x.log ARGUMENT SET TO ", x.log, ":\n", paste(x.lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(any( ! is.finite(if(x.log == "log10"){log10(x.lim)}else{log2(x.lim)}))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT RETURNS INF/NA WITH THE x.log ARGUMENT SET TO ", x.log, "\nAS SCALE COMPUTATION IS ", ifelse(x.log == "log10", "log10", "log2"), ":\n", paste(if(x.log == "log10"){log10(x.lim)}else{log2(x.lim)}, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(x.log != "no" & x.include.zero == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") x.log ARGUMENT SET TO ", x.log, " AND x.include.zero ARGUMENT SET TO TRUE -> x.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-x.include.zero <- FALSE
-}
-# end x.lim management before transfo by x.log
-# y.lim management before transfo by y.log
-if(y.log != "no" & ! is.null(y.lim)){
-if(any(y.lim <= 0)){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(any( ! is.finite(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT RETURNS INF/NA WITH THE y.log ARGUMENT SET TO ", y.log, "\nAS SCALE COMPUTATION IS ", ifelse(y.log == "log10", "log10", "log2"), ":\n", paste(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(y.log != "no" & y.include.zero == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") y.log ARGUMENT SET TO ", y.log, " AND y.include.zero ARGUMENT SET TO TRUE -> y.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-y.include.zero <- FALSE
-}
-# end y.lim management before transfo by y.log
-# end other checkings
-# reserved word checking
-#already done above
-# end reserved word checking
-# end second round of checking and data preparation
-
-
-# package checking
-fun_pack(req.package = c(
-"gridExtra",
-"ggplot2",
-"lemon",
-"scales"
-), lib.path = lib.path)
-# packages Cairo and grid tested by fun_gg_point_rast()
-# end package checking
-
-
-
-
-# main code
-# axes management
-if(is.null(x.lim)){
-if(any(unlist(mapply(FUN = "[[", data1, x, SIMPLIFY = FALSE)) %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE x COLUMN IN data1 CONTAINS -Inf OR Inf VALUES THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-x.lim <- suppressWarnings(range(unlist(mapply(FUN = "[[", data1, x, SIMPLIFY = FALSE)), na.rm = TRUE, finite = TRUE)) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only. y.lim added here. If NULL, ok if y argument has values
-}else if(x.log != "no"){
-x.lim <- get(x.log)(x.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-}
-if(x.log != "no"){
-if(any( ! is.finite(x.lim))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE x.log ARGUMENT SET TO ", x.log, ":\n", paste(x.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(suppressWarnings(all(x.lim %in% c(Inf, -Inf)))){ # happen when x is only NULL
-if(all(unlist(geom) %in% c("geom_vline", "geom_stick"))){
-tempo.cat <- paste0("ERROR IN ", function.name, " NOT POSSIBLE TO DRAW geom_vline OR geom_stick KIND OF LINES ALONE IF x.lim ARGUMENT IS SET TO NULL, SINCE NO X-AXIS DEFINED (", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT MUST BE NULL FOR THESE KIND OF LINES)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, " x.lim ARGUMENT MADE OF NA, -Inf OR Inf ONLY: ", paste(x.lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-x.lim.order <- order(x.lim) # to deal with inverse axis
-# print(x.lim.order)
-x.lim <- sort(x.lim)
-x.lim[1] <- x.lim[1] - abs(x.lim[2] - x.lim[1]) * ifelse(diff(x.lim.order) > 0, x.right.extra.margin, x.left.extra.margin) # diff(x.lim.order) > 0 means not inversed axis
-x.lim[2] <- x.lim[2] + abs(x.lim[2] - x.lim[1]) * ifelse(diff(x.lim.order) > 0, x.left.extra.margin, x.right.extra.margin) # diff(x.lim.order) > 0 means not inversed axis
-if(x.include.zero == TRUE){ # no need to check x.log != "no" because done before
-x.lim <- range(c(x.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-}
-x.lim <- x.lim[x.lim.order]
-if(any(is.na(x.lim))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(is.null(y.lim)){
-if(any(unlist(mapply(FUN = "[[", data1, y, SIMPLIFY = FALSE)) %in% c(Inf, -Inf))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE y COLUMN IN data1 CONTAINS -Inf OR Inf VALUES THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-y.lim <- suppressWarnings(range(unlist(mapply(FUN = "[[", data1, y, SIMPLIFY = FALSE)), na.rm = TRUE, finite = TRUE)) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only. y.lim added here. If NULL, ok if y argument has values
-}else if(y.log != "no"){
-y.lim <- get(y.log)(y.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
-}
-if(y.log != "no"){
-if(any( ! is.finite(y.lim))){
-tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-if(suppressWarnings(all(y.lim %in% c(Inf, -Inf)))){ # happen when y is only NULL
-if(all(unlist(geom) == "geom_vline")){
-tempo.cat <- paste0("ERROR IN ", function.name, " NOT POSSIBLE TO DRAW geom_vline KIND OF LINES ALONE IF y.lim ARGUMENT IS SET TO NULL, SINCE NO Y-AXIS DEFINED (", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT MUST BE NULL FOR THESE KIND OF LINES)")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, " y.lim ARGUMENT MADE OF NA, -Inf OR Inf ONLY: ", paste(y.lim, collapse = " "))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
-y.lim.order <- order(y.lim) # to deal with inverse axis
-y.lim <- sort(y.lim)
-y.lim[1] <- y.lim[1] - abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.bottom.extra.margin, y.top.extra.margin) # diff(y.lim.order) > 0 means not inversed axis
-y.lim[2] <- y.lim[2] + abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.top.extra.margin, y.bottom.extra.margin) # diff(y.lim.order) > 0 means not inversed axis
-if(y.include.zero == TRUE){ # no need to check y.log != "no" because done before
-y.lim <- range(c(y.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
-}
-y.lim <- y.lim[y.lim.order]
-if(any(is.na(y.lim))){
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+fun_open <- function(
+ pdf = TRUE,
+ pdf.path = "working.dir",
+ pdf.name = "graph",
+ width = 7,
+ height = 7,
+ paper = "special",
+ pdf.overwrite = FALSE,
+ rescale = "fixed",
+ remove.read.only = TRUE,
+ return.output = FALSE
+){
+ # AIM
+ # open a pdf or screen (GUI) graphic window and return initial graphic parameters
+ # this order can be used:
+ # fun_width()
+ # fun_open()
+ # fun_prior_plot() # not for ggplot2
+ # plot() or any other plotting
+ # fun_post_plot() if fun_prior_plot() has been used # not for ggplot2
+ # fun_close()
+ # WARNINGS
+ # On Linux, use pdf = TRUE, if (GUI) graphic window is not always available, meaning that X is not installed (clusters for instance). Use X11() in R to test if available
+ # ARGUMENTS:
+ # pdf: logical. Use pdf display? If FALSE, a GUI is opened
+ # pdf.path: where the pdf is saved (do not terminate by / or \\). Write "working.dir" if working directory is required (default). Ignored if pdf == FALSE
+ # pdf.name: name of the pdf file containing the graphs (the .pdf extension is added by the function, if not detected in the name end). Ignored if pdf == FALSE
+ # width: width of the window (in inches)
+ # height: height of the window (in inches)
+ # paper: paper argument of the pdf function (paper format). Only used for pdf(). Either "a4", "letter", "legal", "us", "executive", "a4r", "USr" or "special". If "special", means that the paper dimension will be width and height. With another paper format, if width or height is over the size of the paper, width or height will be modified such that the plot is adjusted to the paper dimension (see $dim in the returned list below to see the modified dimensions). Ignored if pdf == FALSE
+ # pdf.overwrite: logical. Existing pdf can be overwritten? . Ignored if pdf == FALSE
+ # rescale: kind of GUI. Either "R", "fit", or "fixed". Ignored on Mac and Linux OS. See ?windows for details
+ # remove.read.only: logical. remove the read only (R.O.) graphical parameters? If TRUE, the graphical parameters are returned without the R.O. parameters. The returned $ini.par list can be used to set the par() of a new graphical device. If FALSE, graphical parameters are returned with the R.O. parameters, which provides information like text dimension (see ?par() ). The returned $ini.par list can be used to set the par() of a new graphical device, but generate a warning message. Ignored if return.output == FALSE.
+ # return.output: logical. Return output ? If TRUE the output list is displayed
+ # RETURN
+ # a list containing:
+ # $pdf.loc: path of the pdf created
+ # $ini.par: initial par() parameters
+ # $zone.ini: initial window spliting
+ # $dim: dimension of the graphical device (in inches)
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_open(pdf = FALSE, pdf.path = "C:/Users/Gael/Desktop", pdf.name = "graph", width = 7, height = 7, paper = "special", pdf.overwrite = FALSE, return.output = TRUE)
+ # DEBUGGING
+ # pdf = TRUE ; pdf.path = "C:/Users/Gael/Desktop" ; pdf.name = "graphs" ; width = 7 ; height = 7 ; paper = "special" ; pdf.overwrite = FALSE ; rescale = "fixed" ; remove.read.only = TRUE ; return.output = TRUE # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = pdf, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = pdf.path, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = pdf.name, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = height, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = paper, options = c("a4", "letter", "legal", "us", "executive", "a4r", "USr", "special", "A4", "LETTER", "LEGAL", "US"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data =pdf.overwrite, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = rescale, options = c("R", "fit", "fixed"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = remove.read.only, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.output, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if(pdf.path == "working.dir"){
+ pdf.path <- getwd()
+ }else{
+ if(grepl(x = pdf.path, pattern = ".+/$")){
+ pdf.path <- sub(x = pdf.path, pattern = "/$", replacement = "") # remove the last /
+ }else if(grepl(x = pdf.path, pattern = ".+[\\]$")){ # or ".+\\\\$" # cannot be ".+\$" because \$ does not exist contrary to \n
+ pdf.path <- sub(x = pdf.path, pattern = "[\\]$", replacement = "") # remove the last /
+ }
+ if(dir.exists(pdf.path) == FALSE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\npdf.path ARGUMENT DOES NOT CORRESPOND TO EXISTING DIRECTORY\n", pdf.path)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # par.ini recovery
+ # cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
+ if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
+ open.fail <- NULL
+ grDevices::windows()
+ ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ }else if(Sys.info()["sysname"] == "Linux"){
+ if(pdf == TRUE){# cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
+ if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
+ tempo.random.seed <- .Random.seed
+ on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
+ }else{
+ on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
+ }
+ set.seed(NULL)
+ tempo.code <- sample(x = 1:1e7, size = 1)
+ while(file.exists(paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf")) == TRUE){
+ tempo.code <- tempo.code + 1
+ }
+ grDevices::pdf(width = width, height = height, file=paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf"), paper = paper)
+ ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the pdf window
+ file.remove(paste0(pdf.path, "/recover_ini_par", tempo.code, ".pdf")) # remove the pdf file
+ }else{
+ # test if X11 can be opened
+ if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHIS FUNCTION CANNOT BE USED ON LINUX IF A Rplots.pdf FILE ALREADY EXISTS HERE\n", getwd())
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ open.fail <- suppressWarnings(try(grDevices::X11(), silent = TRUE))[] # try to open a X11 window. If open.fail == NULL, no problem, meaning that the X11 window is opened. If open.fail != NULL, a pdf can be opened here paste0(getwd(), "/Rplots.pdf")
+ if(is.null(open.fail)){
+ ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ }else if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
+ file.remove(paste0(getwd(), "/Rplots.pdf")) # remove the pdf file
+ tempo.cat <- ("ERROR IN fun_open()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, EITHER SET THE X GRAPHIC INTERFACE OF THE SYSTEM OR SET THE pdf ARGUMENT OF THE fun_open() FUNCTION TO TRUE AND RERUN")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ }
+ }else{
+ open.fail <- NULL
+ grDevices::quartz()
+ ini.par <- par(no.readonly = remove.read.only) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ }
+ # end par.ini recovery
+ zone.ini <- matrix(1, ncol=1) # to recover the initial parameters for next figure region when device region split into several figure regions
+ if(pdf == TRUE){
+ if(grepl(x = pdf.name, pattern = "\\.pdf$")){
+ pdf.name <- sub(x = pdf.name, pattern = "\\.pdf$", replacement = "") # remove the last .pdf
+ }
+ pdf.loc <- paste0(pdf.path, "/", pdf.name, ".pdf")
+ if(file.exists(pdf.loc) == TRUE & pdf.overwrite == FALSE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\npdf.loc FILE ALREADY EXISTS AND CANNOT BE OVERWRITTEN DUE TO pdf.overwrite ARGUMENT SET TO TRUE\n", pdf.loc)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ grDevices::pdf(width = width, height = height, file=pdf.loc, paper = paper)
+ }
+ }else if(pdf == FALSE){
+ pdf.loc <- NULL
+ if(Sys.info()["sysname"] == "Windows"){ # .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
+ grDevices::windows(width = width, height = height, rescale = rescale)
+ }else if(Sys.info()["sysname"] == "Linux"){
+ if( ! is.null(open.fail)){
+ tempo.cat <- "ERROR IN fun_open()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, EITHER SET THE X GRAPHIC INTERFACE OF THE SYSTEM OR SET THE pdf ARGUMENT OF THE fun_open() FUNCTION TO TRUE AND RERUN"
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ grDevices::X11(width = width, height = height)
+ }
+ }else{
+ grDevices::quartz(width = width, height = height)
+ }
+ }
+ if(return.output == TRUE){
+ output <- list(pdf.loc = pdf.loc, ini.par = ini.par, zone.ini = zone.ini, dim = dev.size())
+ return(output)
+ }
}
-# end axes management
+######## fun_prior_plot() #### set graph param before plotting (erase axes for instance)
-# create a fake categ if NULL to deal with legend display
-if(is.null(categ)){
-categ <- vector("list", length(data1))
-categ[] <- "fake_categ"
-for(i2 in 1:length(data1)){
-data1[[i2]] <- cbind(data1[[i2]], fake_categ = "", stringsAsFactors = TRUE)
-if(geom[[i2]] == "geom_hline" | geom[[i2]] == "geom_vline"){
-data1[[i2]][, "fake_categ"] <- factor(paste0("Line_", 1:nrow(data1[[i2]])))
-}
-}
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL categ ARGUMENT -> FAKE \"fake_categ\" COLUMN ADDED TO EACH DATA FRAME OF data1, AND FILLED WITH \"\"")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-# categ is not NULL anymore
-if(is.null(categ.class.order)){
-categ.class.order <- vector("list", length = length(data1))
-tempo.categ.class.order <- NULL
-for(i2 in 1:length(categ.class.order)){
-categ.class.order[[i2]] <- levels(data1[[i2]][, categ[[i2]]])
-names(categ.class.order)[i2] <- categ[[i2]]
-tempo.categ.class.order <- c(tempo.categ.class.order, ifelse(i2 != 1, "\n", ""), categ.class.order[[i2]])
-}
-if(any(unlist(legend.disp))){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") THE categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR CLASS ORDERING:\n", paste(tempo.categ.class.order, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end create a fake categ if NULL to deal with legend display
-# categ.class.order is not NULL anymore
-
-
-# vector of color with length as in levels(categ) of data1
-if(is.null(color)){
-color <- vector("list", length(data1))
-length.categ.list <- lapply(lapply(mapply(FUN = "[[", data1, categ, SIMPLIFY = FALSE), FUN = unique), FUN = function(x){length(x[ ! is.na(x)])})
-length.categ.list[sapply(categ, FUN = "==", "fake_categ")] <- 1 # when is.null(color), a single color for all the dots or lines of data[[i1]] that contain "fake_categ" category
-total.categ.length <- sum(unlist(length.categ.list), na.rm = TRUE)
-tempo.color <- fun_gg_palette(total.categ.length)
-tempo.count <- 0
-for(i2 in 1:length(data1)){
-color[[i2]] <- tempo.color[(1:length.categ.list[[i2]]) + tempo.count]
-tempo.count <- tempo.count + length.categ.list[[i2]]
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") NULL color ARGUMENT -> COLORS RESPECTIVELY ATTRIBUTED TO EACH CLASS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i2, " OF categ ARGUMENT")), " (", categ[[i2]], ") IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i2, " OF data1 ARGUMENT")), ":\n", paste(color[[i2]], collapse = " "), "\n", paste(if(all(levels(data1[[i2]][, categ[[i2]]]) == "")){'\"\"'}else{levels(data1[[i2]][, categ[[i2]]])}, collapse = " "))
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
+fun_prior_plot <- function(
+ param.reinitial = FALSE,
+ xlog.scale = FALSE,
+ ylog.scale = FALSE,
+ remove.label = TRUE,
+ remove.x.axis = TRUE,
+ remove.y.axis = TRUE,
+ std.x.range = TRUE,
+ std.y.range = TRUE,
+ down.space = 1,
+ left.space = 1,
+ up.space = 1,
+ right.space = 1,
+ orient = 1,
+ dist.legend = 3.5,
+ tick.length = 0.5,
+ box.type = "n",
+ amplif.label = 1,
+ amplif.axis = 1,
+ display.extend = FALSE,
+ return.par = FALSE
+){
+ # AIM
+ # very convenient to erase the axes for post plot axis redrawing using fun_post_plot()
+ # reinitialize and set the graphic parameters before plotting
+ # CANNOT be used if no graphic device already opened
+ # ARGUMENTS
+ # param.reinitial: reinitialize graphic parameters before applying the new ones, as defined by the other arguments? Either TRUE or FALSE
+ # xlog.scale: Log scale for the x-axis? Either TRUE or FALSE. If TRUE, erases the x-axis, except legend, for further drawing by fun_post_plot()(xlog argument of par())
+ # ylog.scale: Log scale for the y-axis? Either TRUE or FALSE. If TRUE, erases the y-axis, except legend, for further drawing by fun_post_plot()(ylog argument of par())
+ # remove.label: remove labels (axis legend) of the two axes? Either TRUE or FALSE (ann argument of par())
+ # remove.x.axis: remove x-axis except legend? Either TRUE or FALSE (control the xaxt argument of par()). Automately set to TRUE if xlog.scale == TRUE
+ # remove.y.axis: remove y-axis except legend? Either TRUE or FALSE (control the yaxt argument of par()). Automately set to TRUE if ylog.scale == TRUE
+ # std.x.range: standard range on the x-axis? TRUE (no range extend) or FALSE (4% range extend). Controls xaxs argument of par() (TRUE is xaxs = "i", FALSE is xaxs = "r")
+ # std.y.range: standard range on the y-axis? TRUE (no range extend) or FALSE (4% range extend). Controls yaxs argument of par() (TRUE is yaxs = "i", FALSE is yaxs = "r")
+ # down.space: lower vertical margin (in inches, mai argument of par())
+ # left.space: left horizontal margin (in inches, mai argument of par())
+ # up.space: upper vertical margin between plot region and grapical window (in inches, mai argument of par())
+ # right.space: right horizontal margin (in inches, mai argument of par())
+ # orient: scale number orientation (las argument of par()). 0, always parallel to the axis; 1, always horizontal; 2, always perpendicular to the axis; 3, always vertical
+ # dist.legend: numeric value that moves axis legends away in inches (first number of mgp argument of par() but in inches thus / 0.2)
+ # tick.length: length of the ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc. 0 means no tick
+ # box.type: bty argument of par(). Either "o", "l", "7", "c", "u", "]", the resulting box resembles the corresponding upper case letter. A value of "n" suppresses the box
+ # amplif.label: increase or decrease the size of the text in legends
+ # amplif.axis: increase or decrease the size of the scale numbers in axis
+ # display.extend: extend display beyond plotting region? Either TRUE or FALSE (xpd argument of par() without NA)
+ # return.par: return graphic parameter modification?
+ # RETURN
+ # return graphic parameter modification
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_prior_plot(param.reinitial = FALSE, xlog.scale = FALSE, ylog.scale = FALSE, remove.label = TRUE, remove.x.axis = TRUE, remove.y.axis = TRUE, std.x.range = TRUE, std.y.range = TRUE, down.space = 1, left.space = 1, up.space = 1, right.space = 1, orient = 1, dist.legend = 4.5, tick.length = 0.5, box.type = "n", amplif.label = 1, amplif.axis = 1, display.extend = FALSE, return.par = FALSE)
+ # DEBUGGING
+ # param.reinitial = FALSE ; xlog.scale = FALSE ; ylog.scale = FALSE ; remove.label = TRUE ; remove.x.axis = TRUE ; remove.y.axis = TRUE ; std.x.range = TRUE ; std.y.range = TRUE ; down.space = 1 ; left.space = 1 ; up.space = 1 ; right.space = 1 ; orient = 1 ; dist.legend = 4.5 ; tick.length = 0.5 ; box.type = "n" ; amplif.label = 1 ; amplif.axis = 1 ; display.extend = FALSE ; return.par = FALSE # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = param.reinitial, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = xlog.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = ylog.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = remove.label, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = remove.x.axis, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = remove.y.axis, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = std.x.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = std.y.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = down.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = up.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = orient, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.type, options = c("o", "l", "7", "c", "u", "]", "n"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = amplif.label, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = amplif.axis, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = display.extend, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.par, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if(is.null(dev.list())){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THIS FUNCTION CANNOT BE USED IF NO GRAPHIC DEVICE ALREADY OPENED (dev.list() IS CURRENTLY NULL)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # par.ini recovery
+ # cannot use pdf(file = NULL), because some small differences between pdf() and other devices. For instance, differences with windows() for par()$fin, par()$pin and par()$plt
+ if(param.reinitial == TRUE){
+ if( ! all(names(dev.cur()) == "null device")){
+ active.wind.nb <- dev.cur()
+ }else{
+ active.wind.nb <- 0
+ }
+ if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
+ grDevices::windows()
+ ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ }else if(Sys.info()["sysname"] == "Linux"){
+ if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THIS FUNCTION CANNOT BE USED ON LINUX WITH param.reinitial SET TO TRUE IF A Rplots.pdf FILE ALREADY EXISTS HERE: ", getwd())
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ open.fail <- suppressWarnings(try(grDevices::X11(), silent = TRUE))[] # try to open a X11 window. If open.fail == NULL, no problem, meaning that the X11 window is opened. If open.fail != NULL, a pdf can be opened here paste0(getwd(), "/Rplots.pdf")
+ if(is.null(open.fail)){
+ ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ }else if(file.exists(paste0(getwd(), "/Rplots.pdf"))){
+ ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ invisible(dev.off()) # close the new window
+ file.remove(paste0(getwd(), "/Rplots.pdf")) # remove the pdf file
+ }else{
+ tempo.cat <- ("ERROR IN fun_prior_plot()\nTHIS FUNCTION CANNOT OPEN GUI ON LINUX OR NON MACOS UNIX SYSTEM\nTO OVERCOME THIS, PLEASE USE A PDF GRAPHIC INTERFACE AND RERUN")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ }else{ # macOS
+ grDevices::quartz()
+ ini.par <- par(no.readonly = FALSE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened)
+ invisible(dev.off()) # close the new window
+ }
+ if( ! all(names(dev.cur()) == "null device")){
+ invisible(dev.set(active.wind.nb)) # go back to the active window if exists
+ par(ini.par) # apply the initial par to current window
+ }
+ }
+ # end par.ini recovery
+ if(remove.x.axis == TRUE){
+ par(xaxt = "n") # suppress the y-axis label
+ }else{
+ par(xaxt = "s")
+ }
+ if(remove.y.axis == TRUE){
+ par(yaxt = "n") # suppress the y-axis label
+ }else{
+ par(yaxt = "s")
+ }
+ if(std.x.range == TRUE){
+ par(xaxs = "i")
+ }else{
+ par(xaxs = "r")
+ }
+ if(std.y.range == TRUE){
+ par(yaxs = "i")
+ }else{
+ par(yaxs = "r")
+ }
+ par(mai = c(down.space, left.space, up.space, right.space), ann = ! remove.label, las = orient, mgp = c(dist.legend/0.2, 1, 0), xpd = display.extend, bty= box.type, cex.lab = amplif.label, cex.axis = amplif.axis)
+ par(tcl = -par()$mgp[2] * tick.length) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
+ if(xlog.scale == TRUE){
+ par(xaxt = "n", xlog = TRUE) # suppress the x-axis label
+ }else{
+ par(xlog = FALSE)
+ }
+ if(ylog.scale == TRUE){
+ par(yaxt = "n", ylog = TRUE) # suppress the y-axis label
+ }else{
+ par(ylog = FALSE)
+ }
+ if(return.par == TRUE){
+ tempo.par <- par()
+ return(tempo.par)
+ }
}
-# end vector of color with length as in levels(categ) of data1
-# color is not NULL anymore
+
+
+######## fun_scale() #### select nice label numbers when setting number of ticks on an axis
-# last check
-for(i1 in 1:length(data1)){
-if(categ[[i1]] != "fake_categ" & length(color[[i1]]) != length(unique(data1[[i1]][, categ[[i1]]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, " LAST CHECK: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE THE LENGTH OF LEVELS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LEVELS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])), "\nREMINDER: A SINGLE COLOR PER CLASS OF CATEG AND A SINGLE CLASS OF CATEG PER COLOR MUST BE RESPECTED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else if(categ[[i1]] == "fake_categ" & length(color[[i1]]) != 1){
-tempo.cat <- paste0("ERROR IN ", function.name, " LAST CHECK: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE LENGTH 1 WHEN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IS NULL\nHERE IT IS COLOR LENGTH ", length(color[[i1]]))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+fun_scale <- function(n, lim, kind = "approx", lib.path = NULL){
+ # AIM
+ # attempt to select nice scale numbers when setting n ticks on a lim axis range
+ # ARGUMENTS
+ # n: desired number of main ticks on the axis (integer above 0)
+ # lim: vector of 2 numbers indicating the limit range of the axis. Order of the 2 values matters (for inverted axis). Can be log transformed values
+ # kind: either "approx" (approximative), "strict" (strict) or "strict.cl" (strict clean). If "approx", use the scales::trans_breaks() function to provide an easy to read scale of approximately n ticks spanning the range of the lim argument. If "strict", cut the range of the lim argument into n + 1 equidistant part and return the n numbers at each boundary. This often generates numbers uneasy to read. If "strict.cl", provide an easy to read scale of exactly n ticks, but sometimes not completely spanning the range of the lim argument
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # RETURN
+ # a vector of numbers
+ # REQUIRED PACKAGES
+ # if kind = "approx":
+ # ggplot2
+ # scales
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_round()
+ # EXAMPLES
+ # approximate number of main ticks
+ # ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "approx") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
+ # strict number of main ticks
+ # ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "strict") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
+ # strict "clean" number of main ticks
+ # ymin = 2 ; ymax = 3.101 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "strict.cl") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
+ # approximate number of main ticks, scale inversion
+ # ymin = 3.101 ; ymax = 2 ; n = 5 ; scale <- fun_scale(n = n, lim = c(ymin, ymax), kind = "approx") ; scale ; par(yaxt = "n", yaxs = "i", las = 1) ; plot(ymin:ymax, ymin:ymax, xlim = range(scale, ymin, ymax)[order(c(ymin, ymax))], ylim = range(scale, ymin, ymax)[order(c(ymin, ymax))], xlab = "DEFAULT SCALE", ylab = "NEW SCALE") ; par(yaxt = "s") ; axis(side = 2, at = scale)
+ # DEBUGGING
+ # n = 9 ; lim = c(2, 3.101) ; kind = "approx" ; lib.path = NULL # for function debugging
+ # n = 10 ; lim = c(1e-4, 1e6) ; kind = "approx" ; lib.path = NULL # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # end initial argument checking
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_round", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_round() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & isTRUE(all.equal(n, 0))){ # isTRUE(all.equal(n, 0)) equivalent to n == 0 but deals with floats (approx ok)
+ tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A NON NULL AND POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE) #
+ }
+ tempo <- fun_check(data = lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & all(diff(lim) == 0L)){ # isTRUE(all.equal(diff(lim), rep(0, length(diff(lim))))) not used because we strictly need zero as a result
+ tempo.cat <- paste0("ERROR IN ", function.name, ": lim ARGUMENT HAS A NULL RANGE (2 IDENTICAL VALUES)")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & any(lim %in% c(Inf, -Inf))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = kind, options = c("approx", "strict", "strict.cl"), length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ lim.rank <- rank(lim) # to deal with inverted axis
+ lim <- sort(lim)
+ if(kind == "approx"){
+ # package checking
+ fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
+ fun_pack(req.package = c("scales"), lib.path = lib.path)
+ # end package checking
+ output <- ggplot2::ggplot_build(ggplot2::ggplot() + ggplot2::scale_y_continuous(
+ breaks = scales::trans_breaks(
+ trans = "identity",
+ inv = "identity",
+ n = n
+ ),
+ limits = lim
+ ))$layout$panel_params[[1]]$y$breaks # pretty() alone is not appropriate: tempo.pret <- pretty(seq(lim[1] ,lim[2], length.out = n)) ; tempo.pret[tempo.pret > = lim[1] & tempo.pret < = lim[2]]. # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks
+ if( ! is.null(attributes(output))){ # layout$panel_params[[1]]$y$breaks can be characters (labels of the axis). In that case, it has attributes that corresponds to positions
+ output <- unlist(attributes(output))
+ }
+ output <- output[ ! is.na(output)]
+ }else if(kind == "strict"){
+ output <- fun_round(seq(lim[1] ,lim[2], length.out = n), 2)
+ }else if(kind == "strict.cl"){
+ tempo.range <- diff(sort(lim))
+ tempo.max <- max(lim)
+ tempo.min <- min(lim)
+ mid <- tempo.min + (tempo.range/2) # middle of axis
+ tempo.inter <- tempo.range / (n + 1) # current interval between two ticks, between 0 and Inf
+ if(tempo.inter == 0L){ # isTRUE(all.equal(tempo.inter, rep(0, length(tempo.inter)))) not used because we strictly need zero as a result
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE INTERVAL BETWEEN TWO TICKS OF THE SCALE IS NULL. MODIFY THE lim OR n ARGUMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ log10.abs.lim <- 200
+ log10.range <- (-log10.abs.lim):log10.abs.lim
+ log10.vec <- 10^log10.range
+ round.vec <- c(5, 4, 3, 2.5, 2, 1.25, 1)
+ dec.table <- outer(log10.vec, round.vec) # table containing the scale units (row: power of ten from -201 to +199, column: the 5, 2.5, 2, 1.25, 1 notches
+
+
+
+ # recover the number of leading zeros in tempo.inter
+ ini.scipen <- options()$scipen
+ options(scipen = -1000) # force scientific format
+ if(any(grepl(pattern = "\\+", x = tempo.inter))){ # tempo.inter > 1
+ power10.exp <- as.integer(substring(text = tempo.inter, first = (regexpr(pattern = "\\+", text = tempo.inter) + 1))) # recover the power of 10. Example recover 08 from 1e+08
+ mantisse <- as.numeric(substr(x = tempo.inter, start = 1, stop = (regexpr(pattern = "\\+", text = tempo.inter) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
+ }else if(any(grepl(pattern = "\\-", x = tempo.inter))){ # tempo.inter < 1
+ power10.exp <- as.integer(substring(text = tempo.inter, first = (regexpr(pattern = "\\-", text = tempo.inter)))) # recover the power of 10. Example recover 08 from 1e+08
+ mantisse <- as.numeric(substr(x = tempo.inter, start = 1, stop = (regexpr(pattern = "\\-", text = tempo.inter) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ tempo.scale <- dec.table[log10.range == power10.exp, ]
+ # new interval
+ inter.select <- NULL
+ for(i1 in 1:length(tempo.scale)){
+ tempo.first.tick <- trunc((tempo.min + tempo.scale[i1]) / tempo.scale[i1]) * (tempo.scale[i1]) # this would be use to have a number not multiple of tempo.scale[i1]: ceiling(tempo.min) + tempo.scale[i1] * 10^power10.exp
+ tempo.last.tick <- tempo.first.tick + tempo.scale[i1] * (n - 1)
+ if((tempo.first.tick >= tempo.min) & (tempo.last.tick <= tempo.max)){
+ inter.select <- tempo.scale[i1]
+ break()
+ }
+ }
+ if(is.null(inter.select)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ options(scipen = ini.scipen) # restore the initial scientific penalty
+ # end new interval
+ # centering the new scale
+ tempo.mid <- trunc((mid + (-1:1) * inter.select) / inter.select) * inter.select # tempo middle tick closest to the middle axis
+ mid.tick <- tempo.mid[which.min(abs(tempo.mid - mid))]
+ if(isTRUE(all.equal(n, rep(1, length(n))))){ # isTRUE(all.equal(n, rep(1, length(n)))) is similar to n == 1L but deals with float
+ output <- mid.tick
+ }else if(isTRUE(all.equal(n, rep(2, length(n))))){ # isTRUE(all.equal(n, rep(0, length(n)))) is similar to n == 2L but deals with float
+ output <- mid.tick
+ tempo.min.dist <- mid.tick - inter.select - tempo.min
+ tempo.max.dist <- tempo.max - mid.tick + inter.select
+ if(tempo.min.dist <= tempo.max.dist){ # distance between lowest tick and bottom axis <= distance between highest tick and top axis. If yes, extra tick but at the top, otherwise at the bottom
+ output <- c(mid.tick, mid.tick + inter.select)
+ }else{
+ output <- c(mid.tick - inter.select, mid.tick)
+ }
+ }else if((n / 2 - trunc(n / 2)) > 0.1){ # > 0.1 to avoid floating point. Because result can only be 0 or 0.5. Thus, > 0.1 means odd number
+ output <- c(mid.tick - (trunc(n / 2):1) * inter.select, mid.tick, mid.tick + (1:trunc(n / 2)) * inter.select)
+ }else if((n / 2 - trunc(n / 2)) < 0.1){ # < 0.1 to avoid floating point. Because result can only be 0 or 0.5. Thus, < 0.1 means even number
+ tempo.min.dist <- mid.tick - trunc(n / 2) * inter.select - tempo.min
+ tempo.max.dist <- tempo.max - mid.tick + trunc(n / 2) * inter.select
+ if(tempo.min.dist <= tempo.max.dist){ # distance between lowest tick and bottom axis <= distance between highest tick and top axis. If yes, extra tick but at the bottom, otherwise at the top
+ output <- c(mid.tick - ((trunc(n / 2) - 1):1) * inter.select, mid.tick, mid.tick + (1:trunc(n / 2)) * inter.select)
+ }else{
+ output <- c(mid.tick - (trunc(n / 2):1) * inter.select, mid.tick, mid.tick + (1:(trunc(n / 2) - 1)) * inter.select)
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end centering the new scale
+ # last check
+ if(min(output) < tempo.min){
+ output <- c(output[-1], max(output) + inter.select) # remove the lowest tick and add a tick at the top
+ }else if( max(output) > tempo.max){
+ output <- c(min(output) - inter.select, output[-length(output)])
+ }
+ if(min(output) < tempo.min | max(output) > tempo.max){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(is.na(output))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5 (NA GENERATION)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end last check
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 6")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(diff(lim.rank) < 0){
+ output <- rev(output)
+ }
+ return(output)
}
+
+
+######## fun_inter_ticks() #### define coordinates of secondary ticks
+
+
+fun_inter_ticks <- function(
+ lim,
+ log = "log10",
+ breaks = NULL,
+ n = NULL,
+ warn.print = TRUE
+){
+ # AIM
+ # define coordinates and values of secondary ticks
+ # ARGUMENTS
+ # lim: vector of 2 numbers indicating the limit range of the axis. Order of the 2 values matters (for inverted axis). If log argument is "log2" or "log10", values in lim must be already log transformed. Thus, negative or zero values are allowed
+ # log: either "log2" (values in the lim argument are log2 transformed) or "log10" (values in the lim argument are log10 transformed), or "no"
+ # breaks: mandatory vector of numbers indicating the main ticks values/positions when log argument is "no". Ignored when log argument is "log2" or "log10"
+ # n: number of secondary ticks between each main tick when log argument is "no". Ignored when log argument is "log2" or "log10"
+ # warn.print: logical. Print potential warning messages at the end of the execution? If FALSE, warning messages are never printed, but can still be recovered in the returned list
+ # RETURN
+ # a list containing
+ # $log: value of the log argument used
+ # $coordinates: the coordinates of the secondary ticks on the axis, between the lim values
+ # $values: the corresponding values associated to each coordinate (with log scale, 2^$values or 10^$values is equivalent to the labels of the axis)
+ # $warn: the potential warning messages. Use cat() for proper display. NULL if no warning
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # no log scale
+ # fun_inter_ticks(lim = c(-4,4), log = "no", breaks = c(-2, 0, 2), n = 3)
+ # fun_inter_ticks(lim = c(10, 0), log = "no", breaks = c(10, 8, 6, 4, 2, 0), n = 4)
+ # log2
+ # fun_inter_ticks(lim = c(-4,4), log = "log2")
+ # log10
+ # fun_inter_ticks(lim = c(-2,3), log = "log10")
+ # DEBUGGING
+ # lim = c(2, 3.101) ; log = "no" ; breaks = NULL ; n = NULL ; warn.print = TRUE # for function debugging
+ # lim = c(0, 26.5) ; log = "no" ; breaks = c(0, 10, 20) ; n = 3 # for function debugging
+ # lim = c(10, 0); log = "no"; breaks = c(10, 8, 6, 4, 2, 0); n = 4 # for function debugging
+ # lim = c(-10, -20); log = "no"; breaks = c(-20, -15, -10); n = 4 # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check"
+ )
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED ", i1, "() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end required function checking
+ # argument primary checking
+ # arg with no default values
+ mandat.args <- c(
+ "lim"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # using fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(breaks)){
+ tempo <- fun_check(data = breaks, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(n)){
+ tempo <- fun_check(data = n, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = warn.print, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA
+ if(any(is.na(lim)) | any(is.na(log)) | any(is.na(breaks)) | any(is.na(n)) | any(is.na(warn.print))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nNO ARGUMENT CAN HAVE NA VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA
+ # management of NULL
+ if(is.null(lim) | is.null(log)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHESE ARGUMENTS\nlim\nlog\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL
+ if(all(diff(lim) == 0L)){ # isTRUE(all.equal(diff(lim), rep(0, length(diff(lim))))) not used because we strictly need zero as a result
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nlim ARGUMENT HAS A NULL RANGE (2 IDENTICAL VALUES): ", paste(lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(any(lim %in% c(Inf, -Inf))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nlim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(log == "no" & is.null(breaks)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT CANNOT BE NULL IF log ARGUMENT IS \"no\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(breaks)){
+ if(length(breaks) < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT MUST HAVE 2 VALUES AT LEAST (OTHERWISE, INTER TICK POSITIONS CANNOT BE COMPUTED): ", paste(breaks, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! isTRUE(all.equal(diff(sort(breaks)), rep(diff(sort(breaks))[1], length(diff(sort(breaks))))))){ # isTRUE(all.equal(n, 0)) equivalent to n == 0 but deals with floats (approx ok)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nbreaks ARGUMENT MUST HAVE EQUIDISTANT VALUES (OTHERWISE, EQUAL NUMBER OF INTER TICK BETWEEN MAIN TICKS CANNOT BE COMPUTED): ", paste(breaks, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if( ! is.null(n)){
+ if(n <= 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nn ARGUMENT MUST BE A POSITIVE AND NON NULL INTEGER: ", paste(n, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end second round of checking and data preparation
+ # main code
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ lim.rank <- rank(lim) # to deal with inverse axis
+ if(log != "no"){
+ ini.scipen <- options()$scipen
+ options(scipen = -1000) # force scientific format
+ power10.exp <- as.integer(substring(text = 10^lim, first = (regexpr(pattern = "\\+|\\-", text = 10^lim)))) # recover the power of 10, i.e., integer part of lim. Example recover 08 from 1e+08. Works for log2
+ # mantisse <- as.numeric(substr(x = 10^lim, start = 1, stop = (regexpr(pattern = "\\+|\\-", text = 10^lim) - 2))) # recover the mantisse. Example recover 1.22 from 1.22e+08
+ options(scipen = ini.scipen) # restore the initial scientific penalty
+ tick.pos <- unique(as.vector(outer(2:10, ifelse(log == "log2", 2, 10)^((power10.exp[1] - ifelse(diff(lim.rank) > 0, 1, -1)):(power10.exp[2] + ifelse(diff(lim.rank) > 0, 1, -1)))))) # use log10(2:10) even if log2: it is to get log values between 0 and 1
+ tick.pos <- sort(tick.pos, decreasing = ifelse(diff(lim.rank) > 0, FALSE, TRUE))
+ if(log == "log2"){
+ tick.values <- tick.pos[tick.pos >= min(2^lim) & tick.pos <= max(2^lim)]
+ tick.pos <- log2(tick.values)
+ }else if(log == "log10"){
+ tick.values <- tick.pos[tick.pos >= min(10^lim) & tick.pos <= max(10^lim)]
+ tick.pos <- log10(tick.values)
+ }
+ }else{
+ # if(length(breaks) > 1){ # not required because already checked above
+ breaks.rank <- rank(c(breaks[1], breaks[length(breaks)]))
+ if(diff(breaks.rank) != diff(lim.rank)){
+ breaks <- sort(breaks, decreasing = ifelse(diff(lim.rank) < 0, TRUE, FALSE))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") VALUES IN breaks ARGUMENT NOT IN THE SAME ORDER AS IN lim ARGUMENT -> VALUES REORDERED AS IN lim: ", paste(breaks, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ breaks.rank <- rank(c(breaks[1], breaks[length(breaks)]))
+ }
+ # }
+ main.tick.dist <- mean(diff(breaks), na.rm = TRUE)
+ tick.dist <- main.tick.dist / (n + 1)
+ tempo.extra.margin <- max(abs(diff(breaks)), na.rm = TRUE)
+ tick.pos <- seq(
+ if(diff(breaks.rank) > 0){breaks[1] - tempo.extra.margin}else{breaks[1] + tempo.extra.margin},
+ if(diff(breaks.rank) > 0){breaks[length(breaks)] + tempo.extra.margin}else{breaks[length(breaks)] - tempo.extra.margin},
+ by = tick.dist
+ )
+ tick.pos <- tick.pos[tick.pos >= min(lim) & tick.pos <= max(lim)]
+ tick.values <- tick.pos
+ }
+ if(any(is.na(tick.pos) | ! is.finite(tick.pos))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, ": NA or Inf GENERATED FOR THE INTER TICK POSITIONS: ", paste(tick.pos, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(tick.pos) == 0L){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO INTER TICKS COMPUTED BETWEEN THE LIMITS INDICATED: ", paste(lim, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ output <- list(log = log, coordinates = tick.pos, values = tick.values, warn = warn)
+ if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE)) # to recover the warning messages, see $warn
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ return(output)
}
-# end last check
+######## fun_post_plot() #### set graph param after plotting (axes redesign for instance)
-# conversion of geom_hline and geom_vline
-for(i1 in 1:length(data1)){
-if(geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline"){
-final.data.frame <- data.frame()
-for(i3 in 1:nrow(data1[[i1]])){
-tempo.data.frame <- rbind(data1[[i1]][i3, ], data1[[i1]][i3, ], stringsAsFactors = TRUE)
-if(geom[[i1]] == "geom_hline"){
-tempo.data.frame[, x[[i1]]] <- x.lim
-}else if(geom[[i1]] == "geom_vline"){
-tempo.data.frame[, y[[i1]]] <- y.lim
-}else{
-tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-# 3 lines below inactivated because I put that above
-# if(is.null(categ[[i1]])){
-# data1[, "fake_categ"] <- paste0("Line_", i3)
-# }
-final.data.frame <- rbind(final.data.frame, tempo.data.frame, stringsAsFactors = TRUE)
-}
-data1[[i1]] <- final.data.frame
-geom[[i1]] <- "geom_line"
-if(length(color[[i1]])== 1L){
-color[[i1]] <- rep(color[[i1]], length(unique(data1[[i1]][ , categ[[i1]]])))
-}else if(length(color[[i1]]) != length(unique(data1[[i1]][ , categ[[i1]]]))){
-tempo.cat <- paste0("ERROR IN ", function.name, " geom_hline AND geom_vline CONVERSION TO FIT THE XLIM AND YLIM LIMITS OF THE DATA: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE THE LENGTH OF LEVELS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LEVELS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])))
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-}
+
+
+fun_post_plot <- function(
+ x.side = 0,
+ x.log.scale = FALSE,
+ x.categ = NULL,
+ x.categ.pos = NULL,
+ x.lab = "",
+ x.axis.size = 1.5,
+ x.label.size = 1.5,
+ x.dist.legend = 0.5,
+ x.nb.inter.tick = 1,
+ y.side = 0,
+ y.log.scale = FALSE,
+ y.categ = NULL,
+ y.categ.pos = NULL,
+ y.lab = "",
+ y.axis.size = 1.5,
+ y.label.size = 1.5,
+ y.dist.legend = 0.5,
+ y.nb.inter.tick = 1,
+ text.angle = 90,
+ tick.length = 0.5,
+ sec.tick.length = 0.3,
+ bg.color = NULL,
+ grid.lwd = NULL,
+ grid.col = "white",
+ corner.text = "",
+ corner.text.size = 1,
+ just.label.add = FALSE,
+ par.reset = FALSE,
+ custom.par = NULL
+){
+ # AIM
+ # redesign axis. If x.side = 0, y.side = 0, the function just adds text at topright of the graph and reset par() for next graphics and provides outputs (see below)
+ # provide also positions for legend or additional text on the graph
+ # use fun_prior_plot() before this function for initial inactivation of the axis drawings
+ # ARGUMENTS
+ # x.side: axis at the bottom (1) or top (3) of the region figure. Write 0 for no change
+ # x.log.scale: Log scale for the x-axis? Either TRUE or FALSE
+ # x.categ: character vector representing the classes (levels()) to specify when the x-axis is qualititative(stripchart, boxplot)
+ # x.categ.pos: position of the classes names (numeric vector of identical length than x.categ). If left NULL, this will be 1:length(levels())
+ # x.lab: label of the x-axis. If x.side == 0 and x.lab != "", then x.lab is printed
+ # x.axis.size: positive numeric. Increase or decrease the size of the x axis numbers. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2. Also control the size of displayed categories
+ # x.label.size: positive numeric. Increase or decrease the size of the x axis legend text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
+ # x.dist.legend: increase the number to move x-axis legends away in inches (first number of mgp argument of par() but in inches)
+ # x.nb.inter.tick: number of secondary ticks between main ticks on x-axis (only if not log scale). 0 means no secondary ticks
+ # y.side: axis at the left (2) or right (4) of the region figure. Write 0 for no change
+ # y.log.scale: Log scale for the y-axis? Either TRUE or FALSE
+ # y.categ: classes (levels()) to specify when the y-axis is qualititative(stripchart, boxplot)
+ # y.categ.pos: position of the classes names (numeric vector of identical length than y.categ). If left NULL, this will be 1:length(levels())
+ # y.lab: label of the y-axis. If y.side == 0 and y.lab != "", then y.lab is printed
+ # y.axis.size: positive numeric. Increase or decrease the size of the y axis numbers. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2. Also control the size of displayed categories
+ # y.label.size: positive numeric. Increase or decrease the size of the y axis legend text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
+ # y.dist.legend: increase the number to move y-axis legends away in inches (first number of mgp argument of par() but in inches)
+ # y.nb.inter.tick: number of secondary ticks between main ticks on y-axis (only if not log scale). 0 means non secondary ticks
+ # text.angle: angle of the text when axis is qualitative
+ # tick.length: length of the main ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
+ # sec.tick.length: length of the secondary ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
+ # bg.color: background color of the plot region. NULL for no color. BEWARE: cover/hide an existing plot !
+ # grid.lwd: if non NULL, activate the grid line (specify the line width)
+ # grid.col: grid line color (only if grid.lwd non NULL)
+ # corner.text: text to add at the top right corner of the window
+ # corner.text.size: positive numeric. Increase or decrease the size of the text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
+ # par.reset: to reset all the graphics parameters. BEWARE: TRUE can generate display problems, mainly in graphic devices with multiple figure regions
+ # just.label.add: just add axis labels (legend)? Either TRUE or FALSE. If TRUE, at least (x.side == 0 & x.lab != "") or (y.side == 0 & y.lab != "") must be set to display the corresponding x.lab or y.lab
+ # custom.par: list that provides the parameters that reset all the graphics parameters. BEWARE: if NULL and par.reset == TRUE, the default par() parameters are used
+ # RETURN
+ # a list containing:
+ # $x.mid.left.dev.region: middle of the left margin of the device region, in coordinates of the x-axis
+ # $x.left.dev.region: left side of the left margin (including the potential margin of the device region), in coordinates of the x-axis
+ # $x.mid.right.dev.region: middle of the right margin of the device region, in coordinates of the x-axis
+ # $x.right.dev.region: right side of the right margin (including the potential margin of the device region), in coordinates of the x-axis
+ # $x.mid.left.fig.region: middle of the left margin of the figure region, in coordinates of the x-axis
+ # $x.left.fig.region: left side of the left margin, in coordinates of the x-axis
+ # $x.mid.right.fig.region: middle of the right margin of the figure region, in coordinates of the x-axis
+ # $x.right.fig.region: right side of the right margin, in coordinates of the x-axis
+ # $x.left.plot.region: left side of the plot region, in coordinates of the x-axis
+ # $x.right.plot.region: right side of the plot region, in coordinates of the x-axis
+ # $x.mid.plot.region: middle of the plot region, in coordinates of the x-axis
+ # $y.mid.bottom.dev.region: middle of the bottom margin of the device region, in coordinates of the y-axis
+ # $y.bottom.dev.region: bottom side of the bottom margin (including the potential margin of the device region), in coordinates of the y-axis
+ # $y.mid.top.dev.region: middle of the top margin of the device region, in coordinates of the y-axis
+ # $y.top.dev.region: top side of the top margin (including the potential margin of the device region), in coordinates of the y-axis
+ # $y.mid.bottom.fig.region: middle of the bottom margin of the figure region, in coordinates of the y-axis
+ # $y.bottom.fig.region: bottom of the bottom margin of the figure region, in coordinates of the y-axis
+ # $y.mid.top.fig.region: middle of the top margin of the figure region, in coordinates of the y-axis
+ # $y.top.fig.region: top of the top margin of the figure region, in coordinates of the y-axis
+ # $y.top.plot.region: top of the plot region, in coordinates of the y-axis
+ # $y.bottom.plot.region: bottom of the plot region, in coordinates of the y-axis
+ # $y.mid.plot.region: middle of the plot region, in coordinates of the y-axis
+ # $text: warning text
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_open() to reinitialize graph parameters if par.reset = TRUE and custom.par = NULL
+ # EXAMPLES
+ # Example of log axis with log y-axis and unmodified x-axis:
+ # prior.par <- fun_prior_plot(param.reinitial = TRUE, xlog.scale = FALSE, ylog.scale = TRUE, remove.label = TRUE, remove.x.axis = FALSE, remove.y.axis = TRUE, down.space = 1, left.space = 1, up.space = 1, right.space = 1, orient = 1, dist.legend = 0.5, tick.length = 0.5, box.type = "n", amplif.label = 1, amplif.axis = 1, display.extend = FALSE, return.par = TRUE) ; plot(1:100, log = "y") ; fun_post_plot(y.side = 2, y.log.scale = prior.par$ylog, x.lab = "Values", y.lab = "TEST", y.axis.size = 1.25, y.label.size = 1.5, y.dist.legend = 0.7, just.label.add = ! prior.par$ann)
+ # Example of log axis with redrawn x-axis and y-axis:
+ # prior.par <- fun_prior_plot(param.reinitial = TRUE) ; plot(1:100) ; fun_post_plot(x.side = 1, x.lab = "Values", y.side = 2, y.lab = "TEST", y.axis.size = 1, y.label.size = 2, y.dist.legend = 0.6)
+ # Example of title easily added to a plot:
+ # plot(1:100) ; para <- fun_post_plot(corner.text = "TITLE ADDED") # try also: par(xpd = TRUE) ; text(x = para$x.mid.left.fig.region, y = para$y.mid.top.fig.region, labels = "TITLE ADDED", cex = 0.5)
+ # example with margins in the device region:
+ # windows(5,5) ; fun_prior_plot(box.type = "o") ; par(mai=c(0.5,0.5,0.5,0.5), omi = c(0.25,0.25,1,0.25), xaxs = "i", yaxs = "i") ; plot(0:10) ; a <- fun_post_plot(x.side = 0, y.side = 0) ; x <- c(a$x.mid.left.dev.region, a$x.left.dev.region, a$x.mid.right.dev.region, a$x.right.dev.region, a$x.mid.left.fig.region, a$x.left.fig.region, a$x.mid.right.fig.region, a$x.right.fig.region, a$x.right.plot.region, a$x.left.plot.region, a$x.mid.plot.region) ; y <- c(a$y.mid.bottom.dev.region, a$y.bottom.dev.region, a$y.mid.top.dev.region, a$y.top.dev.region, a$y.mid.bottom.fig.region, a$y.bottom.fig.region, a$y.mid.top.fig.region, a$y.top.fig.region, a$y.top.plot.region, a$y.bottom.plot.region, a$y.mid.plot.region) ; par(xpd = NA) ; points(x = rep(5, length(y)), y = y, pch = 16, col = "red") ; text(x = rep(5, length(y)), y = y, c("y.mid.bottom.dev.region", "y.bottom.dev.region", "y.mid.top.dev.region", "y.top.dev.region", "y.mid.bottom.fig.region", "y.bottom.fig.region", "y.mid.top.fig.region", "y.top.fig.region", "y.top.plot.region", "y.bottom.plot.region", "y.mid.plot.region"), cex = 0.65, col = grey(0.25)) ; points(y = rep(5, length(x)), x = x, pch = 16, col = "blue") ; text(y = rep(5, length(x)), x = x, c("x.mid.left.dev.region", "x.left.dev.region", "x.mid.right.dev.region", "x.right.dev.region", "x.mid.left.fig.region", "x.left.fig.region", "x.mid.right.fig.region", "x.right.fig.region", "x.right.plot.region", "x.left.plot.region", "x.mid.plot.region"), cex = 0.65, srt = 90, col = grey(0.25))
+ # DEBUGGING
+ # x.side = 0 ; x.log.scale = FALSE ; x.categ = NULL ; x.categ.pos = NULL ; x.lab = "" ; x.axis.size = 1.5 ; x.label.size = 1.5 ; x.dist.legend = 1 ; x.nb.inter.tick = 1 ; y.side = 0 ; y.log.scale = FALSE ; y.categ = NULL ; y.categ.pos = NULL ; y.lab = "" ; y.axis.size = 1.5 ; y.label.size = 1.5 ; y.dist.legend = 0.7 ; y.nb.inter.tick = 1 ; text.angle = 90 ; tick.length = 0.5 ; sec.tick.length = 0.3 ; bg.color = NULL ; grid.lwd = NULL ; grid.col = "white" ; corner.text = "" ; corner.text.size = 1 ; just.label.add = FALSE ; par.reset = FALSE ; custom.par = NULL # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_open", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_open() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = x.side, options = c(0, 1, 3), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.log.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(x.categ)){
+ tempo <- fun_check(data = x.categ, class = "character", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(x.categ.pos)){
+ tempo <- fun_check(data = x.categ.pos, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = x.lab, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.axis.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.label.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.nb.inter.tick, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.side, options = c(0, 2, 4), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.log.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(y.categ)){
+ tempo <- fun_check(data = y.categ, class = "character", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(y.categ.pos)){
+ tempo <- fun_check(data = y.categ.pos, class = "vector", mode = "numeric", fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = y.lab, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.axis.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.label.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.nb.inter.tick, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = text.angle, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = sec.tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(bg.color)){
+ tempo <- fun_check(data = bg.color, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! (bg.color %in% colors() | grepl(pattern = "^#", bg.color))){ # check color
+ tempo.cat <- paste0("ERROR IN ", function.name, ": bg.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # OR A COLOR NAME GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ if( ! is.null(grid.lwd)){
+ tempo <- fun_check(data = grid.lwd, class = "vector", mode = "numeric", neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(grid.col)){
+ tempo <- fun_check(data = grid.col, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! (grid.col %in% colors() | grepl(pattern = "^#", grid.col))){ # check color
+ tempo.cat <- paste0("ERROR IN ", function.name, ": grid.col ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # OR A COLOR NAME GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = corner.text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = corner.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = just.label.add, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = par.reset, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(custom.par)){
+ tempo <- fun_check(data = custom.par, typeof = "list", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ text <- NULL
+ par(tcl = -par()$mgp[2] * tick.length)
+ if(x.log.scale == TRUE){
+ grid.coord.x <- c(10^par("usr")[1], 10^par("usr")[2])
+ }else{
+ grid.coord.x <- c(par("usr")[1], par("usr")[2])
+ }
+ if(y.log.scale == TRUE){
+ grid.coord.y <- c(10^par("usr")[3], 10^par("usr")[4])
+ }else{
+ grid.coord.y <- c(par("usr")[3], par("usr")[4])
+ }
+ if( ! is.null(bg.color)){
+ rect(grid.coord.x[1], grid.coord.y[1], grid.coord.x[2], grid.coord.y[2], col = bg.color, border = NA)
+ }
+ if( ! is.null(grid.lwd)){
+ grid(nx = NA, ny = NULL, col = grid.col, lty = 1, lwd = grid.lwd)
+ }
+ if(x.log.scale == TRUE){
+ x.mid.left.dev.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
+ x.left.dev.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1]) # in x coordinates
+ x.mid.right.dev.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
+ x.right.dev.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2])) # in x coordinates
+ x.mid.left.fig.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
+ x.left.fig.region <- 10^(par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1]) # in x coordinates
+ x.mid.right.fig.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
+ x.right.fig.region <- 10^(par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2])) # in x coordinates
+ x.left.plot.region <- 10^par("usr")[1] # in x coordinates, left of the plot region (according to x scale)
+ x.right.plot.region <- 10^par("usr")[2] # in x coordinates, right of the plot region (according to x scale)
+ x.mid.plot.region <- 10^((par("usr")[2] + par("usr")[1]) / 2) # in x coordinates, right of the plot region (according to x scale)
+ }else{
+ x.mid.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
+ x.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1]) # in x coordinates
+ x.mid.right.dev.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
+ x.right.dev.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) + ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * (1 - par("omd")[2])) # in x coordinates
+ x.mid.left.fig.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] / 2) # in x coordinates, to position axis labeling at the bottom of the graph (according to x scale)
+ x.left.fig.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1]) # in x coordinates
+ x.mid.right.fig.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2]) / 2) # in x coordinates, to position axis labeling at the top of the graph (according to x scale)
+ x.right.fig.region <- (par("usr")[2] + ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * (1 - par("plt")[2])) # in x coordinates
+ x.left.plot.region <- par("usr")[1] # in x coordinates, left of the plot region (according to x scale)
+ x.right.plot.region <- par("usr")[2] # in x coordinates, right of the plot region (according to x scale)
+ x.mid.plot.region <- (par("usr")[2] + par("usr")[1]) / 2 # in x coordinates, right of the plot region (according to x scale)
+ }
+ if(y.log.scale == TRUE){
+ y.mid.bottom.dev.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (par("omd")[3] / 2)) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
+ y.bottom.dev.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * par("omd")[3]) # in y coordinates
+ y.mid.top.dev.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
+ y.top.dev.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4])) # in y coordinates
+ y.mid.bottom.fig.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] / 2) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
+ y.bottom.fig.region <- 10^(par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3]) # in y coordinates
+ y.mid.top.fig.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
+ y.top.fig.region <- 10^(par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4])) # in y coordinates
+ y.top.plot.region <- 10^par("usr")[4] # in y coordinates, top of the plot region (according to y scale)
+ y.bottom.plot.region <- 10^par("usr")[3] # in y coordinates, bottom of the plot region (according to y scale)
+ y.mid.plot.region <- (par("usr")[3] + par("usr")[4]) / 2 # in x coordinates, right of the plot region (according to x scale)
+ }else{
+ y.mid.bottom.dev.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (par("omd")[3] / 2)) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
+ y.bottom.dev.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] - ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * par("omd")[3]) # in y coordinates
+ y.mid.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
+ y.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4])) # in y coordinates
+ y.mid.bottom.fig.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3] / 2) # in y coordinates, to position axis labeling at the bottom of the graph (according to y scale). Ex mid.bottom.space
+ y.bottom.fig.region <- (par("usr")[3] - ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * par("plt")[3]) # in y coordinates
+ y.mid.top.fig.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) / 2) # in y coordinates, to position axis labeling at the top of the graph (according to y scale). Ex mid.top.space
+ y.top.fig.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4])) # in y coordinates
+ y.top.plot.region <- par("usr")[4] # in y coordinates, top of the plot region (according to y scale)
+ y.bottom.plot.region <- par("usr")[3] # in y coordinates, bottom of the plot region (according to y scale)
+ y.mid.plot.region <- ((par("usr")[3] + par("usr")[4]) / 2) # in x coordinates, right of the plot region (according to x scale)
+ }
+ if(any(sapply(FUN = all.equal, c(1, 3), x.side) == TRUE)){
+ par(xpd=FALSE, xaxt="s")
+ if(is.null(x.categ) & x.log.scale == TRUE){
+ if(any(par()$xaxp[1:2] == 0L)){ # any(sapply(FUN = all.equal, par()$xaxp[1:2], 0) == TRUE) not used because we strictly need zero as a result. Beware: write "== TRUE", because the result is otherwise character and a warning message appears using any()
+ if(par()$xaxp[1] == 0L){ # isTRUE(all.equal(par()$xaxp[1], 0)) not used because we strictly need zero as a result
+ par(xaxp = c(10^-30, par()$xaxp[2:3])) # because log10(par()$xaxp[1] == 0) == -Inf
+ }
+ if(par()$xaxp[2] == 0L){ # isTRUE(all.equal(par()$xaxp[1], 0)) not used because we strictly need zero as a result
+ par(xaxp = c(par()$xaxp[1], 10^-30, par()$xaxp[3])) # because log10(par()$xaxp[2] == 0) == -Inf
+ }
+ }
+ axis(side = x.side, at = c(10^par()$usr[1], 10^par()$usr[2]), labels=rep("", 2), lwd=1, lwd.ticks = 0) # draw the axis line
+ mtext(side = x.side, text = x.lab, line = x.dist.legend / 0.2, las = 0, cex = x.label.size)
+ par(tcl = -par()$mgp[2] * sec.tick.length) # length of the secondary ticks are reduced
+ suppressWarnings(rug(10^outer(c((log10(par("xaxp")[1]) -1):log10(par("xaxp")[2])), log10(1:10), "+"), ticksize = NA, side = x.side)) # ticksize = NA to allow the use of par()$tcl value
+ par(tcl = -par()$mgp[2] * tick.length) # back to main ticks
+ axis(side = x.side, at = c(1e-15, 1e-14, 1e-13, 1e-12, 1e-11, 1e-10, 1e-9, 1e-8, 1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 1e-1, 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10), labels = expression(10^-15, 10^-14, 10^-13, 10^-12, 10^-11, 10^-10, 10^-9, 10^-8, 10^-7, 10^-6, 10^-5, 10^-4, 10^-3, 10^-2, 10^-1, 10^0, 10^1, 10^2, 10^3, 10^4, 10^5, 10^6, 10^7, 10^8, 10^9, 10^10), lwd = 0, lwd.ticks = 1, cex.axis = x.axis.size)
+ x.text <- 10^par("usr")[2]
+ }else if(is.null(x.categ) & x.log.scale == FALSE){
+ axis(side=x.side, at=c(par()$usr[1], par()$usr[2]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
+ axis(side=x.side, at=round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), cex.axis = x.axis.size) # axis(side=x.side, at=round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), labels = format(round(seq(par()$xaxp[1], par()$xaxp[2], length.out=par()$xaxp[3]+1), 2), big.mark=','), cex.axis = x.axis.size) # to get the 1000 comma separator
+ mtext(side = x.side, text = x.lab, line = x.dist.legend / 0.2, las = 0, cex = x.label.size)
+ if(x.nb.inter.tick > 0){
+ inter.tick.unit <- (par("xaxp")[2] - par("xaxp")[1]) / par("xaxp")[3]
+ par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
+ suppressWarnings(rug(seq(par("xaxp")[1] - 10 * inter.tick.unit, par("xaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + x.nb.inter.tick)), ticksize = NA, x.side)) # ticksize = NA to allow the use of par()$tcl value
+ par(tcl = -par()$mgp[2] * tick.length) # back to main ticks
+ }
+ x.text <- par("usr")[2]
+ }else if(( ! is.null(x.categ)) & x.log.scale == FALSE){
+ if(is.null(x.categ.pos)){
+ x.categ.pos <- 1:length(x.categ)
+ }else if(length(x.categ.pos) != length(x.categ)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.categ.pos MUST BE THE SAME LENGTH AS x.categ")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ par(xpd = TRUE)
+ if(isTRUE(all.equal(x.side, 1))){ #isTRUE(all.equal(x.side, 1)) is similar to x.side == 1L but deals with float
+ segments(x0 = x.left.plot.region, x1 = x.right.plot.region, y0 = y.bottom.plot.region, y1 = y.bottom.plot.region) # draw the line of the axis
+ text(x = x.categ.pos, y = y.mid.bottom.fig.region, labels = x.categ, srt = text.angle, cex = x.axis.size)
+ }else if(isTRUE(all.equal(x.side, 3))){ #isTRUE(all.equal(x.side, 1)) is similar to x.side == 3L but deals with float
+ segments(x0 = x.left.plot.region, x1 = x.right.plot.region, y0 = y.top.plot.region, y1 = y.top.plot.region) # draw the line of the axis
+ text(x = x.categ.pos, y = y.mid.top.fig.region, labels = x.categ, srt = text.angle, cex = x.axis.size)
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ARGUMENT x.side CAN ONLY BE 1 OR 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ par(xpd = FALSE)
+ x.text <- par("usr")[2]
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": PROBLEM WITH THE x.side (", x.side ,") OR x.log.scale (", x.log.scale,") ARGUMENTS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ x.text <- par("usr")[2]
+ }
+ if(any(sapply(FUN = all.equal, c(2, 4), y.side) == TRUE)){
+ par(xpd=FALSE, yaxt="s")
+ if(is.null(y.categ) & y.log.scale == TRUE){
+ if(any(par()$yaxp[1:2] == 0L)){ # any(sapply(FUN = all.equal, par()$yaxp[1:2], 0) == TRUE) not used because we strictly need zero as a result. Beware: write "== TRUE", because the result is otherwise character and a warning message appears using any()
+ if(par()$yaxp[1] == 0L){ # strict zero needed
+ par(yaxp = c(10^-30, par()$yaxp[2:3])) # because log10(par()$yaxp[1] == 0) == -Inf
+ }
+ if(par()$yaxp[2] == 0L){ # strict zero needed
+ par(yaxp = c(par()$yaxp[1], 10^-30, par()$yaxp[3])) # because log10(par()$yaxp[2] == 0) == -Inf
+ }
+ }
+ axis(side=y.side, at=c(10^par()$usr[3], 10^par()$usr[4]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
+ par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
+ suppressWarnings(rug(10^outer(c((log10(par("yaxp")[1])-1):log10(par("yaxp")[2])), log10(1:10), "+"), ticksize = NA, side = y.side)) # ticksize = NA to allow the use of par()$tcl value
+ par(tcl = -par()$mgp[2] * tick.length) # back to main tick length
+ axis(side = y.side, at = c(1e-15, 1e-14, 1e-13, 1e-12, 1e-11, 1e-10, 1e-9, 1e-8, 1e-7, 1e-6, 1e-5, 1e-4, 1e-3, 1e-2, 1e-1, 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, 1e10), labels = expression(10^-15, 10^-14, 10^-13, 10^-12, 10^-11, 10^-10, 10^-9, 10^-8, 10^-7, 10^-6, 10^-5, 10^-4, 10^-3, 10^-2, 10^-1, 10^0, 10^1, 10^2, 10^3, 10^4, 10^5, 10^6, 10^7, 10^8, 10^9, 10^10), lwd = 0, lwd.ticks = 1, cex.axis = y.axis.size)
+ y.text <- 10^(par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
+ mtext(side = y.side, text = y.lab, line = y.dist.legend / 0.2, las = 0, cex = y.label.size)
+ }else if(is.null(y.categ) & y.log.scale == FALSE){
+ axis(side=y.side, at=c(par()$usr[3], par()$usr[4]), labels=rep("", 2), lwd=1, lwd.ticks=0) # draw the axis line
+ axis(side=y.side, at=round(seq(par()$yaxp[1], par()$yaxp[2], length.out=par()$yaxp[3]+1), 2), cex.axis = y.axis.size)
+ mtext(side = y.side, text = y.lab, line = y.dist.legend / 0.2, las = 0, cex = y.label.size)
+ if(y.nb.inter.tick > 0){
+ inter.tick.unit <- (par("yaxp")[2] - par("yaxp")[1]) / par("yaxp")[3]
+ par(tcl = -par()$mgp[2] * sec.tick.length) # length of the ticks are reduced
+ suppressWarnings(rug(seq(par("yaxp")[1] - 10 * inter.tick.unit, par("yaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + y.nb.inter.tick)), ticksize = NA, side=y.side)) # ticksize = NA to allow the use of par()$tcl value
+ par(tcl = -par()$mgp[2] * tick.length) # back to main tick length
+ }
+ y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
+ }else if(( ! is.null(y.categ)) & y.log.scale == FALSE){
+ if(is.null(y.categ.pos)){
+ y.categ.pos <- 1:length(y.categ)
+ }else if(length(y.categ.pos) != length(y.categ)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.categ.pos MUST BE THE SAME LENGTH AS y.categ")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ axis(side = y.side, at = y.categ.pos, labels = rep("", length(y.categ)), lwd=0, lwd.ticks=1) # draw the line of the axis
+ par(xpd = TRUE)
+ if(isTRUE(all.equal(y.side, 2))){ #isTRUE(all.equal(y.side, 2)) is similar to y.side == 2L but deals with float
+ text(x = x.mid.left.fig.region, y = y.categ.pos, labels = y.categ, srt = text.angle, cex = y.axis.size)
+ }else if(isTRUE(all.equal(y.side, 4))){ # idem
+ text(x = x.mid.right.fig.region, y = y.categ.pos, labels = y.categ, srt = text.angle, cex = y.axis.size)
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ARGUMENT y.side CAN ONLY BE 2 OR 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ par(xpd = FALSE)
+ y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": PROBLEM WITH THE y.side (", y.side ,") OR y.log.scale (", y.log.scale,") ARGUMENTS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ y.text <- (par("usr")[4] + (par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3]) * (1 - par("plt")[4]))
+ }
+ par(xpd=NA)
+ text(x = x.mid.right.fig.region, y = y.text, corner.text, adj=c(1, 1.1), cex = corner.text.size) # text at the topright corner. Replace x.right.fig.region by x.text if text at the right edge of the plot region
+ if(just.label.add == TRUE & isTRUE(all.equal(x.side, 0)) & x.lab != ""){
+ text(x = x.mid.plot.region, y = y.mid.bottom.fig.region, x.lab, adj=c(0.5, 0.5), cex = x.label.size) # x label
+ }
+ if(just.label.add == TRUE & isTRUE(all.equal(y.side, 0)) & y.lab != ""){
+ text(x = y.mid.plot.region, y = x.mid.left.fig.region, y.lab, adj=c(0.5, 0.5), cex = y.label.size) # x label
+ }
+ par(xpd=FALSE)
+ if(par.reset == TRUE){
+ tempo.par <- fun_open(pdf = FALSE, return.output = TRUE)
+ invisible(dev.off()) # close the new window
+ if( ! is.null(custom.par)){
+ if( ! names(custom.par) %in% names(tempo.par$ini.par)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": custom.par ARGUMENT SHOULD HAVE THE NAMES OF THE COMPARTMENT LIST COMING FROM THE par() LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ par(custom.par)
+ text <- c(text, "\nGRAPH PARAMETERS SET TO VALUES DEFINED BY custom.par ARGUMENT\n")
+ }else{
+ par(tempo.par$ini.par)
+ text <- c(text, "\nGRAPH PARAMETERS RESET TO par() DEFAULT VALUES\n")
+ }
+ }
+ output <- list(x.mid.left.dev.region = x.mid.left.dev.region, x.left.dev.region = x.left.dev.region, x.mid.right.dev.region = x.mid.right.dev.region, x.right.dev.region = x.right.dev.region, x.mid.left.fig.region = x.mid.left.fig.region, x.left.fig.region = x.left.fig.region, x.mid.right.fig.region = x.mid.right.fig.region, x.right.fig.region = x.right.fig.region, x.left.plot.region = x.left.plot.region, x.right.plot.region = x.right.plot.region, x.mid.plot.region = x.mid.plot.region, y.mid.bottom.dev.region = y.mid.bottom.dev.region, y.bottom.dev.region = y.bottom.dev.region, y.mid.top.dev.region = y.mid.top.dev.region, y.top.dev.region = y.top.dev.region, y.mid.bottom.fig.region = y.mid.bottom.fig.region, y.bottom.fig.region = y.bottom.fig.region, y.mid.top.fig.region = y.mid.top.fig.region, y.top.fig.region = y.top.fig.region, y.top.plot.region = y.top.plot.region, y.bottom.plot.region = y.bottom.plot.region, y.mid.plot.region = y.mid.plot.region, text = text)
+ return(output)
}
-# end conversion of geom_hline and geom_vline
+######## fun_close() #### close specific graphic windows
-# kind of geom_point (vectorial or raster)
-scatter.kind <- vector("list", length = length(data1)) # list of same length as data1, that will be used to use either ggplot2::geom_point() (vectorial dot layer) or fun_gg_point_rast() (raster dot layer)
-fix.ratio <- FALSE
-if(is.null(raster.threshold)){
-if(raster == TRUE){
-scatter.kind[] <- "fun_gg_point_rast" # not important to fill everything: will be only used when geom == "geom_point"
-fix.ratio <- TRUE
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") RASTER PLOT GENERATED -> ASPECT RATIO OF THE PLOT REGION SET BY THE raster.ratio ARGUMENT (", fun_round(raster.ratio, 2), ") TO AVOID A BUG OF ELLIPSOID DOT DRAWING")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}else{
-scatter.kind[] <- "ggplot2::geom_point"
-}
-}else{
-for(i2 in 1:length(data1)){
-if(geom[[i2]] == "geom_point"){
-if(nrow(data1[[i2]]) <= raster.threshold){
-scatter.kind[[i2]] <- "ggplot2::geom_point"
-}else{
-scatter.kind[[i2]] <- "fun_gg_point_rast"
-fix.ratio <- TRUE
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i2, " OF data1 ARGUMENT")), " LAYER AS RASTER (NOT VECTORIAL)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-}
-if(any(unlist(scatter.kind) == "fun_gg_point_rast")){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") RASTER PLOT GENERATED -> ASPECT RATIO OF THE PLOT REGION SET BY THE raster.ratio ARGUMENT (", fun_round(raster.ratio, 2), ") TO AVOID A BUG OF ELLIPSOID DOT DRAWING")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}
-# end kind of geom_point (vectorial or raster)
-
-
-
-
-# no need loop part
-coord.names <- NULL
-tempo.gg.name <- "gg.indiv.plot."
-tempo.gg.count <- 0
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::xlab(if(is.null(x.lab)){x[[1]]}else{x.lab}))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ylab(if(is.null(y.lab)){y[[1]]}else{y.lab}))
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
-# text angle management
-x.tempo.just <- fun_gg_just(angle = x.text.angle, pos = "bottom", kind = "axis")
-y.tempo.just <- fun_gg_just(angle = y.text.angle, pos = "left", kind = "axis")
-# end text angle management
-add.check <- TRUE
-if( ! is.null(add)){ # if add is NULL, then = 0
-if(grepl(pattern = "ggplot2::theme", add) == TRUE){
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT\n-> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER\n-> article ARGUMENT WILL BE IGNORED\nIT IS RECOMMENDED TO USE \"+ theme(aspect.ratio = raster.ratio)\" IF RASTER MODE IS ACTIVATED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-add.check <- FALSE
-}
-}
-if(add.check == TRUE & article == TRUE){
-# WARNING: not possible to add several times theme(). NO message but the last one overwrites the others
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
-if(grid == TRUE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-line = ggplot2::element_line(size = 0.5),
-axis.line.y.left = ggplot2::element_line(colour = "black"), # draw lines for the y axis
-axis.line.x.bottom = ggplot2::element_line(colour = "black"), # draw lines for the x axis
-panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.minor.x = ggplot2::element_line(colour = "grey90", size = 0.25),
-panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
-axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
-axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
-aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
-))
-}else{
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_line(size = 0.5),
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-axis.line.y.left = ggplot2::element_line(colour = "black"),
-axis.line.x.bottom = ggplot2::element_line(colour = "black"),
-axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
-axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
-aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
-))
-}
-}else if(add.check == TRUE & article == FALSE){
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
-text = ggplot2::element_text(size = text.size),
-plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
-line = ggplot2::element_line(size = 0.5),
-legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
-panel.background = ggplot2::element_rect(fill = "grey95"),
-axis.line.y.left = ggplot2::element_line(colour = "black"),
-axis.line.x.bottom = ggplot2::element_line(colour = "black"),
-panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.minor.x = ggplot2::element_line(colour = "grey90", size = 0.25),
-panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
-panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
-strip.background = ggplot2::element_rect(fill = "white", colour = "black"),
-axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
-axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
-aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
-# do not work -> legend.position = "none" # to remove the legend completely: https://www.datanovia.com/en/blog/how-to-remove-legend-from-a-ggplot/
-))
-}
-# end no need loop part
-
-
-# loop part
-point.count <- 0
-line.count <- 0
-lg.order <- vector(mode = "list", length = 6) # order of the legend
-lg.order <- lapply(lg.order, as.numeric) # order of the legend
-lg.color <- vector(mode = "list", length = 6) # color of the legend
-lg.dot.shape <- vector(mode = "list", length = 6) # etc.
-lg.dot.size <- vector(mode = "list", length = 6) # etc.
-lg.dot.size <- lapply(lg.dot.size, as.numeric) # etc.
-lg.dot.border.size <- vector(mode = "list", length = 6) # etc.
-lg.dot.border.size <- lapply(lg.dot.border.size, as.numeric) # etc.
-lg.dot.border.color <- vector(mode = "list", length = 6) # etc.
-lg.line.size <- vector(mode = "list", length = 6) # etc.
-lg.line.size <- lapply(lg.line.size, as.numeric) # etc.
-lg.line.type <- vector(mode = "list", length = 6) # etc.
-lg.alpha <- vector(mode = "list", length = 6) # etc.
-lg.alpha <- lapply(lg.alpha, as.numeric) # etc.
-for(i1 in 1:length(data1)){
-if(geom[[i1]] == "geom_point"){
-point.count <- point.count + 1
-if(point.count== 1L){
-fin.lg.disp[[1]] <- legend.disp[[point.count + line.count]]
-lg.order[[1]] <- point.count + line.count
-lg.color[[1]] <- color[[i1]] # if color == NULL -> NULL
-lg.dot.shape[[1]] <- dot.shape[[i1]]
-lg.dot.size[[1]] <- dot.size[[i1]]
-lg.dot.border.size[[1]] <- dot.border.size[[i1]]
-lg.dot.border.color[[1]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
-if(plot == TRUE & fin.lg.disp[[1]] == TRUE & dot.shape[[1]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[1]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[1]] <- alpha[[i1]]
-}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], fill = categ[[i1]]), shape = dot.shape[[i1]], size = dot.size[[i1]], stroke = dot.border.size[[i1]], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = if(i5== 1L){TRUE}else{FALSE})) # WARNING: a single color allowed for color argument outside aesthetic, but here a single color for border --> loop could be inactivated but kept for commodity # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
-}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_fill_manual(name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = as.character(color[[i1]]), breaks = class.categ)) # values are the values of fill, breaks reorder the classes according to class.categ in the legend, order argument of guide_legend determines the order of the different aesthetics in the legend (not order of classes). See guide_legend settings of scale_..._manual below
-}
-if(point.count== 2L){
-fin.lg.disp[[2]] <- legend.disp[[point.count + line.count]]
-lg.order[[2]] <- point.count + line.count
-lg.color[[2]] <- color[[i1]] # if color == NULL -> NULL
-lg.dot.shape[[2]] <- dot.shape[[i1]]
-lg.dot.size[[2]] <- dot.size[[i1]]
-lg.dot.border.size[[2]] <- dot.border.size[[i1]]
-lg.dot.border.color[[2]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
-if(plot == TRUE & fin.lg.disp[[2]] == TRUE & dot.shape[[2]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[2]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[2]] <- alpha[[i1]]
-}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], shape = categ[[i1]]), size = dot.size[[i1]], stroke = dot.border.size[[i1]], fill = color[[i1]][i5], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = FALSE)) # WARNING: a single color allowed for fill argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+fun_close <- function(kind = "pdf", return.text = FALSE){
+ # AIM
+ # close only specific graphic windows (devices)
+ # ARGUMENTS:
+ # kind: vector, among c("windows", "quartz", "x11", "X11", "pdf", "bmp", "png", "tiff"), indicating the kind of graphic windows (devices) to close. BEWARE: either "windows", "quartz", "x11" or "X11" means that all the X11 GUI graphics devices will be closed, whatever the OS used
+ # return.text: print text regarding the kind parameter and the devices that were finally closed?
+ # RETURN
+ # text regarding the kind parameter and the devices that were finally closed
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # windows() ; windows() ; pdf() ; dev.list() ; fun_close(kind = c("pdf", "x11"), return.text = TRUE) ; dev.list()
+ # DEBUGGING
+ # kind = c("windows", "pdf") ; return.text = FALSE # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = kind, options = c("windows", "quartz", "x11", "X11", "pdf", "bmp", "png", "tiff"), fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.text, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ text <- paste0("THE REQUIRED KIND OF GRAPHIC DEVICES TO CLOSE ARE ", paste(kind, collapse = " "))
+ if(Sys.info()["sysname"] == "Windows"){ # Note that .Platform$OS.type() only says "unix" for macOS and Linux and "Windows" for Windows
+ if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
+ tempo <- kind %in% c("windows", "quartz", "x11", "X11")
+ kind[tempo] <- "windows" # term are replaced by what is displayed when using a <- dev.list() ; names(a)
+ }
+ }else if(Sys.info()["sysname"] == "Linux"){
+ if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
+ tempo.device <- suppressWarnings(try(X11(), silent = TRUE))[] # open a X11 window to try to recover the X11 system used
+ if( ! is.null(tempo.device)){
+ text <- paste0(text, "\nCANNOT CLOSE GUI GRAPHIC DEVICES AS REQUIRED BECAUSE THIS LINUX SYSTEM DOES NOT HAVE IT")
+ }else{
+ tempo <- kind %in% c("windows", "quartz", "x11", "X11")
+ kind[tempo] <- names(dev.list()[length(dev.list())]) # term are replaced by what is displayed when using a <- dev.list() ; names(a)
+ invisible(dev.off()) # close the X11 opened by tempo
+ }
+ }
+ }else{ # for macOS
+ if(any(kind %in% c("windows", "quartz", "x11", "X11"))){
+ tempo <- kind %in% c("windows", "quartz", "x11", "X11")
+ kind[tempo] <- "quartz" # term are replaced by what is displayed when using a <- dev.list() ; names(a)
+ }
+ }
+ kind <- unique(kind)
+ if(length(dev.list()) != 0){
+ for(i in length(names(dev.list())):1){
+ if(names(dev.list())[i] %in% kind){
+ text <- paste0(text, "\n", names(dev.list())[i], " DEVICE NUMBER ", dev.list()[i], " HAS BEEN CLOSED")
+ invisible(dev.off(dev.list()[i]))
+ }
+ }
+ }
+ if(return.text == TRUE){
+ return(text)
+ }
}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_shape_manual(name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(dot.shape[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of shape, breaks reorder the classes according to class.categ in the legend. See guide_legend settings of scale_..._manual below
+
+################ Standard graphics
+
+
+######## fun_empty_graph() #### text to display for empty graphs
+
+
+
+
+
+fun_empty_graph <- function(
+ text = NULL,
+ text.size = 1,
+ title = NULL,
+ title.size = 1.5
+){
+ # AIM
+ # display an empty plot with a text in the middle of the window (for instance to specify that no plot can be drawn)
+ # ARGUMENTS
+ # text: character string of the message to display
+ # text.size: numeric value of the text size
+ # title: character string of the graph title
+ # title.size: numeric value of the title size (in points)
+ # RETURN
+ # an empty plot
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # simple example
+ # fun_empty_graph(text = "NO GRAPH")
+ # white page
+ # fun_empty_graph() # white page
+ # all the arguments
+ # fun_empty_graph(text = "NO GRAPH", text.size = 2, title = "GRAPH1", title.size = 1)
+ # DEBUGGING
+ # text = "NO GRAPH" ; title = "GRAPH1" ; text.size = 1
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if( ! is.null(text)){
+ tempo <- fun_check(data = text, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(title)){
+ tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = title.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ ini.par <- par(no.readonly = TRUE) # to recover the initial graphical parameters if required (reset). BEWARE: this command alone opens a pdf of GUI window if no window already opened. But here, protected with the code because always a tempo window opened
+ par(ann=FALSE, xaxt="n", yaxt="n", mar = rep(1, 4), bty = "n", xpd = NA)
+ plot(1, 1, type = "n") # no display with type = "n"
+ x.left.dev.region <- (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / ((par("omd")[2] - par("omd")[1]) * (par("plt")[2] - par("plt")[1]))) * par("omd")[1])
+ y.top.dev.region <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par("omd")[4] - par("omd")[3]) * (par("plt")[4] - par("plt")[3]))) * (1 - par("omd")[4]))
+ if( ! is.null(text)){
+ text(x = 1, y = 1, labels = text, cex = text.size)
+ }
+ if( ! is.null(title)){
+ text(x = x.left.dev.region, y = y.top.dev.region, labels = title, adj=c(0, 1), cex = title.size)
+ }
+ par(ini.par)
}
-if(point.count== 3L){
-fin.lg.disp[[3]] <- legend.disp[[point.count + line.count]]
-lg.order[[3]] <- point.count + line.count
-lg.color[[3]] <- color[[i1]] # if color == NULL -> NULL
-lg.dot.shape[[3]] <- dot.shape[[i1]]
-lg.dot.size[[3]] <- dot.size[[i1]]
-lg.dot.border.size[[3]] <- dot.border.size[[i1]]
-lg.dot.border.color[[3]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
-if(plot == TRUE & fin.lg.disp[[3]] == TRUE & dot.shape[[3]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[3]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[3]] <- alpha[[i1]]
-}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], stroke = categ[[i1]]), shape = dot.shape[[i1]], size = dot.size[[i1]], fill = color[[i1]][i5], stroke = dot.border.size[[i1]], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = FALSE)) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+
+
+################ gg graphics
+
+
+######## fun_gg_palette() #### ggplot2 default color palette
+
+
+
+
+
+fun_gg_palette <- function(n, kind = "std"){
+ # AIM
+ # provide colors used by ggplot2
+ # the interest is to use another single color that is not the red one used by default
+ # for ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
+ # ARGUMENTS
+ # n: number of groups on the graph
+ # kind: either "std" for standard gg colors, "dark" for darkened gg colors, or "light" for pastel gg colors
+ # RETURN
+ # the vector of hexadecimal colors
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # output of the function
+ # fun_gg_palette(n = 2)
+ # the ggplot2 palette when asking for 7 different colors
+ # plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7))
+ # selection of the 5th color of the ggplot2 palette made of 7 different colors
+ # plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7)[5])
+ # the ggplot2 palette made of 7 darkened colors
+ # plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7, kind = "dark"))
+ # the ggplot2 palette made of 7 lighten colors
+ # plot(1:7, pch = 16, cex = 5, col = fun_gg_palette(n = 7, kind = "light"))
+ # DEBUGGING
+ # n = 0
+ # kind = "std"
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = n, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & isTRUE(all.equal(n, 0))){ # isTRUE(all.equal(n, 0))) is similar to n == 0 but deals with float
+ tempo.cat <- paste0("ERROR IN ", function.name, ": n ARGUMENT MUST BE A NON ZERO INTEGER. HERE IT IS: ", paste(n, collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ tempo <- fun_check(data = kind, options = c("std", "dark", "light"), length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ hues = seq(15, 375, length = n + 1)
+ hcl(h = hues, l = if(kind == "std"){65}else if(kind == "dark"){35}else if(kind == "light"){85}, c = 100)[1:n]
}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "stroke", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(dot.border.size[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of stroke, breaks reorder the classes according to class.categ in the legend. See guide_legend settings of scale_..._manual below
+
+######## fun_gg_just() #### ggplot2 justification of the axis labeling, depending on angle
+
+
+
+
+
+fun_gg_just <- function(angle, pos, kind = "axis"){
+ # AIM
+ # provide correct justification for text labeling, depending on the chosen angle
+ # WARNINGS
+ # justification behave differently on plot, depending whether it is used for annotayed text or for axis labelling. Indeed the latter has labelling constrained
+ # Of note, a bug in ggplot2: vjust sometimes does not work, i.e., the same justification result is obtained whatever the value used. This is the case with angle = 90, pos = "top", kind = "axis". While everything is fine with angle = 90, pos = "bottom", kind = "axis". At least, everything seems fine for kind = "axis" and pos = c("left", "bottom")
+ # ARGUMENTS
+ # angle: integer value of the text angle, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
+ # pos: where text is? Either "top", "right", "bottom" or "left" of the elements to justify from
+ # kind: kind of text? Either "axis" or "text". In the first case, the pos argument refers to the axis position, and in the second to annotated text (using ggplot2::annotate() or ggplot2::geom_text())
+ # RETURN
+ # a list containing:
+ # $angle: the submitted angle (value potentially reduced to fit the [-360 ; 360] interval, e.g., 460 -> 100, without impact on the final angle displayed)
+ # $pos: the selected position (argument pos)
+ # $kind: the selected kind of text (argument kind)
+ # $hjust: the horizontal justification
+ # $vjust: the vertical justification
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_gg_just(angle = 45, pos = "bottom")
+ # fun_gg_just(angle = (360*2 + 45), pos = "left")
+ # output <- fun_gg_just(angle = 45, pos = "bottom") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.x = ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust))
+ # output <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.y = ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)) + ggplot2::coord_flip()
+ # output1 <- fun_gg_just(angle = 90, pos = "bottom") ; output2 <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; ggplot2::ggplot() + ggplot2::geom_bar(data = obs1, mapping = ggplot2::aes(x = group, y = time), stat = "identity") + ggplot2::theme(axis.text.x = ggplot2::element_text(angle = output1$angle, hjust = output1$hjust, vjust = output1$vjust), axis.text.y = ggplot2::element_text(angle = output2$angle, hjust = output2$hjust, vjust = output2$vjust))
+ # output <- fun_gg_just(angle = -45, pos = "left") ; obs1 <- data.frame(time = 1, km = 1, bird = "pigeon", stringsAsFactors = FALSE) ; ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point() + ggplot2::geom_text(mapping = ggplot2::aes(label = bird), angle = output$angle, hjust = output$hjust, vjust = output$vjust)
+ # obs1 <- data.frame(time = 1:10, km = 1:10, bird = c(NA, NA, NA, "pigeon", NA, "cat", NA, NA, NA, NA), stringsAsFactors = FALSE) ; fun_open(width = 4, height = 4) ; for(i0 in c("text", "axis")){for(i1 in c("top", "right", "bottom", "left")){for(i2 in c(0, 45, 90, 135, 180, 225, 270, 315, 360)){output <- fun_gg_just(angle = i2, pos = i1, kind = i0) ; title <- paste0("kind: ", i0, " | pos: ", i1, " | angle = ", i2, " | hjust: ", output$hjust, " | vjust: ", output$vjust) ; if(i0 == "text"){print(ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point(color = fun_gg_palette(1), alpha = 0.5) + ggplot2::ggtitle(title) + ggplot2::geom_text(mapping = ggplot2::aes(label = bird), angle = output$angle, hjust = output$hjust, vjust = output$vjust) + ggplot2::theme(title = ggplot2::element_text(size = 5)))}else{print(ggplot2::ggplot(data = obs1, mapping = ggplot2::aes(x = time, y = km)) + ggplot2::geom_point(color = fun_gg_palette(1), alpha = 0.5) + ggplot2::ggtitle(title) + ggplot2::geom_text(mapping = ggplot2::aes(label = bird)) + ggplot2::scale_x_continuous(position = ifelse(i1 == "top", "top", "bottom")) + ggplot2::scale_y_continuous(position = ifelse(i1 == "right", "right", "left")) + ggplot2::theme(title = ggplot2::element_text(size = 5), axis.text.x = if(i1 %in% c("top", "bottom")){ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)}, axis.text.y = if(i1 %in% c("right", "left")){ggplot2::element_text(angle = output$angle, hjust = output$hjust, vjust = output$vjust)}))}}}} ; dev.off()
+ # DEBUGGING
+ # angle = 45 ; pos = "left" ; kind = "axis"
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument primary checking
+ # arg with no default values
+ mandat.args <- c(
+ "angle",
+ "pos"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # using fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = angle, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = pos, options = c("left", "top", "right", "bottom"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = kind, options = c("axis", "text"), length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <- c(
+ "angle",
+ "pos",
+ "kind"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # end second round of checking and data preparation
+ # main code
+ # to get angle between -360 and 360
+ while(angle > 360){
+ angle <- angle - 360
+ }
+ while(angle < -360){
+ angle <- angle + 360
+ }
+ # end to get angle between -360 and 360
+ # justifications
+ if(pos %in% c("bottom", "top")){
+ # code below is for if(pos == "bottom"){
+ if(any(sapply(FUN = all.equal, c(-360, -180, 0, 180, 360), angle) == TRUE)){ # equivalent of angle == -360 | angle == -180 | angle == 0 | angle == 180 | angle == 360 but deals with floats
+ hjust <- 0.5
+ if(kind == "text"){
+ if(any(sapply(FUN = all.equal, c(-360, 0, 360), angle) == TRUE)){
+ vjust <- 1
+ }else if(any(sapply(FUN = all.equal, c(-180, 180), angle) == TRUE)){
+ vjust <- 0
+ }
+ }else{
+ vjust <- 0.5
+ }
+ }else if(any(sapply(FUN = all.equal, c(-270, 90), angle) == TRUE)){
+ hjust <- 1
+ vjust <- 0.5
+ }else if(any(sapply(FUN = all.equal, c(-90, 270), angle) == TRUE)){
+ hjust <- 0
+ vjust <- 0.5
+ }else if((angle > -360 & angle < -270) | (angle > 0 & angle < 90)){
+ hjust <- 1
+ vjust <- 1
+ }else if((angle > -270 & angle < -180) | (angle > 90 & angle < 180)){
+ hjust <- 1
+ vjust <- 0
+ }else if((angle > -180 & angle < -90) | (angle > 180 & angle < 270)){
+ hjust <- 0
+ vjust <- 0
+ if(kind == "text" & pos == "top"){
+ hjust <- 1
+ }
+ }else if((angle > -90 & angle < 0) | (angle > 270 & angle < 360)){
+ hjust <- 0
+ vjust <- 1
+ }
+ if(pos == "top"){
+ if( ! ((angle > -180 & angle < -90) | (angle > 180 & angle < 270))){
+ hjust <- 1 - hjust
+ }
+ vjust <- 1 - vjust
+ }
+ }else if(pos %in% c("left", "right")){
+ # code below is for if(pos == "left"){
+ if(any(sapply(FUN = all.equal, c(-270, -90, 90, 270), angle) == TRUE)){ # equivalent of angle == -270 | angle == -90 | angle == 90 | angle == 270 but deals with floats
+ hjust <- 0.5
+ if(kind == "text"){
+ if(any(sapply(FUN = all.equal, c(-90, 90), angle) == TRUE)){
+ vjust <- 0
+ }else if(any(sapply(FUN = all.equal, c(-270, 270), angle) == TRUE)){
+ vjust <- 1
+ }
+ }else{
+ vjust <- 0.5
+ }
+ }else if(any(sapply(FUN = all.equal, c(-360, 0, 360), angle) == TRUE)){
+ hjust <- 1
+ vjust <- 0.5
+ }else if(any(sapply(FUN = all.equal, c(-180, 180), angle) == TRUE)){
+ hjust <- 0
+ vjust <- 0.5
+ }else if((angle > -360 & angle < -270) | (angle > 0 & angle < 90)){
+ hjust <- 1
+ vjust <- 0
+ }else if((angle > -270 & angle < -180) | (angle > 90 & angle < 180)){
+ hjust <- 0
+ vjust <- 0
+ }else if((angle > -180 & angle < -90) | (angle > 180 & angle < 270)){
+ hjust <- 0
+ vjust <- 1
+ }else if((angle > -90 & angle < 0) | (angle > 270 & angle < 360)){
+ hjust <- 1
+ vjust <- 1
+ }
+ if(pos == "right"){
+ hjust <- 1 - hjust
+ if( ! (((angle > -270 & angle < -180) | (angle > 90 & angle < 180)) | ((angle > -180 & angle < -90) | (angle > 180 & angle < 270)))){
+ vjust <- 1 - vjust
+ }
+ }
+ }
+ # end justifications
+ output <- list(angle = angle, pos = pos, kind = kind, hjust = hjust, vjust = vjust)
+ return(output)
}
-}else{
-line.count <- line.count + 1
-if(line.count== 1L){
-fin.lg.disp[[4]] <- legend.disp[[point.count + line.count]]
-lg.order[[4]] <- point.count + line.count
-lg.color[[4]] <- color[[i1]] # if color == NULL -> NULL
-lg.line.size[[4]] <- line.size[[i1]]
-lg.line.type[[4]] <- line.type[[i1]]
-if(plot == TRUE & fin.lg.disp[[4]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[4]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[4]] <- alpha[[i1]]
+
+
+######## fun_gg_get_legend() #### get the legend of ggplot objects
+
+
+
+
+
+fun_gg_get_legend <- function(ggplot_built, fun.name = NULL, lib.path = NULL){
+ # AIM
+ # get legend of ggplot objects
+ # # from https://stackoverflow.com/questions/12539348/ggplot-separate-legend-and-plot
+ # ARGUMENTS
+ # ggplot_built: a ggplot build object
+ # fun.name: single character string indicating the name of the function using fun_gg_get_legend() for warning and error messages. Ignored if NULL
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # RETURN
+ # a list of class c("gtable", "gTree", "grob", "gDesc"), providing legend information of ggplot_built objet, or NULL if the ggplot_built object has no legend
+ # REQUIRED PACKAGES
+ # ggplot2
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # EXAMPLES
+ # Simple example
+ # obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time, fill = group)) ; fun_gg_get_legend(ggplot_built = ggplot2::ggplot_build(p))
+ # Error message because no legend in the ggplot
+ # obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time)) ; fun_gg_get_legend(ggplot_built = ggplot2::ggplot_build(p))
+ # DEBUGGING
+ # obs1 <- data.frame(time = 1:20, group = rep(c("CLASS_1", "CLASS_2"), times = 10), stringsAsFactors = TRUE) ; p <- ggplot2::ggplot() + ggplot2::geom_point(data = obs1, mapping = ggplot2::aes(x = group, y = time)) ; ggplot_built = ggplot2::ggplot_build(p) ; fun.name = NULL ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check",
+ "fun_pack"
+ )
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED ", i1, "() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end required function checking
+ # argument primary checking
+ # arg with no default values
+ mandat.args <- c(
+ "ggplot_built"
+ )
+ tempo <- eval(parse(text = paste0("c(missing(", paste0(mandat.args, collapse = "), missing("), "))")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(sum(tempo, na.rm = TRUE) > 1, "S HAVE", " HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args[tempo], collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # using fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = ggplot_built, class = "ggplot_built", mode = "list", fun.name = function.name) ; eval(ee)
+ if( ! is.null(fun.name)){
+ tempo <- fun_check(data = fun.name, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(arg.check)){
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ }
+ # end using fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking
+ # management of NA
+ if(any(is.na(ggplot_built)) | any(is.na(fun.name)) | any(is.na(lib.path))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": NO ARGUMENT CAN HAVE NA VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA
+ # management of NULL
+ if(is.null(ggplot_built)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nggplot_built ARGUMENT CANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL
+ if( ! is.null(lib.path)){
+ if( ! all(dir.exists(lib.path))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end second round of checking
+ # package checking
+ fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
+ # end package checking
+ # main code
+ win.nb <- dev.cur()
+ pdf(file = NULL)
+ tmp <- ggplot2::ggplot_gtable(ggplot_built)
+ # BEWARE with ggplot_gtable : open a blanck device https://stackoverflow.com/questions/17012518/why-does-this-r-ggplot2-code-bring-up-a-blank-display-device
+ invisible(dev.off())
+ if(win.nb > 1){ # to go back to the previous active device, if == 1 means no opened device
+ dev.set(win.nb)
+ }
+ leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
+ if(length(leg) == 0L){
+ legend <- NULL
+ }else{
+ legend <- tmp$grobs[[leg]]
+ }
+ return(legend)
}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
-ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
-"(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
-x[[i1]],
-ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
-y[[i1]],
-if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
-", linetype = ",
-categ[[i1]],
-"), color = \"",
-color[[i1]][i5],
-"\", size = ",
-line.size[[i1]],
-ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
-ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
-", alpha = ",
-alpha[[i1]],
-", show.legend = ",
-ifelse(i5== 1L, TRUE, FALSE),
-")"
-)))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+
+
+######## fun_gg_point_rast() #### ggplot2 raster scatterplot layer
+
+
+
+
+
+fun_gg_point_rast <- function(
+ data = NULL,
+ mapping = NULL,
+ stat = "identity",
+ position = "identity",
+ ...,
+ na.rm = FALSE,
+ show.legend = NA,
+ inherit.aes = TRUE,
+ raster.width = NULL,
+ raster.height = NULL,
+ raster.dpi = 300,
+ inactivate = TRUE,
+ lib.path = NULL
+){
+ # AIM
+ # equivalent to ggplot2::geom_point() but in raster mode
+ # use it like ggplot2::geom_point() with the main raster.dpi additional argument
+ # WARNINGS
+ # can be long to generate the plot
+ # use a square plot region. Otherwise, the dots will have ellipsoid shape
+ # solve the transparency problems with some GUI
+ # this function is derived from the geom_point_rast() function, created by Viktor Petukhov , and present in the ggrastr package (https://rdrr.io/github/VPetukhov/ggrastr/src/R/geom-point-rast.R, MIT License, Copyright (c) 2017 Viktor Petukhov). Has been placed here to minimize package dependencies
+ # ARGUMENTS
+ # classical arguments of geom_point(), shown here https://rdrr.io/github/VPetukhov/ggrastr/man/geom_point_rast.html
+ # raster.width : width of the result image (in inches). Default: deterined by the current device parameters
+ # raster.height: height of the result image (in inches). Default: deterined by the current device parameters
+ # raster.dpi: resolution of the result image
+ # inactivate: logical. Inactivate the fun.name argument of the fun_check() function? If TRUE, the name of the fun_check() function in error messages coming from this function. Use TRUE if fun_gg_point_rast() is used like this: eval(parse(text = "fun_gg_point_rast"))
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # RETURN
+ # a raster scatter plot
+ # REQUIRED PACKAGES
+ # ggplot2
+ # grid (included in the R installation packages but not automatically loaded)
+ # Cairo
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # EXAMPLES
+ # Two pdf in the current directory
+ # set.seed(1) ; data1 = data.frame(x = rnorm(100000), y = rnorm(10000), stringsAsFactors = TRUE) ; fun_open(pdf.name = "Raster") ; ggplot2::ggplot() + fun_gg_point_rast(data = data1, mapping = ggplot2::aes(x = x, y = y)) ; fun_open(pdf.name = "Vectorial") ; ggplot2::ggplot() + ggplot2::geom_point(data = data1, mapping = ggplot2::aes(x = x, y = y)) ; dev.off() ; dev.off()
+ # DEBUGGING
+ #
+ # function name
+ if(all(inactivate == FALSE)){ # inactivate has to be used here but will be fully checked below
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ }else if(all(inactivate == TRUE)){
+ function.name <- NULL
+ }else{
+ tempo.cat <- paste0("ERROR IN fun_gg_point_rast(): CODE INCONSISTENCY 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if( ! is.null(data)){
+ tempo <- fun_check(data = data, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(mapping)){
+ tempo <- fun_check(data = mapping, class = "uneval", typeof = "list", fun.name = function.name) ; eval(ee) # aes() is tested
+ }
+ # stat and position not tested because too complicate
+ tempo <- fun_check(data = na.rm, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = show.legend, class = "vector", mode = "logical", length = 1, na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = inherit.aes, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(raster.width)){
+ tempo <- fun_check(data = raster.width, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(raster.height)){
+ tempo <- fun_check(data = raster.height, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = raster.dpi, class = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = inactivate, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # package checking
+ fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
+ fun_pack(req.package = c("grid"), lib.path = lib.path)
+ fun_pack(req.package = c("Cairo"), lib.path = lib.path)
+ # end package checking
+ # additional functions
+ DrawGeomPointRast <- function(data, panel_params, coord, na.rm = FALSE, raster.width = NULL, raster.height= NULL, raster.dpi = raster.dpi){
+ if (is.null(raster.width)){
+ raster.width <- par('fin')[1]
+ }
+ if (is.null(raster.height)){
+ raster.height <- par('fin')[2]
+ }
+ prev_dev_id <- dev.cur()
+ p <- ggplot2::GeomPoint$draw_panel(data, panel_params, coord)
+ dev_id <- Cairo::Cairo(type='raster', width = raster.width*raster.dpi, height = raster.height*raster.dpi, dpi = raster.dpi, units = 'px', bg = "transparent")[1]
+ grid::pushViewport(grid::viewport(width = 1, height = 1))
+ grid::grid.points(x = p$x, y = p$y, pch = p$pch, size = p$size,
+ name = p$name, gp = p$gp, vp = p$vp, draw = T)
+ grid::popViewport()
+ cap <- grid::grid.cap()
+ invisible(dev.off(dev_id))
+ invisible(dev.set(prev_dev_id))
+ grid::rasterGrob(cap, x = 0, y = 0, width = 1, height = 1, default.units = "native", just = c("left","bottom"))
+ }
+ # end additional functions
+ # main code
+ GeomPointRast <- ggplot2::ggproto("GeomPointRast", ggplot2::GeomPoint, draw_panel = DrawGeomPointRast)
+ ggplot2::layer(
+ data = data,
+ mapping = mapping,
+ stat = stat,
+ geom = GeomPointRast,
+ position = position,
+ show.legend = show.legend,
+ inherit.aes = inherit.aes,
+ params = list(
+ na.rm = na.rm,
+ raster.width = raster.width,
+ raster.height = raster.height,
+ raster.dpi = raster.dpi,
+ ...
+ )
+ )
+ # end main code
}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "linetype", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(line.type[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, with different values, breaks reorder the classes according to class.categ in the legend
+
+
+######## fun_gg_boxplot() #### ggplot2 boxplot + background dots if required
+
+
+
+
+######## fun_gg_scatter() #### ggplot2 scatterplot + lines (up to 6 overlays totally)
+
+
+
+
+######## fun_gg_heatmap() #### ggplot2 heatmap + overlaid mask if required
+
+
+#test plot.margin = margin(up.space.mds, right.space.mds, down.space.mds, left.space.mds, "inches") to set the dim of the region plot ?
+# if matrix is full of zero (or same value I guess), heatmap is complicate. Test it and error message
+
+fun_gg_heatmap <- function(
+ data1,
+ legend.name1 = "",
+ low.color1 = "blue",
+ mid.color1 = "white",
+ high.color1 = "red",
+ limit1 = NULL,
+ midpoint1 = NULL,
+ data2 = NULL,
+ color2 = "black",
+ alpha2 = 0.5,
+ invert2 = FALSE,
+ text.size = 12,
+ title = "",
+ title.text.size = 12,
+ show.scale = TRUE,
+ rotate = FALSE,
+ return = FALSE,
+ plot = TRUE,
+ add = NULL,
+ warn.print = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # ggplot2 heatmap with the possibility to overlay a mask
+ # see also:
+ # draw : http://www.sthda.com/english/wiki/ggplot2-quick-correlation-matrix-heatmap-r-software-and-data-visualization
+ # same range scale : https://stackoverflow.com/questions/44655723/r-ggplot2-heatmap-fixed-scale-color-between-graphs
+ # for ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
+ # ARGUMENTS
+ # data1: numeric matrix or data frame resulting from the conversion of the numeric matrix by reshape2::melt()
+ # legend.name1: character string of the data1 heatmap scale legend
+ # low.color1: character string of the color (i.e., "blue" or "#0000FF") of the lowest scale value
+ # mid.color1: same as low.color1 but for the middle scale value. If NULL, the middle color is the default color between low.color1 and high.color1. BEWARE: argument midpoint1 is not ignored, even if mid.color1 is NULL, meaning that the default mid color can still be controled
+ # high.color1: same as low.color1 but for the highest scale value
+ # limit1: 2 numeric values defining the lowest and higest color scale values. If NULL, take the range of data1 values
+ # midpoint1: single numeric value defining the value corresponding to the mid.color1 argument. A warning message is returned if midpoint1 does not correspond to the mean of limit1 values, because the color scale is not linear anymore. If NULL, takes the mean of limit1 values. Mean of data1, instead of mean of limit1, can be used here if required
+ # data2: binary mask matrix (made of 0 and 1) of same dimension as data1 or a data frame resulting from the conversion of the binary mask matrix by reshape2::melt(). Value 1 of data2 will correspond to color2 argument (value 0 will be NA color), and the opposite if invert2 argument is TRUE (inverted mask)
+ # color2: color of the 1 values of the binary mask matrix. The 0 values will be color NA
+ # alpha2: numeric value (from 0 to 1) of the mask transparency
+ # invert2: logical. Invert the mask (1 -> 0 and 0 -> 1)?
+ # text.size: numeric value of the size of the texts in scale
+ # title: character string of the graph title
+ # title.text.size: numeric value of the title size (in points)
+ # show.scale: logical. Show color scale?
+ # rotate: logical. Rotate the heatmap 90° clockwise?
+ # return: logical. Return the graph parameters?
+ # plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
+ # add: character string allowing to add more ggplot2 features (dots, lines, themes, etc.). BEWARE: (1) must start with "+" just after the simple or double opening quote (no space, end of line, carriage return, etc., allowed), (2) must finish with ")" just before the simple or double closing quote (no space, end of line, carriage return, etc., allowed) and (3) each function must be preceded by "ggplot2::" (for instance: "ggplot2::coord_flip()). If the character string contains the "ggplot2::theme" string, then internal ggplot2 theme() and theme_classic() functions will be inactivated to be reused by add. BEWARE: handle this argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
+ # warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
+ # lib.path: absolute path of the required packages, if not in the default folders
+ # RETURN
+ # a heatmap if plot argument is TRUE
+ # a list of the graph info if return argument is TRUE:
+ # $data: a list of the graphic info
+ # $axes: a list of the axes info
+ # $scale: the scale info (lowest, mid and highest values)
+ # $warn: the warning messages. Use cat() for proper display. NULL if no warning
+ # REQUIRED PACKAGES
+ # ggplot2
+ # reshape2
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # fun_round()
+ # EXAMPLES
+ # fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), title = "GRAPH 1")
+ # fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), return = TRUE)
+ # fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), legend.name1 = "VALUE", title = "GRAPH 1", text.size = 5, data2 = matrix(rep(c(1,0,0,0), 4), ncol = 4), invert2 = FALSE, return = TRUE)
+ # diagonal matrix
+ # fun_gg_heatmap(data1 = matrix(c(1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1), ncol = 4))
+ # fun_gg_heatmap(data1 = reshape2::melt(matrix(c(1,0,0,0,0,1,0,0,0,0,1,0,0,0,0,1), ncol = 4)))
+ # error message
+ # fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), data2 = matrix(rep(c(1,0,0,0), 5), ncol = 5))
+ # fun_gg_heatmap(data1 = matrix(1:16, ncol = 4), data2 = reshape2::melt(matrix(rep(c(1,0,0,0), 4), ncol = 4)))
+ # fun_gg_heatmap(data1 = reshape2::melt(matrix(1:16, ncol = 4)), data2 = reshape2::melt(matrix(rep(c(1,0,0,0), 4), ncol = 4)))
+ # DEBUGGING
+ # data1 = matrix(1:16, ncol = 4) ; legend.name1 = "" ; low.color1 = "blue" ; mid.color1 = "white" ; high.color1 = "red" ; limit1 = NULL ; midpoint1 = NULL ; data2 = matrix(rep(c(1,0,0,0), 4), ncol = 4) ; color2 = "black" ; alpha2 = 0.5 ; invert2 = FALSE ; text.size = 12 ; title = "" ; title.text.size = 12 ; show.scale = TRUE ; rotate = FALSE ; return = FALSE ; plot = TRUE ; add = NULL ; warn.print = TRUE ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_round", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_round() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # no reserved words required for this function
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if(all(is.matrix(data1))){
+ tempo <- fun_check(data = data1, class = "matrix", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }else if(all(is.data.frame(data1))){
+ tempo <- fun_check(data = data1, class = "data.frame", length = 3, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ # structure of reshape2::melt() data frame
+ tempo <- fun_check(data = data1[, 1], data.name = "COLUMN 1 OF data1 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = data1[, 2], data.name = "COLUMN 2 OF data1 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = data1[, 3], data.name = "COLUMN 3 OF data1 (reshape2::melt() DATA FRAME)", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data1 ARGUMENT MUST BE A NUMERIC MATRIX OR A DATA FRAME OUTPUT OF THE reshape::melt() FUNCTION")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = legend.name1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = low.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (all(low.color1 %in% colors() | grepl(pattern = "^#", low.color1)))){ # check that all strings of low.color1 start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": low.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(mid.color1)){
+ tempo <- fun_check(data = mid.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (all(mid.color1 %in% colors() | grepl(pattern = "^#", mid.color1)))){ # check that all strings of mid.color1 start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": mid.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = high.color1, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (all(high.color1 %in% colors() | grepl(pattern = "^#", high.color1)))){ # check that all strings of high.color1 start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": high.color1 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(limit1)){
+ tempo <- fun_check(data = limit1, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & any(limit1 %in% c(Inf, -Inf))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": limit1 ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ if( ! is.null(midpoint1)){
+ tempo <- fun_check(data = midpoint1, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(data2)){
+ if(all(is.matrix(data2))){
+ tempo <- fun_check(data = data2, class = "matrix", mode = "numeric", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! all(unique(data2) %in% c(0,1))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX IN data2 MUST BE MADE OF 0 AND 1 ONLY (MASK MATRIX)")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & all(is.matrix(data1)) & ! identical(dim(data1), dim(data2))){ # matrix and matrix
+ tempo.cat <- paste0("ERROR IN ", function.name, ": MATRIX DIMENSION IN data2 MUST BE IDENTICAL AS MATRIX DIMENSION IN data1. HERE IT IS RESPECTIVELY:\n", paste(dim(data2), collapse = " "), "\n", paste(dim(data1), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & all(is.data.frame(data1)) & nrow(data1) != prod(dim(data2))){ # reshape2 and matrix
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME IN data2 MUST HAVE ROW NUMBER EQUAL TO PRODUCT OF DIMENSIONS OF data1 MATRIX. HERE IT IS RESPECTIVELY:\n", paste(nrow(data1), collapse = " "), "\n", paste(prod(dim(data2)), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else if(all(is.data.frame(data2))){
+ tempo <- fun_check(data = data2, class = "data.frame", length = 3, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ # structure of reshape2::melt() data frame
+ tempo <- fun_check(data = data2[, 1], data.name = "COLUMN 1 OF data2 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = data2[, 2], data.name = "COLUMN 2 OF data2 (reshape2::melt() DATA FRAME)", typeof = "integer", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = data2[, 3], data.name = "COLUMN 3 OF data2 (reshape2::melt() DATA FRAME)", mode = "numeric", fun.name = function.name) ; eval(ee)
+ }
+ if(tempo$problem == FALSE & ! all(unique(data2[, 3]) %in% c(0,1))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THIRD COLUMN OF DATA FRAME IN data2 MUST BE MADE OF 0 AND 1 ONLY (MASK DATA FRAME)")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & all(is.data.frame(data1)) & ! identical(dim(data1), dim(data2))){ # data frame and data frame
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME DIMENSION IN data2 MUST BE IDENTICAL TO DATA FRAME DIMENSION IN data1. HERE IT IS RESPECTIVELY:\n", paste(dim(data2), collapse = " "), "\n", paste(dim(data1), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & all(is.matrix(data1)) & nrow(data2) != prod(dim(data1))){ # reshape2 and matrix
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DATA FRAME IN data2 MUST HAVE ROW NUMBER EQUAL TO PRODUCT OF DIMENSION OF data1 MATRIX. HERE IT IS RESPECTIVELY:\n", paste(nrow(data2), collapse = " "), "\n", paste(prod(dim(data1)), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE data2 ARGUMENT MUST BE A NUMERIC MATRIX OR A DATA FRAME OUTPUT OF THE reshape::melt() FUNCTION")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = color2, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (all(color2 %in% colors() | grepl(pattern = "^#", color2)))){ # check that all strings of color2 start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": color2 ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors()")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = alpha2, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = invert2, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = show.scale, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(add)){
+ tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! grepl(pattern = "^\\+", add)){ # check that the add string start by +
+ tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & ! grepl(pattern = "ggplot2::", add)){ #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST CONTAIN \"ggplot2::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & ! grepl(pattern = ")$", add)){ # check that the add string finished by )
+ tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # package checking
+ fun_pack(req.package = c("reshape2", "ggplot2"), lib.path = lib.path)
+ # end package checking
+ # main code
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ if(all(is.matrix(data1))){
+ data1 <- reshape2::melt(data1) # transform a matrix into a data frame with 2 coordinates columns and the third intensity column
+ }
+ if(rotate == TRUE){
+ data1[, 1] <- rev(data1[, 1])
+ }
+ if(is.null(limit1)){
+ if(any(data1[, 3] %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE THIRD COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ limit1 <- range(data1[, 3], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE limit1 ARGUMENT IS NULL -> RANGE OF data1 ARGUMENT HAS BEEN TAKEN: ", paste(fun_round(limit1), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ if(suppressWarnings(any(limit1 %in% c(Inf, -Inf)))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED LIMIT CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(is.null(midpoint1)){
+ midpoint1 <- mean(limit1, na.rm = TRUE)
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE midpoint1 ARGUMENT IS NULL -> MEAN OF limit1 ARGUMENT HAS BEEN TAKEN: ", paste(fun_round(midpoint1), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else if(fun_round(midpoint1, 9) != fun_round(mean(limit1), 9)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE midpoint1 ARGUMENT (", fun_round(mean(midpoint1), 9), ") DOES NOT CORRESPOND TO THE MEAN OF THE limit1 ARGUMENT (", fun_round(mean(limit1), 9), "). COLOR SCALE IS NOT LINEAR")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data2)){
+ if(all(is.matrix(data2))){
+ data2 <- reshape2::melt(data2) # transform a matrix into a data frame with 2 coordinates columns and the third intensity column
+ }
+ if(rotate == TRUE){
+ data2[, 1] <- rev(data2[, 1])
+ }
+ data2[, 3] <- factor(data2[, 3]) # to converte continuous scale into discrete scale
+ }
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0 # to facilitate debugging
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggplot())
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data1, mapping = ggplot2::aes_string(x = names(data1)[ifelse(rotate == FALSE, 2, 1)], y = names(data1)[ifelse(rotate == FALSE, 1, 2)], fill = names(data1)[3]), show.legend = show.scale)) # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_fill_gradient2(low = low.color1, high = high.color1, mid = mid.color1, midpoint = midpoint1, limit = limit1, breaks = c(limit1[1], midpoint1, limit1[2]), labels = fun_round(c(limit1[1], midpoint1, limit1[2])), name = legend.name1))
+ if( ! is.null(data2)){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data2, mapping = ggplot2::aes_string(x = names(data2)[ifelse(rotate == FALSE, 2, 1)], y = names(data2)[ifelse(rotate == FALSE, 1, 2)], alpha = names(data2)[3]), fill = color2, show.legend = FALSE))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", values = if(invert2 == FALSE){c(0, alpha2)}else{c(alpha2, 0)}, guide = FALSE))
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_raster(data = data2, mapping = ggplot2::aes_string(x = names(data2)[ifelse(rotate == FALSE, 2, 1)], y = names(data2)[ifelse(rotate == FALSE, 1, 2)], group = names(data2)[3]), fill = data2[, 3], alpha = alpha2, show.legend = FALSE)) # BEWARE: this does not work if NA present, because geom_raster() has a tendency to complete empty spaces, and thus, behave differently than geom_tile(). See https://github.com/tidyverse/ggplot2/issues/3025
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_fixed()) # x = y
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_y_reverse())
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
+ add.check <- TRUE
+ if( ! is.null(add)){ # if add is NULL, then = 0
+ if(grepl(pattern = "ggplot2::theme", add) == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT -> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ add.check <- FALSE
+ }
+ }
+ if(add.check == TRUE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_blank(),
+ axis.title = ggplot2::element_blank(),
+ axis.text = ggplot2::element_blank(),
+ axis.ticks = ggplot2::element_blank(),
+ panel.background = ggplot2::element_blank()
+ ))
+ }
+ if(plot == TRUE){
+ # suppressWarnings(
+ print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))
+ # )
+ }else{
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ if(return == TRUE){
+ output <- ggplot2::ggplot_build(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "))))
+ output <- output$data
+ names(output)[1] <- "heatmap"
+ if( ! is.null(data2)){
+ names(output)[2] <- "mask"
+ }
+ return(list(data = output, axes = output$layout$panel_params[[1]], scale = c(limit1[1], midpoint1, limit1[2]), warn = warn))
+ }
}
-if(line.count== 2L){
-fin.lg.disp[[5]] <- legend.disp[[point.count + line.count]]
-lg.order[[5]] <- point.count + line.count
-lg.color[[5]] <- color[[i1]] # if color == NULL -> NULL
-lg.line.size[[5]] <- line.size[[i1]]
-lg.line.type[[5]] <- line.type[[i1]]
-if(plot == TRUE & fin.lg.disp[[5]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[5]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[5]] <- alpha[[i1]]
+
+
+######## fun_gg_empty_graph() #### text to display for empty graphs
+
+
+
+
+
+fun_gg_empty_graph <- function(
+ text = NULL,
+ text.size = 12,
+ title = NULL,
+ title.size = 8,
+ lib.path = NULL
+){
+ # AIM
+ # display an empty ggplot2 plot with a text in the middle of the window (for instance to specify that no plot can be drawn)
+ # ARGUMENTS
+ # text: character string of the message to display
+ # text.size: numeric value of the text size (in points)
+ # title: character string of the graph title
+ # title.size: numeric value of the title size (in points)
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL
+ # RETURN
+ # an empty plot
+ # REQUIRED PACKAGES
+ # ggplot2
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # EXAMPLES
+ ### simple example
+ # fun_gg_empty_graph(text = "NO GRAPH")
+ ### white page
+ # fun_gg_empty_graph()
+ ### all the arguments
+ # fun_gg_empty_graph(text = "NO GRAPH", text.size = 8, title = "GRAPH1", title.size = 10, lib.path = NULL)
+ # DEBUGGING
+ # text = "NO GRAPH" ; text.size = 12 ; title = "GRAPH1" ; title.size = 8 ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if( ! is.null(text)){
+ tempo <- fun_check(data = text, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(title)){
+ tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = title.size, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # package checking
+ fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
+ # end package checking
+ # main code
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0
+ # no need loop part
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggplot())
+ if( ! is.null(text)){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(data = data.frame(x = 1, y = 1, stringsAsFactors = TRUE), ggplot2::aes(x = x, y = y, label = text), size = text.size))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_void())
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ plot.title = ggplot2::element_text(size = title.size) # stronger than text
+ ))
+ suppressWarnings(print(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ")))))
}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
-ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
-"(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
-x[[i1]],
-ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
-y[[i1]],
-if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
-", alpha = ",
-categ[[i1]],
-"), color = \"",
-color[[i1]][i5],
-"\", size = ",
-line.size[[i1]],
-", linetype = ",
-ifelse(is.numeric(line.type[[i1]]), "", "\""),
-line.type[[i1]],
-ifelse(is.numeric(line.type[[i1]]), "", "\""),
-ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
-ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
-", show.legend = FALSE)"
-)))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+
+
+################ Graphic extraction
+
+
+######## fun_trim() #### display values from a quantitative variable and trim according to defined cut-offs
+
+# Add name of the variable in the graph
+# not max and min for boxplot but 1.5IQR
+fun_trim <- function(
+ data,
+ displayed.nb = NULL,
+ single.value.display = FALSE,
+ trim.method = "",
+ trim.cutoffs = c(0.05,
+ 0.975),
+ interval.scale.disp = TRUE,
+ down.space = 0.75,
+ left.space = 0.75,
+ up.space = 0.3,
+ right.space = 0.25,
+ orient = 1,
+ dist.legend = 0.37,
+ box.type = "l",
+ amplif.label = 1.25,
+ amplif.axis = 1.25,
+ std.x.range = TRUE,
+ std.y.range = TRUE,
+ cex.pt = 0.2,
+ col.box = hsv(0.55,
+ 0.8,
+ 0.8),
+ x.nb.inter.tick = 4,
+ y.nb.inter.tick = 0,
+ tick.length = 1,
+ sec.tick.length = 0.75,
+ corner.text = "",
+ amplif.legend = 1,
+ corner.text.size = 0.75,
+ trim.return = FALSE
+){
+ # AIM
+ # trim and display values from a numeric vector or matrix
+ # plot 4 graphs: stripchart of values, stripchart of rank of values, histogram and normal QQPlot
+ # different kinds of intervals are displayed on the top of graphes to facilitate the analysis of the variable and a trimming setting
+ # the trimming interval chosen is displayed on top of graphs
+ # both trimmed and not trimmed values are returned in a list
+ # ARGUMENTS
+ # data: values to plot (either a numeric vector or a numeric matrix)
+ # displayed.nb: number of values displayed. If NULL, all the values are displayed. Otherwise, if the number of values is over displayed.nb, then displayed.nb values are displayed after random selection
+ # single.value.display: provide the 4 graphs if data is made of a single (potentially repeated value)? If FALSE, an empty graph is displayed if data is made of a single (potentially repeated value). And the return list is made of NULL compartments
+ # trim.method: Write "" if not required. write "mean.sd" if mean +/- sd has to be displayed as a trimming interval (only recommanded for normal distribution). Write "quantile" to display a trimming interval based on quantile cut-offs. No other possibility allowed. See trim.cutoffs below
+ # trim.cutoffs: 2 values cutoff for the trimming interval displayed, each value between 0 and 1. Not used if trim.method == "".The couple of values c(lower, upper) represents the lower and upper boundaries of the trimming interval (in proportion), which represent the interval of distribution kept (between 0 and 1). Example: trim.cutoffs = c(0.05, 0.975). What is strictly kept for the display is ]lower , upper[, boundaries excluded. Using the "mean.sd" method, 0.025 and 0.975 represent 95% CI which is mean +/- 1.96 * sd
+ # interval.scale.disp: display sd and quantiles intervals on top of graphs ?
+ # down.space: lower vertical margin (in inches, mai argument of par())
+ # left.space: left horizontal margin (in inches, mai argument of par())
+ # up.space: upper vertical margin between plot region and grapical window (in inches, mai argument of par())
+ # right.space: right horizontal margin (in inches, mai argument of par())
+ # orient: scale number orientation (las argument of par()). 0, always parallel to the axis; 1, always horizontal; 2, always perpendicular to the axis; 3, always vertical
+ # dist.legend: numeric value that moves axis legends away in inches (first number of mgp argument of par() but in inches thus / 0.2)
+ # box.type: bty argument of par(). Either "o", "l", "7", "c", "u", "]", the resulting box resembles the corresponding upper case letter. A value of "n" suppresses the box
+ # amplif.label: increase or decrease the size of the text in legends
+ # amplif.axis: increase or decrease the size of the scale numbers in axis
+ # std.x.range: standard range on the x-axis? TRUE (no range extend) or FALSE (4% range extend). Controls xaxs argument of par() (TRUE is xaxs = "i", FALSE is xaxs = "r")
+ # std.y.range: standard range on the y-axis? TRUE (no range extend) or FALSE (4% range extend). Controls yaxs argument of par() (TRUE is yaxs = "i", FALSE is yaxs = "r")
+ # cex.pt: size of points in stripcharts (in inches, thus cex.pt will be thereafter / 0.2)
+ # col.box: color of boxplot
+ # x.nb.inter.tick: number of secondary ticks between main ticks on x-axis (only if not log scale). Zero means non secondary ticks
+ # y.nb.inter.tick: number of secondary ticks between main ticks on y-axis (only if not log scale). Zero means non secondary ticks
+ # tick.length: length of the ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc. 0 means no tick
+ # sec.tick.length: length of the secondary ticks (1 means complete the distance between the plot region and the axis numbers, 0.5 means half the length, etc., 0 for no ticks)
+ # corner.text: text to add at the top right corner of the window
+ # amplif.legend: increase or decrease the size of the text of legend
+ # corner.text.size: positive numeric. Increase or decrease the size of the text. Value 1 does not change it, 0.5 decreases by half, 2 increases by 2
+ # trim.return: return the trimmed and non trimmed values? NULL returned for trimmed and non trimmed values if trim.method == ""
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # RETURN
+ # a list containing:
+ # $trim.method: correspond to trim.method above
+ # $trim.cutoffs: correspond to trim.cutoffs above
+ # $real.trim.cutoffs: the two boundary values (in the unit of the numeric vector or numeric matrix analyzed). NULL
+ # $trimmed.values: the values outside of the trimming interval as defined in trim.cutoffs above
+ # $kept.values: the values inside the trimming interval as defined in trim.cutoffs above
+ # EXAMPLES
+ # fun_trim(data = c(1:100, 1:10), displayed.nb = NULL, single.value.display = FALSE, trim.method = "mean.sd", trim.cutoffs = c(0.05, 0.975), interval.scale.disp = TRUE, down.space = 0.75, left.space = 0.75, up.space = 0.3, right.space = 0.25, orient = 1, dist.legend = 0.37, box.type = "l", amplif.label = 1.25, amplif.axis = 1.25, std.x.range = TRUE, std.y.range = TRUE, cex.pt = 0.2, col.box = hsv(0.55, 0.8, 0.8), x.nb.inter.tick = 4, y.nb.inter.tick = 0, tick.length = 0.5, sec.tick.length = 0.3, corner.text = "", amplif.legend = 1, corner.text.size = 0.75, trim.return = TRUE)
+ # DEBUGGING
+ # data = c(1:100, 1:10) ; displayed.nb = NULL ; single.value.display = FALSE ; trim.method = "quantile" ; trim.cutoffs = c(0.05, 0.975) ; interval.scale.disp = TRUE ; down.space = 1 ; left.space = 1 ; up.space = 0.5 ; right.space = 0.25 ; orient = 1 ; dist.legend = 0.5 ; box.type = "l" ; amplif.label = 1 ; amplif.axis = 1 ; std.x.range = TRUE ; std.y.range = TRUE ; cex.pt = 0.1 ; col.box = hsv(0.55, 0.8, 0.8) ; x.nb.inter.tick = 4 ; y.nb.inter.tick = 0 ; tick.length = 0.5 ; sec.tick.length = 0.3 ; corner.text = "" ; amplif.legend = 1 ; corner.text.size = 0.75 ; trim.return = TRUE # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ # argument checking without fun_check()
+ if( ! (all(class(data) == "numeric") | all(class(data) == "integer") | (all(class(data) %in% c("matrix", "array")) & base::mode(data) == "numeric"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data ARGUMENT MUST BE A NUMERIC VECTOR OR NUMERIC MATRIX")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking without fun_check()
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ if( ! is.null(displayed.nb)){
+ tempo <- fun_check(data = displayed.nb, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(displayed.nb < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": displayed.nb ARGUMENT MUST BE A SINGLE INTEGER VALUE GREATER THAN 1 AND NOT: ", paste(displayed.nb, collapse = " "))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = single.value.display, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = trim.method, options = c("", "mean.sd", "quantile"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = trim.cutoffs, class = "vector", mode = "numeric", length = 2, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = interval.scale.disp, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = down.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = left.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = up.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = right.space, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = orient, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dist.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.type, options = c("o", "l", "7", "c", "u", "]", "n"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = amplif.label, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = amplif.axis, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = std.x.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = std.y.range, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = cex.pt, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = col.box, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.nb.inter.tick, class = "integer", length = 1, neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.nb.inter.tick, class = "integer", length = 1, neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = sec.tick.length, class = "vector", mode = "numeric", length = 1, prop = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = corner.text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = amplif.legend, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = corner.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = trim.return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ if(all(is.na(data) | ! is.finite(data))){
+ tempo.cat <- paste0("ERROR IN fun_trim FUNCTION\ndata ARGUMENT CONTAINS ONLY NA OR Inf")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end argument checking
+ # main code
+ if(all(class(data)%in% c("matrix", "array"))){
+ data <- as.vector(data)
+ }
+ na.nb <- NULL
+ if(any(is.na(data))){
+ na.nb <- sum(c(is.na(data)))
+ data <- data[ ! is.na(data)]
+ }
+ color.cut <- hsv(0.75, 1, 1) # color of interval selected
+ col.mean <- hsv(0.25, 1, 0.8) # color of interval using mean+/-sd
+ col.quantile <- "orange" # color of interval using quantiles
+ quantiles.selection <- c(0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 0.75, 0.9, 0.95, 0.975, 0.99) # quantiles used in axis to help for choosing trimming cutoffs
+ if(single.value.display == FALSE & length(unique(data)) == 1L){
+ par(bty = "n", xaxt = "n", yaxt = "n", xpd = TRUE)
+ plot(1, pch = 16, col = "white", xlab = "", ylab = "")
+ text(x = 1, y = 1, paste0("No graphic displayed\nBecause data made of a single different value (", formatC(as.double(table(data))), ")"), cex = 2)
+ output <- list(trim.method = NULL, trim.cutoffs = NULL, real.trim.cutoffs = NULL, trimmed.values = NULL, kept.values = NULL)
+ }else{
+ output <- list(trim.method = trim.method, trim.cutoffs = trim.cutoffs, real.trim.cutoffs = NULL, trimmed.values = NULL, kept.values = NULL)
+ fun.rug <- function(sec.tick.length.f = sec.tick.length, x.nb.inter.tick.f = x.nb.inter.tick, y.nb.inter.tick.f = y.nb.inter.tick){
+ if(x.nb.inter.tick.f > 0){
+ inter.tick.unit <- (par("xaxp")[2] - par("xaxp")[1]) / par("xaxp")[3]
+ par.ini <- par()[c("xpd", "tcl")]
+ par(xpd = FALSE)
+ par(tcl = -par()$mgp[2] * sec.tick.length.f) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
+ suppressWarnings(rug(seq(par("xaxp")[1] - 10 * inter.tick.unit, par("xaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + x.nb.inter.tick.f)), ticksize = NA, side = 1)) # ticksize = NA to allow the use of par()$tcl value
+ par(par.ini)
+ rm(par.ini)
+ }
+ if(y.nb.inter.tick.f > 0){
+ inter.tick.unit <- (par("yaxp")[2] - par("yaxp")[1]) / par("yaxp")[3]
+ par.ini <- par()[c("xpd", "tcl")]
+ par(xpd = FALSE)
+ par(tcl = -par()$mgp[2] * sec.tick.length.f) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
+ suppressWarnings(rug(seq(par("yaxp")[1] - 10 * inter.tick.unit, par("yaxp")[2] + 10 * inter.tick.unit, by = inter.tick.unit / (1 + y.nb.inter.tick.f)), ticksize = NA, side = 2)) # ticksize = NA to allow the use of par()$tcl value
+ par(par.ini)
+ rm(par.ini)
+ }
+ }
+ fun.add.cut <- function(data.f, trim.method.f = trim.method, trim.cutoffs.f = trim.cutoffs, color.cut.f = color.cut, return.f = FALSE){
+ # DEBUGGING
+ # data.f = data ; trim.method.f = "mean.sd"; trim.cutoffs.f = trim.cutoffs ; color.cut.f = color.cut ; return.f = TRUE
+ real.trim.cutoffs.f <- NULL
+ if(trim.method.f != ""){
+ data.f <- sort(data.f)
+ par.ini <- par()$xpd
+ par(xpd = FALSE)
+ if(trim.method.f == "mean.sd"){
+ real.trim.cutoffs.f <- qnorm(trim.cutoffs.f, mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE))
+ abline(v = qnorm(trim.cutoffs.f, mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), col = color.cut.f)
+ segments(qnorm(trim.cutoffs.f[1], mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), par()$usr[4] * 0.75, qnorm(trim.cutoffs.f[2], mean(data.f, na.rm = TRUE), sd(data.f, na.rm = TRUE)), par()$usr[4] * 0.75, col = color.cut.f)
+ }
+ if(trim.method.f == "quantile"){
+ real.trim.cutoffs.f <- quantile(data.f, probs = trim.cutoffs.f, type = 7, na.rm = TRUE)
+ abline(v = quantile(data.f, probs = trim.cutoffs.f, type = 7, na.rm = TRUE), col = color.cut.f)
+ segments(quantile(data.f, probs = trim.cutoffs.f[1], type = 7, na.rm = TRUE), par()$usr[4] * 0.75, quantile(data.f, probs = trim.cutoffs.f[2], type = 7, na.rm = TRUE), par()$usr[4] * 0.75, col = color.cut.f)
+ }
+ par(par.ini)
+ if(return.f == TRUE){
+ trimmed.values.f <- data.f[data.f <= real.trim.cutoffs.f[1] | data.f >= real.trim.cutoffs.f[2]]
+ kept.values.f <- data.f[data.f > real.trim.cutoffs.f[1] & data.f < real.trim.cutoffs.f[2]]
+ }
+ }else{
+ real.trim.cutoffs.f <- NULL
+ trimmed.values.f <- NULL
+ kept.values.f <- NULL
+ }
+ if(return.f == TRUE){
+ output <- list(trim.method = trim.method.f, trim.cutoffs = trim.cutoffs.f, real.trim.cutoffs = real.trim.cutoffs.f, trimmed.values = trimmed.values.f, kept.values = kept.values.f)
+ return(output)
+ }
+ }
+ fun.interval.scale.display <- function(data.f, col.quantile.f = col.quantile, quantiles.selection.f = quantiles.selection, col.mean.f = col.mean){ # intervals on top of graphs
+ par.ini <- par()[c("mgp", "xpd")]
+ par(mgp = c(0.25, 0.25, 0), xpd = NA)
+ axis(side = 3, at = c(par()$usr[1], par()$usr[2]), labels = rep("", 2), col = col.quantile.f, lwd.ticks = 0)
+ par(xpd = FALSE)
+ axis(side = 3, at = quantile(as.vector(data.f), probs = quantiles.selection.f, type = 7, na.rm = TRUE), labels = quantiles.selection.f, col.axis = col.quantile.f, col = col.quantile.f)
+ par(mgp = c(1.75, 1.75, 1.5), xpd = NA)
+ axis(side = 3, at = c(par()$usr[1], par()$usr[2]), labels = rep("", 2), col = col.mean.f, lwd.ticks = 0)
+ par(xpd = FALSE)
+ axis(side = 3, at = m + s * qnorm(quantiles.selection.f), labels = formatC(round(qnorm(quantiles.selection.f), 2)), col.axis = col.mean.f, col = col.mean.f, lwd.ticks = 1)
+ par(par.ini)
+ }
+ zone<-matrix(1:4, ncol=2)
+ layout(zone)
+ par(omi = c(0, 0, 1.5, 0), mai = c(down.space, left.space, up.space, right.space), las = orient, mgp = c(dist.legend / 0.2, 0.5, 0), xpd = FALSE, bty= box.type, cex.lab = amplif.label, cex.axis = amplif.axis, xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
+ par(tcl = -par()$mgp[2] * tick.length) # tcl gives the length of the ticks as proportion of line text, knowing that mgp is in text lines. So the main ticks are a 0.5 of the distance of the axis numbers by default. The sign provides the side of the tick (negative for outside of the plot region)
+ if(is.null(displayed.nb)){
+ sampled.data <- as.vector(data)
+ if(corner.text == ""){
+ corner.text <- paste0("ALL VALUES OF THE DATASET DISPLAYED")
+ }else{
+ corner.text <- paste0(corner.text, "\nALL VALUES OF THE DATASET DISPLAYED")
+ }
+ }else{
+ if(length(as.vector(data)) > displayed.nb){
+ sampled.data <- sample(as.vector(data), displayed.nb, replace = FALSE)
+ if(corner.text == ""){
+ corner.text <- paste0("WARNING: ONLY ", displayed.nb, " VALUES ARE DISPLAYED AMONG THE ", length(as.vector(data)), " VALUES OF THE DATASET ANALYZED")
+ }else{
+ corner.text <- paste0(corner.text, "\nWARNING: ONLY ", displayed.nb, " VALUES ARE DISPLAYED AMONG THE ", length(as.vector(data)), " VALUES OF THE DATASET ANALYZED")
+ }
+ }else{
+ sampled.data <- as.vector(data)
+ if(corner.text == ""){
+ corner.text <- paste0("WARNING: THE DISPLAYED NUMBER OF VALUES PARAMETER ", deparse(substitute(displayed.nb)), " HAS BEEN SET TO ", displayed.nb, " WHICH IS ABOVE THE NUMBER OF VALUES OF THE DATASET ANALYZED -> ALL VALUES DISPLAYED")
+ }else{
+ corner.text <- paste0(corner.text, "\nWARNING: THE DISPLAYED NUMBER OF VALUES PARAMETER ", deparse(substitute(displayed.nb)), " HAS BEEN SET TO ", displayed.nb, " WHICH IS ABOVE THE NUMBER OF VALUES OF THE DATASET ANALYZED -> ALL VALUES DISPLAYED")
+ }
+ }
+ }
+ if( ! is.null(na.nb)){
+ if(corner.text == ""){
+ corner.text <- paste0("WARNING: NUMBER OF NA REMOVED IS ", na.nb)
+ }else{
+ corner.text <- paste0("WARNING: NUMBER OF NA REMOVED IS ", na.nb)
+ }
+ }
+ stripchart(sampled.data, method="jitter", jitter=0.4, vertical=FALSE, ylim=c(0.5, 1.5), group.names = "", xlab = "Value", ylab="", pch=1, cex = cex.pt / 0.2)
+ fun.rug(y.nb.inter.tick.f = 0)
+ boxplot(as.vector(data), horizontal=TRUE, add=TRUE, boxwex = 0.4, staplecol = col.box, whiskcol = col.box, medcol = col.box, boxcol = col.box, range = 0, whisklty = 1)
+ m <- mean(as.vector(data), na.rm = TRUE)
+ s <- sd(as.vector(data), na.rm = TRUE)
+ segments(m, 0.8, m, 1, lwd=2, col="red") # mean
+ segments(m -1.96 * s, 0.9, m + 1.96 * s, 0.9, lwd=1, col="red") # mean
+ graph.xlim <- par()$usr[1:2] # for hist() and qqnorm() below
+ if(interval.scale.disp == TRUE){
+ fun.interval.scale.display(data.f = data)
+ if(corner.text == ""){
+ corner.text <- paste0("MULTIPLYING FACTOR DISPLAYED (MEAN +/- SD) ON SCALES: ", paste(formatC(round(qnorm(quantiles.selection), 2))[-(1:(length(quantiles.selection) - 1) / 2)], collapse = ", "), "\nQUANTILES DISPLAYED ON SCALES: ", paste(quantiles.selection, collapse = ", "))
+ }else{
+ corner.text <- paste0(corner.text, "\nMULTIPLYING FACTOR DISPLAYED (MEAN +/- SD) ON SCALES: ", paste(formatC(round(qnorm(quantiles.selection), 2))[-(1:(length(quantiles.selection) - 1) / 2)], collapse = ", "), "\nQUANTILES DISPLAYED ON SCALES: ", paste(quantiles.selection, collapse = ", "))
+ }
+ }
+ output.tempo <- fun.add.cut(data.f = data, return.f = TRUE) # to recover real.trim.cutoffs
+ if(trim.return == TRUE){
+ output <- output.tempo
+ }
+ par(xpd = NA)
+ if(trim.method != ""){
+ if(corner.text == ""){
+ corner.text <- paste0("SELECTED CUT-OFFS (PROPORTION): ", paste(trim.cutoffs, collapse = ", "), "\nSELECTED CUT-OFFS: ", paste(output.tempo$real.trim.cutoffs, collapse = ", "))
+ }else{
+ corner.text <- paste0(corner.text, "\nSELECTED CUT-OFFS (PROPORTION): ", paste(trim.cutoffs, collapse = ", "), "\nSELECTED CUT-OFFS: ", paste(output.tempo$real.trim.cutoffs, collapse = ", "))
+ }
+ if(interval.scale.disp == TRUE){
+ legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- 1.96sd", paste0("Trimming interval: ", paste0(trim.cutoffs, collapse = " , ")), "Mean +/- sd multiplying factor", "Quantile"), yjust = 0, lty=1, col=c(col.box, "red", color.cut, col.mean, col.quantile), bty="n", cex = amplif.legend)
+ }else{
+ legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- 1.96sd", paste0("Trimming interval: ", paste0(trim.cutoffs, collapse = " , "))), yjust = 0, lty=1, col=c(col.box, "red", color.cut), bty="n", cex = amplif.legend, y.intersp=1.25)
+ }
+ }else{
+ if(interval.scale.disp == TRUE){
+ legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- sd", "Mean +/- sd multiplying factor", "Quantile"), yjust = 0, lty=1, col=c(col.box, "red", col.mean, col.quantile), bty="n", cex = amplif.legend)
+ }else{
+ legend(x = (par("usr")[1] - ((par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1])) * par("plt")[1] - ((par("usr")[2] - par("usr")[1]) / (par("omd")[2] - par("omd")[1])) * par("omd")[1]), y = (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / (par("omd")[4] - par("omd")[3])) * (1 - par("omd")[4]) / 2), legend = c(c("min, Q1, Median, Q3, max"), "mean +/- sd"), yjust = 0, lty=1, col=c(col.box, "red"), bty="n", cex = amplif.legend, y.intersp=1.25)
+ }
+ }
+ par(xpd = FALSE, xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
+ hist(as.vector(data), main = "", xlim = graph.xlim, xlab = "Value", ylab="Density", col = grey(0.25)) # removed: breaks = seq(min(as.vector(data), na.rm = TRUE), max(as.vector(data), na.rm = TRUE), length.out = length(as.vector(data)) / 10)
+ abline(h = par()$usr[3])
+ fun.rug()
+ if(interval.scale.disp == TRUE){
+ fun.interval.scale.display(data.f = data)
+ }
+ fun.add.cut(data.f = data)
+ par(xaxs = ifelse(std.x.range, "i", "r"))
+ stripchart(rank(sampled.data), method="stack", vertical=FALSE, ylim=c(0.99, 1.3), group.names = "", xlab = "Rank of values", ylab="", pch=1, cex = cex.pt / 0.2)
+ fun.rug(y.nb.inter.tick.f = 0)
+ x.text <- par("usr")[2] + (par("usr")[2] - par("usr")[1]) / (par("plt")[2] - par("plt")[1]) * (1 - par("plt")[2]) / 2
+ y.text <- (par("usr")[4] + ((par("usr")[4] - par("usr")[3]) / (par("plt")[4] - par("plt")[3])) * (1 - par("plt")[4]) + ((par("usr")[4] - par("usr")[3]) / ((par()$omd[4] / 2) * ((par("plt")[4] - par("plt")[3])))) * (1 - par("omd")[4])) # BEWARE. Here in "(par()$omd[4] / 2", division by two because there are 2 graphs staked on the y axis, and not one
+ par(xpd=NA)
+ text(x = x.text, y = y.text, paste0(corner.text), adj=c(1, 1.1), cex = corner.text.size) # text at the topright corner
+ par(xpd=FALSE)
+ par(xaxs = ifelse(std.x.range, "i", "r"), yaxs = ifelse(std.y.range, "i", "r"))
+ qqnorm(as.vector(sampled.data), main = "", datax = TRUE, ylab = "Value", pch = 1, col = "red", cex = cex.pt / 0.2)
+ fun.rug()
+ if(diff(quantile(as.vector(data), probs = c(0.25, 0.75), na.rm = TRUE)) != 0){ # otherwise, error generated
+ qqline(as.vector(data), datax = TRUE)
+ }
+ if(interval.scale.disp == TRUE){
+ fun.interval.scale.display(data.f = data)
+ }
+ fun.add.cut(data.f = data)
+ }
+ if(trim.return == TRUE){
+ return(output)
+ }
}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(alpha[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, with different values, breaks reorder the classes according to class.categ in the legend
+
+
+######## fun_segmentation() #### segment a dot cloud on a scatterplot and define the dots from another cloud outside the segmentation
+
+
+fun_segmentation <- function(
+ data1,
+ x1,
+ y1,
+ x.range.split = NULL,
+ x.step.factor = 10,
+ y.range.split = NULL,
+ y.step.factor = 10,
+ error = 0,
+ data2 = NULL,
+ x2,
+ y2,
+ data2.pb.dot = "unknown",
+ xy.cross.kind = "&",
+ plot = FALSE,
+ graph.in.file = FALSE,
+ raster = TRUE,
+ warn.print = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # if data1 is a data frame corresponding to the data set of a scatterplot (with a x column for x-axis values and a y column for the y-axis column), then fun_segmentation() delimits a frame around the dots cloud using a sliding window set by x.range.split and x.step.factor to frame the top and bottom part of the cloud, and set by y.range.split and y.step.factor to frame the left and right part of the cloud
+ # if a second data frame is provided, corresponding to the data set of a scatterplot (with a x column for x-axis values and a y column for the y-axis column), then fun_segmentation() defines the dots of this data frame, outside of the frame of the first data frame
+ # WARNINGS
+ # if dots from data2 look significant on the graph (outside the frame) but are not (not black on the last figure), this is probably because the frame is flat on the zero coordinate (no volume inside the frame at this position). Thus, no way to conclude that data2 dots here are significant. These dots are refered to as "unknown". The pb.dot argument deals with such dots
+ # dots that are sometimes inside and outside the frame, depending on the sliding window, are treated differently: they are removed. Such dots are neither classified as "signif", "non signif" or "unknown", but as "inconsistent"
+ # unknown dots are treated as finally significant, not significant, or unknown (data2.pb.dot argument) for each x-axis and y-axis separately. Then, the union or intersection of significant dots is performed (argument xy.cross.kind). See the example section
+ # ARGUMENTS
+ # data1: a data frame containing a column of x-axis values and a column of y-axis values
+ # x1: character string of the data1 column name for x-axis (first column of data1 by default)
+ # y1: character string of the data1 column name for y-axis (second column of data1 by default)
+ # x.range.split: positive non null numeric value giving the number of interval on the x value range. if x.range is the range of the dots on the x-axis, then abs(diff(x.range) / x.range.split) gives the window size. Window size decreases when range.split increases. In unit of x-axis. Write NULL if not required. At least one of the x.range.split and y.range.split must be non NULL
+ # x.step.factor: positive non null numeric value giving the shift step of the window. If x.step.factor = 1, no overlap during the sliding (when the window slides from position n to position n+1, no overlap between the two positions). If x.step.factor = 2, 50% of overlap (when the window slides from position n to position n+1, the window on position n+1 overlap 50% of the window when it was on position n)
+ # y.range.split: same as x.range.split for the y-axis. At least one of the x.range.split and y.range.split must be non NULL
+ # y.step.factor: same as x.step.factor for the y-axis
+ # error: proportion (from 0 to 1) of false positives (i.e., proportion of dots from data1 outside of the frame). 0.05 means 5% of the dots from data1 outside of the frame
+ # data2: a data frame containing a column of x-axis values and a column of y-axis values, for which outside dots of the data1 cloud has to be determined. Write NULL if not required
+ # x2: character string of the data1 column name for x-axis (first column of data1 by default)
+ # y2: character string of the data1 column name for y-axis (second column of data1 by default)
+ # data2.pb.dot: unknown dots are explain in the warning section above. If "signif", then the unknown dots are finally considered as significant (outside the frame). If "not.signif", then the unknown dots are finally considered as non significant (inside the frame). If "unknown", no conclusion are drawn from these dots. See the examples below
+ # xy.cross.kind: if data2 is non null and if both x.range.split and y.range.split are non null, which dots are finally significants? Write "&" for intersection of outside dots on x and on y. Write "|" for union of outside dots on x and on y. See the examples below
+ # plot: logical. Print graphs that check the frame?
+ # graph.in.file: logical. Graphs sent into a graphic device already opened? If FALSE, GUI are opened for each graph. If TRUE, no GUI are opended. The graphs are displayed on the current active graphic device. Ignored if plot is FALSE
+ # raster: logical. Dots in raster mode? If FALSE, dots from each geom_point from geom argument are in vectorial mode (bigger pdf and long to display if millions of dots). If TRUE, dots from each geom_point from geom argument are in matricial mode (smaller pdf and easy display if millions of dots, but long to generate the layer). If TRUE, the region plot will be square to avoid a bug in fun_gg_point_rast(). If TRUE, solve the transparency problem with some GUI. Not considered if plot is FALSE
+ # warn.print: logical. Print warnings at the end of the execution? No print if no warning messages
+ # lib.path: character vector specifying the absolute pathways of the directories containing the required packages if not in the default directories. Ignored if NULL. Ignored if plot is FALSE
+ # RETURN
+ # several graphs if plot is TRUE
+ # a list containing:
+ # $data1.removed.row.nb: which rows have been removed due to NA; NaN, -Inf or Inf detection in x1 or y1 columns (NULL if no row removed)
+ # $data1.removed.rows: removed rows (NULL if no row removed)
+ # $data2.removed.row.nb: which rows have been removed due to NA; NaN, -Inf or Inf detection in x2 or y2 columns (NULL if no row removed)
+ # $data2.removed.rows: removed rows (NULL if no row removed)
+ # $hframe: x and y coordinates of the bottom and top frames for frame plotting (frame1 for the left step and frame2 for the right step)
+ # $vframe: x and y coordinates of the left and right frames for frame plotting (frame1 for the down step and frame2 for the top step)
+ # $data1.signif.dot: the significant dots of data1 (i.e., dots outside the frame). A good segmentation should not have any data1.signif.dot
+ # $data1.non.signif.dot: the non significant dots of data1 (i.e., dots inside the frame)
+ # $data1.inconsistent.dot: see the warning section above
+ # $data2.signif.dot: the significant dots of data2 if non NULL (i.e., dots outside the frame)
+ # $data2.non.signif.dot: the non significant dots of data2 (i.e., dots inside the frame)
+ # $data2.unknown.dot: the problematic dots of data2 (i.e., data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots). Is systematically NULL except if argument data2.pb.dot = "unknown" and some data2 dots are in such situation. Modifying the segmentation x.range.split, x.step.factor, y.range.split, y.step.factor arguments can solve this problem
+ # $data2.inconsistent.dot: see the warning section above
+ # $axes: the x-axis and y-axis info
+ # $warn: the warning messages. Use cat() for proper display. NULL if no warning
+ # REQUIRED PACKAGES
+ # ggplot2 if plot is TRUE
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # if plot is TRUE:
+ # fun_pack()
+ # fun_open()
+ # fun_gg_palette()
+ # fun_gg_scatter()
+ # fun_gg_empty_graph()
+ # fun_close()
+ # EXAMPLES
+ # example explaining the unknown and inconsistent dots, and the cross
+
+ # set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data1[5:7, 2] <- NA ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; data2[11:13, 1] <- Inf ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = 20, x.step.factor = 10, y.range.split = 23, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "not.signif", xy.cross.kind = "|", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
+ # set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = NULL, x.step.factor = 10, y.range.split = 23, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "unknown", xy.cross.kind = "|", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
+ # set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_segmentation(data1 = data1, x1 = names(data1)[1], y1 = names(data1)[2], x.range.split = 20, x.step.factor = 10, y.range.split = NULL, y.step.factor = 10, error = 0, data2 = data2, x2 = names(data2)[1], y2 = names(data2)[2], data2.pb.dot = "unknown", xy.cross.kind = "&", plot = TRUE, graph.in.file = FALSE, raster = FALSE, lib.path = NULL)
+ # DEBUGGING
+ # set.seed(1) ; data1 = data.frame(x = rnorm(50), y = rnorm(50), stringsAsFactors = TRUE) ; data1[5:7, 2] <- NA ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 5 ; x.step.factor = 10 ; y.range.split = 5 ; y.step.factor = 10 ; error = 0 ; data2 = data.frame(x = rnorm(50, 0, 2), y = rnorm(50, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "unknown" ; xy.cross.kind = "|" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
+ # set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 20 ; x.step.factor = 10 ; y.range.split = 23 ; y.step.factor = 10 ; error = 0 ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "not.signif" ; xy.cross.kind = "|" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
+ # set.seed(1) ; data1 = data.frame(x = rnorm(500), y = rnorm(500), stringsAsFactors = TRUE) ; data2 = data.frame(x = rnorm(500, 0, 2), y = rnorm(500, 0, 2), stringsAsFactors = TRUE) ; set.seed(NULL) ; x1 = names(data1)[1] ; y1 = names(data1)[2] ; x.range.split = 20 ; x.step.factor = 10 ; y.range.split = NULL ; y.step.factor = 10 ; error = 0 ; x2 = names(data2)[1] ; y2 = names(data2)[2] ; data2.pb.dot = "unknown" ; xy.cross.kind = "&" ; plot = TRUE ; graph.in.file = FALSE ; raster = FALSE ; warn.print = TRUE ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ ini.warning.length <- options()$warning.length
+ warn <- NULL
+ warn.count <- 0
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & length(data1) < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OF AT LEAST 2 COLUMNS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = x1, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (x1 %in% names(data1))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x1 ARGUMENT MUST BE A COLUMN NAME OF data1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & x1 %in% names(data1)){
+ tempo <- fun_check(data = data1[, x1], data.name = "x1 COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = y1, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (y1 %in% names(data1))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y1 ARGUMENT MUST BE A COLUMN NAME OF data1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & y1 %in% names(data1)){
+ tempo <- fun_check(data = data1[, y1], data.name = "y1 COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ if(is.null(x.range.split) & is.null(y.range.split)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE OF THE x.range.split AND y.range.split ARGUMENTS MUST BE NON NULL")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(x.range.split)){
+ tempo <- fun_check(data = x.range.split, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & x.range.split < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.range.split ARGUMENT CANNOT BE LOWER THAN 1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ if( ! is.null(y.range.split)){
+ tempo <- fun_check(data = y.range.split, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & y.range.split < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.range.split ARGUMENT CANNOT BE LOWER THAN 1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = x.step.factor, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & x.step.factor < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.step.factor ARGUMENT CANNOT BE LOWER THAN 1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = y.step.factor, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & y.step.factor < 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.step.factor ARGUMENT CANNOT BE LOWER THAN 1")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = error, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(data2)){
+ if(is.null(x2) | is.null(y2)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x2 AND y2 ARGUMENTS CANNOT BE NULL IF data2 ARGUMENT IS NON NULL")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = data2, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & length(data2) < 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data2 ARGUMENT MUST BE A DATA FRAME OF AT LEAST 2 COLUMNS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(x2)){
+ tempo <- fun_check(data = x2, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (x2 %in% names(data2))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x2 ARGUMENT MUST BE A COLUMN NAME OF data2")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & x2 %in% names(data2)){
+ tempo <- fun_check(data = data2[, x2], data.name = "x2 COLUMN OF data2", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ }
+ if( ! is.null(y2)){
+ tempo <- fun_check(data = y2, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & ! (y2 %in% names(data2))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y2 ARGUMENT MUST BE A COLUMN NAME OF data2")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & y2 %in% names(data2)){
+ tempo <- fun_check(data = data2[, y2], data.name = "y2 COLUMN OF data2", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ }
+ }
+ }
+ if( ! is.null(data2)){
+ tempo <- fun_check(data = data2.pb.dot, options = c("signif", "not.signif", "unknown"), length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if( ! (is.null(x.range.split)) & ! (is.null(y.range.split))){
+ tempo <- fun_check(data = xy.cross.kind, options = c("&", "|"), length = 1, fun.name = function.name) ; eval(ee)
+ }
+ tempo <- fun_check(data = plot, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & plot == TRUE){
+ tempo <- fun_check(data = raster, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = graph.in.file, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & graph.in.file == TRUE & is.null(dev.list())){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": \ngraph.in.file PARAMETER SET TO TRUE BUT NO ACTIVE GRAPHIC DEVICE DETECTED")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo$problem == FALSE & graph.in.file == TRUE & ! is.null(dev.list())){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHS PRINTED IN THE CURRENT DEVICE (TYPE ", toupper(names(dev.cur())), ")")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # other required function checking
+ if(plot == TRUE){
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_open", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_open() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_gg_palette", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_palette() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_gg_empty_graph", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_empty_graph() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_gg_scatter", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_gg_scatter() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_close", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_close() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end other required function checking
+ # package checking
+ if(plot == TRUE){
+ fun_pack(req.package = c("ggplot2"), lib.path = lib.path)
+ }
+ # end package checking
+ # main code
+ # na and Inf detection and removal (done now to be sure of the correct length of categ)
+ data1.removed.row.nb <- NULL
+ data1.removed.rows <- NULL
+ data2.removed.row.nb <- NULL
+ data2.removed.rows <- NULL
+ if(any(is.na(data1[, c(x1, y1)])) | any(is.infinite(data1[, x1])) | any(is.infinite(data1[, y1]))){
+ tempo.na <- unlist(lapply(lapply(c(data1[c(x1, y1)]), FUN = is.na), FUN = which))
+ tempo.inf <- unlist(lapply(lapply(c(data1[c(x1, y1)]), FUN = is.infinite), FUN = which))
+ data1.removed.row.nb <- sort(unique(c(tempo.na, tempo.inf)))
+ if(length(data1.removed.row.nb) > 0){
+ data1.removed.rows <- data1[data1.removed.row.nb, ]
+ }
+ if(length(data1.removed.row.nb) == nrow(data1)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE NA, NaN, -Inf OR Inf DETECTED IN EACH ROW OF data1. FUNCTION CANNOT BE USED ON EMPTY DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(data1.removed.row.nb) > 0){
+ data1 <- data1[-data1.removed.row.nb, ]
+ }
+ if(nrow(data1) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x1, y1), collapse = " "), " OF data1 AND CORRESPONDING ROWS REMOVED (SEE $data1.removed.row.nb AND $data1.removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x1, y1), collapse = " "), " OF data1. NO ROW REMOVED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data2)){
+ if(any(is.na(data2[, c(x2, y2)])) | any(is.infinite(data2[, x2])) | any(is.infinite(data2[, y2]))){
+ tempo.na <- unlist(lapply(lapply(c(data2[c(x2, y2)]), FUN = is.na), FUN = which))
+ tempo.inf <- unlist(lapply(lapply(c(data2[c(x2, y2)]), FUN = is.infinite), FUN = which))
+ data2.removed.row.nb <- sort(unique(c(tempo.na, tempo.inf)))
+ if(length(data2.removed.row.nb) > 0){
+ data2.removed.rows <- data2[data2.removed.row.nb, ]
+ }
+ if(length(data2.removed.row.nb) == nrow(data2)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": AT LEAST ONE NA, NaN, -Inf OR Inf DETECTED IN EACH ROW OF data2. FUNCTION CANNOT BE USED ON EMPTY DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(data2.removed.row.nb) > 0){
+ data2 <- data2[-data2.removed.row.nb, ]
+ }
+ if(nrow(data2) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x2, y2), collapse = " "), " OF data2 AND CORRESPONDING ROWS REMOVED (SEE $data2.removed.row.nb AND $data2.removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NO NA, NaN, -Inf OR Inf DETECTED IN COLUMN ", paste(c(x2, y2), collapse = " "), " OF data2. NO ROW REMOVED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end na and Inf detection and removal (done now to be sure of the correct length of categ)
+ # row annotation (dot number)
+ # data1 <- data1[ ! duplicated(data1[, c(x1, y1)]), ] # do not remove the dots that have same x and y values, because they will have different dot number -> not the same position on the matrices (so true for symmetric matrices)
+ data1 <- cbind(data1, DOT_NB = 1:nrow(data1), stringsAsFactors = TRUE)
+ if( ! is.null(data2)){
+ # data2 <- data2[ ! duplicated(data2[, c(x2, y2)]), ] # do not remove the dots that have same x and y values, because they will have different dot number -> not the same position on the matrices (so true for symmetric matrices)
+ data2 <- cbind(data2, DOT_NB = 1:nrow(data2), stringsAsFactors = TRUE)
+ }
+ # end row annotation (dot number)
+
+
+
+
+ # Method using x unit interval
+ # may be create vector of each column to increase speed
+ x.data1.l <- NULL # x coord of the y upper and lower limits defined on the data1 cloud for left step line
+ x.data1.r <- NULL # x coord of the y upper and lower limits defined on the data1 cloud for right step line
+ y.data1.down.limit.l <- NULL # lower limit of the data1 cloud for left step line
+ y.data1.top.limit.l <- NULL # upper limit of the data1 cloud for left step line
+ y.data1.down.limit.r <- NULL # lower limit of the data1 cloud for right step line
+ y.data1.top.limit.r <- NULL # upper limit of the data1 cloud for left step line
+ if(any(data1[, x1] %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE x1 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ x.range <- range(data1[, x1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ if(suppressWarnings(any(x.range %in% c(Inf, -Inf)))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED x.range CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1[, y1] %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE data1 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE y1 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ y.range <- range(data1[, y1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ if(suppressWarnings(any(x.range %in% c(Inf, -Inf)))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED y.range CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ x.range.plot <- range(data1[, x1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ y.range.plot <- range(data1[, y1], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ if( ! is.null(data2)){
+ if(any(data2[, x2] %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE data2 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE x2 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ x.range.plot <- range(data1[, x1], data2[, x2], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ if(any(data2[, y2] %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE data2 ARGUMENT CONTAINS -Inf OR Inf VALUES IN THE y2 COLUMN, THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ y.range.plot <- range(data1[, y1], data2[, y2], na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ }
+ if(suppressWarnings(any(x.range.plot %in% c(Inf, -Inf)))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED x.range.plot CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 (AND data2?) ARGUMENTS ARE NA OR Inf ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(suppressWarnings(any(y.range.plot %in% c(Inf, -Inf)))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " COMPUTED y.range.plot CONTAINS Inf VALUES, BECAUSE VALUES FROM data1 (AND data2?) ARGUMENTS ARE NA OR Inf ONLY")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(x.range.split)){
+ # data.frame ordering to slide the window from small to big values + sliding window definition
+ data1 <- data1[order(data1[, x1], na.last = TRUE), ]
+ if( ! is.null(data2)){
+ data2 <- data2[order(data2[, x2], na.last = TRUE), ]
+ }
+ x.win.size <- abs(diff(x.range) / x.range.split) # in unit of x-axis
+ step <- x.win.size / x.step.factor
+ # end data.frame ordering to slide the window from small to big values + sliding window definition
+ # x-axis sliding and y-axis limits of the data1 cloud -> y significant data2
+ loop.nb <- ceiling((diff(x.range) - x.win.size) / step) # x.win.size + n * step covers the x range if x.win.size + n * step >= diff(x.range), thus if n >= (diff(x.range) - x.win.size) / step
+ y.outside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are upper or lower than the frame
+ y.inside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are not upper or lower than the frame
+ y.data1.median <- median(data1[, y1], na.rm = TRUE) # will be used for sliding windows without data1 in it
+ if( ! is.null(data2)){
+ y.outside.data2.dot.nb <- integer() # vector that will contain the selected 1D coordinates (i.e., dots) of data2 that are upper or lower than the data1 frame
+ y.inside.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are not upper or lower than the data1 frame
+ y.unknown.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are problematic: data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots
+ # recover data2 dots outside the range of data1
+ if(any(data2[, x2] < x.range[1])){
+ y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[data2[, x2] < x.range[1]])
+ #tempo.warn & indicate the interval
+ }
+ if(any(data2[, x2] > x.range[2])){
+ y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[data2[, x2] > x.range[2]])
+ #tempo.warn & indicate the interval
+ }
+ # end recover data2 dots outside the range of data1
+ }
+ # loop.ini.time <- as.numeric(Sys.time())
+ for(i1 in 0:(loop.nb + 1)){
+ min.pos <- x.range[1] + step * i1 # lower position of the sliding window in data1
+ max.pos <- min.pos + x.win.size # upper position of the sliding window in data1
+ x.data1.l <- c(x.data1.l, min.pos, min.pos + step) # min.pos + step to make the steps
+ x.data1.r <- c(x.data1.r, max.pos, max.pos + step) # max.pos + step to make the steps
+ x.data1.dot.here <- data1[, x1] >= min.pos & data1[, x1] < max.pos # is there data1 dot present in the sliding window, considering the x axis?
+ if( ! is.null(data2)){
+ x.data2.dot.here <- data2[, x2] >= min.pos & data2[, x2] < max.pos # is there data2 dot present in the sliding window, considering the x axis?
+ }
+ # recover the data1 dots outside the frame
+ if(any(x.data1.dot.here == TRUE)){
+ tempo.y.data1.top.limit <- quantile(data1[x.data1.dot.here, y1], probs = 1 - error, na.rm = TRUE)
+ tempo.y.data1.down.limit <- quantile(data1[x.data1.dot.here, y1], probs = 0 + error, na.rm = TRUE)
+ y.data1.top.limit.l <- c(y.data1.top.limit.l, tempo.y.data1.top.limit, tempo.y.data1.top.limit)
+ y.data1.down.limit.l <- c(y.data1.down.limit.l, tempo.y.data1.down.limit, tempo.y.data1.down.limit)
+ y.data1.top.limit.r <- c(y.data1.top.limit.r, tempo.y.data1.top.limit, tempo.y.data1.top.limit)
+ y.data1.down.limit.r <- c(y.data1.down.limit.r, tempo.y.data1.down.limit, tempo.y.data1.down.limit)
+ y.data1.dot.signif <- ( ! ((data1[, y1] <= tempo.y.data1.top.limit) & (data1[, y1] >= tempo.y.data1.down.limit))) & x.data1.dot.here # is there data1 dot present in the sliding window, above or below the data1 limits, considering the y axis?
+ y.data1.dot.not.signif <- x.data1.dot.here & ! y.data1.dot.signif
+ y.outside.data1.dot.nb <- c(y.outside.data1.dot.nb, data1$DOT_NB[y.data1.dot.signif]) # recover the row number of data1
+ y.outside.data1.dot.nb <- unique(y.outside.data1.dot.nb)
+ y.inside.data1.dot.nb <- c(y.inside.data1.dot.nb, data1$DOT_NB[y.data1.dot.not.signif])
+ y.inside.data1.dot.nb <- unique(y.inside.data1.dot.nb)
+ }else{
+ y.data1.top.limit.l <- c(y.data1.top.limit.l, y.data1.median, y.data1.median)
+ y.data1.down.limit.l <- c(y.data1.down.limit.l, y.data1.median, y.data1.median)
+ y.data1.top.limit.r <- c(y.data1.top.limit.r, y.data1.median, y.data1.median)
+ y.data1.down.limit.r <- c(y.data1.down.limit.r, y.data1.median, y.data1.median)
+ }
+ # end recover the data1 dots outside the frame
+ # recover the data2 dots outside the frame
+ if( ! is.null(data2)){
+ if(any(x.data1.dot.here == TRUE) & any(x.data2.dot.here == TRUE)){
+ y.data2.dot.signif <- ( ! ((data2[, y2] <= tempo.y.data1.top.limit) & (data2[, y2] >= tempo.y.data1.down.limit))) & x.data2.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the y axis?
+ y.data2.dot.not.signif <- x.data2.dot.here & ! y.data2.dot.signif
+ y.outside.data2.dot.nb <- c(y.outside.data2.dot.nb, data2$DOT_NB[y.data2.dot.signif])
+ y.outside.data2.dot.nb <- unique(y.outside.data2.dot.nb)
+ y.inside.data2.dot.nb <- c(y.inside.data2.dot.nb, data2$DOT_NB[y.data2.dot.not.signif])
+ y.inside.data2.dot.nb <- unique(y.inside.data2.dot.nb)
+ }else if(any(x.data1.dot.here == FALSE) & any(x.data2.dot.here == TRUE)){ # problem: data2 dots in the the window but no data1 dots to generates the quantiles
+ y.unknown.data2.dot.nb <- c(y.unknown.data2.dot.nb, data2$DOT_NB[x.data2.dot.here])
+ y.unknown.data2.dot.nb <- unique(y.unknown.data2.dot.nb)
+ #tempo.warn & indicate the interval
+
+
+
+
+ # tempo.warn <- paste0("FROM FUNCTION ", function.name, ": THE [", round(min.pos, 3), " ; ", round(max.pos, 3), "] INTERVAL DOES NOT CONTAIN data1 X VALUES BUT CONTAINS data2 X VALUES WHICH CANNOT BE EVALUATED.\nTHE CONCERNED data2 ROW NUMBERS ARE:\n", paste(which(x.data1.dot.here == FALSE & x.data2.dot.here == TRUE), collapse = "\n"))
+ # warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end recover the data2 dots outside the frame
+ # if(any(i1 == seq(1, loop.nb, 500))){
+ # loop.fin.time <- as.numeric(Sys.time()) # time of process end
+ # cat(paste0("COMPUTATION TIME OF LOOP ", i1, " / ", loop.nb, ": ", as.character(lubridate::seconds_to_period(round(loop.fin.time - loop.ini.time))), "\n"))
+ # }
+ }
+ if(max.pos < x.range[2]){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE SLIDING WINDOW HAS NOT REACHED THE MAX VALUE OF data1 ON THE X-AXIS: ", max.pos, " VERSUS ", x.range[2])
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ y.incon.data1.dot.nb.final <- unique(c(y.outside.data1.dot.nb[y.outside.data1.dot.nb %in% y.inside.data1.dot.nb], y.inside.data1.dot.nb[y.inside.data1.dot.nb %in% y.outside.data1.dot.nb])) # inconsistent dots: if a row number of y.inside.data1.dot.nb is present in y.outside.data1.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
+ y.outside.data1.dot.nb.final <- y.outside.data1.dot.nb[ ! (y.outside.data1.dot.nb %in% y.incon.data1.dot.nb.final)] # inconsistent dots removed from the outside list
+ y.inside.data1.dot.nb.final <- y.inside.data1.dot.nb[ ! (y.inside.data1.dot.nb %in% y.incon.data1.dot.nb.final)] # inconsistent dots removed from the inside list
+ if( ! is.null(data2)){
+ # if some unknown dots are also inside, and/or outside, they are put in the inside and/or outside. Ok, because then the intersection between inside and outside is treated -> inconsistent dots
+ tempo.unknown.out <- y.unknown.data2.dot.nb[y.unknown.data2.dot.nb %in% y.outside.data2.dot.nb]
+ y.outside.data2.dot.nb <- unique(c(y.outside.data2.dot.nb, tempo.unknown.out)) # if a row number of y.unknown.data2.dot.nb is present in y.outside.data2.dot.nb, it is put into outside
+ tempo.unknown.in <- y.unknown.data2.dot.nb[y.unknown.data2.dot.nb %in% y.inside.data2.dot.nb]
+ y.inside.data2.dot.nb <- unique(c(y.inside.data2.dot.nb, tempo.unknown.in)) # if a row number of y.unknown.data2.dot.nb is present in y.inside.data2.dot.nb, it is put into inside
+ y.unknown.data2.dot.nb.final <- y.unknown.data2.dot.nb[ ! (y.unknown.data2.dot.nb %in% c(y.outside.data2.dot.nb, y.inside.data2.dot.nb))] # then dots also in inside and outside are remove from unknown
+ y.incon.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb[y.outside.data2.dot.nb %in% y.inside.data2.dot.nb], y.inside.data2.dot.nb[y.inside.data2.dot.nb %in% y.outside.data2.dot.nb])) # inconsistent dots: if a row number of y.inside.data2.dot.nb is present in y.outside.data2.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
+ y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb[ ! (y.outside.data2.dot.nb %in% y.incon.data2.dot.nb.final)] # inconsistent dots removed from the outside list
+ y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb[ ! (y.inside.data2.dot.nb %in% y.incon.data2.dot.nb.final)] # inconsistent dots removed from the inside list
+ }
+ # end x-axis sliding and y-axis limits of the data1 cloud -> y significant data2
+ }
+ # end Method using x unit interval
+
+
+
+
+ # Method using y unit interval
+ y.data1.d <- NULL # y coord of the x upper and lower limits defined on the data1 cloud for down step line
+ y.data1.t <- NULL # y coord of the x upper and lower limits defined on the data1 cloud for top step line
+ x.data1.left.limit.d <- NULL # left limit of the data1 cloud for down step line
+ x.data1.right.limit.d <- NULL # right limit of the data1 cloud for down step line
+ x.data1.left.limit.t <- NULL # left limit of the data1 cloud for top step line
+ x.data1.right.limit.t <- NULL # right limit of the data1 cloud for top step line
+ if( ! is.null(y.range.split)){
+ # data.frame ordering to slide the window from small to big values + sliding window definition
+ data1 <- data1[order(data1[, y1], na.last = TRUE), ]
+ if( ! is.null(data2)){
+ data2 <- data2[order(data2[, y2], na.last = TRUE), ]
+ }
+ y.win.size <- abs(diff(y.range) / y.range.split) # in unit of y-axis
+ step <- y.win.size / y.step.factor
+ # end data.frame ordering to slide the window from small to big values + sliding window definition
+ # y-axis sliding and x-axis limits of the data1 cloud -> x significant data2
+ loop.nb <- ceiling((diff(y.range) - y.win.size) / step) # y.win.size + n * step covers the y range if y.win.size + n * step >= diff(y.range), thus if n >= (diff(y.range) - y.win.size) / step
+ x.outside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are upper or lower than the frame
+ x.inside.data1.dot.nb <- integer() # vector that will contain the selected rows numbers of data1 that are not upper or lower than the frame
+ x.data1.median <- median(data1[, x1], na.rm = TRUE) # will be used for sliding window without data1 in it
+ if( ! is.null(data2)){
+ x.outside.data2.dot.nb <- integer() # vector that will contain the selected 1D coordinates (i.e., dots) of data2 that are upper or lower than the data1 frame
+ x.inside.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are not upper or lower than the data1 frame
+ x.unknown.data2.dot.nb <- integer() # vector that will contain the 1D coordinates (i.e., dots) of data2 that are problematic: data2 dots outside of the range of data1, or data2 dots in a sliding window without data1 dots
+ # recover data2 dots outside the range of data1
+ if(any(data2[, y2] < y.range[1])){
+ x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[data2[, y2] < y.range[1]])
+ }
+ if(any(data2[, y2] > y.range[2])){
+ x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[data2[, y2] > y.range[2]])
+ }
+ # end recover data2 dots outside the range of data1
+ }
+ # loop.ini.time <- as.numeric(Sys.time())
+ for(i1 in 0:(loop.nb + 1)){
+ min.pos <- y.range[1] + step * i1 # lower position of the sliding window in data1
+ max.pos <- min.pos + y.win.size # upper position of the sliding window in data1
+ y.data1.d <- c(y.data1.d, min.pos, min.pos + step) # min.pos + step to make the steps
+ y.data1.t <- c(y.data1.t, max.pos, max.pos + step) # max.pos + step to make the steps
+ y.data1.dot.here <- data1[, y1] >= min.pos & data1[, y1] < max.pos # is there data1 dot present in the sliding window, considering the y axis?
+ if( ! is.null(data2)){
+ y.data2.dot.here <- data2[, y2] >= min.pos & data2[, y2] < max.pos # is there data2 dot present in the sliding window, considering the y axis?
+ }
+ # recover the data1 dots outside the frame
+ if(any(y.data1.dot.here == TRUE)){
+ tempo.x.data1.right.limit <- quantile(data1[y.data1.dot.here, x1], probs = 1 - error, na.rm = TRUE)
+ tempo.x.data1.left.limit <- quantile(data1[y.data1.dot.here, x1], probs = 0 + error, na.rm = TRUE)
+ x.data1.right.limit.d <- c(x.data1.right.limit.d, tempo.x.data1.right.limit, tempo.x.data1.right.limit)
+ x.data1.left.limit.d <- c(x.data1.left.limit.d, tempo.x.data1.left.limit, tempo.x.data1.left.limit)
+ x.data1.right.limit.t <- c(x.data1.right.limit.t, tempo.x.data1.right.limit, tempo.x.data1.right.limit)
+ x.data1.left.limit.t <- c(x.data1.left.limit.t, tempo.x.data1.left.limit, tempo.x.data1.left.limit)
+ x.data1.dot.signif <- ( ! ((data1[, x1] <= tempo.x.data1.right.limit) & (data1[, x1] >= tempo.x.data1.left.limit))) & y.data1.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the x axis?
+ x.data1.dot.not.signif <- y.data1.dot.here & ! x.data1.dot.signif
+ x.outside.data1.dot.nb <- c(x.outside.data1.dot.nb, data1$DOT_NB[x.data1.dot.signif]) # recover the row number of data1
+ x.outside.data1.dot.nb <- unique(x.outside.data1.dot.nb)
+ x.inside.data1.dot.nb <- c(x.inside.data1.dot.nb, data1$DOT_NB[x.data1.dot.not.signif])
+ x.inside.data1.dot.nb <- unique(x.inside.data1.dot.nb)
+ }else{
+ x.data1.right.limit.d <- c(x.data1.right.limit.d, x.data1.median, x.data1.median)
+ x.data1.left.limit.d <- c(x.data1.left.limit.d, x.data1.median, x.data1.median)
+ x.data1.right.limit.t <- c(x.data1.right.limit.t, x.data1.median, x.data1.median)
+ x.data1.left.limit.t <- c(x.data1.left.limit.t, x.data1.median, x.data1.median)
+ }
+ # end recover the data1 dots outside the frame
+ # recover the data2 dots outside the frame
+ if( ! is.null(data2)){
+ if(any(y.data1.dot.here == TRUE) & any(y.data2.dot.here == TRUE)){
+ x.data2.dot.signif <- ( ! ((data2[, x2] <= tempo.x.data1.right.limit) & (data2[, x2] >= tempo.x.data1.left.limit))) & y.data2.dot.here # is there data2 dot present in the sliding window, above or below the data1 limits, considering the x axis?
+ x.data2.dot.not.signif <- y.data2.dot.here & ! x.data2.dot.signif
+ x.outside.data2.dot.nb <- c(x.outside.data2.dot.nb, data2$DOT_NB[x.data2.dot.signif])
+ x.outside.data2.dot.nb <- unique(x.outside.data2.dot.nb)
+ x.inside.data2.dot.nb <- c(x.inside.data2.dot.nb, data2$DOT_NB[x.data2.dot.not.signif])
+ x.inside.data2.dot.nb <- unique(x.inside.data2.dot.nb)
+ }else if(any(y.data1.dot.here == FALSE) & any(y.data2.dot.here == TRUE)){ # recover the data2 dots outside the range of the data1 cloud
+ x.unknown.data2.dot.nb <- c(x.unknown.data2.dot.nb, data2$DOT_NB[y.data2.dot.here])
+ x.unknown.data2.dot.nb <- unique(x.unknown.data2.dot.nb)
+
+
+
+ # tempo.warn <- paste0("FROM FUNCTION ", function.name, ": THE [", round(min.pos, 3), " ; ", round(max.pos, 3), "] INTERVAL DOES NOT CONTAIN data1 Y VALUES BUT CONTAINS data2 Y VALUES WHICH CANNOT BE EVALUATED.\nTHE CONCERNED data2 ROW NUMBERS ARE:\n", paste(which(y.data1.dot.here == FALSE & y.data2.dot.here == TRUE), collapse = "\n"))
+ # warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end recover the data2 dots outside the frame
+ # if(any(i1 == seq(1, loop.nb, 500))){
+ # loop.fin.time <- as.numeric(Sys.time()) # time of process end
+ # cat(paste0("COMPUTATION TIME OF LOOP ", i1, " / ", loop.nb, ": ", as.character(lubridate::seconds_to_period(round(loop.fin.time - loop.ini.time))), "\n"))
+ # }
+ }
+ if(max.pos < y.range[2]){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": THE SLIDING WINDOW HAS NOT REACHED THE MAX VALUE OF data1 ON THE Y-AXIS: ", max.pos, " VERSUS ", y.range[2])
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ x.incon.data1.dot.nb.final <- unique(c(x.outside.data1.dot.nb[x.outside.data1.dot.nb %in% x.inside.data1.dot.nb], x.inside.data1.dot.nb[x.inside.data1.dot.nb %in% x.outside.data1.dot.nb])) # inconsistent dots: if a row number of x.inside.data1.dot.nb is present in x.outside.data1.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
+ x.outside.data1.dot.nb.final <- x.outside.data1.dot.nb[ ! (x.outside.data1.dot.nb %in% x.incon.data1.dot.nb.final)] # inconsistent dots removed from the outside list
+ x.inside.data1.dot.nb.final <- x.inside.data1.dot.nb[ ! (x.inside.data1.dot.nb %in% x.incon.data1.dot.nb.final)] # inconsistent dots removed from the inside list
+ if( ! is.null(data2)){
+ # if some unknown dots are also inside, and/or outside, they are put in the inside and/or outside. Ok, because then the intersection between inside and outside is treated -> inconsistent dots
+ tempo.unknown.out <- x.unknown.data2.dot.nb[x.unknown.data2.dot.nb %in% x.outside.data2.dot.nb]
+ x.outside.data2.dot.nb <- unique(c(x.outside.data2.dot.nb, tempo.unknown.out)) # if a row number of x.unknown.data2.dot.nb is present in x.outside.data2.dot.nb, it is put into outside
+ tempo.unknown.in <- x.unknown.data2.dot.nb[x.unknown.data2.dot.nb %in% x.inside.data2.dot.nb]
+ x.inside.data2.dot.nb <- unique(c(x.inside.data2.dot.nb, tempo.unknown.in)) # if a row number of x.unknown.data2.dot.nb is present in x.inside.data2.dot.nb, it is put into inside
+ x.unknown.data2.dot.nb.final <- x.unknown.data2.dot.nb[ ! (x.unknown.data2.dot.nb %in% c(x.outside.data2.dot.nb, x.inside.data2.dot.nb))] # then dots also in inside and outside are remove from unknown
+ x.incon.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb[x.outside.data2.dot.nb %in% x.inside.data2.dot.nb], x.inside.data2.dot.nb[x.inside.data2.dot.nb %in% x.outside.data2.dot.nb])) # inconsistent dots: if a row number of x.inside.data2.dot.nb is present in x.outside.data2.dot.nb (and vice versa), it means that during the sliding, a dot has been sometime inside, sometime outside -> removed from the outside list
+ x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb[ ! (x.outside.data2.dot.nb %in% x.incon.data2.dot.nb.final)] # inconsistent dots removed from the outside list
+ x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb[ ! (x.inside.data2.dot.nb %in% x.incon.data2.dot.nb.final)] # inconsistent dots removed from the inside list
+ }
+ # end y-axis sliding and x-axis limits of the data1 cloud -> x significant data2
+ }
+ # end Method using y unit interval
+
+
+
+ # recovering the frame coordinates
+ hframe = rbind(
+ data.frame(
+ x = if(is.null(x.data1.l)){NULL}else{x.data1.l},
+ y = if(is.null(x.data1.l)){NULL}else{y.data1.down.limit.l},
+ kind = if(is.null(x.data1.l)){NULL}else{"down.frame1"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(x.data1.r)){NULL}else{x.data1.r},
+ y = if(is.null(x.data1.r)){NULL}else{y.data1.down.limit.r},
+ kind = if(is.null(x.data1.r)){NULL}else{"down.frame2"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(x.data1.l)){NULL}else{x.data1.l},
+ y = if(is.null(x.data1.l)){NULL}else{y.data1.top.limit.l},
+ kind = if(is.null(x.data1.l)){NULL}else{"top.frame1"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(x.data1.r)){NULL}else{x.data1.r},
+ y = if(is.null(x.data1.r)){NULL}else{y.data1.top.limit.r},
+ kind = if(is.null(x.data1.r)){NULL}else{"top.frame2"},
+ stringsAsFactors = TRUE
+ ),
+ stringsAsFactors = TRUE
+ )
+ vframe = rbind(
+ data.frame(
+ x = if(is.null(y.data1.d)){NULL}else{x.data1.left.limit.d},
+ y = if(is.null(y.data1.d)){NULL}else{y.data1.d},
+ kind = if(is.null(y.data1.d)){NULL}else{"left.frame1"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(y.data1.t)){NULL}else{x.data1.left.limit.t},
+ y = if(is.null(y.data1.t)){NULL}else{y.data1.t},
+ kind = if(is.null(y.data1.t)){NULL}else{"left.frame2"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(y.data1.d)){NULL}else{x.data1.right.limit.d},
+ y = if(is.null(y.data1.d)){NULL}else{y.data1.d},
+ kind = if(is.null(y.data1.d)){NULL}else{"right.frame1"},
+ stringsAsFactors = TRUE
+ ),
+ data.frame(
+ x = if(is.null(y.data1.t)){NULL}else{x.data1.right.limit.t},
+ y = if(is.null(y.data1.t)){NULL}else{y.data1.t},
+ kind = if(is.null(y.data1.t)){NULL}else{"right.frame2"},
+ stringsAsFactors = TRUE
+ ),
+ stringsAsFactors = TRUE
+ )
+ # end recovering the frame coordinates
+ # recovering the dot coordinates
+ data1.signif.dot <- NULL
+ data1.non.signif.dot <- NULL
+ data1.incon.dot <- NULL
+ data2.signif.dot <- NULL
+ data2.non.signif.dot <- NULL
+ data2.unknown.dot <- NULL
+ data2.incon.dot <- NULL
+ if(( ! is.null(x.range.split)) & ( ! is.null(y.range.split))){
+ # inconsistent dots recovery
+ if(length(unique(c(x.incon.data1.dot.nb.final, y.incon.data1.dot.nb.final))) > 0){
+ data1.incon.dot <- data1[data1$DOT_NB %in% unique(c(x.incon.data1.dot.nb.final, y.incon.data1.dot.nb.final)), ] # if a dot in inconsistent in x or y -> classified as inconsistent (so unique() used)
+ # removal of the inconsistent dot in the other classifications
+ x.inside.data1.dot.nb.final <- x.inside.data1.dot.nb.final[ ! x.inside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ y.inside.data1.dot.nb.final <- y.inside.data1.dot.nb.final[ ! y.inside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ x.outside.data1.dot.nb.final <- x.outside.data1.dot.nb.final[ ! x.outside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ y.outside.data1.dot.nb.final <- y.outside.data1.dot.nb.final[ ! y.outside.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ x.unknown.data1.dot.nb.final <- x.unknown.data1.dot.nb.final[ ! x.unknown.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ y.unknown.data1.dot.nb.final <- y.unknown.data1.dot.nb.final[ ! y.unknown.data1.dot.nb.final %in% data1.incon.dot$DOT_NB]
+ # end removal of the inconsistent dot in the other classifications
+ }
+ if( ! is.null(data2)){
+ if(length(unique(c(x.incon.data2.dot.nb.final, y.incon.data2.dot.nb.final))) > 0){
+ data2.incon.dot <- data2[data2$DOT_NB %in% unique(c(x.incon.data2.dot.nb.final, y.incon.data2.dot.nb.final)), ]
+ # removal of the inconsistent dot in the other classifications
+ x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ x.unknown.data2.dot.nb.final <- x.unknown.data2.dot.nb.final[ ! x.unknown.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ y.unknown.data2.dot.nb.final <- y.unknown.data2.dot.nb.final[ ! y.unknown.data2.dot.nb.final %in% data2.incon.dot$DOT_NB]
+ # end removal of the inconsistent dot in the other classifications
+ }
+ }
+ # end inconsistent dots recovery
+ # unknown dots recovery
+ if( ! is.null(data2)){
+ if(data2.pb.dot == "signif"){
+ x.outside.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
+ x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% x.unknown.data2.dot.nb.final] # remove x.unknown.data2.dot.nb.final from x.inside.data2.dot.nb.final
+ y.outside.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
+ y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% y.unknown.data2.dot.nb.final] # remove y.unknown.data2.dot.nb.final from y.inside.data2.dot.nb.final
+ x.unknown.data2.dot.nb.final <- NULL
+ y.unknown.data2.dot.nb.final <- NULL
+ data2.unknown.dot <- NULL
+ }else if(data2.pb.dot == "not.signif"){
+ x.inside.data2.dot.nb.final <- unique(c(x.inside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
+ x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% x.unknown.data2.dot.nb.final] # remove x.unknown.data2.dot.nb.final from x.outside.data2.dot.nb.final
+ y.inside.data2.dot.nb.final <- unique(c(y.inside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
+ y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% y.unknown.data2.dot.nb.final] # remove y.unknown.data2.dot.nb.final from y.outside.data2.dot.nb.final
+ x.unknown.data2.dot.nb.final <- NULL
+ y.unknown.data2.dot.nb.final <- NULL
+ data2.unknown.dot <- NULL
+ }else if(data2.pb.dot == "unknown"){
+ if(length(unique(c(x.unknown.data2.dot.nb.final, y.unknown.data2.dot.nb.final))) > 0){
+ data2.unknown.dot <- data2[data2$DOT_NB %in% unique(c(x.unknown.data2.dot.nb.final, y.unknown.data2.dot.nb.final)), ] # if a dot in unknown in x or y -> classified as unknown (so unique() used)
+ x.outside.data2.dot.nb.final <- x.outside.data2.dot.nb.final[ ! x.outside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove x.unknown.data2.dot.nb.final from x.outside.data2.dot.nb.final
+ x.inside.data2.dot.nb.final <- x.inside.data2.dot.nb.final[ ! x.inside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove x.unknown.data2.dot.nb.final from x.inside.data2.dot.nb.final
+ y.outside.data2.dot.nb.final <- y.outside.data2.dot.nb.final[ ! y.outside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove y.unknown.data2.dot.nb.final from y.outside.data2.dot.nb.final
+ y.inside.data2.dot.nb.final <- y.inside.data2.dot.nb.final[ ! y.inside.data2.dot.nb.final %in% data2.unknown.dot$DOT_NB] # remove y.unknown.data2.dot.nb.final from y.inside.data2.dot.nb.final
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end unknown dots recovery
+ # sign and non sign dot recovery
+ if(xy.cross.kind == "|"){ # here the problem is to deal with significant dots depending on x and y. Thus I start with that, recover dots finally non significant in outside and put them in inside (when &), and remove from inside the dots in outside
+ if(length(unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final))) > 0){
+ tempo.outside <- unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final)) # union so unique() used
+ tempo.inside <- unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final))
+ tempo.inside <- tempo.inside[ ! tempo.inside %in% tempo.outside]
+ data1.signif.dot <- data1[data1$DOT_NB %in% tempo.outside, ]
+ data1.non.signif.dot <- data1[data1$DOT_NB %in% tempo.inside, ]
+ }else{
+ data1.non.signif.dot <- data1[unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
+ }
+ }else if(xy.cross.kind == "&"){
+ if(sum(x.outside.data1.dot.nb.final %in% y.outside.data1.dot.nb.final) > 0){ # that is intersection
+ tempo.outside <- unique(x.outside.data1.dot.nb.final[x.outside.data1.dot.nb.final %in% y.outside.data1.dot.nb.final]) # intersection
+ tempo.outside.removed <- unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final))[ ! unique(c(x.outside.data1.dot.nb.final, y.outside.data1.dot.nb.final)) %in% tempo.outside]
+ tempo.inside <- unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final))
+ data1.signif.dot <- data1[data1$DOT_NB %in% tempo.outside, ]
+ data1.non.signif.dot <- data1[data1$DOT_NB %in% tempo.inside, ]
+ }else{
+ data1.non.signif.dot <- data1[unique(c(x.inside.data1.dot.nb.final, y.inside.data1.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(data2)){
+ if(xy.cross.kind == "|"){ # here the problem is to deal with significant dots depending on x and y. Thus I start with that, recover dots finally non significant in outside and put them in inside (when &), and remove from inside the dots in outside
+ if(length(unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final))) > 0){
+ tempo.outside <- unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final)) # union so unique() used
+ tempo.inside <- unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final))
+ tempo.inside <- tempo.inside[ ! tempo.inside %in% tempo.outside]
+ data2.signif.dot <- data2[data2$DOT_NB %in% tempo.outside, ]
+ data2.non.signif.dot <- data2[data2$DOT_NB %in% tempo.inside, ]
+ }else{
+ data2.non.signif.dot <- data2[unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
+ }
+ }else if(xy.cross.kind == "&"){
+ if(sum(x.outside.data2.dot.nb.final %in% y.outside.data2.dot.nb.final) > 0){ # that is intersection
+ tempo.outside <- unique(x.outside.data2.dot.nb.final[x.outside.data2.dot.nb.final %in% y.outside.data2.dot.nb.final]) # intersection
+ tempo.outside.removed <- unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final))[ ! unique(c(x.outside.data2.dot.nb.final, y.outside.data2.dot.nb.final)) %in% tempo.outside]
+ tempo.inside <- unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final))
+ data2.signif.dot <- data2[data2$DOT_NB %in% tempo.outside, ]
+ data2.non.signif.dot <- data2[data2$DOT_NB %in% tempo.inside, ]
+ }else{
+ data2.non.signif.dot <- data2[unique(c(x.inside.data2.dot.nb.final, y.inside.data2.dot.nb.final)), ] # if no outside dots, I recover all the inside dots and that's it
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end sign and non sign dot recovery
+ }else if(( ! is.null(x.range.split)) & is.null(y.range.split)){
+ # inconsistent dots recovery
+ if(length(y.incon.data1.dot.nb.final) > 0){
+ data1.incon.dot <- data1[data1$DOT_NB %in% y.incon.data1.dot.nb.final, ]
+ }
+ if( ! is.null(data2)){
+ if(length(y.incon.data2.dot.nb.final) > 0){
+ data2.incon.dot <- data2[data2$DOT_NB %in% y.incon.data2.dot.nb.final, ]
+ }
+ }# end inconsistent dots recovery
+ # unknown dots recovery
+ if( ! is.null(data2)){
+ if(data2.pb.dot == "signif"){
+ y.outside.data2.dot.nb.final <- unique(c(y.outside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
+ }else if(data2.pb.dot == "not.signif"){
+ y.inside.data2.dot.nb.final <- unique(c(y.inside.data2.dot.nb.final, y.unknown.data2.dot.nb.final))
+ }else if(data2.pb.dot == "unknown"){
+ if(length(y.unknown.data2.dot.nb.final) > 0){
+ data2.unknown.dot <- data2[data2$DOT_NB %in% y.unknown.data2.dot.nb.final, ]
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 6")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end unknown dots recovery
+ # sign and non sign dot recovery
+ if(length(y.outside.data1.dot.nb.final) > 0){
+ data1.signif.dot <- data1[data1$DOT_NB %in% y.outside.data1.dot.nb.final, ]
+ }
+ if(length(y.inside.data1.dot.nb.final) > 0){
+ data1.non.signif.dot <- data1[data1$DOT_NB %in% y.inside.data1.dot.nb.final, ]
+ }
+ if( ! is.null(data2)){
+ if(length(y.outside.data2.dot.nb.final) > 0){
+ data2.signif.dot <- data2[data2$DOT_NB %in% y.outside.data2.dot.nb.final, ]
+ }
+ if(length(y.inside.data2.dot.nb.final) > 0){
+ data2.non.signif.dot <- data2[data2$DOT_NB %in% y.inside.data2.dot.nb.final, ]
+ }
+ }
+ # end sign and non sign dot recovery
+ }else if(is.null(x.range.split) & ( ! is.null(y.range.split))){
+ # inconsistent dots recovery
+ if(length(x.incon.data1.dot.nb.final) > 0){
+ data1.incon.dot <- data1[data1$DOT_NB %in% x.incon.data1.dot.nb.final, ]
+ }
+ if( ! is.null(data2)){
+ if(length(x.incon.data2.dot.nb.final) > 0){
+ data2.incon.dot <- data2[data2$DOT_NB %in% x.incon.data2.dot.nb.final, ]
+ }
+ }# end inconsistent dots recovery
+ # unknown dots recovery
+ if( ! is.null(data2)){
+ if(data2.pb.dot == "signif"){
+ x.outside.data2.dot.nb.final <- unique(c(x.outside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
+ }else if(data2.pb.dot == "not.signif"){
+ x.inside.data2.dot.nb.final <- unique(c(x.inside.data2.dot.nb.final, x.unknown.data2.dot.nb.final))
+ }else if(data2.pb.dot == "unknown"){
+ if(length(x.unknown.data2.dot.nb.final) > 0){
+ data2.unknown.dot <- data2[data2$DOT_NB %in% x.unknown.data2.dot.nb.final, ]
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 7")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end unknown dots recovery
+ # sign and non sign dot recovery
+ if(length(x.outside.data1.dot.nb.final) > 0){
+ data1.signif.dot <- data1[data1$DOT_NB %in% x.outside.data1.dot.nb.final, ]
+ }
+ if(length(x.inside.data1.dot.nb.final) > 0){
+ data1.non.signif.dot <- data1[data1$DOT_NB %in% x.inside.data1.dot.nb.final, ]
+ }
+ if( ! is.null(data2)){
+ if(length(x.outside.data2.dot.nb.final) > 0){
+ data2.signif.dot <- data2[data2$DOT_NB %in% x.outside.data2.dot.nb.final, ]
+ }
+ if(length(x.inside.data2.dot.nb.final) > 0){
+ data2.non.signif.dot <- data2[data2$DOT_NB %in% x.inside.data2.dot.nb.final, ]
+ }
+ }
+ # end sign and non sign dot recovery
+ }
+ # end recovering the dot coordinates
+ # verif
+ if(any(data1.signif.dot$DOT_NB %in% data1.non.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", FUNCTION.NAME, ": CODE INCONSISTENCY 8")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1.non.signif.dot$DOT_NB %in% data1.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", FUNCTION.NAME, ": CODE INCONSISTENCY 9")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1.signif.dot$DOT_NB %in% data1.incon.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 10")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1.incon.dot$DOT_NB %in% data1.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 11")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1.non.signif.dot$DOT_NB %in% data1.incon.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 12")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data1.incon.dot$DOT_NB %in% data1.non.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 13")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(data2)){
+ if(any(data2.signif.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 14")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.non.signif.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 15")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.signif.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 16")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.unknown.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 17")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.signif.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 18")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.incon.dot$DOT_NB %in% data2.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 19")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.non.signif.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 20")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.unknown.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 21")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.non.signif.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 22")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.incon.dot$DOT_NB %in% data2.non.signif.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 23")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.unknown.dot$DOT_NB %in% data2.incon.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 24")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(data2.incon.dot$DOT_NB %in% data2.unknown.dot$DOT_NB)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 25")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end verif
+ # plot
+ # recovering the axes data whatever plot or not
+ if(is.null(data2)){
+ axes <- fun_gg_scatter(data1 = list(data1), x = list(x1), y = list(y1), categ = list(NULL), color = list(fun_gg_palette(2)[2]), geom = list("geom_point"), alpha = list(0.5), x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, plot = FALSE, return = TRUE)$axes
+ }else{
+ axes <- fun_gg_scatter(data1 = list(data1, data2), x = list(x1, x2), y = list(y1, y2), categ = list(NULL, NULL), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1]), geom = list("geom_point", "geom_point"), alpha = list(0.5, 0.5), x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, plot = FALSE, return = TRUE)$axes
+ }
+ # end recovering the axes data whatever plot or not
+ if(plot == TRUE){
+ # add a categ for plot legend
+ tempo.df.name <- c("data1", "data1.signif.dot", "data1.incon.dot", "data2", "data2.signif.dot", "data2.unknown.dot", "data2.incon.dot")
+ tempo.class.name <- c("data1", "data1", "data1", "data2", "data2", "data2", "data2")
+ for(i2 in 1:length(tempo.df.name)){
+ if( ! is.null(get(tempo.df.name[i2], env = sys.nframe(), inherit = FALSE))){
+ assign(tempo.df.name[i2], data.frame(get(tempo.df.name[i2], env = sys.nframe(), inherit = FALSE), kind = tempo.class.name[i2]),
+ stringsAsFactors = TRUE)
+ }
+ }
+ # end add a categ for plot legend
+ if(( ! is.null(x.range.split)) & ( ! is.null(y.range.split))){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe), x = list(x1, "x", "x"), y = list(y1, "y", "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data1.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe, data1.signif.dot), x = list(x1, "x", "x", x1), y = list(y1, "y", "y", y1), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data1.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, vframe, data1.incon.dot), x = list(x1, "x", "x", x1), y = list(y1, "y", "y", y1), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME" , "VERT FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, hframe , vframe), x = list(x1, x2, "x", "x"), y = list(y1, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data2.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data2.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2.unknown.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, hframe , vframe), x = list(x1, x2, x2, "x", "x"), y = list(y1, y2, y2, "y", "y"), categ = list("kind", "kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "HORIZ FRAME" , "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 12, title = "DATA2 + DATA2 UNKNOWN DOTS")
+ }
+ }
+ }else if(( ! is.null(x.range.split)) & is.null(y.range.split)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe), x = list(x1, "x"), y = list(y1, "y"), categ = list("kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_path"), alpha = list(0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data1.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, data1.signif.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data1.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, hframe, data1.incon.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "HORIZ FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, hframe), x = list(x1, x2, "x"), y = list(y1, y2, "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data2.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data2.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2.unknown.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, hframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "HORIZ FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.1, 0.15), v = c(0.75, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 8, title = "DATA2 + DATA2 UNKNOWN DOTS")
+ }
+ }
+ }else if(is.null(x.range.split) & ( ! is.null(y.range.split))){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe), x = list(x1, "x"), y = list(y1, "y"), categ = list("kind", "kind"), legend.name = list("DATASET", "VERT FRAME"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_path"), alpha = list(0.5, 0.5), title = "DATA1", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data1.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe, data1.signif.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "VERT FRAME", "SIGNIF DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), "black"), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA1 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data1.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, vframe, data1.incon.dot), x = list(x1, "x", x1), y = list(y1, "y", y1), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "VERT FRAME", "INCONSISTENT DOTS"), color = list(fun_gg_palette(2)[2], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2), fun_gg_palette(7)[6]), geom = list("geom_point", "geom_path", "geom_point"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA1 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA1\nINCONSISTENT DOTS", text.size = 8, title = "DATA1 + DATA1 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, vframe), x = list(x1, x2, "x"), y = list(y1, y2, "y"), categ = list("kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5), title = "DATA1 + DATA2", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(data2.signif.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.signif.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "SIGNIF DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], "black", rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2 DOTS\nOUTSIDE THE FRAMES", text.size = 8, title = "DATA1 + DATA2 + DATA2 SIGNIFICANT DOTS")
+ }
+ if( ! is.null(data2.incon.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.incon.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "INCONSISTENT DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[6], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 INCONSISTENT DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nINCONSISTENT DOTS", text.size = 8, title = "DATA2 + DATA2 INCONSISTENT DOTS")
+ }
+ if( ! is.null(data2.unknown.dot)){
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ tempo.graph <- fun_gg_scatter(data1 = list(data1, data2, data2.unknown.dot, vframe), x = list(x1, x2, x2, "x"), y = list(y1, y2, y2, "y"), categ = list("kind", "kind", "kind", "kind"), legend.name = list("DATASET", "DATASET", "UNKNOWN DOTS", "VERT FRAME"), color = list(fun_gg_palette(2)[2], fun_gg_palette(2)[1], fun_gg_palette(7)[5], rep(hsv(h = c(0.5, 0.6), v = c(0.9, 1)), 2)), geom = list("geom_point", "geom_point", "geom_point", "geom_path"), alpha = list(0.5, 0.5, 0.5, 0.5), title = "DATA1 + DATA2 + DATA2 UNKNOWN DOTS", x.lim = x.range.plot, y.lim = y.range.plot, raster = raster, return = TRUE)
+ if( ! is.null(tempo.graph$warn)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") FROM fun_gg_scatter():\n", tempo.graph$warn)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(graph.in.file == FALSE){
+ fun_open(pdf = FALSE)
+ }
+ fun_gg_empty_graph(text = "NO PLOT\nBECAUSE\nNO DATA2\nUNKNOWN DOTS", text.size = 8, title = "DATA2 + DATA2 UNKNOWN DOTS")
+ }
+ }
+ }
+ }
+ # end plot
+ if(warn.print == TRUE & ! is.null(warn)){
+ options(warning.length = 8170)
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ tempo.list <- list(data1.removed.row.nb = data1.removed.row.nb, data1.removed.rows = data1.removed.rows, data2.removed.row.nb = data2.removed.row.nb, data2.removed.rows = data2.removed.rows, hframe = hframe, vframe = vframe, data1.signif.dot = data1.signif.dot, data1.non.signif.dot = data1.non.signif.dot, data1.inconsistent.dot = data1.incon.dot, data2.signif.dot = data2.signif.dot, data2.non.signif.dot = data2.non.signif.dot, data2.unknown.dot = data2.unknown.dot, data2.inconsistent.dot = data2.incon.dot, axes = axes, warn = warn)
+ return(tempo.list)
}
-if(line.count== 3L){
-fin.lg.disp[[6]] <- legend.disp[[point.count + line.count]]
-lg.order[[6]] <- point.count + line.count
-lg.color[[6]] <- color[[i1]] # if color == NULL -> NULL
-lg.line.size[[6]] <- line.size[[i1]]
-lg.line.type[[6]] <- line.type[[i1]]
-if(plot == TRUE & fin.lg.disp[[6]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-lg.alpha[[6]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
-}else{
-lg.alpha[[6]] <- alpha[[i1]]
+
+
+################ Import
+
+
+######## fun_pack() #### check if R packages are present and import into the working environment
+
+
+fun_pack <- function(
+ req.package,
+ load = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # check if the specified R packages are present in the computer and import them into the working environment
+ # ARGUMENTS
+ # req.package: character vector of package names to import
+ # load: logical. Load the package into the environement (using library())? Interesting if packages are not in default folders or for checking the functions names of packages using search()
+ # lib.path: optional character vector specifying the absolute pathways of the directories containing some of the listed packages in the req.package argument, if not in the default directories. Ignored if NULL
+ # RETURN
+ # nothing
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_pack(req.package = "nopackage")
+ # fun_pack(req.package = "ggplot2")
+ # fun_pack(req.package = "ggplot2", lib.path = "blablabla")
+ # DEBUGGING
+ # req.package = "ggplot2" ; lib.path = "C:/Program Files/R/R-3.5.1/library"
+ # req.package = "serpentine" ; lib.path = "C:/users/gael/appdata/roaming/python/python36/site-packages"
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = req.package, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = load, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ if(is.null(lib.path)){
+ lib.path <- .libPaths() # .libPaths(new = lib.path) # or .libPaths(new = c(.libPaths(), lib.path))
+ }else{
+ .libPaths(new = sub(x = lib.path, pattern = "/$|\\\\$", replacement = "")) # .libPaths(new = ) add path to default path. BEWARE: .libPaths() does not support / at the end of a submitted path. Thus check and replace last / or \\ in path
+ lib.path <- .libPaths()
+ }
+ tempo <- NULL
+ for(i1 in 1:length(req.package)){
+ if( ! req.package[i1] %in% rownames(utils::installed.packages(lib.loc = lib.path))){
+ tempo <- c(tempo, req.package[i1])
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0(
+ "ERROR IN ",
+ function.name,
+ ": PACKAGE",
+ ifelse(length(tempo) == 1L, paste0("\n\n", tempo, "\n\n"), paste0("S\n", paste(tempo, collapse = "\n"), "\n")),
+ "MUST BE INSTALLED IN",
+ ifelse(length(lib.path) == 1L, "", " ONE OF THESE FOLDERS"),
+ ":\n",
+ paste(lib.path, collapse = "\n")
+ )
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(load == TRUE){
+ for(i2 in 1:length(req.package)){
+ suppressMessages(suppressWarnings(suppressPackageStartupMessages(library(req.package[i2], lib.loc = lib.path, quietly = TRUE, character.only = TRUE))))
+ }
+ }
}
-class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
-for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
-tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
-ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
-"(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
-x[[i1]],
-ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
-y[[i1]],
-if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
-", size = ",
-categ[[i1]],
-"), color = \"",
-color[[i1]][i5],
-"\", linetype = ",
-ifelse(is.numeric(line.type[[i1]]), "", "\""),
-line.type[[i1]],
-ifelse(is.numeric(line.type[[i1]]), "", "\""),
-ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
-ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
-", alpha = ",
-alpha[[i1]],
-", show.legend = FALSE)"
-)))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
-coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+
+
+######## fun_python_pack() #### check if python packages are present
+
+
+fun_python_pack <- function(
+ req.package,
+ python.exec.path = NULL,
+ lib.path = NULL,
+ R.lib.path = NULL
+){
+ # AIM
+ # check if the specified python packages are present in the computer (no import)
+ # WARNINGS
+ # for python 3.7. Previous versions return an error "Error in sys$stdout$flush() : attempt to apply non-function"
+ # ARGUMENTS
+ # req.package: character vector of package names to import
+ # python.exec.path: optional character vector specifying the absolute pathways of the executable python file to use (associated to the packages to use). If NULL, the reticulate::import_from_path() function used in fun_python_pack() seeks for an available version of python.exe, and then uses python_config(python_version, required_module, python_versions). But might not be the correct one for the lib.path parameter specified. Thus, it is recommanded to do not leave NULL, notably when using computing clusters
+ # lib.path: optional character vector specifying the absolute pathways of the directories containing some of the listed packages in the req.package argument, if not in the default directories
+ # R.lib.path: absolute path of the reticulate packages, if not in the default folders
+ # RETURN
+ # nothing
+ # REQUIRED PACKAGES
+ # reticulate
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # fun_pack()
+ # EXAMPLES
+ # example of error message
+ # fun_python_pack(req.package = "nopackage")
+ # example without error message (require the installation of the python serpentine package from https://github.com/koszullab/serpentine
+ # fun_python_pack(req.package = "serpentine", python.exec.path = "C:/ProgramData/Anaconda3/python.exe", lib.path = "c:/programdata/anaconda3/lib/site-packages/")
+ # another example of error message
+ # fun_python_pack(req.package = "serpentine", lib.path = "blablabla")
+ # DEBUGGING
+ # req.package = "serpentine" ; python.exec.path = "C:/ProgramData/Anaconda3/python.exe" ; lib.path = "c:/programdata/anaconda3/lib/site-packages/" ; R.lib.path = NULL
+ # req.package = "bad" ; lib.path = NULL ; R.lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(utils::find("fun_pack", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_pack() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = req.package, class = "character", fun.name = function.name) ; eval(ee)
+ if( ! is.null(python.exec.path)){
+ tempo <- fun_check(data = python.exec.path, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(file.exists(python.exec.path))){ # separation to avoid the problem of tempo$problem == FALSE and python.exec.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": FILE PATH INDICATED IN THE python.exec.path ARGUMENT DOES NOT EXISTS:\n", paste(python.exec.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if( ! is.null(R.lib.path)){
+ tempo <- fun_check(data = R.lib.path, class = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(R.lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and R.lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE R.lib.path ARGUMENT DOES NOT EXISTS:\n", paste(R.lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # package checking
+ fun_pack(req.package = "reticulate", lib.path = R.lib.path)
+ # end package checking
+ # main code
+ if(is.null(python.exec.path)){
+ python.exec.path <- reticulate::py_run_string("
+import sys ;
+path_lib = sys.path
+") # python string
+ python.exec.path <- python.exec.path$path_lib
+ }
+ if(is.null(lib.path)){
+ lib.path <- reticulate::py_run_string("
+import sys ;
+path_lib = sys.path
+") # python string
+ lib.path <- lib.path$path_lib
+ }
+ reticulate::use_python(Sys.which(python.exec.path), required = TRUE) # required to avoid the use of erratic python exec by reticulate::import_from_path()
+ for(i1 in 1:length(req.package)){
+ tempo.try <- vector("list", length = length(lib.path))
+ for(i2 in 1:length(lib.path)){
+ tempo.try[[i2]] <- suppressWarnings(try(reticulate::import_from_path(req.package[i1], path = lib.path[i2]), silent = TRUE))
+ tempo.try[[i2]] <- suppressWarnings(try(reticulate::import_from_path(req.package[i1], path = lib.path[i2]), silent = TRUE)) # done twice to avoid the error message about flushing present the first time but not the second time. see https://stackoverflow.com/questions/57357001/reticulate-1-13-error-in-sysstdoutflush-attempt-to-apply-non-function
+ }
+ if(all(sapply(tempo.try, FUN = grepl, pattern = "[Ee]rror"))){
+ print(tempo.try)
+ tempo.cat <- paste0("ERROR IN ", function.name, ": PACKAGE ", req.package[i1], " MUST BE INSTALLED IN THE MENTIONNED DIRECTORY:\n", paste(lib.path, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ } # else{
+ # suppressMessages(suppressWarnings(suppressPackageStartupMessages(assign(req.package[i1], reticulate::import(req.package[i1]))))) # not required because try() already evaluates
+ # }
+ }
}
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "size", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(line.size[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, breaks reorder the classes according to class.categ in the legend
+
+
+################ Print / Exporting results (text & tables)
+
+
+######## fun_report() #### print string or data object into output file
+
+
+# Problem with 1D tables : names over the table not printed. In addition, see how the 2D tables are printed.
+
+fun_report <- function(
+ data,
+ output = "results.txt",
+ path = "C:/Users/Gael/Desktop/",
+ overwrite = FALSE,
+ rownames.kept = FALSE,
+ vector.cat = FALSE,
+ noquote = TRUE,
+ sep = 2
+){
+ # AIM
+ # log file function: print a character string or a data object into a same output file
+ # ARGUMENTS
+ # data: object to print in the output file. If NULL, nothing is done, with no warning
+ # output: name of the output file
+ # path: location of the output file
+ # overwrite: (logical) if output file already exists, defines if the printing is appended (default FALSE) or if the output file content is erased before printing (TRUE)
+ # rownames.kept: (logical) defines whether row names have to be removed or not in small tables (less than length.rows rows)
+ # vector.cat (logical). If TRUE print a vector of length > 1 using cat() instead of capture.output(). Otherwise (default FALSE) the opposite
+ # noquote: (logical). If TRUE no quote are present for the characters
+ # sep: number of separating lines after printed data (must be integer)
+ # RETURN
+ # nothing
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_report()
+ # fun_report(data = 1:3, output = "results.txt", path = "C:/Users/Gael/Desktop", overwrite = TRUE, rownames.kept = FALSE, vector.cat = FALSE, noquote = FALSE, sep = 2)
+ # DEBUGGING
+ # data = 1:3 ; output = "results.txt" ; path = "C:/Users/Gael/Desktop" ; overwrite = TRUE ; rownames.kept = FALSE ; vector.cat = FALSE ; noquote = FALSE ; sep = 2 # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # argument checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = output, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & output == ""){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": output ARGUMENT AS \"\" DOES NOT CORRESPOND TO A VALID FILE NAME")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo <- fun_check(data = path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": path ARGUMENT DOES NOT CORRESPOND TO EXISTING DIRECTORY\n", paste(path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = overwrite, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = rownames.kept, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = vector.cat, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = noquote, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = sep, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # the 4 next lines are inactivated but kept because at a time, I might have a problem with data (solved with data = NULL). These 4 lines are just to know how to detect a missing argument. Important here because if data is not provided, print the code of the data function
+ # arg.user.list <- as.list(match.call(expand.dots = FALSE))[-1] # recover all the arguments provided by the function user (excluding the argument with defaults values not provided by the user. Thus, it is really the list indicated by the user)
+ # default.arg.list <- formals(fun = sys.function(sys.parent())) # list of all the arguments of the function with their default values (not the values of the user !). It seems that ls() as first line of the function provide the names of the arguments (empty, called, etc., or not)
+ # arg.without.default.value <- sapply(default.arg.list, is.symbol) & sapply(sapply(default.arg.list, as.character), identical, "") # logical to detect argument without default values (these are typeof "symbol" and class "name" and empty character
+ # if( ! all(names(default.arg.list)[arg.without.default.value] %in% names(arg.user.list))){ # test that the arguments with no null values are provided by the user
+ # tempo.cat <- paste0("ERROR IN ", function.name, ": VALUE REQUIRED FOR THESE ARGUMENTS WITH NO DEFAULTS VALUES: ", paste(names(default.arg.list)[arg.without.default.value][ ! names(default.arg.list)[arg.without.default.value] %in% names(arg.user.list)], collapse = " "))
+ # stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ # }
+ # end argument checking
+ # main code
+ if( ! is.null(data)){
+ if(all(class(data) == "data.frame") | all(class(data) == "table") | all(class(data) %in% c("matrix", "array"))){ # before R4.0.0, it was all(class(data) %in% c("matrix", "data.frame", "table"))
+ if(rownames.kept == FALSE & all(class(data) == "data.frame") & nrow(data) != 0 & nrow(data) <= 4){ # for data frames with nrows <= 4
+ rownames.output.tables <- ""
+ length.rows <- nrow(data)
+ for(i in 1:length.rows){ # replace the rownames of the first 4 rows by increasing number of spaces (because identical row names not allowed in data frames). This method cannot be extended to more rows as the printed data frame is shifted on the right because of "big empty rownames"
+ rownames.output.tables <- c(rownames.output.tables, paste0(rownames.output.tables[i]," ", collapse=""))
+ }
+ row.names(data) <- rownames.output.tables[1:length.rows]
+ }else if(rownames.kept == FALSE & (all(class(data) == "table") | all(class(data) %in% c("matrix", "array")))){ # before R4.0.0, it was & all(class(data) %in% c("matrix", "table"))
+ rownames(data) <- rep("", nrow(data)) # identical row names allowed in matrices and tables
+ }
+ if(noquote == TRUE){
+ utils::capture.output(noquote(data), file=paste0(path, "/", output), append = ! overwrite)
+ }else{
+ utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
+ }
+ }else if(is.vector(data) & all(class(data) != "list") & (length(data) == 1L | vector.cat == TRUE)){
+ if(noquote == TRUE){
+ cat(noquote(data), file= paste0(path, "/", output), append = ! overwrite)
+ }else{
+ cat(data, file= paste0(path, "/", output), append = ! overwrite)
+ }
+ }else if(all(base::mode(data) == "character")){ # characters (array, list, factor or vector with vector.cat = FALSE)
+ if(noquote == TRUE){
+ utils::capture.output(noquote(data), file=paste0(path, "/", output), append = ! overwrite)
+ }else{
+ utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
+ }
+ }else{ # other object (S4 for instance, which do not like noquote()
+ utils::capture.output(data, file=paste0(path, "/", output), append = ! overwrite)
+ }
+ sep.final <- paste0(rep("\n", sep), collapse = "")
+ write(sep.final, file= paste0(path, "/", output), append = TRUE) # add a sep
+ }
}
+
+
+######## fun_get_message() #### return error/warning/other messages of an expression (that can be exported)
+
+
+fun_get_message <- function(
+ data,
+ kind = "error",
+ header = TRUE,
+ print.no = FALSE,
+ text = NULL,
+ env = NULL
+){
+ # AIM
+ # evaluate an instruction written between "" and return the first of the error, or warning or standard (non error non warning) messages if ever exist
+ # using argument print.no = FALSE, return NULL if no message, which is convenient in some cases
+ # WARNINGS
+ # Only the first message is returned
+ # Always use the env argument when fun_get_message() is used inside functions
+ # The function does not prevent printing if print() is used inside the instruction tested. To prevent that, use tempo <- capture.output(error <- fun_get_message(data = "fun_check(data = 'a', class = mean, neg.values = FALSE, print = TRUE)")). The return of fun_get_message() is assigned into error and the printed messages are captured by capture.output() and assigned into tempo. See the examples
+ # ARGUMENTS
+ # data: character string to evaluate
+ # kind: character string. Either "error" to get error messages, or "warning" to get warning messages, or "message" to get non error and non warning messages
+ # header: logical. Add a header in the returned message?
+ # print.no: logical. Print a message saying that no message reported?
+ # text: character string added to the output message (even if no message exists and print.no is TRUE). Inactivated if header is FALSE
+ # env: the name of an existing environment. NULL if not required
+ # RETURN
+ # the message or NULL if no message and print.no is FALSE
+ # REQUIRED PACKAGES
+ # none
+ # REQUIRED FUNCTIONS FROM CUTE_LITTLE_R_FUNCTION
+ # fun_check()
+ # EXAMPLES
+ # fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "error", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "warning", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)", kind = "message", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "wilcox.test()", kind = "error", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "sum(1)", kind = "error", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "message('ahah')", kind = "error", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "message('ahah')", kind = "message", print.no = TRUE, text = "IN A")
+ # fun_get_message(data = "ggplot2::ggplot(data = data.frame(X = 1:10, stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()", kind = "message", print.no = TRUE, text = "IN FUNCTION 1")
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; fun_get_message(data = 'fun_gg_boxplot(data = obs1, y = "Time", categ = "Group1")', kind = "message", print.no = TRUE, text = "IN FUNCTION 1")
+ # DEBUGGING
+ # data = "wilcox.test(c(1,1,3), c(1, 2, 4), paired = TRUE)" ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
+ # data = "sum(1)" ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Group1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; data = 'fun_gg_boxplot(data1 = obs1, y = "Time", categ = "Group1")' ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
+ # data = "message('ahah')" ; kind = "error" ; header = TRUE ; print.no = TRUE ; text = "IN A" ; env = NULL
+ # data = 'ggplot2::ggplot(data = data.frame(X = "a", stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()' ; kind = "message" ; header = TRUE ; print.no = FALSE ; text = NULL # for function debugging
+ # data = 'ggplot2::ggplot(data = data.frame(X = "a", stringsAsFactors = TRUE), mapping = ggplot2::aes(x = X)) + ggplot2::geom_histogram()' ; kind = "warning" ; header = TRUE ; print.no = FALSE ; text = NULL # for function debugging
+ # data = "emmeans::emmeans(object = emm.rg, specs = contrast.var)" ; kind = "message" ; header = TRUE ; print.no = FALSE ; text = NULL ; env = NULL # for function debugging
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ # end function name
+ # required function checking
+ if(length(utils::find("fun_check", mode = "function")) == 0L){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": REQUIRED fun_check() FUNCTION IS MISSING IN THE R ENVIRONMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # no need to use reserved words to avoid bugs, because it is local, and exists("tempo.warning", inherit = FALSE), never use the scope
+ # argument checking
+ # argument checking with fun_check()
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = kind, options = c("error", "warning", "message"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = print.no, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = header, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(text)){
+ tempo <- fun_check(data = text, class = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ if( ! is.null(env)){
+ tempo <- fun_check(data = env, class = "environment", fun.name = function.name) ; eval(ee) #
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # end argument checking with fun_check()
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument checking
+ # main code
+ pdf(file = NULL) # send plots into a NULL file, no pdf file created
+ window.nb <- dev.cur()
+ invisible(dev.set(window.nb))
+ # last warning cannot be used because suppressWarnings() does not modify last.warning present in the base evironment (created at first warning in a new R session), or warnings() # to reset the warning history : unlockBinding("last.warning", baseenv()) ; assign("last.warning", NULL, envir = baseenv())
+ output <- NULL
+ tempo.error <- try(suppressMessages(suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))), silent = TRUE) # get error message, not warning or messages
+ if(any(class(tempo.error) %in% c("gg", "ggplot"))){
+ tempo.error <- try(suppressMessages(suppressWarnings(ggplot2::ggplot_build(tempo.error))), silent = TRUE)[1]
+ }
+ if(exists("tempo.error", inherit = FALSE) == TRUE){ # inherit = FALSE avoid the portee lexical and thus the declared word
+ if( ! all(class(tempo.error) == "try-error")){ # deal with NULL and S4 objects. Old code: ! (all(class(tempo.error) == "try-error") & any(grepl(x = tempo.error, pattern = "^Error|^error|^ERROR"))) but problem with S4 objects. Old code : if((length(tempo.error) > 0 & ! any(grepl(x = tempo.error, pattern = "^Error|^error|^ERROR"))) | (length(tempo.error) == 0) ){ but problem when tempo.error is a list but added this did not work: | ! all(class(tempo.error) == "character")
+ tempo.error <- NULL
+ }
+ }else{
+ tempo.error <- NULL
+ }
+ if(kind == "error" & ! is.null(tempo.error)){ #
+ if(header == TRUE){
+ tempo.error[1] <- gsub(x = tempo.error[1], pattern = "^Error i|^error i|^ERROR I", replacement = "I")
+ output <- paste0("ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.error[1]) #
+ }else{
+ output <- tempo.error[1] #
+ }
+ }else if(kind == "error" & is.null(tempo.error) & print.no == TRUE){
+ output <- paste0("NO ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ }else if(kind != "error" & ( ! is.null(tempo.error)) & print.no == TRUE){
+ output <- paste0("NO ", ifelse(kind == "warning", "WARNING", "STANDARD (NON ERROR AND NON WARNING)"), " MESSAGE BECAUSE OF ERROR MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ }else if(is.null(tempo.error)){
+ fun.warning.capture <- function(expr){
+ # from demo(error.catching) typed in the R console, coming from ?tryCatch
+ # see also http://mazamascience.com/WorkingWithData/?p=912
+ # return a character string or NULL
+ # expr <- wilcox.test.default(c(1, 1, 3), c(1, 2, 4), paired = TRUE)
+ W <- NULL
+ w.handler <- function(w){ # warning handler
+ W <<- w # send to the above env, i.e., the inside of the fun.warning.capture function
+ invokeRestart("muffleWarning") # here w.handler() muffles all the warnings. See http://romainfrancois.blog.free.fr/index.php?post/2009/05/20/Disable-specific-warnings to muffle specific warnings and print others
+ }
+ output <- list(
+ value = suppressMessages(withCallingHandlers(tryCatch(expr, error = function(e){e}), warning = w.handler)), # BEWARE: w.handler is a function written without (), like in other functions with FUN argument
+ warning = W # processed by w.handler()
+ )
+ return(if(is.null(output$warning)){NULL}else{as.character(output$warning)})
+ }
+ tempo.warn <- fun.warning.capture(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))
+ # warn.options.ini <- options()$warn ; options(warn = 1) ; tempo.warn <- utils::capture.output({tempo <- suppressMessages(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))}, type = "message") ; options(warn = warn.options.ini) # this recover warnings not messages and not errors but does not work in all enviroments
+ tempo.message <- utils::capture.output({
+ tempo <- suppressMessages(suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env})))
+ if(any(class(tempo) %in% c("gg", "ggplot"))){
+ tempo <- ggplot2::ggplot_build(tempo)
+ }else{
+ tempo <- suppressWarnings(eval(parse(text = data), envir = if(is.null(env)){parent.frame()}else{env}))
+ }
+ }, type = "message") # recover messages not warnings and not errors
+ if(kind == "warning" & ! is.null(tempo.warn)){
+ if(length(tempo.warn) > 0){ # to avoid character(0)
+ if( ! any(sapply(tempo.warn, FUN = "grepl", pattern = "() FUNCTION:$"))){
+ tempo.warn <- paste(unique(tempo.warn), collapse = "\n") # if FALSE, means that the tested data is a special function. If TRUE, means that the data is a standard function. In that case, the output of capture.output() is two strings per warning messages: if several warning messages -> identical first string, which is removed in next messages by unique()
+ }else{
+ tempo.warn <- paste(tempo.warn, collapse = "\n")
+ }
+ if(header == TRUE){
+ if(any(grepl(x = tempo.warn[[1]], pattern = "^simpleWarning i"))){
+ tempo.warn[[1]] <- gsub(x = tempo.warn[[1]], pattern = "^Warning i", replacement = "I")
+ }
+ if(any(grepl(x = tempo.warn[[1]], pattern = "^Warning i"))){
+ tempo.warn[[1]] <- gsub(x = tempo.warn[[1]], pattern = "^Warning i", replacement = "I")
+ }
+ output <- paste0("WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.warn) #
+ }else{
+ output <- tempo.warn #
+ }
+ }else{
+ if(print.no == TRUE){
+ output <- paste0("NO WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ } # no need else{} here because output is already NULL at first
+ }
+ }else if(kind == "warning" & is.null(tempo.warn) & print.no == TRUE){
+ output <- paste0("NO WARNING MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ }else if(kind == "message" & exists("tempo.message", inherit = FALSE) == TRUE){ # inherit = FALSE avoid the portee lexical and thus the declared word
+ if(length(tempo.message) > 0){ # if something is returned by capture.ouptput() (only in this env) with a length more than 1
+ if(header == TRUE){
+ output <- paste0("STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text, ":\n", tempo.message) #
+ }else{
+ output <- tempo.message #
+ }
+ }else{
+ if(print.no == TRUE){
+ output <- paste0("NO STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ } # no need else{} here because output is already NULL at first
+ }
+ }else if(kind == "message" & exists("tempo.message", inherit = FALSE) == FALSE & print.no == TRUE){
+ output <- paste0("NO STANDARD (NON ERROR AND NON WARNING) MESSAGE REPORTED", ifelse(is.null(text), "", " "), text)
+ } # no need else{} here because output is already NULL at first
+ } # no need else{} here because output is already NULL at first
+ invisible(dev.off(window.nb)) # end send plots into a NULL file
+ return(output) # do not use cat() because the idea is to reuse the message
}
+
+
+
+
+
+
+
+# Error: class order not good when a class is removed due to NA
+# Error: line 136 in check 20201126 with add argument
+# Solve this: sometimes error messages can be more than the max display (8170). Thus, check every paste0("ERROR IN ", function.name, and trunck the message if to big. In addition, add at the begining of the warning message that it is too long and see the $warn output for complete message. Add also this into fun_scatter
+# add dot.shape ? See with available aesthetic layers
+# rasterise: https://cran.r-project.org/web/packages/ggrastr/vignettes/Raster_geoms.html
+# add horizontal argument and deal any conflict with vertical argument. Start with horizontal = NULL as default. If ! is.null() -> convert vertical if required
+# time for excecution : microbenchmark package. See also in RStudio time per line of code
+
+fun_gg_boxplot <- function(
+ data1,
+ y,
+ categ,
+ categ.class.order = NULL,
+ categ.color = NULL,
+ box.legend.name = NULL,
+ box.fill = FALSE,
+ box.width = 0.5,
+ box.space = 0.1,
+ box.line.size = 0.75,
+ box.notch = FALSE,
+ box.alpha = 1,
+ box.mean = TRUE,
+ box.whisker.kind = "std",
+ box.whisker.width = 0,
+ dot.color = grey(0.25),
+ dot.categ = NULL,
+ dot.categ.class.order = NULL,
+ dot.legend.name = NULL,
+ dot.tidy = FALSE,
+ dot.tidy.bin.nb = 50,
+ dot.jitter = 0.5,
+ dot.seed = 2,
+ dot.size = 3,
+ dot.alpha = 0.5,
+ dot.border.size = 0.5,
+ dot.border.color = NULL,
+ x.lab = NULL,
+ x.angle = 0,
+ y.lab = NULL,
+ y.lim = NULL,
+ y.log = "no",
+ y.tick.nb = NULL,
+ y.second.tick.nb = 1,
+ y.include.zero = FALSE,
+ y.top.extra.margin = 0.05,
+ y.bottom.extra.margin = 0.05,
+ stat.pos = "top",
+ stat.mean = FALSE,
+ stat.size = 4,
+ stat.dist = 5,
+ stat.angle = 0,
+ vertical = TRUE,
+ text.size = 12,
+ title = "",
+ title.text.size = 8,
+ legend.show = TRUE,
+ legend.width = 0.5,
+ article = TRUE,
+ grid = FALSE,
+ add = NULL,
+ return = FALSE,
+ return.ggplot = FALSE,
+ return.gtable = TRUE,
+ plot = TRUE,
+ warn.print = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # Plot ggplot2 boxplots + dots + means
+ # For ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
+ # WARNINGS
+ # Rows containing NA in data1[, c(y, categ)] will be removed before processing, with a warning (see below)
+ # Hinges are not computed like in the classical boxplot() function of R. See https://ggplot2.tidyverse.org/reference/geom_boxplot.html
+ # To have a single box, please create a factor column with a single class and specify the name of this column in the categ argument. For a single set of grouped boxes, create a factor column with a single class and specify this column in categ argument as first element (i.e., as categ1, knowing that categ2 must also be specified in this situation). See categ argument below
+ # The dot.alpha argument can alter the display of the color boxes when using pdf output
+ # Size arguments (box.line.size, dot.size, dot.border.size, stat.size, text.size and title.text.size) are in mm. See Hadley comment in https://stackoverflow.com/questions/17311917/ggplot2-the-unit-of-size. See also http://sape.inf.usi.ch/quick-reference/ggplot2/size). Unit object are not accepted, but conversion can be used (e.g., grid::convertUnit(grid::unit(0.2, "inches"), "mm", valueOnly = TRUE))
+ # Display seems to be done twice on Windows devices (like a blink). However, no double plots on pdf devices. Thus, the blink remains mysterious
+ # To remove boxes and have only dots, use box.alpha = 0
+ # ARGUMENTS
+ # data1: data frame containing one column of quantitative values (see the y argument below) and one or two columns of categories (see the categ argument below). Duplicated column names are not allowed
+ # y: character string of the data1 column name for y-axis (column containing numeric values). Numeric values will be split according to the classes of the column names indicated in the categ argument to generate the boxes and will also be used to plot the dots
+ # categ: vector of character strings of the data1 column name for categories (column of characters or factors). Must be either one or two column names. If a single column name (further referred to as categ1), then one box per class of categ1. If two column names (further referred to as categ1 and categ2), then one box per class of categ2, which form a group of boxes in each class of categ1. WARNING: no empty classes allowed. To have a single box, create a factor column with a single class and specify the name of this column in the categ argument (here, no categ2 in categ argument). For a single set of grouped boxes, create a factor column with a single class and specify this column in categ argument as first element (i.e., as categ1), in addition to the already used category (as categ2 in this situation)
+ # categ.class.order: list indicating the order of the classes of categ1 and categ2 represented on the boxplot (the first compartment for categ1 and and the second for categ2). If categ.class.order == NULL, classes are represented according to the alphabetical order. Some compartments can be NULL and others not. See the categ argument for categ1 and categ2 description
+ # categ.color: vector of color character string for box frames (see the categ argument for categ1 and categ2 description)
+ # If categ.color == NULL, default colors of ggplot2, whatever categ1 and categ2
+ # If categ.color is non-null and only categ1 in categ argument, categ.color can be either:
+ # (1) a single color string. All the boxes will have this color, whatever the number of classes of categ1
+ # (2) a vector of string colors, one for each class of categ1. Each color will be associated according to categ.class.order of categ1
+ # (3) a vector or factor of string colors, like if it was one of the column of data1 data frame. WARNING: a single color per class of categ1 and a single class of categ1 per color must be respected
+ # Color functions, like grey(), hsv(), etc., are also accepted
+ # Positive integers are also accepted instead of character strings, as long as above rules about length are respected. Integers will be processed by fun_gg_palette() using the maximal integer value among all the integers in categ.color (see fun_gg_palette())
+ # If categ.color is non-null and categ1 and categ2 are specified, all the rules described above will apply to categ2 instead of categ1 (colors will be determined for boxes inside a group of boxes)
+ # box.legend.name: character string of the legend title. If box.legend.name is NULL, then box.legend.name <- categ1 if only categ1 is present, and box.legend.name <- categ2 if categ1 and categ2 are present in the categ argument. Write "" if no legend required. See the categ argument for categ1 and categ2 description
+ # box.fill: logical. Fill the box? If TRUE, the categ.color argument will be used to generate filled boxplots (the box frames being black) as well as filled outlier dots (the dot border being controlled by the dot.border.color argument). If all the dots are plotted (argument dot.color other than NULL), they will be over the boxes. If FALSE, the categ.color argument will be used to color the box frames and the outlier dot borders. If all the dots are plotted, they will be beneath the boxes
+ # box.width: single numeric value (from 0 to 1) of width of either boxes or group of boxes
+ # When categ argument has a single categ1 element (i.e., separate boxes. See the categ argument for categ1 and categ2 description), then each class of categ1 is represented by a single box. In that case, box.width argument defines each box width, from 0 (no box width) to 1 (max box width), but also the space between boxes (the code uses 1 - box.width for the box spaces). Of note, xmin and xmax of the fun_gg_boxplot() output report the box boundaries (around x-axis unit 1, 2, 3, etc., for each box)
+ # When categ argument has a two categ1 and categ2 elements (i.e., grouped boxes), box.width argument defines the width allocated for each set of grouped boxes, from 0 (no group width) to 1 (max group width), but also the space between grouped boxes (the code uses 1 - box.width for the spaces). Of note, xmin and xmax of the fun_gg_boxplot() output report the box boundaries (around x-axis unit 1, 2, 3, etc., for each set of grouped box)
+ # box.space: single numeric value (from 0 to 1) indicating the box separation inside grouped boxes, when categ argument has a two categ1 and categ2 elements. 0 means no space and 1 means boxes shrunk to a vertical line. Ignored if categ argument has a single categ1 element
+ # box.line.size: single numeric value of line width of boxes and whiskers in mm
+ # box.notch: logical. Notched boxplot? It TRUE, display notched boxplot, notches corresponding approximately to the 95% confidence interval of the median (the notch interval is exactly 1.58 x Inter Quartile Range (IQR) / sqrt(n), with n the number of values that made the box). If notch intervals between two boxes do not overlap, it can be interpreted as significant median differences
+ # box.alpha: single numeric value (from 0 to 1) of box transparency (full transparent to full opaque, respectively). To remove boxplots, use box.alpha = 0
+ # box.mean: logical. Add mean value? If TRUE, a diamond-shaped dot, with the horizontal diagonal corresponding to the mean value, is displayed over each boxplot
+ # box.whisker.kind: range of the whiskers. Either "no" (no whiskers), or "std" (length of each whisker equal to 1.5 x Inter Quartile Range (IQR)), or "max" (length of the whiskers up or down to the most distant dot)
+ # box.whisker.width: single numeric value (from 0 to 1) of the whisker width, with 0 meaning no whiskers and 1 meaning a width equal to the box width
+ # dot.color: vector of color character string ruling the dot colors and the dot display. See the example section below for easier understanding of the rules described here
+ # If NULL, no dots plotted
+ # If "same", the dots will have the same colors as the respective boxplots
+ # Otherwise, as in the rule (1), (2) or (3) described in the categ.color argument, except that in the possibility (3), the rule "a single color per class of categ and a single class of categ per color", does not have to be respected (for instance, each dot can have a different color). Colors will also depend on the dot.categ argument. If dot.categ is NULL, then colors will be applied to each class of the last column name specified in categ. If dot.categ is non-NULL, colors will be applied to each class of the column name specified in dot.categ. See examples
+ # dot.categ: optional single character string of a column name (further referred to as categ3) of the data1 argument. This column of data1 will be used to generate a legend for dots, in addition to the legend for boxes. See the dot.color argument for details about the way the legend is built using the two dot.categ and dot.color arguments. If NULL, no legend created and the colors of dots will depend on dot.color and categ arguments (as explained in the dot.color argument)
+ # dot.categ.class.order: optional vector of character strings indicating the order of the classes of categ3 (see the dot.categ argument). If dot.categ is non-NULL and dot.categ.class.order is NULL, classes are displayed in the legend according to the alphabetical order. Ignored if dot.categ is NULL
+ # dot.legend.name: optional character string of the legend title for categ3 (see the dot.categ argument). If dot.legend.name == NULL, dot.categ value is used (name of the column in data1). Write "" if no legend required. Ignored if dot.categ is NULL
+ # dot.tidy: logical. Nice dot spreading? If TRUE, use the geom_dotplot() function for a nice representation. WARNING: change the true quantitative coordinates of dots (i.e., y-axis values for vertical display) because of binning. Thus, the gain in aestheticism is associated with a loss in precision that can be very important. If FALSE, dots are randomly spread on the qualitative axis, using the dot.jitter argument (see below) keeping the true quantitative coordinates
+ # dot.tidy.bin.nb: positive integer indicating the number of bins (i.e., nb of separations) of the y.lim range. Each dot will then be put in one of the bin, with a diameter of the width of the bin. In other words, increase the number of bins to have smaller dots. Not considered if dot.tidy is FALSE
+ # dot.jitter: numeric value (from 0 to 1) of random dot horizontal dispersion (for vertical display), with 0 meaning no dispersion and 1 meaning dispersion in the corresponding box width interval. Not considered if dot.tidy is TRUE
+ # dot.seed: integer value that set the random seed. Using the same number will generate the same dot jittering. Write NULL to have different jittering each time the same instruction is run. Ignored if dot.tidy is TRUE
+ # dot.size: numeric value of dot diameter in mm. Not considered if dot.tidy is TRUE
+ # dot.alpha: numeric value (from 0 to 1) of dot transparency (full transparent to full opaque, respectively)
+ # dot.border.size: numeric value of border dot width in mm. Write zero for no dot border. If dot.tidy is TRUE, value 0 remove the border and other values leave the border without size control (geom_doplot() feature)
+ # dot.border.color: single character color string defining the color of the dot border (same color for all the dots, whatever their categories). If dot.border.color == NULL, the border color will be the same as the dot color. A single integer is also accepted instead of a character string, that will be processed by fun_gg_palette()
+ # x.lab: a character string or expression for x-axis legend. If NULL, character string of categ1 (see the categ argument for categ1 and categ2 description)
+ # x.angle: integer value of the text angle for the x-axis numbers, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
+ # y.lab: a character string or expression for y-axis legend. If NULL, character string of the y argument
+ # y.lim: 2 numeric values indicating the range of the y-axis. Order matters (for inverted axis). If NULL, the range of the x column name of data1 will be used.
+ # y.log: either "no", "log2" (values in the y argument column of the data1 data frame will be log2 transformed and y-axis will be log2 scaled) or "log10" (values in the y argument column of the data1 data frame will be log10 transformed and y-axis will be log10 scaled). WARNING: not possible to have horizontal boxes with a log axis, due to a bug in ggplot2 (see https://github.com/tidyverse/ggplot2/issues/881)
+ # y.tick.nb: approximate number of desired values labeling the y-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if y.log is "no", then the number of labeling values is set by ggplot2. If NULL and if y.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the y.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for y.lim = c(9, 1200)). WARNING: if non-NULL and if y.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
+ # y.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if y.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$y.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
+ # y.include.zero: logical. Does y.lim range include 0? Ignored if y.log is "log2" or "log10"
+ # y.top.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to y.lim. If different from 0, add the range of the axis multiplied by y.top.extra.margin (e.g., abs(y.lim[2] - y.lim[1]) * y.top.extra.margin) to the top of y-axis
+ # y.bottom.extra.margin: idem as y.top.extra.margin but to the bottom of y-axis
+ # stat.pos: add the median number above the corresponding box. Either NULL (no number shown), "top" (at the top of the plot region) or "above" (above each box)
+ # stat.mean: logical. Display mean numbers instead of median numbers? Ignored if stat.pos is NULL
+ # stat.size: numeric value of the stat font size in mm. Ignored if stat.pos is NULL
+ # stat.dist: numeric value of the stat distance in percentage of the y-axis range (stat.dist = 5 means move the number displayed at 5% of the y-axis range). Ignored if stat.pos is NULL or "top"
+ # stat.angle: integer value of the angle of stat, using the same rules as in ggplot2. Positive values for counterclockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Negative values for clockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
+ # vertical: logical. Vertical boxes? WARNING: will be automatically set to TRUE if y.log argument is other than "no". Indeed, not possible to have horizontal boxes with a log axis, due to a bug in ggplot2 (see https://github.com/tidyverse/ggplot2/issues/881)
+ # text.size: numeric value of the font size of the (1) axis numbers, (2) axis labels and (3) texts in the graphic legend (in mm)
+ # title: character string of the graph title
+ # title.text.size: numeric value of the title font size in mm
+ # legend.show: logical. Show legend? Not considered if categ argument is NULL, because this already generate no legend, excepted if legend.width argument is non-NULL. In that specific case (categ is NULL, legend.show is TRUE and legend.width is non-NULL), an empty legend space is created. This can be useful when desiring graphs of exactly the same width, whatever they have legends or not
+ # legend.width: single proportion (between 0 and 1) indicating the relative width of the legend sector (on the right of the plot) relative to the width of the plot. Value 1 means that the window device width is split in 2, half for the plot and half for the legend. Value 0 means no room for the legend, which will overlay the plot region. Write NULL to inactivate the legend sector. In such case, ggplot2 will manage the room required for the legend display, meaning that the width of the plotting region can vary between graphs, depending on the text in the legend
+ # article: logical. If TRUE, use an article theme (article like). If FALSE, use a classic related ggplot theme. Use the add argument (e.g., add = "+ggplot2::theme_classic()" for the exact classic ggplot theme
+ # grid: logical. Draw lines in the background to better read the box values? Not considered if article == FALSE (grid systematically present)
+ # add: character string allowing to add more ggplot2 features (dots, lines, themes, facet, etc.). Ignored if NULL
+ # WARNING: (1) the string must start with "+", (2) the string must finish with ")" and (3) each function must be preceded by "ggplot2::". Example: "+ ggplot2::coord_flip() + ggplot2::theme_bw()"
+ # If the character string contains the "ggplot2::theme" string, then the article argument of fun_gg_boxplot() (see above) is ignored with a warning. In addition, some arguments can be overwritten, like x.angle (check all the arguments)
+ # Handle the add argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
+ # WARNING: the call of objects inside the quotes of add can lead to an error if the name of these objects are some of the fun_gg_boxplot() arguments. Indeed, the function will use the internal argument instead of the global environment object. Example article <- "a" in the working environment and add = '+ ggplot2::ggtitle(article)'. The risk here is to have TRUE as title. To solve this, use add = '+ ggplot2::ggtitle(get("article", envir = .GlobalEnv))'
+ # return: logical. Return the graph parameters?
+ # return.ggplot: logical. Return the ggplot object in the output list? Ignored if return argument is FALSE. WARNING: always assign the fun_gg_boxplot() function (e.g., a <- fun_gg_boxplot()) if return.ggplot argument is TRUE, otherwise, double plotting is performed. See $ggplot in the RETURN section below for more details
+ # return.gtable: logical. Return the ggplot object as gtable of grobs in the output list? Ignored if plot argument is FALSE. Indeed, the graph must be plotted to get the grobs dispositions. See $gtable in the RETURN section below for more details
+ # plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
+ # warn.print: logical. Print warnings at the end of the execution? ? If FALSE, warning messages are never printed, but can still be recovered in the returned list. Some of the warning messages (those delivered by the internal ggplot2 functions) are not apparent when using the argument plot = FALSE
+ # lib.path: character string indicating the absolute path of the required packages (see below). if NULL, the function will use the R library default folders
+ # RETURN
+ # A boxplot if plot argument is TRUE
+ # A list of the graph info if return argument is TRUE:
+ # $data: the initial data with graphic information added
+ # $stat: the graphic statistics (mostly equivalent to ggplot_build()$data[[2]])
+ # $removed.row.nb: which rows have been removed due to NA/Inf detection in y and categ columns (NULL if no row removed)
+ # $removed.rows: removed rows (NULL if no row removed)
+ # $plot: the graphic box and dot coordinates
+ # $dots: dot coordinates
+ # $main.box: coordinates of boxes
+ # $median: median coordinates
+ # $sup.whisker: coordinates of top whiskers (y for base and y.end for extremities)
+ # $inf.whisker: coordinates of bottom whiskers (y for base and y.end for extremities)
+ # $sup.whisker.edge: coordinates of top whisker edges (x and xend)
+ # $inf.whisker.edge: coordinates of bottom whisker edges(x and xend)
+ # $mean: diamond mean coordinates (only if box.mean argument is TRUE)
+ # $stat.pos: coordinates of stat numbers (only if stat.pos argument is not NULL)
+ # y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non-NULL or if y.log argument is different from "no")
+ # y.second.tick.values: values of secondary ticks. NULL except if y.second.tick.nb argument is non-NULL or if y.log argument is different from "no")
+ # $panel: the variable names used for the panels (NULL if no panels). WARNING: NA can be present according to ggplot2 upgrade to v3.3.0
+ # $axes: the x-axis and y-axis info
+ # $warn: the warning messages. Use cat() for proper display. NULL if no warning. WARNING: warning messages delivered by the internal ggplot2 functions are not apparent when using the argument plot = FALSE
+ # $ggplot: ggplot object that can be used for reprint (use print(...$ggplot) or update (use ...$ggplot + ggplot2::...). NULL if return.ggplot argument is FALSE. Of note, a non-NULL $ggplot in the output list is sometimes annoying as the manipulation of this list prints the plot
+ # $gtable: gtable object that can be used for reprint (use gridExtra::grid.arrange(...$ggplot) or with additionnal grobs (see the grob decomposition in the examples). NULL if return.ggplot argument is FALSE. Contrary to $ggplot, a non-NULL $gtable in the output list is not annoying as the manipulation of this list does not print the plot
+ # REQUIRED PACKAGES
+ # ggplot2
+ # gridExtra
+ # lemon (in case of use in the add argument)
+ # scales
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # fun_check()
+ # fun_comp_1d()
+ # fun_comp_2d()
+ # fun_gg_just()
+ # fun_gg_palette()
+ # fun_inter_ticks()
+ # fun_name_change()
+ # fun_pack()
+ # fun_round()
+ # fun_scale()
+ # EXAMPLE
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(20, 100, 10), rnorm(20, 200, 50), rnorm(20, 500, 60), rnorm(20, 100, 50)), Categ1 = rep(c("CAT", "DOG"), times = 40), Categ2 = rep(c("A", "B", "C", "D"), each = 20), Color1 = rep(c("coral", "lightblue"), times = 40), Color2 = rep(c("#9F2108", "#306100", "#007479", "#8500C0"), each = 20), stringsAsFactors = TRUE) ; set.seed(NULL) ; fun_gg_boxplot(data1 = obs1, y = "Time", categ = "Categ1")
+ # see http
+ # DEBUGGING
+ # set.seed(1) ; obs1 <- data.frame(Time = c(rnorm(10), rnorm(10) + 2), Categ1 = rep(c("G", "H"), each = 10), stringsAsFactors = TRUE) ; set.seed(NULL) ; obs1$Time[1:10] <- NA ; data1 = obs1 ; y = "Time" ; categ = c("Categ1") ; categ.class.order = NULL ; categ.color = NULL ; box.legend.name = NULL ; box.fill = FALSE ; box.width = 0.5 ; box.space = 0.1 ; box.line.size = 0.75 ; box.notch = FALSE ; box.alpha = 1 ; box.mean = TRUE ; box.whisker.kind = "std" ; box.whisker.width = 0 ; dot.color = grey(0.25) ; dot.categ = NULL ; dot.categ.class.order = NULL ; dot.legend.name = NULL ; dot.tidy = FALSE ; dot.tidy.bin.nb = 50 ; dot.jitter = 0.5 ; dot.seed = 2 ; dot.size = 3 ; dot.alpha = 0.5 ; dot.border.size = 0.5 ; dot.border.color = NULL ; x.lab = NULL ; x.angle = 0 ; y.lab = NULL ; y.lim = NULL ; y.log = "no" ; y.tick.nb = NULL ; y.second.tick.nb = 1 ; y.include.zero = FALSE ; y.top.extra.margin = 0.05 ; y.bottom.extra.margin = 0.05 ; stat.pos = "top" ; stat.mean = FALSE ; stat.size = 4 ; stat.dist = 5 ; stat.angle = 0 ; vertical = TRUE ; text.size = 12 ; title = "" ; title.text.size = 8 ; legend.show = TRUE ; legend.width = 0.5 ; article = TRUE ; grid = FALSE ; add = NULL ; return = FALSE ; return.ggplot = FALSE ; return.gtable = TRUE ; plot = TRUE ; warn.print = FALSE ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots = FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots = FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_comp_2d",
+ "fun_gg_just",
+ "fun_gg_palette",
+ "fun_name_change",
+ "fun_pack",
+ "fun_check",
+ "fun_round",
+ "fun_scale",
+ "fun_inter_ticks"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function")) == 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words to avoid bugs (names of dataframe columns used in this function)
+ reserved.words <- c("categ.check", "categ.color", "dot.color", "dot.categ", "dot.max", "dot.min", "group", "PANEL", "group.check", "MEAN", "tempo.categ1", "tempo.categ2", "text.max.pos", "text.min.pos", "x", "x.y", "y", "y.check", "y_from.dot.max", "ymax", "tidy_group", "binwidth")
+ # end reserved words to avoid bugs (used in this function)
+ # arg with no default values
+ mandat.args <- c(
+ "data1",
+ "y",
+ "categ"
+ )
+ tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), ")")))
+ if(any(tempo)){ # normally no NA for missing() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", "HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = categ, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if( ! is.null(categ.class.order)){
+ tempo <- fun_check(data = categ.class.order, class = "list", fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = categ.class.order, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(box.legend.name)){
+ tempo <- fun_check(data = box.legend.name, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = box.legend.name, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(categ.color)){
+ tempo1 <- fun_check(data = categ.color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = categ.color, class = "factor", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.check.color <- fun_check(data = categ.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, neg.values = FALSE, fun.name = function.name)$problem
+ if(tempo.check.color == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(any(categ.color == 0L, na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = categ.color, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = box.fill, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.space, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.line.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.notch, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.alpha, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.mean, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.whisker.kind, options = c("no", "std", "max"), length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = box.whisker.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(dot.color)){
+ tempo1 <- fun_check(data = dot.color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.color, class = "factor", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.check.color <- fun_check(data = dot.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, neg.values = FALSE, fun.name = function.name)$problem
+ if(tempo.check.color == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(any(dot.color == 0L, na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE A FACTOR OR CHARACTER VECTOR OR POSITVE INTEGER VECTOR") # integer possible because dealt above
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.color, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(dot.categ)){
+ tempo <- fun_check(data = dot.categ, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.categ, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(dot.categ.class.order)){
+ tempo <- fun_check(data = dot.categ.class.order, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.categ.class.order, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(dot.legend.name)){
+ tempo <- fun_check(data = dot.legend.name, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.legend.name, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = dot.tidy, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dot.tidy.bin.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if(dot.tidy.bin.nb == 0L){ # length and NA checked above
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.tidy.bin.nb ARGUMENT MUST BE A NON-NULL AND POSITVE INTEGER VALUE") # integer possible because dealt above
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ tempo <- fun_check(data = dot.jitter, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(dot.seed)){
+ tempo <- fun_check(data = dot.seed, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.seed, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = dot.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dot.alpha, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = dot.border.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(dot.border.color)){
+ tempo1 <- fun_check(data = dot.border.color, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.border.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR STRING STARTING BY #, OR (2) A COLOR NAME GIVEN BY colors(), OR (3) AN INTEGER VALUE")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }else if(tempo1$problem == FALSE & tempo2$problem == TRUE){
+ if( ! all(dot.border.color %in% colors() | grepl(pattern = "^#", dot.border.color), na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.border.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR STRING STARTING BY #, OR (2) A COLOR NAME GIVEN BY colors(), OR (3) AN INTEGER VALUE")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.border.color, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(x.lab)){
+ tempo1 <- fun_check(data = x.lab, class = "expression", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = x.lab, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lab ARGUMENT MUST BE A SINGLE CHARACTER STRING OR EXPRESSION")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x.lab, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = x.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(y.lab)){
+ tempo1 <- fun_check(data = y.lab, class = "expression", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = y.lab, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lab ARGUMENT MUST BE A SINGLE CHARACTER STRING OR EXPRESSION")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.lab, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(y.lim)){
+ tempo <- fun_check(data = y.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if(any(is.infinite(y.lim))){ # normally no NA for is.infinite() output
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.lim, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = y.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(y.tick.nb)){
+ tempo <- fun_check(data = y.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if(y.tick.nb < 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.tick.nb ARGUMENT MUST BE A NON NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(y.second.tick.nb)){
+ tempo <- fun_check(data = y.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if(y.second.tick.nb <= 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.second.tick.nb ARGUMENT MUST BE A NON NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.second.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = y.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.top.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.bottom.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(stat.pos)){
+ tempo <- fun_check(data = stat.pos, options = c("top", "above"), length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = stat.pos, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = stat.mean, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = stat.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = stat.dist, class = "vector", mode = "numeric", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = stat.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = vertical, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = legend.show, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(legend.width)){
+ tempo <- fun_check(data = legend.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = legend.width, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = article, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = grid, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(add)){
+ tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = add, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.ggplot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.gtable, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path), na.rm = TRUE)){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = lib.path, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if(any(arg.check) == TRUE){ # normally no NA
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between == #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length) == 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){ # normally no NA because is.na() used here
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "data1",
+ "y",
+ "categ",
+ "box.fill",
+ "box.width",
+ "box.space",
+ "box.line.size",
+ "box.notch",
+ "box.alpha",
+ "box.mean",
+ "box.whisker.kind",
+ "box.whisker.width",
+ # "dot.color", # inactivated because can be null
+ "dot.tidy",
+ "dot.tidy.bin.nb",
+ "dot.jitter",
+ # "dot.seed", # inactivated because can be null
+ "dot.size",
+ "dot.alpha",
+ "dot.border.size",
+ "x.angle",
+ "y.log",
+ # "y.second.tick.nb", # inactivated because can be null
+ "y.include.zero",
+ "y.top.extra.margin",
+ "y.bottom.extra.margin",
+ # "stat.pos", # inactivated because can be null
+ "stat.mean",
+ "stat.size",
+ "stat.dist",
+ "stat.angle",
+ "vertical",
+ "text.size",
+ "title",
+ "title.text.size",
+ "legend.show",
+ # "legend.width", # inactivated because can be null
+ "article",
+ "grid",
+ "return",
+ "return.ggplot",
+ "return.gtable",
+ "plot",
+ "warn.print"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){# normally no NA with is.null()
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # see also Protocol 100-rev0 Parallelization in R.docx
+ if(exists(".Random.seed", envir = .GlobalEnv)){ # if .Random.seed does not exists, it means that no random operation has been performed yet in any R environment
+ tempo.random.seed <- .Random.seed
+ on.exit(assign(".Random.seed", tempo.random.seed, env = .GlobalEnv))
+ }else{
+ on.exit(set.seed(NULL)) # inactivate seeding -> return to complete randomness
+ }
+ set.seed(dot.seed)
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ # end warning initiation
+ # other checkings
+ if(any(duplicated(names(data1)), na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDUPLICATED COLUMN NAMES OF data1 ARGUMENT NOT ALLOWED:\n", paste(names(data1)[duplicated(names(data1))], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! (y %in% names(data1))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny ARGUMENT MUST BE A COLUMN NAME OF data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ tempo <- fun_check(data = data1[, y], data.name = "y COLUMN OF data1", class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny ARGUMENT MUST BE NUMERIC COLUMN IN data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(length(categ) > 2){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg ARGUMENT CANNOT HAVE MORE THAN 2 COLUMN NAMES OF data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! all(categ %in% names(data1))){ # all() without na.rm -> ok because categ cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg ARGUMENT MUST BE COLUMN NAMES OF data1. HERE IT IS:\n", paste(categ, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(length(dot.categ) > 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT CANNOT HAVE MORE THAN 1 COLUMN NAMES OF data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! all(dot.categ %in% names(data1))){ # all() without na.rm -> ok because dot.categ cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT MUST BE COLUMN NAMES OF data1. HERE IT IS:\n", paste(dot.categ, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # reserved word checking
+ if(any(names(data1) %in% reserved.words, na.rm = TRUE)){
+ if(any(duplicated(names(data1)), na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDUPLICATED COLUMN NAMES OF data1 ARGUMENT NOT ALLOWED:\n", paste(names(data1)[duplicated(names(data1))], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(dot.categ)){
+ if(dot.categ %in% categ){
+ reserved.words <- c(reserved.words, paste0(dot.categ, "_DOT")) # paste0(dot.categ, "_DOT") is added to the reserved words because in such situation, a new column will be added to data1 that is named paste0(dot.categ, "_DOT")
+ }
+ }
+ tempo.output <- fun_name_change(names(data1), reserved.words)
+ for(i2 in 1:length(tempo.output$ini)){ # a loop to be sure to take the good ones
+ names(data1)[names(data1) == tempo.output$ini[i2]] <- tempo.output$post[i2]
+ if(any(y == tempo.output$ini[i2])){ # any() without na.rm -> ok because y cannot be NA (tested above)
+ y[y == tempo.output$ini[i2]] <- tempo.output$post[i2]
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN y ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN y ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # WARNING: names of y argument potentially replaced
+ if(any(categ == tempo.output$ini[i2])){ # any() without na.rm -> ok because categ cannot be NA (tested above)
+ categ[categ == tempo.output$ini[i2]] <- tempo.output$post[i2]
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN categ ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN categ ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # WARNING: names of categ argument potentially replaced
+ if( ! is.null(dot.categ)){
+ if(any(dot.categ == tempo.output$ini[i2])){ # any() without na.rm -> ok because dot.categ cannot be NA (tested above)
+ dot.categ[dot.categ == tempo.output$ini[i2]] <- tempo.output$post[i2]
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN dot.categ ARGUMENT (COLUMN NAMES OF data1 ARGUMENT),\n", tempo.output$ini[i2], " HAS BEEN REPLACED BY ", tempo.output$post[i2], "\nBECAUSE RISK OF BUG AS SOME NAMES IN dot.categ ARGUMENT ARE RESERVED WORD USED BY THE ", function.name, " FUNCTION")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # WARNING: names of dot.categ argument potentially replaced
+ }
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") REGARDING COLUMN NAMES REPLACEMENT, THE NAMES\n", paste(tempo.output$ini, collapse = " "), "\nHAVE BEEN REPLACED BY\n", paste(tempo.output$post, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ if( ! (is.null(add) | is.null(tempo.output$ini))){
+ if(grepl(x = add, pattern = paste(tempo.output$ini, collapse = "|"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF COLUMN NAMES OF data1 IN THE add ARGUMENT STRING, THAT CORRESPOND TO RESERVED STRINGS FOR ", function.name, "\nCOLUMN NAMES HAVE TO BE CHANGED\nTHE PROBLEMATIC COLUMN NAMES ARE SOME OF THESE NAMES:\n", paste(tempo.output$ini, collapse = " "), "\nIN THE DATA FRAME OF data1 AND IN THE STRING OF add ARGUMENT, TRY TO REPLACE NAMES BY:\n", paste(tempo.output$post, collapse = " "), "\n\nFOR INFORMATION, THE RESERVED WORDS ARE:\n", paste(reserved.words, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ }
+ if( ! (is.null(add))){
+ if(any(sapply(X = arg.names, FUN = grepl, x = add), na.rm = TRUE)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NAMES OF ", function.name, " ARGUMENTS DETECTED IN THE add STRING:\n", paste(arg.names[sapply(X = arg.names, FUN = grepl, x = add)], collapse = "\n"), "\nRISK OF WRONG OBJECT USAGE INSIDE ", function.name)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end reserved word checking
+ # verif of add
+ if( ! is.null(add)){
+ if( ! grepl(pattern = "^\\s*\\+", add)){ # check that the add string start by +
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nadd ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! grepl(pattern = "(ggplot2|lemon)\\s*::", add)){ #
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOR EASIER FUNCTION DETECTION, add ARGUMENT MUST CONTAIN \"ggplot2::\" OR \"lemon::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! grepl(pattern = ")\\s*$", add)){ # check that the add string finished by )
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nadd ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end verif of add
+ # management of add containing facet
+ facet.categ <- NULL
+ if( ! is.null(add)){
+ facet.check <- TRUE
+ tempo <- unlist(strsplit(x = add, split = "\\s*\\+\\s*(ggplot2|lemon)\\s*::\\s*")) #
+ tempo <- sub(x = tempo, pattern = "^facet_wrap", replacement = "ggplot2::facet_wrap")
+ tempo <- sub(x = tempo, pattern = "^facet_grid", replacement = "ggplot2::facet_grid")
+ tempo <- sub(x = tempo, pattern = "^facet_rep", replacement = "lemon::facet_rep")
+ if(any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap"), na.rm = TRUE)){
+ tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")])))
+ facet.categ <- names(tempo1$params$facets)
+ tempo.text <- "facet_wrap OR facet_rep_wrap"
+ facet.check <- FALSE
+ }else if(grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")){
+ tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")])))
+ facet.categ <- c(names(tempo1$params$rows), names(tempo1$params$cols))
+ tempo.text <- "facet_grid OR facet_rep_grid"
+ facet.check <- FALSE
+ }
+ if(facet.check == FALSE & ! all(facet.categ %in% names(data1))){ # WARNING: all(facet.categ %in% names(data1)) is TRUE when facet.categ is NULL # all() without na.rm -> ok because facet.categ cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF \"", tempo.text, "\" STRING IN THE add ARGUMENT BUT PROBLEM OF VARIABLE DETECTION (COLUMN NAMES OF data1)\nTHE DETECTED VARIABLES ARE:\n", paste(facet.categ, collapse = " "), "\nTHE data1 COLUMN NAMES ARE:\n", paste(names(data1), collapse = " "), "\nPLEASE REWRITE THE add STRING AND RERUN")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end management of add containing facet
+ # conversion of categ columns in data1 into factors
+ for(i1 in 1:length(categ)){
+ tempo1 <- fun_check(data = data1[, categ[i1]], data.name = paste0("categ NUMBER ", i1, " OF data1"), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = data1[, categ[i1]], data.name = paste0("categ NUMBER ", i1, " OF data1"), class = "factor", na.contain = TRUE, fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\n", paste0("categ NUMBER ", i1, " OF data1"), " MUST BE A FACTOR OR CHARACTER VECTOR")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(tempo1$problem == FALSE){ # character vector
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN categ NUMBER ", i1, " IN data1, THE CHARACTER COLUMN HAS BEEN CONVERTED TO FACTOR, WITH LEVELS ACCORDING TO THE ALPHABETICAL ORDER")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ data1[, categ[i1]] <- factor(data1[, categ[i1]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ }
+ # OK: all the categ columns of data1 are factors from here
+ # end conversion of categ columns in data1 into factors
+
+
+
+ # management of log scale and Inf removal
+ if(any(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf # normally no NA with is.finite0() and is.na()
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN THE ", y, " COLUMN OF THE data1 ARGUMENT AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ data1.ini <- data1 # strictly identical to data1 except that in data1 y is log converted if and only if y.log != "no"
+ if(y.log != "no"){
+ tempo1 <- ! is.finite(data1[, y]) # where are initial NA and Inf
+ data1[, y] <- suppressWarnings(get(y.log)(data1[, y]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ if(any( ! (tempo1 | is.finite(data1[, y])))){ # normally no NA with is.finite
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN THE ", y, " COLUMN OF THE data1 ARGUMENT AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # Inf removal
+ if(any(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf # normally no NA with is.finite
+ removed.row.nb <- which(( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])))
+ removed.rows <- data1.ini[removed.row.nb, ] # here data1.ini used to have the y = O rows that will be removed because of Inf creation after log transformation
+ data1 <- data1[-removed.row.nb, ] #
+ data1.ini <- data1.ini[-removed.row.nb, ] #
+ }else{
+ removed.row.nb <- NULL
+ removed.rows <- data.frame(stringsAsFactors = FALSE)
+ }
+ # From here, data1 and data.ini have no more Inf
+ # end Inf removal
+ if(y.log != "no" & ! is.null(y.lim)){
+ if(any(y.lim <= 0)){ # any() without na.rm -> ok because y.lim cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(any( ! is.finite(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}))){ # normally no NA with is.finite
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT RETURNS INF/NA WITH THE y.log ARGUMENT SET TO ", y.log, "\nAS SCALE COMPUTATION IS ", ifelse(y.log == "log10", "log10", "log2"), ":\n", paste(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(y.log != "no" & y.include.zero == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") y.log ARGUMENT SET TO ", y.log, " AND y.include.zero ARGUMENT SET TO TRUE -> y.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ y.include.zero <- FALSE
+ }
+ if(y.log != "no" & vertical == FALSE){
+ vertical <- TRUE
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") BECAUSE OF A BUG IN ggplot2, CANNOT FLIP BOXES HORIZONTALLY WITH A Y.LOG SCALE -> vertical ARGUMENT RESET TO TRUE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end management of log scale and Inf removal
+ # na detection and removal (done now to be sure of the correct length of categ)
+ column.check <- unique(c(y, categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){dot.categ}, if( ! is.null(facet.categ)){facet.categ})) # dot.categ because can be a 3rd column of data1, categ.color and dot.color will be tested later
+ if(any(is.na(data1[, column.check]))){ # data1 used here instead of data1.ini in case of new NaN created by log conversion (neg values) # normally no NA with is.na
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS OF data1 AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ for(i2 in 1:length(column.check)){
+ if(any(is.na(data1[, column.check[i2]]))){ # normally no NA with is.na
+ tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i2], " OF data1")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n", tempo.warn)))
+ }
+ }
+ tempo <- unique(unlist(lapply(lapply(c(data1[column.check]), FUN = is.na), FUN = which)))
+ removed.row.nb <- c(removed.row.nb, tempo) # removed.row.nb created to remove Inf
+ removed.rows <- rbind(removed.rows, data1.ini[tempo, ], stringsAsFactors = FALSE) # here data1.ini used to have the non NA rows that will be removed because of NAN creation after log transformation (neg values for instance)
+ column.check <- column.check[ ! column.check == y] # remove y to keep quali columns
+ if(length(tempo) != 0){
+ data1 <- data1[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
+ data1.ini <- data1.ini[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers than the former
+ for(i3 in 1:length(column.check)){
+ if(any( ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]]), na.rm = TRUE)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i3], " OF data1, THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA/Inf REMOVAL (IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[, column.check[i3]])[ ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]])], collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ }
+ count.categ <- 0
+ for(i2 in 1:length(column.check)){
+ if(column.check[i2] %in% categ){
+ count.categ <- count.categ + 1
+ }
+ if(column.check[i2] == categ[count.categ]){
+ categ.class.order[count.categ] <- list(levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])]) # remove the absent color in the character vector
+ data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(categ.class.order[[count.categ]]))
+ }
+ if( ! is.null(dot.color) & ! is.null(dot.categ)){ # reminder : dot.categ cannot be a column name of categ anymore (because in that case dot.categ name is changed into "..._DOT"
+ if(column.check[i2] == dot.categ){
+ dot.categ.class.order <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
+ data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(dot.categ.class.order))
+ }
+ }
+ if(column.check[i2] %in% facet.categ){ # works if facet.categ == NULL this method should keep the order of levels when removing some levels
+ tempo.levels <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(as.character(data1[, column.check[i2]]))]
+ data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = tempo.levels)
+ }
+ }
+ }
+ # end na detection and removal (done now to be sure of the correct length of categ)
+ # From here, data1 and data.ini have no more NA or NaN in y, categ, dot.categ (if dot.color != NULL) and facet.categ
+
+
+
+ if( ! is.null(categ.class.order)){
+ if(length(categ.class.order) != length(categ)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.class.order ARGUMENT MUST BE A LIST OF LENGTH EQUAL TO LENGTH OF categ\nHERE IT IS LENGTH: ", length(categ.class.order), " VERSUS ", length(categ))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ for(i3 in 1:length(categ.class.order)){
+ if(is.null(categ.class.order[[i3]])){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE categ.class.order COMPARTMENT ", i3, " IS NULL. ALPHABETICAL ORDER WILL BE APPLIED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ data1[, categ[i3]] <- factor(as.character(data1[, categ[i3]])) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ categ.class.order[[i3]] <- levels(data1[, categ[i3]]) # character vector that will be used later
+ }else{
+ tempo <- fun_check(data = categ.class.order[[i3]], data.name = paste0("COMPARTMENT ", i3 , " OF categ.class.order ARGUMENT"), class = "vector", mode = "character", length = length(levels(data1[, categ[i3]])), fun.name = function.name) # length(data1[, categ[i1]) -> if data1[, categ[i1] was initially character vector, then conversion as factor after the NA removal, thus class number ok. If data1[, categ[i1] was initially factor, no modification after the NA removal, thus class number ok
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(any(duplicated(categ.class.order[[i3]]), na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i3, " OF categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(categ.class.order[[i3]], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! (all(categ.class.order[[i3]] %in% unique(data1[, categ[i3]]), na.rm = TRUE) & all(unique(data1[, categ[i3]]) %in% categ.class.order[[i3]], na.rm = TRUE))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i3, " OF categ.class.order ARGUMENT MUST BE CLASSES OF ELEMENT ", i3, " OF categ ARGUMENT\nHERE IT IS:\n", paste(categ.class.order[[i3]], collapse = " "), "\nFOR COMPARTMENT ", i3, " OF categ.class.order AND IT IS:\n", paste(unique(data1[, categ[i3]]), collapse = " "), "\nFOR COLUMN ", categ[i3], " OF data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ data1[, categ[i3]] <- factor(data1[, categ[i3]], levels = categ.class.order[[i3]]) # reorder the factor
+
+ }
+ names(categ.class.order)[i3] <- categ[i3]
+ }
+ }
+ }else{
+ categ.class.order <- vector("list", length = length(categ))
+ tempo.categ.class.order <- NULL
+ for(i2 in 1:length(categ.class.order)){
+ categ.class.order[[i2]] <- levels(data1[, categ[i2]])
+ names(categ.class.order)[i2] <- categ[i2]
+ tempo.categ.class.order <- c(tempo.categ.class.order, ifelse(i2 != 1, "\n", ""), categ.class.order[[i2]])
+ }
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR BOX ORDERING:\n", paste(tempo.categ.class.order, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # categ.class.order not NULL anymore (list)
+ if(is.null(box.legend.name) & box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE box.legend.name SETTING IS NULL. NAMES OF categ WILL BE USED: ", paste(categ, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ box.legend.name <- categ[length(categ)] # if only categ1, then legend name of categ1, if length(categ) == 2L, then legend name of categ2
+ }
+ # box.legend.name not NULL anymore (character string)
+ # management of categ.color
+ if( ! is.null(categ.color)){
+ # check the nature of color
+ # integer colors into gg_palette
+ tempo.check.color <- fun_check(data = categ.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem
+ if(tempo.check.color == FALSE){
+ # convert integers into colors
+ categ.color <- fun_gg_palette(max(categ.color, na.rm = TRUE))[categ.color]
+ }
+ # end integer colors into gg_palette
+ if( ! (all(categ.color %in% colors() | grepl(pattern = "^#", categ.color)))){ # check that all strings of low.color start by #, # all() without na.rm -> ok because categ.color cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors() OR A COLUMN NAME OF THE data1 PARAMETER: ", paste(unique(categ.color), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(is.na(categ.color)) & box.alpha != 0){ # normally no NA with is.na
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT CONTAINS NA")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end check the nature of color
+ # check the length of color
+ categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
+ if(length(data1[, categ[categ.len]]) == length(levels(data1[, categ[categ.len]])) & length(categ.color) == length(data1[, categ[categ.len]])){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE NUMBER OF CLASSES OF THE COLUMN ", categ[categ.len], " THE NUMBER OF ROWS OF THIS COLUMN AND THE NUMBER OF COLORS OF THE categ.color ARGUMENT ARE ALL EQUAL. BOX COLORS WILL BE ATTRIBUTED ACCORDING THE LEVELS OF ", categ[categ.len], ", NOT ACCORDING TO THE ROWS OF ", categ[categ.len])
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if(length(categ.color) == length(levels(data1[, categ[categ.len]]))){ # here length(categ.color) is equal to the different number of categ
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ data1 <- data.frame(data1, categ.color = data1[, categ[categ.len]], stringsAsFactors = TRUE) # no need stringsAsFactors here for stat.nolog as factors remain factors
+ data1$categ.color <- factor(data1$categ.color, labels = categ.color) # replace the characters of data1[, categ[categ.len]] put in the categ.color column by the categ.color (can be write like this because categ.color is length of levels of data1[, categ[categ.len]])
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", categ[categ.len], " OF categ ARGUMENT, THE FOLLOWING COLORS:\n", paste(categ.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else if(length(categ.color) == length(data1[, categ[categ.len]])){# here length(categ.color) is equal to nrow(data1) -> Modif to have length(categ.color) equal to the different number of categ (length(categ.color) == length(levels(data1[, categ[categ.len]])))
+ data1 <- data.frame(data1, categ.color = categ.color, stringsAsFactors = TRUE)
+ tempo.check <- unique(data1[ , c(categ[categ.len], "categ.color")])
+ if( ! (nrow(tempo.check) == length(unique(categ.color)) & nrow(tempo.check) == length(unique(data1[ , categ[categ.len]])))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT HAS THE LENGTH OF data1 ROW NUMBER\nBUT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF categ ", categ[categ.len], ":\n", paste(unique(mapply(FUN = "paste", data1[ ,categ[categ.len]], data1[ ,"categ.color"])), collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ categ.color <- unique(data1$categ.color[order(data1[, categ[categ.len]])]) # Modif to have length(categ.color) equal to the different number of categ (length(categ.color) == length(levels(data1[, categ[categ.len]])))
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT HAS THE LENGTH OF data1 ROW NUMBER\nCOLORS HAVE BEEN RESPECTIVELY ASSOCIATED TO EACH CLASS OF categ ", categ[categ.len], " AS:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\n", paste(categ.color, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ }else if(length(categ.color) == 1L){
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ data1 <- data.frame(data1, categ.color = categ.color, stringsAsFactors = TRUE)
+ categ.color <- rep(categ.color, length(levels(data1[, categ[categ.len]])))
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") categ.color ARGUMENT HAS LENGTH 1, MEANING THAT ALL THE DIFFERENT CLASSES OF ", categ[categ.len], "\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(categ.color, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ncateg.color ARGUMENT MUST BE (1) LENGTH 1, OR (2) THE LENGTH OF data1 NROWS AFTER NA/Inf REMOVAL, OR (3) THE LENGTH OF THE CLASSES IN THE categ ", categ[categ.len], " COLUMN. HERE IT IS COLOR LENGTH ", length(categ.color), " VERSUS CATEG LENGTH ", length(data1[, categ[categ.len]]), " AND CATEG CLASS LENGTH ", length(unique(data1[, categ[categ.len]])), "\nPRESENCE OF NA/Inf COULD BE THE PROBLEM")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ categ.color <- fun_gg_palette(length(levels(data1[, categ[categ.len]])))
+ data1 <- data.frame(data1, categ.color = data1[, categ[categ.len]], stringsAsFactors = TRUE)
+ data1$categ.color <- factor(data1$categ.color, labels = categ.color) # replace the characters of data1[, categ[categ.len]] put in the categ.color column by the categ.color (can be write like this because categ.color is length of levels of data1[, categ[categ.len]])
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL categ.color ARGUMENT -> COLORS RESPECTIVELY ATTRIBUTED TO EACH CLASS OF ", categ[categ.len], " IN data1:\n", paste(categ.color, collapse = " "), "\n", paste(levels(data1[, categ[categ.len]]), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # categ.color not NULL anymore
+ categ.color <- as.character(categ.color)
+ # categ.color is a character string representing the diff classes
+ data1$categ.color <- factor(data1$categ.color, levels = unique(categ.color)) # ok because if categ.color is a character string, the order make class 1, class 2, etc. unique() because no duplicates allowed
+ # data1$categ.color is a factor with order of levels -> categ.color
+ # end management of categ.color
+ # management of dot.color
+ if( ! is.null(dot.color)){
+ # optional legend of dot colors
+ if( ! is.null(dot.categ)){
+ ini.dot.categ <- dot.categ
+ if( ! dot.categ %in% names(data1)){ # no need to use all() because length(dot.categ) = 1
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT MUST BE A COLUMN NAME OF data1. HERE IT IS:\n", dot.categ)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(dot.categ %in% categ){ # no need to use all() because length(dot.categ) = 1. Do not use dot.categ %in% categ[length(categ)] -> error
+ # management of dot legend if dot.categ %in% categ (because legends with the same name are joined in ggplot2)
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE COLUMN NAME OF data1 INDICATED IN THE dot.categ ARGUMENT (", dot.categ, ") HAS BEEN REPLACED BY ", paste0(dot.categ, "_DOT"), " TO AVOID MERGED LEGEND BY GGPLOT2")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ data1 <- data.frame(data1, dot.categ = data1[, dot.categ], stringsAsFactors = TRUE) # dot.categ is not a column name of data1 (checked above with reserved words)
+ dot.categ <- paste0(dot.categ, "_DOT")
+ names(data1)[names(data1) == "dot.categ"] <- dot.categ # paste0(dot.categ, "_DOT") is not a column name of data1 (checked above with reserved words)
+ # tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ ARGUMENT CANNOT BE A COLUMN NAME OF data1 ALREADY SPECIFIED IN THE categ ARGUMENT:\n", dot.categ, "\nINDEED, dot.categ ARGUMENT IS MADE TO HAVE MULTIPLE DOT COLORS NOT RELATED TO THE BOXPLOT CATEGORIES")
+ # stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ tempo1 <- fun_check(data = data1[, dot.categ], data.name = paste0(dot.categ, " COLUMN OF data1"), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = data1[, dot.categ], data.name = paste0(dot.categ, " COLUMN OF data1"), class = "factor", na.contain = TRUE, fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ COLUMN MUST BE A FACTOR OR CHARACTER VECTOR") #
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ data1[, dot.categ] <- factor(data1[, dot.categ]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ # dot.categ column of data1 is factor from here
+ if( ! is.null(dot.categ.class.order)){
+ if(any(duplicated(dot.categ.class.order), na.rm = TRUE)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(dot.categ.class.order, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! (all(dot.categ.class.order %in% levels(data1[, dot.categ])) & all(levels(data1[, dot.categ]) %in% dot.categ.class.order, na.rm = TRUE))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.categ.class.order ARGUMENT MUST BE CLASSES OF dot.categ ARGUMENT\nHERE IT IS:\n", paste(dot.categ.class.order, collapse = " "), "\nFOR dot.categ.class.order AND IT IS:\n", paste(levels(data1[, dot.categ]), collapse = " "), "\nFOR dot.categ COLUMN (", ini.dot.categ, ") OF data1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
+ }
+ }else{
+ if(all(dot.color == "same") & length(dot.color)== 1L){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
+ dot.categ.class.order <- unlist(categ.class.order[length(categ)])
+ data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order SETTING IS NULL AND dot.color IS \"same\". ORDER OF categ.class.order WILL BE APPLIED FOR LEGEND DISPLAY: ", paste(dot.categ.class.order, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ dot.categ.class.order <- sort(levels(data1[, dot.categ]))
+ data1[, dot.categ] <- factor(data1[, dot.categ], levels = dot.categ.class.order) # reorder the factor
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR LEGEND DISPLAY: ", paste(dot.categ.class.order, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # dot.categ.class.order not NULL anymore (character string) if dot.categ is not NULL
+ if(all(dot.color == "same") & length(dot.color)== 1L){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
+ if( ! identical(ini.dot.categ, categ[length(categ)])){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN dot.color ARGUMENT IS \"same\", THE COLUMN NAME IN dot.categ ARGUMENT MUST BE IDENTICAL TO THE LAST COLUMN NAME IN categ ARGUMENT. HERE IT IS:\ndot.categ: ", paste(ini.dot.categ, collapse = " "), "\ncateg: ", paste(categ, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if( ! fun_comp_1d(unlist(categ.class.order[length(categ)]), dot.categ.class.order)$identical.content){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN dot.color ARGUMENT IS \"same\",\nLAST COMPARTMENT OF categ.class.order ARGUMENT AND dot.categ.class.order ARGUMENT CANNOT BE DIFFERENT:\nLAST COMPARTMENT OF categ.class.order: ", paste(unlist(categ.class.order[length(categ)]), collapse = " "), "\ndot.categ.class.order: ", paste(dot.categ.class.order, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ for(i3 in 1:length(categ)){
+ if(identical(categ[i3], ini.dot.categ) & ! identical(unlist(categ.class.order[i3]), dot.categ.class.order) & identical(sort(unlist(categ.class.order[i3])), sort(dot.categ.class.order))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE dot.categ ARGUMENT SETTING IS PRESENT IN THE categ ARGUMENT SETTING, BUT ORDER OF THE CLASSES IS NOT THE SAME:\ncateg.class.order: ", paste(unlist(categ.class.order[i3]), collapse = " "), "\ndot.categ.class.order: ", paste(dot.categ.class.order, collapse = " "), "\nNOTE THAT ORDER OF categ.class.order IS THE ONE USED FOR THE AXIS REPRESENTATION")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(is.null(dot.legend.name)){
+ dot.legend.name <- if(ini.dot.categ %in% categ[length(categ)]){dot.categ}else{ini.dot.categ} #
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE dot.legend.name SETTING IS NULL -> ", dot.legend.name, " WILL BE USED AS LEGEND TITLE OF DOTS")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # dot.legend.name not NULL anymore (character string)
+ }else{
+ if( ! is.null(dot.categ.class.order)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE dot.categ.class.order ARGUMENT IS NOT NULL, BUT IS THE dot.categ ARGUMENT\n-> dot.categ.class.order NOT CONSIDERED AS NO LEGEND WILL BE DRAWN")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # But dot.categ.class.order will be converted to NULL below (not now)
+ }
+ # end optional legend of dot colors
+ # check the nature of color
+ # integer colors into gg_palette
+ tempo.check.color <- fun_check(data = dot.color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem
+ if(tempo.check.color == FALSE){
+ # convert integers into colors
+ dot.color <- fun_gg_palette(max(dot.color, na.rm = TRUE))[dot.color]
+ }
+ # end integer colors into gg_palette
+ if(all(dot.color == "same") & length(dot.color)== 1L){# all() without na.rm -> ok because dot.color cannot be NA (tested above)
+ dot.color <- categ.color # same color of the dots as the corresponding box color
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT HAS BEEN SET TO \"same\"\nTHUS, DOTS WILL HAVE THE SAME COLORS AS THE CORRESPONDING BOXPLOT")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else if( ! (all(dot.color %in% colors() | grepl(pattern = "^#", dot.color)))){ # check that all strings of low.color start by #, # all() without na.rm -> ok because dot.color cannot be NA (tested above)
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE (1) A HEXADECIMAL COLOR VECTOR STARTING BY #, OR (2) COLOR NAMES GIVEN BY colors(), OR (3) INTEGERS, OR THE STRING \"same\"\nHERE IT IS: ", paste(unique(dot.color), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(any(is.na(dot.color))){ # normally no NA with is.finite
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT CONTAINS NA")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end check the nature of color
+ # check the length of color
+ if( ! is.null(dot.categ)){
+ # optional legend of dot colors
+ if(length(data1[, dot.categ]) == length(levels(data1[, dot.categ])) & length(dot.color) == length(data1[, dot.categ])){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE NUMBER OF CLASSES OF THE COLUMN ", dot.categ, " THE NUMBER OF ROWS OF THIS COLUMN AND THE NUMBER OF COLORS OF THE dot.color ARGUMENT ARE ALL EQUAL. DOT COLORS WILL BE ATTRIBUTED ACCORDING THE LEVELS OF ", dot.categ, ", NOT ACCORDING TO THE ROWS OF ", dot.categ)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if(length(dot.color) > 1 & ! (length(dot.color) == length(unique(data1[, dot.categ])) | length(dot.color) == length(data1[, dot.categ]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nWHEN LENGTH OF THE dot.color ARGUMENT IS MORE THAN 1, IT MUST BE EQUAL TO THE NUMBER OF 1) ROWS OR 2) LEVELS OF dot.categ COLUMN (", dot.categ, "):\ndot.color: ", paste(dot.color, collapse = " "), "\ndot.categ LEVELS: ", paste(levels(data1[, dot.categ]), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else if(length(dot.color) > 1 & length(dot.color) == length(unique(data1[, dot.categ]))){
+ data1 <- data.frame(data1, dot.color = data1[, dot.categ], stringsAsFactors = TRUE)
+ data1$dot.color <- factor(data1$dot.color, labels = dot.color) # do not use labels = unique(dot.color). Otherwise, we can have green1 green2 when dot.color is c("green", "green")
+ }else if(length(dot.color) > 1 & length(dot.color) == length(data1[, dot.categ])){
+ data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
+ }else if(length(dot.color)== 1L){ # to deal with single color. Warning: & length(dot.categ.class.order) > 1 removed because otherwise, the data1 is not with dot.color column when length(dot.categ.class.order) == 1
+ data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
+ }
+ dot.color <- as.character(unique(data1$dot.color[order(data1[, dot.categ])])) # reorder the dot.color character vector
+ if(length(dot.color)== 1L & length(dot.categ.class.order) > 1){ # to deal with single color
+ dot.color <- rep(dot.color, length(dot.categ.class.order))
+ }
+ tempo.check <- unique(data1[ , c(dot.categ, "dot.color")])
+ if(length(unique(data1[ , "dot.color"])) > 1 & ( ! (nrow(tempo.check) == length(unique(data1[ , "dot.color"])) & nrow(tempo.check) == length(unique(data1[ , dot.categ]))))){ # length(unique(data1[ , "dot.color"])) > 1 because if only one color, can be attributed to each class of dot.categ
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF dot.categ (", dot.categ, ") COLUMN:\n", paste(unique(mapply(FUN = "paste", data1[ , dot.categ], data1[ ,"dot.color"])), collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN dot.categ ARGUMENT (", ini.dot.categ, "), THE FOLLOWING COLORS OF DOTS:\n", paste(dot.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(data1[, dot.categ]), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # dot.color is a character string representing the diff classes of dot.categ
+ # data1$dot.color is a factor with order of levels -> dot.categ
+ # end optional legend of dot colors
+ }else{
+ categ.len <- length(categ) # if only categ1, then colors for classes of categ1, if length(categ) == 2L, then colors for classes of categ2
+ if(length(dot.color) == length(levels(data1[, categ[categ.len]]))){ # here length(dot.color) is equal to the different number of categ
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ data1 <- data.frame(data1, dot.color = data1[, categ[categ.len]], stringsAsFactors = TRUE)
+ data1$dot.color <- factor(data1$dot.color, labels = dot.color)
+ if(box.alpha != 0){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", categ[categ.len], " OF categ ARGUMENT, THE FOLLOWING COLORS:\n", paste(dot.color, collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else if(length(dot.color) == length(data1[, categ[categ.len]])){# here length(dot.color) is equal to nrow(data1) -> Modif to have length(dot.color) equal to the different number of categ (length(dot.color) == length(levels(data1[, categ[categ.len]])))
+ data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
+ }else if(length(dot.color)== 1L & ! all(dot.color == "same")){ # all() without na.rm -> ok because dot.color cannot be NA (tested above)
+ # data1[, categ[categ.len]] <- factor(data1[, categ[categ.len]]) # not required because sure that is is a factor
+ data1 <- data.frame(data1, dot.color = dot.color, stringsAsFactors = TRUE)
+ dot.color <- rep(dot.color, length(levels(data1[, categ[categ.len]])))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") dot.color ARGUMENT HAS LENGTH 1, MEANING THAT ALL THE DIFFERENT CLASSES OF ", categ[categ.len], "\n", paste(levels(factor(data1[, categ[categ.len]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(dot.color, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ndot.color ARGUMENT MUST BE (1) LENGTH 1, OR (2) THE LENGTH OF data1 NROWS AFTER NA/Inf REMOVAL, OR (3) THE LENGTH OF THE CLASSES IN THE categ ", categ[categ.len], " COLUMN. HERE IT IS COLOR LENGTH ", length(dot.color), " VERSUS CATEG LENGTH ", length(data1[, categ[categ.len]]), " AND CATEG CLASS LENGTH ", length(unique(data1[, categ[categ.len]])), "\nPRESENCE OF NA/Inf COULD BE THE PROBLEM")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end check the length of color
+ dot.color <- as.character(dot.color)
+ # dot.color is a character string representing the diff classes
+ data1$dot.color <- factor(data1$dot.color, levels = unique(dot.color)) # ok because if dot.color is a character string, the order make class 1, class 2, etc. If dot.color is a column of data1, then levels will be created, without incidence, except if dot.categ specified (see below). unique() because no duplicates allowed
+ # data1$dot.color is a factor with order of levels -> dot.color
+ }
+ # end optional legend of dot colors
+ }else if(is.null(dot.color) & ! (is.null(dot.categ) & is.null(dot.categ.class.order) & is.null(dot.legend.name))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") dot.categ OR dot.categ.class.order OR dot.legend.name ARGUMENT HAS BEEN SPECIFIED BUT dot.color ARGUMENT IS NULL (NO DOT PLOTTED)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # dot.color either NULL (no dot plotted) or character string (potentially representing the diff classes of dot.categ)
+ # data1$dot.color is either NA or a factor (with order of levels -> depending on dot.categ or categ[length(categ)], or other
+ if(is.null(dot.categ)){
+ dot.categ.class.order <- NULL # because not used anyway
+ }
+ # dot.categ.class.order either NULL if dot.categ is NULL (no legend displayed) or character string (potentially representing the diff classes of dot.categ)
+ # end management of dot.color
+ if(is.null(dot.color) & box.fill == FALSE & dot.alpha <= 0.025){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE FOLLOWING ARGUMENTS WERE SET AS:\ndot.color = NULL (NOT ALL DOTS BUT ONLY POTENTIAL OUTLIER DOTS DISPLAYED)\nbox.fill = FALSE (NO FILLING COLOR FOR BOTH BOXES AND POTENTIAL OUTLIER DOTS)\ndot.alpha = ", fun_round(dot.alpha, 4), "\n-> POTENTIAL OUTLIER DOTS MIGHT NOT BE VISIBLE BECAUSE ALMOST TRANSPARENT")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if(is.null(dot.color) & box.fill == FALSE & dot.border.size == 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE FOLLOWING ARGUMENTS WERE SET AS:\ndot.color = NULL (NOT ALL DOTS BUT ONLY POTENTIAL OUTLIER DOTS DISPLAYED)\nbox.fill = FALSE (NO FILLING COLOR FOR BOTH BOXES AND POTENTIAL OUTLIER DOTS)\ndot.border.size = 0 (NO BORDER FOR POTENTIAL OUTLIER DOTS)\n-> THESE SETTINGS ARE NOT ALLOWED BECAUSE THE POTENTIAL OUTLIER DOTS WILL NOT BE VISIBLE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # integer dot.border.color into gg_palette
+ if( ! is.null(dot.border.color)){
+ tempo <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
+ if(tempo$problem == FALSE){ # convert integers into colors
+ dot.border.color <- fun_gg_palette(max(dot.border.color, na.rm = TRUE))[dot.border.color]
+ }
+ }
+ # end integer dot.border.color into gg_palette
+ # na detection and removal (done now to be sure of the correct length of categ)
+ column.check <- c("categ.color", if( ! is.null(dot.color)){"dot.color"}) #
+ if(any(is.na(data1[, column.check]))){ # data1 used here instead of data1.ini in case of new NaN created by log conversion (neg values) # normally no NA with is.na
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS ", paste(column.check, collapse = " "), " OF data1 AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ for(i2 in 1:length(column.check)){
+ if(any(is.na(data1[, column.check[i2]]))){ # normally no NA with is.na
+ tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i2], " OF data1")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n", tempo.warn)))
+ }
+ }
+ tempo <- unique(unlist(lapply(lapply(c(data1[column.check]), FUN = is.na), FUN = which)))
+ removed.row.nb <- c(removed.row.nb, tempo)
+ removed.rows <- rbind(removed.rows, data1[tempo, ], stringsAsFactors = FALSE) # here data1 used because categorical columns tested
+ if(length(tempo) != 0){
+ data1 <- data1[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
+ data1.ini <- data1.ini[-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
+ for(i3 in 1:length(column.check)){
+ if(any( ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]]), na.rm = TRUE)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i3], " OF data1, THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA/Inf REMOVAL (IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[, column.check[i3]])[ ! unique(removed.rows[, column.check[i3]]) %in% unique(data1[, column.check[i3]])], collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ }
+ for(i2 in 1:length(column.check)){
+ if(column.check[i2] == "categ.color"){
+ categ.color <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
+ if(length(categ.color)== 1L & length(unlist(categ.class.order[length(categ)])) > 1){ # to deal with single color
+ categ.color <- rep(categ.color, length(unlist(categ.class.order[length(categ)])))
+ }
+ data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(categ.color))
+ }
+ if(column.check[i2] == "dot.color"){
+ dot.color <- levels(data1[, column.check[i2]])[levels(data1[, column.check[i2]]) %in% unique(data1[, column.check[i2]])] # remove the absent color in the character vector
+ if(length(dot.color)== 1L & length(dot.categ.class.order) > 1){ # to deal with single color. If dot.categ.class.order == NULL (which is systematically the case if dot.categ == NULL), no rep(dot.color, length(dot.categ.class.order)
+ dot.color <- rep(dot.color, length(dot.categ.class.order))
+ }
+ data1[, column.check[i2]] <- factor(as.character(data1[, column.check[i2]]), levels = unique(dot.color))
+ }
+ }
+ }
+ # end na detection and removal (done now to be sure of the correct length of categ)
+ # From here, data1 and data.ini have no more NA or NaN
+ # end other checkings
+ # reserved word checking
+ #already done above
+ # end reserved word checking
+ # end second round of checking and data preparation
+
+
+ # package checking
+ fun_pack(req.package = c(
+ "ggplot2",
+ "gridExtra",
+ "lemon",
+ "scales"
+ ), lib.path = lib.path)
+ # end package checking
+
+
+
+
+
+ # main code
+ # y coordinates recovery (create ini.box.coord, dot.coord and modify data1)
+ if(length(categ)== 1L){
+ # width commputations
+ box.width2 <- box.width
+ box.space <- 0 # to inactivate the shrink that add space between grouped boxes, because no grouped boxes here
+ # end width commputations
+ # data1 check categ order for dots coordinates recovery
+ data1 <- data.frame(data1, categ.check = data1[, categ[1]], stringsAsFactors = TRUE)
+ data1$categ.check <- as.integer(data1$categ.check) # to check that data1[, categ[1]] and dot.coord$group are similar, during merging
+ # end data1 check categ order for dots coordinates recovery
+ # per box dots coordinates recovery
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[1]), stroke = dot.border.size, size = dot.size, alpha = dot.alpha, shape = 21))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(is.null(categ.color)){rep(NA, length(unique(data1[, categ[1]])))}else if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[1]])))}else{categ.color})) # categ.color used for dot colors because at that stage, we do not care about colors
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[1]), coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf})) # fill because this is what is used with geom_box # to easily have the equivalent of the grouped boxes
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[1]])))}else{categ.color}))
+ # end per box dots coordinates recovery
+ }else if(length(categ) == 2L){
+ # width commputations
+ box.width2 <- box.width / length(unique(data1[, categ[length(categ)]])) # real width of each box in x-axis unit, among the set of grouped box. Not relevant if no grouped boxes length(categ)== 1L
+ # end width commputations
+ # data1 check categ order for dots coordinates recovery
+ tempo.factor <- paste0(data1[order(data1[, categ[2]], data1[, categ[1]]), categ[2]], "_", data1[order(data1[, categ[2]], data1[, categ[1]]), categ[1]])
+ data1 <- data.frame(data1[order(data1[, categ[2]], data1[, categ[1]]), ], categ.check = factor(tempo.factor, levels = unique(tempo.factor)), stringsAsFactors = TRUE)
+ data1$categ.check <- as.integer(data1$categ.check)
+ # end data1 check categ order for dots coordinates recovery
+ # per box dots coordinates recovery
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[2]), stroke = dot.border.size, size = dot.size, alpha = dot.alpha, shape = 21))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(is.null(categ.color)){rep(NA, length(unique(data1[, categ[2]])))}else if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[2]])))}else{categ.color})) # categ.color used for dot colors because at that stage, we do not care about colors
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[2]), coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf})) # fill because this is what is used with geom_box # to easily have the equivalent of the grouped boxes
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[2]])))}else{categ.color}))
+ # end per box dots coordinates recovery
+ }else{
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if( ! is.null(stat.pos)){
+ stat.just <- fun_gg_just(
+ angle = stat.angle,
+ pos = ifelse(
+ vertical == TRUE,
+ ifelse(stat.pos == "top", "bottom", "top"), # "bottom" because we want justification for text that are below the ref point which is the top of the graph. The opposite for "above"
+ ifelse(stat.pos == "top", "left", "right") # "left" because we want justification for text that are on the left of the ref point which is the right border of the graph. The opposite for "above"
+ ),
+ kind = "text"
+ )
+ }
+ # has in fact no interest because ggplot2 does not create room for geom_text()
+ tempo.data.max <- data1[which.max(data1[, y]), ]
+ tempo.data.max <- data.frame(tempo.data.max, label = formatC(tempo.data.max[, y], digit = 2, drop0trailing = TRUE, format = "f"), stringsAsFactors = TRUE)
+ # end has in fact no interest because ggplot2 does not create room for geom_text()
+ tempo.graph.info.ini <- ggplot2::ggplot_build(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if( ! is.null(stat.pos)){' + ggplot2::geom_text(data = tempo.data.max, mapping = ggplot2::aes_string(x = 1, y = y, label = "label"), size = stat.size, color = "black", angle = stat.angle, hjust = stat.just$hjust, vjust = stat.just$vjust)'})))) # added here to have room for annotation
+ dot.coord <- tempo.graph.info.ini$data[[1]]
+ dot.coord$x <- as.numeric(dot.coord$x) # because weird class
+ dot.coord$PANEL <- as.numeric(dot.coord$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
+ tempo.mean <- aggregate(x = dot.coord$y, by = list(dot.coord$group, dot.coord$PANEL), FUN = mean, na.rm = TRUE)
+ names(tempo.mean)[names(tempo.mean) == "x"] <- "MEAN"
+ names(tempo.mean)[names(tempo.mean) == "Group.1"] <- "BOX"
+ names(tempo.mean)[names(tempo.mean) == "Group.2"] <- "PANEL"
+ dot.coord <- data.frame(
+ dot.coord[order(dot.coord$group, dot.coord$y), ], # dot.coord$PANEL deals below
+ y.check = as.double(data1[order(data1$categ.check, data1[, y]), y]),
+ categ.check = data1[order(data1$categ.check, data1[, y]), "categ.check"],
+ dot.color = if(is.null(dot.color)){NA}else{data1[order(data1$categ.check, data1[, y]), "dot.color"]},
+ data1[order(data1$categ.check, data1[, y]), ][categ], # avoid the renaming below
+ stringsAsFactors = TRUE
+ ) # y.check to be sure that the order is the same between the y of data1 and the y of dot.coord
+ # names(dot.coord)[names(dot.coord) == "tempo.categ1"] <- categ[1]
+ if( ! is.null(dot.categ)){
+ dot.coord <- data.frame(dot.coord, data1[order(data1$categ.check, data1[, y]), ][dot.categ], stringsAsFactors = TRUE) # avoid the renaming
+ }
+ if( ! is.null(facet.categ)){
+ dot.coord <- data.frame(dot.coord, data1[order(data1$categ.check, data1[, y]), ][facet.categ], stringsAsFactors = TRUE) # for facet panels
+ tempo.test <- NULL
+ for(i2 in 1:length(facet.categ)){
+ tempo.test <- paste0(tempo.test, ".", formatC(as.numeric(dot.coord[, facet.categ[i2]]), width = nchar(max(as.numeric(dot.coord[, facet.categ[i2]]), na.rm = TRUE)), flag = "0")) # convert factor into numeric with leading zero for proper ranking # merge the formatC() to create a new factor. The convertion to integer should recreate the correct group number. Here as.numeric is used and not as.integer in case of numeric in facet.categ (because comes from add and not checked by fun_check, contrary to categ)
+ }
+ tempo.test <- as.integer(factor(tempo.test))
+ if( ! identical(as.integer(dot.coord$PANEL), tempo.test)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nas.integer(dot.coord$PANEL) AND tempo.test MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(dot.tidy == TRUE){
+ if( ! is.null(dot.categ)){
+ dot.coord <- data.frame(dot.coord, tidy_group = data1[order(data1$categ.check, data1[, y]), ][, dot.categ], stringsAsFactors = TRUE) # avoid the renaming
+ # tidy_group_coord is to be able to fuse table when creating the table for dot coordinates
+ if(dot.categ %in% categ){
+ dot.coord <- data.frame(dot.coord, tidy_group_coord = dot.coord$group, stringsAsFactors = TRUE)
+ }else{
+ dot.coord <- data.frame(dot.coord, tidy_group_coord = as.integer(factor(paste0(
+ formatC(as.integer(dot.coord[, categ[1]]), width = nchar(max(as.integer(dot.coord[, categ[1]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
+ ".",
+ if(length(categ) == 2L){formatC(as.integer(dot.coord[, categ[2]]), width = nchar(max(as.integer(dot.coord[, categ[2]]), na.rm = TRUE)), flag = "0")}, # convert factor into numeric with leading zero for proper ranking
+ if(length(categ) == 2L){"."},
+ formatC(as.integer(dot.coord[, dot.categ]), width = nchar(max(as.integer(dot.coord[, dot.categ]), na.rm = TRUE)), flag = "0") # convert factor into numeric with leading zero for proper ranking
+ )), stringsAsFactors = TRUE) # merge the 2 or 3 formatC() to create a new factor. The convertion to integer should recreate the correct group number
+ ) # for tidy dot plots
+ }
+ }else{
+ dot.coord <- data.frame(dot.coord, tidy_group = if(length(categ)== 1L){
+ dot.coord[, categ]}else{as.integer(factor(paste0(
+ formatC(as.integer(dot.coord[, categ[1]]), width = nchar(max(as.integer(dot.coord[, categ[1]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
+ ".",
+ formatC(as.integer(dot.coord[, categ[2]]), width = nchar(max(as.integer(dot.coord[, categ[2]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
+ )), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
+ }) # for tidy dot plots
+ # tidy_group_coord is to be able to fuse table when creating the table for dot coordinates
+ dot.coord <- data.frame(dot.coord, tidy_group_coord = dot.coord$group, stringsAsFactors = TRUE)
+ }
+ }
+ if( ! (identical(dot.coord$y, dot.coord$y.check) & identical(dot.coord$group, dot.coord$categ.check))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\n(dot.coord$y AND dot.coord$y.check) AS WELL AS (dot.coord$group AND dot.coord$categ.check) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ if( ! identical(tempo.mean[order(tempo.mean$BOX, tempo.mean$PANEL), ]$BOX, unique(dot.coord[order(dot.coord$group, dot.coord$PANEL), c("group", "PANEL")])$group)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\n(tempo.mean$BOX, tempo.mean$PANEL) AND (dot.coord$group, dot.coord$PANEL) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ tempo <- unique(dot.coord[order(dot.coord$group, dot.coord$PANEL), c(categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){if(dot.categ != ini.dot.categ){dot.categ}}, if( ! is.null(facet.categ)){facet.categ}), drop = FALSE])
+ # names(tempo) <- paste0(names(tempo), ".mean")
+ tempo.mean <- data.frame(tempo.mean[order(tempo.mean$BOX, tempo.mean$PANEL), ], tempo, stringsAsFactors = TRUE)
+ }
+ }
+ # at that stage, categ color and dot color are correctly attributed in data1, box.coord and dot.coord
+ # end y dot coordinates recovery (create ini.box.coord, dot.coord and modify data1)
+ # ylim range
+ if(is.null(y.lim)){
+ y.lim <- tempo.graph.info.ini$layout$panel_params[[1]]$y.range # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ if(any(( ! is.finite(y.lim)) | is.na(y.lim)) | length(y.lim) != 2){ # kept but normally no more Inf in data1 # normally no NA with is.finite, etc.
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\ntempo.graph.info.ini$layout$panel_params[[1]]$y.range[1] CONTAINS NA OR Inf OR HAS LENGTH 1")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else if(y.log != "no"){
+ y.lim <- get(y.log)(y.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ }
+ if(y.log != "no"){
+ # normally this control is not necessary anymore
+ if(any( ! is.finite(y.lim))){ # normally no NA with is.finite
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ if(suppressWarnings(all(y.lim %in% c(Inf, -Inf)))){ # all() without na.rm -> ok because y.lim cannot be NA (tested above)
+ # normally this control is not necessary anymore
+ tempo.cat <- paste0("ERROR IN ", function.name, " y.lim CONTAINS Inf VALUES, MAYBE BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY OR BECAUSE OF LOG SCALE REQUIREMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ if(suppressWarnings(any(is.na(y.lim)))){ # normally no NA with is.na
+ # normally this control is not necessary anymore
+ tempo.cat <- paste0("ERROR IN ", function.name, " y.lim CONTAINS NA OR NaN VALUES, MAYBE BECAUSE VALUES FROM data1 ARGUMENTS ARE NA OR Inf ONLY OR BECAUSE OF LOG SCALE REQUIREMENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ y.lim.order <- order(y.lim) # to deal with inverse axis
+ y.lim <- sort(y.lim)
+ y.lim[1] <- y.lim[1] - abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.bottom.extra.margin, y.top.extra.margin) # diff(y.lim.order) > 0 medians not inversed axis
+ y.lim[2] <- y.lim[2] + abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.top.extra.margin, y.bottom.extra.margin) # diff(y.lim.order) > 0 medians not inversed axis
+ if(y.include.zero == TRUE){ # no need to check y.log != "no" because done before
+ y.lim <- range(c(y.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ }
+ y.lim <- y.lim[y.lim.order]
+ if(any(is.na(y.lim))){ # normally no NA with is.na
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 2")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end ylim range
+
+
+
+
+
+
+ # drawing
+ # constant part
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add is directly put here to deal with additional variable of data, like when using facet_grid. No problem if add is a theme, will be dealt below
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::xlab(if(is.null(x.lab)){categ[1]}else{x.lab}))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ylab(if(is.null(y.lab)){y}else{y.lab}))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
+ # text angle management
+ axis.just <- fun_gg_just(angle = x.angle, pos = ifelse(vertical == TRUE, "bottom", "left"), kind = "axis")
+ # end text angle management
+ add.check <- TRUE
+ if( ! is.null(add)){ # if add is NULL, then = 0
+ if(grepl(pattern = "ggplot2\\s*::\\s*theme", add) == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT\n-> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER\n-> article ARGUMENT WILL BE IGNORED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ add.check <- FALSE
+ }
+ }
+ if(add.check == TRUE & article == TRUE){
+ # WARNING: not possible to add theme()several times. NO message but the last one overwrites the others
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
+ if(grid == TRUE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_line(size = 0.5),
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ axis.line.y.left = ggplot2::element_line(colour = "black"), # draw lines for the y axis
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"), # draw lines for the x axis
+ panel.grid.major.x = if(vertical == TRUE){NULL}else{ggplot2::element_line(colour = "grey85", size = 0.75)},
+ panel.grid.major.y = if(vertical == TRUE){ggplot2::element_line(colour = "grey85", size = 0.75)}else{NULL},
+ panel.grid.minor.y = if(vertical == TRUE){ggplot2::element_line(colour = "grey90", size = 0.25)}else{NULL},
+ axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
+ axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)},
+ strip.background = ggplot2::element_rect(fill = NA, colour = NA) # for facet background
+ ))
+ }else{
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_line(size = 0.5),
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ axis.line.y.left = ggplot2::element_line(colour = "black"),
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"),
+ axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
+ axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)},
+ strip.background = ggplot2::element_rect(fill = NA, colour = NA)
+ ))
+ }
+ }else if(add.check == TRUE & article == FALSE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_line(size = 0.5),
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ panel.background = ggplot2::element_rect(fill = "grey95"),
+ axis.line.y.left = ggplot2::element_line(colour = "black"),
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"),
+ panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.minor.x = ggplot2::element_blank(),
+ panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
+ strip.background = ggplot2::element_rect(fill = NA, colour = NA),
+ axis.text.x = if(vertical == TRUE){ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}else{NULL},
+ axis.text.y = if(vertical == TRUE){NULL}else{ggplot2::element_text(angle = axis.just$angle, hjust = axis.just$hjust, vjust = axis.just$vjust)}
+ ))
+ }
+ # Contrary to fun_gg_bar(), cannot plot the boxplot right now, because I need the dots plotted first
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, group = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), color = NA, width = box.width, fill = NA)) # this is to set the graph (i.e., a blanck boxplot to be able to use x coordinates to plot dots before boxes)
+ # end constant part
+
+
+
+
+ # graphic info recovery (including means)
+ tempo.graph.info <- ggplot2::ggplot_build(eval(parse(text = paste0(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, notch = box.notch, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}) + ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color})')))) # will be recovered later again, when ylim will be considered
+ tempo.yx.ratio <- (tempo.graph.info$layout$panel_params[[1]]$y.range[2] - tempo.graph.info$layout$panel_params[[1]]$y.range[1]) / (tempo.graph.info$layout$panel_params[[1]]$x.range[2] - tempo.graph.info$layout$panel_params[[1]]$x.range[1])
+ box.coord <- tempo.graph.info$data[[2]] # to have the summary statistics of the plot. Contrary to ini.box.plot, now integrates ylim Here because can be required for stat.pos when just box are plotted
+ box.coord$x <- as.numeric(box.coord$x) # because x is of special class that block comparison of values using identical
+ box.coord$PANEL <- as.numeric(box.coord$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
+ box.coord <- box.coord[order(box.coord$group, box.coord$PANEL), ]
+ if( ! (identical(tempo.mean$BOX, box.coord$group) & identical(tempo.mean$PANEL, box.coord$PANEL))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nidentical(tempo.mean$BOX, box.coord$group) & identical(tempo.mean$PANEL, box.coord$PANEL) DO NOT HAVE THE SAME VALUE ORDER")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ # tempo <- c(categ, if( ! is.null(dot.color) & ! is.null(dot.categ)){if(dot.categ != ini.dot.categ){dot.categ}}, if( ! is.null(facet.categ)){facet.categ})
+ if(any(names(tempo.mean) %in% names(box.coord), na.rm = TRUE)){
+ names(tempo.mean)[names(tempo.mean) %in% names(box.coord)] <- paste0(names(tempo.mean)[names(tempo.mean) %in% names(box.coord)], ".mean")
+ }
+ box.coord <- data.frame(box.coord, tempo.mean, stringsAsFactors = TRUE)
+ }
+ # end graphic info recovery (including means)
+
+
+
+ # stat output (will also serve for boxplot and mean display)
+ # x not added now (to do not have them in stat.nolog)
+ stat <- data.frame(
+ MIN = box.coord$ymin_final,
+ QUART1 = box.coord$lower,
+ MEDIAN = box.coord$middle,
+ MEAN = box.coord$MEAN,
+ QUART3 = box.coord$upper,
+ MAX = box.coord$ymax_final,
+ WHISK_INF = box.coord$ymin,
+ BOX_INF = box.coord$lower,
+ NOTCH_INF = box.coord$notchlower,
+ NOTCH_SUP = box.coord$notchupper,
+ BOX_SUP = box.coord$upper,
+ WHISK_SUP = box.coord$ymax,
+ OUTLIERS = box.coord["outliers"],
+ tempo.mean[colnames(tempo.mean) != "MEAN"],
+ COLOR = box.coord$fill,
+ stringsAsFactors = TRUE
+ ) # box.coord["outliers"] written like this because it is a list. X coordinates not put now because several features to set
+ names(stat)[names(stat) == "outliers"] <- "OUTLIERS"
+ stat.nolog <- stat # stat.nolog ini will serve for outputs
+ if(y.log != "no"){
+ stat.nolog[c("MIN", "QUART1", "MEDIAN", "MEAN", "QUART3", "MAX", "WHISK_INF", "BOX_INF", "NOTCH_INF", "NOTCH_SUP", "BOX_SUP", "WHISK_SUP")] <- ifelse(y.log == "log2", 2, 10)^(stat.nolog[c("MIN", "QUART1", "MEDIAN", "MEAN", "QUART3", "MAX", "WHISK_INF", "BOX_INF", "NOTCH_INF", "NOTCH_SUP", "BOX_SUP", "WHISK_SUP")])
+ stat.nolog$OUTLIERS <- lapply(stat.nolog$OUTLIERS, FUN = function(X){ifelse(y.log == "log2", 2, 10)^X})
+ }
+ # end stat output (will also serve for boxplot and mean display)
+
+
+
+
+
+
+ # x coordinates management (for random plotting and for stat display)
+ # width commputations
+ width.ini <- c(box.coord$xmax - box.coord$xmin)[1] # all the box widths are equal here. Only the first one taken
+ width.correct <- width.ini * box.space / 2
+ if( ! (identical(stat$BOX, box.coord$group) & identical(stat$PANEL, box.coord$PANEL))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nidentical(stat$BOX, box.coord$group) & identical(stat$PANEL, box.coord$PANEL) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ stat <- data.frame(
+ stat,
+ X = box.coord$x,
+ X_BOX_INF = box.coord$xmin + width.correct,
+ X_BOX_SUP = box.coord$xmax - width.correct,
+ X_NOTCH_INF = box.coord$x - (box.coord$x - (box.coord$xmin + width.correct)) / 2,
+ X_NOTCH_SUP = box.coord$x + (box.coord$x - (box.coord$xmin + width.correct)) / 2,
+ X_WHISK_INF = box.coord$x - (box.coord$x - (box.coord$xmin + width.correct)) * box.whisker.width,
+ X_WHISK_SUP = box.coord$x + (box.coord$x - (box.coord$xmin + width.correct)) * box.whisker.width,
+ # tempo.mean[colnames(tempo.mean) != "MEAN"], # already added above
+ stringsAsFactors = TRUE
+ )
+ stat$COLOR <- factor(stat$COLOR, levels = unique(categ.color))
+ if( ! all(stat$NOTCH_SUP < stat$BOX_SUP & stat$NOTCH_INF > stat$BOX_INF, na.rm = TRUE) & box.notch == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") SOME NOTCHES ARE BEYOND BOX HINGES. TRY ARGUMENT box.notch = FALSE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ dot.jitter <- c((box.coord$xmax - width.correct) - (box.coord$xmin + width.correct))[1] * dot.jitter # real dot.jitter. (box.coord$xmin + width.correct) - (box.coord$xmax - width.correct))[1] is the width of the box. Is equivalent to (box.coord$x - (box.coord$xmin + width.correct))[1] * 2
+ # end width commputations
+ if( ! is.null(dot.color)){
+ # random dots
+ if(dot.tidy == FALSE){
+ dot.coord.rd1 <- merge(dot.coord, box.coord[c("fill", "PANEL", "group", "x")], by = c("PANEL", "group"), sort = FALSE) # rd for random. Send the coord of the boxes into the coord data.frame of the dots (in the column x.y). WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in box.coord. Thus, no need to consider fill column
+ if(nrow(dot.coord.rd1) != nrow(dot.coord)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.rd1 DATA FRAME. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ sampled.dot.jitter <- if(nrow(dot.coord.rd1)== 1L){runif(n = nrow(dot.coord.rd1), min = - dot.jitter / 2, max = dot.jitter / 2)}else{sample(x = runif(n = nrow(dot.coord.rd1), min = - dot.jitter / 2, max = dot.jitter / 2), size = nrow(dot.coord.rd1), replace = FALSE)}
+ dot.coord.rd2 <- data.frame(dot.coord.rd1, dot.x = dot.coord.rd1$x.y + sampled.dot.jitter, stringsAsFactors = TRUE) # set the dot.jitter thanks to runif and dot.jitter range. Then, send the coord of the boxes into the coord data.frame of the dots (in the column x.y)
+ if(length(categ)== 1L){
+ tempo.data1 <- unique(data.frame(data1[categ[1]], group = as.integer(data1[, categ[1]]), stringsAsFactors = TRUE)) # categ[1] is factor
+ names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
+ verif <- paste0(categ[1], ".check")
+ }else if(length(categ) == 2L){
+ tempo.data1 <- unique(
+ data.frame(
+ data1[c(categ[1], categ[2])],
+ group = as.integer(factor(paste0(
+ formatC(as.integer(data1[, categ[2]]), width = nchar(max(as.integer(data1[, categ[2]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
+ ".",
+ formatC(as.integer(data1[, categ[1]]), width = nchar(max(as.integer(data1[, categ[1]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
+ )), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
+ )
+ ) # categ[2] first if categ[2] is used to make the categories in ggplot and categ[1] is used to make the x-axis
+ names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
+ names(tempo.data1)[names(tempo.data1) == categ[2]] <- paste0(categ[2], ".check")
+ verif <- c(paste0(categ[1], ".check"), paste0(categ[2], ".check"))
+ }else{
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ dot.coord.rd3 <- merge(dot.coord.rd2, tempo.data1, by = intersect("group", "group"), sort = FALSE) # send the factors of data1 into coord. WARNING: I have replaced by = "group" by intersect("group", "group") because of an error due to wrong group group merging in dot.coord.rd3
+ if(nrow(dot.coord.rd3) != nrow(dot.coord) | ( ! fun_comp_2d(dot.coord.rd3[categ], dot.coord.rd3[verif])$identical.content)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.rd3 DATA FRAME. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end random dots
+ }
+ # tidy dots
+ # coordinates are recovered during plotting (see dot.coord.tidy1 below)
+ # end tidy dots
+ }
+ # end x coordinates management (for random plotting and for stat display)
+
+
+
+
+
+ # boxplot display before dot display if box.fill = TRUE
+ coord.names <- NULL
+ # creation of the data frame for (main box + legend) and data frame for means
+ if(box.notch == FALSE){
+ for(i3 in 1:length(categ)){
+ if(i3== 1L){
+ tempo.polygon <- data.frame(GROUPX = c(t(stat[, rep(categ[i3], 5)])), stringsAsFactors = TRUE)
+ }else{
+ tempo.polygon <- cbind(tempo.polygon, c(t(stat[, rep(categ[i3], 5)])), stringsAsFactors = TRUE)
+ }
+ }
+ names(tempo.polygon) <- categ
+ tempo.polygon <- data.frame(X = c(t(stat[, c("X_BOX_INF", "X_BOX_SUP", "X_BOX_SUP", "X_BOX_INF", "X_BOX_INF")])), Y = c(t(stat[, c("BOX_INF", "BOX_INF", "BOX_SUP", "BOX_SUP", "BOX_INF")])), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), BOX = as.character(c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX")]))), tempo.polygon, stringsAsFactors = TRUE)
+ if( ! is.null(facet.categ)){
+ for(i4 in 1:length(facet.categ)){
+ tempo.polygon <- data.frame(tempo.polygon, c(t(stat[, c(facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4])])), stringsAsFactors = TRUE)
+ names(tempo.polygon)[length(names(tempo.polygon))] <- facet.categ[i4]
+ }
+ }
+ }else{
+ for(i3 in 1:length(categ)){
+ if(i3== 1L){
+ tempo.polygon <- data.frame(GROUPX = c(t(stat[, rep(categ[i3], 11)])), stringsAsFactors = TRUE)
+ }else{
+ tempo.polygon <- cbind(tempo.polygon, c(t(stat[, rep(categ[i3], 11)])), stringsAsFactors = TRUE)
+ }
+ }
+ names(tempo.polygon) <- categ
+ tempo.polygon <- data.frame(X = c(t(stat[, c("X_BOX_INF", "X_BOX_SUP", "X_BOX_SUP", "X_NOTCH_SUP", "X_BOX_SUP", "X_BOX_SUP", "X_BOX_INF", "X_BOX_INF", "X_NOTCH_INF", "X_BOX_INF", "X_BOX_INF")])), Y = c(t(stat[, c("BOX_INF", "BOX_INF", "NOTCH_INF", "MEDIAN", "NOTCH_SUP", "BOX_SUP", "BOX_SUP", "NOTCH_SUP", "MEDIAN", "NOTCH_INF", "BOX_INF")])), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), BOX = as.character(c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX", "BOX")]))), tempo.polygon, stringsAsFactors = TRUE)
+ if( ! is.null(facet.categ)){
+ for(i4 in 1:length(facet.categ)){
+ tempo.polygon <- data.frame(tempo.polygon, c(t(stat[, c(facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4], facet.categ[i4])])), stringsAsFactors = TRUE)
+ names(tempo.polygon)[length(names(tempo.polygon))] <- facet.categ[i4]
+ }
+ }
+ }
+ tempo.polygon$COLOR <- factor(tempo.polygon$COLOR, levels = unique(categ.color))
+ if( ! is.null(categ.class.order)){
+ for(i3 in 1:length(categ)){
+ tempo.polygon[, categ[i3]] <- factor(tempo.polygon[, categ[i3]], levels = categ.class.order[[i3]])
+ }
+ }
+ # modified name of dot.categ column (e.g., "Categ1_DOT") must be included for boxplot using ridy dots
+ if( ! is.null(dot.color) & ! is.null(dot.categ)){
+ if(dot.categ != ini.dot.categ){
+ tempo.polygon <- data.frame(tempo.polygon, GROUPX = tempo.polygon[, ini.dot.categ], stringsAsFactors = TRUE)
+ names(tempo.polygon)[names(tempo.polygon) == "GROUPX"] <- dot.categ
+
+ }
+ }
+ tempo.diamon.mean <- data.frame(X = c(t(stat[, c("X", "X_NOTCH_INF", "X", "X_NOTCH_SUP", "X")])), Y = c(t(cbind(stat["MEAN"] - (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stat["MEAN"], stat["MEAN"] + (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stat["MEAN"], stat["MEAN"] - (stat[, "X"] - stat[, "X_NOTCH_INF"]) * tempo.yx.ratio, stringsAsFactors = TRUE))), COLOR = c(t(stat[, c("COLOR", "COLOR", "COLOR", "COLOR", "COLOR")])), GROUP = c(t(stat[, c("BOX", "BOX", "BOX", "BOX", "BOX")])), stringsAsFactors = TRUE) # stringsAsFactors = TRUE for cbind() because stat["MEAN"] is a data frame. Otherwise, stringsAsFactors is not an argument for cbind() on vectors
+ if( ! is.null(facet.categ)){
+ for(i3 in 1:length(facet.categ)){
+ tempo.diamon.mean <- data.frame(tempo.diamon.mean, c(t(stat[, c(facet.categ[i3], facet.categ[i3], facet.categ[i3], facet.categ[i3], facet.categ[i3])])), stringsAsFactors = TRUE)
+ names(tempo.diamon.mean)[length(names(tempo.diamon.mean))] <- facet.categ[i3]
+ }
+ }
+ tempo.diamon.mean$COLOR <- factor(tempo.diamon.mean$COLOR, levels = unique(categ.color))
+ # end creation of the data frame for (main box + legend) and data frame for means
+ if(box.fill == TRUE){
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[length(categ)], fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, size = box.line.size, notch = box.notch, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}, alpha = box.alpha, outlier.shape = if( ! is.null(dot.color)){NA}else{21}, outlier.color = if( ! is.null(dot.color)){NA}else{dot.border.color}, outlier.fill = if( ! is.null(dot.color)){NA}else{NULL}, outlier.size = if( ! is.null(dot.color)){NA}else{dot.size}, outlier.stroke = if( ! is.null(dot.color)){NA}else{dot.border.size}, outlier.alpha = if( ! is.null(dot.color)){NA}else{dot.alpha})) # the color, size, etc. of the outliers are dealt here. outlier.color = NA to do not plot outliers when dots are already plotted. Finally, boxplot redrawn (see below)
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_polygon(
+ data = tempo.polygon,
+ mapping = ggplot2::aes_string(x = "X", y = "Y", group = "BOX", fill = categ[length(categ)], color = categ[length(categ)]),
+ size = box.line.size,
+ alpha = box.alpha # works only for fill, not for color
+ ))
+ coord.names <- c(coord.names, "main.box")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "sup.whisker")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_INF, yend = WHISK_INF, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "inf.whisker")
+ if(box.whisker.width > 0){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha, lineend = "round")) #
+ coord.names <- c(coord.names, "sup.whisker.edge")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_INF, yend = WHISK_INF, group = categ[length(categ)]), color = "black", size = box.line.size, alpha = box.alpha, lineend = "round")) #
+ coord.names <- c(coord.names, "inf.whisker.edge")
+ }
+ if(box.mean == TRUE){
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = stat, mapping = ggplot2::aes_string(x = "X", y = "MEAN", group = categ[length(categ)]), shape = 23, stroke = box.line.size * 2, fill = stat$COLOR, size = box.mean.size, color = "black", alpha = box.alpha)) # group used in aesthetic to do not have it in the legend
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_polygon(
+ data = tempo.diamon.mean,
+ mapping = ggplot2::aes(x = X, y = Y, group = GROUP),
+ fill = tempo.diamon.mean[, "COLOR"],
+ color = hsv(0, 0, 0, alpha = box.alpha), # outline of the polygon in black but with alpha
+ size = box.line.size,
+ alpha = box.alpha
+ ))
+ coord.names <- c(coord.names, "mean")
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = if(box.notch == FALSE){X_BOX_INF}else{X_NOTCH_INF}, xend = if(box.notch == FALSE){X_BOX_SUP}else{X_NOTCH_SUP}, y = MEDIAN, yend = MEDIAN, group = categ[length(categ)]), color = "black", size = box.line.size * 2, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "median")
+ }
+ # end boxplot display before dot display if box.fill = TRUE
+
+
+
+
+
+
+ # dot display
+ if( ! is.null(dot.color)){
+ if(dot.tidy == FALSE){
+ if(is.null(dot.categ)){
+ if(dot.border.size == 0){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
+ data = dot.coord.rd3,
+ mapping = ggplot2::aes_string(x = "dot.x", y = "y", group = categ[length(categ)]),
+ size = dot.size,
+ shape = 19,
+ color = dot.coord.rd3$dot.color,
+ alpha = dot.alpha
+ )) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
+ }else{
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
+ data = dot.coord.rd3,
+ mapping = ggplot2::aes_string(x = "dot.x", y = "y", group = categ[length(categ)]),
+ shape = 21,
+ stroke = dot.border.size,
+ color = if(is.null(dot.border.color)){dot.coord.rd3$dot.color}else{rep(dot.border.color, nrow(dot.coord.rd3))},
+ size = dot.size,
+ fill = dot.coord.rd3$dot.color,
+ alpha = dot.alpha
+ )) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
+ }
+ }else{
+ if(dot.border.size == 0){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
+ data = dot.coord.rd3,
+ mapping = ggplot2::aes_string(x = "dot.x", y = "y", alpha = dot.categ),
+ size = dot.size,
+ shape = 19,
+ color = dot.coord.rd3$dot.color
+ )) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
+ }else{
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(
+ data = dot.coord.rd3,
+ mapping = ggplot2::aes_string(x = "dot.x", y = "y", alpha = dot.categ),
+ size = dot.size,
+ shape = 21,
+ stroke = dot.border.size,
+ color = if(is.null(dot.border.color)){dot.coord.rd3$dot.color}else{rep(dot.border.color, nrow(dot.coord.rd3))},
+ fill = dot.coord.rd3$dot.color
+ )) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = dot.legend.name, values = rep(dot.alpha, length(dot.categ.class.order)), guide = ggplot2::guide_legend(override.aes = list(fill = dot.color, color = if(is.null(dot.border.color)){dot.color}else{dot.border.color}, stroke = dot.border.size, alpha = dot.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ }
+ coord.names <- c(coord.names, "dots")
+ }else if(dot.tidy == TRUE){
+ # here plot using group -> no scale
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_dotplot(
+ data = dot.coord,
+ mapping = ggplot2::aes_string(x = categ[1], y = "y", group = "group"), # not dot.categ here because the classes of dot.categ create new separations
+ position = ggplot2::position_dodge(width = box.width),
+ binpositions = "all",
+ binaxis = "y",
+ stackdir = "center",
+ alpha = dot.alpha,
+ fill = dot.coord$dot.color,
+ stroke = dot.border.size,
+ color = if(is.null(dot.border.color)){dot.coord$dot.color}else{rep(dot.border.color, nrow(dot.coord))},
+ show.legend = FALSE, # WARNING: do not use show.legend = TRUE because it uses the arguments outside aes() as aesthetics (here color and fill). Thus I must find a way using ggplot2::scale_discrete_manual()
+ binwidth = (y.lim[2] - y.lim[1]) / dot.tidy.bin.nb
+ )) # geom_dotplot ggplot2 v3.3.0: I had to remove rev() in fill and color # very weird behavior of geom_dotplot ggplot2 v3.2.1, (1) because with aes group = (to avoid legend), the dot plotting is not good in term of coordinates, and (2) because data1 seems reorderer according to x = categ[1] before plotting. Thus, I have to use fill = dot.coord[rev(order(dot.coord[, categ[1]], decreasing = TRUE)), "dot.color"] to have the good corresponding colors # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
+ coord.names <- c(coord.names, "dots")
+ if( ! is.null(dot.categ)){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_dotplot(
+ data = dot.coord,
+ mapping = ggplot2::aes_string(x = categ[1], y = "y", alpha = dot.categ), # not dot.categ here because the classes of dot.categ create new separations
+ position = ggplot2::position_dodge(width = box.width),
+ binpositions = "all",
+ binaxis = "y",
+ stackdir = "center",
+ fill = NA,
+ stroke = NA,
+ color = NA,
+ # WARNING: do not use show.legend = TRUE because it uses the arguments outside aes() as aesthetics (here color and fill). Thus I must find a way using ggplot2::scale_discrete_manual()
+ binwidth = (y.lim[2] - y.lim[1]) / dot.tidy.bin.nb
+ )) # geom_dotplot ggplot2 v3.3.0: I had to remove rev() in fill and color # very weird behavior of geom_dotplot ggplot2 v3.2.1, (1) because with aes group = (to avoid legend), the dot plotting is not good in term of coordinates, and (2) because data1 seems reorderer according to x = categ[1] before plotting. Thus, I have to use fill = dot.coord[rev(order(dot.coord[, categ[1]], decreasing = TRUE)), "dot.color"] to have the good corresponding colors # show.legend option do not remove the legend, only the aesthetic of the legend (dot, line, etc.)
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "linetype", name = dot.legend.name, values = rep(1, length(categ.color)))) # values = rep("black", length(categ.color)) are the values of color (which is the border color of dots), and this modify the border color on the plot. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ coord.names <- c(coord.names, "bad_remove")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = dot.legend.name, values = rep(dot.alpha, length(dot.categ.class.order)), labels = dot.categ.class.order, guide = ggplot2::guide_legend(title = if(ini.dot.categ == categ[length(categ)]){dot.categ}else{ini.dot.categ}, override.aes = list(fill = levels(dot.coord$dot.color), color = if(is.null(dot.border.color)){levels(dot.coord$dot.color)}else{dot.border.color}, stroke = dot.border.size, alpha = dot.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ }
+ # coordinates of tidy dots
+ tempo.coord <- ggplot2::ggplot_build(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "))))$data # to have the tidy dot coordinates
+ if(length(which(sapply(X = tempo.coord, FUN = function(X){any(names(X) == "binwidth", na.rm = TRUE)}))) != 1){ # detect the compartment of tempo.coord which is the binned data frame
+ # if(length(which(sapply(tempo.coord, FUN = nrow) == nrow(data1))) > if(is.null(dot.categ)){1}else{2}){ # this does not work if only one dot per class, thus replaced by above # if(is.null(dot.categ)){1}else{2} because 1 dotplot if dot.categ is NULL and 2 dotplots if not, with the second being a blank dotplot with wrong coordinates. Thus take the first in that situation
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nEITHER MORE THAN 1 OR NO COMPARTMENT HAVING A DATA FRAME WITH binwidth AS COLUMN NAME IN THE tempo.coord LIST (FOR TIDY DOT COORDINATES). CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ # dot.coord.tidy1 <- tempo.coord[[which(sapply(tempo.coord, FUN = nrow) == nrow(data1))[1]]] # this does not work if only one dot per class, thus replaced by above # the second being a blank dotplot with wrong coordinates. Thus take the first whatever situation
+ dot.coord.tidy1 <- tempo.coord[[which(sapply(X = tempo.coord, FUN = function(X){any(names(X) == "binwidth", na.rm = TRUE)}))]] # detect the compartment of tempo.coord which is the binned data frame
+ dot.coord.tidy1$x <- as.numeric(dot.coord.tidy1$x) # because weird class
+ dot.coord.tidy1$PANEL <- as.numeric(dot.coord.tidy1$PANEL) # because numbers as levels. But may be a problem is facet are reordered ?
+ }
+ # tempo.box.coord <- merge(box.coord, unique(dot.coord[, c("PANEL", "group", categ)]), by = c("PANEL", "group"), sort = FALSE) # not required anymore because box.coord already contains categ do not add dot.categ and tidy_group_coord here because the coordinates are for stats. Add the categ in box.coord. WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in box.coord. Thus, no need to consider fill column
+ # below inactivated because not true when dealing with dot.categ different from categ
+ # if(nrow(tempo.box.coord) != nrow(box.coord)){
+ # tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT tempo.box.coord DATA FRAME. CODE HAS TO BE MODIFIED")
+ # stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ # }
+ dot.coord.tidy2 <- merge(dot.coord.tidy1, box.coord[c("fill", "PANEL", "group", "x", categ)], by = c("PANEL", "group"), sort = FALSE) # send the coord of the boxes into the coord data.frame of the dots (in the column x.y).WARNING: by = c("PANEL", "group") without fill column because PANEL & group columns are enough as only one value of x column per group number in tempo.box.coord. Thus, no need to consider fill colum # DANGER: from here the fill.y and x.y (from tempo.box.coord) are not good in dot.coord.tidy2. It is ok because Categ1 Categ2 from tempo.box.coord are ok with the group column from dot.coord.tidy1. This is due to the fact that dot.coord.tidy resulting from geom_dotplot does not make the same groups as the other functions
+ if(nrow(dot.coord.tidy2) != nrow(dot.coord)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.tidy2 DATA FRAME. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # From here, check for dot.coord.tidy3 which wil be important for stat over the plot. WARNING: dot.categ has nothing to do here for stat coordinates. Thus, not in tempo.data1
+ if(length(categ)== 1L){
+ tempo.data1 <- unique(data.frame(data1[categ[1]], group = as.integer(data1[, categ[1]]), stringsAsFactors = TRUE)) # categ[1] is factor
+ names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
+ verif <- paste0(categ[1], ".check")
+ }else if(length(categ) == 2L){
+ tempo.data1 <- unique(
+ data.frame(
+ data1[c(categ[1], categ[2])],
+ group = as.integer(factor(paste0(
+ formatC(as.integer(data1[, categ[2]]), width = nchar(max(as.integer(data1[, categ[2]]), na.rm = TRUE)), flag = "0"), # convert factor into numeric with leading zero for proper ranking
+ ".",
+ formatC(as.integer(data1[, categ[1]]), width = nchar(max(as.integer(data1[, categ[1]]), na.rm = TRUE)), flag = "0")# convert factor into numeric with leading zero for proper ranking
+ )), stringsAsFactors = TRUE) # merge the 2 formatC() to create a new factor. The convertion to integer should recreate the correct group number
+ )
+ ) # categ[2] first if categ[2] is used to make the categories in ggplot and categ[1] is used to make the x-axis
+ names(tempo.data1)[names(tempo.data1) == categ[1]] <- paste0(categ[1], ".check")
+ names(tempo.data1)[names(tempo.data1) == categ[2]] <- paste0(categ[2], ".check")
+ verif <- c(paste0(categ[1], ".check"), paste0(categ[2], ".check"))
+ }else{
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ dot.coord.tidy3 <- merge(dot.coord.tidy2, tempo.data1, by = intersect("group", "group"), sort = FALSE) # send the factors of data1 into coord. WARNING: I have tested intersect("group", "group") instead of by = "group". May be come back to by = "group" in case of error. But I did this because of an error in dot.coord.rd3 above
+ if(nrow(dot.coord.tidy3) != nrow(dot.coord) | ( ! fun_comp_2d(dot.coord.tidy3[categ], dot.coord.tidy3[verif])$identical.content)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nTHE merge() FUNCTION DID NOT RETURN A CORRECT dot.coord.tidy3 DATA FRAME. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end coordinates of tidy dots
+ }
+ }
+ # end dot display
+
+
+
+ # boxplot display (if box.fill = FALSE, otherwise, already plotted above)
+ if(box.fill == TRUE){
+ # overcome "work only for the filling of boxes, not for the frame. See https://github.com/tidyverse/ggplot2/issues/252"
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color}, guide = ggplot2::guide_legend(order = 1))) #, guide = ggplot2::guide_legend(override.aes = list(fill = levels(tempo.polygon$COLOR), color = "black")))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = rep(hsv(0, 0, 0, alpha = box.alpha), length(unique(data1[, categ[length(categ)]]))), guide = ggplot2::guide_legend(order = 1))) # , guide = ggplot2::guide_legend(override.aes = list(color = "black", alpha = box.alpha)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor # outline of the polygon in black but with alpha
+ }else{
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_boxplot(data = data1, mapping = ggplot2::aes_string(x = categ[1], y = y, color = categ[length(categ)], fill = categ[length(categ)]), position = ggplot2::position_dodge(width = NULL), width = box.width, size = box.line.size, notch = box.notch, alpha = box.alpha, coef = if(box.whisker.kind == "no"){0}else if(box.whisker.kind == "std"){1.5}else if(box.whisker.kind == "max"){Inf}, outlier.shape = if( ! is.null(dot.color)){NA}else{21}, outlier.color = if( ! is.null(dot.color)){NA}else{if(dot.border.size == 0){NA}else{dot.border.color}}, outlier.fill = if( ! is.null(dot.color)){NA}else{NULL}, outlier.size = if( ! is.null(dot.color)){NA}else{dot.size}, outlier.stroke = if( ! is.null(dot.color)){NA}else{dot.border.size}, outlier.alpha = if( ! is.null(dot.color)){NA}else{dot.alpha})) # the color, size, etc. of the outliers are dealt here. outlier.color = NA to do not plot outliers when dots are already plotted
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_path(
+ data = tempo.polygon,
+ mapping = ggplot2::aes_string(x = "X", y = "Y", group = "BOX", color = categ[length(categ)]),
+ size = box.line.size,
+ alpha = box.alpha,
+ lineend = "round",
+ linejoin = "round"
+ ))
+ coord.names <- c(coord.names, "main.box")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = if(box.notch == FALSE){X_BOX_INF}else{X_NOTCH_INF}, xend = if(box.notch == FALSE){X_BOX_SUP}else{X_NOTCH_SUP}, y = MEDIAN, yend = MEDIAN, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size * 2, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "median")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "sup.whisker")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X, xend = X, y = BOX_INF, yend = WHISK_INF, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha)) #
+ coord.names <- c(coord.names, "inf.whisker")
+ if(box.whisker.width > 0){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_SUP, yend = WHISK_SUP, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha, lineend = "round")) #
+ coord.names <- c(coord.names, "sup.whisker.edge")
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_segment(data = stat, mapping = ggplot2::aes(x = X_WHISK_INF, xend = X_WHISK_SUP, y = WHISK_INF, yend = WHISK_INF, group = categ[length(categ)]), color = stat$COLOR, size = box.line.size, alpha = box.alpha, lineend = "round")) #
+ coord.names <- c(coord.names, "inf.whisker.edge")
+ }
+ if(box.mean == TRUE){
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_point(data = stat, mapping = ggplot2::aes_string(x = "X", y = "MEAN", group = categ[length(categ)]), shape = 23, stroke = box.line.size * 2, color = stat$COLOR, size = box.mean.size, fill = NA, alpha = box.alpha)) # group used in aesthetic to do not have it in the legend. Here ggplot2::scale_discrete_manual() cannot be used because of the group easthetic
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_path(
+ data = tempo.diamon.mean,
+ mapping = ggplot2::aes(x = X, y = Y, group = GROUP),
+ color = tempo.diamon.mean[, "COLOR"],
+ size = box.line.size,
+ alpha = box.alpha,
+ lineend = "round",
+ linejoin = "round"
+ ))
+ coord.names <- c(coord.names, "mean")
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "fill", name = box.legend.name, values = rep(NA, length(unique(data1[, categ[length(categ)]]))))) #, guide = ggplot2::guide_legend(override.aes = list(color = categ.color)))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "color", name = box.legend.name, values = if(length(categ.color)== 1L){rep(categ.color, length(unique(data1[, categ[length(categ)]])))}else{categ.color}, guide = ggplot2::guide_legend(override.aes = list(alpha = if(plot == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list()) == 0L & Sys.info()["sysname"] == "Windows"))){1}else{box.alpha})))) # , guide = ggplot2::guide_legend(override.aes = list(color = as.character(categ.color))))) # values are the values of color (which is the border color in geom_box. WARNING: values = categ.color takes the numbers to make the colors if categ.color is a factor
+ if(plot == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list()) == 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(box.alpha == 0){ # remove box legend because no boxes drawn
+ # add this after the scale_xxx_manual() for boxplots
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none")) # inactivate the legend
+ }
+ # end boxplot display (if box.fill = FALSE, otherwise, already plotted above)
+
+
+
+
+ # stat display
+ # layer after dots but ok, behind dots on the plot
+ if( ! is.null(stat.pos)){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NUMBERS DISPLAYED ARE ", ifelse(stat.mean == FALSE, "MEDIANS", "MEANS"))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ if(stat.pos == "top"){
+ tempo.stat <- data.frame(stat, Y = y.lim[2], stringsAsFactors = TRUE) # I had to create a data frame for geom_tex() so that facet is taken into account, (ggplot2::annotate() does not deal with facet because no data and mapping arguments). Of note, facet.categ is in tempo.stat, via tempo.mean, via dot.coord
+ if(stat.mean == FALSE){tempo.stat$MEDIAN <- formatC(stat.nolog$MEDIAN, digit = 2, drop0trailing = TRUE, format = "f")}else{tempo.stat$MEAN <- formatC(stat.nolog$MEAN, digit = 2, drop0trailing = TRUE, format = "f")}
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
+ data = tempo.stat,
+ mapping = ggplot2::aes_string(x = "X", y = "Y", label = ifelse(stat.mean == FALSE, "MEDIAN", "MEAN")),
+ size = stat.size,
+ color = "black",
+ angle = stat.angle,
+ hjust = stat.just$hjust,
+ vjust = stat.just$vjust
+ )) # stat$X used here because identical to stat.nolog but has the X. WARNING: no need of order() for labels because box.coord$x set the order. For justification, see https://stackoverflow.com/questions/7263849/what-do-hjust-and-vjust-do-when-making-a-plot-using-ggplot
+ coord.names <- c(coord.names, "stat.pos")
+ }else if(stat.pos == "above"){
+ # stat coordinates
+ if( ! is.null(dot.color)){ # for text just above max dot
+ if(dot.tidy == FALSE){
+ tempo.stat.ini <- dot.coord.rd3
+ }else if(dot.tidy == TRUE){
+ tempo.stat.ini <- dot.coord.tidy3
+ tempo.stat.ini$x.y <- tempo.stat.ini$x.x # this is just to be able to use tempo.stat.ini$x.y for untidy or tidy dots (remember that dot.coord.tidy3$x.y is not good, see above)
+ }
+ stat.coord1 <- aggregate(x = tempo.stat.ini["y"], by = {x.env <- if(length(categ)== 1L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]])}else if(length(categ) == 2L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]], tempo.stat.ini[, categ[2]])} ; names(x.env) <- if(length(categ)== 1L){c("group", "PANEL", "x.y", categ[1])}else if(length(categ) == 2L){c("group", "PANEL", "x.y", categ[1], categ[2])} ; x.env}, FUN = min, na.rm = TRUE)
+ names(stat.coord1)[names(stat.coord1) == "y"] <- "dot.min"
+ stat.coord2 <- aggregate(x = tempo.stat.ini["y"], by = {x.env <- if(length(categ)== 1L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]])}else if(length(categ) == 2L){list(tempo.stat.ini$group, tempo.stat.ini$PANEL, tempo.stat.ini$x.y, tempo.stat.ini[, categ[1]], tempo.stat.ini[, categ[2]])} ; names(x.env) <- if(length(categ)== 1L){c("group", "PANEL", "x.y", categ[1])}else if(length(categ) == 2L){c("group", "PANEL", "x.y", categ[1], categ[2])} ; x.env}, FUN = max, na.rm = TRUE)
+ names(stat.coord2) <- paste0(names(stat.coord2), "_from.dot.max")
+ names(stat.coord2)[names(stat.coord2) == "y_from.dot.max"] <- "dot.max"
+ stat.coord3 <- cbind(box.coord[order(box.coord$group, box.coord$PANEL), ], stat.coord1[order(stat.coord1$group, stat.coord1$x.y), ], stat.coord2[order(stat.coord2$group, stat.coord2$x.y), ], stringsAsFactors = TRUE) #
+ if( ! all(identical(round(stat.coord3$x, 9), round(as.numeric(stat.coord3$x.y), 9)), na.rm = TRUE)){ # as.numeric() because stat.coord3$x is class "mapped_discrete" "numeric"
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nFUSION OF box.coord, stat.coord1 AND stat.coord2 ACCORDING TO box.coord$x, stat.coord1$x.y AND stat.coord2$x.y IS NOT CORRECT. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # text.coord <- stat.coord3[, c("x", "group", "dot.min", "dot.max")]
+ # names(text.coord)[names(text.coord) == "dot.min"] <- "text.min.pos"
+ #names(text.coord)[names(text.coord) == "dot.max"] <- "text.max.pos"
+ box.coord <- box.coord[order(box.coord$x, box.coord$group, box.coord$PANEL), ]
+ # text.coord <- text.coord[order(text.coord$x), ] # to be sure to have the two objects in the same order for x. WARNING: cannot add identical(as.integer(text.coord$group), as.integer(box.coord$group)) because with error, the correspondence between x and group is not the same
+ stat.coord3 <- stat.coord3[order(stat.coord3$x, stat.coord3$group, stat.coord3$PANEL), ] # to be sure to have the two objects in the same order for x. WARNING: cannot add identical(as.integer(text.coord$group), as.integer(box.coord$group)) because with error, the correspondence between x and group is not the same
+ if( ! (identical(box.coord$x, stat.coord3$x) & identical(box.coord$group, stat.coord3$group) & identical(box.coord$PANEL, stat.coord3$PANEL))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\ntext.coord AND box.coord DO NOT HAVE THE SAME x, group AND PANEL COLUMN CONTENT")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }else{
+ stat.coord3 <- box.coord
+ }
+ stat.coord3 <- data.frame(
+ stat.coord3,
+ Y = stat.coord3[, ifelse(
+ is.null(dot.color),
+ ifelse(diff(y.lim) > 0, "ymax", "ymin"),
+ ifelse(diff(y.lim) > 0, "ymax_final", "ymin_final")
+ )],
+ stringsAsFactors = TRUE
+ ) # ymax is top whisker, ymax_final is top dot
+ # stat.coord3 <- data.frame(stat.coord3, Y = vector("numeric", length = nrow(stat.coord3)), stringsAsFactors = TRUE)
+ # check.Y <- as.logical(stat.coord3$Y) # convert everything in Y into FALSE (because Y is full of zero)
+ # end stat coordinates
+ # stat display
+ # performed twice: first for y values >=0, then y values < 0, because only a single value allowed for hjust anf vjust
+ if(stat.mean == FALSE){
+ tempo.center.ref <- "middle"
+ }else{
+ tempo.center.ref <- "MEAN"
+ }
+ # if(is.null(dot.color)){
+ # tempo.low.ref <- "ymin"
+ # tempo.high.ref <- "ymax"
+ # }else{
+ # tempo.low.ref <- "ymin_final"
+ # tempo.high.ref <- "ymax_final"
+ # }
+ # tempo.log.high <- if(diff(y.lim) > 0){stat.coord3[, tempo.center.ref] >= 0}else{stat.coord3[, tempo.center.ref] < 0}
+ # tempo.log.low <- if(diff(y.lim) > 0){stat.coord3[, tempo.center.ref] < 0}else{stat.coord3[, tempo.center.ref] >= 0}
+ # stat.coord3$Y[tempo.log.high] <- stat.coord3[tempo.log.high, tempo.high.ref]
+ # stat.coord3$Y[tempo.log.low] <- stat.coord3[tempo.log.low, tempo.low.ref]
+ # add distance
+ stat.coord3$Y <- stat.coord3$Y + diff(y.lim) * stat.dist / 100
+ # end add distance
+ # correct median or mean text format
+ if(y.log != "no"){
+ stat.coord3[, tempo.center.ref] <- ifelse(y.log == "log2", 2, 10)^(stat.coord3[, tempo.center.ref])
+ }
+ stat.coord3[, tempo.center.ref] <- formatC(stat.coord3[, tempo.center.ref], digit = 2, drop0trailing = TRUE, format = "f")
+ # end correct median or mean text format
+ # if(any(tempo.log.high) == TRUE){
+ # tempo.stat <- stat.coord3[tempo.log.high,]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
+ data = stat.coord3,
+ mapping = ggplot2::aes_string(x = "x", y = "Y", label = tempo.center.ref),
+ size = stat.size,
+ color = "black",
+ angle = stat.angle,
+ hjust = stat.just$hjust,
+ vjust = stat.just$vjust
+ )) # WARNING: no need of order() for labels because box.coord$x set the order
+ coord.names <- c(coord.names, "stat.pos")
+ # }
+ # if(any(tempo.log.low) == TRUE){
+ # tempo.stat <- stat.coord3[tempo.log.low,]
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::geom_text(
+ # data = tempo.stat,
+ # mapping = ggplot2::aes_string(x = "x", y = "Y", label = tempo.center.ref),
+ # size = stat.size,
+ # color = "black",
+ # hjust = ifelse(vertical == TRUE, 0.5, 0.5 + stat.dist),
+ # vjust = ifelse(vertical == TRUE, 0.5 + stat.dist, 0.5)
+ # )) # WARNING: no need of order() for labels because box.coord$x set the order
+ # coord.names <- c(coord.names, "stat.pos.negative")
+ # }
+ # end stat display
+ }else{
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 5")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ }
+ # end stat display
+ # legend management
+ if(legend.show == FALSE){ # must be here because must be before bef.final.plot <-
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none", alpha = "none")) # inactivate the initial legend
+ }
+ # end legend management
+
+
+
+ # y scale management (cannot be before dot plot management)
+ # the rescaling aspect is complicated and not intuitive. See:
+ # explaination: https://github.com/tidyverse/ggplot2/issues/3948
+ # the oob argument of scale_y_continuous() https://ggplot2.tidyverse.org/reference/scale_continuous.html
+ # see also https://github.com/rstudio/cheatsheets/blob/master/data-visualization-2.1.pdf
+ # secondary ticks
+ bef.final.plot <- ggplot2::ggplot_build(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + if(vertical == TRUE){ggplot2::scale_y_continuous(expand = c(0, 0), limits = sort(y.lim), oob = scales::rescale_none)}else{ggplot2::coord_flip(ylim = y.lim)}')))) # here I do not need the x-axis and y-axis orientation, I just need the number of main ticks and the legend. I DI NOT UNDERSTAND THE COMMENT HERE BECAUSE WE NEED COORD_FLiP
+ tempo.coord <- bef.final.plot$layout$panel_params[[1]]
+ # y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non NULL or if y.log argument is different from "no")
+ if(y.log != "no"){ # integer main ticks for log2 and log10
+ tempo.scale <- (as.integer(min(y.lim, na.rm = TRUE)) - 1):(as.integer(max(y.lim, na.rm = TRUE)) + 1)
+ }else{
+ tempo <- if(is.null(attributes(tempo.coord$y$breaks))){tempo.coord$y$breaks}else{unlist(attributes(tempo.coord$y$breaks))}
+ if(all(is.na(tempo))){# all() without na.rm -> ok because is.na() cannot be NA
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$y$breaks")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ tempo.scale <- fun_scale(lim = y.lim, n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks. If fact: n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) replaced by n = ifelse(is.null(y.tick.nb), 4, y.tick.nb))
+ }
+ y.second.tick.values <- NULL
+ y.second.tick.pos <- NULL
+ if(y.log != "no"){
+ tempo <- fun_inter_ticks(lim = y.lim, log = y.log)
+ y.second.tick.values <- tempo$values
+ y.second.tick.pos <- tempo$coordinates
+ # if(vertical == TRUE){ # do not remove in case the bug is fixed
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", y = y.second.tick.pos, yend = y.second.tick.pos, x = tempo.coord$x.range[1], xend = tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80))
+ # }else{ # not working because of the ggplot2 bug
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", x = y.second.tick.pos, xend = y.second.tick.pos, y = tempo.coord$y.range[1], yend = tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80))
+ # }
+ coord.names <- c(coord.names, "y.second.tick.positions")
+ }else if(( ! is.null(y.second.tick.nb)) & y.log == "no"){
+ # if(y.second.tick.nb > 0){ #inactivated because already checked before
+ if(length(tempo.scale) < 2){
+ tempo.cat1 <- c("y.tick.nb", "y.second.tick.nb")
+ tempo.cat2 <- sapply(list(y.tick.nb, y.second.tick.nb), FUN = paste0, collapse = " ")
+ tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE Y-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ tempo <- fun_inter_ticks(lim = y.lim, log = y.log, breaks = tempo.scale, n = y.second.tick.nb)
+ }
+ y.second.tick.values <- tempo$values
+ y.second.tick.pos <- tempo$coordinates
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
+ geom = "segment",
+ y = y.second.tick.pos,
+ yend = y.second.tick.pos,
+ x = if(vertical == TRUE){tempo.coord$x.range[1]}else{tempo.coord$y.range[1]},
+ xend = if(vertical == TRUE){tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80}else{tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80}
+ ))
+ coord.names <- c(coord.names, "y.second.tick.positions")
+ }
+ # end y.second.tick.positions
+ # for the ggplot2 bug with y.log, this does not work: eval(parse(text = ifelse(vertical == FALSE & y.log == "log10", "ggplot2::scale_x_continuous", "ggplot2::scale_y_continuous")))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_y_continuous(
+ breaks = tempo.scale,
+ minor_breaks = y.second.tick.pos,
+ labels = if(y.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(y.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(y.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 6") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)}, # == in stop() to be able to add several messages between ==
+ expand = c(0, 0), # remove space after after axis limits
+ limits = sort(y.lim), # NA indicate that limits must correspond to data limits but ylim() already used
+ oob = scales::rescale_none,
+ trans = ifelse(diff(y.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_y_reverse() but create the problem of y-axis label disappearance with y.lim decreasing. Thus, do not use. Use ylim() below and after this
+ ))
+ if(vertical == TRUE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_cartesian(ylim = y.lim)) # problem of ggplot2::ylim() is that it redraws new breaks # coord_cartesian(ylim = y.lim)) not used because bug -> y-axis label disappearance with y.lim decreasing I DO NOT UNDERSTAND THIS MESSAGE WHILE I USE COORD_CARTESIAN # clip = "off" to have secondary ticks outside plot region does not work
+ }else{
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_flip(ylim = y.lim)) # clip = "off" to have secondary ticks outside plot region does not work # create the problem of y-axis label disappearance with y.lim decreasing. IDEM ABOVE
+
+ }
+ # end y scale management (cannot be before dot plot management)
+
+
+ # legend management
+ if( ! is.null(legend.width)){
+ legend.final <- fun_gg_get_legend(ggplot_built = bef.final.plot, fun.name = function.name, lib.path = lib.path) # get legend
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::guides(fill = "none", color = "none", alpha = "none")) # inactivate the initial legend
+ if(is.null(legend.final) & plot == TRUE){ # even if any(unlist(legend.disp)) is TRUE
+ legend.final <- ggplot2::ggplot()+ggplot2::theme_void() # empty graph instead of legend
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON NULL legend.width ARGUMENT\n")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end legend management
+
+
+ # drawing
+ fin.plot <- suppressMessages(suppressWarnings(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ")))))
+ grob.save <- NULL
+ if(plot == TRUE){
+ # following lines inactivated because of problem in warn.recov and message.recov
+ # assign("env_fun_get_message", new.env())
+ # assign("tempo.gg.name", tempo.gg.name, envir = env_fun_get_message)
+ # assign("tempo.gg.count", tempo.gg.count, envir = env_fun_get_message)
+ # assign("add", add, envir = env_fun_get_message)
+ # two next line: for the moment, I cannot prevent the warning printing
+ # warn.recov <- fun_get_message(paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}), kind = "warning", header = FALSE, print.no = FALSE, env = env_fun_get_message) # for recovering warnings printed by ggplot() functions
+ # message.recov <- fun_get_message('print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))', kind = "message", header = FALSE, print.no = FALSE, env = env_fun_get_message) # for recovering messages printed by ggplot() functions
+ # if( ! (return == TRUE & return.ggplot == TRUE)){ # because return() plots when return.ggplot is TRUE # finally not used -> see return.ggplot description
+ if(is.null(legend.width)){
+ grob.save <- suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot)))
+ }else{
+ grob.save <-suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot, legend.final, ncol=2, widths=c(1, legend.width))))
+ }
+ # }
+ # suppressMessages(suppressWarnings(print(eval(parse(text = paste(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), if(is.null(add)){NULL}else{add}))))))
+ }else{
+ # following lines inactivated because of problem in warn.recov and message.recov
+ # message.recov <- NULL
+ # warn.recov <- NULL
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end drawing
+
+
+
+ # output
+ # following lines inactivated because of problem in warn.recov and message.recov
+ # if( ! (is.null(warn) & is.null(warn.recov) & is.null(message.recov))){
+ # warn <- paste0(warn, "\n\n", if(length(warn.recov) > 0 | length(message.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", ifelse( ! is.null(warn.recov), unique(message.recov), ""), ifelse( ! is.null(message.recov), unique(message.recov), ""), collapse = "\n\n"), "\n\n")})
+ # }else if( ! (is.null(warn) & is.null(warn.recov)) & is.null(message.recov)){
+ # warn <- paste0(warn, "\n\n", if(length(warn.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", unique(warn.recov), collapse = "\n\n"), "\n\n")})
+ # }else if( ! (is.null(warn) & is.null(message.recov)) & is.null(warn.recov)){
+ # warn <- paste0(warn, "\n\n", if(length(message.recov) > 0){paste0(paste0("MESSAGES FROM ggplot2 FUNCTIONS: ", unique(message.recov), collapse = "\n\n"), "\n\n")})
+ # }
+ if(warn.print == TRUE & ! is.null(warn)){
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ }
+ on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
+ if(return == TRUE){
+ tempo.output <- ggplot2::ggplot_build(fin.plot)
+ tempo.output$data <- tempo.output$data[-1] # remove the first data because corresponds to the initial empty boxplot
+ if(length(tempo.output$data) != length(coord.names)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nlength(tempo.output$data) AND length(coord.names) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ names(tempo.output$data) <- coord.names
+ tempo.output$data <- tempo.output$data[coord.names != "bad_remove"]
+ }
+ tempo <- tempo.output$layout$panel_params[[1]]
+ output <- list(
+ data = data1.ini,
+ stat = stat.nolog,
+ removed.row.nb = removed.row.nb,
+ removed.rows = removed.rows,
+ plot = c(tempo.output$data, y.second.tick.values = list(y.second.tick.values)),
+ panel = facet.categ,
+ axes = list(
+ x.range = tempo$x.range,
+ x.labels = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{tempo$x$scale$get_labels()}, # is.null(attributes(tempo$x$breaks)) test if it is number (TRUE) or character (FALSE)
+ x.positions = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{unlist(attributes(tempo$x$breaks))},
+ y.range = tempo$y.range,
+ y.labels = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{tempo$y$scale$get_labels()},
+ y.positions = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{unlist(attributes(tempo$y$breaks))}
+ ),
+ warn = paste0("\n", warn, "\n\n"),
+ ggplot = if(return.ggplot == TRUE){fin.plot}else{NULL}, # fin.plot plots the graph if return == TRUE
+ gtable = if(return.gtable == TRUE){grob.save}else{NULL}
+ )
+ return(output) # this plots the graph if return.ggplot is TRUE and if no assignment
+ }
+ # end output
+ # end main code
}
-# end loop part
+# add density
+# rasterise all kind: https://cran.r-project.org/web/packages/ggrastr/vignettes/Raster_geoms.html
-# legend display
-tempo.legend.final <- 'ggplot2::guides(
+
+fun_gg_scatter <- function(
+ data1,
+ x,
+ y,
+ categ = NULL,
+ categ.class.order = NULL,
+ color = NULL,
+ geom = "geom_point",
+ geom.step.dir = "hv",
+ geom.stick.base = NULL,
+ alpha = 0.5,
+ dot.size = 2,
+ dot.shape = 21,
+ dot.border.size = 0.5,
+ dot.border.color = NULL,
+ line.size = 0.5,
+ line.type = "solid",
+ x.lim = NULL,
+ x.lab = NULL,
+ x.log = "no",
+ x.tick.nb = NULL,
+ x.second.tick.nb = NULL,
+ x.include.zero = FALSE,
+ x.left.extra.margin = 0.05,
+ x.right.extra.margin = 0.05,
+ x.text.angle = 0,
+ y.lim = NULL,
+ y.lab = NULL,
+ y.log = "no",
+ y.tick.nb = NULL,
+ y.second.tick.nb = NULL,
+ y.include.zero = FALSE,
+ y.top.extra.margin = 0.05,
+ y.bottom.extra.margin = 0.05,
+ y.text.angle = 0,
+ raster = FALSE,
+ raster.ratio = 1,
+ raster.threshold = NULL,
+ text.size = 12,
+ title = "",
+ title.text.size = 12,
+ legend.show = TRUE,
+ legend.width = 0.5,
+ legend.name = NULL,
+ article = TRUE,
+ grid = FALSE,
+ add = NULL,
+ return = FALSE,
+ return.ggplot = FALSE,
+ return.gtable = TRUE,
+ plot = TRUE,
+ warn.print = FALSE,
+ lib.path = NULL
+){
+ # AIM
+ # Plot ggplot2 scatterplot with the possibility to overlay dots from up to 3 different data frames (-> three different legends) and lines from up to 3 different data frames (-> three different legends) -> up to 6 overlays totally
+ # For ggplot2 specifications, see: https://ggplot2.tidyverse.org/articles/ggplot2-specs.html
+ # WARNINGS
+ # Rows containing NA in data1[, c(x, y, categ)] will be removed before processing, with a warning (see below)
+ # Size arguments (dot.size, dot.border.size, line.size, text.size and title.text.size) are in mm. See Hadley comment in https://stackoverflow.com/questions/17311917/ggplot2-the-unit-of-size. See also http://sape.inf.usi.ch/quick-reference/ggplot2/size). Unit object are not accepted, but conversion can be used (e.g., grid::convertUnit(grid::unit(0.2, "inches"), "mm", valueOnly = TRUE))
+ # ARGUMENTS
+ # data1: a dataframe compatible with ggplot2, or a list of data frames. Order matters for the order of the legend and for the layer staking (starting from below to top)
+ # x: single character string of the data1 column name for x-axis coordinates. If data1 is a list, then x must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Write NULL for each "geom_hline" in geom argument
+ # y: single character string of the data1 column name for y-axis coordinates. If data1 is a list, then y must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Write NULL for each "geom_vline" in geom argument
+ # categ: either NULL or a single character string or a list of single character strings, indicating the data1 column names to use for categories which creates legend display
+ # If categ == NULL, no categories -> no legend displayed
+ # If data1 is a data frame, categ must be a single character string of the data1 column name for categories
+ # If data1 is a list, then categ must be a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL (no legend display for these compartments), and other not
+ # categ.class.order: either (1) NULL or (2) a vector of character strings or (3) a list of these vectors, setting the order of the classes of categ in the legend display
+ # If categ.class.order is NULL, classes are represented according to the alphabetical order
+ # If data1 is a data frame, categ.class.order must be a vector of character strings specifying the different classes in the categ column name of data1
+ # If data1 is a list, then categ.class.order must be a list of vector of character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL (alphabetical order for these compartments), and other not
+ # color: either (1) NULL, or (2) a vector of character strings or integers, or (3) a list of vectors of character strings or integers
+ # If color is NULL, default colors of ggplot2
+ # If data1 is a data frame, color argument can be either:
+ # (1) a single color string. All the dots of the corresponding data1 will have this color, whatever the categ value (NULL or not)
+ # (2) if categ is non-null, a vector of string colors, one for each class of categ. Each color will be associated according to the categ.class.order argument if specified, or to the alphabetical order of categ classes otherwise
+ # (3) if categ is non-null, a vector or factor of string colors, like if it was one of the column of data1 data frame. WARNING: a single color per class of categ and a single class of categ per color must be respected
+ # Positive integers are also accepted instead of character strings, as long as above rules about length are respected. Integers will be processed by fun_gg_palette() using the max integer value among all the integers in color (see fun_gg_palette())
+ # If data1 is a list, then color argument must be either:
+ # (1) a list of character strings or integers, of same size as data1, with compartment 1 related to compartment 1 of data1, etc.
+ # (2) a single character string or a single integer
+ # With a list (first possibility), the rules described for when data1 is a data frame apply to each compartment of the list. Some of the compartments can be NULL. In that case, a different grey color will be used for each NULL compartment. With a single value (second possibility), the same color will be used for all the dots and lines, whatever the data1 list
+ # geom: single character string of the kind of plot, or a list of single character strings
+ # Either:
+ # "geom_point" (scatterplot)
+ # "geom_line" (coordinates plotted then line connection, from the lowest to highest x coordinates first and from the lowest to highest y coordinates thenafter)
+ # "geom_path" (coordinates plotted then line connection respecting the row order in data1)
+ # "geom_step" coordinates plotted then line connection respecting the row order in data1 but drawn in steps). See the geom.step.dir argument
+ # "geom_hline" (horizontal line, no x value provided)
+ # "geom_vline" (vertical line, no y value provided)
+ # "geom_stick" (dots as vertical bars)
+ # If data1 is a list, then geom must be either:
+ # (1) a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc.
+ # (2) a single character string. In that case the same kind of plot will apply for the different compartments of the data1 list
+ # WARNING concerning "geom_hline" or "geom_vline":
+ # (1) x or y argument must be NULL, respectively
+ # (2) x.lim or y.lim argument must NOT be NULL, respectively, if only these kind of lines are drawn (if other geom present, then x.lim = NULL and y.lim = NULL will generate x.lim and y.lim defined by these other geom, which is not possible with "geom_hline" or "geom_vline" alone)
+ # (3) the function will draw n lines for n values in the x argument column name of the data1 data frame. If several colors required, the categ argument must be specified and the corresponding categ column name must exist in the data1 data frame with a different class name for each row
+ # geom.step.dir: single character string indicating the direction when using "geom_step" of the geom argument, or a list of single character strings
+ # Either:
+ # "vh" (vertical then horizontal)
+ # "hv" (horizontal then vertical)
+ # "mid" (step half-way between adjacent x-values)
+ # See https://ggplot2.tidyverse.org/reference/geom_path.html
+ # If data1 is a list, then geom.step.dir must be either:
+ # (1) a list of single character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. The value in compartments related to other geom values than "geom_step" will be ignored
+ # (2) a single character string, which will be used for all the "geom_step" values of the geom argument, whatever the data1 list
+ # geom.stick.base: either (1) NULL or (2) a single numeric value or (3) a list of single numeric values, setting the base of the sticks when using "geom_stick" of the geom argument
+ # If geom.stick.base is NULL, the bottom limit of the y-axis is taken as the base
+ # If data1 is a list, then geom.stick.base must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the values in compartments related to other geom values than "geom_stick" will be ignored. With a single value (latter possibility), the same base will be used for all the sticks, whatever the data1 list
+ # Warning: the y-axis limits are not modified by the value of geom.stick.base, meaning that this value can be outside of the range of y.lim. Add the value of geom.stick.base also in the y.lim argument if required
+ # Warning: if geom.stick.base is NULL, the bottom limit of the y-axis is taken as the base. Thus, be careful with inverted y-axis
+ # alpha: single numeric value (from 0 to 1) of transparency. If data1 is a list, then alpha must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. In that case the same transparency will apply for the different compartments of the data1 list
+ # dot.size: single numeric value of dot shape radius? in mm. If data1 is a list, then dot.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.size will be used for all the dots, whatever the data1 list
+ # dot.shape: value indicating the shape of the dots (see https://ggplot2.tidyverse.org/articles/ggplot2-specs.html) If data1 is a list, then dot.shape must be either (1) a list of single shape values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single shape value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.shape will be used for all the dots, whatever the data1 list
+ # dot.border.size: single numeric value of border dot width in mm. Write zero for no dot border. If data1 is a list, then dot.border.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to lines will be ignored. With a single value (latter possibility), the same dot.border.size will be used for all the dots, whatever the data1 list
+ # dot.border.color: single character color string defining the color of the dot border (same border color for all the dots, whatever their categories). If dot.border.color == NULL, the border color will be the same as the dot color. A single integer is also accepted instead of a character string, that will be processed by fun_gg_palette()
+ # line.size: single numeric value of line width in mm. If data1 is a list, then line.size must be either (1) a list of single numeric values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single numeric value. With a list (former possibility), the value in compartments related to dots will be ignored. With a single value (latter possibility), the same line.size will be used for all the lines, whatever the data1 list
+ # line.type: value indicating the kind of lines (see https://ggplot2.tidyverse.org/articles/ggplot2-specs.html) If data1 is a list, then line.type must be either (1) a list of single line kind values, of same size as data1, with compartment 1 related to compartment 1 of data1, etc., or (2) a single line kind value. With a list (former possibility), the value in compartments related to dots will be ignored. With a single value (latter possibility), the same line.type will be used for all the lines, whatever the data1 list
+ # x.lim: 2 numeric values setting the x-axis range. Order of the 2 values matters (for inverted axis). If NULL, the range of the x column name of data1 will be used
+ # x.lab: a character string or expression for x-axis label. If NULL, will use the first value of x (x column name of the first data frame in data1). Warning message if the elements in x are different between data frames in data1
+ # x.log: either "no", "log2" (values in the x column name of the data1 data frame will be log2 transformed and x-axis will be log2 scaled) or "log10" (values in the x column name of the data1 data frame will be log10 transformed and x-axis will be log10 scaled)
+ # x.tick.nb: approximate number of desired values labeling the x-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if x.log is "no", then the number of labeling values is set by ggplot2. If NULL and if x.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the x.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for x.lim = c(9, 1200)). WARNING: if non-NULL and if x.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
+ # x.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if x.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$x.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
+ # x.include.zero: logical. Does x.lim range include 0? Ignored if x.log is "log2" or "log10"
+ # x.left.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to x.lim. If different from 0, add the range of the axis multiplied by x.left.extra.margin (e.g., abs(x.lim[2] - x.lim[1]) * x.left.extra.margin) to the left of x-axis
+ # x.right.extra.margin: idem as x.left.extra.margin but to the right of x-axis
+ # x.text.angle: integer value of the text angle for the x-axis labeling values, using the same rules as in ggplot2. Use positive value for clockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Use negative values for counterclockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
+ # y.lim: 2 numeric values setting the y-axis range. Order of the 2 values matters (for inverted axis). If NULL, the range of the y column name of data1 will be used
+ # y.lab: a character string or expression for y-axis label. If NULL, will use the first value of y (y column name of the first data frame in data1). Warning message if the elements in y are different between data frames in data1
+ # y.log: either "no", "log2" (values in the y column name of the data1 data frame will be log2 transformed and y-axis will be log2 scaled) or "log10" (values in the y column name of the data1 data frame will be log10 transformed and y-axis will be log10 scaled)
+ # y.tick.nb: approximate number of desired values labeling the y-axis (i.e., main ticks, see the n argument of the the cute::fun_scale() function). If NULL and if y.log is "no", then the number of labeling values is set by ggplot2. If NULL and if y.log is "log2" or "log10", then the number of labeling values corresponds to all the exposant integers in the y.lim range (e.g., 10^1, 10^2 and 10^3, meaning 3 main ticks for y.lim = c(9, 1200)). WARNING: if non-NULL and if y.log is "log2" or "log10", labeling can be difficult to read (e.g., ..., 10^2, 10^2.5, 10^3, ...)
+ # y.second.tick.nb: number of desired secondary ticks between main ticks. Ignored if y.log is other than "no" (log scale plotted). Use argument return = TRUE and see $plot$y.second.tick.values to have the values associated to secondary ticks. IF NULL, no secondary ticks
+ # y.include.zero: logical. Does y.lim range include 0? Ignored if y.log is "log2" or "log10"
+ # y.top.extra.margin: single proportion (between 0 and 1) indicating if extra margins must be added to y.lim. If different from 0, add the range of the axis multiplied by y.top.extra.margin (e.g., abs(y.lim[2] - y.lim[1]) * y.top.extra.margin) to the top of y-axis
+ # y.bottom.extra.margin: idem as y.top.extra.margin but to the bottom of y-axis
+ # y.text.angle: integer value of the text angle for the y-axis labeling values, using the same rules as in ggplot2. Use positive value for clockwise rotation: 0 for horizontal, 90 for vertical, 180 for upside down etc. Use negative values for counterclockwise rotation: 0 for horizontal, -90 for vertical, -180 for upside down etc.
+ # raster: logical. Dots in raster mode? If FALSE, dots from each "geom_point" from geom argument are plotted in vectorial mode (bigger pdf and long to display if lots of dots). If TRUE, dots from each "geom_point" from geom argument are plotted in matricial mode (smaller pdf and easy display if lots of dots, but it takes time to generate the layer). If TRUE, the raster.ratio argument is used to avoid an ellipsoid representation of the dots. If TRUE, solve the transparency problem with some GUI. Overriden by the non-NULL raster.threshold argument
+ # raster.ratio: single numeric value indicating the height / width ratio of the graphic device used (for instance provided by the $dim compartment in the output of the fun_open() function). The default value is 1 because by default R opens a square graphic device. But this argument has to be set when using other device dimensions. Ignored if raster == FALSE
+ # raster.threshold: positive integer value indicating the limit of the dot number above which "geom_point" layers from the geom argument switch from vectorial mode to matricial mode (see the raster argument). If any layer is matricial, then the raster.ratio argument is used to avoid an ellipsoid representation of the dots. If non-NULL, it overrides the raster argument
+ # text.size: numeric value of the font size of the (1) axis numbers and axis legends and (2) texts in the graphic legend (in mm)
+ # title: character string of the graph title
+ # title.text.size: numeric value of the title font size in mm
+ # legend.show: logical. Show legend? Not considered if categ argument is NULL, because this already generate no legend, excepted if legend.width argument is non-NULL. In that specific case (categ is NULL, legend.show is TRUE and legend.width is non-NULL), an empty legend space is created. This can be useful when desiring graphs of exactly the same width, whatever they have legends or not
+ # legend.width: single proportion (between 0 and 1) indicating the relative width of the legend sector (on the right of the plot) relative to the width of the plot. Value 1 means that the window device width is split in 2, half for the plot and half for the legend. Value 0 means no room for the legend, which will overlay the plot region. Write NULL to inactivate the legend sector. In such case, ggplot2 will manage the room required for the legend display, meaning that the width of the plotting region can vary between graphs, depending on the text in the legend
+ # legend.name: character string of the legend title. If legend.name is NULL and categ argument is not NULL, then legend.name <- categ. If data1 is a list, then legend.name must be a list of character strings, of same size as data1, with compartment 1 related to compartment 1 of data1, etc. Some of the list compartments can be NULL, and other not
+ # article: logical. If TRUE, use an article theme (article like). If FALSE, use a classic related ggplot theme. Use the add argument (e.g., add = "+ggplot2::theme_classic()" for the exact classic ggplot theme
+ # grid: logical. Draw lines in the background to better read the box values? Not considered if article == FALSE (grid systematically present)
+ # add: character string allowing to add more ggplot2 features (dots, lines, themes, facet, etc.). Ignored if NULL
+ # WARNING: (1) the string must start with "+", (2) the string must finish with ")" and (3) each function must be preceded by "ggplot2::". Example: "+ ggplot2::coord_flip() + ggplot2::theme_bw()"
+ # If the character string contains the "ggplot2::theme" string, then the article argument of fun_gg_scatter() (see above) is ignored with a warning. In addition, some arguments can be overwritten, like x.angle (check all the arguments)
+ # Handle the add argument with caution since added functions can create conflicts with the preexisting internal ggplot2 functions
+ # WARNING: the call of objects inside the quotes of add can lead to an error if the name of these objects are some of the fun_gg_scatter() arguments. Indeed, the function will use the internal argument instead of the global environment object. Example article <- "a" in the working environment and add = '+ ggplot2::ggtitle(article)'. The risk here is to have TRUE as title. To solve this, use add = '+ ggplot2::ggtitle(get("article", envir = .GlobalEnv))'
+ # return: logical. Return the graph parameters?
+ # return.ggplot: logical. Return the ggplot object in the output list? Ignored if return argument is FALSE. WARNING: always assign the fun_gg_scatter() function (e.g., a <- fun_gg_scatter()) if return.ggplot argument is TRUE, otherwise, double plotting is performed. See $ggplot in the RETURN section below for more details
+ # return.gtable: logical. Return the ggplot object as gtable of grobs in the output list? Ignored if plot argument is FALSE. Indeed, the graph must be plotted to get the grobs dispositions. See $gtable in the RETURN section below for more details
+ # plot: logical. Plot the graphic? If FALSE and return argument is TRUE, graphical parameters and associated warnings are provided without plotting
+ # warn.print: logical. Print warnings at the end of the execution? ? If FALSE, warning messages are never printed, but can still be recovered in the returned list. Some of the warning messages (those delivered by the internal ggplot2 functions) are not apparent when using the argument plot = FALSE
+ # lib.path: character string indicating the absolute path of the required packages (see below). if NULL, the function will use the R library default folders
+ # RETURN
+ # a scatter plot if plot argument is TRUE
+ # a list of the graph info if return argument is TRUE:
+ # $data: the initial data with graphic information added. WARNING: if the x.log or y.log argument is not "no", x or y argument column of the data1 data frame are log2 or log10 converted in $data, respectively. Use 2^values or 10^$values to recover the initial values
+ # $removed.row.nb: a list of the removed rows numbers in data frames (because of NA). NULL if no row removed
+ # $removed.rows: a list of the removed rows in data frames (because of NA). NULL if no row removed
+ # $plot: the graphic box and dot coordinates
+ # $dots: dot coordinates
+ # y.second.tick.positions: coordinates of secondary ticks (only if y.second.tick.nb argument is non-null or if y.log argument is different from "no")
+ # y.second.tick.values: values of secondary ticks. NULL except if y.second.tick.nb argument is non-null or if y.log argument is different from "no")
+ # $panel: the variable names used for the panels (NULL if no panels). WARNING: NA can be present according to ggplot2 upgrade to v3.3.0
+ # $axes: the x-axis and y-axis info
+ # $warn: the warning messages. Use cat() for proper display. NULL if no warning. WARNING: warning messages delivered by the internal ggplot2 functions are not apparent when using the argument plot = FALSE
+ # $ggplot: ggplot object that can be used for reprint (use print($ggplot) or update (use $ggplot + ggplot2::...). NULL if return.ggplot argument is FALSE. Of note, a non-null $ggplot in the output list is sometimes annoying as the manipulation of this list prints the plot
+ # $gtable: gtable object that can be used for reprint (use gridExtra::grid.arrange(...$ggplot) or with additionnal grobs (see the grob decomposition in the examples). NULL if return.ggplot argument is FALSE. Contrary to $ggplot, a non-NULL $gtable in the output list is not annoying as the manipulation of this list does not print the plot
+ # REQUIRED PACKAGES
+ # ggplot2
+ # gridExtra
+ # lemon (in case of use in the add argument)
+ # scales
+ # if raster plots are drawn (see the raster and raster.threshold arguments):
+ # Cairo
+ # grid
+ # REQUIRED FUNCTIONS FROM THE cute PACKAGE
+ # fun_gg_empty_graph()
+ # fun_gg_palette()
+ # fun_gg_point_rast()
+ # fun_pack()
+ # fun_check()
+ # fun_round()
+ # fun_scale()
+ # fun_inter_ticks()
+ # EXAMPLES
+ # set.seed(1) ; obs1 <- data.frame(Km = c(2, 1, 6, 5, 4, 7), Time = c(2, 1, 6, 5, 4, 7)^2, Car = c("TUUT", "TUUT", "TUUT", "WIIM", "WIIM", "WIIM"), Color1 = rep(c("coral", "lightblue"), each = 3), stringsAsFactors = TRUE) ; fun_gg_scatter(data1 = obs1, x = "Km", y = "Time")
+ # DEBUGGING
+ # set.seed(1) ; obs1 <- data.frame(km = rnorm(1000, 10, 3), time = rnorm(1000, 10, 3), group1 = rep(c("A1", "A2"), 500), stringsAsFactors = TRUE) ; obs2 <-data.frame(km = rnorm(1000, 15, 3), time = rnorm(1000, 15, 3), group2 = rep(c("G1", "G2"), 500), stringsAsFactors = TRUE) ; set.seed(NULL) ; obs1$km[2:3] <- NA ; data1 = list(L1 = obs1, L2 = obs2) ; x = list(L1 = "km", L2 = "km") ; y = list(L1 = "time", L2 = "time") ; categ = list(L1 = "group1", L2 = "group2") ; categ = NULL ; categ.class.order = NULL ; color = NULL ; geom = "geom_point" ; geom.step.dir = "hv" ; geom.stick.base = NULL ; alpha = 0.5 ; dot.size = 2 ; dot.shape = 21 ; dot.border.size = 0.5 ; dot.border.color = NULL ; line.size = 0.5 ; line.type = "solid" ; x.lim = NULL ; x.lab = NULL ; x.log = "no" ; x.tick.nb = NULL ; x.second.tick.nb = NULL ; x.include.zero = FALSE ; x.left.extra.margin = 0.05 ; x.right.extra.margin = 0.05 ; x.text.angle = 0 ; y.lim = NULL ; y.lab = NULL ; y.log = "no" ; y.tick.nb = NULL ; y.second.tick.nb = NULL ; y.include.zero = FALSE ; y.top.extra.margin = 0.05 ; y.bottom.extra.margin = 0.05 ; y.text.angle = 0 ; raster = FALSE ; raster.ratio = 1 ; raster.threshold = NULL ; text.size = 12 ; title = "" ; title.text.size = 12 ; legend.show = TRUE ; legend.width = 0.5 ; legend.name = NULL ; article = TRUE ; grid = FALSE ; add = NULL ; return = FALSE ; return.ggplot = FALSE ; return.gtable = TRUE ; plot = TRUE ; warn.print = FALSE ; lib.path = NULL
+ # function name
+ function.name <- paste0(as.list(match.call(expand.dots=FALSE))[[1]], "()")
+ arg.names <- names(formals(fun = sys.function(sys.parent(n = 2)))) # names of all the arguments
+ arg.user.setting <- as.list(match.call(expand.dots=FALSE))[-1] # list of the argument settings (excluding default values not provided by the user)
+ # end function name
+ # required function checking
+ req.function <- c(
+ "fun_check",
+ "fun_gg_just",
+ "fun_gg_empty_graph",
+ "fun_gg_palette",
+ "fun_gg_point_rast",
+ "fun_round",
+ "fun_pack",
+ "fun_scale",
+ "fun_inter_ticks"
+ )
+ tempo <- NULL
+ for(i1 in req.function){
+ if(length(find(i1, mode = "function"))== 0L){
+ tempo <- c(tempo, i1)
+ }
+ }
+ if( ! is.null(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nREQUIRED cute FUNCTION", ifelse(length(tempo) > 1, "S ARE", " IS"), " MISSING IN THE R ENVIRONMENT:\n", paste0(tempo, collapse = "()\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end required function checking
+ # reserved words to avoid bugs (used in this function)
+ reserved.words <- c("fake_x", "fake_y", "fake_categ")
+ # end reserved words to avoid bugs (used in this function)
+ # arg with no default values
+ mandat.args <- c(
+ "data1",
+ "x",
+ "y"
+ )
+ tempo <- eval(parse(text = paste0("missing(", paste0(mandat.args, collapse = ") | missing("), ")")))
+ if(any(tempo)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nFOLLOWING ARGUMENT", ifelse(length(mandat.args) > 1, "S HAVE", "HAS"), " NO DEFAULT VALUE AND REQUIRE ONE:\n", paste0(mandat.args, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end arg with no default values
+ # argument primary checking
+ arg.check <- NULL #
+ text.check <- NULL #
+ checked.arg.names <- NULL # for function debbuging: used by r_debugging_tools
+ ee <- expression(arg.check <- c(arg.check, tempo$problem) , text.check <- c(text.check, tempo$text) , checked.arg.names <- c(checked.arg.names, tempo$object.name))
+ tempo1 <- fun_check(data = data1, class = "data.frame", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = data1, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OR A LIST OF DATA FRAMES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(x)){
+ tempo1 <- fun_check(data = x, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = x, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(y)){
+ tempo1 <- fun_check(data = y, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = y, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(categ)){
+ tempo1 <- fun_check(data = categ, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = categ, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = categ, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(categ.class.order)){
+ if(is.null(categ)){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT IS NOT NULL, BUT categ IS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = categ.class.order, class = "vector", mode = "character", fun.name = function.name)
+ tempo2 <- fun_check(data = categ.class.order, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT MUST BE A VECTOR OF CHARACTER STRINGS OR A LIST OF VECTOR OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = categ.class.order, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(legend.name)){
+ tempo1 <- fun_check(data = legend.name, class = "vector", mode = "character", na.contain = TRUE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = legend.name, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = legend.name, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(color)){
+ tempo1 <- fun_check(data = color, class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = color, class = "factor", na.contain = TRUE, fun.name = function.name)
+ tempo3 <- fun_check(data = color, class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)
+ tempo4 <- fun_check(data = color, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo4$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE & tempo3$problem == TRUE & tempo4$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT MUST BE A VECTOR (OF CHARACTER STRINGS OR INTEGERS) OR A FACTOR OR A LIST OF THESE POSSIBILITIES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = color, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo1 <- fun_check(data = geom, class = "vector", mode = "character", na.contain = FALSE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = geom, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT MUST BE A SINGLE CHARACTER STRING OR A LIST OF CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = geom.step.dir, options = c("vh", "hv", "mid"), na.contain = FALSE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = geom.step.dir, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT MUST BE A SINGLE CHARACTER STRING (\"vh\" OR \"hv\" OR \"mid\") OR A LIST OF THESE CHARACTER STRINGS")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(geom.stick.base)){
+ tempo1 <- fun_check(data = geom.stick.base, class = "vector", mode = "numeric", na.contain = FALSE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = color, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = geom.stick.base, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo1 <- fun_check(data = alpha, prop = TRUE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = alpha, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT MUST BE A SINGLE NUMERIC VALUE BETWEEN 0 AND 1 OR A LIST OF SUCH VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = dot.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.size, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = dot.shape, class = "vector", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.shape, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT MUST BE A SINGLE SHAPE VALUE OR A LIST OF SINGLE SHAPE VALUES (SEE https://ggplot2.tidyverse.org/articles/ggplot2-specs.html)")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = dot.border.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.border.size, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(dot.border.color)){
+ tempo1 <- fun_check(data = dot.border.color, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = dot.border.color, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ # integer colors -> gg_palette
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color MUST BE A SINGLE CHARACTER STRING OF COLOR OR A SINGLE INTEGER VALUE")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = dot.border.color, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo1 <- fun_check(data = line.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name)
+ tempo2 <- fun_check(data = line.size, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo2$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT MUST BE A SINGLE NUMERIC VALUE OR A LIST OF SINGLE NUMERIC VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ tempo1 <- fun_check(data = line.type, class = "vector", typeof = "integer", double.as.integer.allowed = FALSE, length = 1, fun.name = function.name)
+ tempo2 <- fun_check(data = line.type, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ tempo3 <- fun_check(data = line.type, class = "list", na.contain = TRUE, fun.name = function.name)
+ checked.arg.names <- c(checked.arg.names, tempo3$object.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE & tempo3$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT MUST BE A SINGLE LINE KIND VALUE OR A LIST OF SINGLE LINE KIND VALUES (SEE https://ggplot2.tidyverse.org/articles/ggplot2-specs.html)")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ if( ! is.null(x.lim)){
+ tempo <- fun_check(data = x.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & any(x.lim %in% c(Inf, -Inf))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x.lim, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(x.lab)){
+ if(all(class(x.lab) %in% "expression")){ # to deal with math symbols
+ tempo <- fun_check(data = x.lab, class = "expression", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ tempo <- fun_check(data = x.lab, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x.lab, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = x.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(x.tick.nb)){
+ tempo <- fun_check(data = x.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & x.tick.nb < 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(x.second.tick.nb)){
+ tempo <- fun_check(data = x.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & x.second.tick.nb <= 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x.second.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = x.second.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = x.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.left.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.right.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = x.text.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(y.lim)){
+ tempo <- fun_check(data = y.lim, class = "vector", mode = "numeric", length = 2, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & any(y.lim %in% c(Inf, -Inf))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.lim ARGUMENT CANNOT CONTAIN -Inf OR Inf VALUES")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.lim, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(y.lab)){
+ if(all(class(y.lab) %in% "expression")){ # to deal with math symbols
+ tempo <- fun_check(data = y.lab, class = "expression", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ tempo <- fun_check(data = y.lab, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.lab, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = y.log, options = c("no", "log2", "log10"), length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(y.tick.nb)){
+ tempo <- fun_check(data = y.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & y.tick.nb < 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if( ! is.null(y.second.tick.nb)){
+ tempo <- fun_check(data = y.second.tick.nb, class = "vector", typeof = "integer", length = 1, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE & y.second.tick.nb <= 0){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y.second.tick.nb ARGUMENT MUST BE A NON-NULL POSITIVE INTEGER")
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = y.second.tick.nb, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = y.include.zero, class = "vector", mode = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.top.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.bottom.extra.margin, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = y.text.angle, class = "vector", typeof = "integer", double.as.integer.allowed = TRUE, length = 1, neg.values = TRUE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = raster, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = raster.ratio, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ if( ! is.null(raster.threshold)){
+ tempo <- fun_check(data = raster.threshold, class = "vector", typeof = "integer", neg.values = FALSE, double.as.integer.allowed = TRUE, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = raster.threshold, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = title.text.size, class = "vector", mode = "numeric", length = 1, neg.values = FALSE, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = legend.show, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(legend.width)){
+ tempo <- fun_check(data = legend.width, prop = TRUE, length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = legend.width, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = article, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = grid, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(add)){
+ tempo <- fun_check(data = add, class = "vector", mode = "character", length = 1, fun.name = function.name) ; eval(ee)
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = add, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ tempo <- fun_check(data = return, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.ggplot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = return.gtable, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = plot, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ tempo <- fun_check(data = warn.print, class = "logical", length = 1, fun.name = function.name) ; eval(ee)
+ if( ! is.null(lib.path)){
+ tempo <- fun_check(data = lib.path, class = "vector", mode = "character", fun.name = function.name) ; eval(ee)
+ if(tempo$problem == FALSE){
+ if( ! all(dir.exists(lib.path))){ # separation to avoid the problem of tempo$problem == FALSE and lib.path == NA
+ tempo.cat <- paste0("ERROR IN ", function.name, ": DIRECTORY PATH INDICATED IN THE lib.path ARGUMENT DOES NOT EXISTS:\n", paste(lib.path, collapse = "\n"))
+ text.check <- c(text.check, tempo.cat)
+ arg.check <- c(arg.check, TRUE)
+ }
+ }
+ }else{
+ # no fun_check test here, it is just for checked.arg.names
+ tempo <- fun_check(data = lib.path, class = "vector")
+ checked.arg.names <- c(checked.arg.names, tempo$object.name)
+ }
+ if(any(arg.check) == TRUE){
+ stop(paste0("\n\n================\n\n", paste(text.check[arg.check], collapse = "\n"), "\n\n================\n\n"), call. = FALSE) #
+ }
+ # source("C:/Users/Gael/Documents/Git_versions_to_use/debugging_tools_for_r_dev-v1.7/r_debugging_tools-v1.7.R") ; eval(parse(text = str_basic_arg_check_dev)) ; eval(parse(text = str_arg_check_with_fun_check_dev)) # activate this line and use the function (with no arguments left as NULL) to check arguments status and if they have been checked using fun_check()
+ # end argument primary checking
+
+
+ # second round of checking and data preparation
+ # management of NA arguments
+ tempo.arg <- names(arg.user.setting) # values provided by the user
+ tempo.log <- suppressWarnings(sapply(lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.na), FUN = any)) & lapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = length)== 1L # no argument provided by the user can be just NA
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT JUST BE NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NA arguments
+ # management of NULL arguments
+ tempo.arg <-c(
+ "data1",
+ # "x", # inactivated because of hline or vline
+ # "y", # inactivated because of hline or vline
+ "geom",
+ "geom.step.dir",
+ # "geom.stick.base", # inactivated because can be null
+ "alpha",
+ "dot.size",
+ "dot.shape",
+ "dot.border.size",
+ "line.size",
+ "line.type",
+ "x.log",
+ "x.include.zero",
+ "x.left.extra.margin",
+ "x.right.extra.margin",
+ "x.text.angle",
+ "y.log",
+ "y.include.zero",
+ "y.top.extra.margin",
+ "y.bottom.extra.margin",
+ "y.text.angle",
+ "raster",
+ "raster.ratio",
+ "text.size",
+ "title",
+ "title.text.size",
+ "legend.show",
+ # "legend.width", # inactivated because can be null
+ "article",
+ "grid",
+ "return",
+ "return.ggplot",
+ "return.gtable",
+ "plot",
+ "warn.print"
+ )
+ tempo.log <- sapply(lapply(tempo.arg, FUN = get, env = sys.nframe(), inherit = FALSE), FUN = is.null)
+ if(any(tempo.log) == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ":\n", ifelse(sum(tempo.log, na.rm = TRUE) > 1, "THESE ARGUMENTS\n", "THIS ARGUMENT\n"), paste0(tempo.arg[tempo.log], collapse = "\n"),"\nCANNOT BE NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n"), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }
+ # end management of NULL arguments
+ # code that protects set.seed() in the global environment
+ # end code that protects set.seed() in the global environment
+ # warning initiation
+ ini.warning.length <- options()$warning.length
+ options(warning.length = 8170)
+ warn <- NULL
+ warn.count <- 0
+ # end warning initiation
+ # other checkings
+ # check list lengths (and names of data1 compartments if present)
+ list.color <- NULL
+ list.geom <- NULL
+ list.geom.step.dir <- NULL
+ list.geom.stick.base <- NULL
+ list.alpha <- NULL
+ list.dot.size <- NULL
+ list.dot.shape <- NULL
+ list.dot.border.size <- NULL
+ list.dot.border.color <- NULL
+ list.line.size <- NULL
+ list.line.type <- NULL
+ if(all(class(data1) == "list")){
+ if(length(data1) > 6){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A LIST OF 6 DATA FRAMES MAXIMUM (6 OVERLAYS MAX)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(is.null(names(data1))){
+ names(data1) <- paste0("L", 1:length(data1))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL NAME COMPARTMENT OF data1 LIST -> NAMES RESPECTIVELY ATTRIBUTED TO EACH COMPARTMENT:\n", paste(names(data1), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ if( ! is.null(x)){
+ if( ! (all(class(x) == "list") & length(data1) == length(x))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }else{
+ x <- vector("list", length(data1))
+ }
+ if( ! is.null(y)){
+ if( ! (all(class(y) == "list") & length(data1) == length(y))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }else{
+ y <- vector("list", length(data1))
+ }
+ if( ! is.null(categ)){
+ if( ! (all(class(categ) == "list") & length(data1) == length(categ))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if( ! is.null(categ.class.order)){
+ if( ! (all(class(categ.class.order) == "list") & length(data1) == length(categ.class.order))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if( ! is.null(color)){
+ if( ! ((all(class(color) == "list") & length(data1) == length(color)) | ((all(mode(color) == "character") | all(mode(color) == "numeric")) & length(color)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER STRING OR INTEGER")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if((all(mode(color) == "character") | all(mode(color) == "numeric")) & length(color)== 1L){ # convert the single value into a list of single value
+ list.color <- vector(mode = "list", length = length(data1))
+ list.color[] <- color
+ }
+ }
+ if( ! ((all(class(geom) == "list") & length(data1) == length(geom)) | (all(mode(geom) == "character") & length(geom)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(geom) == "character") & length(geom)== 1L){ # convert the single value into a list of single value
+ list.geom <- vector(mode = "list", length = length(data1))
+ list.geom[] <- geom
+ }
+ if( ! ((all(class(geom.step.dir) == "list") & length(data1) == length(geom.step.dir)) | (all(mode(geom.step.dir) == "character") & length(geom.step.dir)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(geom.step.dir) == "character") & length(geom.step.dir)== 1L){ # convert the single value into a list of single value
+ list.geom.step.dir <- vector(mode = "list", length = length(data1))
+ list.geom.step.dir[] <- geom.step.dir
+ }
+ if( ! is.null(geom.stick.base)){
+ if( ! ((all(class(geom.stick.base) == "list") & length(data1) == length(geom.stick.base)) | (all(mode(geom.stick.base) == "numeric") & length(geom.stick.base)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(geom.stick.base) == "numeric") & length(geom.stick.base)== 1L){ # convert the single value into a list of single value
+ list.geom.stick.base <- vector(mode = "list", length = length(data1))
+ list.geom.stick.base[] <- geom.stick.base
+ }
+ }
+ if( ! ((all(class(alpha) == "list") & length(data1) == length(alpha)) | (all(mode(alpha) == "numeric") & length(alpha)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(alpha) == "numeric") & length(alpha)== 1L){ # convert the single value into a list of single value
+ list.alpha <- vector(mode = "list", length = length(data1))
+ list.alpha[] <- alpha
+ }
+ if( ! ((all(class(dot.size) == "list") & length(data1) == length(dot.size)) | (all(mode(dot.size) == "numeric") & length(dot.size)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(dot.size) == "numeric") & length(dot.size)== 1L){ # convert the single value into a list of single value
+ list.dot.size <- vector(mode = "list", length = length(data1))
+ list.dot.size[] <- dot.size
+ }
+ if( ! ((all(class(dot.shape) == "list") & length(data1) == length(dot.shape)) | (all(mode(dot.shape) != "list") & length(dot.shape)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE SHAPE VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(dot.shape) != "list") & length(dot.shape)== 1L){ # convert the single value into a list of single value
+ list.dot.shape <- vector(mode = "list", length = length(data1))
+ list.dot.shape[] <- dot.shape
+ }
+ if( ! ((all(class(dot.border.size) == "list") & length(data1) == length(dot.border.size)) | (all(mode(dot.border.size) == "numeric") & length(dot.border.size)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(dot.border.size) == "numeric") & length(dot.border.size)== 1L){ # convert the single value into a list of single value
+ list.dot.border.size <- vector(mode = "list", length = length(data1))
+ list.dot.border.size[] <- dot.border.size
+ }
+ if( ! is.null(dot.border.color)){
+ if( ! ((all(class(dot.border.color) == "list") & length(data1) == length(dot.border.color)) | ((all(mode(dot.border.color) == "character") | all(mode(dot.border.color) == "numeric")) & length(dot.border.color)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE CHARACTER STRING OR INTEGER")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if((all(mode(dot.border.color) == "character") | all(mode(dot.border.color) == "numeric")) & length(dot.border.color)== 1L){ # convert the single value into a list of single value
+ list.dot.border.color <- vector(mode = "list", length = length(data1))
+ list.dot.border.color[] <- dot.border.color
+ }
+ }
+ if( ! ((all(class(line.size) == "list") & length(data1) == length(line.size)) | (all(mode(line.size) == "numeric") & length(line.size)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE NUMERIC VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(line.size) == "numeric") & length(line.size)== 1L){ # convert the single value into a list of single value
+ list.line.size <- vector(mode = "list", length = length(data1))
+ list.line.size[] <- line.size
+ }
+ if( ! ((all(class(line.type) == "list") & length(data1) == length(line.type)) | (all(mode(line.type) != "list") & length(line.type)== 1L))){ # list of same length as data1 or single value
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST, OR A SINGLE LINE KIND VALUE")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(all(mode(line.type) != "list") & length(line.type)== 1L){ # convert the single value into a list of single value
+ list.line.type <- vector(mode = "list", length = length(data1))
+ list.line.type[] <- line.type
+ }
+ if( ! is.null(legend.name)){
+ if( ! (all(class(legend.name) == "list") & length(data1) == length(legend.name))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT MUST BE A LIST OF SAME LENGTH AS data1 IF data1 IS A LIST")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }
+ # end check list lengths (and names of data1 compartments if present)
+ # conversion into lists
+ if(all(is.data.frame(data1))){
+ data1 <- list(L1 = data1)
+ if(all(class(x) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": x ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ x <- list(L1 = x)
+ }
+ if(all(class(y) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": y ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ y <- list(L1 = y)
+ }
+ if( ! is.null(categ)){
+ if(all(class(categ) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ categ <- list(L1 = categ)
+ }
+ }
+ if( ! is.null(categ.class.order)){
+ if(all(class(categ.class.order) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": categ.class.order ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ categ.class.order <- list(L1 = categ.class.order)
+ }
+ }
+ if( ! is.null(color)){
+ if(all(class(color) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": color ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ color <- list(L1 = color)
+ }
+ }
+ if(all(class(geom) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ geom <- list(L1 = geom)
+ }
+ if(all(class(geom.step.dir) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.step.dir ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ geom.step.dir <- list(L1 = geom.step.dir)
+ }
+ if( ! is.null(geom.stick.base)){
+ if(all(class(geom.stick.base) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom.stick.base ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ geom.stick.base <- list(L1 = geom.stick.base)
+ }
+ }
+ if(all(class(alpha) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": alpha ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ alpha <- list(L1 = alpha)
+ }
+ if(all(class(dot.size) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ dot.size <- list(L1 = dot.size)
+ }
+ if(all(class(dot.shape) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.shape ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ dot.shape <- list(L1 = dot.shape)
+ }
+ if(all(class(dot.border.size) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ dot.border.size <- list(L1 = dot.border.size)
+ }
+ if( ! is.null(dot.border.color)){
+ if(all(class(dot.border.color) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": dot.border.color ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ dot.border.color <- list(L1 = dot.border.color)
+ }
+ }
+ if(all(class(line.size) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.size ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ line.size <- list(L1 = line.size)
+ }
+ if(all(class(line.type) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": line.type ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ line.type <- list(L1 = line.type)
+ }
+ if( ! is.null(legend.name)){
+ if(all(class(legend.name) == "list")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": legend.name ARGUMENT CANNOT BE A LIST IF data1 IS A DATA FRAME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ legend.name <- list(L1 = legend.name)
+ }
+ }
+ }else if( ! all(sapply(data1, FUN = "class") == "data.frame")){ # if not a data frame, data1 can only be a list, as tested above
+ tempo.cat <- paste0("ERROR IN ", function.name, ": data1 ARGUMENT MUST BE A DATA FRAME OR A LIST OF DATA FRAMES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ # single value converted into list now reattributed to the argument name
+ if( ! is.null(color)){
+ if( ! is.null(list.color)){
+ color <- list.color
+ }
+ }
+ if( ! is.null(list.geom)){
+ geom <- list.geom
+ }
+ if( ! is.null(list.geom.step.dir)){
+ geom.step.dir <- list.geom.step.dir
+ }
+ if( ! is.null(geom.stick.base)){
+ if( ! is.null(list.geom.stick.base)){
+ geom.stick.base <- list.geom.stick.base
+ }
+ }
+ if( ! is.null(list.alpha)){
+ alpha <- list.alpha
+ }
+ if( ! is.null(list.dot.size)){
+ dot.size <- list.dot.size
+ }
+ if( ! is.null(list.dot.shape)){
+ dot.shape <- list.dot.shape
+ }
+ if( ! is.null(list.dot.border.size)){
+ dot.border.size <- list.dot.border.size
+ }
+ if( ! is.null(dot.border.color)){
+ if( ! is.null(list.dot.border.color)){
+ dot.border.color <- list.dot.border.color
+ }
+ }
+ if( ! is.null(list.line.size)){
+ line.size <- list.line.size
+ }
+ if( ! is.null(list.line.type)){
+ line.type <- list.line.type
+ }
+ # end single value converted into list now reattributed to the argument name
+ # data, x, y, geom, alpha, dot.size, shape, dot.border.size, line.size, line.type, legend.name are list now
+ # if non-null, categ, categ.class.order, legend.name, color, dot.border.color are list now
+ # end conversion into lists
+ # verif of add
+ if( ! is.null(add)){
+ if( ! grepl(pattern = "^\\s*\\+", add)){ # check that the add string start by +
+ tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST START WITH \"+\": ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+
+ }else if( ! grepl(pattern = "(ggplot2|lemon)\\s*::", add)){ #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": FOR EASIER FUNCTION DETECTION, add ARGUMENT MUST CONTAIN \"ggplot2::\" OR \"lemon::\" IN FRONT OF EACH GGPLOT2 FUNCTION: ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if( ! grepl(pattern = ")\\s*$", add)){ # check that the add string finished by )
+ tempo.cat <- paste0("ERROR IN ", function.name, ": add ARGUMENT MUST FINISH BY \")\": ", paste(unique(add), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ # end verif of add
+ # management of add containing facet
+ facet.categ <- NULL
+ if( ! is.null(add)){
+ facet.check <- TRUE
+ tempo <- unlist(strsplit(x = add, split = "\\s*\\+\\s*(ggplot2|lemon)\\s*::\\s*")) #
+ tempo <- sub(x = tempo, pattern = "^facet_wrap", replacement = "ggplot2::facet_wrap")
+ tempo <- sub(x = tempo, pattern = "^facet_grid", replacement = "ggplot2::facet_grid")
+ tempo <- sub(x = tempo, pattern = "^facet_rep", replacement = "lemon::facet_rep")
+ if(length(data1) > 1 & (any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")) | grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nfacet PANELS CANNOT BE USED IF MORE THAN ONE DATA FRAME IN THE data1 ARGUMENT\nPLEASE REWRITE THE add STRING AND RERUN")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ if(any(grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap"))){
+ tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_wrap|lemon::facet_rep_wrap")])))
+ facet.categ <- list(names(tempo1$params$facets)) # list of length 1
+ tempo.text <- "facet_wrap OR facet_rep_wrap"
+ facet.check <- FALSE
+ }else if(grepl(x = add, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")){
+ tempo1 <- suppressWarnings(eval(parse(text = tempo[grepl(x = tempo, pattern = "ggplot2::facet_grid|lemon::facet_rep_grid")])))
+ facet.categ <- list(c(names(tempo1$params$rows), names(tempo1$params$cols))) # list of length 1
+ tempo.text <- "facet_grid OR facet_rep_grid"
+ facet.check <- FALSE
+ }
+ if(facet.check == FALSE & ! all(facet.categ %in% names(data1[[1]]))){ # WARNING: all(facet.categ %in% names(data1)) is TRUE when facet.categ is NULL
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF \"", tempo.text, "\" STRING IN THE add ARGUMENT BUT PROBLEM OF VARIABLE DETECTION (COLUMN NAMES OF data1)\nTHE DETECTED VARIABLES ARE:\n", paste(facet.categ, collapse = " "), "\nTHE data1 COLUMN NAMES ARE:\n", paste(names(data1[[1]]), collapse = " "), "\nPLEASE REWRITE THE add STRING AND RERUN")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }
+ # if facet.categ is not NULL, it is a list of length 1 now
+ # end management of add containing facet
+ # legend name filling
+ if(is.null(legend.name) & ! is.null(categ)){
+ legend.name <- categ
+ }else if(is.null(legend.name) & is.null(categ)){
+ legend.name <- vector("list", length(data1)) # null list
+ }
+ # legend.name not NULL anymore (list)
+ # end legend name filling
+ # ini categ for legend display
+ fin.lg.disp <- vector("list", 6) # will be used at the end to display or not legends
+ fin.lg.disp[] <- FALSE
+ legend.disp <- vector("list", length(data1))
+ if(is.null(categ) | legend.show == FALSE){
+ legend.disp[] <- FALSE
+ }else{
+ for(i2 in 1:length(data1)){
+ if(is.null(categ[[i2]])){
+ legend.disp[[i2]] <- FALSE
+ }else{
+ legend.disp[[i2]] <- TRUE
+ }
+ }
+ }
+ # end ini categ for legend display
+ # integer colors into gg_palette
+ tempo.check.color <- NULL
+ for(i1 in 1:length(data1)){
+ if(any(is.na(color[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), ": color ARGUMENT CANNOT CONTAIN NA")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo.check.color <- c(tempo.check.color, fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "integer", double.as.integer.allowed = TRUE, na.contain = TRUE, fun.name = function.name)$problem)
+ }
+ tempo.check.color <- ! tempo.check.color # invert TRUE and FALSE because if integer, then problem = FALSE
+ if(any(tempo.check.color == TRUE)){ # convert integers into colors
+ tempo.integer <- unlist(color[tempo.check.color])
+ tempo.color <- fun_gg_palette(max(tempo.integer, na.rm = TRUE))
+ for(i1 in 1:length(data1)){
+ if(tempo.check.color[i1] == TRUE){
+ color[[i1]] <-tempo.color[color[[i1]]]
+ }
+ }
+ }
+ # end integer colors into gg_palette
+ # loop (checking inside list compartment)
+ compart.null.color <- 0 # will be used to attribute a color when color is non-null but a compartment of color is NULL
+ data1.ini <- data1 # to report NA removal
+ removed.row.nb <- vector("list", length = length(data1)) # to report NA removal. Contains NULL
+ removed.rows <- vector("list", length = length(data1)) # to report NA removal. Contains NULL
+ for(i1 in 1:length(data1)){
+ tempo <- fun_check(data = data1[[i1]], data.name = ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "data.frame", na.contain = TRUE, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ # reserved word checking
+ if(any(names(data1[[i1]]) %in% reserved.words)){ # I do not use fun_name_change() because cannot control y before creating "fake_y". But ok because reserved are not that common
+ tempo.cat <- paste0("ERROR IN ", function.name, ": COLUMN NAMES OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " ARGUMENT CANNOT BE ONE OF THESE WORDS\n", paste(reserved.words, collapse = " "), "\nTHESE ARE RESERVED FOR THE ", function.name, " FUNCTION")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if( ! (is.null(add))){
+ if(any(sapply(X = reserved.words, FUN = grepl, x = add))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nDETECTION OF COLUMN NAMES OF data1 IN THE add ARGUMENT STRING, THAT CORRESPOND TO RESERVED STRINGS FOR ", function.name, "\nFOLLOWING COLUMN NAMES HAVE TO BE CHANGED:\n", paste(arg.names[sapply(X = reserved.words, FUN = grepl, x = add)], collapse = "\n"), "\nFOR INFORMATION, THE RESERVED WORDS ARE:\n", paste(reserved.words, collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(any(sapply(X = arg.names, FUN = grepl, x = add))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NAMES OF ", function.name, " ARGUMENTS DETECTED IN THE add STRING:\n", paste(arg.names[sapply(X = arg.names, FUN = grepl, x = add)], collapse = "\n"), "\nRISK OF WRONG OBJECT USAGE INSIDE ", function.name)
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end reserved word checking
+ # check of geom now because required for y argument
+ tempo <- fun_check(data = geom[[i1]], data.name = ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), options = c("geom_point", "geom_line", "geom_path", "geom_step", "geom_hline", "geom_vline", "geom_stick"), length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(geom[[i1]] == "geom_step" & is.null(geom.step.dir[[i1]])){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(geom.step.dir)== 1L, "geom.step.dir", paste0("ELEMENT ", i1, " OF geom.step.dir ARGUMENT")), ": geom.step.dir ARGUMENT CANNOT BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("ELEMENT ", i1, " OF geom")), " ARGUMENT IS \"geom_step\"\nHERE geom.step.dir ARGUMENT IS: ", paste(geom.step.dir[[i1]], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(geom[[i1]] == "geom_step" & ! is.null(geom.step.dir[[i1]])){
+ tempo <- fun_check(data = geom.step.dir[[i1]], data.name = ifelse(length(geom.step.dir)== 1L, "geom.step.dir", paste0("geom.step.dir NUMBER ", i1)), options = c("vh", "hv", "mid"), length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if( ! (is.null(geom.stick.base))){
+ if(geom[[i1]] == "geom_stick" & ! is.null(geom.stick.base[[i1]])){
+ tempo <- fun_check(data = geom.stick.base[[i1]], data.name = ifelse(length(geom.stick.base)== 1L, "geom.stick.base", paste0("geom.stick.base NUMBER ", i1)), mode = "numeric", length = 1, na.contain = FALSE, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }
+ # end check of geom now because required for y argument
+ if(is.null(x[[i1]])){
+ if(all(geom[[i1]] != "geom_hline")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": x ARGUMENT CANNOT BE NULL EXCEPT IF ", ifelse(length(geom)== 1L, "x", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_hline\"\nHERE geom ARGUMENT IS: ", paste(geom[[i1]], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ x[[i1]] <- "fake_x"
+ data1[[i1]] <- cbind(data1[[i1]], fake_x = NA, stringsAsFactors = TRUE)
+ data1[[i1]][, "fake_x"] <- as.numeric(data1[[i1]][, "fake_x"])
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT ASSOCIATED TO ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT ", geom[[i1]], " -> FAKE COLUMN ADDED TO DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", NAMED \"fake_x\" FOR FINAL DRAWING")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(all(geom[[i1]] == "geom_hline")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": x ARGUMENT MUST BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_hline\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo <- fun_check(data = x[[i1]], data.name = ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), class = "vector", mode = "character", length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(is.null(y[[i1]])){
+ if(all(geom[[i1]] != "geom_vline")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": y ARGUMENT CANNOT BE NULL EXCEPT IF ", ifelse(length(geom)== 1L, "y", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_vline\"\nHERE geom ARGUMENT IS: ", paste(geom[[i1]], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ y[[i1]] <- "fake_y"
+ data1[[i1]] <- cbind(data1[[i1]], fake_y = NA, stringsAsFactors = TRUE)
+ data1[[i1]][, "fake_y"] <- as.numeric(data1[[i1]][, "fake_y"])
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT ASSOCIATED TO ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT ", geom[[i1]], " -> FAKE COLUMN ADDED TO DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", NAMED \"fake_y\" FOR FINAL DRAWING")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else{
+ if(all(geom[[i1]] == "geom_vline")){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ": y ARGUMENT MUST BE NULL IF ", ifelse(length(geom)== 1L, "geom", paste0("geom NUMBER ", i1)), " ARGUMENT IS \"geom_vline\"")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo <- fun_check(data = y[[i1]], data.name = ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), class = "vector", mode = "character", length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ # x[[i1]] and y[[i1]] not NULL anymore
+ if( ! (x[[i1]] %in% names(data1[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(x[[i1]], collapse = " "))))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if( ! (y[[i1]] %in% names(data1[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(y[[i1]], collapse = " "))))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo <- fun_check(data = data1[[i1]][, x[[i1]]], data.name = ifelse(length(x)== 1L, "x ARGUMENT (AS COLUMN NAME OF data1 DATA FRAME)", paste0("ELEMENT ", i1, " OF x ARGUMENT", " (AS COLUMN NAME OF data1 DATA FRAME NUMBER ", i1, ")")), class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo <- fun_check(data = data1[[i1]][, y[[i1]]], data.name = ifelse(length(y)== 1L, "y ARGUMENT (AS COLUMN NAME OF data1 DATA FRAME)", paste0("ELEMENT ", i1, " OF y ARGUMENT", " (AS COLUMN NAME OF data1 DATA FRAME NUMBER ", i1, ")")), class = "vector", mode = "numeric", na.contain = TRUE, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(x[[i1]] == "fake_x" & y[[i1]] == "fake_y"){ # because the code cannot accept to be both "fake_x" and "fake_y" at the same time
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 2\nTHE CODE CANNOT ACCEPT x AND y TO BE \"fake_x\" AND \"fake_y\" IN THE SAME DATA FRAME ", i1, " ")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+
+ if(( ! is.null(categ)) & ( ! is.null(categ[[i1]]))){ # is.null(categ[[i1]]) works even if categ is NULL # is.null(categ[[i1]]) works even if categ is NULL # if categ[[i1]] = NULL, fake_categ will be created later on
+ tempo <- fun_check(data = categ[[i1]], data.name = ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")),, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if( ! (categ[[i1]] %in% names(data1[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT MUST BE A COLUMN NAME OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT\nHERE IT IS: ", paste(categ[[i1]], collapse = " "))))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ tempo1 <- fun_check(data = data1[[i1]][, categ[[i1]]], data.name = ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name)
+ tempo2 <- fun_check(data = data1[[i1]][, categ[[i1]]], data.name = ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), class = "factor", na.contain = TRUE, fun.name = function.name)
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " MUST BE A FACTOR OR CHARACTER VECTOR")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(tempo1$problem == FALSE){
+ data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE CHARACTER COLUMN HAS BEEN CONVERTED TO FACTOR")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+
+ }
+ if(geom[[i1]] == "geom_vline" | geom[[i1]] == "geom_hline"){
+ if(length(unique(data1[[i1]][, categ[[i1]]])) != nrow(data1[[i1]])){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(geom)== 1L, "geom OF data1 ARGUMENT", paste0("geom NUMBER ", i1, " OF DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " ARGUMENT IS ", geom[[i1]], ", MEANING THAT ", ifelse(length(categ)== 1L, "categ OF data1 ARGUMENT", paste0("ELEMENT ", i1, " OF categ ARGUMENT IN DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " MUST HAVE A DIFFERENT CLASS PER LINE OF data1 (ONE x VALUE PER CLASS)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }else if(( ! is.null(categ)) & is.null(categ[[i1]])){ # is.null(categ[[i1]]) works even if categ is NULL # if categ[[i1]] = NULL, fake_categ will be created. WARNING: is.null(categ[[i1]]) means no legend display (see above), because categ has not been precised. This also means a single color for data1[[i1]]
+ if(length(color[[i1]]) > 1){ # 0 means is.null(color[[i1]]) or is.null(color) and 1 is ok -> single color for data1[[i1]]
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT BUT CORRESPONDING COLORS IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " HAS LENGTH OVER 1\n", paste(color[[i1]], collapse = " "), "\nWHICH IS NOT COMPATIBLE WITH NULL CATEG -> COLOR RESET TO A SINGLE COLOR")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ color[i1] <- list(NULL) # will provide a single color below # Warning color[[i1]] <- NULL removes the compartment
+ }
+ categ[[i1]] <- "fake_categ"
+ data1[[i1]] <- cbind(data1[[i1]], fake_categ = "", stringsAsFactors = TRUE)
+ # inactivated because give a different color to different "Line_" categ while a single color for all the data1[[i1]] required. Thus, put back after the color management
+ # if(geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline"){
+ # data1[[i1]][, "fake_categ"] <- paste0("Line_", 1:nrow(data1[[i1]]))
+ # }else{
+ data1[[i1]][, "fake_categ"] <- data1[[i1]][, "fake_categ"] # as.numeric("") create a vector of NA but class numeric
+ # }
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ")), " ARGUMENT -> FOR DATA FRAME ", ifelse(length(data1)== 1L, "data1 ARGUMENT:", paste0("NUMBER ", i1, " OF data1 ARGUMENT:")), "\n- FAKE \"fake_categ\" COLUMN ADDED FILLED WITH \"\"(OR WITH \"Line_...\" FOR LINES)\n- SINGLE COLOR USED FOR PLOTTING\n- NO LEGEND DISPLAYED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # OK: if categ is not NULL, all the non-null categ columns of data1 are factors from here
+
+ # management of log scale and Inf removal
+ if(x[[i1]] != "fake_x"){
+ if(any(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN ", ifelse(length(categ)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(y[[i1]] != "fake_y"){
+ if(any(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[, y])) & ( ! is.na(data1[, y])) detects only Inf
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PRESENCE OF -Inf OR Inf VALUES IN ", ifelse(length(categ)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # log conversion
+ if(x.log != "no"){
+ tempo1 <- ! is.finite(data1[[i1]][, x[[i1]]]) # where are initial NA and Inf
+ data1[[i1]][, x[[i1]]] <- suppressWarnings(get(x.log)(data1[[i1]][, x[[i1]]]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ if(any( ! (tempo1 | is.finite(data1[[i1]][, x[[i1]]])))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN ", ifelse(length(categ)== 1L, "x", paste0("ELEMENT ", i1, " OF x ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(y.log != "no"){
+ tempo1 <- ! is.finite(data1[[i1]][, y[[i1]]]) # where are initial NA and Inf
+ data1[[i1]][, y[[i1]]] <- suppressWarnings(get(y.log)(data1[[i1]][, y[[i1]]]))# no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ if(any( ! (tempo1 | is.finite(data1[[i1]][, y[[i1]]])))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LOG CONVERSION INTRODUCED -Inf OR Inf OR NaN VALUES IN ", ifelse(length(categ)== 1L, "y", paste0("ELEMENT ", i1, " OF y ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # Inf removal
+ # removed.row.nb[[i1]] <- NULL # already NULL and Warning this removes the compartment
+ removed.rows[[i1]] <- data.frame(stringsAsFactors = FALSE)
+ if(any(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]])) detects only Inf
+ removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], which(( ! is.finite(data1[[i1]][, x[[i1]]])) & ( ! is.na(data1[[i1]][, x[[i1]]]))))
+ }
+ if(any(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])))){ # is.finite also detects NA: ( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]])) detects only Inf
+ removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], which(( ! is.finite(data1[[i1]][, y[[i1]]])) & ( ! is.na(data1[[i1]][, y[[i1]]]))))
+ }
+ if( ! is.null(removed.row.nb[[i1]])){
+ removed.row.nb[[i1]] <- unique(removed.row.nb[[i1]]) # to remove the duplicated positions (NA in both x and y)
+ removed.rows[[i1]] <- rbind(removed.rows[[i1]], data1.ini[[i1]][removed.row.nb[[i1]], ]) # here data1.ini used to have the y = O rows that will be removed because of Inf creation after log transformation
+ data1[[i1]] <- data1[[i1]][-removed.row.nb[[i1]], ]
+ data1.ini[[i1]] <- data1.ini[[i1]][-removed.row.nb[[i1]], ] #
+ }
+ # From here, data1 and data.ini have no more Inf
+ # end Inf removal
+ # x.lim and y.lim dealt later on, after the end f the loop
+ # end management of log scale and Inf removal
+ # na detection and removal
+ column.check <- unique(unlist(c( # unlist because creates a list
+ if(x[[i1]] == "fake_x"){NULL}else{x[[i1]]},
+ if(y[[i1]] == "fake_y"){NULL}else{y[[i1]]},
+ if( ! is.null(categ)){if(is.null(categ[[i1]])){NULL}else{categ[[i1]]}},
+ if( ! is.null(facet.categ)){if(is.null(facet.categ[[i1]])){NULL}else{facet.categ[[i1]]}}
+ ))) # dot.categ because can be a 3rd column of data1
+ if(any(is.na(data1[[i1]][, column.check]))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NA DETECTED IN COLUMNS ", paste(column.check, collapse = " "), " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " AND CORRESPONDING ROWS REMOVED (SEE $removed.row.nb AND $removed.rows)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ for(i3 in 1:length(column.check)){
+ if(any(is.na(data1[[i1]][, column.check[i3]]))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("NA REMOVAL DUE TO COLUMN ", column.check[i3], " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ tempo <- unique(unlist(lapply(lapply(c(data1[[i1]][column.check]), FUN = is.na), FUN = which)))
+ removed.row.nb[[i1]] <- c(removed.row.nb[[i1]], tempo)
+ removed.rows[[i1]] <- rbind(removed.rows[[i1]], data1.ini[[i1]][tempo, ]) # # tempo used because removed.row.nb is not empty. Here data1.ini used to have the non NA rows that will be removed because of NAN creation after log transformation (neg values for instance)
+ column.check <- column.check[ ! (column.check == x[[i1]] | column.check == y[[i1]])] # remove x and y to keep quali columns
+ if(length(tempo) != 0){
+ data1[[i1]] <- data1[[i1]][-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers thant the former
+ data1.ini[[i1]] <- data1.ini[[i1]][-tempo, ] # WARNING tempo here and not removed.row.nb because the latter contain more numbers than the former
+ for(i4 in 1:length(column.check)){
+ if(any( ! unique(removed.rows[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]]))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN COLUMN ", column.check[i4], " OF ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE FOLLOWING CLASSES HAVE DISAPPEARED AFTER NA REMOVAL\n(IF COLUMN USED IN THE PLOT, THIS CLASS WILL NOT BE DISPLAYED):\n", paste(unique(removed.rows[[i1]][, column.check[i4]])[ ! unique(removed.rows[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]])], collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ tempo.levels <- levels(data1[[i1]][, column.check[i4]])[levels(data1[[i1]][, column.check[i4]]) %in% unique(as.character(data1[[i1]][, column.check[i4]]))]
+ data1[[i1]][, column.check[i4]] <- factor(as.character(data1[[i1]][, column.check[i4]]), levels = tempo.levels)
+ if(column.check[i4] %in% categ[[i1]] & ! is.null(categ.class.order)){
+ categ.class.order[[i1]] <- levels(data1[[i1]][, column.check[i4]])[levels(data1[[i1]][, column.check[i4]]) %in% unique(data1[[i1]][, column.check[i4]])] # remove the absent class in the categ.class.order vector
+ data1[[i1]][, column.check[i4]] <- factor(as.character(data1[[i1]][, column.check[i4]]), levels = unique(categ.class.order[[i1]]))
+ }
+ }
+ }
+ }
+ }
+ # end na detection and removal
+ # From here, data1 and data.ini have no more NA or NaN in x, y, categ (if categ != NULL) and facet.categ (if categ != NULL)
+ if( ! is.null(categ.class.order)){
+ # the following check will be done several times but I prefer to keep it here, after the creation of categ
+ if(is.null(categ[[i1]]) & ! is.null(categ.class.order[[i1]])){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ ARGUMENT CANNOT BE NULL IF COMPARTMENT ", i1, " OF categ.class.order ARGUMENT IS NOT NULL: ", paste(categ.class.order, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ if(is.null(categ.class.order[[i1]])){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE categ.class.order COMPARTMENT ", i1, " IS NULL. ALPHABETICAL ORDER WILL BE APPLIED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ data1[[i1]][, categ[[i1]]] <- factor(as.character(data1[[i1]][, categ[[i1]]])) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ categ.class.order[[i1]] <- levels(data1[[i1]][, categ[[i1]]]) # character vector that will be used later
+ }else{
+ tempo <- fun_check(data = categ.class.order[[i1]], data.name = paste0("COMPARTMENT ", i1 , " OF categ.class.order ARGUMENT"), class = "vector", mode = "character", length = length(levels(data1[[i1]][, categ[[i1]]])), fun.name = function.name) # length(data1[, categ[i1]) -> if data1[, categ[i1] was initially character vector, then conversion as factor after the NA removal, thus class number ok. If data1[, categ[i1] was initially factor, no modification after the NA removal, thus class number ok
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(any(duplicated(categ.class.order[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ.class.order ARGUMENT CANNOT HAVE DUPLICATED CLASSES: ", paste(categ.class.order[[i1]], collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if( ! (all(categ.class.order[[i1]] %in% unique(data1[[i1]][, categ[[i1]]])) & all(unique(data1[[i1]][, categ[[i1]]]) %in% categ.class.order[[i1]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nCOMPARTMENT ", i1, " OF categ.class.order ARGUMENT MUST BE CLASSES OF COMPARTMENT ", i1, " OF categ ARGUMENT\nHERE IT IS:\n", paste(categ.class.order[[i1]], collapse = " "), "\nFOR COMPARTMENT ", i1, " OF categ.class.order AND IT IS:\n", paste(unique(data1[[i1]][, categ[[i1]]]), collapse = " "), "\nFOR COLUMN ", categ[[i1]], " OF data1 NUMBER ", i1)
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]], levels = categ.class.order[[i1]]) # reorder the factor
+ }
+ names(categ.class.order)[i1] <- categ[[i1]]
+ }
+ }
+ # OK: if categ.class.order is not NULL, all the NULL categ.class.order columns of data1 are character from here
+
+ if( ! is.null(legend.name[[i1]])){
+ tempo <- fun_check(data = legend.name[[i1]], data.name = ifelse(length(legend.name)== 1L, "legend.name", paste0("legend.name NUMBER ", i1)),, class = "vector", mode = "character", length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if( ! is.null(color)){ # if color is NULL, will be filled later on
+ # check the nature of color
+ if(is.null(color[[i1]])){
+ compart.null.color <- compart.null.color + 1
+ color[[i1]] <- grey(compart.null.color / 8) # cannot be more than 7 overlays. Thus 7 different greys. 8/8 is excluded because white dots
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL COLOR IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " ASSOCIATED TO ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", SINGLE COLOR ", paste(color[[i1]], collapse = " "), " HAS BEEN ATTRIBUTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ tempo1 <- fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "vector", mode = "character", na.contain = TRUE, fun.name = function.name) # na.contain = TRUE in case of colum of data1
+ tempo2 <- fun_check(data = color[[i1]], data.name = ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), class = "factor", na.contain = TRUE, fun.name = function.name) # idem
+ if(tempo1$problem == TRUE & tempo2$problem == TRUE){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A FACTOR OR CHARACTER VECTOR OR INTEGER VECTOR") # integer possible because dealt above
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if( ! (all(color[[i1]] %in% colors() | grepl(pattern = "^#", color[[i1]])))){ # check that all strings of low.color start by #
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A HEXADECIMAL COLOR VECTOR STARTING BY # AND/OR COLOR NAMES GIVEN BY colors() OR A COLUMN NAME OF THE data1 PARAMETER: ", paste(unique(color[[i1]]), collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(any(is.na(color[[i1]]))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), ", THE COLORS:\n", paste(unique(color[[i1]]), collapse = " "), "\nCONTAINS NA")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # end check the nature of color
+ # check the length of color
+ if(is.null(categ) & length(color[[i1]]) != 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A SINGLE COLOR IF categ IS NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if( ! is.null(categ)){
+ # No problem of NA management by ggplot2 because already removed
+ if(categ[[i1]] == "fake_categ" & length(color[[i1]]) != 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE A SINGLE COLOR IF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IS NULL")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(length(color[[i1]]) == length(unique(data1[[i1]][, categ[[i1]]]))){ # here length(color) is equal to the different number of categ
+ data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", THE FOLLOWING COLORS:\n", paste(color[[i1]], collapse = " "), "\nHAVE BEEN ATTRIBUTED TO THESE CLASSES:\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else if(length(color[[i1]]) == length(data1[[i1]][, categ[[i1]]])){# here length(color) is equal to nrow(data1[[i1]]) -> Modif to have length(color) equal to the different number of categ (length(color) == length(levels(data1[[i1]][, categ[[i1]]])))
+ data1[[i1]] <- cbind(data1[[i1]], color = color[[i1]], stringsAsFactors = TRUE)
+ tempo.check <- unique(data1[[i1]][ , c(categ[[i1]], "color")])
+ if( ! (nrow(data1[[i1]]) == length(color[[i1]]) & nrow(tempo.check) == length(unique(data1[[i1]][ , categ[[i1]]])))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color")), " ARGUMENT HAS THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nBUT IS INCORRECTLY ASSOCIATED TO EACH CLASS OF THIS categ:\n", paste(unique(mapply(FUN = "paste", data1[[i1]][ ,categ[[i1]]], data1[[i1]][ ,"color"])), collapse = "\n"))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ color[[i1]] <- unique(color[[i1]][order(data1[[i1]][, categ[[i1]]])]) # Modif to have length(color) equal to the different number of categ (length(color) == length(levels(data1[[i1]][, categ[[i1]]])))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count, ") FROM FUNCTION ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " HAS THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " COLUMN VALUES\nCOLORS HAVE BEEN RESPECTIVELY ASSOCIATED TO EACH CLASS OF categ AS:\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "), "\n", paste(color[[i1]], collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else if(length(color[[i1]])== 1L){
+ data1[[i1]][, categ[[i1]]] <- factor(data1[[i1]][, categ[[i1]]]) # if already a factor, change nothing, if characters, levels according to alphabetical order
+ color[[i1]] <- rep(color[[i1]], length(levels(data1[[i1]][, categ[[i1]]])))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") IN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), ", COLOR HAS LENGTH 1 MEANING THAT ALL THE DIFFERENT CLASSES OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), "\n", paste(levels(factor(data1[[i1]][, categ[[i1]]])), collapse = " "), "\nWILL HAVE THE SAME COLOR\n", paste(color[[i1]], collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST BE\n(1) LENGTH 1\nOR (2) THE LENGTH OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), " COLUMN VALUES\nOR (3) THE LENGTH OF THE CLASSES IN THIS COLUMN\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LENGTH ", length(data1[[i1]][, categ[[i1]]]), " AND CATEG CLASS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])), "\nPRESENCE OF NA IN THE COLUMN x, y OR categ OF data1 COULD BE THE PROBLEME")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }
+ if((geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline") & ! is.null(categ[[i1]])){ # add here after the color management, to deal with the different lines to plot inside any data[[i1]]
+ if(categ[[i1]] == "fake_categ"){
+ data1[[i1]][, "fake_categ"] <- factor(paste0("Line_", formatC(1:nrow(data1[[i2]]), width = nchar(nrow(data1[[i2]])), flag = "0")))
+ }
+ }
+ tempo <- fun_check(data = alpha[[i1]], data.name = ifelse(length(alpha)== 1L, "alpha", paste0("alpha NUMBER ", i1)), prop = TRUE, length = 1, fun.name = function.name)
+ if(tempo$problem == TRUE){
+ stop(paste0("\n\n================\n\n", tempo$text, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ # end loop (checking inside list compartment)
+ if(length(data1) > 1){
+ if(length(unique(unlist(x)[ ! x == "fake_x"])) > 1){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE x ARGUMENT DOES NOT CONTAIN IDENTICAL COLUMN NAMES:\n", paste(unlist(x), collapse = " "), "\nX-AXIS OVERLAYING DIFFERENT VARIABLES?")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(length(data1) > 1){
+ if(length(unique(unlist(y)[ ! y == "fake_y"])) > 1){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE y ARGUMENT DOES NOT CONTAIN IDENTICAL COLUMN NAMES:\n", paste(unlist(y), collapse = " "), "\nY-AXIS OVERLAYING DIFFERENT VARIABLES?")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ if(sum(geom %in% "geom_point") > 3){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT HAVE MORE THAN THREE \"geom_point\" ELEMENTS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(length(geom) - sum(geom %in% "geom_point") > 3){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": geom ARGUMENT CANNOT HAVE MORE THAN THREE LINE ELEMENTS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ # x.lim management before transfo by x.log
+ if(x.log != "no" & ! is.null(x.lim)){
+ if(any(x.lim <= 0)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE x.log ARGUMENT SET TO ", x.log, ":\n", paste(x.lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(any( ! is.finite(if(x.log == "log10"){log10(x.lim)}else{log2(x.lim)}))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT RETURNS INF/NA WITH THE x.log ARGUMENT SET TO ", x.log, "\nAS SCALE COMPUTATION IS ", ifelse(x.log == "log10", "log10", "log2"), ":\n", paste(if(x.log == "log10"){log10(x.lim)}else{log2(x.lim)}, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(x.log != "no" & x.include.zero == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") x.log ARGUMENT SET TO ", x.log, " AND x.include.zero ARGUMENT SET TO TRUE -> x.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ x.include.zero <- FALSE
+ }
+ # end x.lim management before transfo by x.log
+ # y.lim management before transfo by y.log
+ if(y.log != "no" & ! is.null(y.lim)){
+ if(any(y.lim <= 0)){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(any( ! is.finite(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT RETURNS INF/NA WITH THE y.log ARGUMENT SET TO ", y.log, "\nAS SCALE COMPUTATION IS ", ifelse(y.log == "log10", "log10", "log2"), ":\n", paste(if(y.log == "log10"){log10(y.lim)}else{log2(y.lim)}, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(y.log != "no" & y.include.zero == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") y.log ARGUMENT SET TO ", y.log, " AND y.include.zero ARGUMENT SET TO TRUE -> y.include.zero ARGUMENT RESET TO FALSE BECAUSE 0 VALUE CANNOT BE REPRESENTED IN LOG SCALE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ y.include.zero <- FALSE
+ }
+ # end y.lim management before transfo by y.log
+ # end other checkings
+ # reserved word checking
+ #already done above
+ # end reserved word checking
+ # end second round of checking and data preparation
+
+
+ # package checking
+ fun_pack(req.package = c(
+ "gridExtra",
+ "ggplot2",
+ "lemon",
+ "scales"
+ ), lib.path = lib.path)
+ # packages Cairo and grid tested by fun_gg_point_rast()
+ # end package checking
+
+
+
+
+ # main code
+ # axes management
+ if(is.null(x.lim)){
+ if(any(unlist(mapply(FUN = "[[", data1, x, SIMPLIFY = FALSE)) %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE x COLUMN IN data1 CONTAINS -Inf OR Inf VALUES THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ x.lim <- suppressWarnings(range(unlist(mapply(FUN = "[[", data1, x, SIMPLIFY = FALSE)), na.rm = TRUE, finite = TRUE)) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only. y.lim added here. If NULL, ok if y argument has values
+ }else if(x.log != "no"){
+ x.lim <- get(x.log)(x.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ }
+ if(x.log != "no"){
+ if(any( ! is.finite(x.lim))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nx.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE x.log ARGUMENT SET TO ", x.log, ":\n", paste(x.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(suppressWarnings(all(x.lim %in% c(Inf, -Inf)))){ # happen when x is only NULL
+ if(all(unlist(geom) %in% c("geom_vline", "geom_stick"))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " NOT POSSIBLE TO DRAW geom_vline OR geom_stick KIND OF LINES ALONE IF x.lim ARGUMENT IS SET TO NULL, SINCE NO X-AXIS DEFINED (", ifelse(length(x)== 1L, "x", paste0("ELEMENT ", i1, " OF x")), " ARGUMENT MUST BE NULL FOR THESE KIND OF LINES)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, " x.lim ARGUMENT MADE OF NA, -Inf OR Inf ONLY: ", paste(x.lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ x.lim.order <- order(x.lim) # to deal with inverse axis
+ # print(x.lim.order)
+ x.lim <- sort(x.lim)
+ x.lim[1] <- x.lim[1] - abs(x.lim[2] - x.lim[1]) * ifelse(diff(x.lim.order) > 0, x.right.extra.margin, x.left.extra.margin) # diff(x.lim.order) > 0 means not inversed axis
+ x.lim[2] <- x.lim[2] + abs(x.lim[2] - x.lim[1]) * ifelse(diff(x.lim.order) > 0, x.left.extra.margin, x.right.extra.margin) # diff(x.lim.order) > 0 means not inversed axis
+ if(x.include.zero == TRUE){ # no need to check x.log != "no" because done before
+ x.lim <- range(c(x.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ }
+ x.lim <- x.lim[x.lim.order]
+ if(any(is.na(x.lim))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 3")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(is.null(y.lim)){
+ if(any(unlist(mapply(FUN = "[[", data1, y, SIMPLIFY = FALSE)) %in% c(Inf, -Inf))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE y COLUMN IN data1 CONTAINS -Inf OR Inf VALUES THAT WILL NOT BE CONSIDERED IN THE PLOT RANGE")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ y.lim <- suppressWarnings(range(unlist(mapply(FUN = "[[", data1, y, SIMPLIFY = FALSE)), na.rm = TRUE, finite = TRUE)) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only. y.lim added here. If NULL, ok if y argument has values
+ }else if(y.log != "no"){
+ y.lim <- get(y.log)(y.lim) # no env = sys.nframe(), inherit = FALSE in get() because look for function in the classical scope
+ }
+ if(y.log != "no"){
+ if(any( ! is.finite(y.lim))){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\ny.lim ARGUMENT CANNOT HAVE ZERO OR NEGATIVE VALUES WITH THE y.log ARGUMENT SET TO ", y.log, ":\n", paste(y.lim, collapse = " "), "\nPLEASE, CHECK DATA VALUES (PRESENCE OF ZERO OR INF VALUES)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ if(suppressWarnings(all(y.lim %in% c(Inf, -Inf)))){ # happen when y is only NULL
+ if(all(unlist(geom) == "geom_vline")){
+ tempo.cat <- paste0("ERROR IN ", function.name, " NOT POSSIBLE TO DRAW geom_vline KIND OF LINES ALONE IF y.lim ARGUMENT IS SET TO NULL, SINCE NO Y-AXIS DEFINED (", ifelse(length(y)== 1L, "y", paste0("ELEMENT ", i1, " OF y")), " ARGUMENT MUST BE NULL FOR THESE KIND OF LINES)")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, " y.lim ARGUMENT MADE OF NA, -Inf OR Inf ONLY: ", paste(y.lim, collapse = " "))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ y.lim.order <- order(y.lim) # to deal with inverse axis
+ y.lim <- sort(y.lim)
+ y.lim[1] <- y.lim[1] - abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.bottom.extra.margin, y.top.extra.margin) # diff(y.lim.order) > 0 means not inversed axis
+ y.lim[2] <- y.lim[2] + abs(y.lim[2] - y.lim[1]) * ifelse(diff(y.lim.order) > 0, y.top.extra.margin, y.bottom.extra.margin) # diff(y.lim.order) > 0 means not inversed axis
+ if(y.include.zero == TRUE){ # no need to check y.log != "no" because done before
+ y.lim <- range(c(y.lim, 0), na.rm = TRUE, finite = TRUE) # finite = TRUE removes all the -Inf and Inf except if only this. In that case, whatever the -Inf and/or Inf present, output -Inf;Inf range. Idem with NA only
+ }
+ y.lim <- y.lim[y.lim.order]
+ if(any(is.na(y.lim))){
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 4")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ # end axes management
+
+
+
+
+ # create a fake categ if NULL to deal with legend display
+ if(is.null(categ)){
+ categ <- vector("list", length(data1))
+ categ[] <- "fake_categ"
+ for(i2 in 1:length(data1)){
+ data1[[i2]] <- cbind(data1[[i2]], fake_categ = "", stringsAsFactors = TRUE)
+ if(geom[[i2]] == "geom_hline" | geom[[i2]] == "geom_vline"){
+ data1[[i2]][, "fake_categ"] <- factor(paste0("Line_", 1:nrow(data1[[i2]])))
+ }
+ }
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL categ ARGUMENT -> FAKE \"fake_categ\" COLUMN ADDED TO EACH DATA FRAME OF data1, AND FILLED WITH \"\"")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ # categ is not NULL anymore
+ if(is.null(categ.class.order)){
+ categ.class.order <- vector("list", length = length(data1))
+ tempo.categ.class.order <- NULL
+ for(i2 in 1:length(categ.class.order)){
+ categ.class.order[[i2]] <- levels(data1[[i2]][, categ[[i2]]])
+ names(categ.class.order)[i2] <- categ[[i2]]
+ tempo.categ.class.order <- c(tempo.categ.class.order, ifelse(i2 != 1, "\n", ""), categ.class.order[[i2]])
+ }
+ if(any(unlist(legend.disp))){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") THE categ.class.order SETTING IS NULL. ALPHABETICAL ORDER WILL BE APPLIED FOR CLASS ORDERING:\n", paste(tempo.categ.class.order, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end create a fake categ if NULL to deal with legend display
+ # categ.class.order is not NULL anymore
+
+
+ # vector of color with length as in levels(categ) of data1
+ if(is.null(color)){
+ color <- vector("list", length(data1))
+ length.categ.list <- lapply(lapply(mapply(FUN = "[[", data1, categ, SIMPLIFY = FALSE), FUN = unique), FUN = function(x){length(x[ ! is.na(x)])})
+ length.categ.list[sapply(categ, FUN = "==", "fake_categ")] <- 1 # when is.null(color), a single color for all the dots or lines of data[[i1]] that contain "fake_categ" category
+ total.categ.length <- sum(unlist(length.categ.list), na.rm = TRUE)
+ tempo.color <- fun_gg_palette(total.categ.length)
+ tempo.count <- 0
+ for(i2 in 1:length(data1)){
+ color[[i2]] <- tempo.color[(1:length.categ.list[[i2]]) + tempo.count]
+ tempo.count <- tempo.count + length.categ.list[[i2]]
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") NULL color ARGUMENT -> COLORS RESPECTIVELY ATTRIBUTED TO EACH CLASS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i2, " OF categ ARGUMENT")), " (", categ[[i2]], ") IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i2, " OF data1 ARGUMENT")), ":\n", paste(color[[i2]], collapse = " "), "\n", paste(if(all(levels(data1[[i2]][, categ[[i2]]]) == "")){'\"\"'}else{levels(data1[[i2]][, categ[[i2]]])}, collapse = " "))
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end vector of color with length as in levels(categ) of data1
+ # color is not NULL anymore
+
+
+
+
+
+ # last check
+ for(i1 in 1:length(data1)){
+ if(categ[[i1]] != "fake_categ" & length(color[[i1]]) != length(unique(data1[[i1]][, categ[[i1]]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " LAST CHECK: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE THE LENGTH OF LEVELS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LEVELS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])), "\nREMINDER: A SINGLE COLOR PER CLASS OF CATEG AND A SINGLE CLASS OF CATEG PER COLOR MUST BE RESPECTED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else if(categ[[i1]] == "fake_categ" & length(color[[i1]]) != 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, " LAST CHECK: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE LENGTH 1 WHEN ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IS NULL\nHERE IT IS COLOR LENGTH ", length(color[[i1]]))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ # end last check
+
+
+
+
+
+ # conversion of geom_hline and geom_vline
+ for(i1 in 1:length(data1)){
+ if(geom[[i1]] == "geom_hline" | geom[[i1]] == "geom_vline"){
+ final.data.frame <- data.frame()
+ for(i3 in 1:nrow(data1[[i1]])){
+ tempo.data.frame <- rbind(data1[[i1]][i3, ], data1[[i1]][i3, ], stringsAsFactors = TRUE)
+ if(geom[[i1]] == "geom_hline"){
+ tempo.data.frame[, x[[i1]]] <- x.lim
+ }else if(geom[[i1]] == "geom_vline"){
+ tempo.data.frame[, y[[i1]]] <- y.lim
+ }else{
+ tempo.cat <- paste0("ERROR IN ", function.name, ": CODE INCONSISTENCY 5")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ # 3 lines below inactivated because I put that above
+ # if(is.null(categ[[i1]])){
+ # data1[, "fake_categ"] <- paste0("Line_", i3)
+ # }
+ final.data.frame <- rbind(final.data.frame, tempo.data.frame, stringsAsFactors = TRUE)
+ }
+ data1[[i1]] <- final.data.frame
+ geom[[i1]] <- "geom_line"
+ if(length(color[[i1]])== 1L){
+ color[[i1]] <- rep(color[[i1]], length(unique(data1[[i1]][ , categ[[i1]]])))
+ }else if(length(color[[i1]]) != length(unique(data1[[i1]][ , categ[[i1]]]))){
+ tempo.cat <- paste0("ERROR IN ", function.name, " geom_hline AND geom_vline CONVERSION TO FIT THE XLIM AND YLIM LIMITS OF THE DATA: ", ifelse(length(color)== 1L, "color", paste0("ELEMENT NUMBER ", i1, " OF color ARGUMENT")), " MUST HAVE THE LENGTH OF LEVELS OF ", ifelse(length(categ)== 1L, "categ", paste0("ELEMENT ", i1, " OF categ ARGUMENT")), " IN ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i1, " OF data1 ARGUMENT")), "\nHERE IT IS COLOR LENGTH ", length(color[[i1]]), " VERSUS CATEG LEVELS LENGTH ", length(unique(data1[[i1]][, categ[[i1]]])))
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ }
+ }
+ # end conversion of geom_hline and geom_vline
+
+
+
+
+ # kind of geom_point (vectorial or raster)
+ scatter.kind <- vector("list", length = length(data1)) # list of same length as data1, that will be used to use either ggplot2::geom_point() (vectorial dot layer) or fun_gg_point_rast() (raster dot layer)
+ fix.ratio <- FALSE
+ if(is.null(raster.threshold)){
+ if(raster == TRUE){
+ scatter.kind[] <- "fun_gg_point_rast" # not important to fill everything: will be only used when geom == "geom_point"
+ fix.ratio <- TRUE
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") RASTER PLOT GENERATED -> ASPECT RATIO OF THE PLOT REGION SET BY THE raster.ratio ARGUMENT (", fun_round(raster.ratio, 2), ") TO AVOID A BUG OF ELLIPSOID DOT DRAWING")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }else{
+ scatter.kind[] <- "ggplot2::geom_point"
+ }
+ }else{
+ for(i2 in 1:length(data1)){
+ if(geom[[i2]] == "geom_point"){
+ if(nrow(data1[[i2]]) <= raster.threshold){
+ scatter.kind[[i2]] <- "ggplot2::geom_point"
+ }else{
+ scatter.kind[[i2]] <- "fun_gg_point_rast"
+ fix.ratio <- TRUE
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") ", ifelse(length(data1)== 1L, "data1 ARGUMENT", paste0("DATA FRAME NUMBER ", i2, " OF data1 ARGUMENT")), " LAYER AS RASTER (NOT VECTORIAL)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ }
+ if(any(unlist(scatter.kind) == "fun_gg_point_rast")){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") RASTER PLOT GENERATED -> ASPECT RATIO OF THE PLOT REGION SET BY THE raster.ratio ARGUMENT (", fun_round(raster.ratio, 2), ") TO AVOID A BUG OF ELLIPSOID DOT DRAWING")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }
+ # end kind of geom_point (vectorial or raster)
+
+
+
+
+ # no need loop part
+ coord.names <- NULL
+ tempo.gg.name <- "gg.indiv.plot."
+ tempo.gg.count <- 0
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::ggplot()", if(is.null(add)){""}else{add})))) # add added here to have the facets
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::xlab(if(is.null(x.lab)){x[[1]]}else{x.lab}))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ylab(if(is.null(y.lab)){y[[1]]}else{y.lab}))
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::ggtitle(title))
+ # text angle management
+ x.tempo.just <- fun_gg_just(angle = x.text.angle, pos = "bottom", kind = "axis")
+ y.tempo.just <- fun_gg_just(angle = y.text.angle, pos = "left", kind = "axis")
+ # end text angle management
+ add.check <- TRUE
+ if( ! is.null(add)){ # if add is NULL, then = 0
+ if(grepl(pattern = "ggplot2::theme", add) == TRUE){
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") \"ggplot2::theme\" STRING DETECTED IN THE add ARGUMENT\n-> INTERNAL GGPLOT2 THEME FUNCTIONS theme() AND theme_classic() HAVE BEEN INACTIVATED, TO BE USED BY THE USER\n-> article ARGUMENT WILL BE IGNORED\nIT IS RECOMMENDED TO USE \"+ theme(aspect.ratio = raster.ratio)\" IF RASTER MODE IS ACTIVATED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ add.check <- FALSE
+ }
+ }
+ if(add.check == TRUE & article == TRUE){
+ # WARNING: not possible to add several times theme(). NO message but the last one overwrites the others
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::theme_classic(base_size = text.size))
+ if(grid == TRUE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ line = ggplot2::element_line(size = 0.5),
+ axis.line.y.left = ggplot2::element_line(colour = "black"), # draw lines for the y axis
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"), # draw lines for the x axis
+ panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.minor.x = ggplot2::element_line(colour = "grey90", size = 0.25),
+ panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
+ axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
+ axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
+ aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
+ ))
+ }else{
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_line(size = 0.5),
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ axis.line.y.left = ggplot2::element_line(colour = "black"),
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"),
+ axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
+ axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
+ aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
+ ))
+ }
+ }else if(add.check == TRUE & article == FALSE){
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), m.gg <- ggplot2::theme(
+ text = ggplot2::element_text(size = text.size),
+ plot.title = ggplot2::element_text(size = title.text.size), # stronger than text
+ line = ggplot2::element_line(size = 0.5),
+ legend.key = ggplot2::element_rect(color = "white", size = 1.5), # size of the frame of the legend
+ panel.background = ggplot2::element_rect(fill = "grey95"),
+ axis.line.y.left = ggplot2::element_line(colour = "black"),
+ axis.line.x.bottom = ggplot2::element_line(colour = "black"),
+ panel.grid.major.x = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.minor.x = ggplot2::element_line(colour = "grey90", size = 0.25),
+ panel.grid.major.y = ggplot2::element_line(colour = "grey85", size = 0.75),
+ panel.grid.minor.y = ggplot2::element_line(colour = "grey90", size = 0.25),
+ strip.background = ggplot2::element_rect(fill = "white", colour = "black"),
+ axis.text.x = ggplot2::element_text(angle = x.tempo.just$angle, hjust = x.tempo.just$hjust, vjust = x.tempo.just$vjust),
+ axis.text.y = ggplot2::element_text(angle = y.tempo.just$angle, hjust = y.tempo.just$hjust, vjust = y.tempo.just$vjust),
+ aspect.ratio = if(fix.ratio == TRUE){raster.ratio}else{NULL} # for raster
+ # do not work -> legend.position = "none" # to remove the legend completely: https://www.datanovia.com/en/blog/how-to-remove-legend-from-a-ggplot/
+ ))
+ }
+ # end no need loop part
+
+
+ # loop part
+ point.count <- 0
+ line.count <- 0
+ lg.order <- vector(mode = "list", length = 6) # order of the legend
+ lg.order <- lapply(lg.order, as.numeric) # order of the legend
+ lg.color <- vector(mode = "list", length = 6) # color of the legend
+ lg.dot.shape <- vector(mode = "list", length = 6) # etc.
+ lg.dot.size <- vector(mode = "list", length = 6) # etc.
+ lg.dot.size <- lapply(lg.dot.size, as.numeric) # etc.
+ lg.dot.border.size <- vector(mode = "list", length = 6) # etc.
+ lg.dot.border.size <- lapply(lg.dot.border.size, as.numeric) # etc.
+ lg.dot.border.color <- vector(mode = "list", length = 6) # etc.
+ lg.line.size <- vector(mode = "list", length = 6) # etc.
+ lg.line.size <- lapply(lg.line.size, as.numeric) # etc.
+ lg.line.type <- vector(mode = "list", length = 6) # etc.
+ lg.alpha <- vector(mode = "list", length = 6) # etc.
+ lg.alpha <- lapply(lg.alpha, as.numeric) # etc.
+ for(i1 in 1:length(data1)){
+ if(geom[[i1]] == "geom_point"){
+ point.count <- point.count + 1
+ if(point.count== 1L){
+ fin.lg.disp[[1]] <- legend.disp[[point.count + line.count]]
+ lg.order[[1]] <- point.count + line.count
+ lg.color[[1]] <- color[[i1]] # if color == NULL -> NULL
+ lg.dot.shape[[1]] <- dot.shape[[i1]]
+ lg.dot.size[[1]] <- dot.size[[i1]]
+ lg.dot.border.size[[1]] <- dot.border.size[[i1]]
+ lg.dot.border.color[[1]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
+ if(plot == TRUE & fin.lg.disp[[1]] == TRUE & dot.shape[[1]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[1]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[1]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], fill = categ[[i1]]), shape = dot.shape[[i1]], size = dot.size[[i1]], stroke = dot.border.size[[i1]], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = if(i5== 1L){TRUE}else{FALSE})) # WARNING: a single color allowed for color argument outside aesthetic, but here a single color for border --> loop could be inactivated but kept for commodity # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_fill_manual(name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = as.character(color[[i1]]), breaks = class.categ)) # values are the values of fill, breaks reorder the classes according to class.categ in the legend, order argument of guide_legend determines the order of the different aesthetics in the legend (not order of classes). See guide_legend settings of scale_..._manual below
+ }
+ if(point.count== 2L){
+ fin.lg.disp[[2]] <- legend.disp[[point.count + line.count]]
+ lg.order[[2]] <- point.count + line.count
+ lg.color[[2]] <- color[[i1]] # if color == NULL -> NULL
+ lg.dot.shape[[2]] <- dot.shape[[i1]]
+ lg.dot.size[[2]] <- dot.size[[i1]]
+ lg.dot.border.size[[2]] <- dot.border.size[[i1]]
+ lg.dot.border.color[[2]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
+ if(plot == TRUE & fin.lg.disp[[2]] == TRUE & dot.shape[[2]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[2]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[2]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], shape = categ[[i1]]), size = dot.size[[i1]], stroke = dot.border.size[[i1]], fill = color[[i1]][i5], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = FALSE)) # WARNING: a single color allowed for fill argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_shape_manual(name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(dot.shape[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of shape, breaks reorder the classes according to class.categ in the legend. See guide_legend settings of scale_..._manual below
+
+ }
+ if(point.count== 3L){
+ fin.lg.disp[[3]] <- legend.disp[[point.count + line.count]]
+ lg.order[[3]] <- point.count + line.count
+ lg.color[[3]] <- color[[i1]] # if color == NULL -> NULL
+ lg.dot.shape[[3]] <- dot.shape[[i1]]
+ lg.dot.size[[3]] <- dot.size[[i1]]
+ lg.dot.border.size[[3]] <- dot.border.size[[i1]]
+ lg.dot.border.color[[3]] <- dot.border.color[[i1]] # if dot.border.color == NULL -> NULL
+ if(plot == TRUE & fin.lg.disp[[3]] == TRUE & dot.shape[[3]] %in% 0:14 & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE DOTS (DOT LAYER NUMBER ", point.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[3]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[3]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = scatter.kind[[i1]]))(data = tempo.data.frame, mapping = ggplot2::aes_string(x = x[[i1]], y = y[[i1]], stroke = categ[[i1]]), shape = dot.shape[[i1]], size = dot.size[[i1]], fill = color[[i1]][i5], stroke = dot.border.size[[i1]], color = if(dot.shape[[i1]] %in% 21:24 & ! is.null(dot.border.color)){dot.border.color[[i1]]}else{color[[i1]][i5]}, alpha = alpha[[i1]], show.legend = FALSE)) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "stroke", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(dot.border.size[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of stroke, breaks reorder the classes according to class.categ in the legend. See guide_legend settings of scale_..._manual below
+
+ }
+ }else{
+ line.count <- line.count + 1
+ if(line.count== 1L){
+ fin.lg.disp[[4]] <- legend.disp[[point.count + line.count]]
+ lg.order[[4]] <- point.count + line.count
+ lg.color[[4]] <- color[[i1]] # if color == NULL -> NULL
+ lg.line.size[[4]] <- line.size[[i1]]
+ lg.line.type[[4]] <- line.type[[i1]]
+ if(plot == TRUE & fin.lg.disp[[4]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[4]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[4]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
+ ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
+ "(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
+ x[[i1]],
+ ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
+ y[[i1]],
+ if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
+ ", linetype = ",
+ categ[[i1]],
+ "), color = \"",
+ color[[i1]][i5],
+ "\", size = ",
+ line.size[[i1]],
+ ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
+ ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
+ ", alpha = ",
+ alpha[[i1]],
+ ", show.legend = ",
+ ifelse(i5== 1L, TRUE, FALSE),
+ ")"
+ )))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "linetype", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(line.type[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, with different values, breaks reorder the classes according to class.categ in the legend
+ }
+ if(line.count== 2L){
+ fin.lg.disp[[5]] <- legend.disp[[point.count + line.count]]
+ lg.order[[5]] <- point.count + line.count
+ lg.color[[5]] <- color[[i1]] # if color == NULL -> NULL
+ lg.line.size[[5]] <- line.size[[i1]]
+ lg.line.type[[5]] <- line.type[[i1]]
+ if(plot == TRUE & fin.lg.disp[[5]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[5]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[5]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
+ ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
+ "(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
+ x[[i1]],
+ ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
+ y[[i1]],
+ if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
+ ", alpha = ",
+ categ[[i1]],
+ "), color = \"",
+ color[[i1]][i5],
+ "\", size = ",
+ line.size[[i1]],
+ ", linetype = ",
+ ifelse(is.numeric(line.type[[i1]]), "", "\""),
+ line.type[[i1]],
+ ifelse(is.numeric(line.type[[i1]]), "", "\""),
+ ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
+ ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
+ ", show.legend = FALSE)"
+ )))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "alpha", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(alpha[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, with different values, breaks reorder the classes according to class.categ in the legend
+ }
+ if(line.count== 3L){
+ fin.lg.disp[[6]] <- legend.disp[[point.count + line.count]]
+ lg.order[[6]] <- point.count + line.count
+ lg.color[[6]] <- color[[i1]] # if color == NULL -> NULL
+ lg.line.size[[6]] <- line.size[[i1]]
+ lg.line.type[[6]] <- line.type[[i1]]
+ if(plot == TRUE & fin.lg.disp[[6]] == TRUE & ((length(dev.list()) > 0 & names(dev.cur()) == "windows") | (length(dev.list())== 0L & Sys.info()["sysname"] == "Windows"))){ # if any Graph device already open and this device is "windows", or if no Graph device opened yet and we are on windows system -> prevention of alpha legend bug on windows using value 1
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") GRAPHIC DEVICE USED ON A WINDOWS SYSTEM ->\nTRANSPARENCY OF THE LINES (LINE LAYER NUMBER ", line.count, ") IS INACTIVATED IN THE LEGEND TO PREVENT A WINDOWS DEPENDENT BUG (SEE https://github.com/tidyverse/ggplot2/issues/2452)\nTO OVERCOME THIS ON WINDOWS, USE ANOTHER DEVICE (pdf() FOR INSTANCE)")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ lg.alpha[[6]] <- 1 # to avoid a bug on windows: if alpha argument is different from 1 for lines (transparency), then lines are not correctly displayed in the legend when using the R GUI (bug https://github.com/tidyverse/ggplot2/issues/2452). No bug when using a pdf
+ }else{
+ lg.alpha[[6]] <- alpha[[i1]]
+ }
+ class.categ <- levels(factor(data1[[i1]][, categ[[i1]]]))
+ for(i5 in 1:length(color[[i1]])){ # or length(class.categ). It is the same because already checked that lengths are the same
+ tempo.data.frame <- data1[[i1]][data1[[i1]][, categ[[i1]]] == class.categ[i5], ]
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = paste0("ggplot2::", # no CR here te0("ggpl
+ ifelse(geom[[i1]] == 'geom_stick', 'geom_segment', geom[[i1]]), # geom_segment because geom_stick converted to geom_segment for plotting
+ "(data = tempo.data.frame, mapping = ggplot2::aes(x = ",
+ x[[i1]],
+ ifelse(geom[[i1]] == 'geom_stick', ", yend = ", ", y = "),
+ y[[i1]],
+ if(geom[[i1]] == 'geom_stick'){paste0(', xend = ', x[[i1]], ', y = ', ifelse(is.null(geom.stick.base), y.lim[1], geom.stick.base[[i1]]))},
+ ", size = ",
+ categ[[i1]],
+ "), color = \"",
+ color[[i1]][i5],
+ "\", linetype = ",
+ ifelse(is.numeric(line.type[[i1]]), "", "\""),
+ line.type[[i1]],
+ ifelse(is.numeric(line.type[[i1]]), "", "\""),
+ ifelse(geom[[i1]] == 'geom_path', ', lineend = \"round\"', ''),
+ ifelse(geom[[i1]] == 'geom_step', paste0(', direction = \"', geom.step.dir[[i1]], '\"'), ''),
+ ", alpha = ",
+ alpha[[i1]],
+ ", show.legend = FALSE)"
+ )))) # WARNING: a single color allowed for color argument outside aesthetic, hence the loop # legend.show option do not remove the legend, only the aesthetic of the legend (dot, line, etc.). Used here to avoid multiple layers of legend which corrupt transparency
+ coord.names <- c(coord.names, paste0(geom[[i1]], ".", class.categ[i5]))
+ }
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_discrete_manual(aesthetics = "size", name = if(is.null(legend.name)){NULL}else{legend.name[[i1]]}, values = rep(line.size[[i1]], length(color[[i1]])), breaks = class.categ)) # values are the values of linetype. 1 means solid. Regarding the alpha bug, I have tried different things without success: alpha in guide alone, in geom alone, in both, breaks reorder the classes according to class.categ in the legend
+ }
+ }
+ }
+ # end loop part
+
+
+
+
+ # legend display
+ tempo.legend.final <- 'ggplot2::guides(
fill = if(fin.lg.disp[[1]] == TRUE){
ggplot2::guide_legend(
order = lg.order[[1]],
@@ -13495,25 +13501,25 @@ shape = NA
}
)' # clip = "off" to have secondary ticks outside plot region does not work
if( ! is.null(legend.width)){
-if(any(unlist(legend.disp))){ # means some TRUE
-tempo.graph.info <- suppressMessages(ggplot2::ggplot_build(eval(parse(text = paste0(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + ', tempo.legend.final))))) # will be recovered later again, when ylim will be considered
-legend.final <- fun_gg_get_legend(ggplot_built = tempo.graph.info, fun.name = function.name) # get legend
-fin.lg.disp[] <- FALSE # remove all the legends. Must be done even if fin.lg.disp is not appearing in the code thenafter. Otherwise twice the legend
-if(is.null(legend.final) & plot == TRUE){ # even if any(unlist(legend.disp)) is TRUE
-legend.final <- fun_gg_empty_graph() # empty graph instead of legend
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON-NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON-NULL legend.width ARGUMENT\n")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
-}else if(plot == TRUE){ # means all FALSE
-legend.final <- ggplot2::ggplot()+ggplot2::theme_void() # empty graph instead of legend
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON-NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON-NULL legend.width ARGUMENT\n")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
-}
+ if(any(unlist(legend.disp))){ # means some TRUE
+ tempo.graph.info <- suppressMessages(ggplot2::ggplot_build(eval(parse(text = paste0(paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + "), ' + ', tempo.legend.final))))) # will be recovered later again, when ylim will be considered
+ legend.final <- fun_gg_get_legend(ggplot_built = tempo.graph.info, fun.name = function.name) # get legend
+ fin.lg.disp[] <- FALSE # remove all the legends. Must be done even if fin.lg.disp is not appearing in the code thenafter. Otherwise twice the legend
+ if(is.null(legend.final) & plot == TRUE){ # even if any(unlist(legend.disp)) is TRUE
+ legend.final <- fun_gg_empty_graph() # empty graph instead of legend
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON-NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON-NULL legend.width ARGUMENT\n")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
+ }else if(plot == TRUE){ # means all FALSE
+ legend.final <- ggplot2::ggplot()+ggplot2::theme_void() # empty graph instead of legend
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") LEGEND REQUESTED (NON-NULL categ ARGUMENT OR legend.show ARGUMENT SET TO TRUE)\nBUT IT SEEMS THAT THE PLOT HAS NO LEGEND -> EMPTY LEGEND SPACE CREATED BECAUSE OF THE NON-NULL legend.width ARGUMENT\n")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ }
}
if( ! any(unlist(legend.disp))){
-fin.lg.disp[] <- FALSE # remove all the legends. Must be done even if fin.lg.disp is not appearing in the code thenafter. Otherwise twice the legend
+ fin.lg.disp[] <- FALSE # remove all the legends. Must be done even if fin.lg.disp is not appearing in the code thenafter. Otherwise twice the legend
}
assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(text = tempo.legend.final)))
# end legend display
@@ -13526,135 +13532,136 @@ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), eval(parse(t
tempo.coord <- suppressMessages(ggplot2::ggplot_build(eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ", ' + ggplot2::scale_x_continuous(expand = c(0, 0), limits = sort(x.lim), oob = scales::rescale_none) + ggplot2::scale_y_continuous(expand = c(0, 0), limits = sort(y.lim), oob = scales::rescale_none)'))))$layout$panel_params[[1]]) # here I do not need the x-axis and y-axis orientation, I just need the number of main ticks
# x.second.tick.positions # coordinates of secondary ticks (only if x.second.tick.nb argument is non-null or if x.log argument is different from "no")
if(x.log != "no"){ # integer main ticks for log2 and log10
-tempo.scale <- (as.integer(min(x.lim, na.rm = TRUE)) - 1):(as.integer(max(x.lim, na.rm = TRUE)) + 1)
-}else{
-tempo <- if(is.null(attributes(tempo.coord$x$breaks))){tempo.coord$x$breaks}else{unlist(attributes(tempo.coord$x$breaks))}
-if(all(is.na(tempo))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$x$breaks")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(length(unique(x.lim)) <= 1){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nIT SEEMS THAT X-AXIS VALUES HAVE A NULL RANGE: ", paste(x.lim, collapse = " "), "\nPLEASE, USE THE x.lim ARGUMENT WITH 2 DIFFERENT VALUES TO SOLVE THIS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-tempo.scale <- fun_scale(lim = x.lim, n = ifelse(is.null(x.tick.nb), length(tempo[ ! is.na(tempo)]), x.tick.nb)) # in ggplot 3.3.0, tempo.coord$x.major_source replaced by tempo.coord$x$breaks. If fact: n = ifelse(is.null(x.tick.nb), length(tempo[ ! is.na(tempo)]), x.tick.nb)) replaced by n = ifelse(is.null(x.tick.nb), 4, x.tick.nb))
-}
+ tempo.scale <- (as.integer(min(x.lim, na.rm = TRUE)) - 1):(as.integer(max(x.lim, na.rm = TRUE)) + 1)
+}else{
+ tempo <- if(is.null(attributes(tempo.coord$x$breaks))){tempo.coord$x$breaks}else{unlist(attributes(tempo.coord$x$breaks))}
+ if(all(is.na(tempo))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$x$breaks")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(length(unique(x.lim)) <= 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nIT SEEMS THAT X-AXIS VALUES HAVE A NULL RANGE: ", paste(x.lim, collapse = " "), "\nPLEASE, USE THE x.lim ARGUMENT WITH 2 DIFFERENT VALUES TO SOLVE THIS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ tempo.scale <- fun_scale(lim = x.lim, n = ifelse(is.null(x.tick.nb), length(tempo[ ! is.na(tempo)]), x.tick.nb)) # in ggplot 3.3.0, tempo.coord$x.major_source replaced by tempo.coord$x$breaks. If fact: n = ifelse(is.null(x.tick.nb), length(tempo[ ! is.na(tempo)]), x.tick.nb)) replaced by n = ifelse(is.null(x.tick.nb), 4, x.tick.nb))
+ }
}
x.second.tick.values <- NULL
x.second.tick.pos <- NULL
if(x.log != "no"){
-tempo <- fun_inter_ticks(lim = x.lim, log = x.log)
-x.second.tick.values <- tempo$values
-x.second.tick.pos <- tempo$coordinates
-# if(vertical == TRUE){ # do not remove in case the bug is fixed
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
-geom = "segment", x = x.second.tick.pos,
-xend = x.second.tick.pos,
-y = if(diff(y.lim) > 0){tempo.coord$y.range[1]}else{tempo.coord$y.range[2]},
-yend = if(diff(y.lim) > 0){tempo.coord$y.range[1] + abs(diff(tempo.coord$y.range)) / 80}else{tempo.coord$y.range[2] - abs(diff(tempo.coord$y.range)) / 80}
-))
-# }else{ # not working because of the ggplot2 bug
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", y = x.second.tick.pos, yend = x.second.tick.pos, x = tempo.coord$x.range[1], xend = tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80))
-# }
-coord.names <- c(coord.names, "x.second.tick.positions")
+ tempo <- fun_inter_ticks(lim = x.lim, log = x.log)
+ x.second.tick.values <- tempo$values
+ x.second.tick.pos <- tempo$coordinates
+ # if(vertical == TRUE){ # do not remove in case the bug is fixed
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
+ geom = "segment",
+ x = x.second.tick.pos,
+ xend = x.second.tick.pos,
+ y = if(diff(y.lim) > 0){tempo.coord$y.range[1]}else{tempo.coord$y.range[2]},
+ yend = if(diff(y.lim) > 0){tempo.coord$y.range[1] + abs(diff(tempo.coord$y.range)) / 80}else{tempo.coord$y.range[2] - abs(diff(tempo.coord$y.range)) / 80}
+ ))
+ # }else{ # not working because of the ggplot2 bug
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", y = x.second.tick.pos, yend = x.second.tick.pos, x = tempo.coord$x.range[1], xend = tempo.coord$x.range[1] + diff(tempo.coord$x.range) / 80))
+ # }
+ coord.names <- c(coord.names, "x.second.tick.positions")
}else if(( ! is.null(x.second.tick.nb)) & x.log == "no"){
-# if(x.second.tick.nb > 0){ #inactivated because already checked before
-if(length(tempo.scale) < 2){
-tempo.cat1 <- c("x.tick.nb", "x.second.tick.nb")
-tempo.cat2 <- sapply(list(x.tick.nb, x.second.tick.nb), FUN = paste0, collapse = " ")
-tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE X-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-tempo <- fun_inter_ticks(lim = x.lim, log = x.log, breaks = tempo.scale, n = x.second.tick.nb)
-}
-x.second.tick.values <- tempo$values
-x.second.tick.pos <- tempo$coordinates
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
-geom = "segment",
-x = x.second.tick.pos,
-xend = x.second.tick.pos,
-y = if(diff(y.lim) > 0){tempo.coord$y.range[1]}else{tempo.coord$y.range[2]},
-yend = if(diff(y.lim) > 0){tempo.coord$y.range[1] + abs(diff(tempo.coord$y.range)) / 80}else{tempo.coord$y.range[2] - abs(diff(tempo.coord$y.range)) / 80}
-))
-coord.names <- c(coord.names, "x.second.tick.positions")
+ # if(x.second.tick.nb > 0){ #inactivated because already checked before
+ if(length(tempo.scale) < 2){
+ tempo.cat1 <- c("x.tick.nb", "x.second.tick.nb")
+ tempo.cat2 <- sapply(list(x.tick.nb, x.second.tick.nb), FUN = paste0, collapse = " ")
+ tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE X-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ tempo <- fun_inter_ticks(lim = x.lim, log = x.log, breaks = tempo.scale, n = x.second.tick.nb)
+ }
+ x.second.tick.values <- tempo$values
+ x.second.tick.pos <- tempo$coordinates
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
+ geom = "segment",
+ x = x.second.tick.pos,
+ xend = x.second.tick.pos,
+ y = if(diff(y.lim) > 0){tempo.coord$y.range[1]}else{tempo.coord$y.range[2]},
+ yend = if(diff(y.lim) > 0){tempo.coord$y.range[1] + abs(diff(tempo.coord$y.range)) / 80}else{tempo.coord$y.range[2] - abs(diff(tempo.coord$y.range)) / 80}
+ ))
+ coord.names <- c(coord.names, "x.second.tick.positions")
}
# for the ggplot2 bug with x.log, this does not work: eval(parse(text = ifelse(vertical == FALSE & x.log == "log10", "ggplot2::scale_x_continuous", "ggplot2::scale_x_continuous")))
assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_x_continuous(
-breaks = tempo.scale,
-minor_breaks = x.second.tick.pos,
-labels = if(x.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(x.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(x.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 10") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)},
-expand = c(0, 0), # remove space after after axis limits
-limits = sort(x.lim), # NA indicate that limits must correspond to data limits but xlim() already used
-oob = scales::rescale_none,
-trans = ifelse(diff(x.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_x_reverse() but create the problem of x-axis label disappearance with x.lim decreasing. Thus, do not use. Use xlim() below and after this
+ breaks = tempo.scale,
+ minor_breaks = x.second.tick.pos,
+ labels = if(x.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(x.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(x.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 10") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)},
+ expand = c(0, 0), # remove space after after axis limits
+ limits = sort(x.lim), # NA indicate that limits must correspond to data limits but xlim() already used
+ oob = scales::rescale_none,
+ trans = ifelse(diff(x.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_x_reverse() but create the problem of x-axis label disappearance with x.lim decreasing. Thus, do not use. Use xlim() below and after this
))
# end x.second.tick.positions
# y.second.tick.positions # coordinates of secondary ticks (only if y.second.tick.nb argument is non-null or if y.log argument is different from "no")
if(y.log != "no"){ # integer main ticks for log2 and log10
-tempo.scale <- (as.integer(min(y.lim, na.rm = TRUE)) - 1):(as.integer(max(y.lim, na.rm = TRUE)) + 1)
-}else{
-tempo <- if(is.null(attributes(tempo.coord$y$breaks))){tempo.coord$y$breaks}else{unlist(attributes(tempo.coord$y$breaks))}
-if(all(is.na(tempo))){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$y$breaks")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}
-if(length(unique(y.lim)) <= 1){
-tempo.cat <- paste0("ERROR IN ", function.name, "\nIT SEEMS THAT Y-AXIS VALUES HAVE A NULL RANGE: ", paste(y.lim, collapse = " "), "\nPLEASE, USE THE y.lim ARGUMENT WITH 2 DIFFERENT VALUES TO SOLVE THIS")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-tempo.scale <- fun_scale(lim = y.lim, n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks. If fact: n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) replaced by n = ifelse(is.null(y.tick.nb), 4, y.tick.nb))
-}
+ tempo.scale <- (as.integer(min(y.lim, na.rm = TRUE)) - 1):(as.integer(max(y.lim, na.rm = TRUE)) + 1)
+}else{
+ tempo <- if(is.null(attributes(tempo.coord$y$breaks))){tempo.coord$y$breaks}else{unlist(attributes(tempo.coord$y$breaks))}
+ if(all(is.na(tempo))){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nONLY NA IN tempo.coord$y$breaks")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }
+ if(length(unique(y.lim)) <= 1){
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nIT SEEMS THAT Y-AXIS VALUES HAVE A NULL RANGE: ", paste(y.lim, collapse = " "), "\nPLEASE, USE THE y.lim ARGUMENT WITH 2 DIFFERENT VALUES TO SOLVE THIS")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ tempo.scale <- fun_scale(lim = y.lim, n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) # in ggplot 3.3.0, tempo.coord$y.major_source replaced by tempo.coord$y$breaks. If fact: n = ifelse(is.null(y.tick.nb), length(tempo[ ! is.na(tempo)]), y.tick.nb)) replaced by n = ifelse(is.null(y.tick.nb), 4, y.tick.nb))
+ }
}
y.second.tick.values <- NULL
y.second.tick.pos <- NULL
if(y.log != "no"){
-tempo <- fun_inter_ticks(lim = y.lim, log = y.log)
-y.second.tick.values <- tempo$values
-y.second.tick.pos <- tempo$coordinates
-# if(vertical == TRUE){ # do not remove in case the bug is fixed
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
-geom = "segment",
-y = y.second.tick.pos,
-yend = y.second.tick.pos,
-x = if(diff(x.lim) > 0){tempo.coord$x.range[1]}else{tempo.coord$x.range[2]},
-xend = if(diff(x.lim) > 0){tempo.coord$x.range[1] + abs(diff(tempo.coord$x.range)) / 80}else{tempo.coord$x.range[2] - abs(diff(tempo.coord$x.range)) / 80}
-))
-# }else{ # not working because of the ggplot2 bug
-# assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", x = y.second.tick.pos, xend = y.second.tick.pos, y = tempo.coord$y.range[1], yend = tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80))
-# }
-coord.names <- c(coord.names, "y.second.tick.positions")
+ tempo <- fun_inter_ticks(lim = y.lim, log = y.log)
+ y.second.tick.values <- tempo$values
+ y.second.tick.pos <- tempo$coordinates
+ # if(vertical == TRUE){ # do not remove in case the bug is fixed
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
+ geom = "segment",
+ y = y.second.tick.pos,
+ yend = y.second.tick.pos,
+ x = if(diff(x.lim) > 0){tempo.coord$x.range[1]}else{tempo.coord$x.range[2]},
+ xend = if(diff(x.lim) > 0){tempo.coord$x.range[1] + abs(diff(tempo.coord$x.range)) / 80}else{tempo.coord$x.range[2] - abs(diff(tempo.coord$x.range)) / 80}
+ ))
+ # }else{ # not working because of the ggplot2 bug
+ # assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(geom = "segment", x = y.second.tick.pos, xend = y.second.tick.pos, y = tempo.coord$y.range[1], yend = tempo.coord$y.range[1] + diff(tempo.coord$y.range) / 80))
+ # }
+ coord.names <- c(coord.names, "y.second.tick.positions")
}else if(( ! is.null(y.second.tick.nb)) & y.log == "no"){
-# if(y.second.tick.nb > 0){ #inactivated because already checked before
-if(length(tempo.scale) < 2){
-tempo.cat1 <- c("y.tick.nb", "y.second.tick.nb")
-tempo.cat2 <- sapply(list(y.tick.nb, y.second.tick.nb), FUN = paste0, collapse = " ")
-tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
-tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE Y-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
-}else{
-tempo <- fun_inter_ticks(lim = y.lim, log = y.log, breaks = tempo.scale, n = y.second.tick.nb)
-}
-y.second.tick.values <- tempo$values
-y.second.tick.pos <- tempo$coordinates
-assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
-geom = "segment",
-y = y.second.tick.pos,
-yend = y.second.tick.pos,
-x = if(diff(x.lim) > 0){tempo.coord$x.range[1]}else{tempo.coord$x.range[2]},
-xend = if(diff(x.lim) > 0){tempo.coord$x.range[1] + abs(diff(tempo.coord$x.range)) / 80}else{tempo.coord$x.range[2] - abs(diff(tempo.coord$x.range)) / 80}
-))
-coord.names <- c(coord.names, "y.second.tick.positions")
+ # if(y.second.tick.nb > 0){ #inactivated because already checked before
+ if(length(tempo.scale) < 2){
+ tempo.cat1 <- c("y.tick.nb", "y.second.tick.nb")
+ tempo.cat2 <- sapply(list(y.tick.nb, y.second.tick.nb), FUN = paste0, collapse = " ")
+ tempo.sep <- sapply(mapply(" ", max(nchar(tempo.cat1)) - nchar(tempo.cat1) + 3, FUN = rep, SIMPLIFY = FALSE), FUN = paste0, collapse = "")
+ tempo.cat <- paste0("ERROR IN ", function.name, "\nTHE NUMBER OF GENERATED TICKS FOR THE Y-AXIS IS NOT CORRECT: ", length(tempo.scale), "\nUSING THESE ARGUMENT SETTINGS (NO DISPLAY MEANS NULL VALUE):\n", paste0(tempo.cat1, tempo.sep, tempo.cat2, collapse = "\n"), "\nPLEASE, TEST OTHER VALUES")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE) # == in stop() to be able to add several messages between ==
+ }else{
+ tempo <- fun_inter_ticks(lim = y.lim, log = y.log, breaks = tempo.scale, n = y.second.tick.nb)
+ }
+ y.second.tick.values <- tempo$values
+ y.second.tick.pos <- tempo$coordinates
+ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::annotate(
+ geom = "segment",
+ y = y.second.tick.pos,
+ yend = y.second.tick.pos,
+ x = if(diff(x.lim) > 0){tempo.coord$x.range[1]}else{tempo.coord$x.range[2]},
+ xend = if(diff(x.lim) > 0){tempo.coord$x.range[1] + abs(diff(tempo.coord$x.range)) / 80}else{tempo.coord$x.range[2] - abs(diff(tempo.coord$x.range)) / 80}
+ ))
+ coord.names <- c(coord.names, "y.second.tick.positions")
}
# for the ggplot2 bug with y.log, this does not work: eval(parse(text = ifelse(vertical == FALSE & y.log == "log10", "ggplot2::scale_x_continuous", "ggplot2::scale_y_continuous")))
assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::scale_y_continuous(
-breaks = tempo.scale,
-minor_breaks = y.second.tick.pos,
-labels = if(y.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(y.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(y.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 10") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)},
-expand = c(0, 0), # remove space after axis limits
-limits = sort(y.lim), # NA indicate that limits must correspond to data limits but ylim() already used
-oob = scales::rescale_none,
-trans = ifelse(diff(y.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_y_reverse() but create the problem of y-axis label disappearance with y.lim decreasing. Thus, do not use. Use ylim() below and after this
+ breaks = tempo.scale,
+ minor_breaks = y.second.tick.pos,
+ labels = if(y.log == "log10"){scales::trans_format("identity", scales::math_format(10^.x))}else if(y.log == "log2"){scales::trans_format("identity", scales::math_format(2^.x))}else if(y.log == "no"){ggplot2::waiver()}else{tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, "\nCODE INCONSISTENCY 10") ; stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)},
+ expand = c(0, 0), # remove space after axis limits
+ limits = sort(y.lim), # NA indicate that limits must correspond to data limits but ylim() already used
+ oob = scales::rescale_none,
+ trans = ifelse(diff(y.lim) < 0, "reverse", "identity") # equivalent to ggplot2::scale_y_reverse() but create the problem of y-axis label disappearance with y.lim decreasing. Thus, do not use. Use ylim() below and after this
))
# end y.second.tick.positions
assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coord_cartesian(xlim = x.lim, ylim = y.lim)) # clip = "off" to have secondary ticks outside plot region. The problem is that points out of bounds are also drawn outside the plot region. Thus, I cannot use it # at that stage, x.lim and y.lim not NULL anymore
@@ -13667,15 +13674,15 @@ assign(paste0(tempo.gg.name, tempo.gg.count <- tempo.gg.count + 1), ggplot2::coo
fin.plot <- eval(parse(text = paste(paste0(tempo.gg.name, 1:tempo.gg.count), collapse = " + ")))
grob.save <- NULL
if(plot == TRUE){
-if( ! is.null(legend.width)){ # any(unlist(legend.disp)) == TRUE removed to have empty legend space # not & any(unlist(fin.lg.disp)) == TRUE here because converted to FALSE
-grob.save <- suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot, legend.final, ncol=2, widths=c(1, legend.width))))
-}else{
-grob.save <- suppressMessages(suppressWarnings(print(fin.plot)))
-}
+ if( ! is.null(legend.width)){ # any(unlist(legend.disp)) == TRUE removed to have empty legend space # not & any(unlist(fin.lg.disp)) == TRUE here because converted to FALSE
+ grob.save <- suppressMessages(suppressWarnings(gridExtra::grid.arrange(fin.plot, legend.final, ncol=2, widths=c(1, legend.width))))
+ }else{
+ grob.save <- suppressMessages(suppressWarnings(print(fin.plot)))
+ }
}else{
-warn.count <- warn.count + 1
-tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
-warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
+ warn.count <- warn.count + 1
+ tempo.warn <- paste0("(", warn.count,") PLOT NOT SHOWN AS REQUESTED")
+ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn)))
}
# end drawing
@@ -13683,49 +13690,49 @@ warn <- paste0(ifelse(is.null(warn), tempo.warn, paste0(warn, "\n\n", tempo.warn
# output
if(warn.print == TRUE & ! is.null(warn)){
-on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
+ on.exit(warning(paste0("FROM ", function.name, ":\n\n", warn), call. = FALSE))
}
on.exit(exp = options(warning.length = ini.warning.length), add = TRUE)
if(return == TRUE){
-output <- suppressMessages(ggplot2::ggplot_build(fin.plot))
-# output$data <- output$data[-1] # yes for boxplot but not for scatter # remove the first data because corresponds to the initial empty boxplot
-if(length(output$data) != length(coord.names)){
-tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, ": length(output$data) AND length(coord.names) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
-stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
-}else{
-names(output$data) <- coord.names
-}
-if(is.null(unlist(removed.row.nb))){
-removed.row.nb <- NULL
-removed.rows <- NULL
-}else{
-for(i3 in 1:length(data1)){
-if( ! is.null(removed.row.nb[[i3]])){
-removed.row.nb[[i3]] <- sort(removed.row.nb[[i3]])
-removed.rows[[i3]] <- data1.ini[[i3]][removed.row.nb[[i3]], ]
-}
-}
-}
-tempo <- output$layout$panel_params[[1]]
-output <- list(
-data = data1,
-removed.row.nb = removed.row.nb,
-removed.rows = removed.rows,
-plot = c(output$data, x.second.tick.values = list(x.second.tick.values), y.second.tick.values = list(y.second.tick.values)),
-panel = facet.categ,
-axes = list(
-x.range = tempo$x.range,
-x.labels = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{tempo$x$scale$get_labels()}, # is.null(attributes(tempo$x$breaks)) test if it is number (TRUE) or character (FALSE)
-x.positions = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{unlist(attributes(tempo$x$breaks))},
-y.range = tempo$y.range,
-y.labels = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{tempo$y$scale$get_labels()},
-y.positions = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{unlist(attributes(tempo$y$breaks))}
-),
-warn = paste0("\n", warn, "\n\n"),
-ggplot = if(return.ggplot == TRUE){fin.plot}else{NULL}, # fin.plot plots the graph if return == TRUE
-gtable = if(return.gtable == TRUE){grob.save}else{NULL} #
-)
-return(output) # this plots the graph if return.ggplot is TRUE and if no assignment
+ output <- suppressMessages(ggplot2::ggplot_build(fin.plot))
+ # output$data <- output$data[-1] # yes for boxplot but not for scatter # remove the first data because corresponds to the initial empty boxplot
+ if(length(output$data) != length(coord.names)){
+ tempo.cat <- paste0("INTERNAL CODE ERROR IN ", function.name, ": length(output$data) AND length(coord.names) MUST BE IDENTICAL. CODE HAS TO BE MODIFIED")
+ stop(paste0("\n\n================\n\n", tempo.cat, "\n\n================\n\n", ifelse(is.null(warn), "", paste0("IN ADDITION\nWARNING", ifelse(warn.count > 1, "S", ""), ":\n\n", warn))), call. = FALSE)
+ }else{
+ names(output$data) <- coord.names
+ }
+ if(is.null(unlist(removed.row.nb))){
+ removed.row.nb <- NULL
+ removed.rows <- NULL
+ }else{
+ for(i3 in 1:length(data1)){
+ if( ! is.null(removed.row.nb[[i3]])){
+ removed.row.nb[[i3]] <- sort(removed.row.nb[[i3]])
+ removed.rows[[i3]] <- data1.ini[[i3]][removed.row.nb[[i3]], ]
+ }
+ }
+ }
+ tempo <- output$layout$panel_params[[1]]
+ output <- list(
+ data = data1,
+ removed.row.nb = removed.row.nb,
+ removed.rows = removed.rows,
+ plot = c(output$data, x.second.tick.values = list(x.second.tick.values), y.second.tick.values = list(y.second.tick.values)),
+ panel = facet.categ,
+ axes = list(
+ x.range = tempo$x.range,
+ x.labels = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{tempo$x$scale$get_labels()}, # is.null(attributes(tempo$x$breaks)) test if it is number (TRUE) or character (FALSE)
+ x.positions = if(is.null(attributes(tempo$x$breaks))){tempo$x$breaks}else{unlist(attributes(tempo$x$breaks))},
+ y.range = tempo$y.range,
+ y.labels = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{tempo$y$scale$get_labels()},
+ y.positions = if(is.null(attributes(tempo$y$breaks))){tempo$y$breaks}else{unlist(attributes(tempo$y$breaks))}
+ ),
+ warn = paste0("\n", warn, "\n\n"),
+ ggplot = if(return.ggplot == TRUE){fin.plot}else{NULL}, # fin.plot plots the graph if return == TRUE
+ gtable = if(return.gtable == TRUE){grob.save}else{NULL} #
+ )
+ return(output) # this plots the graph if return.ggplot is TRUE and if no assignment
}
# end output
# end main code
@@ -13733,3 +13740,7 @@ return(output) # this plots the graph if return.ggplot is TRUE and if no assignm
+
+
+
+
diff --git a/cute_little_R_functions.docx b/cute_little_R_functions.docx
index e0905868b371d07c5e89ecf67192f21afbdaf85e..a42532236df2c47feaf0ecd05f13141efb27063b 100755
GIT binary patch
delta 426086
zcmV)lK%c+2yB?~(9<b603em$EB{tIq00F?0+zA;`>r&%Lw(kEEao!=vy{E&DF$SAk
zL2bIjfT?0{3hHo~OHAz`5-DwKkd^MtG~nr(cRJ5-p6ukxlx)ctE(RoFEuy2#mSs7W
z`qsBDxz@`6{F6osEDyi`*MHvcy4)G`SfGf$M3aFED}Psg!5t>0{=EYl+)-HyFXucw
z?`p1~_xj6l!Hrd?ONC+&<7?Dke38x7oz+j17wS#l8t=0}f7s@tN0ojnI;%Y@ZU_Ap
zece7)-i_ybYN#){wiR85@0A83*bOhSG8zheW!DZR{3rT`c+T(IuN;Pt!~Ei^V7?v$
z4`kQtkALR0x9HXMhu!ERcaPn~UC)osyYJUGUI&On`&7mke9rL0N7?i3$REFAeeD@f
z1THc9#&d>{{LN@2J?aJHA6{r`Vv}3neAxvWQy18h>}8kO3@@=NhrO|A-1j@Ld&*w~
zAN0pp^Imtg<KT95E#RvcTq{IN9KrH+gSqI^zJC_wp0j)03Ams)z8;Z&_C!BEVx9m2
zDJlJ;+4P3HL-@B(?sqjCyXSkQwY8&9^{r3arRdNhYtw<Rra$=}J7g~MUBEwH;OY~6
z*4AIy+%ji(Y0xp;4)0gtBw^1LM?6qcpFySL>4$K@2f|~5oU(iKy=EY%k2*4W$XxKo
z{(t{Knt$a#^?7&H*UIIS{apU@0DdQN%-6$;j<I0n>H-e0hJ#xj-Ovr^x4^O&O4GAG
z-*fei!$1Ps(^Gr>h3`}6HoULA-rSe{_|4t#s$YJZ`h$;^Jt^1sl6+9J@8z+-#C^hZ
zMiF>I1U?iL$SeL39Lo$2P_etRPaRF$eSg6u6L+i>A?D!fvPF)rP7f}R&rUDMBpJpL
zu=#1D)i^W{^#k9xMSO~+|Nb}g$lr%mZgcJznF`g|+&=_I5y0_<GM_YS<eTUFTs;Aa
z@gIT%nPcuBf&;-Z_Yc9bMBu=P(D)Bw{W4hN6fHJI>zf<R!>w6Ul>Ed@QFth4Q-9=e
zpNr`3kB-*#Uo=y1Hh1YT&3oR-n1aah>CxHK{PH4X6jONC*0&CthqI<A`3YJUdDF&o
z;;zI>V4wO-DYi=_V{^l<-$m>}f0!Hp1r~DU?&@mCQ{BOhS-x%cJpb#Rr`pVS)v8!A
zlduv_f<#NCcw+E4=9`Vpef<j`4}bFm$yC>qHL>D*Hzr4A@<u=SlOt5ilDSK&2=yg@
zT6niVHFl0#BjL4Acx!FFakw!{qLZf^i4G5$D=({6m~>5JqCRupKKSiDLJY0*^Rx3^
z0#ZbLFL+8je$@GRrno8A>+_n=JLG3yeO@7Q(1#D@U#JK)*kxJLe%~LK{aVC7_FGhR
zs<60A_DF}RDq+E$ekq5t7PJNJv2vwKyr9d3r}Xio=E+5?5`95aSU^YbXfUi)$j`Lb
z|4g3$OIbg-_x5$-Gg)Ym&V$jsNr5TZIaoV7Twj7IlV1)Ue<y-ycChK|eZlWMm$^04
zdeB_rGtwRe*OEDd`2w_DSgDaC?;!(=Fq>zu34z@)F}&C0z@D}fH+I%{j+$^sG5MGu
zBTAlb?8NYpx$?4lelw@%W88%a5E2I-0E;1GSCeF!{K4291K%${PMA5r<LMP9O@B3~
z%beQ}eP7d0f3i~f{b`wAg5dw#dUGEJxAXk7BdtifjH({(KPP(Gs&B4s?|&e1jUqWW
z9`}MaUulb4z|^WE7!xZllPlrP%N4rPr$TvfbfqJ>l={OSRl;-h3w>>Y;_))^z$toi
ziarv9XtL1P2OHmk0?A!t?f2dGwZj94v^9^Ma|sb1e}}v5*B1+m<(J0c=JxD3;)~%?
zudka6AwIIzFpn#g!=uLjwzcs;mO_z=I=5U~y=|Z0n^m#j227gL;nn)~=4NB3zP;LG
z0<gEj9z6Y#exc=MrS17zqYR+XF=eIlu|%S%t7SPCiTSH#_xJZTv2x?{PKOC;WTLTF
z-&tMTe^~v8w!T(m-}lrC4O~;|4{`69JluP#3vU(YzDzE&p;(|;XtCH?JKXw69(5!Z
z!WE+CaqV-^0Mtg`7puMKB?A+x)>S?KX~=KST6vu>2WuOR^|gb99E6AF$N}b3<U$O<
zC6|51+$yZh6LG@=rwi5lWcR7uqVePXSR$hMe?1khbg6=I>icCzdx#ZJu1IZP8KvPV
zxiZwlS}_P;X|x^YQpee*H#bkq&g-Mm1<hF8-rhbKk;v1WlQ`YXTj<giY~Dg6Zerfn
zw-2DJ(e%6}Psa@W<h<2)0CBVC?OFMP?=&UvxeyfIalS`CUxz*M_wg$Be>uClJUKpX
ze?6{$KlW22Pd&D$wJav~E=F)$t&Q#G*s{lYo1VAL_4>}1=~*>riemH8zfeP;MZa_T
zt!{r@(Hj~58+|6cPFKN2eMViTpM}fK+dkKKytB5s6^<|Z>dZ;J>@w)0)$;-_zQX*$
z6&RJODLoqhuoWH~P4vufI#R_B?|ZHnf6jL_=rGwM;NJ!3ZSy*j5kd!wM~@&n)@V#W
zl<3&>Ly3+5<<8b@uuKg8C9oMYSKD_Fx)g3Se04eO>xsHy8tc)SS&YjIq!O3xVIH$y
zo*j@MUGn3f<cA`qOHT<_|6KoUUNk;8N=6}$E|T4S%$Z-w`czQG;#1t!p9tRze{Mkv
zw1djE@LXSVXM25XeS2fal8`4A7eoC!Rwq6p&z1*wHW~+O$>NxVD4%HG(EPB{(`1Gd
z7lXLIp446B?B*ZLQG^E6A4)G-q%wPGu~EcD?VjCy^<0+)$zP;Ehi`VdX9NAUKlBuG
zzc;_M&Ra(${%8Tp^8ow0@x1r^f6W5&>mU6u>u-&wIm2gxJ0AIJOa@=9Gn@NcpOOLb
zNu3$$hUY81ISOq;MyGYsv{re7T0Uh<EswiAk2hU=`++UGE(?qICXKCJF?N62<Bp4u
z83%EEDrJ=@{grB%gIy&|IkhU`ZNd$GBW*8E+n<oOk+w6W?H#1;m(aFhe`s4wVF~(<
z39n5|p{CCr>XSPvJZOzA2^GwgsF?c-6!8MWwGSzvatkSQamw65%A5dYeG@q|HO|~z
zGdF~s`Fc|qa=&w6>D~xVUE(t72(J&hf?(RxAhd3gsTX8wEhxy;lT7^ynL0(Lu4l{C
zV^RjT>oV$o5OU4|DZ^|5e-CSmnZOzv;ewcTAvk#Bbx7k7!w;`!=sGg=1nwFbe@K&|
z8+l@AMZILg@B5GC12Z(t7BcjUerN%PZXiQXGBidz(q!oMJTWwsC|%dljDG$q9&Fz#
zJZXa(>YkJOAu$>KOarWZr`eecStV^6zR({l6Tg16TM&}=qolo#f22K0+FMB4DU!CW
zoXitxH;}YnLE4QtX$yWI_K9@3PhhiH&D*ZmV*%`YE^U4W()Z%@y@B*Sfynhb(szpV
zy_q}u_Vr+pz!xWQm_H1IEKbRrNXZiz+gL+NPLYyLfiNb0ypEbpkex}Dhv>B)hPI!3
zH(o$Ze<+jm5p_}4e-|4tduY^Kjlix9f~kQB2or(}ZHmL8j-dT6hD%;S?psLiljPo5
zLUPA~&vz>Ld_r~Qwd$|OU9asqhJKMf7iLea{F`LYEi{eiQ`;12z1lR!(!Vz{jUkkK
z?U>g6xhVjo#I6y7Ev$zosH2!5nrL8WBx4dA8XxMH!|*Wif7<b)7aG{I7keEf`h|%e
zYXy<$GbMT~7JMV3ciY#3Y2M5M%bRg9A84DJHAZ4~TO*2-F1>@k@&2F}m{mqXHQ|%r
z2GsYIHrn-vVVDH8z@6u?l{+RRMTTFP;jy%Fis5nLF-3;2n;Sxge?7wsJ_vF_?=TJ^
z_h8Vw(MGs7f6ZOycs=SzCOXpfqI6xy`o<}`-axuek*+hYZ(JU_?op*FTU)hx2)yj9
z{lEW99G^S4_nu_LY<KoR%<wQ$K00FozQ$_DDZa+E>@@j0<7&s{;%jZuHq3%+E)xEN
zJAE3VU}rPoIT34n0R&Rm1C6Y1yF_e1MJivI%CRhRe~QYtk;+q~@{G$Omy61IKL;F6
zrDVMu{Y?qmrfaH>_8BIy3m)P68fbvOj<vBaa+0;TaRoC?*3MO3Bru;B+<g$P&c#}#
zNxp`LBOch>lGzMx`P%@Rt)fx=N_xh+$0>ToZcu5`bFS(hfu7r*j~$_&J$*1+hMifj
zuZE;cfBiPmVrd#eX_g-LdJHUT<m)%$Yb=VK;%i)gOp~v%DDn-68+47nW^XVmi|h;D
z(H~3ij`_nZ3H?SedJ9Q^Q4Y9p8D?-VPS{xJI7Qgl5Iar6&bSnCc?etjo|}hUkuW4e
zp_IKpuo)uSuyz@Ng7t{E2h!$u!>Hlay%)Ire;$c^VIs$x%Sj^tw1Y&RB9UXw<r@*X
z!(ARUD2=#0%H%M|p$dk=b0pO9+>k~yWP7P-;8Nj|4hxv_92g}-8ef>kv7&O4#&^&j
zPm{*6qVh#FK4$Fewl-neDD|)7-7#rG&z85tY8NO>>~rZEtJt$EhxP>C*5^1r<H{SK
ze}WDA7N=4yU7V!S`r0zqE@Bg>Oqw_~%;U;a6Q?iG_Re!8>Cu387_`kp`dpMg8(2y>
zO`o`yH=o<4SnIW>Xaj2r-zcJR@42d*gV3-I@)9!~;<7%oh2Yxwb$rNmmJr?O`y6R|
zaoWa;LZt0%X&Xxl--x!kCdeGHHVlJMfAZz(81_@_yKUQF+H-Kx?zLo%HG|V+je9Ak
zN!D02_(o*i7akYaV7x|@Et$%awhc)`vNP?G8`kE68Pi&U#a(8CN=E2|Ho5gD1n5VD
zgW;9|i*q?v5>9b>9ZLz*<noM537069&Y^xDQt2e~VON5{a~SdI4fDfQM2)wuf40fB
zf-!6!E-k{oUY>I+O`e_p@4>^>!^P<wD-WmX9OEcy(s{<Ehs#6fUZ9w`Hr$v4Qg>O~
z3n1x(3m10!vYUEI=5R=)8(I!W;$E1zu>^6NxG@ttP2$eD1~FMP-se)8u*3%;>|B`L
z%7(bZ=m~AZ1~cQ}i3yts<}&GZe*%o3yb*6>CE^rsH?Y@7n!Jsrh;L+YJ3g1W*e?vm
zL1dpfu&d+5vL)Ksi*q#=8%}Zc`ZD&d#?<O;QmZqrrJPQ!z8IDHg*nzpo(q#FmiSGR
zCw7XS&umki^=d;GYy95G&^2A<t_>&Vz}$rsAcNZVWZ$Pl6Pd7m@5~61f3p|oY^)_L
z24|zen~A}@ZXQ?48fn(;On`jK4(@c&0)91{u48H8G@IfQ-F&K>65Q8X!0Q>87ABYI
zI-To2^_d!P_1EdZYEy%_cDBPUn!=NicorreERI9s$(DFBE{>Zbo_`GFs9+&t9XJLH
zCbo4Wq?LA|R=|@s3DpZaf7N95t~R55q@~4a3CqByXlY{^JJsZumaq)$F~@Lr*DuYU
zDGw|@f(}#HKI>@sj_8{y!_@XD@U|s%!f*b;9EA?+;slMgUDE`Ob-HO1G}d;#h@j^$
z*?_any+B<HI<|bX<nhqxhT$+lq_(>CzlOJcI>fy#UcuT}!Zpp>f7nbnP1eQ|uD4_D
z8~zaQu8;%D4o5`(L~f{a+Yx-A&2E=#@Of<_*lz<*Fc<lIVgANSt||WB#IQ)3{EfTG
zy&ZptmTivsJ8o<LiD(osNy>vhq)=@Uey4#!htQHzQQse8Y1iUJj-_2l<k=EAmUg`z
zk#`wYJ=)I^k%uGFe<W^M+jB{e%3CzU--Nxf*bCV^TlU7S<=&3Hv1Fkjw2ftEQ?$Lc
zwuH2eB@5Y<EM#0nH+TCZqQ5Ap1G#(1o(r=l7JN;!XJZN3Gq*<XhPfd$dS7q!D&h6=
z(DyMkarM^*&~{z$9--v70S%O=sPgWZ{z!QAMqeCu358)3f9g1<O=QOy$)<CpOPqmP
z;q5jW#|zW@23DXUy=P1BxVhaM(Ys_F*y+<0tK_12q#2H;cI|Bu$L9mrl*lAj8-d@$
zKERN%ZE{AeLI}7%AfoRLlpdlzCw`kq{0kF5Rsc>B|2A$44R0|j@&-|njl9%1jx^bE
zug?W;D7rXfe`0asBx5$#uwB`FI-4S@SDVUMocKnjvOad3hqk<3=9uJ+a{?ox9VQ8d
z?df2!U^;8-*fX?Eyg+ed{@!zO3t<oz+<AfrU<xCrFU;v!&xo9!EvI8W<6Ck1P<46!
zsND{0Cch(ZFU;Fm(KyZ9%UIEf(Ti-N7a7-UPDd}ke+z8^lU34s0CN}&V9gI2Mv|0G
z>fFz?*Z)iuEE-CzP+6Q<vD9#iSnHVMJfHQZ`0UjNF_s#>kwGjNP-PBm;lAN+O(&*H
zo**t&H0;G>;~IRR`U6GU9yZT=1$AR_AyRj?)SYp0;Y3eQrmlNTO6_%j5Yo;8RSTvD
zA|Ol%e=b(<sR%S>ui8wEa4yXz%~YTA00WoM7`3drR1rrXykT%$YaQmI5xzK+W1-<R
zlW*jd$&WtOw>}x-iwWe~(RzLDXcoOEPyZe}WUd@R#w8HgXe*%NEnD@cDMbrocwoK;
z*ef0If$+4zp0fK9`%fKpWb%-?APAWL|ACxPe+TvO2jd|Kdu&=o*#G5cWY2!kPd?h$
zJjKl|s~e+-2=DXq44>HkBYL#YNB_a6#>SRe*H~!Y-f2qSa4KT#WAC`Wv%a~}jPp+p
zNB^XAJMVAGAj$oG{&IA6dT@Dsc6u>SYb?WaBR_q-2lE8f5;ugTJa;NzPRbic8}<4=
zf6m)W=j~{1{nJr%;dz^I-aRE)-K+wQb;%9qw>>IuFBBC@pXs^aJL?w#1&@8?7i{c=
z(>&`7Cf#)W1%2+P#%A6A)a-AVWU|R`G<G(^M>%sAj0`_diOf_Q>X$Pc_M-vyhtiv@
z+=|B8GHA0C5zZA7!K(p{Q}j%kueTZ-e>;mCurp-7@eG+aKR6}x;T>A>dEw$j?D;Id
zj9@ZZ=7pxrcf({C$EQ+OiPGLxmG~^^sBV=o<<zPWGMc^j!nD8}{y{XqA)2?e8dx)I
zkMlNX-nO?I>pNrP$j;kgYx5{}Fg9<=)6Kkvhq9lyaK=3I|3q@h+S>lwW@r#Be>9;<
z(ioE&{KE9Seel~e&I(c7<;3@-B9`OM{yRZn2qpy6^TBjrASA9`b~EU}v%6ND`*P*3
zUfZahBeoD*3u0@ISr`J&wEkdbkC#dFEe@@N`u1A=XlZAovDR4AhS2uhIaoHnDNDwn
zLkhDNysWw0d8i=Eywj8ymu1)kf9>_Pzbm_pDUGkBgdxBXV9z&^td&iDpFr&+4Tg*G
z6(R?z$x_p#u3y$WUHQ4AJ{QKaaePJv^nvSe((H5J=TB2a{b=i8+lcnFG*0q#<c%dj
z%}>Nii4ruB?Z@|%Y&9FL)_dgzq<Ys=x48uO#N2T^1aNSo;@YQKU}k)Le}AjqY|LEY
zO@5NE`1o7S_#iqU^{D+YL3lBf;$VHfbr5ELFWEg?5;^1~;EMFP>$Qg(f}Vr`i|I10
zQE44+-&4wG*XP%xPqZc5<Kodo^w!SiW^-q0XRbyrm~G%^L>7vZo|e}M(G2CazOC8$
znerMQGFQ%2UK7tD@dOU^fBbJ8!Z%Iz@Bw_bSE7!?0yV1rea?efOk%_7T+X`&fQ=Th
z$<My}yfPBka`yPE*TenHmlA-_OrBjb06sH$6N(_K&j^(v1QCMo1VNLqL-xq&)yWA$
z5uunN6d`Ghw-Jko#q6=DT|0(IL?mX9M4Lp5SVSylkHx;=eT~GUf9A<W3!#Wm%+ht{
zgBzdKsH}N`VjU)G)c5&4a}l42&m7>B_aVH9o;89o2Vm6HSPKEE>9S^euOUq$GP6f!
zKx>Y^OX>v(NCada0LePFJG&|A9=SZfLOdcKvvs+lZz^&lVljIxg56ky<^<@~ARN<#
zBc$)dftNk(a*^bTf2ez3)Kz{j7QYn}X<6SjAnYg7JpPsgux+o^5p2s3Hro?GK^}Iz
zUz16CfSSv2&6&6DhU^e7q@pvJ+Gi3$iXeSAkba|H;19__G8gLnZ1y(bF@mBrEyWb{
zrH|6NldZ)No(Ru(=1}TO9%AqoP@oZM=n@*>_n2ZrldC6*f2W8(Cb-$Rj0eAxV{;5>
zm)MnTax+jQ81!!RSIE7mx?I0^nd9}S4_8yaBD4|O=|KBf(`vMQ9YuQduLguFZ5mK$
zgl4kF_2>+O^}oh+(Eo;(LqyS}H_&I_FcR=!B^dZVf*3)ZI*2s^NB9eY95Z{3!E4|f
zVbuaYW(SA6f7-xtz_{#dT5fyH%>kG(I#38O|D}0;dVKmNxsw6XCnD}$5U0seowLAY
zg1O&mhwTeSZ_O70i<jC38`ZQ!PeFZo*!xSf?7i@mVx|V(Q?Si>=%`f*yMx2+=`%Gu
z2M*Wvvp%mzj`YpKepfVbSk*Kww4JBJjHNbp6c^=sf2A5)AbX@_kCnpcg+AbPwe9U_
z)4NpBhP3aKa6dC+fNM*;S2d&1BNE1)3^$>KJ26cqrehve&HfJ6j;K~6(dnU1u)~N3
zG1euqQIN9z)NcDe_Y2zgXTozzl^Qu~LwkWy1MP9=Y<Lx3fCYD+-~kM;1hYu}dpeY4
z4H*I%e<B^7bZn!8q1;4Uxyrb|0H2Sj!pLF>(~(FT@zyAbWE0_w@J$E47h3Nya!#Nh
z)E?QWt<}~-ty1n$CA^1nr4r&))2%&JA|)yW9hR(n06kzvpHQC%+L@4AE3~;_HS(3;
zgN~^}1as^VkMvgby-{eh4Q{kK7ztB(O!pJNe-COZG4zTV(Cl9dzmoVNb8N=%%)30g
zIw+Uv$Yt~V%hk8m=_UF)2)UFX2gd&H$PeJgq(Sr{`qDw4(Y(84<S-JcRejK@k(OuH
zmB6eBSK_UC0%nQSs3PV4fg;f19X<;;>elbWZgV&Mj#VIDS-NA^*@V}fPT%MH9!S}%
ze=*pV$!^Is_AH&UK1Ulk;hCk1xkf&+?ocF(AriI|%=2Wzo;wx_1S^6yRj@+%?s-s!
zGh^%xei*DvS$Ync7}*v9oHBrIlzA7jPnhr=atGyaL?gj-r~;D$u$xQ?v{<c?Q?7`^
z^$nY~0toOUbP>AgK^Ic0c8SkB8mfMce>8=)g%e1Ih@84-EFfzHf@F149%o`zeuX%;
zt@8|*bWFmbiRM*DiC7WFK^nr`0oc?OfsQ~=0qFB~HksQ4CngdjbO)R;Az*zKmR|t=
z<JC^5%giE7k4Z^840wDLM%JfGX2T+3f5tsh3V+YIlwq7Xb`%y^J>QC=VAS{2e-JJc
z`@S9d#Mh%0P?HK7im^$DOo-XiJ^kmjQlI8NureLx50`sx7LyFprFV?{lL-#%G1}@L
z_?(ml>roGq8o-F5#P#kx2^yeAE}?;ckI_J5)Z{0GO~UI2$WC_*-m+H-cMustSbr9?
zfGFfM7%4EiW4?%n-34k<YR|5Te?tE8k981*wG*2T%t8F%v@&#XVZRNiZ&D%jX;rdT
zslbmLbp`noqa+0oo$)G=+4s+^`Gh0T@~qtrJJvAwxa@KUIdahsw{UipdmIU&nHk7n
zpna`AAM|Gmj6{3MJgg6)CM<->7bbP?Y7vec=k}EMEkU!4@E&tzU{^Vie~&(w(gWz*
z%SDzV03SY{NdF04a_#Sg!T@d%<c^M5zzBxE-@ehR#z`7l8lu|lv=oFgf-;hhjLhiw
zHOnLfVTc_1FjYFrfi}uitB+B}KZk6I*~xKXDnu9Li}{k7UkLk^NZvDb0nhlUXJ5P%
z81>~P61*|)fC84DfQ{Rhf5qhLfX+ZL7>B04*=Z9ixDcOd;d9i*Y$7_QKhTEI-BgH-
z4pf&5@1M;5mHfr7B=1@<TeLw8z!*@PYOV4Bf5`@isitW|-W1ZJB{Q)G>qzLc`%Gx7
zs;3D%GOjG)J7!sAG#)z+@Wsp^qaJMWFoZVnygO==O`$OO9x_bBf6#={Hd-|Vgu1Q=
zN})>1^-BCoVGAu2<ul)pLVe~sv4SATwhjy0#+Alk-+_N*lUq%}=e;K(ykQergD|)f
zMo{#xH!4rvtgTB5eJ&UXIE*t!Y|Y8(L|Yo>!y6q5XS87()O!=7jM8dH5uGUKR$rlX
zw*OzPgG=)5>f(~@f49iD=3$GR9g)+sQ*zq+(!4zWbBp}Bd2-ddAp2LB<oJR#^-vBj
z&(2FoVMt-=P?%|a7h|vQrV$>%JQk(=k1%aKqNNyl0W;hTNBk9$L=@7|qT~6V(sTl$
z4QAA|;3Fe-N`Ux4^#{t5#Q-98vFelx)U0^!IM-c{H#%ewf3(-verA}KgYtv&laBn1
zn)OZRCQPpFn$Bv8mE9-{(ztR~=RssKs@~GN((8aK2{+(PTD*~-A1Rzrh7kLR{dBM&
zwF`V1^hcFSkwF%wNE*&huzeVHpWl@xu_bWDNKv8!J<PGtXO21>n(`=ZlWGGtzPELx
ze@4Qjby{pJe>spTtjj2bOr5tbug*`A!x3-k;H^n~4?4}*@EMr&*<-KK*!qOnTYl_C
zZe@JY#JGvD$`=fyVrA`IDs3z?HwTejgy!-?^YOkM|Do7}+8swnl7_B^D)s!(IEAoe
z*a{=`)#%MAN|~EVe1vpC*fz{2w;tPF<QYYQ1?wJ-f1(SaQx<pJB7ZFkjT;Q#dD@Sv
zhE2PGx*=YV{t}`(HGk><Tpgdc4$1fC!GAZuw6LBXp_dZ$PI-`rU@H<`v54;I>h$39
z`0Vt89G#ziBbQ%W#2F|S5g_s(A`<xz(Y9=8YyGwP?fVJR8v-sRz`??NSfLbls;lp=
zSIFn-e}1WOX4PiWjCL~}wS^12!i*cIjyjMclo875KpC1Bb(QL)wnoIIgSf-i{?(T+
z$ERPA<PdV{KrU3qkT9>6bfkYxVEXf@C_Z{0D&z27*mY&}9vWAS^1?qqT3sBSGup|8
zlxRu6pHUkYJ9}K%5<fV(C%PHQ*r<S!Ol%Eof31;x?b(VHUkrI0<?#Xf5OxT=bYK^7
zn-ywGp<?NuqkQOz3L(V_z2lxsSa3)5sD01X@IbsGUem!V)6X2K5}}s?^aT5D;6V+4
zGUXoW4>6c3{UHt!huL$X=~|{AJQr>r_OC|V6^Q>3dx*XCu{R10Mwz>9Pnc#<wt?YD
ze=P^>ArBj|AS&M(m5F=L2$f#DO#VPTS%V_oa)i)I<@W>t3n#5}I;{Rgj%|BGWujg-
z;&o4l@ZM$Ax6N!||GOc%1z{TPF9XrAXm)cT+bOQW6;P(LlVz%f_=ZZhyK6U1%20~)
zz-}`ot}uX@O=1<(bv(9RbmjLYZ#Di5f71-)2KLRi`yK?eCVFAVYkoU2(5y(EHccyx
zQdlFKgWh0iRt#qKm<q|}mdR~41X6jCsa&)6ANELn!Yt2*Z9c7UBHJt(zGj;WiuL-u
z=Flkbx=*EIYvoew{QT^k9G}KpGVNw*MC5^>pzv!H;fc0$9(5aSl7p|!^X9>2f9st5
zxdlx{q4(%X>-5XzS5n_4rK+9Xkqcfm$q}?KO_Dm%W?BJ7Kf^~_JcC0j-Y-WY{v-Gi
z>`nZ^F^$s^D^$t+(<-pBnD5u3e99zxQ=QaY#*wrS<^`}NO<(&RiklXAurUXuN4YMM
z5JRGVlr}Zq%x!P4fc{1?X&_sDe>NMr|Ax&d2n3pfVHu*vG!`?)Kmw#gik{}ZA>>7#
zHf=oE&QlTfH|p)I1Qkv97l`fY+1Oxj{q>;r{nA#iG*1slg(L^f)2wGHS0>EY=AT=@
zaCXOpf&=JB@csO3|D^S;d3k&=#?tY{$Ro@{c`>?g=)R%*_6FRC?i<oTf8JHnzq1Eg
z>RBcTv@C!IA%Q%sOQlC^Ma)w~=k|lj=WEubCcKt(Coj)wpiivT5y5&KDhv`k)Q`5C
zzvtrCh6Zi#W=VYScDxf0nuy4$=qY|6wY;pJA{y}MDW(Io6l;wTXb3bcYFqMU_^`_h
zmRlk+bB0VRI^mZ0P%>!ne_}WPc}5}n`ugTt!#HNo!eH`rgTe4n4ltN_Ca8u}#9)sG
zE>$%RL>b1<oMdrHHrPX7qX5_|L0-)>O6HL15N)qV+cR=*UI?_Et<BB)76R=lpe-+^
zOi(n`n@pcp?o%OI8FnEhd!v;or`+?T)DOHI751!^V|q8aU{I-&f6>|cb*WO(|GhnS
zb4`-PbV>36=rEbz;2QhB%nRdB<n=>4mq=zSgb!4I@QfJ3(%I0(64X3F^2rp44COd|
zR{OF)$!c2dM$75p5`EI^SXTH<&BSc@{0NTDPEOALa(w!w;Q2AL=O@aO(J(3mY-X7S
zkk>cWb!7Q&R1SIte|^MhKHNNbZPl=`lij|jpbx!5Ok1nER)`m-h^sV1H7?z4&#~(Z
zih-}qi%c^+^S}ri_2l%79JY>{S0|UYjL{T)h8ug6v(wh@kD!<cnJq3kForS)&2%KM
zI4bn-Mu+}S;c3H(MQ(ot#qcU;m=Sr1PvesD(jv^)Onc4(f0qg7ey1I_FBrWwU&Q?N
zR0H4dnZJ&!{J4PidR2sJ#RULbe$t9xlfa2ML>#^a4mAb`J$+r+kGx045I68k>0u<7
z*NOtc>E+d*h}7e9Sh;ktyLzRqMY;NgGJo}uI|i8gt*5TFX(*w2+O^`>1Ibohb~ESz
zD)a&F%j-55f7jyLWp^ug_1g9-yiA;64%g0074msTu9Z~c`i4r+xz^%f%l+Y6m5R1n
zsiV9*s=^(5d@s;es|M-7?AbG}v<55Id~qaEQeT#;0dhHVdAc?Q@;SmVOBhP#aQ)Z?
z9|U&WvJioWH%&LpzF?YvX+AnsAS{ak%N`AAhZ%0vf1s3o=6KY%J24f_;YvIfj4-7L
zLt|yu>_(2yub6vx?4`K;f$8~KVXt(0b#hXwCg4^$MKE`DVq$APJymvdJ~L@Ga;bei
zcu6Q|4n^)g--i_p*sE0$Upt7RSaO^-q6;g>b24h8O(Z8WYY>Ns!&GtT1&Vc;KpY|t
zGsR&rf9NscIfcU^h8S{%!=FgW;k~}k9*_}{5!3SY#(T0O`62mbNq#-5gjYQ7Lm=i2
zh*StVL>wXxGsYn-fY#aqHgZESA{es;qwh(DFhm$;48ykeRe142h6u#G0Z~Cg65<eX
zm@y971H6TdX9OdHF=H^AO{Wot2*Ye)D2Kfpe_p)&1_Wa6yp^!z2cd{i%ovJ;0E*cV
zhzP{29hTtE6Fh+G3PdBKF=I4#I(?t(-$EoJ67zt>>zje+BQYWz(@Fq?tV8PidC304
zvc`ADTh@0C2zy2g$7KRqF^Me?K(mb)sp!;hsART;cA|o|aBTm`PD2D@_p;nJTMp@e
ze+L0{yBW$_6vHaRn?4nKOlA12QPF?U+{U$?$@F=)AI_*{qZ|m+vJna&?K%4Ph{J_|
zvdEYpSS(v7YvCv?CCn-RF=f11a)Bt!6oqQoXS|ItL>OiY!yb2ww}Ou#%ohlK7Iah>
zVTdry6o!4l`v^h=VWuE-`QXN9H7aXif4htt^?iQNTtp@!GgD;pK7@tPb4DCyi$mB*
zC&Z%W$Qo2lA|?@&nPM`aHAmkj^@^AJiBQZL6j`TsXSRw#C?XUycBs7qCVe6bbLLS)
zTa7|isv#1yN=AW7Ad9&iAP)2F3V=AwvqPYz0<>_$<MBQf&?yM>?B?d#b(kuLe|dA5
zCWoryqnOQ*M7z6dx4ba_ub}mefXoRXZSou<5)qjfL_!`N;t}zf7d+aOIz%KQGB1dP
z>^j6F;xR9HoSSHeAVrYo1*Bn~6G9RpnHNaTNWVjPB0Tc~&v7Omq7qS=8&r;x@(`7X
z%KW(G+Vw#|A|P|)lSip~2ucKHe=4BVuv6zOaG7B4ciLh5g3(*^MU_Bil+RbNG%JK9
z=JO%YmOr1*X7Z6g5J~%ee^~bIKe1fDDml73C3~dwM=$tesY*0CdXD|55`9K}A7P8I
zO$D}IyG(-+p_R(-Xe|~5c6oALLVBdC9`a`bg#}e5qZX`xH*fz3B@-p{fBiwHg|HC{
zVQ9jjEJmhVD2S1j*4HjCTdNX1Fm2{Z21Xqz<-u@ETTW3vQv;%8{iJyI>^~ZLRnnsm
z<-qAufkGP@6j@J1R#A@%Ss@4ogo1_w`4d@>be3TKD!DlR_Wk7e=x_S{QS;=YHHtgT
z(rBpAP@!ev3Wi$FLb6<Hf1RJ7os;9!_;v|e<9Ti7OIG$|Eg1Ozcy*zwfm<DYS%2^1
z_^<^ka&!iqvRiORbM_->p8kzA&%a!KYn@(_)?W`=-!BO)D3Whi7nkH~^Up2PB>T;a
z;{$SV(!99%5qv*C+dpZ2YhLO@(Kp8r$1eZaLW(F7s_D2we(tExf26LT!$k?67d#Qt
zJUt}(nr8>c&CAxI{ehQUs&%19(1Poc;|p`i-5=0PEQG@NJ5@3=sG^tP`y|Ove^qPJ
z3_%jYBjcT}0<*b^KZG*u)26Mp29cxXA}R-rN~8!rsd&E}@%WG6N3b{X2ggb=39f?F
z&N)rf0qogU8yP;?e@S*OyDaSO5icNS%A>uA=#6ihEKt`alHD;uebQr6(higQb9K*H
zpud*`8o<@60(f)(E*8+TR{z-mEgwb?)6cv|15I?$Lx%7+k$vWPn*U5Nm`Q{pLNOgE
zCSj;8a4I?jcv~<GCY}lE-K;IfFm=Rwg5#9q3i%fqAGeuJf5~HE+nMA+)A=KLC~Urv
zJZS3QlZWP)ASch8RdSa+BUbe^@xi$8WO6u=`=2~E4l-4K$LIk@58{H1QG=}LDxRR3
zp-|?4&VO;riEhrLtz5^uX+|3wj`x&{w)J{+XQertiX``SNVe7=bvGaP`mOxknXOr>
z(~2d2DAFtOe`8jy#E(Q;C4S5*mH3fIbV~fRnJOiILR_OHj>i;AawJBd#1F;PNfEQa
zkTEl5Ok9jhiIh*zW-q(Jl2}MF<e+()^(^HoW^Gm@wNd%!=E+s-qSyeMJw;Eg&c+v0
z<4dxSETh_!aitV}o&3r};WRj|l;TP$Mr<)+8_@x-f0PygXn9>JE$$K(;xISMq*o^=
zxKLUo6lQ(bfUr^Nt__(v$?}_QfQN2hQG-9xnhk;wL6{u~A6qa)WFj)NL+0$&YQ!XB
zG6hWjlE{ic&>`s30bMNl20@3QO9gZ`$14YL!yrrn;D*gJ5op<Y6*dIu^DqQBvlT2A
zov7(be}}0m>{hN{CSaSvY-S{`awuy-ThJbx$dV}@b5)5KbeZtfRBxUzs}iXVsVzHd
zi#H1fhO0m#%;ou23+tTb=l(%1G*&uc)?eBd$Rm0Ch0?NL_9tZ(>JGxPg@T$~yT!J%
z=B3I{2LFo#h_D(!4titlUL^soi5>}6{J1!vf8a*?<=56lYg9^Mn%7-^J%);ju(-l>
zvO8#=o}OKj{g%l@n$d+$`+&BiG|ct?OqWv>oTRZ}0!^p)4Le}LL`sSk3nq{Ta<A2~
zV4?s(%SWVWM!`gZlhqN4*&)$o)c4>l2v%x0#M4F$q7TuR4*C?+#<#0&J)hqr>JWA5
zf1nPuFY;dc&mtM(DGCB716z-6l>QTOxT}5QE>)y2xMGgJZOK$kGUjT(S$oiWaHQk&
z8|^Ez;LZ~~0E-c!i_m?4=tks;AbrN;N<1{w@M(fH0kmbm|L2Ba@2LpD;_(7cc@#>X
zP##epQ696BM^L0imr3-3v-5(v{s}=$e-@N}%-A2L0Km)ot^r|QyG;HtN{fxP_WnHw
z2*h-c<wxM$#C8qK<o=>#a6<M%{e=8Xd;QOZYTVlMOb7aj(@L=CHgyyeCEWc3^bAZ_
z#l%_pK!pi~D<+eW-;x{V(1B!jN3HT{v!a=Sl)`vy4$4Ivjjd0qGD5MKAWsZ{f4%NA
ze*_~YEou_>sQv#p*lBwo_SwETKK*jiA_rfa=gou5);VdNw9F~6E5Pa5B|ytIWiK|O
z#^6aNQfHs6p3!_%a*#iry86WY=@b!$`O~Qd3qN@NH2T#AP{0MtpGKe|&=6?L9;=T0
z`>;u9C5!<^hwUDe69>axC?}4Z8+|ClV~{t~2bFG2orx2p?J0{G5J$WH{b7yalMp!>
ze}m!LZ89`cB|sci!x&YfABCld@lKIZsOaK*>)`n4xOGU%$F)|~l<bou{ROkNZPxRh
z(L7HS(WCeC&U0QkGQ}VW20_qeK$ii1Qv_P7KuhLPO}MA=k3tvqU$H}xgBm|(-P3=K
zJX6#034KsBo6rYcQU(E^bB3!q?rS!3e=VUoMb#u67<MqGX_Xu&+}5!}dVDeVMSlb{
zPl;bRx{etN{K%?))^<SAJ3B1<+Kgc4Z)N>B(ZxlR16^Enai2A*=;9&`d{DaqU0eiO
zb`%*FYD^Ykj5js2RTxnTLF5r93OgoFuAg9`#_Y*{RrN*@x6wp<G%KiL>>y1We*hfz
zF^f(+_Ga-}&%4cUIXgZ1Th<eX-Z@%J=$)f?{;Www?;L61gKB2<&Jk#7EkMUCz&W|h
zFU^m}1!pvq&}l4~(>UHJ>*6oRmj_>4hvf1sq?6;L<QlbEg`QZZZl2qf<gBL`T})K@
z=whOa`K(Dr7ZYjVgX(eIC<}p>f0m0mmB>8dA*MSixm~E$dZR~&MnRO?T=WJ$EfJ~(
zrgTUZ(=R0`m9A2Kxx2b5Sipi|yRHA}beO0)-PM3ACRbgpTzuY#X8o%$kO63z)$>*D
z$2ca1bJaYALfM-r_Zi7Oq#oF$;6Hl7ABj^g-7!aT0oyG{IR^Tr7bvDde+(~!L5~T~
zftL>N{eSkpy*q6r+xM$flj%8PwL^oSv3ogpVOw@MW8)l*(`Rmy<y97it%;BrNjUMH
zneTpUSCxbW0%OPMx${Ri76MdNzx{rVE%rma3>$LtKStDdO>vpJE`5X-qH);J<z~A0
zX4CWG|3e!5mJjL|ykI|of8ab$2ja)n2fWA8ts|Z!pb~gJvDt%szQ_O!`19^=Y&m?t
z!k0I<6VnoJfm)oLu%g^(L9PmNz(fwHDx+nu)_dl%FhRu`svH%0{93(EJkvH9>1{Wf
zjV3l@Oxn<DqwHC^Pu!xlxda7Ut=lG=8f!;yod;IME_~>e$%<&vf1#Zwi<no2D-OT;
zM6hXplPjHkYM)>02VZOyxAR-~MQOKRIF(g;f0oTh{1PbsuuES8)nERhlz;h$QvcN-
zfdL@@lsce%EsOy9a%u+1zrhfYucRp;UkhVEzMPr^>H)%!N|Qi+SQ!QCGgD(gUQp&h
z^bctes7L2I$d{^ie=yENU<*4Kml<Ig*uuaT#<$V$sMtXs0$Z4TRzf*r4l6#LBUvac
z&L6F)#`_DQ_G8rtxh#tLM4~&9n9q{|9T~ovnTApwNM#x=$}Q)?;D+rGo3R0e#$;qe
zwtOQ){y4yd>S2UKP~yPhOd3KQ&cq*2e9~=auPb3(ql^qIf8`K{YCATci+<Vvp!Z1s
zL%U19APb#mp*iwbXM|LQMG`C`et5)16>;%C9+!DC+)`DPWFFSZ7$#vjt(pK>6b6Qp
zw~8ZU>K8(2Bqdis*W1}j$w~%G(I88K8Zk(UK~jueRhIm*p-fc(hS7H{E3E?1;$@|(
z8xdAjEzn4~f3`;WZ;Spb9-%H1rBw&7QA{egzEX+{l6yXY9cd%38Vdyp2g#+S181VD
zKo?b(28E;XD$!|ms9`zY#FZtZV`iB0Cq%BdDQ=x96usIO->L|wR!=Mp4O*#T7jtpe
z@83jr44RP#j0?it&h0;#;fJ-Ls@_&|xTov<6iTqre=kbeXn<wKzp&6RO}3Jts8k$(
zSn0P4ZkE-nNsZ#^a<yq4+G+)7O3zSlNKHfjp>&M}m1tMn>i3lV#Imz0NpdVZ!-xt-
zRIuy}9pCA*11vj3IkPTDcf}R(#T9111RkBh?>F`E1#Ukq8On|!p)0G5$V%C36^snP
z;*KR}e-b((s>$sZ*bN+xSl*FT-Vyq+Dq7u<ye*isjXB$xvyF;YUKB0#FrtSM9Y2*N
zf6Upg3c%2HjXB#@09rg}J84#Ix_g2d+tpHAP`cy=;E~L2F$XpwZ+nF~K*`zK%V+u|
zCTPo{J&DQLf=gj)w&LI`=2?p~#VqS66{5+Me-*F6EbBBm1dx8#TaYVN4s)zmK+Q~!
z^~z&Fo|{+w#2}9#eHtfrR82f%>NEy6FtCBC)94{cpB-T8G|HKE`I{@QS1(ST7A~4!
z4yS&$ifX%<Cym@^9mrZX0W)QflpuA?pI*iVrXGiCvcUz`17^fxMl5E;qDak)vWAW$
ze{>w7JzrV!$Bfvj01Q2n2(&5yEuIl8Q_f~{VyhuQ5q_D$PkNSw|0954QYK{ZmdJ%v
zH$XNZ|Asr4jRW|CN6Pn^%N0Sce1B=w>h)Y88ZAQ1_`{69H82rL_E~R2KHXy%D(A0F
zt~%YbbDe{zub=5XnfHg8R_PvC<=7)Sf6UNfhWbA0`vE&ZzXk#gfrdaU4bYZWmvVzn
zRh4LC*BK`|HRKyu-YLoUjc`R$uD(#qWpwaJgq6$&WXK-}_Nsyg3mPg}(J}gQ(j@*l
z@~$}gX1q8RN5_#)`n_u%daBAa^2_e|->V51tdLB3&nqVm(Qk)-JG64q$_?1Te}6!+
z<@&8KR{^geiYu8Wg3>1ixW%)Q#GqYF;Xt&l6>Zdhta`DooM;`i!h2wb?}#>ckxqCc
z>bs_BghySMKEjIu^=V`1s-|~7XfCxnPCd$C;W_Z&F4S>thuDk_U~szZ*(Y@f;nxMy
ztB1Y&1pWpN*K?tJpEL2t6Q6Y3e>vd(@3wYQ;{H|a4V)3hrtIUDJa&08bocuFd;=#3
zB~A{KQVF`GRaC-YrM%QeWTQN(kI2_TBT-|~L!CtaEwqxV0;UFK^ulnDR!xKS&2l(p
zPPG)DsM%zpK9ZBDb9%~0W@xRR@_}x4bhD%9sj}pcjRz4}2rL9vQ6n=+e-o5YJR;a$
z4{UQ?LUVyOw8ct@w){241V6BtAwTtbt?J4_x<I->AH~v5MIQw!VMzXHN+PfjSO}~V
z0W0pKki&-2OM!4(AKWCygr8#IOhf`Cx+xHDWpGm{Xs_z1knl%81rkQ;gn@nv^i!Z3
zR$20|qB)7RuoVRw7F{FIe-LO0wATRIl0lJ6yW2tQec8%R)HG?*^lMPBp)%+j{0`?k
z8;JIHHPs2Pme>Vv`Dd{;-T}16gAK2f@1}q6Ong;E`S!1O{vX8o7FH|Bv#1jrG;rcF
zf8yF+F3Km-*<M1m8~f%(_(MjVq5nZA3mNGAD1iAcY6q~VVw)E;f8q_oS7*^NXEB_a
zli3P?w6~D6pg(V=O+|km8YyU`JewV$HH<(@F?}T3yd4CGGxkl#2DPXG%>@#z*u?t+
z`xOl^RFlvED<j89K7s^hP}P<-i>J|@U!HZ2+vjo`9nt1Z)6DzGWh{&$rg;LO&?wLL
zgeoU(&?rZv9Bn1Ef0fWEN1$cTA{PXrZQ3yS;;Ct`5P6=t8AHt&YDP~-{#N@co(_x@
zf6#%L6~lXGXn6WTG&{3B%F2;Gnw@BNqBVfl0Ggc$wCq{r0<&}Mnzp}!#W}qXjZHK*
z(b&x2YSGwy4Z;RvQ}DWH+L_`rXkuo2fR!V7G%?Y{M9m*HfB%Y_m|0r_*#Ipc_Kz4=
z2HBK^^3xAeE4iTIvS>iKvdJHTb3l#(o5KKciMUJ$myDme!+*vl!ldCx?IHLOeCdJD
zV}`^3=Q<NW;xL4v;FBVunk$uK>QYFGMQ9>4(}5<pXTEW(B!nX(50RH1Q$+M3`m#aa
z2y%uDXAmExe_VRV#f|`dFeyUL9EK%{p^aS4N?L?p2JoBuWn{E(0Jet5G;eb2`KCXC
zb~#YuorlmA^1})M64Ra#XsG~g>^ip=8!5ggBA5t0gkBoZ^VyG??3&e@Q*^q~z7(P>
zMNM?4=w6bKz00F>{eqn6ef_wvpODi_7+fJbS%gBKf8R<jkN%tAkcSZ!2=)qqePMcO
zbbgg;`A`pXNh5iuayE-pLPg%GaZg=p_{^=<$-hNj<$@bkbH}G{3x*kCn(C@)s%i@{
zPZe28W<HZu2)c^X@DPvL;E@Y6fTRf6{`xCv0TP?%8;XS<ifrI~ATsT_V2H>J5E(aa
z0epx&e?(py$a|J;D{4w225wkmGT(l2u_+r?TwGmV=;ZPgZiR}AkNwVN_bO6!p{$I#
zvI~*qesnFyH+C=kME|6BN&oL3b#nZn-D`7ex(9_Eoo-)0(|frX?ucqd1xN{}y&=%j
z0U9u9w3ZNUX=@0z8iEc%mjdXPH945t?WX-Je+gaD=nW>eLu|%GT6)xvL5BQsq^CG=
z35%^8ILzap@rg6>#}hvn4q-Gp(cOW_L*%Uoc}vEM7S<60->;Gh+WpQk(fJF#+rNS@
zoc-N)Pmg@<#5Vsqk@<q3p?T1}Q2h!$19I2%0yRMNrXbJ|Xs;Tbo|}tvsDgkMivebB
zf5}uDhswHDP*hf$Nav5EFx}0`fpNRAX-^4E-mUHCCQ2DJc@cStyi_Cn;tj~atHsjR
z;j*wSMXq{kM({=ZL>R&C^YfP*!WW$@nBo^Rg5i1~QNyl}fh8O_h1G4s8lD%zIkkq-
zZ-v${TEhr51R4UZKtPk`ZlG7lKJ=Jze~s`<4}P%;2(ljKmqVhEm;)f9d4`SwL>?lq
z?2dtRy?fUGkQA>Dd;|QtWXmFm-8~U$|NP=S2S<Y(i3jZ@1R4SjfrdbP7SOmwG!^`4
zh$HMW)<Ub*TnTwB3iDKeh9#m1v~+-m?c%BcG?3LK9?Y56<RQC56sCc~JUY=4e|c!W
zVE>%gAth8LP-jJ;A<$k2Xw<zEJk?W-+t6x;MLiF~ow7YCyc#(J2LG4gMP!4B&MXq+
z1TfK(LiDAFJ^?uq1{zV{HGj+ogVNm;wH>*vJ`8FLVGb;8nkF{P(CT&KnYO`5Z@bZK
zG_$8&j4o7xb|vA4I*p=UuEJ2)f2EJP&|px8RKbu7e8N*nGNgv@xHUB|WI7E%vQ3$M
zEh_CeZq>aUw~i}X^@w2~D1s72knfNIqn4NfLezc{xq}{AaX^SXgl@aewN7sDICrK&
zEo}9O+#(C9l|cO{w-AY`$gLaz4P874Gz1y~?bW<FRZq&uT6wQ7_=b+Ff0>)av`OnA
zR;d-nUvTCdf-y_`IUZ`cWBwGHpUus$8-&>p<S@&|@>$Za_sBP9=<?mr;wD)2SlI^W
zV)x87eU1h(dVJ0!l@05{#-M8kUk+S<@81y5aLus~zCQMY855sX3wl{|fa<DnJ{ygl
zhJv4T(fwEU*u(Y|RlweSe@-{~1F?tLO9Ol0HyQA6g+Q3<dy-voYA#gGt^M77#C`eY
zeP-1RwK}qQgn5byD#%n3Z7HB_$;`22r{p{jR*lz0=QCiFafc-GNks0t_#^@iftCW$
zp2rA%XEGi$Hy*12Jrh(+In4`WIWv@uv7F^tn-B1W>Bw_w^xPf~e=dfZ&WJh0+$&-(
zIR!qrEX@~$8^UenaP@z%F*2%hsbDW2A53@sIf&Mmc~T{&n^XY6@857WVt^?pk>f!5
zkv0Der}fD8EUI7>3QGsh<kn(AVX6G9L1B!y&+eGp;H=g8&e^US=dJG3;+nJgCfNce
zQlW|pHY(UZY0m+!e-skA1xNl(5^$$F4k7RmcrUT+W-x}8@;|SD^BtX6zlk}7uYeae
z0fhv<=<slZe7`r1d#Ie}3nK=XQt}%E`h^<(MM7)Ds0}}(3ijbKKv-q3YNZ)<sivl%
z2WBEvy#s6@?-w;(+unL>YqvJH-MZb{w#}(+TU*<<v9)d6Zr}d?U-IRBGs$FTlF3Y-
z%$(=kd(S=3m=4G*i1f{ARX*gEWtL4p6@=Po+yUMo?~hK@0f09T76AWPL=;@fu!m5z
zT@FdxVVq#0`j&!RXu{?E?&BH}%Kg-Yi-3j{YP3?`(*kSBBZ7&0mW@t>e6*;ef%n!n
zsQ%+i*<+k#Hx8zX`FSij+}HVIN05<b(>=lR@{@V-``m8Vq)PX*kpG*6mTR#;2e;!h
za@v$gK=`+TQ6J6&AAqtTh0dRcnl`KdNn<6YbsI+rtwJy*^)fhvYD;8P0)4gIyaZLv
zzp0fhJs8&MD7CV5k(|pXU?n;w)E6ni#K2U7Vh|jjh56!QFsUyaYR>AoEWUBJeK4<f
z-WWwSex>!P{mEH`I`8gKZj}C(t|u<qv|y4)2WMLN!-w|a27rECGwkmd6WEpdu!{#4
z?4FII2|Pd=Uu^a9rdUb@Tlw)?zcA`kakEEs^NyRtBD5~~<qi+g`(v)&Ay(eUTde+g
zx{xtJI=T=mO+AI@dJR!@-?<eNUVmwHCJk4mH$i8LqSM&new;NDmrRMSjWHsE-!A)q
z(;4lT4+HRPNbYr~8K-z}_Tclxa8@~AhczO)l?86&XA0@bDB})CF)isxUS&k9y{O|p
z6BcT?2pwUGqYHbaZqCk2zz$7QC`Z%hmb*n^=wI^UZ^9Lr4a#%s@)iT-6&bOXgtd*f
zH&9^oh&kI&d=Tnf0(~Xs|APUmNN0b}xxvaGK-RG-)BPu1DcgWy&m-+bpE7q!m+uk>
zd(GRQu6!Lw$BZJfO8|ZN6=S|_iYoEvGv+h4Kx?2=JMER?O^~O{nhoKT*}#Tf+NE}O
z_U$RH%5d1;DhbxhUs!F(Kqv16wEtKK)QxS-4F1J*n>OfLl|_sKJ+hZ)tb<C=NMbzj
zO=6g_Y=m)FI6CI^JCfqI``7!a7b4gM%BNDk*d6A1$?!EyR^8Kj7km~~x=1KqjVcqS
zkVCD-suhC&jPO?;!1C1(Gu`@uJb}OGpyZ=NxPBPYPD6~mepuK1>$Es+LEdfcaJYr`
zZ!JleNPX@rehZ;{z<;Fo4~{nH+97Z>FJi35mQyl5=;t@*3#Ap3^z)}7Kk+w03I)MP
zIfR7c3#A2^XYGsY^$M3&4e>K0%7<#4Bl=uzQcDXOYQj+N*k-*^&d<33Ax13Ds7Zsy
z9-kEl!Iom=Jid?@VzBf1JHg662nPWqB@;m<@ya=oloXa(5vF4QjTGn214aPR<Y(g&
z@-@p@F5+KhDP#>qhFH#~h_a}N6uxH}-XrI%xhCcmL2$g$Gw4|-%M^9R^yA_r#Hw!n
zyRfCGhrc0h3ar+yf?!y24|B?U&e`-oC8DEJQjU2D{&`NGRZnO>3}YuqKf!_7Q{t`p
zVCWd6ZVHeW)FPK~?-U;VJ)r<-pj1X2dTwC$ph|4P>BxTV8Bqi?p(DYQGEapD?uF?7
z3lR1WHKjGBz_;)CZOi*L-_k1vHIboTJk_q+xU{N#<=1b84>c**$N0>29{DvNxIR_*
zdjj-C7zcS!zmyqYH=Mf(_-}EwSzO^!NJa2RK`j=UrPN*D4-1^y<^$R9%}geW52%Bl
z;|hlsc~F16y=!a-`=ksWTa+(2#<tbgm+L4G#}^I=<9>(@e=!lBh?pWjkBFx~Addx6
zF{Fo;el~EVdm(HQF|u|cnpJR`$X|S5EMdE1iuNGSD=&Ubai-%euF`wq!iDcR-@_~S
zk!z^b7XfMvbr`2l1jtE*G99Cs3d&qU*UwmrA-Go4qU0v2f*Tg5e5Jc!9&1ix!(cP~
z1%W;4+8cBh!sb{T%pp>#0$Uj@Ae>_~dZm40h}>GI^C)o$*C|MyQD!)ByJ;Ff`(US2
zoV(U|g8m-AL5`hFnfu5`#5*^Y8777EA6>!Bp)V)LX)8Pjpsim2c*!Hgf3s8)tH)`M
znm0D=fdA$|fv}w4b~vBt3w7VnP4qPIp-(jN=5L{?qWF8ams`$ExI1sAF^FCzU8@FZ
zOsxtR<w0y7pVUBjFbQhoPhi{Kyg;%y{ak3MJNZ7-r$(5;;RDgXtH8$~9e7?&s<1^*
zXOmw_B9@->fDcphGy38JnJsM)SHjrxh|%CAzLA3xhn6=JyI&1Y&=1|^J|wxIV%io@
zug;P-e1)P*^|s508>kYM0P!8H1{s&=$Nm}PCaMpa#3oaVBJ#RylD+vE_NK=`%k<Tz
zrE~qwxq{)&$f~zr;gFNF>|V)MZ1d<P8rB4dt0AqMK&9u&%bSctcN0b@R(VR_iu5M)
zL!)!kL*wJ)JR5wgI8sr&{O17}RqU->*iKxR8%K=|kFiuem&BXzQ{%a%)n$E}eC1uO
zj+bsFX98&u-B1peTTZ1!3+55(3ClFPbX4CRZ1V5$ED4_JhlT5r>q-9DzUR!>lQLlP
z4BF-ZkObE~@_Qho!>~W7F7svvExdN2k=mGC$s2}MHVfadx;%&7_<?+o?-jnz@F=ZK
zbJ<xTaLTC-d6n!FOZ!zNo0rDpCgdceH{GY?F<U>qy;UNqK82Dt{)p(!J7`e(Zq?;{
zxHI1~WqEPl;emxyi+lRqs!KJ~0;rcN(Q{q`J9KCX|CW`o4v&c>0Ri|ID)>5Ga}>JO
zpm&PHA8}<Et^<YmtGKBRQc#wrE0zM*L<1Vc{X=v6Xlo|n6$x?RyQxWf)7966NNF2f
z$aZ9=!b7zgmzp$jge@!?%w_E2pM(cb3h8SsEY}<|p21VlL-<!7Jk8K-X2Y7scnGaP
zEc8karOE3CGmv=DUks)!0(q9HrE|3Yp@c*%k0e@s6|)6)=;QhZK{xnZK6&wEbK}5*
zcmhtTz%qj2&_{5!H;|fSHyL4*L2owKksd<3!^Y72bUKIvk;)dY#M#&3BJ+?|qyf27
z2lXscQ9zq^`{%$@6kSZ+u%9X!H)<RRIjpORleJw#9jbBM7MlQ;+0F>r(={NumLPwZ
z4zkwy8aj^IX;v%BYFitW6c$ANZ2m)$sYs<B2Nj@p^PWY&$TbFf8PlBjSS|B!KS-6D
z_)5EYh5`rzUF;VwXaQR*?+}zCZ{XEOB3)=|8NC@3e-;`ZmtZS7v(+y~;}8L)lGL-%
zSvwBtlo&_;R8YC%?#W_c>|<m2{mM|Au*#5-N;OLbx$<U<&1H$pc{4=dNP@&B6mPyk
zua4Ocr(4*ZzP)cJ2t-swR<qS?M>RvsJsoQ<Vqhgc*m-6+V*SC#6swr~rsFGy!_k$5
zI{ndho_QcY-6H1i7(|k7Xh01NahP2LqV}HBS{nQOeceZYlgQ=pRlN(irV7#|B_`zU
za#=DpZhrq=a7~@K?}(#kIhi~!QHb4w%GKO{VQwrQwvcN?SB=yQG41B*V;v$E4m7MG
zc%`p@#(ekiSM39H-iNG&y;2>}Vt2TCZX4nEgzwdYDw%^83J&^JfB6U$+D&dUmBRuw
z*AFg3?@;>uSn1g--Z(P95EVe)5s#1GPqiQCuYDvEpZQ-NShBPZGCDYgC79QjPY6*(
z@Z_@{7EL1r*SML2NsLCKJrd{HOZp2j*_t_gmWiWi5vW31mZH{DioOkvOfM`<%cDhy
zbdkCSw32n^AF&wdCHnySJhsGyABU~8v$DfO57(P3CNFSmDnFnZIERjvZ<{k!3DSwF
zvjhZl$waB;P!^Gman$ze!)geJv~#TlNW!pBxycWWj1~Y)f9{S8!bk4pbz1wCwszv7
zMX&dR_SwtqQ}&MZt#SM<DBjDKZba4RUW%bLM+i|6xOZR`n+TXxk&lZ*onBkxTk#+e
z$>GdA*=9Tnxc(Cj<+e58`{bvKE`<MGn=53DtVUXdi=-n%Joq>0yw6{W50J=4*4600
zAB;9Rl?{9JacqA)kWN^IbV?6+Z@P+&+|=K1O~RL5KD_SK-%l6a-}!GO*$#mrTTLfB
zMc#0%wQ-%2=uiOrLfoISDxRI}=6-L&Q?93q$(bT6?xo*lu>ZE?n%;IY?XQN4C}W@g
zY~)cBy=u26?l?b1R+YY854PxumMGQ@y+I?L2f+=wjDbtfJM>6faf7niUbLI`uv^%W
z8&<v(%cPCZRA#Mj_LiK2JM`u^HVl?vbugp&u8n6v>Jq5@JXMQ`d>aH_tXd|9^RhM#
z9(-CXuVOc9NRy{+cKn%!WvtOnj_eJ~hl(xMZr8u7pAs(-J5^|Z;uD@}#0@Kb-md6_
z;+)qfsYuPW9m4xs0>3+Ff0+u{$|{+eCS>--+lvv5r{zEa0lwn7jzT^tomTOf8N)}9
zB|)2qA^?9YANQ%_q?|q}85!B1P*fACSY_uh_CUB3pLLmwt(2HVV(Z@K-Z|G2EIR#F
zbyd~&2bu)g-z!Oe$^FJK4$?|e#i+|1Gxmo3gL8yGp(bV)u+GOT7LHqQq8DmlyJcmw
z<!Zy7tIuy~GcAU*HFUbHW73p$jLzOC9v*5sIRW9DmuYQ}G*NWY=cV(wy?yfQan3l1
znvIX~1JtHezY4D&<0_7PQ7cWj7`4M2kRNM1W>U@c>sKj^=AC5Ou2}w5J#6LO6r&CY
zVL-c;*SPIK3F@`bBI;rp=&tWKqLq&@G@jn?$n&TIn5D>ZKeu`3|4zFxLC-=Z35cf5
z%L4-I7Q8f^2>7Md>QPgw2#Cq=4*IRJOmXG2r5U0`s|G*UNpMo3@F_aY+@PU_y~Wh&
zrHIO{Vq$-QK%;7+<D?O66D&h4`MG$ka<^6NSoVO$dbx_H9ga05^%6YL{H57wcvxxE
zH(jO+9NHX7t#XOP(9XEMUp76n#iL0<EC)Q98jSW@yIQ>RZLneb-K$)dNi!KrLe&pS
zcFp|fI+w<hTBY(ZP2TlH1;_@t%aMpPY3-+5T2CA&b;DnX9A`+~9~FySmIEv2>Q)sq
zjH>qB-Lpph5a=KIls3OvfMUPu12xDdA3Y7Y2;3?WvR6KkP4nz9&B?f1KB!TlMnJoe
zP1A<e=?^&t?8Q2uguFI}J!dV24Yf@pxZhV}_@y=_#ghpKC7PEh`D@PLiQZt2?z#km
zd20IlR49e&@!+Tfl~Xv<VFKRQl^raaYz4%}M{*vDaLQ6%vb{&AV=$P^8y=!$(M?9N
zD0S~z?Q->bssluhwh7w~LRzEwfp91Uq622n6k}aczc5escMsDG)aW?kQWPadQA}}0
z5#`>6WN?A1pm<ovnhI+*FH4JQv@iki8+8b>gjpqu84@lOf*JD9zF)4yJXqbr4boO1
z=)O2&a$1eVc%RN-CQihstTsfU9_WQfn@>>&6?s*}l{7xISowN&L|aeyK#sbVvfQ^>
z+EM?H<eJkpb<01Lwt7K!-&XDvZ22{P)3JM&rVcjbi6TK+Yd5P`UoV~ScHhCyA9VJb
zk!!j|;c3lBK~7qR6+mFa!}2)kWykart$yb4kq0|q4h;yt#|2$O78p$Yr9Tn8aXa^|
zqZtY;KHg2?6mHHhM7y;B2TXd&1^nF+ohk(N5C~M*ktDiYDr|~z&8P@5p0R5W4w<sT
zKp`h1_#+053m3kB7MK6lM`FE!XYPQC_>MxLep0sgw}JqFfB2Y+gCl7aHFd2a0tg&u
zG9S)H@niJglKg>-0<(|)XU;nVcgRUR8EhST08>ar$Z)=0#7?r`2Dssbm{(z#!XL{s
zkeyQ=zUw#HW5|R5Rzp?O9Jcm5Og{U<x+!{t91PT8ZuXsekOqs(^uXg5(^&BGJL_&k
z?}gDYcO-e=8EaG8dA3B*h?dX^S2A>pn=sD@2-K|k5OkN0!^@R7|KEQv@BPdowRw-P
zbR{D4^BWr*EAjZ%TfnQW@e4b$=lzt@!=Q^b1))akjg?nXM_r|@R4HvXY|;h-!6TGV
zF|rMjdSu7uVP46O{of{2hfy)&hADcM!b-|}<w{%kP7eo3-r45ul`_M=YLoi`7uiey
z^49bikFw~F?{in<&?mydlCz{qW<P`=>ymjH`(m1~4EG^n2Y@%@34gseYLQ{l^RNhc
z8JjE56Iq^Xufn6IYpQg!FtMSYl$qoZOew|aoFWXf6PibK#rb4rTqzoFzG@z04mr$S
zsV5*)FNC@$F+e5V;wZyPVauKoV#TiQ`NELM_rWr!B$AeIDr+@Ti^}0V^TC+9*iyvG
zeF)<80AoRm1E6n43+9^#N(ZF%Uv>Db<5q5Zap=fSN;t~J7t!@%Ojid3m2T&A>evMh
zI`a!XsYYTSv{5bNi0LVbhYRfAwdy)L7+SJu@NOs!Vr$<hove`bWN(jLEqTqA&w%fB
zD?S_SdogRw062w#5B$!w#gcGx9b-kOpww$t3q>;pV7N#31Y$@seZ^?b&tPq#3|k-D
zH8cH$T>HXi(xG7itq&oFDLD8#utzdmYCrSLhzG{WLsmhl4wj_z9?D^W?nKP2cA*@Z
zm0mOYy9%uUYUZX)L=acG7=o7((PBO4H#cHgzV4`8RhCWsH|Hd}qIy}Fr*W65@_INr
z`-jm=U?EKyC#={lembG4_iZGITY;6&euHdplJOkBjL63N;?xL*^&R;oQD&v1&f+N6
zeLzs|6gbihRH2c^i1AQ|#i3L6mMG}q!NhNB`-*{|FHg((IA@YY{)~#gK`U`?nls%R
z(rAYUCf;u)dY5^+&F7duUN{u5lg+6o#2=soEA0JPFCOGl`*E*23l|R)`G+qrE{qyY
z$Lidbf1w-XFfx)5Jk5-74oLS<x!i{70s@YWUPz6xlWw-k(zHVGjmEpxv&!0Ay4+eD
zDn0$^T1M{O24-Qrld>~1tO?H>;D==OU%O-)bN|CcpC;a}At(5Q+MZuBRKFPhZuSGv
zhjnp?=kv^UpVl5lZg=?qFiK%1Ihl96sv-#(FJ>7!gyL{<W`6kTz#@FP@a3iF^ML+c
zUVC*#<0=$2P`|4W)74iS_KtaPN71OtK72`|hhds*fRXxkt4Z4C?^Y=b`nT1satym;
z0gG&z;L#!8V|FE?o{TR4kwKmIZA%+)c&%^H3QFwc32osi(kf%jzfmA~08<BObXs_z
zD{d0ME4fJs6lg+xh}7DL7P|)C(w(j-ZC)`MJQmrjg}2gIC6B2aI&dfwNp8<?OF5`5
zws>HU8PS`{j?>85s<T`dw3K^rr=Powcx^J@T7Gk!8^rW<gI`{nZ7H80=c5Ilr#~GZ
zzTG60-eL^!X}66#s6i*U`++v2C>}@ZxRU>PK>Q|O$%i*V%t!0#B!0x5PRTtIpN@jY
znzB~!R@=tlve`<!jrWK+cIT&Bfp1?rzop<<TtcB^783}*P+T;Bx<_voyE(_3_8)!f
z+usZc^CkB*PCckw;PbXv>H?CRPHFKK*oIdZpv)EqFcM)*rK06XwC$MtqgCs74IccD
zxf<o(JP)2NySI03A-bgn1vF)TANP!F^Mb7Kv^QH_*2dlvL)WXW^<!`56*jEdN^ne{
zY$x+!nj|8WAAy8hqP6@W3fNN7V01Tu#xsnv)|$<!{@kt<7E9HWh5~wUaF@E5rD>l&
zcoOximp3+mCKcixVQckTaKvV_5g5kaa!F@hlO%_+0c<k$@MUeyGfp<}W6FIE%a$CX
z-XVRshPbC6ueP-fn^N^PJ54ITa<wPHhX`L2iN6c86Wb&noYg`{+Anxco?G{sRI=3w
zGKVEcj>q<@E5$jLa|0yaGkQ9wsv=%3M;W<eI5m|fvB%d(>~cR?=Fvq31AqOj>%o<i
z&{vHbx)+@|R)+ClTrJwt4=5SfM3thiEg7C;DGE0_UIU|&MSzgcs;M}*P>!WY1wVvp
zq%iv~Si$fF%2yDX7+6!_V0k*l$SGOb|E;65e!7W0n$+D%WCiHoEQzX@Sj*mt>BVEw
zodbJJNa;D_)OU2JxhmEj&hxo~K>=0Nfr>$as+<V%wQi2EFU!zz{szy1)PbluJj=~K
z{OB>eh0^9iu>ev#l79=Pjhg^)*AQg$Fg?b8!m%GU7cjOLMoAJDTZGNk^%z?EhGS}$
zg>!J6-EKT=7y}MvM{jL{EB^Z<wC0(1F`0x`=SPtCZCyn|lbEd+_5;<Pzvy5`?+$kR
ze~ejOG(>5f{=Sy9X+a5y!w>9oBs_}j!QNJw1l*y2+i!IW^FXyv-+hI!F)jO)wqvtL
zhm)&1T}fein+~Ca{rErKa^6@0%|!1ZDr_gzl3!X-4#1ps3yufbKQuG<UeY`LNZONQ
zGaCY{1ErR*b3kc*@!$$}sGh5LryM3tt=O8gj+S<Ugru`EjzD&*nIaB<X1}aFUPt;>
zMd5T^d_c2_=zw-CtZr1vy{JER7v1uQ{Y0P*j4D{88>PSs4tJJ)AiSPBn{t@t&n8+A
zC_H*5Y`}{RJu=hSwuUN$mDOfbeY%s-Mu+TK@kV=lcNPqppg~g%dMk%i@Z%Z^G@Va8
z#4-Oc8i0XxNQJ!B<eIqaJ!}eMOI(GfiDV%mIk~TbiMsJ8P3u8vx#7A}#W(%!1N|ao
z4;hVV-fj~02-!0ro{#7sE@~fx%~3_>CO8H+1;FECdh3k}=qo(@8^EF5+*pk~WIRZS
z9~zC3l}&%sscz%%*gUR%a;h|h9f|(e=8?UoXw4<S+~8nQGd<J1#6pD$q!VkRC`JfW
zO;o;6c?1ZkDL9y2cEy3~7f@`gigel{W79};YpiltfEmeB&JK+_?aNY;h?YgkOW8vX
z0&*Fw(wyESNM(+0A6>kHGNQZmpq!AE>%CcvJacf~OqjH>)W3E^1{7uYX&zT9G7nni
znvv>&QlotEp*fk8`<GHW{_9@M9Y#}hMaUB`Yg&9u^g2)e8tH#81gyuX(2b$gB&4*<
zzVDvtDI&>g9v(J_rs@v940WvX0%tWkfTZ4az@ed@oo;G|H+-uw#$mOwar>{R62qvr
zopL?Wf2aNv@ys1g=GSjn0O^y+oUVGop)=a$x&@ny%j&QP{IExSDu&5$A#WChUFg1N
zi*1Evk^@s>w_L`4dB%`Vy}$ZA8U2L3e3d+lI@<cH(6J(ObW@zisoEZwC1XkiK%D<x
zAIAwvo=f%8ZP7=Qhum+ch$JaSESesKeIm+kg>@nnN0D?3whyRI>eq#Ne|J*xX;q9y
zDHLm^O=Wb|{8kQn5~>c3n-KCTrhfmTazkNB1$C~btBh`Ec`DCFuHoo6qfM*qbOTm=
zQz5YM>banSKkUFOv`;EtiSeWg@BzwyuABnlxF7^_K~bDa0}uJf;IbqAQt*XM6GZ9p
zxU7BK!c%1T@Dj<=OjaPBe<huTY(P65qh^cPAs-G52v<?Vj7|sSE>+7M@Epkb8+#`e
zg?ba-iV6(~zWhOFGfCWvBiw)@OQmVr`F{LZ=`^2SN;Z@lGothqFU3_4xD%$_9UN3r
zVTw{@l5-Vu4IS>pS+gcQRQW<8)mUKUkfY{gUz5pW3Mo<1Dw!Jp7NAH&Nh@;$8G_5)
znevSk5Fq{o1#=)r+T}A0V=OBbS&*aDbj*4dYP8>%C&n2EK1Ws-KGB~YBsY|Bsw`dM
z?BtdkFMF<C<mFhY6T1Qh$QyOSDG9}v+r+HwQy&9UBiGRHPc|>G({0&Y9b*@P8|Sx^
z5SxQGJ3wf3w~2ULQT)U6Jo)p|Y}Hdm{wz7P&e@7$!!+coZ8&4s&>5)1@)l`_!T+cQ
z^!uZ^h4BwQeuq-;xV)}j3c~KBiS^c%Lx1?2?`$F4s#pXby1=(N@e0DO?nxsGc!GTy
zCrlLGFDOSr1Vdz=p6=j_b`esA#2H3~n?#lS*^Y)1=5WApOIcYVaU=VSJta;Q3}mZ4
z9QX)bRm7Jo?}fo4ayuii+xss?U5Y5l8O`Gs{-EJ8G$O5{e@VBLcb)iYTC-L3T6kG2
zJ?-mu99?<|go0NW*jQguvgJo_e#}sAdz^4=ip{+{?<+1Ix}^E{r5!mSD^Kg47hd9c
zXtZf~OcBUsRS2y3anat}s9EkBL9zTK`Ej26cV1!#S;WP@sbO{QJ~i_XHhJn=D-uBr
z=q<t`zHn%<X~WQ`$1Ae&zlCfF?tJ*Yv3Gfmr@8NZz+*d{q~@XuEjqUKbi9V++{$VO
z_*H19Lo1L_Z*#pm&{-n6e3`z&thV#bLTZwzwxup<8YA>o`D!2Zm%Fg1$=2{fPHn>3
zjg9P>2L!qwHLBDQn9#-e|B@$X)}BK(?M=uPgU)}0mxL@kP&C5LugJFq;{DH1D@W+3
z(IkO}yYU0#x_R>pUQ{(|WQS><!4T%_<;WOCZG$Zw{*z&ah5G$@BkD6~g2<M@)Ab(*
z4xSZLhv1+=Ad+*z`M(jZW(H8qUa!gD;9~vDsogfj)Kp&nrw2=%wnPY*$8Js8<jm)F
zrW+FstuyYF2b*jE9{1Zbd1WJ}g#iy`7C4~tvku6iM3+(8Jkn|>ojTFwTCr+iN9V>=
zfs~5h@*B!t(VrAn_52at)Off1r%*q?(>QbhJ$NZfR&1ghpKnwiQm{<Q^o{t8ZS3wV
zNq=YE>(VUr(|)ahx9sjcGxNdQ%%xJWG6lbzRD8i`H<{&ApC+=#AtbY#dGqre9e5(!
z%tYl5dPK6AV@f%<sT8Ilvo_&WK?H%%%cdEl%^7he%9Pqlj%Gd#>1x<vRJ$gU%bp+$
zC;c)+G83Aj2qxLb{ObLa`h^A*Rj%W-t+^^<_9o7NV4rTCa{2C+r`GN=bee-K(cJqR
z?byzXs4tWyefT;bl0fT9)-=QP07(Q(e617yoVo#VCS?00ufiggh_OJU$GXhp8Jrt2
z2?IgD6T$01EYY4#NTW%q%YRV_qAvo{_hhuaZY9qg3{LT^9JU6+4xTF>2widzQ=)D%
zl38{l95U@{?vIGK57;+WCd8B5KI~YnbOh<e+fd$b>o*?FkiB%QN^!e@v#z^AIA0ty
zFlH-eaPo4c&7`8xppOE{mA0%KiD*B0Bw+FLiMCDKCWc>Y)}+@6QHN5pS#<QfC(ly>
z6sKJZrT$%BIKsCZX@9&A`}s>PqQ@!^_qWA?NfPGo%mu<yibPMKH5uG8e9Hc-*IVdB
z+<c#c@#;GiNMO-&I)3Q@xM(cwYfl6$&eLbGW6)l`mv=$<=?UroQayHXISizZOeiA?
z4iI9`uWapBy(-r_5pe;7E@E=2oO_NKo(3Jh>TGBdTp}~GoY@y7iel{r{!NXg@CRdg
zg}Q%ZnD(NmY=;n8bPXnaw`)n&O7!HBIgWguq79(Rza`v39Rs(I)+N{H3pFkFd$HWc
zKawq|qM?@!ILBjALiW$9-;%9gh4@iAXoTmvC<Q?RbVZ3>JvTaAx#TQapoH>5!_~Fa
zV`%F)gycx5bRZMnyMR4UW|Sp)@m^MwkQ5QA!pM>TNtkQ}_F_iWG)t!wLnQl+PPh#J
zdN`uk?K!Uky=%ua4Ac1Bwq;<971WH%R{nDdE;!mq86y<~TbRvBbYIN`?Oz22PjBdL
zqAK5MC1F+4@>#!o=ngM?;$FDAa^(Qrm}wG{S>@lWnGTf+3bSd7zn@wEYvZ&M$#P`4
zyHGl;^$>tM2)RCVIScuxa`k3y@5Ji4DeOZcFbnoo|Gpm(ZF_uVA|FhT5S#NNOLO;?
zLT66pqVQjfqo(n4(SJfuw$prJ&_`oVWus7{`RMU94@o#EoMR^yP5rOrgMcy{pGwZ@
zLmRYJ2D%s(kqdq&=18o0#|Hh^0(S%I8?=*5(xEOD0dp!A{JEp7$y*Kq-slO`9zEbI
zAOa8<LiB65c?>#9CU1_mY)u_b9@V0nguVR*C2O;SAOm9|7qBON1EI6|uesQPN&D#h
z0+ii`xPZ6IRz$MUJ_&(5opSVhfEYqAGv38g+uJRay<b)H$B`#?Xb)j8@^&)XTNL9Y
z_|tW8O2k<^G0MIj<bI7St0O0vJTWQZ4`4TB5qmw|m5=N1pjS8XZE{Rhd#ztP+KfZ2
zBnevGP8>GTY6ss>4Y<2sxg1b#aWTpacvv~C{>A^m5<|gwX2eJV1^iW$1i}KW`!~Wm
z=zm83(RWb<EfL=6L<y0s|4(6qtFcB{uCSl0MwT*<p0_?Q;^M9q@L0RlbgmjUyS(wc
z<3F+oN!9lm{zQpACPo{K38fyveQ`i>r1rardS<7&=f<8AhSh{UI{sYkd~%Hf!vakv
zsq%01O1sh}id*Qx`mHXa^ZV!)hxD<NUko%nL=xJL@@4vdS*^$;Na_nQkoLspY)Y6{
z-U=)}rUZuL_zkCkrx>Q!+<{Lxim{vIFBiiRylF9#lwN75GiQ6~9Kpd*DjDd3?+?UT
zn)FKpQMoMDWA5`l)gX?D(<y2<Q_T0C3v<PbK7a5c>;CL^>dB4GEJZUnKu7xaxopYJ
z?=12I)?V+eg68VH$kMkl8p@!JU)s8mdkn(?3+cPTrl(JUlK7<KR~2fvhW}Rre;+xo
z$0Ob8gAr<BR099Tz>vVzz`VL|Wyy%pfk^e7chSGsu~0WiehMH(9=R4@$IQKeg*NOn
z=uFum5A`(iy=%P3G~<qK9N<AsEWv4<mv1{m?aT~fMMW2diW4`AO3BbDWU+?cdyw}e
zl!1ZthExTF`2LaNuqo@I!{#Ov+-W{ch91|#e25H+?6(Wo{1A+!5=F3w2x^R&2|}QU
z-jbZ|$@*Z<AW%Z}yjEiTLt&w}qiW?(jD{LYO)T%n5XYe`{2YGcvR(!ip!9D-G{*??
zXLjUZ)XWvJa%8jcT7hp{?yk8pym9R>>ByHZxjJFst-zPx0f*gWtP}jUwIZ#QV{hp&
zp;=ID!Y5W3jHW7Q;H#+z30dL?16pZt0?$?c&sQ2^*-?K6W7C}~MEypZq>kvPF*x8;
zaG&T)SRnkLoFQ@X2Ak~QO-QOrPnJ&7{F|CF#Q`Wqq9ga8a8#p`8Fqw|ujMf;{=ZZ|
zz))fpAVEmd<N9B+UvCj>I*;6Rz1M~{_}@)Rp_Ge6+uncT?f|^22HKg6V<w#}$4Yaj
zr%%ztlhRpUe<HsqM&3LZekb>2t)o@vtAT^ZQfP%is7><lDBf-8sS}-UXrDo7g<&r0
z{@ctNte64p*KjY6TjWZ+oS+0UDnp>h3jZ#qa23YEp$^@$MQVtb2O_#{>4xi?z<vy-
zqXDNq-yigP3EZKCH8aL-?CmmQ`OTrYMkN3-$1Fsx@f1p(uh#N=58e-}FH0H|4|vdV
z<PAG$)!U$E$;8j^4=5hT2GOl=NV8p(jXxA$f`i1KnN#jdLG?(@k&8FC1P}o8mzr+K
znq;I0^C?O=3s@AZW059p0@K$vZte;#h!sTvTE@b>67z`7!pM?S=5Z7LoF1*M3HJKf
zW4E{N2D5x>5|`cU!WSsm%b+c#4(;^wRDTKy_a*7I$s4d0eDk}{y2py~P*RC6rDssq
zvX5%e!Hz~-pEsEQ*>xM+#Xi7NhnIyU6w<rU7^5n&biTCy`JtTDPoQ@OC83n~t~Z#}
zQ%ECgob_+?U#(6r>3o5^kE;tQB&hmbF$Le%sJm}<h5SA`9VJ&f>BJIIq!%P-y?R(0
zg?OjBc5Dvekj?zU9b=VpGFb=d$XUPp)0>&j#`@aYJ19BJ@9##E_oV<^6_*@MJ&&E>
zw=xGpEKSi0!UgX?RS6c*XqD@D$e4-ly*Dz#mdTWj+P1;l_4cB6)p8oo8uu>Rnk9Mk
zAuuo)fdgL;nb|vrSjei5^*PxYBq*^+o}Z9Tf79+bHqBotmbK0xb3r`6lhu$85Rt0>
zoh3?+kT#EzGZXcY<pHda{emv)i^FkH+v2xkC7EM21lOR~Uw2sAvXu`bM)*3GN;D{g
zO!K{~#6=0Jz=8vz=Pejiy$R)z5?sDc(Gg^^AJ4t$+|qnc3Gf9k;DlJL6K(bm0k^Y8
zvWP3m`*TENPG)YXj|=M*Q5^T*Rw6G^8Cj291O?A-muC#zv}I%twDo%!Ohqd;|A^kq
zK6Ea^j&SiBs;M4Fa^ghN;wy@m%sO_NGD5x>)H-MzJd;TA{BYaj;tfY~fFPuM*2jtD
zcKhkZ%j1jvOAVH^?IwOiBXgsxc>C6*v;i+iovZ-sfS8*-U|RcvLmbM@^{EKt=777Y
zP^GklmD~wP6cFTR;3Q>pAWpjInV(3dJix%mmJ!wwoBP&HI{R=89%#}7BKA<8K<Q=j
zPg6<o5pH`DzSW3xGQtJ2A-)|VyAAt6Hy4-x7a#PvLqh41>zu-CK+HRTv5v}EOR-aJ
zobJ|$RKaGxb9QU}rzntv%`o3}%`Vn|H6_5|z!d=i9(-<wmlNo-CWawz7XJtZsM2$|
z!}zmw?x$&M!u<A7I{G&5T@)4f;e-*W<90;s5QYELSfZ0Q&bfjQM9#k~Yr&vFjNpr7
zShR<Vzx*?4ofip?TQnz)jRIA5{hLYy(s$?8ngarlY%puG-T+U!SpVKaHEvHRCES!0
z_mBd_{YsR9cfY2wPibeT2i<OpD$+X=udP-k8rl2sZ?Fy*hEdclF!<M5vdohJ=A>zw
zC9}d(FhXKwPWB7FvsDLQ$N@9rQE}+e$p}qMX1T><x>x~EQSuv|6Nv9Dg)ZI$uN8?l
zPl1n&n3XA<D*50CT#7?RPt75{tZ<(o?(b264<*`*!Qc2t!)!Ex@nL(k3ReT)0)YVn
zI_gqW;ii!`7?)#v<)z~g;d}Gk&jkVwYi#wAEN?Fw%qad!$HeQQKx5-HF|B5-{@ZnH
z*+;3rV;nXyEkT=!;okRWPE4Hn+|sVWHm?GXeeBte*_^Gv{#OL*8mb=W<Mt3Tp$F`i
zNA_YzAYdN!gS_)R+I?eQ|K(8GJ%CLvzSj%^*_=K)k3xj^l%t-Mh*hBlSIJ8|hb-5w
zdZA!hqy%o&w#7)l%67n|QG!N#rv3)?^<QuqJm{D(i4wDCyctHJS{N@oG4;bp1{m>F
z<DamFJGNLE=2-qQi6SuI{zAgLQUFx12S*9V13SUus#eN-V&teq<Er@b$2Mi^l%yRP
z1wP`iXYj4zga3D|?jya^sn8rSN~m#xGJ2D|`3={8XpOhrHa361<omnnVfPVr_|Nne
zKc1GepQ?_p0`EKqSF+TmB@)9t?u^<!gw6k`7$C=C>Z?uH>xoJ%gc!&;Zh;hC=s~fg
z*-OV)@?=tAI{5D-#Fw?esxxigH73UI8?N#h#F_YpJ41%V=A^$U!cE&4QB6OK@_xe_
zASv;VysZy*YZhWAsY*B77uZGPa?Ngb4;ef3fbgcs$ABRYGUz<xz;0N)Wz9+{U9xJk
z>K2*TEfsrYCU(^<;USwK0l+oOVvC0KJk+GyD|<yj><IlLJyN>VYt!k%otj(#UfQM^
z4}s<yG<RF2aq@eUvenOvmAy|Vbu0T^X6@VsnJQe%_EjXaId61s*Cs8`5#+D4aRGM~
zM3}x-bx5D2EEXO*HbMNt@-5g((VjR;$F`b0+mbyqGa4P1?!IbSz<W!yzEJr7+MYtr
z9VHK0W&+xTc;HIUJp^X~#D`H#I3$P(3BlM{cnZs5V)zBc!J`*nW>h$beCS3QuC@m1
zA@MI-xo@WBzya=QjZv<d@GKTQ8vC5-*J2zcbklo(@hMbHcFt*4<xtocyIcS1&Q>%0
zSTWbrqF69u;B+k?1J<0Y9<-8S$x}1a@#seXM04xO8)u|K(Tzro)alDpr1{RZIelFR
zF)g7V{nJd$;GMd)v<{1gScF_>>cmIUTLh(%@$231ms$3+*+I?nFJk1RS%c2h71_em
zB_BBmDzKk~K7J)4F?sy54gA5S6e-jeD}^>R91_lgSUC0lKx$LNxKK!^kgLC`?Za*a
zNRT$$<Lyo~Y6{Fh4j;~*O*D?lVTt4B5GPu=tD*d_aRt8%gkmN)y14F<p{GazgHmP6
zsNpT?pt_-s;IljK8Y+~0#fcccbeWCU{KHEu_(mSgAN&sM;yww-=0Z9j9(a=KPxox5
z?rL$EL=ztm2of$tQ=y!&>_G}bQ5S$Z?%zyHl;<eITaNggWkPOR&}7MCc0GQ0#=_A?
z<(Q4ov>+DwDvu<$EjYctxI7sBmBDq1%;DH>onEM{H>QAWX~Ut<2z$gO_md4zt)zh0
zDv^BExr27!R(Qt)NU_WZO_3lN=R`W`tH8NGiRz01l)*-|7<Qk3Ske+H!5`wW`5$Zk
zU2$;=-NGKpNXP`X_}-PNj`;Dm_)R!B6dm7St%Oc5zN|pnRvG7w(_Kj8)EhwcG~1TV
z=1Q$9WNIJihTE(Ke|H<daTK{@7pdD>>JuwP7f#G)1PoGJR%T^Lcf;Gqn__mRHo`$~
zUM)ERxDz)jhqf^=c~1O-N^m}bhjla0tMj*NmQ{bc;|hwvejn3dK<L<S!IL^sxDK7m
zf`&W^zYUnYq4<aN<TD@1_kmJz8C(teQ^SQ$bOx6(4i6WQt`D3T1{g<7;U5pOzx(FN
zi|mo>VxW43bQ(^v&q=2v&&a{qt1_UH_FKLH<rhzw4VLPD0vbA4J+?#eYsqli%exr=
z9A%#JHG&<6zB`Q<z%!yR!TwmzH0^MO-0V*`N(v)Ns38bjhYT8ruK^vyH#~Nkz3n*b
zN8ca0C3Vv4e61?jr`cD1Bu~_WlKn@3s2##doMoeqW!Kr`ygXI0z<TZr?(oi|Un}|m
zP)`?0TT;vESTe_`Ms6<3m9TnuS#Pkq2JR<>DBGl|H+;zEz3;ZpTVoC<<5jH8h$~t+
z<L*_;&teo2o0*dJ$FZfNQdq3&;<ALfPJ*X_3wm<d;Qg_~LCB+pay-s5;ahT$&z8=H
zKWn2)IwReCfNMd(m}Ytb``mR2O1HKJe4p@f{L}jntLC&j6uE}l*{X(My&2E_!52-6
zI1Rh8junxS9Z@D<?WTT%!bj&DRJNqzvKU1d*wN^MZGk6q0c7jJ`YPT7)|0o=z}YNa
zI~nFrRL~L-;>;R;Tv@c$q~qsM6>zfE*k-=KM6*9OsTUa~+CN!XbzBXjoXxTUbDXS}
z=L<>SR)$v)Utd3eU92(ZJ`v`}L|c_sy&5Lgs|>n_cPJ)3*=O$M$LK$oiZK3erre>x
zcM2KxOmv0arUI}zXP7W0X!6eeleBi-pzlObb%yV@mR|qF6xLmRNRj(|kG&1e+2@tv
zi$C#)<aprZt`XoP<=+G}h{72GL%Imds|x*deAH{S6>`H9)8q%_pFHUzJ_S^-`&|*m
zKf3Q08&ATItJnolGtJ92Dl5nP?yy<7rGNq{>L2HT_crwAExvFGZ_k}duSR7#bMGB?
zVV6jqWXi|?j!<;%3cKjc`JbXwK30^ynBVWAsPYseoyO)H+)_S-0OR1}ssw7I12GHD
z2EZR;iDyjCT!jfV=_-lVB8mw#2$pI342`0j36uhFs&{-)MD<L-Krj~xbbD=^_IaQl
z(DP<ASmE=^iP;_v@L$^=>-=${_A`Dgw(F>G+T;d(`aJne^HS*Q<DUS#=g>GAmp_M{
zuBbi2jc<47;c(qCz|RsSXxXuaG*Sh)UGew**&$@TtNBG)v=d1~B$ZK<0X;iafzY0L
z{!(Ux++X`<ZMbNvxx>VObH)hyx)?bov<v3vWy2F-nPG-3wVE+u*wc*7ioUzyNKHT3
zkrFN377+3M+u&3=Tjft;1O^W?0rU&C;RYds&VxRbt%&+`fM5V=ep;8jeP293M*MrM
zP+NdAHNAi^NrOnfdgiLc2BZWAF6~d{V^G{UDl&P~WQDeU@w^xj$k*7G`BFq(dEwGk
z@xN1=xh+Dtfj{BS(Xi_Bw8EfpvFb>s^s7&jE}o^$AOCL*++AT&X{Z21luq*RTokB*
zG|F40ZoOm1*HPOb(s1uX9ln!F>!%)C;@_imi8Uob!Ut+Kdv1^9lhn2(*P`-AJo+jt
zO_313GVTPFewLD;b|k|ZYrzVI4A=B3i+cs0B;qZejwFmQ%-kqGD9CD9kEFB33=c5U
zu41^Y{&M7r<A_^=V1#5_pp+{G5Wa=Lk|5ZB9fTudZf1Q8BVyu|`jvJX`vaQ0cs?Pb
z#5L9}tysuK$fQ+$ng*i@EVt>(6~d)<gG>o4sQ-+}+V@A(NdDDyGOG%;<;Q9D;;QG1
zQ|F4GpWGvMW&dHP&ljgHq}sSIYH6N#4v}=4tO@-reg)OG&%cz5ejlcsjCutkQ~Z6*
z51*O{<D`F{gbHGfP{fVj#Yyv+D1n%|PK1F-PIFppA_VzwGt&~*U0?+t$s4TOJ0TJh
z){#_5W-KtH1SO}^SU(g62$guLqRXLe>}mB^*1A4wYoP}W;e@}&mgsDnM=snGWN!vL
zQ!inj6$?8AL{Q0a2OK+nzI_0lclQ=i8JV#+(lj{J68k99F=xg8nwi*SkU}6bG9N6F
z2kS&k9cA)uZ6i6dA_ArMPyPamAf~GCK6hvI&afIFQw)0d;9o;XLG|DNVV}O_ixsuZ
zMz?<J2l2y@ljTauH#x}Aey__D?&?jFLbxcn^)k(~u>PX;MUtZF0vNb@^&}XSV<a*i
zYFn4^r;A(KNihP2D@OZ?I+7a11kT6n2`3vU+49Ns?*(Z(8^2EWw^Q#uw;S<b7J&fq
zAV_~={I+O(a<>>6{w@3E#}Qj;lRE9OmpBZ{wr}$e6-*R#v(~{7c$lByz$~&OH!}z=
z$>zTf)UP(&U`>-pJgwT3eRCJ~ObtWsVdsNonQzNTI_zA8jy&eo1X9eSFkC#cN|3s)
zSz!k7hJ{5ocp6)g!NTiaJzo{C6UHb8S8yylN5TJ4X}WJ3Qv4jP$a#sqy_qOg`S77}
z<t_fn#{p13H!fPZ_{qvqt_lak%zl)8n0uA3!dLgc9b$J(m9F<)ZYdQn!Gt9$TP+Q&
zh%m)IbZ}GxZ~hRm53+mmaCn&2D*MP_Qx6WxAG{nJinI#l4dbHiH1d6U-v#*qo%{?P
zeJD_M^{aQ%*x`FjSKg!)!klYbP?6c!&Y4;;2iOi^fKaCsSk)K8j+%dotx@lp*!e?0
zFvmsDnS*d?YvYJXS@ngZ({USqKh+qAL#UBpWr-rob6h!P8MqOtKw*rDDYDNLINC!e
z=}U&$&<#@`!cU;lPU(<}f>Uh|tc-K;I(I4f5-6!};|=p;M`8N#QPz>{_v&TA90l-u
z03E|VkjP)=@n00Zgx}dJ&w*{&;p*$tAK#US2hCi=O=uy56^85WzD4%z3P60at&U<j
z<12>V6+kc$yla;iL&8MIbcjxHvAq5HFGGx`G7McAH4ay_<-)yf9EK^5u0}%HQ>94%
zg@i(mrT7bO931Zd|2Q}3`2nyOuN6vSHaWdD?B9^O4e+v+Bx2rT*%6HU$5_bnZ$G`C
zw!6^!JA(6%{`w+h%E_48+_cMrBR_RQrRxtX5fy|Zbz96bJy58;5N<We5fs%K-1=9K
za;ELJT<gFQw}aC~kAZ^Et(n-w_le%N?s4I*!`Hcn?_25kIVG^s_wvA|3QtN(Z0~Eu
zR-G%g1W6QrxLEA>|12Cdz-Z5#I(2Ci9U0g%7Z5c{FhJBrXferNszc$@i(?vL5|?^4
z6io0G+S=X~tQu_^#5DHX6lxM_z=#~0ib_Kf$g2PW9Srd5tDQd(T5vE^Jm7L`%X+k&
zH`(Oku1mi=+sT)HM**E?HfoJ(I*7Pg#hq-H)Q83}8lyD@nkXcJ6*@~g+4g5Nzu2q0
zyvG^z0A3p7nXu6Jd(_g9_xozZB^^fUk=_q?SSwEF>CbAjJ_dI(<6}AFv%Fu91A8>p
z;e>YB$x6by->V;tr(Cglm6rlE{3`BfB#AVRNw<9~?(l5=D}cKaf4UhF2rMuQ%I;p}
z7l=%0^X}k^J9}`e*!g0xdL;%Eq}n}7eRoI?OEoo*mM}%cuCQTA5I0|g4rDHMe;fNf
zN-VDWm2sJhbu;r+iFPAM=)_N5-vPGEGz>9bg^p}Wl&B@Qoz%WqBP8u5m*n~Mq@YTC
zJ3{ygJDMo441k#c@|Q2KNMBG8JyPFb@owG4N?@$T3Lk<l+#3y=c)yN4^-IXvI2gse
z;yi!p_@L8nn4Qj)Y5|*pbP)-c_)j@IQ-Z2VQd;Xc$Wq+ra6Wvmtg-fP$vLS`Ds<BP
zo(E26UvG9VBzB5*x++5k3VvrsX^fp(Hy243Hx``LAi$=Uz~4|Y%TVUC{<Ho^m|dt6
zjGbfwh4#d4{}g>0)A)CiL;)IKiT-t)mNbce3=n#fVk9;2cRmFxh-F+jRxpJvX{tXx
zc@O@d!w&=X<Gnk5npsEhe6Unq3agEU8n{~It)m=1SP0T-8Hj&36E<xMcdsH@Qp;RH
zf>R{1901|<`q1sWV{Sc{;zXMLvCM8HIV|)OtbCu3o6NEemil%cYJyDLX=|fOES(8E
zbiTB{@#6h1RmYo=mBPyz1v^lB_}AQMFO$sn#^-jMchg=W_YS=^?R@eI&#O@GD6P=?
zf1+H9m07!<cqbQ4NmXrp9tF$%V^=!xoC|eQO~5(MP~zXwYrPr_YX^`S<ZXWOQRm!B
z8_@drPY7Vwi`km(*AtQIWB9wnQ+)rot5hNogx;9rG~{SLp0iZsd`(K3OP#r^<RX1(
z<#IbY$Wz272)oz5<^g*3^*sL*`D?T$(}Ssjr*2)=yz*6%B;V9Z0u0&$NroTalQ5Zo
znE_sm6>_~hhyLMzrvXj!KI~ha+S0gC@*0f!k^8N{=dyx=zq$$SaB;!Taij_DYu+iL
zfzf<x#wbN85su%1g3^?xlBU&MSM;&bFyc)uV2fB#y`wWyrJq?@HZP<l#3`EH-PqyH
zFQM}|lJ4C|I*&qXo34ACb79IH`7|1)0$?-DpBf%9*-KL)NHK15Rr{I7bqq27cS135
zl`+@cA)eh)^Y=@=4Q?y4a8c?2-bts0_v9^mjBgn-^^%~@rJ6<t8i|<5V2BqS;dK3B
zCW*ifrs`fOFX6+|p##*vU?ggdAOrNx?}Dqp8~>1f(J?|jo#u(UA%Ue}>g(Oy4q%*0
zw?APx#n1LBK`uU)2HQ_xKzeZ8R;mJxo{%Tl#7GfMs5Z)~qdiDSByLZ<b?T#-twC|e
zmg@MTFC`WCA@rN3kcSaB-!~7(YZwTLGDmS!^K+Rx-xr3e1p4*c^tl6-T+eBuqd>`f
zo3}_};0lTMSEdE3TvzcQ28M<2y1+FhKIryeDu|q{p{hgRQj@)b>sg`Vy~4e~3N@br
zmS^~%rgB(=xf45;05=LdoaN94$&H9~vd}Ceccl93OXW04dt&Q?nZ<hgsq?GYDO+1S
z(O@{#+*9*3NP*NKDQl;Z0n&-w^)6P!L-fgca=H}{cbr@7{@GnS8YlcMJFq-sR**v3
zxUG7&tVdWI<Vi#O8nuaOEZ>aK+s2ql=rtaXSFyg3M*pzBy{H-GiR8U}XvXV;)j1gt
zha2wVV_J&|kI36gZ1Mi>;N`a`^PhrBEyv2rN<0-))cwYgH-QUYikfzFG=7MopL+7l
z$FmzxpnD-u^6q)~;`li?#(-XMqy~0y_Q|i_@49*#kH=_z@2Cctu?%*4#Hm;3=-Wpv
z`;Z~@%?|6dwu+NHe%Go#W_nDlznWeHJL@=bbge6@XUtm^T6r)`3%2GN$m6w{=FUGl
z@0_kTxN<RQWm%TjW(FdM@$z>OnO^m<*S5Fz#JQ<S;rc~<B|mxDv;eBtbOB#VEu$zr
ztVTtMk;@lwf6eh8IEUY-!8!z-Xi`ICQC_n*GP2MJ;#E`KyDvSWydc>)w3^?F=_Y<<
z6Rr(Y#PNLN64co2SFYT!rk}7U`NiaKD-(r`6#aV#hB7a5|11K-Mg(=bG;*V`<0{`#
zlXu4<E6M9>M^+)}a~+sx8`$-5A(QonWC}l~TpVj>;{hZLOSISA5$Dpz+bCMB%L#({
zX`jTM!5{WNli;xX)fbb%vKCHaPO(}#tQ)qHh$$`~Y;eucDz0ChZYx1MjF{E83CvET
zjRa=?4_WUNok<sU3wNxJZFX$iwvCQ$^GVXNZQJG(JL%X)$F|Wq>G%7_IsX{{m=`tb
zqONvT?YY*RdoFw5;7EVBWcT~*+50V&wuC*K69W&qkp*oQ9rTxmjoWAhK&GmGWr{pP
z%7}Rg;!}hDn}8_=p}$Ew`=5@CUm>*wKV~Ap4EnE~V)>OcT1Vr&a{dbPJvx7;D`$3H
zO};iGaaAdC%kmUu&vm<N9?zPdd`(9H`@Lf2*8gu6{r!nPI|t55cSV)6WHa8;f`|K;
zE{Q~?AaOeM#mI;aEvOs-c|lBWh0e}*zoS`pPr$9dSqT<FKHqo5Qu|Zq)xkR4&-;O7
z2`LKF)RR{weLN2PkErlN1l;R~t!FQR{?j40_^07qCB!$mN)t&GU-4a_$RABoUlN(T
zQA!5=2-@YR3kL84pk%s-N2dtn^br+yhLp4|O*H&ay4j@F3|j#L*?3mX3TMRBG0=Vp
zCF|E}6c%ZZ^-Z?)tztZhEm(epk~=^=>bG+jS&O#XS3~~j)2Nk+DJ>w>&}2GA$!ZLo
z1PlY|>Gz#@%eVW0{>E+=GxerC@8D}Kp;p?XborXMiNuqkydQoZII0)9bUB5>6lA7^
z)H2&lZ?KE-i*f)5S7VwyPZa&~@otU=>_{q&;ow9rRU*~Ar=QkxwpEHW+H5eu59Yla
zF3twu4{^&sgO4$m?;RRHxxC6CB>H{ACFY2>zs^*Y+@iN9d4IgNhaE8^JgMM5u^|ow
z)A{b-1YYDROCV6%!~ty}eb;xXnN(Vp(AB7}jU+`>zI_7Rm!<WR<oOqa!1@D2=r%Te
zbgR%R|7=cm$nC&&qxg?N?bajyoUti!d#xXc3VXVdE?pV{8iP013y5pUApstC@|<*I
z2FxzK32S6~-UihX!~Opg1UB1l!7l2&o~1&P8L}^3p<>p9S$awha;1a!VrfPI?~eH|
z5iIagfUxIv)V{1z&)oP+FOX0ZR^z*zg_zZmb&YZHs<<1U*7bS5`D5l`JThnyMrjd}
zR2d^9T@sm(cWp|t)}qCwxAb9w%)iDO{1f);0rat&=nb3WrVz?R>?EjpP4V(3erMf%
zmAKt6MQ=&Zu_Y3;^CjSyLdNy5J<}%IL#Yt+09hOJ%MemOiImG59Pur4%p|=@{hb$}
z0K%mj%gccl!@d%rpa^Qx%`h1KO@{XSbN=cn$g{Mz=U?K-Raz?ZHUXn55<Tv=X$Ya|
zRkKR!jGeqcrW6-SWwYZrbO@0Aart)4q*wRbe1o;R?X)?0_fK>m!LeNmGGDX~cccOj
zut>A`dP^I4ozEyE@ELBB-xRE&YxctUfcLL`62Q&)+|pYM6&%Ru$7eyFO8WVOQN{f+
zMlN$uc<@zBFYkkLjNvxj|5yGCk?yAJ117ny;c$JIc4reA@OH@YWZq`@G;nfj8poLT
z$_)AUw;CdQ0<?%=qntE;PqP4omvB+Q>7(rZS@Zj(zG6~Ft)2)2L1-%}zT;OZazWd3
z845<CaNcGb4Q9OC9Qj$8BC)N%68=gX7nPl(ahy3&XByy(p5dJYoKSYpKBD}MdYw5w
zE=@NiV$piQQLx>p%$+PFOUnwG3yKLDxY*9%dq5I6)wN3c6T`yKgk9i}Xq*fnl8;Gl
zY`Qheb4xl$pBn4Y0AGxs*aSZ!Y)<N*W#Z0$pk@xH!Bx;0zy99Yl|(p16X$W%9u%ny
z+8!jTOR^RKGr>$oihXAytXVFacdcQR*n_f4Mt&_-Z1bH@*!A(ECBBFlN@3D&XoINZ
z4Dq`YJYwijbLIQQ*_Lb0%P}m#@clUhA=(37E=H+gH;hA{2X;>Mq`*amMsK5_olaW;
zwRN??61BgN!ftB2*}gF4(`a_=$TG8AX1p`Nx+H>c<b{1<)1J=2-S<(t5s>odgar%I
zGv%ZOf3M9+m5ZM?Wxy8Ueplo<_DTK)*q#8IfjVX_-NHXNc?kUpcIF4D2Ot0vI6G@;
zEFhe9cmF@Uu8;O)=;pE!-5O7|&uf5AXyY%VonxeW>v5i!YUd!GbkCasyzqaV&hY46
zs9qVk&x~rMl|Yc9uojH@Yl~20?0V=uwW@+FMX(g4)9~XhR<c{2w@j6W^5veoO>Syb
zDTzwo36KM7{4(-03X=AM@zKK;JV~pY-kj;gwaq~KiMYr7FZW~ml~i}f-(Fq_;fm$f
zn+VMq%U6<c%_rzVz$)HJwC@~8SA20#)fl%BD!C)1(mq)Tr{fp13q$27TK->Rm#tGh
zYnNZ1s58SG#Qy{Mfl|TF;VFV8HdE?5518>?J4&sDoQJ%9hYy;@`NlJ94o@V$|NjVG
zkG(q!?d>1|qWAe77ud0R347F!(0t!7snj=t4t*@{v_)bCv*r#@_JRe^nzxD1Q%49%
z1y2PDAs5ApS@}R%@q5`=8+J^&TEWn-qw5`eX<^Bw`s&nW#W2YPJShO?8sTDzn=~Fm
z@hYXFzs%2Cn~|d@e^Se2To}2pl8Rk%`yhq+**|%Hwb`l0K$RL3^K8HK8I}I%S}7?@
zDRXc6u*bb7Q#;$IeCe-x*IWy<SSmQdqmk<WayHvoT!<E9dt`%x5}Evy)NJu;U*Bbm
z`0>L{x^mYw?I*w?aI<?BIC;4rdQ$+x#&5^^KC0_Sb}k&Q{U-77I6ldqgHP2l>g497
zAD$Jgm0@$zq&$JT?m+u^&WPA9RL1kP;{Mgd){1=PckOzcJDFYPZJ%AUtzI?S#`l}-
z3wC`^_037i6H)P_QlR$acLk*zUXN#8jqs-@v?`X?dIs!C2QXZ4)>>m`S5}@4qi2Y<
zPm&0DXSPNb(_N6rj%znW5(c8Ck8MiLuglL8Nl&MMg@dVX(Vo}+Ip4ij!Do<*bqW9T
zH>^HFGL=(E8YGUyXMIHS$wyl^-G!)SXUes<SB~!vdf2$nUq<`*Dc<f0&UYkMo2IHs
zyC#Vq5dxr1<<TrNR;1BoY1zK<;d@fq239tug_k^ONc4lp-e2aMdmlc6d`wf~YB6sZ
zXujgk93WN>llq-p(c=jsViX60`SwsS4O^*lUhBU;_G-Po_SVUXKgbC>j)$&5TUlwV
zkb`S*B%P5kjQP0r(GaRs&iCxRU-m{csgenb03+>}W7He}{}`JJZ<o`0=HB#_<c&BQ
za`oD=EjNsU^vGi2D>q^1b7Il~z`a@S=6MT$Y`%0Le!$2}V4^DsI|^M#fMJ^}zSzt0
zj7`z{tMJ|j3Z>oXj`G)I-MOeg#GgZ+kAKpsF@!bO&_VFhtq?EMIeBkd>2cHuZUHW~
zS?9>K_nyuU^}yCfjhgy-=9UP@{fo9=VKG52sIQKB3eN}c|5|=p=F)xqGM~_wtNzNa
zj%OG@fZ4ZziyPk`>ye|bRg<BB8s|_5{r^$8t(q?i=L%+U`oVv5jOywd`Mvz;1aZMw
z;3=1?dG&1zP-OrYKY&v_&?K@(b+LX_VU)lmpMJ%#*f}7&=ORg+!%r({bn!PVC{gNc
zD^_?fPnWx5aqtl-#F&dnsoP*^SL`M(_Igt)4JAC?VGrCupG?Q_ncBf37{ud?)uCeh
z4A+suX*aFoYZdC3$(G6TsFLUUJt<`>e!(qYP8Sew(esU-y6i<wsa#z;&D+@jk&19j
z(2HU4LjEkHqkB)&o=_yjD$HuMPn_aM-91kkaTNJ)dQ?&4M>1LHzr0S_Ibxw)pol5F
zdGXnN8Jz?<y^)1z8>2PvWIonOO9OU3+*}J*xh6;h(|weL-M8Qux>eRwe>Q)v1B1Cl
zl9o6TJE;w7y7T>}W57~p=|aAdV8i7$xk7Rf;ig;)vAG|HQ#I`Mua1p}eN9p@`6p#u
z+cY{E=;9^5CldnP8Ve7k6b(kY>28eQ@(lKEZomm~9I4)LwQ!#N*_zBOCjBIx!3KtA
z#P0^NK??f3uvep!VZ26-XZ(9lvlk*qMt1S7=_E~ah5lr9S8a0keI3r#QVF9QQPzMD
zx9geMEyqJc@b8Fsez>YU+~|mxtQaDU9z--W?>3{d^j#Ib(_V(c&4g`6V<I}^2>E41
zf8ip)R%h@6ahk$^>oF~?SVhj|!M-Dk&CzHpX7n8{_U@Xff}aUO9y7f?F5`Ynd#Ern
zaVrHMuy!(nz{0aNLX)jccou^kmH&xl>M!)Z`0JN6=%G8}n}g}=CPj<aXo-t&DbM5a
za5AAq=;DBH_Rk^UZ87S}Wv^q0gZ#(+nM%3CyW_!Vh5$V)$J@b7SXd7qvX5T|FYZYV
zCX(i`Ci|7tMVfD?CbjkxJodDE2j3=3G=1S2Ku8$<TOli%Q#%<sJ9)#xbnQ1iC#lNn
zCzfac`^QL)PujK{Xb`WF+Nc}Da*brbO4vrgoH%#<%h(%at=;os;-$*2Kn6j*Y$i&Z
zQ``{syK=tefhC44OMQ0x_~5gv3vdL^$urkl@tMOr;OOixXIuZw{?kHNoWw_X0s(7>
zXb79%IYT|;PAEMmbZ_XS{rf;)NHC$(*Na~t;c~LWj|m=~mswrmpcmP3swEn&zhvRa
zcpdQFH4)ph^mB_iBkKDyLn2=qE_jVb#Y3WQnldtm!ei6$3MIu$FG*d%p7I=TVow?;
zfI~}BT5s*MKwt(g^>q6g_v7wUX}r+Bc2!IcQ~dRB`}`fOuQL{VYhA6dguimC*6G*A
zCKdvE=<KOgUhMJy&@v%$kDD_AlJ>K03pMGUpP3+D%EF+1A+J(hx1EZZG9H4gTs%9{
z`$Rnm;xl<=zgCo;3C7ua_*6nbO~jR7gybN4gKh=)4`nUaDD#-&ar>lZx_eW>->|0f
zgF@6Kx2QRGJQ?~sGE09I?&OI`S}~ggEmjt9a49txcIM37R*tFs#DG{_0J`jR*v+h|
z@iB5mqkU^M89bMVl^e~WBia{JAX=_dmD26_eMWn!ca*MBjxroudlDL@F&k8Ym?${A
z+gwN_mqahk(LA>~f)TC#H&ToO0JMFjr@L#}ZhfD%MBV%%TSQNHQY$&!Q?2=ze%3F*
zvqp`KkH!g#=VnIaSK4`iEzhvyL79_}W>ruFeB2}PJLdgg?g6FNCd<Hg>rp*IelzEw
z&uG^kEM~5O_{)qOV^nj&ck|GsQ}FjX3GCB`zH2`Q1xT-Fh_|mPK2Irotz+aCcpH5f
zK^;M_jVZV1DgcMbIiT2|!?H~RF7s=%ROo9axI}WeX9KQ7XYg}CLi7!oGnC|ijEJ+`
z@Tdd*#)Jkm#M#HwK{eL&=l;|_!8HO7V+?#2#U`BAblV~H!tL)vCA~^QH7l`zB8
z-8Q9h6x0~x5qgjV)5>h8Ex9>fYJ9@QSnvcheqF7ZB^<wjw=A_dI}+y~?ykB66jI2D
zHH5G=TE_j2_<$ZktG;Kd!z*2GCrw(bHBpK$4)EuhN;zkh980v>Bca*;2-iFt^0x9A
zL>Z6E$?}Cu16;H*(85gB>3F5#Pr7zeERv`sY0*I->z^krIGG-jf?(PYyP0?asiYBa
z&?qllmZVZ@t62&wPcqzyY|fSklB*Me-~5u=UyjIDZ9w5>gvN|h5q&VSOOo>;6k2Hm
z*O8?i;+=lkWQ3(NcJmjc(lp7c%A0v6A=hl$i}7;wvCfT8OD5Uz6==y)pQjzTmEtA{
zYjfno$dVs6PLedIepw!?2dL_CO(%w}mbM^Tqp7<ceTG7+IQ<H5QU0bAE-c*@=K(cn
z6fKJZuyZ3t@IK|wBp6b`U$rfd+MoZ|0jA6rOHTJ)+cS*)KsJjbcO{Y(oAn_<7aMPf
z8>gzQPm)zk7^B*KTrLMRa=l<IcNnht8~)OrLX=)Fa`s~mg*Rj?aLMn_G7+T5lo&F)
z5yWe!Jl|?3Wrf1UH55ZNXDp2p%*3xHAejM-+VIS;+223fKRp^56H@vYU7|=uf6Fm>
z*a=MEpn}sxkWjtmQH0^V;j?F2CJ7)uWJ<)+7`{>^YwQMoC^K?Ix5>O5G30Te1vJWv
z_>2v@UIb*CZ$+NO+n|{Zt1FHuKfI8#Fe%Rt`IGd>W5hezb}ZL%kSNp39>5he2FC;T
zk_%ALh{_bS>x<ODQP6Y}U{n;lc$sWhb$vKj*<{trcGUatupF~Lqmt9zc90f-f@MZY
z%kj`Ji!xKMn|>aqkQdA1YqX4lQqd@@7gZG~Ex`|v$2F8pS~~oSNK&)-{rY>TTv98(
zfy6XgrG_WJ%KAW`n-c-*hW2h{DAgH2!1EQiuv`C+1(<b*r;g;tAjIrs%VLH!K??AU
zABXhWY|M%xe1+QoGpbrYsZ6O!oNfjwLDkeTF8J;-v4sztuv@W=i_BCh4#kPT!~zi;
z$D&4FN>PmxD1Z_K6BJg7TjSzB*?7rcf@b*5(38c@C4=ZHcSh@I)Ry8QAcP;?*NV2H
zbILabbEJh%tMx}1RrSs&v%WEQSeo2O#X3?fGfM0e5+eAP&g*i~le2fRkq1fZM8V*5
zZ~R)nrjmoH<Y=UPsvXja(r5Y{BV*Pt`K6ik_!3cyt+pd2!gHS$YYhFjR2o@Fa~_$V
z<SrwTyW%`4QJWrBjdpf6z|i2Ct6LKR0ri*%Y05?-2K$yNJK|{Jz;CICL{1v-YCaLo
zBs&5s;_$Q<&4QJPr@vIF1yUZthrPa&z1Qk;2nkv4`-Wm!OvKn?-^g;Ectuo6DMS=~
zmcUE6w+zb$LAYWlk%TQh{maD--c5eCWVOmppVLmwsV~+UWzeXV0MI+)hM4o2-H8<f
z5mrseQ<DmQ8M0{)37KI^gpQ2-2w<&b1dMX)(2(zKHN@9(NDjqCT#N)7xmRo6CICCs
z*WPP&jg*5T#U2G|Mu*swujL7*keh~b9Rk^pZtKUz#^^C+p5Dou-xi8L7+%4;9LoPj
z_b`<^uLVWqam3_+fD<h+AU2NY4uGKi*$uNSl@v3*Fq8Pj-#30LCzF_$PY;N6D%M7~
z(DQX>%%dNmt}px85l8?LX}9$5w!>Yo<8&DnZsfG{V+R5HXm!$BgMxS!t1G0%5-c78
zrFffPg~~Bx-n*vRoY3rm%bJp8puT$fFw8K{H3=>;gCacarBby#v5G~B63$b`HxwJ5
z6Jvzg5?+N|F##G=sX<_`3jo3D`IGjf;kSivEQZ3)WtL?c4ql%HAw5Y$&;(w0Pu(&H
zhut85)2PAECKzHS#w6&}O5*Vbf9871sY_VY_m8z1M!!u&$Sz@$(WGclO(i&SwhBv0
zGqcPP$EDKVqit2DXt&R@-T{f4CU;w;;h#vW#9}CCpZVkSO}JY}lK{|oSfbOIXi_mc
zo=dM{#ZHd(Q;qb0gS?lnkw1S_<}{-%(G{9UOLY@Us=Rq00m~Q1QmroR7ip>=%mSdK
zy|t*8Di$~McG@(B(FP+x*XuE_*oEa%HU3}MKXtN0NBOL&`Et^=!?(~ib8`%FPYle;
zhnvkWREY&(<@1&F{8y>#!>KV^X1h2`Db+aHS=DXu#Pf#}xlWlHH=nM4{7wQ%u8^uJ
zS~5~Wf{gfC>X<WzRvjVNE#~;=JI();kw9~jkKgQuuYpd3a6!p$FF$SE;mtWuQ(k?b
zZgsi#F1!R0BM4jo&=QCd1;U6Kbx>kZ;Rwnu06lRis+?a{Sp@ko7@OesussN>adcja
z_j!RZzNfTMz&|{8CdGAm{LWh~5tZrf>4l<9#hS9z&zvrf$Gd|l+(1qc_57$vR1MMM
zT}!HlY)!>-qIJQRdJY)kuS<Xp*eYa0w0qUg&OKH;q!au;&brj`rf#7wU-z@B{s<cY
z1Ar1W2t#mV5U5M~wR^^;+Cra?&O{d<6dp;5kMFCG5uU<-VN$%S-NcJDRWJ_Rs9y>c
zC5$c3NCsAhM^qESLA<HPLT?rr@uNK5l@49HR3aGVv_8R$S4C@+=Gkkypy5R;-t~!>
zLrnz@v4OBD!-7qTZrg}$m3gk90N^A$uO0M|uU?VFFcjdbx<Ep|aJKe!EaVjE82ne<
z2!I;A{uM~@y;X&Q`Y<REmr6<fBsEzVMZ0HoOqS73h+Q5Yu14G73D&Xe5X!Heyf8S#
zo}2wY^#o6=Qy%T8a+v-hd=O;K9>Xo~vp8}<;eqrIQe9p3TNKQ{);j<_{?Rco36$su
ztKZE^2OsT?Pmwo2U3WqulAOa4P@_(EdK$rX+k+MhXli=$KV`Z39{SHkr(zdR${A_C
zVO)kEiC^kQ+W6*nYWJ|qdsnR3z7%=p_)l^vr!B$N$EzS73sccP@=K)iSjm~0%}}Nu
z&fgwI!yM_xVZnL;v_6dU|D+V5ly}iD_wpE6t}pFGoeM*W_w?h<4MljAf$SURt@e7q
z)JLA5F1jRV`gO|0_dVQ_i+QNi{?r7a8MBZDw&;o9dhz|V#3cD?Oa!hP96sc1?f&h4
zU7i7?2eOJb-`9qwA6VLTF9m8l?i1%6AIiF?iW&ZrnF7qMJJ#AgbYEgb&*%S)QegoJ
zOI-W+C5x=+!)X^Ib!y|Z8$FlFmZL`#qdiidMJl{p>kBGgEM<oqU2yRVo=JGowcK=S
z>OV3uax645HU71*$Xm13Gj%PBfRncrvEao5(H)SME_QaKSK8IS%M*#?)b9p<?Wks1
zMVMEwfa26s<&y``4ll<-y4vHfV=inPmZX+IK=E4oeP0Jbe`=q6tS=(g=%G!*<r!y(
zv%Eq~?DvLAaI#vmk?fB7&kyxva7kg5T@hjXlM|Lie<V?RzUenDdo|PN=eLMgw~8GM
zfk9G@=2f^F6@nGLvtiBF`y2bUUt5hX7n$z`0B1{{LlskkmydzxP)Fof1>HU5VG_;#
zS+R555enG}ezhUm=&U8lxf7vQza^ZYs&sdzMC8eQ<3se0wx6iK#+3NXV#Y3A$mb|-
zlyPIzs{zY&w!h7Qw$5ZWf7|a0&LBUxhBGkhK9NtuNUUbN26G+1CO<gMm$poiKcO>$
z0jkCE)#&_<s#2=Z8GiFpRq&?clCbT9?ygOsdsg5gP8pl>rhC3RO!)aZ_!&NASqDUm
zzkSc4k~mkw<&6s~gD=@V`rTrRtUSR$bx?_F{p*2$yyLp=%CdA44(Y4_ku$U4mwo|`
zD8{6Q)e=b@D4HP4FNg?MQNa(Gv{WQ#ZoqYcV`CnK#L!OMxamoM0D|`KeGw=+ZVSX3
z%i}yc8`)rLu8Ms>H8ChU<a6cUlok6_MOg;QRA2gqtcy?~oHm=sEJXIu9-$oyirx&k
zgIMhy|0Jtd_M$pawlvx2H9+57mhb0j-_x@Nw?LGH;D{8+XwRC<mf*anlyHzy3Sj<;
zcE2EyK*@#PSHl@hZ&K-cilILP85Cb{QcDxog^hH0Lycwj5A4J|HK0R7fTxE<BFJBe
zAYDaeheD!M#d-<Y{32<`)YA7DKB4DqmjLb6X~YijkJsk{dc7UOJ#jr|x@YkF&T~jO
zIUB#pC2U``Xiv)1Q|y}XAzYD9(bKsXvQNvM>x^{zwmgWA>HM0%JZXnbV2J7cW0a}#
z_PEt#SG`{U_oH_EY;g)V(W?9h!1a00G`a0a8hgFo?uyl$8Ms&#s?+|_P9!TuS+8BA
zo_siBjaA*2Lb&fTQ#Do@oi*k@`DI4l8#J2_?@N>-XmA9K)E<vF(t9{jp?f<{1Wd9h
zPI@9ty%r<VIDQj%@)NrF76n=^n*thsXOIw0)Sf8gCeR9wgI_)C!g(D5m~0B~6F>A=
z2aDF8^s01~M)>HBGqarTy2&RNt9of{PS;PK<Y%HFe)XKQ&UNE<<t8!s9S(6#_>TGg
zx>1)_<QqdToYpxKIU^T183)G>JKhht;t;T3>k*=^e#nO~6|%C)?$?u=a`{*<#r8R;
z+YgpD`-iB?VJrN$Ij)TvfYxyZbbY1)myKacrE>hUwWKcDED68#_6`RdozmbK$|ItL
z1+aJ#%kX?`fr`X>?f3yxHZ1}+;0Rwby~s=}zb{W^D>s|8SLnU%kzo{De;GZxAo((H
z`JrZxY)1^Gl>#X&L<lIMU;r*Il|$>g#p3}=gDNauryg~kyVPH1J~S-Y?+{yJ0Ry))
zp7-g8Q3PjJf9gX?_`cKr2OIGgN$t%LBU9#Dk`Odj7O-`~rBHln+^VRF)d|)_sY*(K
zOuMP4jX;Q-`Ps0f`5_-_?|`ZKZo>)wGTLE(gygk*UU$NRM*Rg@#{1l2AqXGJ+Kbp=
z?ImHD0UWvY;oV0b&#^u&(AbR^42XIpw!0r4S0-TNYQ-juA{AWoyFH*z*CnV|_jdbd
z%s;?Zb*FjC<$ljgIiO}R(?|U23qX<)^pwGpX$wAHu?YN{tC@Rh<lKXD;*wP+teCW<
zWK9;(A#20CDdvhF1+|B6!V7zj&6vZ<KtiMKLIB7HRr&h5!hTj`S5f^3OS2FtxN${d
zy4m}X40TIHCNX18T=xDEgb{0@YqxTUtzu@&00k3areMox*Gf@XBwmUNG?y8m6)=Di
z9z1_?=8y}sh=i06ArUa(PapVBS&JEa6ivob^OxIr<zwQ!8;3VN%2S^O+j73b^C_{i
z3R{Y$y_U8N@6b-ESd%ZWgLw(B+OL1mAUZ21V;VGi?{iXPox7R4iSDW)?Q>=#^NfKD
zS5q*DVjs`yNSnh$yF3;!PjH`8Y`wjo2w%+xVB3=zA|!8EM%`oMaZ;c!mboV#`aXSw
z!S%kC+%ju1&aX>((O3A|tJI(~6#X+a;T#%iz*hyU6fp!XfOyNuTNCNpI2&+N3t!XY
znwTryTeh#G4lV>M8TX>RMj6ODhg7s|zYNv2VD`hK!z9`rri!>f;~T{xATF*zRE9fI
zBbaQY2fs^mRI-nt+$Q$D*S*$C<0vUbvrUG?ut1erseG3bd~J;Uv~$%Whq|uxhi!^P
zWrIdEbfh#?r7*?@<G4Dw)jJG{KXf2S;v4K+xZHWg4R5x3B36wdK>zoSr0=yG?B|XK
z+9S<=8LdwFj>f_r4RVKsC=PY)ERpj0i`Zw&alrbf?QCi_tfYRV&0NAttg#AWLA7IC
zo<T`lUhu2waRURBwgv|)q9-pW#}`_^2MMxJX001$uj4?bR9*&IN)R(I+Uc&%A6hsN
z-8~<Zh}&Egf&l?xLf}F{7I2rgr#sn*7{=2laeRp%l<WdwcpYnpaBp@{@rhZtI;;Am
z!7DCgqPU<o=r+mB7cjrqD<&0AsQWr6ws?@^B+!QagJ0y2ZszfWWv8d+#K65tm44vk
zP$fC7Tu=VP8)hWXUYrox=g!cV<yDaJ6QjRVpbC)(0RASp>w||agrXy)4t65x)C!DN
zXV9gJ5UAk(T8vKja@nY-Y$XLs9JcEEZYLr|vTG}gZ+a4$cDHyHI*^OW+Bz{Ac&^RK
zip6%XaOL(5B}S)h9+Ih~*cuX4^R$1&Ya}$ND2#I^<;2&TthPQ>2q(QM9m$<N{RwZ5
zbls)_hq0Z@HxT>I!haQft<HFCJ2(FQ6P5MyHz5`;y_<mx7)ud#s#jiy<)9LZxpfB}
z*L@TVf7WetpQ80(zw$<~4Ya6}?}zA*e*|pAwrocoL~!p$bQ+K7?TD>})iIfSE_}!J
z)uD}OW2)OD0hJI%cDeqsc+*}HUUR7A`FD})k3?NfK}Pj8>*ckSkUp?cbyj&c!%Jnl
zhzMR32MdJ3?I3)Z*U9G?DS#p><Azjm#H^NL*xJpdHnFA-6Kr*!A#t>h41BTYY(DY2
zjtjJAEs>_47$h`NuAU8W;K&@YO^{&)sel>lT4U8keEP8dAjHsI2)l)FCK^QHCnB19
zVaZOSKHwPVj)p?b86nIaaV@*-MG^4c>N8}G1QLvto2(g5e?}SP&*VIuhXhL9rYflh
zoleLp#*yO$M_gY?A6{{@0h}NDF{M_`AxasPha@Z(!CK0nholU^D^rA~%8EpdAyI^u
zIv|^=D@_wvX*py%cLjb7adi_3r$KK$&kzUd+#yNXU~g2UmQh4XBMd~sX}PQbX`2;_
z)tS~PB%rS(2`yOFzHJ_VuBN?DH|k?kVTzB`VC2VcTn)woUFwzKk}SrFdFH!wkf!QC
zP}L&Rq0=DJf+qmM2`bD6v(OFek`W*(q&CBc601o0tc&&Y;+2lwosz-OTRU=sL5#?C
za2V*G{c>6~D4xLdXwkQ&yp^KhE~A(+@xybeMwdhU>!$i<;WA-KmEzT#!(VJ2++GW`
zcHZoq+Rm{%#sc!9B%-9q)M<B$1MN%E^>!zF+f7hi<|kjFi!VcA4&dt~BtT)8;A2<>
zHytw>8XQc$KT#&48kEmpd5-_BER^~=g?jpm+MXwOHP)J}fs2IKn^LVGdCOp*Ti*{>
z+#l=oE4A<xw4<LN^F3Rq9!DXi#XUH5<}PW(r42vela>Pqhp<=Ru|WoPDT25?n!elv
z(I){LfOu|HlvZJW+^iODJ>j{c4cOI08WbS3mBh2h_Gs^j%FT^bU1GdzAw(MT+&l=|
zf&4Tswez*xex;bZlx05>E2p?I0T;hZ*v=&3E2*k`Kiydoffl?}w+D@g@0c7p7lSlE
z<qH}wp6uE%w-NL)(V5+G(=ch(^7EzI#i_0r;8_)iuv5>f^WU!gE)*~+i1P6^pkAs@
zN)}(_sb;5+ni3g<w1SfXDt#Fv36p5)9a|Y!9N*L+xhw=O6_o7a+dX2OCjfE6{k(e%
zapF~So+kocbOw`2C-EHw-(=v#t4Kn#_C|6!zJ^p!AP#AzHw{$!kBgp7W=DRqn0+e1
z2ok;H&`bn8*ln_y^cSO9@wMofuu-c)lCK3SiCL=wHz=7p_0Nb1xQz_kh`wHt-$loN
ze>>hjQ_s5;X&ICi!z5Pl(7*i#iO_;ufmsO$gASofiywb}@yFL{AP$TuUTPE~HH#hC
zq1bv=_LQ=n80*EwCxisfY;N4nry>%vESXI5n>;aeP(1Lr=5+WROp!9JXZbCUoYx8n
zaQ1}x9u1ST$cAp7*zne8o1eu1Je~#;>6=Lb2N<ZFqxNC*oNII5Dr2vzxhEkaZM-ca
z^EdJzLo13T<4}IfaC@}lN(6cw_DqM*OTeyv>WhlD2D@xwS)8ayHWB@}lNKxro;As6
z1U7VvC&U3Q^qe2F9|K~(?`q9@=6PWS!jSQc5~+O6n$@z9)XQ*LMUUS`f#riPZ)u(Y
zkj@3Y@Y=FD&0()fJ`oo(VWnapR&gZ=v1AkswPQ*0Odn{OCfg6Ptx|F7EAfX!wzJwl
zsR+|33kr$?V?<OVw~ow;?fSjdxjRLNw!Gi)dzA!~aJGV2w&U0^k!mkAqzVbpjNAa4
zgl?3!2?umxF_8vi&<R*Lbn2VGcZB-@2l6L#h)(-aHstwt&jwwqy(q{r7_BYZa-oVz
zNWU`~qHkqQ8#k?TB@FeC(Mh&;E^KE(q$WIH>v`k2x@q=IhxmYL^T@1ny#zfA<!bhB
zZyLBrT9f|SILB~S8WQhZV!Q2QMdnB@GXV*Di2<pW(Jul`cQFSyf9V3DhQciYJKMX=
zn*u{816CK?*^1(5x2(YIEv|Be1~EmW%#=pXWwRS{pDH<4zj!ZMbELUJ+G?RL@3kma
z`2dBPbdg@EVqQtl!d~z+Oju`_r}sgL7ji3su1&)ydXPuZK7Ii%6Br&00UNM4AIdP`
zLzek{*Q0J3_YOp-cisN<)chTwGmbTTCC2Cj*G`o?!>VPHe&u!&Cez?<>vwDk@Med1
z%HCx=yKZRl1>$P{iuXo;i)Jg_yQ}gJPQP+=*K5ajr*vbW5v+%Tp?YIMvx}<UzLUj*
zt)na04+Vmm^E=gkF{$=#pn)k-l;r|+0(P%32i&w)>i}>4VFhGe#hDp^N*$)k?gwE~
z!l@EHJ<UJ^GE`)TNM5nO6--&url3LkDC#k?I-h6YTkyMeyt((X&sAPQkWkLD=r>|p
z==`8TWHdGjK}n&y=7XKeyERK^Pw4L#2`UMBABL4GmB<NFbM@44s8rvINHvTgRAIQu
zQ(J_T<IvbqT{c$p$P~B$WkSj;vfNIFuR#t%{IuwfX&ZO-s@Z%>$W^Wp#ryg~k!wxv
zLPNjPGIU5%s)hBq(efkig;X2mt4|3UDb=7Lib+*6@K+WE73emW+{scq2@vIi$c0g#
zC@DG#U210ZiHD40<|e8HQS-xJJOyM4@eQ}H1N)lYAYnX^l^8$(>U*8DmXBqbNGC0u
z+0DF!^xIi&I#*$<P*1_`LJNYMm=%+CFD-&>SAWZ62VJzE=fbak`{b&A>TR(#SUIHM
zk;@dhnZOWClMfGjOX48XIyAc+tum{(80HTsxXWNuPxrtIcEA2u2#@lD=p!%Jf4MX-
z{szP<==vUy>#zs~h+cZx^*M<ixj1ROGp?A<Ju=(X^eQ~$i+3N-w)D%iTxy+;G2?ol
z8cTyyLG7XOs-cguv!75ZUA1$Qx2Cht(i`$P_!zm&jn>Iw;oW2(hWQcz3dAHDl8wnf
zqss_W(tjtK$=S^RJmhoMmt8s@zr(f3X2~;3XE-;veY}SQ;HImzpD*(p?!io3orhaq
zF9TUBLO7IjVn+m(r*Prz2L;b2Eq%GmIL`j~+6rU(?qR%xANpf8clJ>|l{54II<`$3
zt&PI7a}G40*!K=I$_dBj!a08nDE?)#`9ydZjCEwa4_}WPcudn58YV&Ds^xR%hQLN=
zmL{L9$9Gx?u$SGd(|S!DGp-5mV{K2S&TKQVq6Hi(z18$gCX0ApMk63dA<#*I7A3c0
zu;f&Q4Vt*p!0c30Z}EQk<Y4=EevgwerbmDGbSu^I@A%%mK5tK?4?37D%Y@H6TQbgN
zlz+wEF`td8$01D2d#uRM-S+s?TYO<Xx{?LPBn`O_pg>02DZ>;cxJcnH{_g(tI{&6{
z#eao;B=qE>h&#L+HK$O+DWRZIEwB3W=aE=ab2#5pN@e(xE`GW*IxDj<zw*wFL8<o+
z*3-CTljoiAkkEITj!A)1+Vc*VBE-01w$RLy)Gtxblp_{3bJ8Um*ZWAxxu_U%Qu;T|
z<?_oq;9#_TItjHzq3t<EHuSmorbl$-53eqU2$ey#8~$AW%*ukT(z0pHqn!Ioezf>u
z;~ytOvzQ^Z7cbuSo@Kf|Fb00bT#%eVhrh8d<M<GNv7a1Uu#(Vjry1JkhZi6ZZW*?=
z=(1j@1AILTcwLw#{{rGS7x3iUK_szdJz$a+0I`<4k0`xK1SMywa$sr&Wx^WCun9>K
zaUNlS-AIgQoreI`-@o-!5ctzng0oB#vIZ_s>M|E|+_14#9vnrAV1ERmU<kvY@)^E$
zZe6sN(-5P=k5aE@jQ5mHPc>(vC_*xZlwSG>3C?gRIiE~6&R41f0_yDP`JQ18=fa5y
z0RpL)R<%$x*i;O$c77kp4_iIP0|vFLJ3T<cZWsHl&({-5T5`vYv<~86e))`DQdmQW
z*AW9^Bo(Mj?*<3`SGYrFe&}VL)$u|*#$N^wn9L!|@~EL4A=95NNo|g@ZEAmxppH_N
zzo{>;SFEa*M@L}KKW{;wT4=q>W@53w0qze5Lj7&*k=ln5O0jT7Yr=F9E53v=U5P_g
z(HfUMtR6D`iR#GZ?$Ks6LwU*T<bofrFD4Dhp+&tnO{>Q)REt+pVL05`r3@NVrMx$)
zgYZ(^3HJlxMGgt_8u^1uv6%;ZE6#HKSy;GN#3y+l5w5k{zAv25E{R4jbZ+{EfbUIF
z;@iwQZ<>h$8+7QK!O2G9`GuQuzi4RtzWu<Sp+o}vd46Zs`>=C3{y=fuzDDWYVH}Y|
zK`?wHG!l~*L3)|YK9rJXzq;TP$7f$f(_$ItT~6P+_9BoMxT|`tN%T2)@sfhm&F4fz
zQf*waV!3N-boN3p0xFI-${9t`2#B-P%sT&er%1vqNkE$)>gn&9HjRkuP7L?QZib4J
zgu^B%of47i(1-C7<GMB@Vyr6`wGP*`3AZ4dq|a``{W0i-QVDcdsh<ts8M4>Q|HrHn
z6acw_F!GMz-A!al9Ci-N$$any{*vA^f6s*7ES6_o_srF*8icaBNOk`?4zRnLV6;2E
z5r~Z>ttL#DY8WK9>|I}8J#~TE1Ht}1X}zcRyF))k22$1kbX5YSZbu?SQLB&TIqIKX
zUAOH8JdS>8`_7Po6JT1B$R9T;w7pP+h1h*gjPW{zF<))OjAOuQ+WmtXtvgssCVNPY
zlDu*Iugozq?aN9tKqUZ!Kd0-9Xm{Bdj^lUgY1>^TydQvSUB+{<LA*OTgl6BlRnm2j
zN!XMagiq7QG|h$~YI?^^PlXL>d)f0YCav9lA98?(N)S+d*Ey0>H5X}GbizC$*l)@M
zi#GxrQI(et!qc)$L*~QjkELiVVzFs{i{{mA7-@U`R%R9sNa%trbh2vTeUsJpuwZ7z
z&X7&*xSO~MvSjy8^I3*>dnyza%49XVl*hb`Xc*3bXOPAL+VbZ|NrO?tcbpw{LgJSu
zHEb}JxqBVKA__agUPIx0Yn>pmWvjZYaE+n7sT_w$(pq{s-I2AoObgimK9;V%A(jr2
z@m)^6Y+_CjkPcy*0-lj)4+xh@F-5iav5+LjB1$cuG%Kk;=k^%ipy>|D<Jt6_b*(KJ
zAAC$wUeq)#a`Y}_iC&#{Z@COx*kQPILW#sKan^m{cnk$#;DJN&A<(Bz)X;n3js+)-
zS&KjOX|jM;6tlQHr%Wz)wST$b@8Pr7vQ$(i-D+F`##K76&m0#*S6aYFIltK+O8WjL
z2#le=;ec4Jw)t-qJ_Rs+b-;pe!iIWT6<8r=GDft#DZ_mp10I!To%O6SoFVz>5M#qU
zL#J(aZ%vM_=*(L;eAa#I#W#=iV+QnbH~fO}JWDI7j;>k2evAot0t>X6ycCO4(grfu
zc7I4f^JJ%6!5_uWW)3g^PEPW49t*dzyOY`T1V7cKrfQ*{{OuY0{`;W}A=buXm^=2|
zaET9yht$mn7Q4wjAWGriz*m_Bs$o!}I^qY=z85570hxV-+eY48f*0&lh8g=~j(gll
z7$ty|7;06J=5B&1-w-Y#jxsbPEhcINP{;wmlz)W7qN#BFY08K!d!N?E8}qyM*o@i0
zrO{?Lu{1}Gvz;?OiYUNmk4QX!*|L?Z%|$!_&e9QqUSp%i(R}|WLQc9gUJaIc2crj`
z7zaC-*=LK-GN;+UqN4~ajA$t$N?xya`RBdiy)pK#fHm}Zg(|NdtrgeeTz=_Oz$rj;
zh`M7b+nlBwsW?ibjTiXxpt0or83PaKGW+#;>i0a?<@tG5T8IJ;83u!dYQoxb5CK!C
z9m|<h3AHL#JtrdTpZ1Lj#3(Q*tv9n7Ax6+A)8g6<8}<3HYnD916zI&RqOsyhl#FMd
z<Xo_Dfg4T9RW1W{FKLHSnG%9MTL>T)ah)-7{aGg>V3A<-v(5+sU&k_Q>JVQ?GB^(Q
zq9*r1hbdH*k{G?;iOGTF!KpqW8q!F&NH}lo5$x9xjbw0xRIqS1i3KbSa(Rig%IIxf
z%>EBB(yQLI<d8rx%}hQmL=Cc9T$1|gznx>B97QoL>*~2=6kUr1PZK+guZMuYM^}tb
zJ)QzYRlLM!2Lfz;wzs(y+vAtD2kMt6IKyBVN<Tyb{?<7?h4xAGet}ipSCn*BLX;J0
z;J<syP=2ql!r{knYP`9H;dlyc#gCn8nv)CYZkMTh-{Huo&eAxcC}u?PxNz>Qc+(oV
z+0%HSRM<24B`BCmrcnWUh}_|b;gMK7bL67OWb~Ma$7GLb{y~&-!a?#zLG9KfkTA$;
z4NAg6enw=X;WbbngE|Q1|27j#-n!HfS{$&7!%cyQ_`NgqOKomvhSxVJYE)xZUxI0O
zyf>{(z7G@|d}#KKKvwERvq8e-9)$Yg4xF)^A(@ZZP0xiZHb7>VK3Z&GZd)+slk)f2
zQ+)_1-rvy@%fCL~4n7|8rv1+JKg^#tckYJ$JG#AAs`Okt-VB9G)uG^=OalyY`oP-r
zPEM^}ug6PiX1yNM;cEdOqcz<d-|YlCd)i+Z+q$~50ci}lc{f7~&wRsSGtV1L1n*sM
zH?4@4y083>AprWL=tfWT<<($HVeLc#>W9{cwQ!sIgJpSz6%S_TX~~YC^Whh&CBVJo
zln?I@+r2ZGH^#PhY+b<Nm3$8YVWYudEZcK+h93HljMJ{69-Y+z1t`g9l=GX5R-8>c
zV<7>Kf_wToklmow7>nK3jz>jYu96?a2ZyiYM6}%a@c>@og`>{r;7tMCR#N9jmvt<S
zK?;i7Uq05Vj7*{IOl)>AkA|j?W(|-1GnK<?EX;ZEe!I*IUEm=Y!TBnaUNd^a(aQMi
zFiqtJ_^q46(s)I!e3c1kKCF<X&6OF0DUi#gaz$w@1-Rx_4@lv&#)W#7xhSA<BTJ|Q
zvxF|zVgPz>;kMV}BqRJ!9z9M@414+HxrF4X_^CpQp$2KoC@1|RrR9yPzRwckEt_n0
zj|h&Q{n|t*e#f(Y89K2%t)Qmeo{r_*Er8ckN*C`bBr5n!XEe^n$wR~n|D8#904_!F
z5Z?T3$%yXV<WWcY6e~jB(WU7!>iov4r_T9H9w1hXWyQk@Fl}&``&OxmaOK<BFZDb>
z^xpA~-&R7uT#|92dlphO_<vPWBa>kF_|-vjFl!O3C-az3J#hq`cWua$+bZX^?g$$a
zXg`vZGd)n>NF!<iujbTN*{aHTe6gBjlbE`I$jAiUvoZfbCR1k7NvX{br0WiVGl_R4
ze(ZDX-FDoM6CA+oAppPhvsFBq^x`-O|3lzrWE1)Gu&m)7bEpS!e_`>0RBEsDPW-6p
zVwcOV^KkIU&YuD$5KsF=A8-)N;fc`U?UuH(Nz-@U1>Pu-D?&;PkbMK<_IIZ(wZ7C%
zgqkp433q*EACy7j2@2u$62bwXu1Nt<{)`eJDLj1$>Au$+68A&3gk(#rq|5Q!N{3zs
zXtgE+D^{I}Xv<A!58~@=+Ch6bq1Rg9t>><g^qU*!AM>Q37??Jh5u3~pHO<|DcLn2C
z)wI7ax#ofp>AM@~@tlQm&vLR|_Iq3yL#vbg;Aj)UcQ1<th+O<94z2(c=l9mIVTY-m
zD)C`|FutOVJx||nHS+1D9bN9rwAR9KVGj(uWNi%n9RnQ4#3lwIeng~6bF_Wq`SZH-
zg-3e$C^%gMM?C{g$z2o(nPchl><i61@j4X7`9yU$VEn=Hk3$DHToESrEs;48K2HRS
z8go+sKK>4Z>1CO+GGM!4lay)q^l`mQn|?jX4A-$7Epc#^lCqOs>e?jRfZ6~eQ1@F#
z1!z8a)fhUoQ*W{aVPc}2*hJ-7yW!7&fHVFnQxJQ=RM*X1P<x9}CWBsps}~I~@JrkA
zt7^pE_q>mA_|qV8?Cbe-`BCcrl7M|s;$!G{kkgB<FT@I9IEN;sh>Z*H`GelUL1zdf
zv0KE#zI2sC2=2Bc7w*}3qunT)VEb(SlME}JiByO=&PG0R-j>`XqnxTWb*S?v%?j}6
zncoYHcwqc0)2&n^ewRAeDM2=)08gxt!pz<E{l>~qc8s1S=}+qd4$vrYQkGFGH(o=~
zB3ceAUtePY4qzM}Q$dp+H0M>+g;V_~N&B`mol_&EFd~Bc<A{ZX|GvZd-dpHCfJrLK
zaCzK(Ik8oIiu`-uje8wO9Y+{x<4QMk(kY%9m@!hv#-JOkWB=ezEX9r4{t}Zf;NAz|
zAbs-&Zan1c>Z$biq`Uj}W2_d6GDZ+D=Ek1*?GHY{3C&O-Ft^taZOL!&q^dG+LB2lX
zq<QnD4IOw&H|gFk=%%!F%bMl$i=6GX;>2R0vN#h}b_L5u!|=3g{?0EG4-9kwXaFf1
zg9uM$1phmW%1OUZ=X**l;i%xCOzjx)DykxwJYEkC9armn;mE0Eraa9np+W{zO^h6c
zygD(!vURo!v5{Mt4NfX}9)~(a|7qZ;@8;&JXSt`QD8?mdr`+k_8QSsJuH&z%Pyj>y
zKpqGN*$~3zvtmJr86WTTAcl0Igzvp=W`AX?>{G@3)q_xU6};$9{nKFJnD4D$0ezFn
zez;V<N*Gp=OUL&4Bb3X2_j|Y+>z;%u>!J;yg9OfG<osdKgYNWT9JZ}2cD$m4GX66%
zyw3RrKH0+|y)74r6UPAJb|wL;JK4}gC}GVGM+$X#^Jy1lRym8ew9$wq)wYK<@DQ<i
z@>EjXc&6U0UY%VRKA_K<S)Z#}1!P!7s(wD}de`cjx0T?&-=ToWBv{47d*Qt0I|%_4
zPUlRVN;-E+Px~lICJa9l(KI>CN;9=}|9!~l`W)me2WzIEx@irs;rDKTvU#@qYrHd=
z;{Lxy+yKAdi_d@;*pew>2gmZ&r(gf;W+pwZzPkVd7GY|p-Ia)ZBk(QBrBzt;W>T5`
z;B0r#isx|ND1S{L3V?-~|2YU?s~Lr38BXSWWS`X>G+wHpr237co6vv=d}v%ELc{RD
zWWjFL?GK2AV`oakTQojQ<K_L?Umcmc<}qRO(P3w)!ai^NHA3KGh${#av*TDpt2ESt
zx32eL`r%r9c<iP*bB6-O*<jayO-{%fKk-f4=6jK}E$PU$*>pkGi3Kr0_t#3pDEKh~
zqF%%;GI70;wb%je_k&+17^te&XfZf(@vxz=o9Y8p!(?v`@aWvC$8+k(IAt^W!V+`5
z`NFN+ZQs$k#qMc7Vf9Sh)&9qn5$Z`4_|KG4){U1fU#)evbV%Ahw9<WcARV$nhJ5ca
zWb^kMX_u4B!{a0-z)tO)<BHYywZuNgq&DY+e5}Atr`n7n)wNtJcAz8Nc^tlYl2qbt
ziOD|KmQU$+X{X8Ct0r(i(zHs~#@TB7miBWAh<~>I`P!v_(TU6e#xXj+c-GbjgN6J$
zfO_E;JE$#@4G4!~PN~+w`#=m0B^on!TKp56Dl=lY^p*GqID3?Dq$t#x^lD4q><nc5
zkw17^&Qme>?y)%$v!?AAM{3+aUd!%8zRebT1ohR!k)!xh*gQsuKk$1B7*g%7?}exz
z3#_*fxk3(k*L6aYHK&nd2uMdlL(qnl(`g}4#?$|~Hi9|dv_rt65v*D95ju@nY>n#T
zxwd?K2`cCXu|c}7pi=UI?DLgQZcCZ)DBhmmoRW#5){0fGZ%iP2Wk5OW=n<Ec+YeGc
z+!GZhaX|VKttN578U0i#62m8nKui)VCe;B`Uj<v4>BWLHL<FT&`~f&kMD~t^4Hw>b
z5RoOVxHbpr{I*n1RmxD2U-^B5Y*+^l77j&_q2*sXZu^Cjtit$D=@VXo!yh}i|AR*c
zc=E7O!7pJ=z_>id{+c}Oa8XYdDS)IB?JE{~K{5lkVFEJ_Y#%z#xMFpx*$xopDns~L
z!gtIkv)nz#@QPE+h!1i&_@jzIK=FLapYXm6$v=mGpwiIun3r3+ypH!DAe5BRPLWVk
zDHp$97cyBQi7>p#zjR~^J0m!LMEU<cm8aP&%26dI3_lL`1r@X(V#3jBZQq%0Osa6F
zOp4?x|FPfM_aK0$A~>2>@o1PpDW@4M9*Bf6-j^-GLU_%*gjcfR|C$nv9~;@dUw^P`
zA{phjT5J6Bl7wITFl$wSIgBuSFB_c?+W_JGPjqFkzzc6(^p1pTm7q8uMLt?2#3yV7
zFxhfH_>vAG&0<Hd*r?(6rhdEgU&(kU?p;)-m3iyy)8k_OaurLAzV=Ypvy?p6F`4Q_
zAJ)00#Bbybdh<iy$oFv2W$U$8G+SK+$>BW~tVVqd8dWbRY|Oe<eKV9VC65<MbCyJ{
zcY55Nd`O(0W(0)vZTETk6B?c&a}pvgft{x{k_vwR)w4e7#cGvpSSpuE=z}B+Qi;bJ
zV_~bAiRj=qcmo;J=Fw+i+OF;g6$vJuoTEg{d;>0JH~UAcVjJ+q&taVcbl2l)UXx_7
zXdSG)<<ZZ-WmOHyHTKkT@4Tj9EN%nGIA|gM!%(8%%sq%kqVp^Ou;z8`FKpq?HSt>a
z_J`#hxHh`25qPw}6uUY!wEHe~<@2Z9j+Zkw`2bm|o}vymY?`}zp8W$%=7n+L*GN>N
zFuO~cP0?Y_BG0CqfVA`f;p&~DGl{mS(b%?aTOHfB)k()TzSy>H+qP{d9iwCG&pGGb
zG46lIe5g?m^-^Qkn!DDXYtF#jqPZ>cLZ6h%<AnJ&zWuY)M#@H>h2xRe|53Q~f+(SB
zFb0D*DCFvDuYTjRV(!{JbSHpAd}fHv6!aVVmuIGXj{TG4wU(CMmkE)Ab8X|l$4*8<
z$XwgC_t~n+b062mD<3S?_0(BVihl>74!>B;6<QyeaeZH%UGBeD6%zXw&;b=w^o##*
zDQ2&kfi)RX<9=%Si#j5*V3DlX7XQb1j`?(D)_d8#HRV`u%5>|=!`WRH>`r=!=>Mbl
z17ZEY^?p`}9n%7C`p(;XyPIOYNG+KB)R0}p6B9Vp4Gj<t4F{x=Rn}Kb@i6Xyxa4Fa
zS}P)25%rKNI4$f3mY^$W&yXrv3wICo5ntoT_65Apclm&>i(qhQ*=Gl*e=@8ZRowXQ
z3q|T)sscso-23M>fyj8<<@Ej0d~X`1)s;?;Ppi1|-C~?z;eF71y8n;Hk+<snzx@?v
z3dO`|nQ-WG2?&-7Bzi!py$p2`+_}(yh?&|2`+JDyM|ZGRcIRo_WmY&5@2=C`+Lpa1
zHPt2QIrTLhAf}pJW{=&eD1kqg3N*_$`roDh-<2;n#WU`Y4WqY^w{FQ6+T0jry4Bh#
zC0%m;;I+80?oz2;xgx(dWW<;;?Vg8U`A>#Jl=kaw4giGis=5|Qghz8Z*;A_UsaXBV
z`u=is4}Sr_w#V0do#Au2qucxa4t?Qf*J16x{mEtMon=bS&r3C2%5HgOnL)d>@r+B5
zw^9@4d)2v;^W8wUHkZY$U;c5_vw;+M0@qGx1Gs}UscB+{?4lX0oEbWE1V0St;s;AC
zxhVw`22jnYR-K9)gH@}mR>otLBfOwo{00MIjY)#KK?&B#P8zG5N)?lDat|z`>b~pG
zBp~s2oShWZ59)%7sID<4Wmm?(2ioc}1Y5y7L*65YsTDq(AIoA`EPTZ+k~m97BzQzp
zO6trMSRgHuP_%YI$32TbPlMch0w=9ImW-tu09-_)W+&gg`x^RR(3+()%Kk+PEA$&)
zV8VT?YT&>3b~o0DJ~>luB9iVqp=5T?^cVM?TAZR!V0Sy3^C(=#Y8mJDaeN|%?V!8Z
z`v^XH7w~zlYp|=x1tabPCCdX|mRX>=HxIQ?ky#xPP?rFaR;sy#l&)`9cHZdLXHEF_
z2Po)%b*A=x7l1{V+>d>uBlAok`SV$%aG0d<FH(%Ssv`@L5rdxEh7F61yVM^l+|HA|
zV<xh)>Od5x@lYA;rGFm+8!`(1Nyd0G6x|os%r>grwu^&hWjPc+{1JPE{nS+Xk!2u!
zNoL)(rkh37QLqjeJvC*GVA6mZ5|vqUK$dWczV+337LZn%*M?bErF``~b_ky1X#wKP
z9_3NF66C14{faF@Eep$6d$OHDwKbTuI1#dJX?7DqQ)ag~XT(&$f+9Lq4tiu-(@bb%
zCoF37V{DcE*O%Alg$9jsl`x7%9(jyRATNUj%wJ-iHd}F?0l}PQo^4OY6~P>KfCBo>
zle6M(zFc5b9CP9<wP@5KH#|p&)YkFh1+Lhy9r9OFMKc<u4nkXl_29ZF+iDWR-OS9N
zOp3@ippIek1|a=^<#ZuI38qx_Uum3#5o`y-Ib4+Ei*|1ERzAXaLyx$~7>?}7(Nls_
zgDoxia4+axWpP{8kS~%e=Xu_2fFy<TII`=JZM;g%cLD}xjwRuXY#g`7e=sV6u<?II
zLWxfI=yD`OuV6&{xqBfo+Ina-HLxZOxGOuWoQKnyB82WRoC2@ZJLg@^>n!|AwCNdL
z6~u*=#hl}*f|YbN5Pydi*aY{|UXab(0}TnErq}93I)Ynuf`p%R9j9(p0SZ?xmb^z@
z-$@$wu2@L!+j|>KCw^o)M#8Gp<lFp_dlJ|$$OYnml+Bt^5Gs;VX!PgdN5J5;POMP^
z%l~)4D~&Ux86PJ6Xu9E|5O`sixMvC1M)SK_kRG*-gWCn%75WF|ny%%*onSvGO4BR2
z|6rZQ=37~AsAL8t>M%9^58w?_3vV38=CffaNFghkBAEu#E#;@9^k{RWP8_V!Vn?ex
z77ajh2F~F~yyMiSy<;vSR}HmV-eFkvwdrouF~d;EhzcR6hGdpcS=L=H=?(N=qzZb%
zmL|=jJ0OP5LlxRJI_qd%_#M(gm_Ncijn!lXLmmLgO_89GTC`5FyhOy1*fd*wIJISO
z4puF+ghhYyW5gK>TA8MOkp<c;&&{6}%ezn{cOt1ZA)w~OAZMGF+wxyc2}Gd(ZIS%{
zq~x5%<s~iBJr!%wrQ&{9R`?~8y{V()r%IQ`#d(v04H2g7lzDMg7ywOwj_V0Q+Y?*~
zW<wg$c;<RzKZvD*YB5@veIzcq+<FrI7|$lEeBmZdQ*$>IYY7p$0L??Yv^@p_+yMF@
zB4La0JmFH%5BLZPk1^GNgLva`B(yR^`@BV9|4<olXC5>KGH0G7J=*zMMc!ySIe|g%
zN#c$Vod;GTaxqaKE`Xk}Gz|EJtpD6>-6B^6ut=4r->3$gdQ#${vyu{>u5vf(PwIiv
zwsRROsFG^^-nvCC!&GzP;bn^)=AeG*rky{OgXqmjC;~DH0??54j1ick9S0@PLy`*b
zctC<mX(TMSUi0JD9l=#)LNc^t(2eUvkxev;f~P7CuKE;SJ7D~?k@+>(yqjf!LK!@v
zjP1g4oO=2?f*F4J7R1uGN>Wsc;>k~O5<ye~3i-r+5g1y{(Ov03aOtMXKEad=i75YA
zHK9A-wm=&8^FWl!g2|j^BqvPsH!&#d$MukuK0S~REjkQq=X6h;di4HJEotPrwc1r4
zI+m<8o&=N$kEmC&;^oZ8-#VJg-$S!Hf`(ONJS?6vp}TCi4wG6GP(CMz>#^sQE{Zwn
z^3(0;rMY$2F)_P13TUvh(`t=9ykAEz7v&=K#;nh7T%KW;V2$GtX~j#Xl*E&;Eb!Vj
z;S4_c>T1Mb{|<?#z+YPZC$DAC&T9yj-ajZ28VC3SMf$1PH%4+NL8US4P}yR0-x1q<
zzhpDDJ2<|G$+y;!`j1UT@9<Ey7CjEXANsA47uMOB+Rx~`7U*n7=zL1#qCo%9p=^2k
zoaY`inT;%NCG%pg8}IjJm`4g!A4ZOKL_n@EzhhPo8G#Se6c`ItLzOLKEW^=A4fQ}q
znaWBA42b0u>=+bEdX#vY(lht`Dz2iz{*Mj}xlt4?!;?uaIxyhoASW9R(J~af9<x~l
zMI54eu!_xf_n5&cdixm^Ru{8*a9oo!YdjHY3jo}j7`Igi_!(!g^|F_C>Q`+h8)x&V
zGd{*MP>5RPYHUs$5GsX<om=@CQ^(3-(_w!Vfl`RWBxcm+^`=96xG$(yujwCBGza3C
zD>gC&LzAwdS0p{TrB^PL>G|IWWL}$>DkA0I;F(|KmoPx=5<!Dyr}#_IFt)D;2(i}K
zkTG-k3?STK2AoC`kNQ|7ngAySjo%*zsYux$V+(e%r!8;TB}AyK75FG11coHUBp4jR
z#-%VYlgJ@VUPeHh(kxsW5sQSgfRxjwDk!4ePr#1>!&cNk6lV%_T>S?N3m6<{0ftD`
zT4ci;Qn%w!hpi~TNKgbP2`vmLH~|FPf@w6k55V7N4E6*b59RM;RHRfuELXp#{ushW
zMh=zd=>DN7f)n6kPfy9TW2$Q}F{y;iwA=D*y>{XJPLD(!^Iw(Zz@cyPbre@Kvz}gE
zXl$@>F`UGV5hbkj42Q1n*y~I6FaM`gx}>V(cv3KgVyX;_WNKsV%Sy|ZW%@8a+{ABy
z#|kZ}wYUyZ_(Ycz95>&a9`l{iK$Ru}2Q-ZH0J;!WJ#417I6V)@n`gk!YR-)BW(z}!
zU_>kc6NOHf68VTSSqHC8`+W0^(i3sT5oI>q5kV;(IA(3-%1Dm8l?Jx!-R7=-EZn=&
zZw_S-Uj@;>R)*_(l&?l`j|9KIsoDYXUx@bOeuxfS=+OXm?QqGB^>tqVhrNsw>utFb
z<2in(ybiD-K*U<Ve#3_vXkd7sKIypOmYy}SWLQ3VzxnCu){#ndW}Zuwb1XsiUx!d#
z-Fvp~#vF)#KRD0nx(A?TjV)#;vO;0olXN6dwj(d15cWgx;N-2V5$ine8~~_If0uBt
z93pC<%BK-u?gf8Ls>J+7HM{>61=(qZ*c@-7&Cx4Fu$-Z(1bZeTDD_8?I|chP*&cn@
zg(1y?={`qlv!5cwgoXK!Yz*Ds>wnGjraTU86223;>KXW7SL^>()cDm>XY0MH<Zfgk
zAlp(&W&d~geP74(n7Ix3@{2o*muZwpCS5(n%dR8+Z{LbGB`8MnmnfV@v%DV7;Y>z!
z7_r7u`vT8u2lEQO&qwA{-ZP!vJ4Np)*g!~L>7(wY>kiu(dU3Z!gaZ+2gJ!}Qy`EOS
zm}F2KRT<*#e5~X^>WX8iY}fATg90gSZGRS+Z?fdjhjjUe7l3g&gy`?4MznUQGQnPC
zM9IeDqO*&R%LG)XVMXjgIF!9aY{oFC*0wf18hNzfe>#8)(9pDhogbO?GQ-(s&P&br
zg@aBU8<Fn?(<K*(f`s9cA^a%~NuAz!Sucl>6$2nrSp>8+>-JFSkp;p#<2=g}Jc0C+
z_w~l9alJCu0MYb|#>T1E73i(NPGGpFe*`awY)xE@TkAiEY{RAJ)7FA(*MDyP!}W4c
zoka(R{(YNY!)Bst#58|T3}qJGbkjwGL2Q;@qOGm1N33?HHrpw?%(OEQLlkOCilSV&
z>`VXv=He(RL0}E=4W3Ky`*?9cR>x<)9i^S6r4jl7(<N?Ax!sCCtjs=<XFAT_Og@<{
zOCt~THLV&JMKoudm_uj>D6waoC~tmug<)qpuT0x*DDhlQ?F}f<_i1TIUfZ%F&O%I+
zl$>2LpQo)V@RQ;`xmGpp1U4&Ao?b2mcsiFV`%mm>q32<N9YT1j;lseIxuXRTf^6Hy
zDZNI3Xm?qXV)nNgan+(y)`E@}q<k#nb%NWHq?`Cieu$Ma28Np*hz_&ccha}5d&5Q;
zXV2~X87G;SJV=Gk_3M^}2a+#`sL3Q{#snIq;rFWTbG&9SZK9FH8gyQ=v$y>)!^-X9
zD|VeH4+SSng%2q5Sgl<Z^H^*c8;<W_vYtY~@#{=Sx-6BHtkYf?tb+G?WW1DBdXB2&
zR_K*2rs8`Soc<>x$}M^jql0qP8I^4qOTK%BVsX3W_O1qT<O9e`bl2u(1!N~kW1WzE
z8l{5SQs#yUOFe$Pv#IH2LI$T^+rYDjykf_DbVB_-gX6;4<y2y$^@cAKC1Vzu@pm3z
zX2?|E@RS}Qd_N&YS{wq`z?p~hJ?OJA(9FO&MYaQQ`^*2ml`MhnSs?}^nTa6n@A>@t
z7J?h*d9%A-B&OtC#zIPS$@iY$K78*A@hH;Yu6mWy_Bw~!R^qujzGyNa(WPBB-C7~r
zh7l;#@kMHrTZv5%q5&R$oSme#lvgwWa}4o5q;4+hW$*_=*6!e4ptG;W<oNZkH26ce
zL$|ZOZRn%Y6kae^RQ;`_FRDoVKBj92`#mUy@Y}@{5T_e1_J^~LQ#-@meYiz~8M08Y
z<5yO=q%~nuLV4O?XZsw$`?&E#hRj4t{yiP7&cO13Wj(oWWlAXhd$nzEH}5OJv~Upn
zyDe{SFX834GuNa@KI`;iY5pYj&XTuB4Qehxp8@{@9N5Mp9GtkR1X)NCR7etUC7T9n
zJqoE;r9qk$E$HtUM88W*`?@k(N!Nbygz7k3vRNtgpI=g6G}r}mGjGCCKu)J7jH#N}
zBnvM>4}4TTeR=B^%@E^UHV4LlA+(`uC>10nMvv^lrB{D4y^!LwoLiCmQ+H~P-0{M0
zrqLlb0@2i;q3M%?AP=)xYHUXL9hwCbqENPSbIWI{_NCLT@@ikgV(2`R+l)(fB2dGT
zwsP5DY9)c<Rp4;St#X+HvF=}yuqDLP_EkFAK6MtPH<TI5#GN~w+7jx3PHYpSSMxg|
zYo+~7Ww>75#QFEJwr968_hz31dGJTbPSvjRW8}q_f1D~G!5jZLbKT0T%<uOG4<yvr
zyOwPu7F!+CHT5}SiJ(MI+_YWNk$lso)j5`$D+8Zi4K*O_Fd6I=8yjq8_s;F3L<<@B
zu&OTrDTPClFu%9t!jU%sf12ihx6XG~WNlN!VuZL8nBeJG$RE*p5|d-89CgKE;%>>2
zV-&XxU_Dobq0iLlAgs(4Y|y$9L-S-){Zky`(|5;dE1W&ZDb<CbCUgO+L9i}#M1m#^
z#*eA%t>;1cBLxTZ&Rpw@{20$2BQ9AKL`pS!tqOT$PZQ7(IzR&iaMgRd@&Cx1NgVTt
zL@5m}^%(CHuLvX4?UM5~kiK#@MLBwWS3ZFZc`&bzmV3|PysJuI>Ub{s&>Y1BA4@0>
z6Z^=H${Ab3bXdY%N=5g1eS8))32IUmW!0@{6u9QISqZ}wM>rs?l|K|DiH2GwSjQk+
zC)LyE^JoVb@+r>(^yGnSRU&`dwtCFVp?<w&C$uYUV&Fmt3s9dMhQu)}*E2PWoHSqP
z^RbRWp>>YYAklfYZ%B3pIS0rbNW?kEIlHn2X_^Wgrka;(vy{YV8koYM5ucet4-C0l
zibrCpAM~~1ji=j83@Aw?%@)vgb<#DVHsl>A1Q&v;T^0QUP{-~fHz4Uw#05VI$(Y(P
z<Lkg(p3w%cweXE!w&b-LM5||i#3bm+jqf;1^BIn-1K2n0k}mxVve+TXlP~c@toMPY
ztDI6!Dds?`%D>`Ugrqb>1_t~u-=5BCAkFs4bNyXmOym0>jO{o3!{qrgWt$0`D(1rP
zonNs#aQ=D$wmC3aBcX=3tPKy$$oT2^#hQ2wsiu~6?If$2T~1#zT@~IDmI;P1#a0X0
zTJ%RfB<-1?5Z39Er&mxl?4>@SP&BZ&j-dF!ue9TH;&SHJg=c7}&4+(jo9>YAjO(Im
zmB;Q#0Gj;Ep<k<?{C4tsQfspWC@!nH#D#G#tb-X)l0q3V)Cd8=yp!QDg<@T`@Jij;
zgUn#BDXEqtvFn;Y<r(Rk$2L4#uUO(Nin{A)-ZVgudT~k#!U5;di@h__3Mhi{cBM>6
znifn(at}AnT*Xw?Y-!bWjWyMWSo=-gSsTRJJ5n>?Klz7rTGCoLF}jB2Y2lBJ_}7(C
z{bMuW2@8!@6Ynb#^j`0}x&ITL3aaf4CXNVf`Vz_MECri(yW7OA#fOP20)lQn%xxfM
zdz=`jT!PAM61GJeyd5g#vTHT9Q|yA;E)c2`Kc9$dK&(HrarJt#!E_ccmQ79m#O(U`
zLSy0ETZ*9vVSEqZrP11*Icre+tcX%U0-X<Fy!VuHS0Zun(GW@sdQJ$^Tx3oHPYQ0m
zAK;(|dkY+%7--r4j1(z$N=yGp0N!!tc?ivBk+}VZo=&Prvv++FHu~An;zsnc5SZKD
z3fZ3u=WU|ADBEb(1}`InBv`|5wK+tPy=Ppn0Gp6EG8yB!b3<v_qML!giy-(VxWEMv
zkVxf7f^Bom*HVBtnc7DPA~|`2F7j$~j=fP|0BFnuY7>$XMb=JxWfV?Ndo`3B|LPE^
ze7d23y0ssHd_JGg^9|6pf6sV$_j|vXzR6`y1CYsbyV9@0&<;Jq-F+X^7FDNyVOqWu
zF&_7aHZn<2xGNL@qE_c5>H?T?dL#kFQClC0RjW4C(oqj}oBXF#_Tm}yK{tm%Y^ymn
zK=AWX4sYr-;73D0zP{KYJIA+l4b`$A@&vSjIpA=!B%7Ju+{dR{Nn?#|T-I)#a)NN^
zo){MpE<4<cYLZQpPF%|QXtjcdt}Gfhe9l;_Z@hC%x)-<kPyOSBc~3$^Oa_1`1vr{6
zw8&@CYcS@XBe*<DePvfYhuPI|wzlElI|o6S(Cj6w(n0q1>omoisz^H|`#zx?sH*$M
z%sak8WlL$0iI69MYHb9VMju7I-GKHfDa(g`IIJ#$ky=z+qEb)oY6rY)<dB=yJldVh
zFEyHcB{lAa=H9OddAT_l3K4)h`4q-kJX_kYRoKy%w@ES29=S_cuMje7VWaL}7`c*9
zvN9%x9=1E)xF%j&VvjzzHL|83C9B<-2DLPjQA+*(e%L#9XH$3e8APROQBxrS<!!H7
z5X<eIJ%6>X>jbyj62?dDO)xgJFI^aT6sqEIS||gTI)(9;q&R&9SOl1ykCgZ6-39$d
zja*>9wI0~@Z1!=rGX9lCUPZWco91|zS+egLq(_Mw-R+)JFHUYsCgQZ#RfqU#_#p*1
zS@?W;X3b%q_~_&~Z~N5QX&KT9tHqLj1$Fd$Ost|fA-<m=ZBm5k22UcAqCmTT9Ewqx
znfg?%y9Sk^4R0uf%?nUePj5Ko{xrLM>~ud~+QQS`)=fb<<dvb~)xfx5?w@lV{Chq8
zy3C`eQJuwgJfbr$$~==OxlaNvOVwU2pK1RnhnLX1bzKdS<?_zftVr^Ne{`%L?oO3!
zm-Srk%3<o2gx#*gemLUW<MrU5InN1oPO561#K)r>X}#XwY!g7oQ&NGGp|+)o;Ovn{
z0`j0L`EY^M7g9!I!hCu-{ldtAl5g5TXzX_ouQm9dNY2SeHfD(p(jAHrgCob(aW%nQ
z^9&3`1)OYy#?STs1Dk&>7w$sAs@pD&tK85*n$AC-p4@EmCEcYgcr4}5a?5N=_|Z^F
zBfiQC7P9JYIzX7Y5WRSA4Or?xZaVdZ=l0cnGjjS4YM|vr4MYg4+Cmg*Nf7By6u$v6
zMt<DkuTnZW3YdyhH*xdGQvP4N4Zf>is$NJ2-fB;{+AV9lS9KPZ9*xCRdCFO5%e*YW
z&jlZb=02WGew<xoTAobiC4Wj;z7Wz$s0HciRFatXG65eSsX8tX_cDU9?bdD;aDLcr
z*Q?S;1=b`y)P|81g0k;9*DFwOFbSv)JNCpm<amd$-ZLN)+C{tMElUaWsMznWsL@O~
zMr;F;8^1XRdcShkQTMhX^~eC7s^)+luZ^iqM%jd-8x9Joy!$5}Y#`Q~=`MH%iBA94
zBR_g=M8IvsFzSKU#&BYX@2%>G<Yc$IFlm%HRy0$>(>u=oM<iwY>-%yfB^6q?4YY8C
z_z!m2#bm&>ssa}C?>Ag(872%4Az38eA?)pY%6Bkq$Ty^$lK;@Qa!Nwm{Kg`Q-jEhp
zq7cYRZcJuDSF!ejsWcWm8k<nCE;VZkGeSNBtW7ZMa*<Z5lIzBhvb#=eRlBNSvHXx%
z6$sal1pzoOaP|pSo|%{Z6U7XSuh54hj5)qbOkCr{v=%a$(b#jyXU>Hw&K%L838OdS
z;i?O%`kUU-9?;C*k3E!&anm8>({snEe}y{Oyel}8-fky0PIu3MoqJ^D9RDAep}Kj1
z7*%H^kgfvfF-u<kzY*U_mcA5*+AlNALUeU&so}nl8_q++L&G&?UA2y2R)XHs<v*ch
z5nX{@iR8=s<PV>M@yPaTE5Z;lcApsvXG${W*jm|-qCv276dAJ4%kyfxEyNF2M5HsG
zBpL~!ZcN>grmDV5-?L`zBg(W%R@{~V7ipqPP1xvxnEyimG6N>kd@=(yJohL}Z~&ZT
zRGoDg(WWSdqpNM_0IZkD7n>eIG_p}1Vo$OwNr^YUFomEq>TMk}omjNBiFDefp_;3T
zqYLfVI)^7Dq7m(WM+%`xN~z@=Io98^;Um_X@W!2I+a6!?&)L+&gNrygz*noA%Vy{6
z$D=1Ac|J`75(hbTC>hZ;=G42#p3l7h;{nQm4uU&<$!X2CTj<bruJmiX?ykNb@yjxZ
z09OK9e=6bsMR+|eFgim7+nqmz=k%cTx7h8-k#S$C@%O^zTC>48p(M`9cZ?%(qwb#-
zl|d}0HUmHlBElH~!!~JG^24s$!)~SN(clf;+w}q}&)aqK($7=Q-+E%-dbaI?-`$?h
zKRb?@;0=DsU`}6&K3rX4_eo&J`(5r>#x%rS*<OprSH^(DM9zD5-klOvqS;>7*;<0d
za9N^40|hL?7s>+4Q=*VhlP<^qF&&5`lE_a!1Ta50@$}0EI*PBB%XGl73XvM}!a^lk
zh8A1nAG4zYZws^4`D>38K{cN#KS>rHlD?yL_l^wgXkD?9Ic6OPpg!I=Z3D>1iX_kC
z0)1e2XJ@WrSAKQ88SgXg2TtJk%*_35V^8rj4)mbb(<9OTb$8W$H?ylOBaplw@&cIq
z^?Pq3PC0xc)uvM5fQpbKWLt!&dB>wKc)nIWdGu~_cSr_u`a<s06LyUZ%??LJV;f-H
z(`4|<)LM@%+g2{0j}jsBX908L+!_-D<Z7$HwtOXWpsJWmw|94FV^)0DTlAA$lU(f=
z{;kErC%GrR(A;^&)DK{<Z0LJ95&=RFgAP@xoua^ke?X2Y)E|E=$8MlpR>fSBVt$%&
zRtq=}`nL0_X6DtV;ZhA%Ls1-~%I~U{?{orF^1$Wsl7D&=>*SsPD)^*#k*7B=2UF%$
z=43mUCv?Zil__4~jx?I%$#<b?cG;jCh*-cDm4}TCv;RPdV8TLPxnd><Y>a&gg&SFP
zn19Ui@=%agSi$zlMb$s?d2#fFxt2??g`%wHD9xrRrD8R<U){tu#;%)6vbnv2Z|+}@
zJQ<rVbO5T5nb*KgSn?4#!BC&V=|1DR9&TK_VHooDlffRA^Ki04PdYn;s_sbY9t5$o
z`VYe+Y*ZExHMN6Pl*YpV$S91e2kh*txXM;jS>d;wS2<7k6?|cg$FYq>*kfW4c9BAU
z*k(I^BD__8mF$v8z21wavcDIZt_hkDz{6&fjtwUIbQAcn%kM$ekVyWnlY!WWgyBjS
zmp0Pj>VEJmZ&|2&$~;lHC#&P`i@783;MbI026Gw&q1ocpU_LwoOsyPKv%ua*UNd>1
zYr*YupR*VYR>HsayU;Mj#viCY!oF+dQ#1x)C$G*9EK@QM-$kc=Ti2g3ubGx;c(^zs
z_o+TH=RJ%Wjai5F!JLkEJ$kU%9KyC_bi+R`c}nj;3bt+DAXNs_WR!e!xykZC4me>%
zhYX2BXdqyj(AD4oNXBP7FroE5EvhF61um}6Glb|}lZwpT(Rq-%(&ibZ3hiPNLSjPn
zy?G@01z8|*D*t-(^9u4(3=|uo(aRLipRCPvCifDcKqwf*Ox)wS)L;0+yk6Z>Fmt`)
zDF!rW4RpSwgCB$fm@r5nz?oH<HT~!ZHN&v{gp9CcXli2t9>OIbb!y@2Oy^-;$^pp4
zwM4asaab1M|AEQd814U%7N8}tLIjO!CPU(pepLLY+P+GbOSvblN<dDPlfvr6QKEK_
zN`k8;A`ypcD}@;m42i{W)->=1c!<#wA4%{bUoe{8k-*lmtxp1@x+L86HSsr2IEkNW
z1VBsO4l%W^ubU4Hsiiga1^D?1NwB?}kBf+}wWwa=yQpK7jLoB*pdIC@I^37$Cxy4)
z?zfJi?o;9Y%n60NB-YQT>r&QgPuC~s`{|0Wm;*rfGk%P$M6J+pIK|i$lmu~hh#*lx
zGw)25jlb5R0d4jnyF9Brw)xujVzGGP1#pI(C^lpBD&54#+Xzgto$Pm(aud84F&}ee
zeKX89XJ2k_uX(mETeu)*K=$*#N8A#>NKn9D$hVr`|Jt#%fRDCQZgy6vX;Bz5ka{<&
zkF=`c4zbc1%0arG$N0?I8VVg`MhL<JCxtTu7JHm7h)^M$L<3JWvG$5M5LI*p0N6&%
zg)q5Y<(}a-BXYBptu@w2G}QONj5ZPJ=<v>bzLe>Pz+C~`#6$b4FZ{C41>4*j!*N1;
zIX-CU&+KJ4EGuo?<d~NGgzRJMN#}^7CJiTwzLFTsbOq2S;aF4g8OjAAp_c8Y6`y&d
z!8F40_;v9YV5_$;oXq7e*K){~fcE?+SekIE645GJ|7N?cF6?HM%D>s2Y^Ti_@E&&V
z9`C>HeeSz8^Nw!xh=J)b3P%U5M9PjrwS5rF;(H=PJh|H1@%0OZC!-zf){fb6GK}h_
zM!{PS!@eD^a<tOw8y1uNfXOr$OnjzhHl7QDA|CKMj3$A&#;S&`hrYu20K`j0#zHT;
ztPmJui@iiX((<`uSx9Lr>J3S7RO9y2golkjjiS=9H7=24e=S#AH8h6I{yGrAr()>%
zM7Y-Sy2ch3H+Av~f|~Oh!#8EP2Lwr>ufhQ*mK4G5lRkI75x0f8gQtUFUZI76GtB8q
z#Rga0iS=a3lHo(&-aWc61Cp7;nNU4_>?motgc%H)2JJS9(-&YQI1YZ2HP+2jRZyJF
zu-%QSu=oVY)VagaQx}U1)OQA`e`QSd^Wd;03&A33T1!IgiXN_n8;y(S_m7nIMJG?8
zK#VAGOdSeSdof>v#C!;JQiL3702hP?5Wr;JN?zrco#yx5DJ~U<0#MeG0$q{({?x1q
zfnVhfxiVRyJ`tvV0#eU$q5H6nhALoirVAZePMyG{+YNsuGss8G&r3Vn&f&O~mtbXq
zmbBjP;h`|}4WWqx;{5-@MfF2s2Oy61^<iQCMPOlNqTDpi{S>um1u!>QBvi2(gjNro
zW!&@)C){edKLKfq0T44FWX5{1(T+oXJ!*URM4{T|u1vrRcrI}ztpa=p6Xlivw1F@W
zJEMnYt>p{2I5C@Ts6jJ`S3*x%PS{Y-zE`XF=9*FebqgwnV+beIYED|Ctg@Hk#`~DP
zI@#aL^1jIatajrboBziQ+*VtQS$(Wk+lo)!$8Rhq<NsR+4Zz+_lHu#%<>Pw)dhzJy
z$o%mZoVC3J_pgS3cS{%kUx?fOv~e*odruZl@30z^rT0D-;@AB9-4#u&)GozL*`(Nk
zwPg)w(MlSR^M_n#sZXdS;z4A(hoBePO#4Z;h!u_(GHR^E%{b}N3lTR76JjocR(2GY
ztVUpFX8+B5EnuSl%_}=<JvsfDdt^4lacg~-ve#bs6?rMl3Yw6gGN@i<&{9|F0%RpX
z!9W~_aYI#t8~d^4!5dG_v`GKYQlzYoTFLZZ6ekOzPkKJ^3u@_TwkQIoJC-{Z8fRjd
z?(l#?FUU~|*>%zgeM;*?@	J;YEkvTv$}}k03EDrT`@%U2tx~2RlYa0Rw}Avx>Yk
z6t%C@s0mCn%uIVE&9+Lo_M3zn(Ff4<HH2q4m!8<zsIxx~v-H^7N=*flEkM;&IC5+m
zf7mJw+3JUC?UzJ`Ud0nJ1FvXn7X-L?uKg1I)|&pYBw$P-Mk$#l=ulA?MAN)Bo_*fc
z6ZLHNBLNgnGA`ziLzJ=q^krMxif~Cz-JCCm)TEn5e`Q&%v@y4^K2P4pKP{y=IspAC
zVsK()AQP!|d-SR6{l2K3xb7WPtKwO&yt+2}5_9bJZ#dhHxMosRf?0V*`1%k;fuC+S
z2n$(x`zcK<wZz7aQ2wwjpM7{5Cd#A$qKwCf9DvJcCsvxRrLB<8wcG6rJM!BWC4i+$
zAfE3Z#=ooU@80PS_cCX%?0Yu7s9kC;B)k*^TWFKmnPZOo`))%%j4I2g1~z73t#ZiJ
z9vFPMP=nCJd-nB8#!tk0lHa~3Z}Lf<gwHAHFUPfkL_YJl`a8bZy1Q|Vt-E%T071{g
zbHM#`CP=%5jELuEN8ZN==nUrCise@0NpqaHZ=-LcJj(YAdr7@akMTyvYyBtBetaZ^
z!G0w9El)MUL(<@rAvJwZC9>|;co%w{*o&aNynC(!bLIO`<9HZ~jj8h2kGEgA3_wwW
zbv&x0##KSjb=RwC$#~Z*(l@_W&o%Le5D<i!s@#@D^LG0?(4;w7fQ)OQ+N^F)FLr6)
z<GpJWG-zo;8eyi;WJpL{fshp8P8@GxbaoJlTO3<IVVrCNk|HF5DM{5<mpQ3P&Sg|l
zx&93ulfYj6=IE9jkeA)#=<W<5k7WFa*kQAQC&2t`+pKckDunelqP_bPbD}Re8W0ru
z6G7AfIcH&WefhFTUx@R(SS&uRaMHp6Be11#@uy|gwzVTT>FwH+fC2tGroMDiA+q!S
zvNKp_HKvP?>l7bvoayf935dgb)A^dB#pz@#`Ybk9AW8fg6@dNj*S3XaCMO>yx~HJf
zLBI2NGmYvU<=z%Ab%pnBakDM<3{W*mP#bqws}~t{!R<ym&HnIfy1f^**htrqh6JUQ
zHm9`8xTjaBfPD4N?ToX*YJb^z_9frLdY*}0*BA#wM}#Rgdi|F}%xMEcS~Yc$F{6y2
zUdzo~ml*G3;%@cO!QSHq?4RNOh2~Q!LZsSg1BHOSijH0V#b})Dnt6NEJpgFBMTnTQ
z*T=>*!BxtFEuo)?07r4>;k}`+VU#e}<A>nOqbei=gqiZiPG$`u8e1Mq!Jse9^G|yj
z2HleqCVId?YF}exTQH-N{KNiG7prEeQSE*p<{);95%anB@<1OOC4xVqC#j(JKMh8@
z)SFy!+{e7kK--ut+ZhWJfNdvC#IFc1mmxhd4YT(H1YHJijgoj(ATH3o(e^(vqm6Qo
z+@@_fLg+Kaz~x%#T(OMRAlEy?qQPQ6*xnhNlM)Y>ON2F<GTokIZ#0Vil6tgfWsQJ}
z7^?_7z2x>d+r%{m8wQT!W}|2(e^HEjvz%_({c8M2n58oJFHjGV7hmc>#?dSn<Fm7`
zYN%&qF_NUO4SxK(Dqd?F(h91&*jhmy;T$fCvtFt?OX$oBU>+aZgJC4O7sNF0RVmr$
zs0y}N5uM<Si_^tFX`r8vO#G<W;7$@G*c1FNAQMP4N}lXa5G@-lk1He6K=4nQYM1^^
z$`$*qy$gd4(Xj`Z(0Qh#wCh<$7Q9xcxyI-yxO2?F8w6u5(0Jv|48X9zQH}YeQaN^j
zMq)<K&cBzWw!hgq{L0shU#sy${jKyE>?+3JnV>uw7h1*y88@TfH+;}MU7>_Th8lEk
z?GXSm9SMht*QJjhB6Js@;9b%140=dIv{jXio$kk))ocU^n;6226D3s%Ba0x@1_}?s
zJtZTapBhs{2W?e5lHs~BvsN35cs~as-1XDsy=w!m*F5|dkxM;j;S6Wy`0sCvEI!Xy
z$@XH8NHgT^u<QWlNos+t<n+r|2?X>LfyAB5gm3oHf?FFVcP)r<ylDSR0E|dg1^xq{
z@1N7Tet;Omf3S^_dN3l?zC?FI+|+Y3Kwl?Gh+b8J{7QAi*8*K&rQT%qWbDpNo*VNs
z^3{h7CV-_TV^c5mRW*IwThz3x_-wEYFu5QJ>WrKu6`piAlMOj1V_>gE&)CUZyoT+N
zgqfavCc|$L1=VO`2Ol5BeIf&|3&g(I5IzO0iv8Hi6q8MBjErzdfi~n1SyL^y#5r2m
z#EE_wO@{xeLJJdH@(|StKTt|A1cwpE1|wbr#>t0$C|BZ5IWUs7lod9<g}kF2d|OO0
zMmlwsRG|FjAnhO*V%4w^#qJ}Au`QZu0<%Z-{-2Xlib%gN&<#*s^E@KcUkT?hHeG;!
zU?9$9#YvdXWb^T80e&$*8NPl?F{Tw_g5SVbLQ;5ARW*ThP}BGS*5?-%3W><TiPH$V
z0+1pFyH=qWsi2D!z#ioi{4hvk-39feM}~+$5W|2~8^`JYnfVrX3?N?;QMKupZOS!I
zs~g&jlk1`Ys53X%5oaXz31i^}fAUe)rQ5$CD3|<^qti%ujhYAU;%wW+syIkUE*$sl
zMR#LW3gT%h<7tw7<|OE_c2Y;*k@N*`AY_30g=U?L@=n5Y#EFPcJ=9b17Cy$|<xY?6
zi^y-&H8k;UxH~K>Dc-x5E!<-Ij0j|Nlq{ywzf(49nPQmQvXF0LOt96D4n+Z6azdyD
z0!Y(rsWuVHCY<k&Qwfb}=O78Ir}q0Bt6S7U_CU~t*YL@EI&2imHQ##{mqkxoP_Q*4
zC3ls&9htT+>T*!Jb7c39j#Z`o^J3g|p;2C8ds;CzIiMl$Fhxo6?o-CUdH1M%IQecW
zr2XveXAe4Bo$vdLmpmIPr*;6?dB7V3V8A)cgC;?9B!(pN>Zh<Z=>sY;$ofHURyP~4
zc0Ew*8*I9EB3Jczojsyu=x?_=rIchWF*h&zjwoGaATjIMCj29Ly9juKoUOU>mKR6R
z1&~^gqMcQ!x4k^!Ept5KJtFAfa1FhsFpp?FL1jUG$f`=v{&2E+%}s#N@&0~niRCvx
zx@?S~DsR%tcUrxmk}U60JyEc-DWH6iWE{eOJH&!VPp3I5Sx=|As?S)a<1+Jz!5(v`
zr2$s<LJaN-<YlIFzXf=7yln$QgKmHJ7onc8qaa<^{j+959$#ue5zCA<@`g*Wx-~B7
z+r}zqD0!LY5#FT*KJx(eB>tIAnG-qevMWwDnq0thjw}xUZ1g_2sttq!nzyOxCL-&!
z$ICzF>o}hUtj7MZ4m*O+U+bNn^^dldi{}JvzxsJ$ZKMzvy`jEr)~L7mtq>7R_UED}
zxV1Fhtl&mVu1f`KB=l-q6qIw<z8gPKgI;Gy2N!q;TXc;T!j1u3jH1`rPbU~ly`pQN
z=O3tAq$hv>adt`rh(@x}PZrY3-d%GPN9*z3Y;<4CYVjU_eIe$UrG$`uKTv;@KF4t6
z*6>M?KmB$1F~sEr&Cl>(0!wI?YEk-hh-fj>;*qV(%sY|$IfJLoNlS=~q<F`a-(8_#
zE7+*_0OjYEmm&m&aS))+8p-l3JV3HMcC~wboF!h=yFDSE%gVmT+iiV5Fp94P3<)%M
z+5~%$N#vG1{gs&gI#sm!P<n6sh>s0<oDXLvpt-t8O2Lj-=WGTy_=rg>*h_#@Arorz
z@6_G;9=}8OCQ-_yr&tq18g~t8ZW-4;9U9gv*MsFySHlFn^re&Z)hR6P6aA~Tr(25I
z!1Lol8EIEw!LyeVXRgV^YWzhtLNzicR4Q<y`iw*JBql$!ZsTz%$ZhGSQx1Imf=>+D
zU)(STjP=nP&{^KT6Q^x*KKI`~()z}`kT<gk$#F(uIz=LQy%x4qen8QloQGeDQM7&^
z6ctzkG@yTrnXWc}6IQ9YkWC4yf(oqUihWHQe_E*7M|5+a%#TD-4cU+p$4Sn*R3g-H
zFiAQtf+;%Q?^np335%Iik*lYKImpj||HRgLlcO;=SzQJG)VgiUFBO-+0lU~`m^K1=
z!0ztp4TX%nI1Qh_%7$Y`y*pZSmWzO&A1+@wp#l3k*ojp)+bk~5$U=Rf-AiTRKp+IC
z?bVSJCaNN1pxta;o$V&~^x;rshyISE4>N4Mvqf%$optWk#bx>sl9e$QCRD@~x5=ZW
z__gD(RlD%VwHu=j$J`%Kf5yP^dNzFxV=f>U-am>p!;ZL=IJ01OogZT|uJr9BN~|E;
z%mJ5OTXjcwRh4D^X8c$SpCj3+W3CJJ+q+(a7qhFWyYY92Bd-%_*aY~fvlLe=%17&r
z*gHvo2$;)9%7oy#{Vd;_gltQaeEJt<jp@xupOTPU?4c=9pCUWV;*rT&H}i+mkYar#
zaAV=+2`(dN33Dw4*6<Hl`sB{P@xK|<J^=G!`w6!J-`vKpy`7x(*?h(uPg~@;I1jBn
z6aF2Q!3zvGVb-M&I7^vdo%sp^)6v{6W*J7lj{NfF+}}kUnW2Qee-JuL@tk#RikTnH
zz|o7YjhEhCShSgn_zvW+g1ISDt{mp)$F}MvD6$>)=gN`t&Heuxk9?)PfV|he0RXbT
z-`>}s!U*Lk0y^l^iQY3WH+N5ydv6d|i?n_Sr2~}Xfxh<#TXyQi$ubBMS?LPhJ2B(H
z^UQ`}59P;yI!3=@F1TeMp$ox9%&o|ul+(z4S3o(XFJ0P1Ves>Do{O!`_QWqK3Cd26
zPpwmks?)H7FyqH*225|2{y8@uG60+@f6vxNwF#B95~@C1@2gRUGhjfY*kN_EOET(W
zi>LUgwCLh(4tQW%b=1!)>@PU9*avGcnoSx1<0l=X*4cgq;8G(~*YhO3n7DkmMYRE!
z3_SqfiV%O!tvTV5EL?7qKN!GHANE)50SUVr_*x<0Ep>!hiaU^XakrakZUH(Gt2FjD
zFIonBI9t=7XaCW&l1-%$PmzbV3qjx=*E6Rpny0U5caXyg%i+CAr;(EFyu?Ta>W0DA
zlxpz^Y%vaX1($&eW}hHp)jaVDIPYP1Y3HqE)GTrv6z#&<K|b2Y{&lg-EhWOR<r{&a
z9|3U$&S5scypJ<s_dN(_VFgTorEraMhudb;w)Z%4!|D=l1V&O2#YWxY@2+FiWtnZP
zu&(m)gq$c%=Hh-)M@e>Dbi4+sS-yfxz!uQQtT=HpwC~{3?!IRry{Rnn-Ai$8eUMQR
zL};Jx3>fhq!~8|ta1e07yIlcKb#i<z_mMqWa$B~E<drY4gYxa9JqBRDH1v}`d@R0y
zSn;~6p$ar8P+a=dv4rt_^5-{<nlj=PNti9UG<!_|=B~+SeVLxzMnfupInQiK1bX0+
zp7SYBm>4{~0=vh$dyx_0K~|sQ774s~+5Px4%oe4`ZY0Vaw_M<msw9cnqdYquOy)vo
ztldMN=w*F?Pd1d%76VEuh4^=MeqO(&tU9M~>4v{ku?y4bQ48p8EbK2NU6kG^SD-6F
z;;jQtj>JijEP}vcQj>k;s>PuUozfx5=!}tGACPz01=NK_*i&l$iO8q^;r8Ayzx<s^
zdGFJ<6kMvvSJjw&$Qq*tw!?N~rM)x96Px{r0?oI;ZK;MVdjQ#YGgF$*$Or2#V<jU#
z&sC^dTE-1<V6GS28huNJdGC1d&E*{UH1GN~bgV0@hjF=X0_FbL9wf@yy8BenT;UAg
zd!;Z1?%}2<0{1oELx-Phl_?vINg;$RDUL@|HslPh(`b)*p<`;Kj69lxQqZ_N@%fx6
z{>Qy`{3t(sB!HfiU7YEk@!=h-EBpw^m*oblneo+F;RXeKO6+Cq*@qpyJ2ZdokN^D&
zlA{V`$*o}c0-DtX6Z8U6>*cC|4yrtT3Vn{zYmv#h<FP06x}hL*w*QY_sx2o#UlKx2
zX`YhSgNa#J+SkPIm3LGhs+LYGuM_U!_L+MnZA^De7XZXQaD5or1s>E2;rsXZjIOm|
zl#K4brOhcJE4W#91ya;NdJvCeqOQx&hoRx#8#R--WrUbkcxjy>%Fiahqm9c8_EhB-
zSGKFsPjgBZT}4aS*_?Xza)u9yRL?D!!*Z0CwkecCQ84&AB(aoP?<sroh|~rwZ|Zz4
zq2#B(NdRsC&Ri(|C6l`nZfoS*A?w3Nv==hrQJ44>ZHlghfM9X7=&~g7C1}t5LTqrZ
zJh{%7Cgi_0!K0`OI}b`;D4TlLPE$R7RExKR<XaOA&<V&h-n_^z0h2gEOv;n1xR>&O
z<n*o}V`qI&^6~UZXM8DMJhN{Os<_{2eI;AJ@Bm$UC{A>BxLM;!5U#j~GKy<QGv&6j
zaNjB{>8Ce_G^Zn6gUfsU%~VyiCQ~ND^V$@enk!~1h7d{jRM6b^2_{}%b*M|6WC~g#
z^YG7uQScJp{u(V}vU$B>H%@7nzup2}XZic6Vi<9TMrKd0^5b*Y)Va0ih}gTJ8g@)N
zuK<!xEXVS{CNVg88CVI&?ua7)Cd*A*efAuwGk$&MLzS_O?SfLy<g>m*K4Q*&fX>z@
zC{FGAgSqmh!mn>t)2~RrcrMzu=-~IG+C2{){-H?b5imRqnIE;k5opZ~4RpGzc6(~w
zIodnQE@@%JVc|KO6BQAr^kH_4sy{x0+5k{=(bE-)88M_Yg&H-6*_?RkpH86$CzE)o
zvC?hpMu}|*w|R>6HjT4=f=nF+1luv7blMq*GcB&Hxj9*<i`v9X>&Q+&hAiQz9P{{^
z+|9d{JrVW1OawgmHZJn{HSL=6=eW1nU^pL3J6!mq*bo22bM^-pa3C6k1w5`5F94De
z51mq;DCYv02?VH6)Te^cogo;dJrI0v?topX-JiE~ejj7P$A$0L(Xvm<m$VKYZ%Wc)
zAp02wT|~6t4<0c>thW*Q4*EJAVseD~A4`VD9vZmUBOK5WBz3?;a<&d$7%Pobt-p&f
z?&6rS4-&3i^Ob5I=m6a4Pb)T%2>}v3v{5r(W7d}r8Pj5W<d8YFjPAXtA<c|Hviq+D
z4D~lqws}aB3}F8fH<_x!eelrsP?Ts}rivPWWs2>6l)0=}6^9Fc#1I1R81tc$#KC?v
zIryxTv)>97^7aFQ+9rg%!b1^AuM%s<<4uIGzy2e)o1ETfvsN)f&><1_2e7E5O}&tF
z*F%)<d+542hzDyjF?zKWa!JY)Bf<g66JNZ&Q(79mXbk*q4&XWYTbFvgW61TvvxU0N
z`W%1?66j{|o$}Y)B#B~f0WE7}1f0gY3>QjWH00$#jcx6|^DJLck^#-=UZ{8Yn7RI>
zA3AoOq+z3xsl$PI184qw9#GsbY?Ppz7C=EF6G+N~#VU>*BHMKL^ls?rM0$yUnY<^%
z-GvSJNl{Fi1q1h)A_A%I3XJ^Q0IK~o@Pcy;|1*)<WJ3RjN9j2|pzTXsx86Id20yWL
z=prZ-P5ZrnrqH|vUCxKj;CblDvE{TWhs4G@_d{<0_glDuyVeHW1`vk#KK~f2By}C#
zS;NR+$!*E)On<V;*S?5q4W1oKDLk&svxGJIw+U1Zjx*&BLp1*0e8*&xsH83ODYiu|
z(Rbc5=TR9JwQ?dZuE-DjB=vb!P%gCm46}9BoHMsqA3%vYSZgr0el@-K@BDsxdU?;*
z9dpN<9Nw@s`7s8?6`<o`CuS%=lg(fJCIYWOC5)Z=lcQGdh8a|^xS+H^p5m-OaA%2Z
zhq&=MlOn1+VYBk}(H9?vXnGU7lWM0?yQ|kN79QsqiO(CpN7rwn&rYz<L+k}R`LY`D
zFFet0-K)FhIk9`4mG1kGx-VKtXR86`$w2mtg!UJl0oCYFqVr5o!BTREw!ju5!grLf
z#)s)0(cC8s6WW9c^-+AU+hs6&Oa$3%fPFchO1Dw;3$4w8QmN4$sP~r~{x9Nu4e6H%
zSV|?FX)_fyM4pf04W3hbrR~4HRELk!@B-Cu3tKScuNA>}k$3Nolq}Rsyv23G^$Hgj
zYD%%m7?`a*$N-I}Z-w8vfCxa??vIrpWipqjI>M_aFdl<*rCv7Ka20}P?}By*iFg>Y
zv8qSIujmf6Ft`dtWjtdt?7?Boi@3cpS_5KXyxji$4JEZx9<+x@9_T=Zgz>|>Qwt8^
z?~nc8^ZVgc?1z*xuax;WH~s6kKO->>=?;MRE9JstJ3w!?%5#6?N$+RNXe<)E=lzR-
znms8&lyI_I@D4dl@Dy1gmDpKDx5HR<&4vJ^=m_+A@I@CNE_^%EL{vcyYgC+3RN6zb
z!%#ji&#zNF;X7|EWBYh559d8YZ>LKyiOhJ|!6JqO<&0Z6ViV?%HCr~^3e&he*CaFE
zDIT7TdjJ?#-0qWoa*o@;%N7_SG{pYuW2O1U-+x+=aU>l}R&7Z!RulZM!FFUCW4UtI
zsj<e3o+8%TwcA*%y%2Nt*5>Le8SmD~C_QA~a_(&V3-y>z`t?5($8xn-w<Qu@KVI|D
z6=z$}W`x0&SC4h*s<Yj5?zI@4Ci&31IHs_Ttblm^SjCIHZA*Ux@zQ03o!#B&%Uq#0
z`%khm{AO+-w54CmS?@G`=MwJR$h5vmsh^>(554A1G?)e0f6a4KWdRBr5t+lZwr(b3
zQwQe+oLyT%5h%xFF1u_0x=TW^^%PGz;Pn)ph33J<Fq?D-MKRIz6z9U!N5JI>|19+e
zSpmp~v4U9jz^4MhL3P_cz~#mA4y?HP6v3_J+xrq`HiypnxM7P6mc0_fs|<e8V}@zQ
zLaAWX6Cht8!}_{cNr>u3U?HxMrWgzX3Gmoc)e#`O!8Z^;C)<A4!AjU3S^42m{lxp`
zFy^b~AnTW;tT-XGu^Y!x>j=>Np0gwnk_JR=t9ZF`vg<0e`6mcLxPiYc*ymSSXjvSf
zdBWRs1yCmaE)032NgnigOXtXCWt8WTC0g|wAD&{Qs`U$;|375CV{m5S(k>j^wr$(V
z#J24`vH8TdClgO>+cqZ?+xEoC$=>gI>pNB7zg6qsz3N`weWAN$C;`(A#W4E$Lnvg2
zMPEDd+}DI36<M-QLcGR!8_Hwk5z#NxUV)Vv=6vW7B=HHHXK?@CBFqbE3@+hYYQUIN
zfZFe5ppn||VBd#&hKw<S3-DqH5dF`guzFMf@1gRsdy6{RyU4=*NIv?%fnonG6fwSo
zm#KdbAy2qDKrKVNMT*G&n3StcsU7*^ym{&V_x<4y0W@yXH)!zUV#B^9|Ci|88R(Do
z$;od_S(|!u<z4CR+(ZbQQ4g$Q&LpM*Gx`5(cxV%!<|ftL(|vt<==wqM$-urQeI6OD
zsn+_%ScE0NIaX7S8&`1{RkVS(A511yrl2Vl$uQV?O^eyjyZ>Yjs)=8rH|`11UXZub
zCC-)iQS!h>(X|Hi@5r8sXPw|Fr@%D-uBD^`NkeeACwAiiEJ+n%#(u*H%v@t)&1{N-
z(F=l|*5Vl|mT>l?DYeQp>g{zTlop}RJWJq@@;3cJ_v#aCDa+MyYqpK9P}}g**|ssc
z$X&NAjVCzi%X61u!#WzPIOf5Yl8TREQp?#3fb$q!i`iKwoUP4``mgbH-3!96wHzCu
zSFX%;xaXAPoZw(fgCj@yLVcz_+%&RY3*nZ%2#Dl}TT|OHZ`}%uu!-uG1^Xm5gO~Km
zT~Il9K8nd%wOh!*AN;;Q_<*022T=YN$7BSl{{rq>(8G(@*@?38fBs2hX!q(!h>`%I
zG0_G0Bdb>|6d;RCI_64`LYm?%R3#Gv@)Ij1#!UprgL`_-3kiTAXBVwXV5&vknD$Lp
zk)C{<{j^h)$CC}(_qkxPf$^A1mN}DMsEg7OpTaZ!i}LHXRDmq-*AA&k3>XzmkQfBN
z=ot2|WM|tYeNUOqm1h!GNL}53A$ef_15MQOyKg}@N={u!(0NTwrG-TF)g=4lz;y>H
z2ZJZWq$9w8Qj7=~JxU%!TYgNc943pNJwYr9sk9!1&AkdvBr4s)zfp|lssAth+VjYl
zY@b#r+(4t=tCqe0+<{AL>4)+ae?q9K`Gi}8n&drgW0D)HE0IMKsV}j0pGt;ctY_Mf
zQ=9ycb7ZHXPB$18nDP7_xdqQbOsz9~5Hl8>6?O}3eGljkW=ozXrO<fy9!4@>et8j0
zdRTH6k7F%OW*qcJO>7zQ{WGnpQ{r_qT(tAe5J>yCEa(wo%V!1uPo*vv_&Tv@&@U4L
zIOcJ9$Vhmt7A2kwg0yj0pJV;bjR1AerzcPHH)_w%r@9Xy2h0l2v@3nVvj+NwaL3X7
zTQ}V@{tz@=yg{(Y3nur$Y;BuXAHqu-UT24=nKab``@~0DrKVqIk5yx9FSE`GrQP6<
zCVf)0)czg4;$X)+ol0p2)DXvp{+RyOG*&I4=fBsiQ=NVYm5peIu@VP+#Mr3q|LHaQ
zqIE9CfGG#|=cuAEu(hZ%{2V07(F!!7fZ+%%&6FX+1ZSt(2vc+Mlh9{cv0SW3!Chhi
zP{P0u*j8xsNV5(#h8T7tTSfN#eIXPW#<gg8A;Fo18Yc1D)0GhO|2u@fJR!g&;Ed>s
zI42o3jovRum)jQJjc#<Rq+aG}+PKPRD`*7~qKX78u0(>Gj7j1!dEX~f!1>0z7s5&m
z3S~`err}M&i=qF~Ej3wftfoUO-?D*=zdnKgdct_^DeT^AACgyf$}DiMjFnb3aAJl!
zk!E5prxYn)R2!@T_Mt<v15tLKpKr^X$t+;7=2d#FGk^ZblQ|VG9`h~>VGE3kzpl#2
zXaX_Fh!n0jjHY3I`j}LLj%lfum_7wAPVY8=>=XhQegGPTi-wXcvMF$bxZs(~4ehb?
zxO0ouiu$f;{g$+dEV1A(wy0Wjf8AerT$e`1Dl)})r)v9?8FX}M^prqT=qAcOF4C2W
zBSG=1L$UOa4ta)W{l8;hIz*#B4c3wo{lHFnU}K`f9ht?ePNVAT`ah<&VOYitmzkwZ
z(yS|T=AbbXenS>%ZfQ^5g>H(#DxH-U*k1;u34p`^8$Zxhq*QyK0QIKjVZ8QI0I5rM
zbW8REdLEyctmA%$1pY~Qob<z6nc^+RreJ+uDFIlB5lGtiI(ZoM++G$5xQ(!_3!q!E
z3Nwrkvt#)zoR5syxFE4mnNz;W-WP)gdhM>bci%*-I2D6a9O^c%YfMsGc|HsYjLtAv
z4R<}OwE!HC7jEzEE4rUA*7QZ}+J~FKJWd(RV~q9_&mvB7+`Jn|G+5xkZg0Z%7O>6h
z!xsRgMD3*9zCNW!k<b;DL{3r(0NzXuB{_Zk`kKJ#Y8~)#gq&o1jz>XDptZyW2pHiw
z_|<7NL9~HyP=<E?G)Z5AQgzRo(1xr?`#J&v>UawW_F~+Y&a`;o!}x=(E0%^o`o0wi
zkgrfDYyWUHla<m`HA)>PBiwd#2|n{Z(+-o8C2~`3T(uAqI4LT8HkJMP0j!DsBS~Zm
zS{DV`x1G214n7BB4Z>j(sOiNlT*Pq3EZuDTs=7$POg`B$C(u`vXcJ*dTgYJWnnkw$
zDH)Q?@l#)6ZWt%6q7scrbIagI@$9R_YwvnYlv}*DBO%^qd;~YPmf4+#dpIY2#^&5+
zlEK4$_Xj>1sNX3bA{81T9Wb7$=2ZI#%?If0@6Q?__f`1I6D^Cx=Ka>xg)@LT&M{v2
z{Gjs+zP%rcmr}t{n*mw5#;=4I1V%_iEIAcrk_*d1i9zT*?=d9~c)||$Cyjy?7WM@X
zRFE;brkd_|BRim-AzX;J0N2yw0Mi`*1d04baquDI&MlhZJyJnO9Oz&l_Cu((*Q_cf
z4?Llp#1=$MoTgR@&ss-K0re<#xwj-9&k2Sn7CfRRh$@zQiq?$)>`HYB7%13h__~s!
zaEG`QAktAk@6}EmGX3p_O^6r$TE>Oil;Z8nPoPGzE{g1LuTrw;lngvkP>)<7Y7Xkw
z>PJ}6g$&e@6`#8k0iJC!GV}pVq!DLWBu~}A5Pjx;*4oSk2USSOyA>g|4Sk3O{*rcD
zG@r~(#KHOH^(xvkYDJaaTTe_8o&ByHYA6sX(f7lzdl_XnnlKPb#8+_c@A)L_i>^26
zFD$0@XJ|tA%gz2-jY%>b8e!~+s<4Sqv?RD!VqO0ntnAROR^Z~jJ%`?A0G&cxJIcY{
zZe<g%37c|8JIf7Gbvfe~b$$yRHb{2E?wcrQ`GDTAsiEWnsOwo3!?_aE2^U?^qF@|>
z{&Z-<m_<l<GqsNiJ5k*<XzpPxiJ1*^x(5;jmTg%I#vcJu7e^EK0>ApeB(Zh(L!CTn
zE^1VSyxG^2GvM9bnWrv+5ujG^*$*nEwY5pcnEh54r`q+bMHisTOjPjK#ZVcmbbLdg
zD4;?*!|v$2Nj$9y(xRi*=xlR`+%6N%Z~}SFnu-Bx3$bQUEe#UZ*T#3}87@cecmL%f
z;ww*ESK6GhU;M_AxEp8s_iY-MkZ$6w8U~;K7bfB?Z*oFyDc-2xn2g+9P-Ot(|2GCz
z1F8+xdrV0jvb+x_WaI>6ab1Q>Qv4Q5U#;7au&awTYf^}T$YEkXz+^(Jl<=C<US%$8
zv@>hG|D$#>tJPc%4ZDOdW4d(N0V`!@T~#n+oDh@-C<#vnLZ)ntYJ4)PmWA{0vI1Kb
zO&f8;7BGjgy~#*RlyI7E-NL%|E=$C8eL7g%Vq%Tz=arZPHa0^$Vu(J1OvjJaujIW5
zRL#w6qwUGVDslei3n1sA@ms=c;BSb$w@Z~yG6suOX&wFdKCLs$*a9E<7v^BmUU)9^
zVmpCAvtdgJ%d(JU62=KiiELleVQs`QWZ*J>izdErW0`LBxTT_no+NcDMqwJqYQqWP
z9yzMwAcG(?i7Jt~nkGO#k2fWLd=aYB%0leo9d@?38uQ`T;!HT}82A{t@Eyv9M5kE8
zx;kg)bjP2m>hd)#|2@;2%GN?#`Z?wN0ugbbb49@zwU{-HLNj3QN3EbKXv;)p<K0>z
zaX@_>kiq&rUSW~NyMT;FGoj9$M$vh)f(J#uSGoVS;~^-GmhS?&LM}xzM}@kMY5EJc
z4!cQLR@YuM*h66*bJ=dByJAe2jCX7RE)>VG8JtL$UGdG@?CD$|A(S<nz87hrY{MET
z^wvLg*Hp57D<+1=`o}@Xj;r(HynI}@%P`y%LQ$+H;4lAK0#3QUlkeyT@iOr)_87gA
zZ3+acjwaGL_HkMjcA27Z=M4X*qF@?msf^`Q>#q%7(?5$~_>E*vQQ<W2cQQ-%ehIA&
z?PP4WUvrT@Z-BYOh^fW$`mRMMg%(>N6TTe<9$U{oZL1doCM({yMpFE>T|yQ7tX<yL
zsM=P(?`!t{PFNIb4zkgNCHG>us+sbvqs2Mbq8~1e!;8mjMJHHZmp5{I^h@$M3v@2B
z@SLWV-p}+V+UQA2gWAnfUU5s2x!F6v+h5jDu9(#IApYy=PDm5r4N6GGIRSRS6}+L}
zxv$lV6=$?E%?G)1J=+14L+gtrYN^|qFwRG2*XeSGZD@s|Od~YL&yTD{c|5FS#84^y
zzNeB|PE++yt;^if-+{fKvLh?M72QVUwB4H<2i^)J;fJf~iFC6y*l64fA4e5fFJ;*I
z*@}MZ2V?)A5Kx9LEcl?~WMClh;)2C0j3I*T_q?rK6%1@Lvnb5&Opz4LDFZ-M4GW9R
z|DV`!&D;f|`u^3U6v%9hZYZ&rkazLk+=6{gWGv30v!+wiG+zsJa|w>zX0Jw)ZvbS=
z0>1|TLz13yor)W=u2bE8IFW8jv;}O;VrL<)hy5%<4n?Py2!#lR3l$phd31o~G8-n;
z?59+92Sq)HR(}!2-XF;)zh_iJp|`-6QxZcLd(w5}PC~h>6<29T<@aPdNv0TA8UnQE
zqwKLlh#oPTQfn1QTkr$8tAxmc!@bG-U^BWVvx;(0`H)~?h|ffLl$!`ID)(P`xsf3$
z5IXWV5r#X+IXT>G{+5Zrzc81w(<*NaFfgCB>?CaocVH0P$$j&YHBJwTEZ+tv=yWi)
zJ<97qSn}DwY{`_nkLF~$Y8EzTmR=r4dy*^LiI1#Qetw<&u8_l1l4;+YhY?S)z1%HJ
z>fXA^)_!HzTC8D(c27@kFlHq`mLA*QMsvR;+L3`jEEmA0@IdkaheM-$ePk{SjtOAX
z?HMLW1$%5#(u*$X{QnA{Q{#WlS@}j_-Ht36zv9L$uN4VZyKP|a?crrnF*BqGkWj0R
ze@dH*dGaj(P&)~YYA9Li%9;~IqI0CT7JpaKo=4D(?f=?=ANU?*Gh$@P!Ra~jpE4?+
zSnnUixf)>a6#4~7l=oBnb0mO8BBV>EJb(f;%}e$~LM}$cp3&XYT7OH>?ja%EZ!(CP
za-p*IEaaK3%nodTmlA`>G4Y3?NdMu&&GWJz8QkAmM~kWt9|xLAUMh+Yydy#7f5>Z>
zn#|?)p5?iNdP|?L(q&qRo3FO`Mj2RJGTwS;2pdO+GqicYz0N~HOeL|F0v#i(@Ry}`
zzTt#k0WRBaqf=&oBjW@+-*DP?+0d_j*eDXMj2igWrfh9n#{l38d2NQexL-wLJ>dR4
z2T(qtg~rQw6b@E0msqCWa6`;S`q=7^tz+uivEuT@aJ#MXwp;0ynb1)zM$m#K?TOB5
zfWc@0pv(cvaO81#Sqy6uTi}zwi8d#eoZ*Z!+?6WbR4|~VmD@^o1VuJ@6!J)byHMy^
z`}un7h_(b}|NL%Wr~|AT>R!yMSv@59iXq>TOM2p*&t%++@`iurub&HMXz=gw+dX8d
z8ZRt;xu=^s5!VZ^@<zQM;)FjccKa&=le0GnkWYbD)Ra0ZW2;<7K|zu2yiWqdMQbu@
za~K&BnoTTA?EoR(HVS=c9595%pn@>Wkh)Ed^n>0JK$u7gk4`g1y{MZSmBvt5utoh&
zCM$ybdye&x)DnkV1xifuS9I|c8-%X(r#R3JF3VpIbd30>4S1@x2YvW6Dwaa6<TBC~
zPyw9Cg{wGF5}W=<QIl7dRXuYn8LKQHZLSDpI)S=|j#ZWKQ?;^q7jFhsUg#9ky?zEN
z=r%mdnKgn6EV+&^&?8jFh$jpYOy^o_jy7m0h>vjl=<~_#m%UEWXeHW`rlIf-OB-4w
zxsdeeG|z|Nl#+~)P8MftC={H>7Tu1<j02aRyE%FaKoj>Pk(j6#cHWUGKF;@Ek2=DI
z^kqZrU=@nSFL4awB7AkZLOn%gz3YZ;;9cuOdCWqqTjMWmYqeqh(>$<4{Vu~GFAA*)
z#S!1eVe8i-2?aYB+R2aJ{<ox*XXvde;B8@eA<{c=edF(*=H;bLwKJ4(2PP)rJ)_%U
z#m_0hj6lt~OS?>~UOzUMDUEL19e5tb6(R)^h%gzBrJYO@IUAjPRm=>-J1bt*cSA|u
z(cew^G}yjQrP}N@%1+&bb7|<3o$#<>nl`Ue7Th22(Q@wG-P<E&p?-7g{b5Kuytv^0
zBz57BNBP7&jBv>fok5gXKoblBMaCIftL`%>AD*(OS@2fJLtT?FsW1so6iJHOf{l><
z?5?>JuD84CXIjCWNxrB-9nfE_;`{wnQDJ*GuBSHj>JiuP!)k=(Nm8u)=IF4A8}txo
zk$M2r#pL`dOOth~i;WZ^41<}id?~9R8v`6fOSDnESGn5lJc8sbz;zN|*5GMW>FFB?
zHbU~;ikbR!c-h*=PQV#Nt>$VnHvDCI)yrq=g1iPt2O)oZur)2%HS~$GulFOr4sh*C
zpAMtxzv&G>I$%mCPjgR1KSV_|(HW14X=Z3MHS5iV1d*D$%vr~6)UtYzajsJ|x@u8x
z9>~n-%f5xSwF;C$EZXDeZ#qbv2{vfwuGFbh2`{BWs*rcg9RgJgC%TWP<6%{J^Kf((
zX*MeO*o0ZMet9xk!ezvv)th{2{2pd#Q1jFEfKL0r@MZIN9VGN8FVN>+u_JxX`0igi
zR+Axme1^l8fqrc+hAvpJB{IBRWxh`U{@q}|m!EqG@blZ^A2L%EsD}b-Ex-iF4*6%i
z|L$iutzt}Z4fvVmUkD7v`_ILRWtdD)W~w}JD3ceMk7ksh%AM83A@&(uE^i8NpH(VW
zwd=h=Kg-?wpU@dW1KmKRf|LjsL0z=0_(m8?UmfR{B}(7y7V6MEfvI)5jz50lAm6@v
zqad+QpaJ2DnUWdJa7Ce;S+v$3&Lka4JT|*ums8nK`ss8%Pscv*dY;~#_J3_BG#!?$
zIvL5jfK&)U2@8AGq3OwpNYY4>ND?J7X(3Qlx1Xcs`md(B6ojXo$@fZO=uv4m36Po>
zcw7o*2`5xC2HEO_9?QRouGwNbuItE{884QUfQxm$y=2ypy8cm7XKN<*0{UtYpGwhY
z`QWn=$aDjUkegolHpa|-h63yG|F=?$0x<l;It^})EP^6&as3=@L}by}tf3uIZ6)<!
zrUK&_tP%a!RCYlPT4+K<@wNj2MNW3(_nbfA4hG9-3%?owLX6@k@@g;#Rxo;2%QufD
z2`@o-GlH20`cawX1D%D$u?f#XIJ;QNUbU#R7;A&hgSRv;T$@iJ;&_2oR`Z9Tc@_<=
zOdVDuBa;?X0yvrVPeO`59{QxyR@m#Vy`aAVRav`%A~#LmoF&&-mm%ESdw@^Jw}cUM
zQ6qsxNLgE?%eKO}zORutD6xM;IAreU(jGB<vsJii+WT_Oy`)svrlBF$dFKAF76>f}
zXwpt6q0~C~w#L{9<6EvWP{rxj0`OcONZr5VDS*KA1Yjy^&O63=DwfC|efD%AiSy#{
z-^g=zQsp74+yD}70rvX$xDS%yGE?~|_khIJEF66LPyqZoC;UEb{mxK&N{0ATma|mV
zzw?YWFgMT+=D?a#3D6?rQ&;=bUVeIHSPTulG@B#bI_MQSXkT?vu+MLHVRx1U7#5WZ
z{cTDF0OL{q5Xfvc{_g9?{#3#l+4#n{nb%YM!wAI8@-G!L6Hioz)`!F=p<4|L=g8|L
zC5p^E6KQ#}X%oPCeDl5b+enqvfk>YDL=U1oe2k$X{{g4k9%t`kVvq^U=)pd{9~kQm
zdk1!7gB-2TLlSxuWEm}BOj%<;q_si>57%-FfxRUEO%C$vs+1FJ`Yb>DXFOa*wt8lz
zr(2aB)o+}w%*ASet#GK|bhq*IH=D+6I9t-_ZvO1;IK7A!+VVFLrdY;ec{CUI6IpXF
zX>M^NTHp2T^6(?#;p{X&tis&2RtiuxTxLH3JEfBK<BIi=b)SFqiX=fm!2dL$=YS_|
zzp~9m*<>Z0)D6bInSZb!S9??egafa8w=9POB>9ixxVrNt7B03-yKwztpJ~$x8aqx?
zmQ;5JDg!7cupulV0aKx+iZNmUdF@}Buc6X+u9KJ8GFy+jX**Pl-1ZWCN1mN7ec9gh
z!*BK({9QDacr8hxQn;ND4KQ6p_$HJ1WeUx5<gmLLe{l;rwCeo+X|kaB?qb2&9!b@v
z0Y6?@w?xlmf+s_%2;P8U7Gwml?uU0b0q4z<7YvI8^5O*;z@px>Vk>f4dhxdPNaks^
z{BwoP*^EvK0&Pu5z(9wuh`27=6`mi4K^y_*?@HipNFSe^T@WwaKz#(b`+CQ3AqvSd
zSn&{bWy~!_e!NO{lo%2G552SR&tF4g0C`F$ttHw4M~AsGd6-wJT%O3>GLrY(Y~rjm
z3}~}q<E6MLq=@s;C}8v+PJpdA_}cwZHyROw*<jNnfWN2F8#<8OU5a@YBeFoIW+nSv
z^rPdIGN)+xKJ}NJ&HKM2!V2*rXw+>cMtD?W2|pAvCjv5wNRV}@9L-kTA;T}~*zT*}
zU3J_l$n3X?G&OISgYx&eq%}Lu?W0dS2iDKx*-RQzan(9`FUPVqs!*)+5<<HPs7W1)
zT<WcZ8H&A}0K~od<GJpiL086C)`OR46s+INoyH7uCtA8%E25|i-iLD3q3+L9*51XA
zCl)GQeCXdIG1pn16gfN&IlihE4zt7z<ux<lm_YG?=uHf@x|(X$)t8%kv0#Moe41TV
z4BTb^n9A_ax&L5nUO_hh4&fG_5<gf3c)SH>G~J>#NJa?99qY$LoBR?()8}owi;L13
ztbF=LVE7(CWTjPqg(A6eu1yQhET$a1ZRpQ98r}PFuZh90ysg`LXbJBq7YWG@{;Gr}
z`$C!n9u7B?#PV(8oS<y7!`3p>v{->XLMj`5uT00*^8}O)IsK;30fh-d%?FPaR5+DE
z%aUd9a79dvzElh%vMVU(QgvS!;kXaSFzZQF;NM%ofLS&o*-JD~+)i!0XyBI45X(_e
z+o~=IREp|p{&kawD#Q^FSuMAl#^?)`)RfQxMAXtb{*{^GTj~V6d)Kyo;y!amaPxeq
zkZ4%6gde*y5DpU?A-TM9|6DbH#!Bs+lJluV;K7%0H=t<khS>a)s%JGlqY_aHO2OZa
z>o5lMugK!QC-@WXiXVKifjq<y%|5H!NbA+5d23LMR{11JZd>|_CS<+G1-rSH`}upa
zrpHPpssY6N2J`u5pFwuYwgCck#>vryT@K|oC2;(#b2RVKb0O3(Egs73zE|YS*}U!~
z?n}zX#oxD^`gky=F7K|DiZ+v<MB3Y(0s=La260aGYA!wrbO)(FEudct3PNV;C7&H<
zX+H^Ptgoo@hP0oG!#4^U&<;uxcUyG_oMJg{uuSV#Y47C3@k9@NF?g=~I1AA1V~WU(
z3O7KSmzML+0LQ_1h?~B*E!ZRjVV7Z_jVtJ_cr~(7{lj;#1!8K7JkC<ZK?r8ZS@rOG
z=FJ}52>rxojd}%TCn{qx^0!~YR8PB0F4Jy8aA0|s&z(b4DNK7*%)O4p@BW$vC|WGr
z{Gjv2($3eJx`ph(6z+}I0|vi&iqtOi3vs9HGO1i?=r<JAVYbZd)&uruQ-Av3EpTpB
zsfa;Ak`3?y!JXZwK>gy=HksLeOyW9rmPuR2V+%P$hNe=s7sIqGZ;F<6TA_;kyz9$t
z<-reNC*(_a#?3ZA*x2=9pSIK#@Vfon@nk{m;nP-Zxd7>(CC-dqb;siJ@NR#To?ScZ
zRi?jBze_zZzyGkfvDc%{DiLLSLo8z<P^2+R2#LwD*p31*{QB7}OAlsN6LGsoIEBhk
z)Q`fhDWVreT}&BPloBFs9c|Ih!}qs}02+9KOhZ+;=|UOTcZ0`>?E^<4aGOdy&lFur
zc<kwUYUqypUx?vhg6-0Sg~+3lA?*ef%V&LGw83D>&xOERh-VDvVU$@?VT`r@6$ryC
zsPSbJ02uU{5cWN~F_e>oDPLyZ&ab?iHZF9N?e|dk`e0GG_uqk8EI37pT0bCzQ+=1k
zP@={~7w`#+HckkoeXdc`fB?F366BPc<b=fQaN0rAG|nzJuPKC-K|yz0yH>J8zsa8!
zzeiYs`lxfQhY7GrX#PRI3a{OoKpdXqvK3#24cBqOb5Dv_Je8*~YH9Om)d}h^MPYx{
zmql)02G;F?{q>D2h7q{23l~{}EPS=U7^S#RZs7{$bZqqEPYc8yz*FR@;<}>-TuY{Y
z!{BZR-|>^@0W0P<tEp2qX%?><qEeu&s*PZsy$*?>6Yz5SKvPV3@_yKHruv8L{a}b1
zv~!KbWh>vnPS~&NQPM+^_j0e&K$}*><1l(sA3^A09adI5ugA|pW0%3`8W<VdnDkx!
zI1kr^`;CBT{RxI(AP83RK;h+Im|DCfehrj?s3!+Xs(bw!D#fKBoQV%NGc?Rc(i7;n
zoa!5&-=_R}%8n3^lH(E7r(uotaqY-yI1X-ydD@LEa0s@9#?Ro(>aYqHm7%UrsnI%0
z|4)EIj`(kY@{Nau<wbGsi8BpmhSNfGqD%ss@$qye!>A-GqQ#ew-yzd1g5pBs9&3##
zRa#4$w>j$vzG|b`pq!FLKJ*MHOKpbZ8{bHb)+)oJ2zoC#KJZxyj4e;Dw-JC1J`q8)
z5a~X-So}6oH<({t2XN^={zgn9YCcOIO4j=&@C3}c-SDS?m=zyIhss4JR#%3&;r$8Y
z^a}}HuQ?2mfzR>H{O%H0Sm+T@^Moo(g4S=9aLHyAwP}c#ITdh^4ZyFau9kSLi`@M#
zP+aZtxZYaHw4ezX&ja~Zi8?5x*jV9Cr7gg^W<u0UpiKrEZ%KTYNgqQM*n$kAQDmfA
zj0D@*FVSVp;`d*MCGfWN6+T6Xp}4QdRI-f_pt%=#1u;7Q0>iK)B<4<ND$excOayzA
zYgTjJPK~gaxtB+)-jizmPr_i`KJ+|nAS<-B<6KnGtYZgbn(rNBQ#b~cs}P6R#4LLD
z{#Fzu59XZty;uzM&CZ@(CgmMFVGrZ`4%Xtm{r2buznEtAUDEk^w7@ZqkA`Ad%;hx-
z>27k^3Xlx;1XdokU$ko%fZ4IS?n4Z~VY<NNIsqJ&+CNDUnI*9~zS!2<2m6~rHwcEx
ze<MSh9k7El4|wJ~jn^1R#xM(Flv@#Cv5Vt@EuhpIP7xS>z#S%ueQ%UCL=s!5UYUwp
zy-5f&S!4cb1r3T10WgIEsF<;9S!x=A0de0gZZi6ffKdPQG<=gW=!UVM{zs7CmZH1>
zZM-*pQ_5L<bY`;jScJ;DPiK#8qIGrvIn{uR|8@*3>$jc}jP}bvEpf9&jAlLg)OVrG
zSmqJOd25$Pp{a{V6#2+>#ua&RgxZns(T~-6z3nbgEO|16{g=(57trXFi&+83o$Lfg
zo51|r%fn(Efrvl-=NitRKi>_w++FI(D$ASwMl~(LU2y89X@?q5b?Gt#?gq#iMjp@=
zx?>DWVa2(TB1?=<Bxg8w|IWo(;Rfm;vz&|4fMolugzv)S@0sunz$+aU52Cj5h1k!|
zEn@yKz7$1AhuCGFA`EN?XKrv*?t_4o7Xw0mIn(DL2o7f251k(&0qzMJ`q~>}QCHS^
zY@c*zCFhz!9yMFBwaZSv`1af1XdDHo6%#~<Q*vgk)v3R{s9$X7VYzc=`kfVXe{mx<
zddtrZ(?z1E#i%U22YpWwzT`X$>TwQ);^B8}dPBmwGd|rH{1aKSEy$uc4)<6Rg@@V3
zeJdvCJ8U>GluIXvf2|3&!T#_kLz(;g*ARDN-EJq+_?!3l;R#R<>>mBM{2%f}XbH7m
zrqPFPoo&ep>*X9%pDES{J2EV-TFgOsbOx`xgo1LsUZmDf`})uN@dNcG^#6+47**@j
zS_FORjYJG%8V_-6s$ljd)3O>5frqr-p4@3#KeT}eH`8$L9$A1@^r{q~S^pxkL}Si>
znaMdJN;t>aX~nyUl!*d}72%Kh?tXU`Q$Zb0f6HX*Ls`u7*Kg)!9eg2Ik95A+(bvF6
z*GpO??e)8OLHitVlcE&c5<UfhK!Bqln2VvX=P4{RTjE`dn4EmmzP}d#XtQ~bGC0U4
zhfjWowtt~0Oej27K<r|aJ5ggWxD@1x3|J5o5R<Xs{2f5ZtPEY`BfL;lulamQE=QzQ
zg(glO)gWP~Op0XVx+1(<MZm9<`RKCbn>#ud5g960d)>elm37D~JkKj#@DIm?N~nt|
z8!lCm?Sq|Y@yPP0tlU2EeJvQCJdxZc3RJ!&+-b2GhpG@~Ac#e2_5zNs2VH(F+NLXi
zJ4LcCo`Jq>lYx^GYRZmIV`R=fm{UM61^kSy)y!B#zxj<}M-o+k++&|tjZIvZx!b1m
z7i(zi@>_4NRD_UBFX7sGK{pXvO_*#O(MH4vTIxa56Ol3+x$J9*L+z75;Gk9(VIbPK
zJ}Uc-A4Q@_eJf_(`L|a~)JTbK^U)dIXR~ah>$tzwdHjP~)}*_@w2Ww8P*R{aD-A{*
zW%EcQCLAiE{1jf6%*rLA1C%uw$70Ztm}bxT(=((jDpuG9l)~>#q8xsT<2<+t03S{8
zfU(j3L)oTMDfKf+2bsspuV-OS<ombRllJ|h<TIbHOd~_Rt)FzPp}tpoUS_f^2sUil
z*qVfV^7;s2Ne(iyRIz)Cv;9E6M({3-fR#=wJe$hpGdaDd*%|SGNH@t;$@~=j<ben3
zW4&>5Oy2u0TY_LRm}nB3iN>_P0*_#|+|a+%Y(Mdu_}p{lU*h?8IDOt3Qt-MReo?10
zmawn0aC}X(UT3eX^)gD<fT|*9XPQWJ9?ZQw&EMNWY;Q}IJ)ztow}%1mxMs;ykD`WL
zw|`KyiW+Zxrx_mZr`jGeNVmwycw_ueDoXTRh>cS1`L5~R5d<Chs<MV~tSpn-tU7Y_
zc&NhOO`E)LU;6rBYV<lda=hyujb+u3II(ic`~#0S`a`sJRp+lIRc&zA%WC($jq#ps
zT;Y>hhbGLBJIF)&=sRF>u4FP?+xl5f0u%|T3^c{~bAhYQQj)3s$1rQq#&z>~=o0=0
za&O)pFvoe@KNbAyEuKT>qZR7;T@AJ3OWmN_KSLp>uDHbHV=G{VeIqS7;O_;2yQ+<G
zf(+i3IXp2jK70w{)5h%W$nxTZQUKId#*@l#Z!RCESC7*x_ce-3%fzm?-kF0<|DjLw
ztwX`RGsAD)CO??0FXS!UD=$Cbx0Z0mEI%k7@H^BV2GKYotaWphmV5R1=Vk)P!h951
z(>u5+>jG%d@Zuckw{XVPuZhPTfhH}QH0Y;9n!)O2x3&^LTD+MaU#b@MGvXNWKlaZ6
zt+q<J+%}=~8yg;u?B{!z&wEFEwo~ws>kD_@2~UZxk@o0rA`UnXhcBq@H?h!{xG_yy
zLy~8YuOGiDNI&ReTB<GnOLZ?m)<E+Kh?ldL5j#gb{izY$xZjn8!P%lE>x#S9R4|&U
zFQ>CmLi-2!r;xh$hskFSe@L-EvmnfdW~E_{kgBeVY&fG=0F()+q5irAU)+-%VRRa(
z>RL|kdj;3V3H8hTR$Mfd@K6ArjvEXwk1Dsq(!MntG5j*1!37JQbM+D<R!6BsS`-jX
zYJi^mI^}#N@0EP@e#I!#*(}q$%5Ud(U#k#IJKG=rn_?W_z#vx+oue&VI^<q>!va<3
zGMc7K)nf2PSYH<S|Ip+F$d=N)`;#w*Z~@O*Q-7oS+!421U4}v=dvhknMnmt-sHvrw
zQ28g|8W7zglKe_dUgf-HN0Wq1)&9;A=%xE5*nCcX#@8zbFliJ7Co-56e^DFI(X_D}
zaO5uF=&{xV935liZw~ttT(;kLsESPd=VHr&?~#tpxL!;zR__IIyT>6j#_9<Xw@&S;
zHCaBqv+(-HS=ZQzv%<J7RZHK6VcDr?lB`VDrlo;w6Oe@JnMvMEx8-11%$F|3@v`L1
zdgVlpfPr41x^Vt7lL_BrjYJlDgIx5-%{^$0K5EzZ;L@$F>A(vtohz>a$1sn&%jYWK
zV$L$R7I%moM~^>{N?~pZW>fmRiaObQ%`ACRE8?O0IAf38U}LamF*7vkfq`6mzcW=@
z$mG`~IAeM&-V!+*K4U80A*D2uy7-U$n(Ok8XzDHQ*%$c{nDj9yS>dpHlx!ecO7@#p
zKX3(^type8T@6s}z|S$LO~5BoJP$mL)eT~N{1I~pK1;dnsKl%I(}-Mzn))-V#MD^8
zF?RH|32F++cz-{8ysBC|e$)+#fR){^uv3>$Fj8WZ$$9sJ7J?HqVNNy>fR(|qw3fYz
z5;nP$`KCks{Y^&)r@mocNO~n_yg!aMc@$dedr?Owj0@EtUQadOyQd9(3Lsj+rUERt
zlU!#jge(EN<w>r8(xyl0vJ4F`Ha+1?1G8@XBwZS}&4e&0G)($$e#uQsPy1qO>fzFN
z1WHAfwLWPpY4n!^bfE2i%G&c?ad1FSD^NTRk+5MHUQ2^QOoJ{CmjgWt!N6DS3}G4(
zeX$cRN1*>C3F^lm<QdsMQ((89Pf+0PfI5{ilD-k()X?J3x1KF4$;dTs$Zi{8>FQ^4
zkZxY>-<Q*Z_*ZWIhpgWnLuLosBui?Aw7t5@xC{rxAB$>u<~CN`1&L;cS_cHsl^$DC
z!-s9)qC*`NeFkVoh~ub%T<a|<=u_UD`xlz>LEaZFH;*VAhlz(+FVGY9p+)lHKGfRk
zzmn$vDw;#kTbj4;-rDiR%;BDx$;5Edp9gu&!gOEL{fB~HaIlTMD{2%(q+TdHC}8}q
z6EkrYEBQ4GMcxQmyzoW?X77@P=$X^)B9-iTdudDI3Dw&lFXsWZo2OK6ztaWWY!KxK
zjUFd!2{g%7q8u7ZqfAhSyGProqPI|QjA+McT~6bmME+4G9;VJyx{+nTi-HP(+b%C=
z$clpELrTJSi?U_dih_~>O@41Wj;M!Zv~19YTPmCN#anVzqyW3jV$nf^phswRggo0N
zrI_oxD9+>F?haF4A^8Lh!Q%XSt;;GOsm%++_Sm|O0>$(Tc(Q8npggFiW~NVzm80}9
z^g8zhR`d(n5R1Q?`oW9aprZ>tA(W$kp}G`cw376IrS2W6zUMn-Ac%qL5b*Lia{CxK
z7xM)!1wkjKEF-vwA(f{?A|0m;mSzNrfffS0F1og9{_Ki6{FQ<#u0Sy%Fsjgo@Nk(?
z1Z0KdDxKmyv%Va<))6<(_&Mo^net3&AH?D;n}~gdkj4y*?`g7H1=sp<H|_bkzncYN
zc8)Oo$fVk3U<&n7pTm8TqW$yxC={ejYSUMq3^pByb{)cqm;m9H0D-0)Q)gdPrn(O}
z6_%{K>-D;pyaoXavCmv_3tdT)sv(_`lZb#eGZqp_HKmpr>oL|3RmXP&;&PW&utgv*
zi5@W+afuSHquYK^BKha;e>Ly_dqL=k!58tWxmaEv$=+O`akYP{X@*bt=n=<jRuEjM
zv}^1Qxad>lWjVyQ)LQ`Gf<$uBAX1}3F6~o=P*D1Ok-U2Ase2ASzZ{X?xC7^kq@UM!
zCkp7Bv!r5c^+AWG_bMXtzSbM1{|@|Da)<E>`A+=gdd-}%mf=xVkecc`u0-f7G%dC#
zn)N$-{0<JgXxx|>pl4O`K`P{Z`IL&>DSD@Y>-W$7NggyXz#SM6=x=d#0(+D!4FgSI
zx5CA`1_XpG_g{YWer?64mZ_C(F4cGll#C!J3`kU=sJ<fwG#-UQVB~HMRgtLqN7S#o
zS$nX5T&MRV$%&erCHry=RWurOUz_Yu34?Xx0m_ULpT(>xAAwrVYVl%JGqp{N=`j?C
z{Ui5bD!(RMc<*+yTB@2w<aKD!x7jP|G(LkAwR5AWiYu%Y1Ehq8gt07FM}z^}5%8-m
z^)h18T0atv{(OFx>%L^@{Q$)UYWyi9Gw@ENt<g-NKf!hrDQ|U^T=3rS0uE7oEJTuT
z+>f_HR2>g_(l2Kxqt`RN>hN%JTs5j1q{Psv48X^|qgHvn4s}xA_>-S6Mz0Pt`Tr%n
zBe+&WV8|u{;OwhAkWtCB4L$`8*BhVR1wpa6J7;yM9+obY#Gy~-hPg=rEgV}?4!W>0
z;f}JZUMMk52vWcqQT4YjS|m0OkB}Kz!|V(}#4bbvZAs&UP|Om+G{SY%L{V>`Oa{<P
z7n~L^cO&4+lH~isrj*x*4?t>{D(CcIbw~?raHB7c`kvqKy-e^t@p3x>e`9?$k~obU
zwB<cY8h6Us^B}z#*=na9G@b}f0HQio-yWIs*h+9a*Z#H4V;e?!jaIwZdCqATENItt
zcV4M!6?BrmFt<ikXB20X7kIN6*W1?y@80G(`yAvt6-xLTf(=$qKr58O)s?Pc((t}E
z__sOC!yf;`!rpSi)lVdGngQaWQ-5|j~mrWX(>T^MFQJ#_Z~I}gSu@Z&D$HWHbG
zvfIaZCAlwZXF00wsjT>K-MsDJlzyM=14y$ERQeW^bh*l@t9VlxXHywrXzwlE+^>up
zbvliMhY}rdI|{Wf9UAjP?7YI^eYAf{y&M%|!mGgZcZ2;j7*^WMTkew#s0KvyJziGe
zj^Z(|XhC?Ajh_@;{q0l-I=WoX14&Bl7~jBs*vHU=u<5HX=1hqb&bK+E0EATh(<o08
zTAH=#@I8=?Mtbo0;Rh`-GSe>LagF?F4LfJ_KYpbV`Vov;C9Frg;-N#ngvQMu$#y<^
zhOPoDbg%0LVq53#*)zYi*!8FWnjQoHqdB6!9>2LAF9W3zL8REyNw#-ZG;v9Xx%3jq
zMN5RO-PdQARo+PUzaSo9sS~{oWt7cWZg)-Ny8I(|!uN9gFL;ZDsD#Y!RQ}`H{AcMe
z=o(<=G!C^n*l6kRHbc8vz>9}YKSJ&S1-@W!HI`X#REcP+d3=x$;H%gX0Wx|U)^f+6
ziRvzVJr*Ky9?qO+;U3NCa%b$mqdZ)**z#B?$2br=5M1x6+<8i~E@GTA(e}}zZ=mmD
z6r{yP_ZcJZQM0KuCyRX9mo<O*v)~%VZWH}34{{rnl#9__qf`9_vo2d5O;i^(PpE1B
z8l?pzD&fNj+pvw*>YJl!2E<^1-Wuf86R=%6?_R=Vw2YaZ_37?UB=kyTRCLbP3~tFJ
z{^=ZkOLJ$Mc1zN$pVel`a5>D@C1Fa8)5aTQ@H={0kHkvu<6N+%cc$KA{Hbk<32o4d
zz5d?jq2oiLb<2P3<>eXMzS@4%6#-rkz>v<pjF!7t3P655(&a$40Ol%B9AReXqoLoT
zHOJqjzBfBO%3_*)B)x~j_z`g$>%+o-)R#=UUZ~i8WBE<SI_mfkMFFU^64sR@3=m8k
ztT&2}vsD*rg;7uJL;rFWqjQVHlsmQIISy}8`s}}K(PL~!E1FUKVgdno#Ao+F?kPuW
zPpskFo*$tDTYNOEz`h*?T~qc<_00wggMQT40uSdc-!`oFm)_jQF-8YOS>!4Iag2G?
z|B1S>pTX*;qZwI8NNrY3GnNh%NU;zbQdc9=l?GPhAbrw4>KC(M0k2JqrU(40>r~7F
zs_3U&kCPeF^r=H*BVOa-=g12;rcVFm$tm49#RezA`yV+oVr0Klye#H_jjj%#&m^!F
z_?M-STpqI{pocIMsHtYoX=X(RCW#odVyYBkZ-3#>@!pLhWC6~B-_sk{c<ekof+}&q
zq7^o!Ou>VQb8DwRS*~HIwRa^k$2Bj#k8T|%m-(*|QnI}F6A4maTmM@}H)vhJ*?#J1
z{P?}jS}kl`4rD|tYWMVW+th<i^HBH;(|x9J4zmDbI*%V}aoBHDDpe6?&edqs`-yZh
zP?_hsw&{@giA1A36-Rmu&{Ui9dE*R=3L@E2VCfQ!I#NsJ8@fCqcT}-Xp-b_a$YJAK
z&UCBMx_U5$JD>Kj`?Eh#AwhG1meJ22lRq>ETbN#U31nLzPs|jtS1h9+bmO=+&j4^Y
z#^0v$aa2AJx<AfsE^Jkg&i~hz(t?6|%=STNYV@cnd@cVN<1~bKn@+7wT>`HpjB@U2
zY5Icy)?ddL@YhlHdkK^S{?sPAotOr~;!t!Pkq$uRhbzN*D)he)0J->b?g{muTaH^`
zI^7HK+`1CsqhFuUx$3@B;7vL%(iZPk)3YX$&G+W$XkZuv>>!w=#4Db*XOc?ItFsW!
z#9$ajj#21~JUohtheSD4L`4Eh3igMWaUr|AEH})k168ZHIZkLV%4Z4or&ts8ydm4X
zp%-M}jw=Cqda#p~%zK<_S7sUrtL1q|4t^n!<@RuCg}83#S|gZNvfeP5nSkE18}r_@
zz~lHrE^KhKbl6Ojvk(4oSOmh{k7j)XjcylVXmXSHW`k$os=$}KPY-)dZ<{8TZ%8SY
zn3~MP$0-8c1GxM0Jw;kWJ((({a(qOYJy6<&w1Fj#A8}TT#u;d2f%}m90b2uts;dUP
zfF7E!o6kVIOpoLZ5qtl{#F5V;*y3Q`G3(KkaXZT~S++?&7`iR8JM#B?`WzfSCq6@N
z_lt%cCBy6dEL*&|=V;c<xrN^|PS~!NUtNS4Cj70c=nl^tX|L}1Fe6F8DoCR%qhe@w
zm$xwt-U*q^zeRw;5-j2}^N_gfY61!*?=5k+@+K|?^R*5x-3Bzh-CkH9$wiWn++V%i
zhz{p^3C(%OBBlK{eCzt`Is_f~jaYVreRN+rU=u8MlSyh=o&F6V;h}3jVdGEEG0}Mn
zcOs8gP$9h!VH)(VENCqdb2`qoIr#momsC9e5L91g(6S_aL$R+|tyXH;<1b(_c9zMq
zrJR2&=3y8#>ER-^VhH(0a-_K)ID|4#oiUB4*_X=Y@9g;?my*yfhVWkoKt?aR7w)#*
zYYt0X)%>*Cf6zDtl6(+}`=#GAC|ILFbp4-h8#+7W&v-oO0g91exj-2t;|9Clmg+UU
ze|`H`UWMITMsgbF^}A4w4Zx@O+DF1lc?nzL1sgNJRW5)I`2dHJvYlw*luh}7Yl&3O
zeC*>)0qcK3o12g92AndNg~IO{8)KpvSAu9N`_d`4TY6EAhkbPN-56u^K5~0oP@`QC
zqo7eC<|7F{*D*!qC`M8BG*u$qE?ub^Vh9v(R7@GF@PFL4u1qMvt%u)L>P3v(I#*xa
zPelVFKwFbUwG$3J_otlhT1z@MnWOqIVc+e6Y`)~8H!WM|S^bq_o7|Cqq5dHeD;`z&
z{C!p>Z@G)RC}nlcSZL(1!dIIw*R2&4CPZk)I%u*<R%9jKE5|hFZ+9v>KXq6l{FxA+
z%_CQlH+rt5Ym5H?sh)qhl_*XpSH8~VvYA1N0FqdR2A-wsL?4Qi;%LUR=IiO~MsX?o
zxn6Jx!!mqO34`3Mgv@N_XMX2k%cFP*?G8F9ck1;W{@9{~h*4vI3f%7iG9$iL<1e&d
z%1X#H>hMdf@3_BPi8^;syKEjWFS3&hwP=WAFIvY?c-9XB<%-`y0O(FZ$F<d@?Y)%t
zQSTn>-B$*3o9?u`Fc?0Wim8+!b%XLe#(FKRdISTXbHI}fJK(bwZ|!9CPtKBoAYo0K
z<oYU`V!>;+G~Kku4vdtF&Pj|z<;y<U6F^GInFNDH&<Yf;3RwcjgwaH_7{jZWf)4rB
zB|X;LRzNXO-KsYOOkt2N8eEQ)Y+w+bRPD#OvRLU{U+Y@4E_0#fQjwMFFl<*R6O7+c
zcv5jV2&$CvcnhV(`^^mI53}9{{FS=rQ6I`_7E6H*@u3B3os<0}8U<S}%K)+*Wagh{
zcHaxjaDT@24@h~eYL*%rbWcgqD3+JU-w%4q`K2HO{pUj6#<2`qIIXPjrtYN8+3w2^
znqS7$F;`2lv9Sp8J4^*vMh0V@CzZQUbF#gJug7*mR)fns=qM(13EcG0uFbuc<ja5(
z2$^LaRg9y^{B&TTo&A8!!oZZUyeO+exC3l9tm=#}Mb&vbOpWKB?om29YpikEMI)y#
zOzdW$fqgdjnsYQ>LH~hita;2ZBHj}5Wod}suSg4Dz5B8#+0-%6qYkg~40C5kQkA2U
zsdKp@*edjv<djP%8P)qL%*Clih@A;2Y@G-DUu^SI10)kvsJqcCY5FMjvrZl~Q`g-I
zoVKXLbF}vXSNti87HLYp20C#@PL;kzdIsFUBMZx2a#j~l{;p5|u1Z2EV|Ac)Q1X~6
zMwCf~e4a;oa|k@Om;CG^!>SnT!%qH|P4#-(9GcY#T7t!TB~7R>7R969Uvq%7B1TZa
zpD4qp@y(qUa>R(P5&Q?dXM_OOIM(K><OwOalrPeD@vAS>U6z#dqksPt{~Y5VrP5vi
zGpRG8&5MN6OB!DMe$wCBZV|xu3I|Y;J7XNpsxo07%_1R~rQcaOugsDJJ7~{-kv=W=
z%9wjmKVExs3bi}4Wd3N=bqkA>h$F9gc+2_Ip3xz3=E)p-e$#STphWq5gV?-&2a~{Y
zKvReuwc!T@kiJ4OTl-J#1ec9B>(09_un!iJ6oCgB!E6DPnkNspqlK`s5;rDyuNp+M
z!#tYlTXO-gp#M`ek@GISPL-B7%<LoaDvNtl$b+~V4nWTy_?=f}KRPs@z(apC<69J<
zb#|wgba<cNjU1xdL)VEEf;C{pG!VNl4dZzNoe1f9@(HqsB>1Dj!ZD)f{=sS=*fxaw
zGi;R1NY|&wcr`Ld&51$PCY?SC4U`LX&&~eJZu%!FaeNrEmb#Win&)`KGz%h<kyw{%
zCAg%($4<EWlWHqd8u`N60ZV+935{~)rn}73qU(?n%%U9|SNUSKrF`UvMAU~Rx`tJ>
z5Vm;osukOpan&@YEhAKfi=8nc5Yu_j6WSS#S~X^uo!9}_jhnfijv&=M40v&$)W0Hr
z^8h|p+k%V+frBkhEG}gemJBaD@6XuI{M{0B)uLWFTS{)L8{g=o@v|bHrrt8EgD;rz
z{324mvG1)cP_`#IKi0M7U2L?)<jD2?H)Di^{@F8n^EkD2zqZ#ug#r;p;Mi21I#H6Y
zUu5^3y^wML?UXwCmwoI4N@8IHaUdCb$>=#i=}WD031DsAr;JjPod(}jL7OJSMGS8f
znZwR`Mtp}v)F%$TCfD=NYT9R?qXbnJR(o<%PPKhJ(q1OaV0>YdwT>LLbVNkONPUuJ
zb9z}e4dDgS(};G<-~THYNNEHL`1pEfhdg$*mPZ~iNvamR<S+Z@9Y^DqKDr68wGd-=
zcVgC{P$v>jlH2+yJ>&8(Dp~(^C>c<2E<2xO;i<)9Aze`u_v1YJFJ`r!?Gs35nIOo+
z*PelKrXs)MgK(0f-V4H_O?euu0d0pWN6GhF%}YJmQWIH3LCiDY1ea2?lx+2hzWl*G
zIay;cZ>tkX;FHJO(MG!Pi{}PcKC7fzEG`d6k=$Ga;rFEfn-%rIJ~J~DGNoBm&(}RO
z02AOzv3Q17elc+<pTg7MSh^o~M`GS=)sFF@M41Ywo|4u3zJ}9!b3>YY`K#0&=)X?A
z{Bizvj@7GI+<XsY!gb2DMC>v!Q?`55m2<g3KnQ%UvZ})|xa@Jj%pM8dBx~+?l)Z=9
z*j4O3W|Ej7$}sET__5yZ>AHKH9m6AjjLToWsM!sP<m-vMkDNtcz39-5Fb31M|C}dJ
zN8i<a>@r^fbS_Z-6@MAu@28sUuij8TOyOJ#h8<-oP+<ntF(bZt;ovW+?-&XKb<9`$
zUZbRg!Rj-)YRwbU_WT-_W3`hEm8JSaVf2C~jqkRmUHb%c+q#Xz#8KQ+Us5_ve&!;~
z3aEwifB1UG;M&@#YctM?bz<AL?c~I^ZF9%AofDndwr$(Ct<LklUw3s?S9Ps_t7_G*
z+CT1ljCsvD#~$tt4WDRa0L!bCBZ`vRdbnR{d7$*d_mh?uR@^#X>$eL*W1H}u4F#v#
z_gkjm3&Gx$*E{kRKv|1|+ph0Rq~+a~jXL>_lj8$r-G`kj`QB7JbAkg_P&||V2;29f
zz4(q9b$do+xNs14)s}s0hk+W|$W~4{TJ5$qqf5BYw30|LVWv>kf^1zb9hB}FM%yK0
z;t?{(&uavYYW>KK-+ahu)VDspf5QReh-=8bFov+*-KCBJK-45?`mbz#fB_2OJejk*
z&S)2I;rm?XeHy#%Y09ij^DOK$r3@wgX4|bofvF+Jh!BT6V;pIapv@prt>2qWW=jpN
z0;9SdQ{|;56z)U2@`-Q2hKHaZx4{U1Yg2nW(x#fFgCC;c&&wbDD$o=NVaHAK$oz~v
ziZUqcFc2&fz##*Btq+`!b^NtZ;e~_IgzR7T0HeVV+WQqrHk$1wdsz5W;{83ko;Y-<
zW6?8{`VaW18nsNFHby=#WrevrztyPMh3Gj9hFCnr5u4%SrE<0M&9s|vGv+I8IilF%
zQ3BLZ#E{aP4u6}9WU|(Om-D)taHeI&iU@Kz1Z{8BfN{3_@f;uV;2aoMj3G$XlS$fm
zfqltF9=$*z|B(nG<EHP9Rzl)3%xWR=m1_n%fF>iu26!Q;_c(DAE8(SHr=rEvKys1!
z_eT^&i@MCf2dR<m^wHCuubC13OvYrxGcLIo9k;%I%cQYwRfHvDAf>LXN${Njh^=|U
zv81yYVCKM!Vzr?-iMNeAekf2LGSa}WqWeONC<V#-UP+??4+GZ3hHr6`qq_#yGh71a
zzDQ2Jkon&k464HP3d8C)JF|1G&Ikr!R@~Gbvs$^gc|&G#Cl(H0*#0*rYw`V}A1bo3
z+N)4}38^L(vS^b@g4;nJbiJHw(ZF^szI(DPAe-4qidNdUQ1p*_Lhg@*;M`ao@aKp;
z)v#t|JBr$RjoqxeSBuE4QM6yl7{!D6={-n$F5%SJj7vJnVf#s~U$fm?hlqmR_n^w0
z7S4Rdc)Bm^`-;E<AnW8XC3<@i^KNKAf0rLcKx7JjO&vwoQehh5#436lhq%5-*$4Xq
za+mIi@$QqaAmNnx&=8_qQ|TSN`hPndey;7}G7&_=%DarS9fLTBe9yHbp0s1$xK=Km
z6Opq0fyi<J9!or2g@{}hAB{KSheZ?@Dq`GbwvBb)Bljj6eu(RzC+>&#SxUW>pKAzP
zG^^BZ_G$I}WP78l$$Cxr{k^Kop0HE_V3~()A7JB;B9XrCJ8F$1{EJz4JvGX?XOc;F
z3vvlnCt4Fst}+nEUdDlm%0eswD+s&lO;c-koD}>0?jHCd;QPzhJ*1xzdymu+w3V={
z=>m8W)l%{Bk8hz(;@I>aI={M-PQL6ThDRW_LC6bp=uX1THvwc!5-9ULhK2JOfc%`J
zIu1&F3`{8~<H%tsZ3Ia7FOlUtT{afKac_K`fH4a%m3N0A9}XqH$l*;91p)k|{|Q-g
zp=W3-_QS&;D#qWs3xnT*6vdc_a%vcME3QqqdV(16($7_G0zB|xkU6%3u$X)1{A+d<
zHl|kHh`bCZd6Pn!hc&uN;^y^;fKHnAimHW@qykJvmzu4y+B{4gm0HtI-Sp4fCH8@c
zYP2_SEM~V|<Q>_pJ#b`8p61dSyqsS*x{x&!R^rKEm>Z_bw{`{7GL6$wy*#&d0YxD_
zi%5K-P(@RgaD}2&ol9C&LYIZF{^U?zji}_CVz8MI0!VUN`wVLV4GhFOfXTkqFhYHv
z8V=ZAh=mC!9Gh)jxDxYkzV<OMQ+q{Qt|yrVn&<2glL+NX*Qo(_<A)?)8HZ)akl($v
zPs?wz@g&0n39)3#B+$qvFbXMe^6?ZBN&5LTnu_sc5{a8Er9^UZ<XY}TT*~o867l*G
z?wjQ!$i|Wk((Z25OkkMEfQh);6Bm&f0`UOOx`p4^DHO$Kjk>ZMTYe(9#1ZX$J>f*2
zrvLkCScq2PFGZ3xB0at{;S3y5*$!7;SY1Z<O(B$OVJa5cvEoP1dbmuo*{0=-QGB;u
zzsO<EdUJSa$)8d}!`7vUa2^dCQ0qCUF!YpP-?ua{3Nj1X72s?Tu<lhM6Q=p|7KO<-
zf8UIVf2_<`P`Bju&%)^hgN;V>Xjsf)t7v!y_7jR%ilD@re)_)angF%a!7#g#r3R@A
zAz<;M`%@<5TJ<!3jag+hv{MB`GJ!S$j6~j;n?I2@Gv&S&|LuQu@j;ukqbT3pl8H+E
z0h<_Z4`4PX_xsKK+vfAO{ufzmYRI!N{O0=eST0V(yVkKz+(VzFD)w=-2*`QZ4bCW9
zr~W@tKtrx-dr|Mc=cDRQ%jb+!aE)vGGRGy0eIRDbm_Nj(4#&=(!<70(F8=(-93u22
z#`uTg>m0yxNMecPla!{@&z7^Rs(f|D`|4DZ4B+(X@AH(!e#$TX5&SW7oY@x*?6~G_
zToz$Hzu_7i5BIS&m-zSZn%yrIHhU1CLmM)`HRkK?QhANS0@INXsm%9Y_ELKZ|Lz>k
zbrx!JV@WPUY-5pdR(8tTTw^arpHU*FcZrHH+aLY8ttj+}zR2=G`CcNTNfS_;s~q_~
z1AzG^d{VnD-1wUmMK)Z^MmckUF#yfbTKCFvERSr|X8=g+T$c;Mz6y$tkUM9U>=WV9
zt2rpxb4R{SGN{Xl3eDfnoEvb8nz2Zp_S%6q+9Pjor#toa3_a-Jh>$VKwz+PR{9r+{
z_g0s}Puef}bBEtt*{-m8ESJ#@;!vqu1t1!@6~vLhEs?oHQQ3#Wel|ZYXrX<EDGqvM
zV2nNmjrDa0Jr%ZHe)>ngI9*=5m4}3M{Mswd<jXZyOo!|u#2N{aV;Y7*_2iThR=DcP
z2kIwOp6SQ>dn)5`$fCj|${_>d3X+N)5tT&E&@>ufsy%d(zj#L*<<X?Q3+Nt~0Dv1>
zfF3Lqc5eu~r4)71?at#j>zTr^iCOvFy}5BQ)}Crr+*t)`@X$AuC|U1LC2P$cR<;fi
zAN)56COc{{TDXE}7Ot-b#W1c`(j=40Jb>9GpdQYQXOV#ck7EX_n@J*SIf2g+rWC<E
z$|&wtDv!SCb_3R7DtIA8_>x)$08pTHDFi1^!ufKQ%UrYt#Q@M+9-~{f)jkIiORv<M
z^0)^2TpV4y|4E<(i_}BJyRd8K(w`FV_T)NyBuL`Mt*sUFdDKXKnBNIHd2ng{WsN|z
zfEn@CJ3qz_e>5c?I?Ut94$0^a-2x?`Cf(ubU2WsPVC+8RIa`WlL(#V`3b1jJdR{{Z
z`sZ4)8>(zU>eOyyoC&7lccrM=B=zPkCF_#)JkqtVPp%qnG_neg6P|-u3q3So<xWaG
zcA`#VBn-&=CyPpMgE{wUH=8PB{NYM9lmI8uYh)X)r8hvDQN`tcJaMs~y~ubg{erpp
zJbx?9|817beX-TDQSLS-0C*mPl2zei7<=dVHyYEB9O)R9Fu)5E(N&sabI6;iL{crc
za3Jo-ua+=jE}A^5$y@yAclC>lnbtwi9;nORp9QE|YnA!1+llxEQ9Y>YiV5_vF*)8<
zO?e{vwkSTEO#$5Vz;yvJA>?Rq*dHS)6PwKa8s8*K#UH5$`!`m$7{FY>l5*J?CX5AZ
z3&~9sQ*T+X11zs}NdM}RD@Ur)-p1n=Z%l2SwI})6t%qRy-Yxn0j?2)*^;qxz9HsC|
zy{upYMmX1oDN%WaFY@;3>CR6R-;Q0%nlvJS<$a#G?kg_!vpKE#vDJ9mwbq=5P@8r=
zjW+u1El;oNb2pc<9?)h_Z5Sip6N?&QOk<@PE}GT3G%3Fpe;WV4wVZ4zrg<liQLnZ^
zMYIh(dyhheFGbbgh6-piqgd&R<A!uuW&sy!QXcQh`BK3}I*jkk6;+dX|LROiW8bR$
z?J*n%KMzEGwKWZ=a}jv^s}?I<&t1|iLoGUfasg{|^$h?}@86dh8mCEGi)cWTG+aeM
zG|rxdPej4ZIxUWky63EUzlKHXw<;6(7)^Z1w2c|(QCL;hBhi>a4l+Cf{wf3eU&7}e
ztIuNuM{?`<_DzIJ%5cr69YACLM42Xj`b0|0UH22i&-*NsK2jm((;AUJfI@hKxVSTh
z3@}a@&+!9vB17NrUW|sq5_OoB7B{7Zq%#aAHi#1()ibwkv)ufq24WVExX%-=iqS`>
zU!y{~m+Wi}>8^cl5LOp3VO4YL^rohb=kq->yKd!^x8X3$=(jX7J&RPW9^#nKvmAbY
zHw&tbL3zKOm3o>!mG(kq5=KnY_#Ua#KwWcr>=B@9Esmp*U-9SeU^{HA+=Z1^Zrjw^
zI<>?C%M`e{-3Q`(K^WJYy2+~fWjrCV(BW##_vl7D;ydizOAIUfj?#XLnnQTkXKpv(
zSlJ<M%2@LgC_>)?XB!_$=qjBuFeF@oB*o^grNZR#pw=cW<y(@uje<u)v^kMUrr!&L
zr5|7(G7F2}2TPolj_@JNcVkTKwyOem#Xj8Xt;<<GwBUqKnMWHBoI?YR_pNAJce+o$
z3dTXW_*15(j*dR9D2%NGj^h`F5lv8XccA2n+G5J*KwKZ@yX@Hbx>tSe0KF1e)@b|H
zzf3xP)Xv9H_^W}k16gvBd`P$TEh+M1{}`a&Qib)#io)=4sC2wI55F)1|JZZ=rJH&O
zm2nqF`eM<Z@$|DNl6&cCkc0-MTP}{Ox$P+jd_9Oorke*p>fJUCn)QCl7ttfMLpsbY
zxG(o0NIk?NGE-r*ytgtgetcH!#tV(q55|li0^we=bI7=kr4+r{Ljw6vg5SlF^DIEB
z{v0|=9Xuuv`}HgcgPHGCXelCcT&<GasESfVf!2eqpia_wIR6gW%!o+W6O`$fPr<%N
zNh{-D){}GjLW4kz#@}>#M^Dyxu0*%3%B8t)a*|rFNA53()1ZJ&cUz%rple$W*2TK}
zJx?`Tng~Htr&NP5g*OfzvA>z>1+)Nx*&hZs*=qFpIubG1z5~j4LjW-L%zpZ5uhfSQ
zidN5clhJj6bL6=j*vSAPuMW%4mp<l4Ne&iT<7PZvaL<F@O+|_f8)miLsPGYy@+KQ^
z{~YLm!Ni`AFN^f=5Frb!x=*!mrgVosMb^61tXzI{30KDj?lsiBgPG+1SOutnDO@j=
z9WF$j!%^3GneDWd2<+asmf+Zt%Z)1SUmjTGBX0f45dt4g7GVf@P7IvG(BA^hIbUh6
z@I^&rN0Z|x$XNp9jpIDyeo<;L;K-AE_{wTa8t>ALe(q$7d{N1n0Z`a=xru4&E^EiJ
z@vte92-k{nq1HRcG%i_?NdXXB8%IAkw)mF{n~S*0=D^z=<hLM|L}WqXV3QR#hu{*I
z{T<ET_l_)M^En2$Z##xV*;||36+|<x1;@cMdrN9jvU0EufN2kyPU1Ud!^P|1>gXw_
zq3F-~x;#<yXC{AGOR|v!#N6iPj_#Q)I!4qkjDrTyTGK>JhYIKs8Q`^Jl<5N!ddmgm
zW0I(Ow*x5^hUN+zkv}^`NB-Xm*pU5Qe^`Q3)^VdXUGd$k>WJsgus6cEhEEPWfBpfJ
zymc)7KJ0IboRA#KL}D4Cz{LuF-U)_a_l_y%2LY>|9&!bY+&j1@G9+tXbd^!{B)2hD
z3##>hZ?5)%=niE5Y5;$Kv{7;o9K~>}=(W<{lPc^rmbK5yhW}g0u@^$K{P+~k)5uC6
z|2#c&iTS(^qJU@&vt2n@WXN6hs@|Hd$t`&;t5YLOPEL5?z5{1sdqB5j>UHAGKWh?v
zqkX_WVgL9TDqzJwM906o^s`Y)R7*G(85`=92-eH(;U@+_>0w3`N&;JaJoi`svs79>
zph2SeyTo#e8YZFRjjSxBDkI-JfP$jT;0Ij>F`j(EzFs`S{W;NZD96#%GmA>|6m=cX
zEBiAVYU>A72^=cfgwZASlln%q9%~UP_f|=?Yx7~q)|6JQP4b<E&4!S%(TtWI{*sB!
zWod3p(=`b&PHLZA^T1r7I_(&wZ!W+B!+A4M;*OpdtHe^UEcT@G<x$cibMx6jB(%=&
zQYg)0FKBgan4?jv69+T<@9nBLlwl6UgjpS>71kmh9nE^0O9Lm&Ckn0@UM-JMKuuMp
zw<h;>kCB^UG1}EzPh0@=j>y3a4UraFc1(jC3&1|N7$^$zsHH3s)Qp)p<yc-aGHeio
z`%lrpaqjOCJ^ah=c&R=&E>q(IV95J$gQG<%d~>B%foz%9KV+3rs~i-;AWdXiPkv|6
ztfh|R{Buo`hi12tP3w=1vVun03#gHY2Eh30ov<9V>t^u$7yEUe2-PS9Y(!#zRCtMV
z_5km0jc>=fJcPPTub*ijBS3g~+%XP7tYSMr)IetmNKFQ;(5eMELGI&=6iOcT76f8Y
zGs#y>&|bgA?JrAN$|f}{%<^m3LQ(pRf0rJh5&~A=^>>xaSS@c{ry22R1|)(@#%Esf
zRyPA^8|iS?bVzD%WqdL0a8{C=H2%pPume<EQ=<iE!b7Mee?1IXL%yeW3bkKi<TPiP
zbv|qrHAza3v!%1wX{yTMt%tOFx?p+Vd{E=1j+0Zj6I|w~6tNY~#nmz%b}Ezl@^;z*
zli16+!7rqqMcYFD*4Gr@FM`Kh=)y50WTsu1aK*g};kjBO#PcvLX@%Mls?;10>Hv5d
zf*et&Ohz6bEBP!3iI|7!deCmP-L|cD&{CJ9&kF?HTJ}qrDqb7U+1)b1`o_j#`<G4e
zMqP1_r%XmGOobV?$>Ve~6QnOqvU=?I$w=4-!weWkXYCUl`V$?4y`Ff29o?+x#&(oF
zp>VE(8x<*?H}0K%L_mts=WOBm2?50KjaVYNqk|EQMkCM9-?nXSq2CcqPgOEZzEl1#
z+KJ?Noyr5155RXLD3E(FUMKx03{6ZT0^|%7n9i_E5xr-u*??}@1=WvSZZ?ooJc^WK
zdYGyo__>yxftBJH`0wH;+Cz%Zm`xI*AD_)IRHqD_ktx~o9j9>P;n7n&0h+@bK3&hV
zii-M=*zv#Z@4@!y9lnwSbpss_B>ipNf*J2j%NU`Qz|*KCev!za>X!j2fm6rEl!E9}
zCA^9A%_;knOEC>zrtz+XI;ZjW_>n4Fr_@6p1l?4w_-<Ob3((`mR7bzZ*_(gR<E>rB
z+r9oCO0$zh9Xc;X)dd7<0<tYCx8SaekxPErL8Go|BQU1yf-eW0xlI$M6%Qk6;h)+}
z<m3GbXNZi@SQx>_tuN_OCV7mF(op8mSP*0fN0&FkbAdyhDENnt9WOpvbyfUT3%al|
zYH0?H4gdyM2a?q&fpX`d!ix)6Bll=*lfB)`eaMh#8)o+n236`~2e9oh%J;qeQPx?K
zGW^I-O#5wdu$cxpD+FSc>&|lwt7%`rI^gWvt8rqlU2haC_G}|);wGt`dO5e+RR!>v
zSi7M!ex1K<Jo*!YoAp4=o>J$%!K7o5IpiPIGUAqU^&2C^HjhD?A(a(J>uTJ&1XA-~
z#T^(8&gmmRYb<5c0PN8TNMkY0PchC6xw5QIIx3*vSIEy38-j8fy~l%02WVH%oQHe1
z76==bWqW+b(5&O<E3A8H9g@(}F26Tk>2vtdM$b`LUomcSsn0Kc!^hG`;>OS|%M#uA
zh^Bv*z7P*FqwA6XZ1w39P>w9Q$U~Jf{<?KHB3L(AIQ)3v0Vr($mTQ4mw#kxkrCGPy
zlK#&J9JI-$A>6m@S*lzGNc=y#-wW@7<Eq+)IZktvKkVWC@y1u3$c<gmiXz^h<z<Py
z`EyDvh4+*+(j6sEI<Q(&(?g$Hb^vw<efpR;`Ze!+hJN|$W;zXoHYtiyZs-5#{ZRn+
zN>>2|_!=7JjDwEu9c8s4CCmWZNLXueyx6O>hC-NVx>+JCQ&9>jzhb11UAN-V3LSkc
z)k6Q|^KqI>A5J2Oo9c5io(CT=#kIjbuawlqx>(>Xxcy&~aYg*cqG%ds2+}9IZZ-$V
zS^fnmDG>pykcuAK@-;T2;e$#v(JMf&GM|LE&X7r;>J)Y|+F$JBD!fv-7PP5)s`LOb
zU5`7y^fR&gxyy!6nb$LR^RfB<0n&Rl4LW~&zPE<=p{Q9ZB=2R?O*#A%%1s9P<0L5z
zCoi_=_7l%^+~|7E`-Zw9_)<;obAho;2gnSqk@LY{A4=1l;Cf?O;5*Nek#&H;N03Ep
zzqZ)(FXPqNTueE&c<c;^9RJmRfDs($%8Jc37&aP3;Glbfz$y+S&mBH~`>=~>oiAg<
zI(1X%ZFTPQ)S4{6Z-sB*hXgghYs_OJ)#N<aUyci<e&L<Jtgfi3yN9^-^{HMZ<|>xT
zR66{qI-g=CCGYVs(oE<PMehJX&X_dP(6-w^EA^+*gup<OOpUu8oyJ(oVRLGgC}nbw
zJ3A4T65Dbk4VaF0C&|r@Sb}3dep<3L*ilD1?st7(E-%_Azhu*D>}8ud2Z>joj_leh
zfpYbLL`Qkgyo*=%>k`OlT%!{t{bI}s59{bf#1>^K&Us@m!?by8K{dci+8gB2ZXi5O
z{q=aaBSo|=Z=+(qR%j${n~U>#spEMZeaGg$w%4V+vD)m|v7A)RB6rm9+gxGqrf;uI
z-S(o<QoY#nE0=0=xI$lXBAe4M{DcMa=IyKNu1$nde&5cn_>s@*Ieg=n^ULN;x_2nY
z&E&V|<?N+NgiRZp>kj~x%h$QMzA<oP^=HkA7)`Vr&4sdYB*cZN_>-w5oA(1B;OX)g
z)}cQ~N*aC~5J%EZIiJHgm|2{mWz>I$`u|O4*VG*T=)ELK_6DvhYs4=~9h1TVwYsd=
zS2eGeEqUR8y2jL1`?L(~E!u%arm7zG#4uVi1CUP$HdXxoJ3PW`te;RwBe>ET3QA%r
zbEA%8R$oRvo0WYGYvj_9V0osWNM|&kOv9-g9rqj7WGK~U_H4HC?#D|zOs6j6{b86f
zN~>x5H)!Uyk}0)F*)S`LIaBs9|56R%N`-qR<+k9qdigdytrS$hMdIcdTS(3%7H%a1
zlxGtK;!n?`&&(@lG>fq9hZ!)f`Hejp>ByUs8Fh`~umfc<4h-YYM_zZ?f=C6s9b|m_
zBh7=idoU1_F_SfTNN@&lO^%|MlzS-<V|XyY4S%Oe4ezEUXz6%~Pdl*WfbI(@ALFb<
zgkvmvBtwcbHlo3g^Xp+WWY0>2Gx9|Ns@N?JX(+f#jCUT%Z`XCr>|1bu5u?~b=C8uP
zXy$9aCwa*5OE<hlqfl|<hj)X~Dsa)xa%<){Xc09CdL>ChZmR++2<Oh;XjTm|z&zF~
zz#L?gOCHkC&-jm7x|Ipe8-Ism6_5B}iY=BJhBCLvEwQ8mUvci}DSy(_!DU4N5*{O$
zpBX9EZHAxz{!^e_o^dT4aRqCY#RmmaUKzwZefkBQp`Dw7k;AJz{7Q;^9fVOBCVH3S
zgHsQ`=XMn0_PP`8;Jg~>k!xSSe!=GLMp3@4p9D-lrhd}2pZk+%EA|hlIJ0u351peX
zPREB`COG5O2fh=jw4W=SRqo^(K-Tmkd@U@a0wm-UEB$_>3{KZ^o=#Ey+kz6-^<49L
zOjJ|?JPKLqsE@BKZG&ZKjO=?~s|C+?JO%ILix=jF=5opBLK)Fn+Dn-uzfBCCPPNwp
zWg`CXbKv!7kLA%UsU*|xYYee;0(c3sRN@XFmr4nMPd1>G5isGi3<-|}*m@%9v+UoZ
zBP#5^goHvghk_@N%QD!zT5s~tZ&Do0gadnCt=*eALL`GSV*?<L1aF5@mv8SRdB!o8
zN2Vckly6@E+dWn+3AigkRR8$T2%deej%2Y$jdFbajx9xwl&<AUH_f;!tBevX+H_~j
zj+q{oS>zjgFzo9QPk&*6<#ReJM-<0v6VyT#E7ziu00?S5n4G{C;FPY}r){cG>b~K}
z4zb9MXG>i(;}v*PeG$R_nnOXX=smx3Q;pLZ+E_esHq6=ZozcL@P~>>;{hw=>l-Z<i
za!4?dre|n0%0K0!?kvmub-SEp^=T{itbF@dS^HQ5NLhhLJK#nNQ>DCcnw9ptkP)N%
zxlHS!(W|%1_7PhUA{AfIcx3O&Jp7vhqi;nYJH%8+yG9M63Ql`d9VZAEbJV-nG_?C#
z{&96<%W(tPQLnxxT2K*q@p|zStBk4au8xXZ9P9n`uI!`ohgvz&UBpO!mEdGKIk2L%
zvLs#>6ODd40@V9aleIGp>}+^K;nw>HSpFd-wtnd=jr)#(87Jqy(w>G);=sW)n3DPo
ztc>{lxsui(i|P>cJn~`SFVD$^+HUoK9b=zbT7Uh;2<eFiE*W7xH0EV~pvh#dRuP3+
z2a5w#arowj3-p5XMspZ-3k0N?@j;SD`4A2zf_dGf1+cLvXV;pt@e34v;x81$4rYVK
zyR2p#$ySI05+*szbHgUsCUMj$Q=R%0cQ5W@BKy;y82#=E@1P{7%-FvqH8Z49>7l8y
z@7P2qqmIn{Y0mi?B1fDOe@;->KhkN}D$Q#wnC;hL>^y9tndmVnrVT8cWNuMrLY`h2
zwV-}c255%}1mW3S@iw;Gtw;R0xO$pY1*<g!Wx&vD7moqhEY~^OU-`6bEVSP3{{gMJ
zkAKNzZcdiuAL1#$c?H6EZ%KPQhOaALhm-EqDo^QM*0{AF7&a-)wc=Y;;<j@9tAb@B
zlK4>`De^|Awo+<}k4L>}-a4x$#;}eh^52aQP_4jnZgEG7a(PiS$&oX6|NWU;uI{FJ
zct6ESfR{O6IZMpZ`QYr>E+rH=x}A^a%F(7?6t(|x`BSUWVs+M_PC&vPcZx(lA5F7@
zR#2ACnJlqJUPUX2a(4-3M55P7Vms;<JHf`Cel3^7u0s?BTRF}06f&s}^Q*^$fl@>O
z%&S2|*>ChYVC9(@`ccd9U}YM2UoxUHj@5c-Y>*Ci6UVT8EE$jEF-%L#mTip2HrOb>
zYJBATJX)5q(XQS}1a2SJT@qlD*!z3D_~E$B^$n%ApapHrs%7~0e$01FfK{``+X_Hl
zk_0QvvFM8K^%-__;Cm-V7+ahF=Z^tIyYG5Om~@gq!p?a6+sSq%Yw$`da%v@W13|Y)
z0#3GLsTplB`Nt!-GPh%n{kY@<3Umro`G;qTjyKm9fMt2j0#}R{xVl!bO}0#njs&Cg
z>Ned#X;-6Tr%I^3d^?wCOxNVe3Z23{%S7kDX~CR2JlZctK67hsvS+r9SOCHcYZ3mn
zZ$>&q<Y4i^o#X5L<q`tyJ#JMs4Cqd4aum=`8m2wag%$yN0Ra+r5U=fL|2oUgn%B{u
zOOvSE@t7kDURuD;NjB?wwEO#Hf<%%-u1c4Gc_I{~o_nA|-%E@M_ca;6)2G&4C!Q<B
zoKE@nYhc5V?vYT;mFV7<1Y~=aN^5@7wb~#Q^=_|XTQIcX!ZwEjnH_o-tos-Asqs6v
zWVF3r^z@EgI%PsYfZ5wjcbAwdJkV<lt?*N5x^m&;9cyA;9Xd()Qb-<rsX^%nM&)Z7
zB~BUpeB>mDVB|U}+PA#SKM>yB{JtS;ck>~><aisbf$ML&0U^b00+`%&0hy}^sSkiN
zz;Xu=(b>n3#EXQfOQsr9hb23|SI*6<&KWd<_ZY7i8@R;YVmB&TI7ELks;L&e3-%j)
zpmI=8)?q{hwJs!;+rKurZuy6LERg<_PJ6FHfTaX+a}sxd1!LJ6(5H_aNf9k3zT_?B
zlHP1ne!@-s@^GnI00dc0i&WPC53@2D?$7q(QoIW%>BM3&!s|wI`%l?UBk>b{(EVXb
zc6p8eQ<`J#CcG!OUOTvRMlyX%D|^IM{kgxIScr{bpNUF8@MAOGn2-4=%^c>hG@I%Z
zsF7XzCVhQL12KMY*~jfy1EQJT_cy9ZG@h|ERHs`Fn*k?+nJ`_!&2cx9<ZD;X5T@o8
z^O2?9E}C^apLXo*?f-yMegG6*q0bRi?aP%$3(<ddC1b)8jvLt_ieW<4_!EfkmRZhb
zyiDb$<+hm_Fe7O0Nqa-inpjp}leinu<M?$*Zvi8mzzH>IAIj)sQ@;J)z1rNVlDA09
z(vGia0I013clVA<m=6fqyQ;^|e}BZAfLH2_q%77S^B!Z9?u^7osk^c4m=dx77in5-
zt)HBUjrkR$7bF!llH(iM>0~QrzpqR$Ji_<Xy2N48ON?N)kWhKm1`Yz91Et1lm_cC5
zu_(EdO-;wh{aEi<v79JE0_CFwpk3G04pROf%o9HPW8skh>7@;rHJTVTxSjbwx`nc3
zPeskCH!XkUna^+SB21S}x(;F%LCfEr^BLf6x3dhZY;1scgg66k3%|CjA)7z93%@?A
zY9ON0;aV1;DTE6De+*kp(%+vXZ+1kp^-vtEfQ7o~Rl|#6gwIgl)~8UwP$Vr+v*LlQ
z5%B-%uqCZNv=&FmSt!dIWQ$^|jWBoY8@E)>4lANTTqhuwa(5#IP*OwJp1`o0M9Y+#
zjK=;%)~{D5-6=MRBtx?c{Gb^cmC^r2GoGewP4dKax~4(^J?V?r`#gjzs%*#dHK9h(
zLRZcw6q+~v?`|Dn9dp8&mCnyBn(c<p{|}By*Y`j}E_?eA$G|}UhhtEGa7=DT_m+#z
zq-C>iEK1x@FT)2u>h$J0V^=CK=6)jAOOBNIg>CD`Ih)~Oo=w>_OE@Z^#WP;ka4vml
zNr1568P3TQ5j~@nSee0;+Pb|Bdlv+h&z~p)gZMivlEU2}4o{L$v}lv#x?vXDS{>L-
zXYTWX=$t**LCyU0$h7co3-UWoHoI5<kn;WTLx#=63CCx@cYQ_PyyS|?%;qdvvvSv~
zc-z5gNGEl{iA>+TAuk*NZ0sT0bl9g`w5OjW!iLX3=^`^dbo&=6*P{h})nTkH>X2Ug
zDg>1g!D{SAS9pRr;70Q+V_FYwlsWWR@uOG*%?-!1nqd3j=q8H_em-5M9A5zGf`8r1
zAd5l~iSPgqk=Z;AL$r@sga*w{gV)0vn=JS4%&8)V4q7gH_OcKFJAZ=*G0Bi&z|tSe
z(ErQq2=sgK*JrYKvf3BLzko|?Eb3gfsdB`&@(Rp)ttVrJ&DT_>L|J_PQPYIfUJ;TB
zU6eIHU}cJgtiDK7YV=<cR6zca{c0iV&OO=3T+-+U8*K-3-v9M$_<zNq8<xHeteoqU
z9TL6&L|qn_hxh^}!C>Q?+=BS3b6o$1I7^-gR?k{D8QDNZsUn~y{oo?w;QxrrRo(V+
z=5<J^Jo9x>5Z^XBG#~-R5c8kx1^F1r;4L#hop|Ou>SC~^lh0JV>OX1BL`jkPqBxRt
z!tfzBJ<u{{YT;yVQ|`6@X`qF$Q7DcqyU{_BH`0j!xRd?GV&2N6wGw5@H}bGmFyzH_
zk|m6=yj?#xP8^A;JY2LK66tu*Y=im%FPs1u?Y2i3iaEpwG1_41l2+c#&`P5LSwwTK
z_G7Y<yfADq%QBPWK~}_#3Lu{Bn^c8G+Vw1M)dY58`mM_`(;);0<xE~X>hlKl+f>1X
za35(v(QKtZZ8@x%jU?Mj+Zj&0Zp8kquo8>mXkW-zai?W<#IQ6rnEihTG5u3uCg{xE
z_srIy#CkF!G(<1@_KkEd)86H%-r?zm-5y!^+)7&Y<1LNBaC{)ZM8RW!j3{lR16o1*
zLSp4gWy-QR-W_76NU!TW!S_JVM9EEgaT5?q<8y;=>{3E9x2JAiYZE|BKBDO(#iypd
z_d`6=@xiDZ{%@+`x@_Jr);<T$2wgzyqPZP{+<LkFAvka`Nr)-o)*R?n+O8wQz(zYD
zA;6-QersCZG}<XXZ&kX-FMksPamOB+iJi@E7_?(p_V?v<2GX=jQh-g66_EU~)si4t
zJOuM-i7MOC=2Bg*<xhn7>{DfX8lh~#?y%PbA-mxs$uC<&#LqIc65(S##XkfoyVo{4
znGU<|gWDoZm@5?gyP)aygZ~wG0hoN#&D1l_RH8o3;%PpA^cndy#5pp?R*m1L7$uEj
z=JpF}4{_0{TGtr?-ETm=2QXk^M~HCx>U1~@7+uRD&w&T#xT(9Wvc;<25>s=|gG`Sz
zPDqgi6ZjIvv6q+TLUB;F`m0J7cCFrou7v#7T<FDn0bAIv+I<gcA^rRvSI8Id+nvbu
zteD#HrsKGdCl%OT&py?A^@Oogp>5~PZLj=t77DMX#GZ@GfzgUM48X@g0y4ZH{MVfa
z;LHNN+S4u1+0#D@-3dz)J$Flec&-~uPY_Qv)s&<*&ynjz;d=0#<fZ?GMdEG$zg5Op
z&8Aw@&#*!@b(BP#Xx*zxW;rU+jeTzh+KFWGMw4d>2>z2w-L`d4Yw6ZwbP_cOu=F3}
zsZjwe-}csNz-J^>^M0M)2E;BKMTbDRT_A(VA2t?ahf7Omj5f8htxA>g*_ywq^lI_o
zP-U-*n%?cePJ$4`8uIwXCC&+zbIMJ!k8l%9S7C85vy-u8kl|kLqfHrE;#Pb@R-oW4
zUDXXXmUpXsoU<H<TSw#9HwU@Bo5~P1EFyM<nwzXwfZ^QkNcD0(*~NdP@sr#T>tJxX
z1nUFu&m&c1stlvXx$pyEtznX11o7SpAK3XtdeAa=!h&P6%hzWLhx~P0`E7Z|J5L=n
zJrA~HgnziIB!)toUJAFVkol_Aw2jH=`l%8CQ*fE@z|WPUW<;8Y)Dm!$(b~<o9UC@F
zQ{#VC0g}fkuulGp21B4qm#2a3t4)^H<+8LDF&<pis%5;huU{q=%#OPPZKJQK71WiU
z_3+NxLdcXT|I5c}xcGW83mb-eEh?8wR78fXvLP=m_d90))cf$eKJzBfdOGZC7aG_c
zj5htf$aqI!F&G(_oD5T$EBP|#uXTTqN(bcBF#8j+_@!f^K}_9XdWLtEHZgcuKcXKl
z3ckf9r8)dl^bB0zc86<m!;e{sL6^Mh%l_wPzG@`@fg6rW;jW81s=m}Pw|90EwT+8s
z;D7;y->EU#B<`(ir#6-PZ@#rOe-R;-LGSa(ovs8V(4%KlxGFQ>jt9HVc4Cf7dj;?p
zrKUh5oIyngRoH@`Pv*|_WTbwfu5bP!PO~b%MF|p@jKIBs;(Lh3MGBu~hWmf8E%AWs
zrEZ<V2)aB~qUXy;nV~%fJI8O2-oC%o>vvUU5XYgeORRqBB)Duk7zXr6N24e)UZTC~
zqbpf|PGJ}^T7_JZE1{#E#87;F0o$AfHQ`}msL^K%7%DrRisIm_w=Pm96Qv{@FUd%d
zAUF@!EWWtud6SBC<JGOd<~8Mk232vb7gt-QR>}g0^RI;a+<(H#2}3{wBf>x#pHlkH
z9qFVWRV^PqQQQ;PoRqfX2W@heu^PDutPXPr!QoLecC0gZ<oeHV)X$cR0Cm6Nj!^#{
z-Y`Dv<jNosj@@kv8%#aHLxf7vh*S?7e%3Xc6Q<n8iQ_p)^`+Yj9fVJQc`U1;k9@;h
zPz6|k4l9$0U~dPH1EsD!WywY5U&YVgY*MYBB^?Rp))E!PDMswfh+Bm5<ti%S9`n`_
zw#I1_jvkVzv3i(#d@x0~0;>LJ+K03G2OV;;IS+c)M2BJ)Was_bf)e|p;i7pNM~t?w
zb(m4c&s|y1MPP<v(j0h#?b|0W%<yynup0mRzVJEs+>@x=DK}QyG&2b8(}r`c?!-r5
zy<}#q>TNN5a$3HFieJe<My7>QTTD-PTkUqzEzq-ORzy~VVn_iP7vb=3w+Q@-cjz`Y
z7KyDzXFWYh{E*NhOYXj*uY%XpG-&!AS=<~h`gPwVT4)Oi|LeKhmTl_0FcAA`V~Z=e
ze*QWblm|OEv=c|{W8L>Y0|+MW{HT~aeBV$qyncNyJbjV_Z{{|$Ygr&ZQtY%vVS9e0
zh5bI%8Vi$w*e@BtQe;0i2pz}?pW|Myy~R!b`t^meXG6ZV)DZ3!6f!y^I*(vi-VC{8
z<7oz1E@{r`vdnB4SUeOm4-L6BA<ERdvR-Xi%{~hQ?#I=oK;jsOeG5^n#!)+K%nvo&
zCUr!*Yp%dw`nsouuA8<C1r$ffk_}1VmQT<s!TDT@*;N9N%^=;1=vdlsUKJOh#vQ+U
z$?_?F|K_C8OBh>3A*X<qq#A#hAiWN3BTS|Xi$SzO@6oxkX{TaUWs7plkO?FCDD;@V
zey2wlz@g}oV#y@dVhOnuNv0Peo0hR-D4xMUd7@QPCAhR*B{LF<9Ep-_m3p1h^>w31
zaglksK4k+axIFnA@Yh~*Os`ftTb4M>pAtkY$sZ6MquTgxIdU@;Vu1YKOZZx=vKNhv
zYsv|{TbqK?&POO_koExiPV&&h;i77R*pD)X*oZQIvfxF|2O4e#C&~f~?bQE!4msD+
z7M?mF^k}&5#9Z#7Six36wGiod1)Sp~1F3EZI^qjZtZXeKtNVwk7jA<%oM@_hTgTkF
z>0bM*WG(89LHc(<Cm!}M=~X9cIhIG5;6R-E@IP7gI1q@jays~#9k=I+)<|Wy1(iQJ
zBf@ns0qxp-EQtm&D=3qbdiLGzFStyFM*W4xLbCF@)l!)S4!?Uip6IUa3T?ly27_D|
z#41$)9VahnUHO~ZeX$ZWyUQISS&ykrnVhpHNDZyoO#Bd?Ndv-$v^e9U$Y_-in$axk
z(M+0fz`#tWX&K6D6k@HTp#z?^l>+r~QmhS5+nOwxI$Ih7+zIxy2N&P=2}rACKPB}U
z+F%WM_fuu++SrahkZf+YqDE-YuA<-7eEkam^^T}J|A)dapXx}}^F~#e=9sMw0i>Vl
zqWl`sF{+8y@Z{3{3&y~*$a1<Zzc=T_d`?%I@d|Y9CjIJR261b~I&d;LI;O1}6%8lb
zXt#67bP``Og9BIhfr58$ui14l?Xpd}xU0`c))&DxB<+Q0>jXZwbgY!lf6_-$$c3_i
zoq_A!pYa?Os>)aOGm{XF&miU~FG@dy89QIl?pOL}HQrKTKGt=Bt*(@pHElK025-0`
zWm9HXsEO_r_~}i#=pM-CjM}(~!rVOdDf;uK>-Slcm-x%V*^`eRmbTvG`|{jh_8UGP
zc{s3Z)GjsFok)Cx7c!q+RvhFV?EQBDi+x|#-w`aiH?8}l`A0swXuD0}a%DS%G6wb3
z))%A@;sED##s;Y?PAt+)3)4H2`2c>Km-*3yk!gErkINLcZvkuGmS+RXSqYyU@cbjr
zP2H*^g%5=fom)F@$Z7o@i5e<%y3HU<!QYJnd^grBA3X4G(6p|To90WWYR?ydVCz2a
zWc)69mrw)eelML6mQF)1qlp)EvqZx7`BMQH^sG-Sv&1)fn}v!LZt`sV1_dsjS(T54
zLd7B+6%-oSil?~Bn1#0Pj*c^3ypi3<KSk~i0Z(jU{W>3E{fK|k9m2D~+ZH>RhL@}1
zIld>IpVBC|TBq=4cQ246J`8gJ!r)YKmajE`In)n8QDsV1xKD9lhJmTgqZNBwUXVZ|
z1W?W#pmXbv&Qjc?EWrpoL6Me$5i{+N>s0)(iZ4W(l8Xa~tTa=6GskAo7k_054s`A~
z7C9@R{8GfYcZ+8_>oxAvzS_-uc&%><803kzw#^M6I$;mHU>9g~$LJCUG&<0*x<7jy
zfZVq96n0N5C(_k|$UPNu+Nui{|4Kse>J?CPp^*`8n6!2MpUKRN302nf2j>Y4i{**Y
zKaGpwRM^<yLD)W*gBoug5d=0Qqr#8muz#oA+Z-n)kt1aMQ3m}vX61qq&@*;MZw5KY
z`$H=Vo-IySC>Aq^yD|az<+&s@tT<>Hd$pZ}v-AHFOM4AH*C#tKGOp{z>O??~>HQU}
z-PVcKbkf%A5_}99k52P1?re|{lRve}^HE@`c%gWz(@53a4vx;QuFAZi9<IKL;_>3!
zx^*9I+h<Z~#b|L78oKB}6TB~}wZVR?Zq2g&yPvgWP(HQ3<mUsJQ_3B!t>KCtdTG(g
z5r}K~X_Lu)*`RooYFH|xqwd7uU+6}G+09;$B_iw?mvZf@PYPO95|ZA@7u+b~MTd#z
zmQq?73;j32Vox^xB~?1r^=$a68II!}1V0Zi>1OeqW)dtpr2H>q+aMpQbh?y2GnQEw
z*qxzQ2UngBw;8Z*c>bBp7(ZxOt{!QmTNTLB+Y9p`hyg&&^3D09J{fBT`D(zZZ!8+o
zHy9S{p2v{K5KJM%_GwzclK1lj6Uu7Gos<BPjd3#E;ssk4FnDl)F7c5vM-GDaUk~6+
zlqW;}q>|Cw)kYYG8IGh!dj?lNV29;_CvGp{ikLVI0%Y$$Ias6f#K<{8(3fcbCg^D2
zU4>HHJH`1?uSUylRR1gu(%&ysNL*9MHVR9I<;U{2WD5O&d(^IAIG)pDY7H2Ej&I*6
zq(|Q#RF-s0w99TEBmk;NM;PaF`a{Gh`)?F2$=ECl?Lm5x?tbZVimb}XbI00+A03`d
z;YmqD3ZMWj4?U0+z$67M)yt>YlQwXr;3Fc-^+v)!mm+yc{}EZ59DAr#{2mdm^l!W>
zg$NmQIMCWT;<3%$E0B61f(b$onL%#=3Lft!N{-cwOJK91yY9;esq5(PTmE)?>KxRk
z0cmYKb$GL7o}bj7N=Lo+%j41=GGoQp>WR&wE#O+j|Ajb&c~CLs+Mr6G|JY!uflQ9}
zYLf#MKxvjk>wRZCG0|m`%?N~dtSc)YZw3-Y=XoR~HJ1mTJ6x%_FVa!jp>I%$PCaN!
zqU8}F403VnuWk(tJ!!gDEdDHWsV<(%v>3w`(kBgBSwH{|dtj$pp3E-ajkL3?*a$fd
z3-DZ=sr0vjHAUNm_E0HfwW?*|i^RHX1o7)#VStoJyZQL(sw}~`QT}x^9=zq+enEwF
zJBzY2J)LN#wAz3Jnc_jS?l4zXe>gao#$uULYwg64M5}_y;I=th57OUZl`?X;l`ngg
z>DRU@^`fKj<}>~&wAWqsG*95e(`|<x4S0xq#67)wRLoPhw-&hC(V;$Nx6s68y&sFH
z<Qr!reM+=Ml3k>X*JznEOkS+I)sUZLV439x8nQJ7=9EzldNcD?Un#BE-q2sI<E?+W
zSLPf2;Pyz!rNrzDH)t+Q_(X2t^POxUYJNaONbcJMGPoidRIf>7^b;;Jz@?>;1ehR6
zJH(-Ikjoe+Q$iX&+`*>^nG}i=*rx@lxVW6Qz|OJBHUw)-NNhMGFK?r&4r@^8I3DSK
z6rS-zZFuIO!wk24)VMYxl;j{6$wPzNI0Rt7snX($?K3^c;f-otR>o>a`Qpu?ee#gl
zZVcdJo2t?ZKX=cSt|$16z6@O20FuUs;Yi$q4pgJkub`ZeBs0n8=<qeQkar<jP4!@)
zaB1^Ci9@R8%EUaJuWkgf41;i9$$`TR!HDnv(!m8$4stRPS9C~D#K;sV3X@_a67P3>
zp->Wr&!5e-h<`18WPEr#GU41B${^YeLp(SW9r>l=^9VYUrDb7rUf38&0U8j4FHi?t
zA1HK`7n-isc@`Z~5|cmd#Ueu4EzEJ$)^?-v&^P+;1|xy28Men1>Mk1DMa-%o;42rb
zmtw{n2M=(Jqf|Obau`9@=xEqNlW00ttt;caQ}{==prl6{YxC&icK#H3ZXnO6g`V6%
znJnzX<=MBRjfNk1Ylw7c19nj9U1~$v3=N7hL~FH~RaivBMenI-IQ0&}hBw+NF2H{x
zWYc-GJ$B3fX|nN;?YP2Eql|d>5NNZ?xvP5(efQGsK;8s}LbKajqZU~v2GT$jq%Gv6
zzs0~8JSIv-BT<*i+gP%Pel~-W;*{>8%Eb*eNSd{(8!*6#+__ga0OU!YpF0sn^ZuG|
znf>+bvw0A7&)izwW<9Gcu3jUfz2CxvdBwpRN)5O6ZLK$%H+3r`hJf8x4BU9oFr8sf
z@;q!&tVWg7=5gx%b)vWT)l5s@Eq(Bj3%kiJw(l=VK17j8?9UJK^xIV0{CpCZO<AE)
zCmAb7a<dtIGR5mq2e6Ns>KhH|O7G5wnb|&^e;G2wmbFC@kvB=kwL8yh9656EoL*7d
z^pkKU49H?KoIE*88WdIqveC>@3HA-jNj}H8;ZWG?ySmF8FK;gdM}^Km0s5_yIXnpC
zyo5HDfdK>2SZj^YbWS>}Jyp4=s9RsLXzzf3<Vr1ER%B|i4PaZnLIL6UIk5-4Z;2s(
z`%ml@rs+-)l^%v<8|wwo3Xa1S(mSm<Ytb@@Pi4lZIM7uhn6BzE5>S(5qxzk#LCmtc
zxQvYPRqX)(BVe483!=U7o=rsLTOr=Za;wL?(8Aab@UOy`xIZJ)MebhQX`?u7^XOCd
zR?dV+n&B0U1_Ts<jht?cvx{dtqbtxd#fv9l?Ds5SH7^hDi`7$b@wzyidQ?r5EhwY1
zSDPP(%Q(NeL?=c3%-Y1<1+jZTD5lb^ON;W2U4Y0y|H$}TgibTQ`vGDv5+!N-&WSIV
zZkECua9`$+kX?qj_FvbPW#vs&R??A_apjDR(bXSX09bWP#lF+O>oe&?-_jl}RqcMY
zv5*(QA8MP83_!0v`{2$~^P4I)?P`g|C9T_hrI=9#-PBuqzlTs4es3KuN)1LvCZ7Rs
zz8GT-aqOgD;3m_yTa?ySPsGZ+QVU~xahgyZ;%<tUnLMxBnY@%QeJnnHpw7hddT5sA
zjJ4ev1Jcyu_UKu;vlAWXKmB5qvFVmA9}SB=CM?)_i?*$|>ff;!?ozE}4$Z>uvVAZK
zjC3gkdatlzOAV!`I<dYP^`l^ce;bb=esQ_Qz#b+_>WAiK?FM+Jv}TON%CxtzQ*0hf
zE9LN&j)6<?1JUO4(do3ZCDEgqJ8p8BP<Yl|0!~k7<BND^!>v)HC>K}FjQX=i*u1Rx
z@LDl4&BYH5Z(n|K$=Bzyt@U^}Sjeu9Ve-JIG=G0t0bEWMaA)kIC=uKjlvOU$M`4%0
z4XghY-<#B`Rr_XqeMqVRnY#DO@6MH-%$_AWH4D>T;~ir&(ew~bkvXEL%KVGVE#ax-
z0c2J41$XOc49Y7NqnG{brSGva%$XaFt5ad53)!fHmGQJ$y`qLJ|E6G`E_7G4^|4-#
zxQtyi)OlOsDP7`Uid5{>tYDM6RD9RFu^D<}hG~&n;!;{9b|`K09}h1}3o(hh3^0_5
ziMmx+DG8Mg8CRS&p{1LcQl;06d#2ah1H^-ZRV*?k;|WhA0c<iCRC9I?(|XTz`8D7C
zh)w-r+|3xHg+ui0qL2!rc<SL|qU?;EK)&n16YR2_w>J+tPFBgr#3QTaWXHP-hk|-w
zLadsIzT98adv`a^HbGKdK$ivqBJ*l=IGk!PZw7-`(ocY|)dU`zZnvOW1$*sJz~m#I
z<}LaP5hEHbJ~OfCBhvjZ@!T{{bH>z1m~SGmZMAif-@<w=6;;x|@Tj#_bAEDF)6fNB
z6QfqCpJJWHyJDC&_FEex44Cv>EKi89KCnI@tjbECGR9@<>iTSTyKHKak1bk3PT9gI
zSW5TX$3EyV5_<{q>+TNx_fX+)0NVj3b-7p|0vqv7(1oO*jjM@&dI3$Q7hdBr%p1J`
z5%!*5ywd42b26(7EguX><~%0Qao>dFE%r|`xqnzu_Onsd9tc^A@kR~W&S=&eA^9LV
z7%}MQAkDrU_1VLma;DvHU>ts-N=_hFPhAyQTwI4INd-z>S4sHGKU<F_0k|=p^xTKZ
zQk^c^cB%TK)PFoq3un!fbfkT?D;KeIh<(E3;A7Sj<fd?09!9}$M>*S&cM<~w1am(I
z`8@nc3Z!G$KEOHpvU@rD(Ylyd+B>@A+dkXTxf`nJ=csg@`%Y66$7F@k`TJ@_X8sX^
znHGcCvC;;QF~Ba&<d(L20!Cmx@H~9q#a^kBiq;%zY2-P3JTJFhPv!)L<5^P}gLx7w
zI5QV$w)R%3Lc?@@CpSr#)Exckmn=(q<o%)Yds4>K6#!6W-3=tA^idr9hY<*cKf6)9
zbXXLtSrye|-j+IG+ajBFRZdStp)2Oglr_6(rPTO$D=u}R4Q-V50k_mO-5Yw3h48Qc
z7iI4hURn69dsfVfRk3YWP_b=QP;tdJW^CKGZQH5Xwox%V_1|Zo)4RKG`uT3w&AM50
zKI<FrZ@go?a~2;hu2knSTt7*gNxCr;^xI(W%!zD37To(MAX;<s^1Co4x|%N=U(}k`
zx$UOgI*86+I$4`LPadiY1gy6_U9z#EC&~Yh#PvdOke1IZz9CE7p!X}jZXS$#Uub`F
z=!lZ?+fEsFU=LhS1hxRssGqf=ISIqxMiOPqT0S(WdwF0g#_1`SrCH)AQZQb(t5Q=w
z7yqTNHP{tk%u75{BE?c9ek~d<FLdmIv!sJ`>T+fXDXX{NGm+iT%K1B%3xeH-R)-?C
zEn9E1D2qcm3d$AVbJM}OI!sc$E1y+wc5W<27;9y&S*K=>nzILB&a;%7C{LNnPhBVg
zpEG1-LQ*VeEQKB3E5$<!Q3jR$+I!SNi=S%Og9st0h~aKAM)M&lNo`%qK~iUIyjk~S
zVZ5DHbm*>i%ksEBmO|)0*Uc@wQuQWIvquN~$+fLdxoj~itXJySvsS^E?|fIe0#(**
z-~P$pagfy9Bp{7g&!y*wS-rMQ-4Rx0ysQ-Z+>$Tk(6QqDPyN;U5+l~y--gq}g=@oR
zbB!gR`6qB$)BU+2BB){H>=u;&`{NT~hY2bT?eyf!j5eRqt$-ZV54>B4KeH0;MRgZj
z{6l6JFWE(YnX-aE@TAGJ|MPWKLew%t765H>_nJ>G1vgcLweGj^^muNHOhC%OJCtqO
zGUosXm-K~@MN^e=nbQm>MkJW(A8dN4#~m?vN>CO5A*x7X2L$JVJP496*K0o#8`lFv
zM4RoGsehLmnNZYL@Svk<%+}tnDldoy#MG?3dcvrB*K1V9)0=?`z~Wzvj217pcLH&C
zEkQvuhY@cIw%)&NHQ1XL=pfz#=bpYYzA(aX=B#{A#5URkWor4~;@mnd@t|1`B7<0-
z_g`@~e;s%jKR2M;xPV!u@HXu=I%%O&Co4A;9^=&PbN3&C4q!p}GZ|6jsz0~i>92^#
z22C#W`0O2zP9)mx0N!;CI68Sc6?lJ_i}d+XqfxlINe0Xn#Oq7)>_3(){csB0{x<KL
z_^I(pEMaJ<{|+6x4-}S5lP!b{J3en1<XR=Q4KWOXoYVy^(79|Bt8DeBno2lB4P%nS
z>?qyD08v=(YFOyg?`gQK^jnlj*IJKBD{E^jyCD11_7NuR1_+<LyUn+q_f}70HQ0#H
zJh#G&-!xR&eCG7IimfIa|CoPCdr$LfyeH5ycs8r7NU4Nvr%R%U&ZFa}kE8f#3~8UE
zWh)J9eEMHPz2=OInL=tFX2=WyG68`qCI~xAgE~sIKX+;_o{c%@ziixAK!LK!rhAP?
zg)o4?D;Q)Nl;2kW!4-XwY#1JNFkSF})9Q6F^XGjq7f0Tr1q-F=gqXC=-I#p2)q#Ur
z)%dgU@l*kF<1D{Zl;3;1J{u%mDvwf%oUOssu+vNQSi};p-u90w4O!tS6RY4wp4d~c
zR-4zAC{I4a&kyCj{i~yb=gvUnKPeM4o{PW}*Een~codZ~A<mp;4d*;8__c+{D=}}m
ztvyH!zon(?G%is%HGlSuV??CN6?L{Jka4>gT$d{Wnf$T$c6)=cM~!}AwIVR<PYzV_
z0yqKW!zt^-KopjIJE1=&F5$2Rqz3r7WIhLtkL2kO8oKzaJt-IH9YC8dDRR|vyWbsR
zH8kg2cNK2HKNLaZcEs!brEFf-$L^9?*5(TIDQrZ}$38Lb{A5E{_wo;NjJ}~r^(T^;
z^1_Ejg;fgfHJr6(mqh(MN?0@f5FwGUzj7Z57oDk;aWdosS$>@>G1nah!*`+DGg?P6
zSAP|%lZRJ;M??Yo>`xqLvN!qR=>~0=9ibF7v}~L3ywXz#<@4fjec;S7<VWb~T_5_;
zn>oe16V3)HF4X+;@vCiTiY3mAL2uPZuJV{(xm7F>o^Vb+WyPc(+eD|ao&++PV^7-c
z{?6&>194i-<hnl6mxVKSe=YOsAqVU^*q#>6eF=7>fkMETTrQnu;}3Q$rn<|-s;U`X
z_6wm=S5g6*2#AP)xPtnL%%f`T7WpanfSYzZ2?otv2y-p;t&#TsLJmsN4&JRL^4X$}
z7x9PLg9BCE)|xOKa&MM<k~<Y#c6qTb=$EufiyaN^!tiB4q~%YUZEgj^_uy^mn)o)C
zWEx;;YP^AoRVGuj9f(v?OR+mx?xnaK96$TXg9El@Yn<&TX^Q&GVVVU~F}Av#ofx5~
zc7+tw->9KChPFJnZCAC#mUtu4tlC|HK{a$jyipk<2z+T5tY!%m?Q{E!T-3!k8jb%(
z5CsEVu!>;(@5T+9u3$k-pgdZo_rO+)Koxc?TXjUwc<xW4D#-?8I*c}SwmtIzEWa8(
zglwMOr(OFJdA^n6xveC)j`M{c72=&d3PLa@xd%C;Svx^13<Jv{iMGi^%&>v*R3q9|
zTzDfX70B^ot^-J{1MNCt`Uec&txo`9zN0lt_13xXMRdJojW9Fx_vKzr4nWk{jJeQZ
zSeAqsg6#LZ)RgFw5G6M|NI?JTnP@3_MlnyETB<}`XOlKSeFm$Ire)$v&ChX=@v$^P
zOGr^WaQep;6{cvz(bmss?0$S_N@?h-`EpOftI!ZSgLG|1^<Ih=9(wl|ffH)L07&ne
zNNCM33WF~rJ=JcKt){|z-~bV)9&wp8Csv-2SW-e5x|7EDld5diwfI~HuYPahKN#Ge
znS&ni?ypL6LbV+%8Bq}H7s?@`RrWHOvJIE*GWkDWu}O>s1C7LN{cP$v=UM~{Qt1oa
zc-hCxVd=Ici!Qc0j3Qb$##BgCeo)$3&P9gps|w$`cVz7~4HaTNvjA+CJ1hJc6qDqw
z5Z6nu=>nks+Qsk5rsz&NhrYe_w?^x?*@ffeK%gJ&H>mGEV*Z7bDqOeUMS@T2w0IBb
zloUM4Cv&{74v6IDrmn@Z^k~$}7$2})<_3Et1@pZpg+P~f|4<vjf7|>)vSs?JoXR`i
z5h_xZ;9uVCHQJ9=aR&$>W$Hlf87LHyCmMOWGc+%&@UqebyZYG8h}`9?OVgfSxc}Na
z3y-f7nP$qll}DoPMHWTsL*>5#8-Yx9B~&=h*9>oT))McAHww$l^!*mjl7UmuYE$>M
zV5(hXOwo_@hTz5!TMISPPzrQ;lz#bk**~>3b|eNM9`Mh`f(26L4gcQeQT(Pf@ii5V
zul1+p<<GVwJ=h+ou|DhI7$OM5tU_SuY!zR8)koU9ov~v|csj8%B4lYXbt1KM7Z}a=
zboT^iJH}9tOFzQq44OFA4*b&F&R{yq*iD_d!3TeA*7AY=?KZxt7hCCKb)Rt3J=urZ
z_54!%vG@~_<rD};n$jHL8P_Z~L-mP3MW{niQtRwtJQZ!wGF5){r`Hyt=bhOP3-+hO
z{Ecnj;45Ds>8?scn$&!v?jQet6WSy^VN3EBAO)Bn==iVAac7SkmvS5$=hT|lX$~W{
zwN8jnXcl}k{hVIJ^A87OusJyS9l7q!QMNBiR%#$X1PV;*&CPw6L-en|O4cm;4(wK;
zcgGVI!pD_;Ybt`@#JKnlOHyUUduPQb(c#H|M-O9+wP4}t=K2lAyM%l<b~jd^`aYk#
zWV7o;X9Jxu%bj9EY1{cMp`Y*ajoNV-BJY8i!HYvK{Pb^oQxat7xy}c{H7?2cTjA_j
z_nLL`2{Ev+#y}387he^f$HG&QXmz7Fk*i*w1-}v~tW=~<@azem!Zc(&kG(HhfgLg7
z2Em^O0XqgowkRO^dICm!Ay@i7n`=XVEdOM9&S=?$DhF!%HNth~M`={C#HWn%TVyVI
z4FHf)-PFO7FR;5;EBDAbU7l7aEaiybbW_*&cL08hAYXLbWONBuZ6+F&!h`N|akD+R
zI%07hQifQ};nS+~9E{l4=tl(Uv~lXwO{VEM?|EyPTiQ{m0fXvvwd-O{2+u-Ojn4;E
zgpRn7p|`?KWRZHnIv*JxuU1+UzxGQHZX6GcNZXL#Wn(Alph}0_&mrePh$jNg{u#8d
z3`jl-Jq|;{)AX6a#QHLD`i#30)~ZU^M5^mdInC-ELoq4b&1gEYqe%-PK6tVzLDfy}
z%xOk~BCcPP*GP4`utZqkGC|y$ah)o17#)Pe+skT9>YO^#!Pzz?VT#99T@2oLnVNGY
zn9DKl8GP~YJ#yCWMSu9>91i*L$LP=%z(4GE6)cVepOILyB0I+qJmAD4&!E4|--Ag=
zps*8ln0-8HYN_oaYAvq$jlfCzhxfB77GxRziR6|cGRK{hp>!|?$qfh=L-#WZaZf+P
z-_=x?1?9E&LX6VF(JMYh!t57}xD>BTFLOE3gXr{l;tA`1h1_aR$6=^Jm77oi^0DP#
zzu3dX++fY~8T!Oh23&x;I2hlo>14mup?2%nXZmCK){YpisLfg$*=d6PL=vX^#80N?
z1M~NHc??+lAYxIBP@AA$Og?>n5Zo)DOJzYSztS?W1MI(%2H`TssBZ&2zSu??c+__%
z68g++?$uxhzP@ANiI`$7)}DZIACX)>t(Q~d$HERFscWhPFXko0`45JjY`RqRfgoHW
zLNWfYH?7F(2MC5Rx9H7PjFI-WinLF@1i@Unv{s6H8R$QV-JN=4&w1oj>t`c5S9_7E
z4q7^0ta!YLZuQUb(s7LNi9m-zZ;$mU5E@L08%ipCqnt$@2L8+p#2gX?U0ml1$`sJA
ztnI1<hC7Tc|DAZ_vu*Z790RRCQyxo-D)=1a#1&mSL%~5O8FI_a^C0wFl%r+~Vo)e~
zVrmW}u*K5CGx>}OPX8gi>1qL$1q&Ucrg+=|H$;Pj->Ul|ZQJdtuHw5hdstUp?_xS%
z)D22n-JTzgUGUZQ`iq;~b+e5j_oIb)%ZXe1v6Hs*GafxLO5@t1|GIDN{eWImbN(F%
zaMh}qJiMuu6x%Vn@+W@?PGmI8z6B4TKmQ6}8ejbyN^r=abm;`^ixNd0%r}LwD|K^A
z12cfvy8h%`5#ni0NciUQRf_=QHvid$d*3#wo(%I;%a3qj_rS5T6ZxDe?~TNE-0%ju
z%duOsclvCK718w|p`GloeTHf}V&qRgFpf^NPX4s}f&BDYni%sj8u1B_^S#rVQ~SmL
zuASve$g0!v!H{O=p>3EP?co7SMjqGLF4$dj(o~XZP{<Q*yAR@7871XcdQM+*BQ|t2
zyaQ9v;msDpC`1P*x-cH%wQ93(0j>Z7{JANK^*5=ie!Yclx@vm4B4q;3*QqA}Uq)t)
z)Z2h$oqXqqRr}vPW`id!E5{;Q9VDGNUuars|A`o@xN!KploR%+-7d>GuH_KEcT406
z#)?go_}^#r*17}47+@jcbJveiKpI-M0U&)y?wnP?zw+Uy$)0jd{qhW+InMIALUCW>
z-~%f)=k%;eIn6Up`u*_|Mq3hy`+bt<b>;O{Rm$q2$-nG}I#+CO#D^k_pjDJ_e(Wq}
z8d4}I89Pxwx&IDzNiw&<HgulLNEnHmUrhs-3;CbR-;JS)Y%M`696=xd1;_KAiQSb}
ziBKbTu1kL*jns$y{%<5cU;9bKIa-5>g^_b9i74&jtt6-cHQnhzkN=wiMaw~LZ-bdN
zoB?(D)y`zBaN<H)3?A!UE0}YN*{?Haj%Fyqt)jncB*FC8mxrD=>s70F`hoM0y=!TL
z<UwfyBYV;-%w*lhb?cIalL)u8Jm_{j&wlmbqq>pkX>GhXWR<y)a^&~A){c8xL=}`n
zKonDfo;hY$Xy<WPu|yr-<P<02nA=u)K%Ma@wgqFgvQYSMLsMhD0VB<4WQic2>?Wx;
zez}kUK`e{nB&w^0QeC!!tMVg*8MZs#85+HBXv+Zk<aVTO5C0-!zN_P3eU!Fu*>U;}
zR5~mb(8m{_4uNE%?yRU1S`OHY1E<7cU@xV&l9K6j87Ft1s@KTuR9sc*Tfa$bcyU3h
z!=4D(ATc&pNffnTT$#s%>LHl?2NJ(Ggo~>PXsKk5^1g0b$A!wn4ZRLmkq{fKh)Lh8
zV5g#ty|P~(Lh|iT`K&#Mts83k8^iB)Ay!Mm3!S5z6;B++FqYScn|O*MnHp#g0Bmto
zjqx!1QA9~jCE2&=EX5%??I1j@T_M9bl(^B<`-~;>uyDh)<@6d;>Ewb`LD)$>lzA-{
zzH!ALHSSqGtTIIDwY8H_ocanlOH{^i{^@8G6sEOC55>H><04O=txk;HoX=M0=az=}
z*pH5j5?5uX*U!pIF)?=~EK<-8faR#Y$v?rNOAYWLB&6!qnvPdE8EGt9_+-aMW22aU
zrj*88ofv2acDLNKE?R1M^OAR?*Ah=hBahG1?7Uq2s04UmPt#a{L&Ew{$>5cdr}AAC
zPKrklOc$CNW7*W+BhCZJU?69T>5b`1vLL?ZvNgCUac|~ClUC&GP~IEBa-IJ?6vKpL
zR?lCw@Ii-RcAZKOpSk1Yo7yq<f>W|Obaebh(7jk(O&`VyjJ+ZAyL3mo`VDADInIku
zjIMAPs)0i{lBwZqJacC=?WKjcrL*#k09g$BEU}N}%%%xXQAetzOAU1|Ue-f-mTC?y
z$H|?WK0@t<L6MsCiGn)-108M5BPiIxUpY7lQqzvACita7(@;_dI*%x36qi9`R@LcP
z>(GMA2R5tnY(0jSjtAo=U!5|6W{)IFd7CXJWri9fbU?9%sY{FY;88Gp%Ro)YV>^Ff
zvmb7fm*?1OOt-kRKI8=(l2T>HJs*(;gU4Vx?)jnP!lzQPNP`QY$|<|O&?2+N!u=d&
z!74FG(2${ssYzSpu8=I4zcDzSm;3&+Ey;HYn@#f-`}}{JX7=<Kp)Z_<i_uo!ZMVi+
zzHr3s9)PZ^uPTa2$W`U*t~sy1m91XjwH&Q)ME}1$Gjo~)VUhlT8R77d?KKJfX8zQW
zPak``rz}QZL%cc9vGqin+~P|!Yx5X6bB2mGUGsj_#W8ag7V|^i&%%$tnx3SbBoVl6
zrNd56Q@hR;yU(_#!npIE_`B1skFicB+;(NX3KAiv=|UEyC+vUomDdu$Fe7CP^cA#<
zplWB2bB8)R2q2(<V#N~16?Cb6O*T&ymCE?h>ITKJb1u<|rU%tG$DBO|Ip(h8uH$bu
zob0$UWDgK<z4rsuTaeoG^~InLDk9k^t))k$qMa@dDi9Eq1C?go9<6V0YPJKm&R`_l
zb0mYvSYqQfu+2T+iRp49Z2h5?QNBlR5eXG4k5_k5<%M1$DF}^1$F^2)^VFdc=PP(0
zBvDj=d<Uq1{uh^QHAZ<VRuMB>*VWN+WY{$_K8x@k$3^9j))Locjif>a365{om!|EP
zUf8KYQj+)H7)i;g*C@_%{&SLRyVi+RZl2w}GBpChR0@9fEe`Pv_@gnCW+*A@@ZH@%
z+5qB&1l}1G`D=5J4OAZ<yt5!VS|GQCCiew^Krz+6dtq}E;@{RcLZLAF+gE{1G(jOX
zgHD0WVvd5mxZgP7nt3n?V!+Y0R*XSs6X9=x>Uq9BEiP1fHrr^uZx{hnHuT&vei53t
z(wgR_i?TjAVWynXSm8j#k*q~;#+a&r5*;^pjOBhw_Hb%q!DeACFLikM{Pu-Z_02ed
zX%EZrk0l03VN&Jc^P$GOyHI+*(^mef@uo@}`$eQN`UW}rham7$-V#?Tg4xa|S&Nb2
z<hu#P#xBo<Z67R`NVdlk9As<hZt~4emF}yNx##u&!{ZZ#qkgIU6EEng-`j@d2l?d8
zp*_lsXe<vw$3#1@5MJ~d$wZnnTwmi5#lxR2+M&r!A0m{VJu%Lr*p3k(%*(kCoHh1D
zD0aB6UDx$|I(^CsNoiRJ4BUlsi)1L2{043t)7))YuzY?JnUhMf+PtGs-nljXu$w27
zcC9qbQV(}+#iG4ygbp5kub6;Afvfav3Tdy=-8+d~(^?1CwH+CJHr{keXGjzu7Rxp)
zA4E^?4d#0F&gDC)|A*g4L+-9OWwfs0<B@ZWC&k?S;i%x!vW3S7*OI83=nr#~_=I9!
zYqIU(W6#Ba_!UMU`!$qJt#XD<`9ZK{^#3vYK(NewjusXFXS67570~@`*gfdb>e9sc
zMg7rRAv&f%??TAuB>%RX<LXdyfP@-xaS*LMxE_R_cr_3u?m?Uu^$#a5){5pSnybi0
zv-zs`4MZ`P*IZJ;^g#ULesKr-$*`VESL6oAAyK2B6<TSUJ%`$)_n=|}s4-q%;;{`j
zRG6TcghX9qWi)Dll9W#wjk5VubTC{5r6;SbRfb^u)0B}w?(f5>f-O;cYq^b}LGzfA
zCWS~OQRnQ6NF`@y4=hs9g|LoPDYi`&ySodtK5)>LIE&SW&zv>PwV9#bOS~i*S1n&i
z{jjrN4JoT|NM;?dFy%sLGM4I7{#s)!HLVn>*ohL0HJ}U(aZPys>sd31J3=EM8T$Lh
z^BG&uq;&`s++`+UNV?pR@qk}fCQ1#RH`?rU_*xG9f!|V#?A5PPDM$a|K)~-kPhG>`
z;*3o3-Nom1V6Ui~b4=#M@v^GQ`6yVNsg_uMTfg7L88rYVfJ#`e_Lu{w0Os>h!JF>*
z3X4H70t(f~I}yF^x`GYU0>=N#_7pjM#mxluw~-m_GpH4A0BPWgkcp>A2MXo+YyTg8
z+$3#hq=W%<=b-JO5>t3Kt`X$sYGaUuvz=IHJ}sX#KX9!PD~}y7Ht`b>9Eyn4xkzgM
zsDGxXzAmMKgXpfD%rNu=TvfbtbBshZKtU0+T~=47@eF@nY%V`f?uNwjKP31lF`+rO
zYuLX0ZwFRB%&ie<DqWah*Y_7{G5$DvV{<GxLFc@2mf7;alo3>IdP<Aw+S0bSXP1bm
z*`QO4hBYY2c7^*}+JfRj#qvV=Z!o#xk1kUVB_u`B81$~_^{zX&%6f~Et~<coI)-7{
zmX}U6+r;CjKEIs%@oi1Fm%XSfCP(qUKtv$+SGH7WWzAE<blCaAZ&9Vh=H_m+C=q*B
z)6W0=Q&5U(!$?NFwi3Y8Kg8>8;-Ohw0H#ckkOS!E?>sK2t<Uz%3LSYOV2V_dZkTCe
zQirHrPujh`{v4C)uYPU=q*f8WwxdGqL}5*$NpWK!+C#jgaIyy$jwppW<2WG5=BK+h
zUX8RdvAkqPhJajsf5`Bg93|~O$~B?SC*9Ud4gL;}I7^C5$RkaHJJL(tz-N_!O|gTn
z2+@Xlr(duXo*yk~Bkui!7b%R6KWAC`<@PAECf3n`O6k1VfKp2nD9<Bg`|@9gDjO`4
z?!E#u#LREEr8z&rR<XuuThVVymCYCQYkn#Sb*g%CV&NQo31!aW<4uCS%ZqrH_G<aF
zYK5moM$GF5%8c~a;X|37j$&_v^RQLhWGrkzUvz>w(GnrBwtyPUSQ*bA4E}GW262)*
z9Kn?g|2o?c+ddHQwQ1KC?Q+Edb-erJSu~L!RDLP$T!m>?<ZA%7)8kx`ze!4h`u}OA
zLU#(FkJK>w56fJSTK+GVDVKFkuP{lKH1IW%Zj~jBPY$ndw=<67DJ7PuKGY7ezFJpb
zj~|*7(xQF>6AIcMz&!qUfrux<>#N?9*<1{V<_Z(gPw!lHLoxdpkcKk3J$38Q7DfjD
z(PQ{7UqYv*Zmni@F4~}dDV)Ei@AojBuSEAElkIjY6SBT=p)ReNaX6yfXld4K+c<9B
zh0YF@Qwi$Q3ol!NjqH7Gx#CGkVx}}?$uW2qkt=H$aa0G4M5hMb3r43lNGe^qB~Y%L
ze|CAHgZM6sTQ#?K{hCg#E;2Hd!=xsd`ulsI<QtEUsl40sE>yk`KfZ>51$1~Sex~*I
z!nE0BIx9Ok52!W5{=Fuf<cY<D+#qVh>R&egu<%P(UZ}t9XE$jDJ*Jz2jhqvk=beLA
z#l8!ZgxCj&*?dNq80)C(TNyIAC~k0>M&1m>mUftZ8Dl}&I77?(Jcu(F96beW5VtGr
z_qad=uT_VF<S7#~x@{Aot`JX5F5$JYA@l`})-}T-b!b7Ww(sqpnvpS^f+lzbTSa-r
z^x|3{GY+@?WSc^TBK)Sx0)N|#M!+K4|JmvncK8LX_s(5CV^2EBG@Rk(IOjK#!4eY4
zJVU<`yV5P52$SXnLv%9QgDQ$<?3ddxO@4E-eY8qL&tp=g!RDLD54|4u8Vb`+zL)P<
zV{eB_&=-Jjg0|QpGDF+K7#>1F8#@YwQ*$z`=3pQ4<=kQPMf}3XBClqUR$JPG<K^5Q
zQT7J_A~fvZ!pty4s*cx@H+1K#S`?+N^QG>JD!-hi$A~v3)^nSkDx05zQ=VX%MuuV&
zllKiCsiJ8XFty%sq>d3)HG}VxX;Gn)kMrg8epGkuS0$KZDt>2k`u()qNG`AD+el#9
zCEf_2Z<w`coy6z*bI^)yZEweqtsWmv9U%ePLP*d?;Tk2{I)Xfg5xrnHCCSoM=t8zT
zq4Sb|Nz*UFLu8j-m%zuH%dSO)dlSbvQ>mpgSJv#^)o-@(&$B|(#x?FxqLC3E4&ufM
z+$jYIl8+n5SO>sTi2LKvv33TeKmPgd>NFUJUMO}L`?JKU#Jm*mvRuf!6+W5rGr(f8
z{;~%1YA;IuIb=~5Xt!~`Dq~TkP_aNd-A{(!MB%$nF!x35JN6sQaoibCz?Q)quH&I>
z+e*$>B=`JIfcVgu_8T3e<AP*0hj?r%m`~)JcE1x+g1eCe&6E4(_hDsrHA_qv+AExx
z1e1PsC3JQM5G((`2e^qg?XHJx_Q2E*8K>%AztoKojh(FIP)4hZ0L~20FmdxLpSvTa
zT?hk0Qq)&_;#EsqsFlnyRi6}7vr<^T$c_x&1g8oeOK4$&L6^2?B$z=v0i3AGNuNJ#
zV#)Cmym@+xkjH)!DQ}Be%%^?u5xY11NnO{PQC)e#2F>5O6a6E&yCZ(TzXBjkxQ*1p
zH^i=S0y>WA8BNKrJ|(BF?ych>>-_T8aKcaXjs9cXp5u@>BKT}vu^g;7ssEgLKE7w*
zxCNROI&c&n2y>z*MJmmo2~WCt8O<>jFVyJH2eipZ*w^Mq=9|z+RtrnbP&Tm|8J)Bm
z@s<OE#IZK3_6O~=o`Elhz`v9yaFk!PFp_5%36XT*zb~yn7Av)>_%YSdV8XQeh|tFz
zjW3*cRvmtI{a$j<%rQ%A9!6MnQyKW#=aN2ru-ly|leSh8`Hg9hLOXM-E^hjdyjg)+
zi7GO9(v&olpc70purS}3q+Y*2ZFu`w9%2zC-Ir*YK9bvAb!}e-XjD3x$*xcER$;zx
zeu#t`Q$}s7z>(Z&6h4P94V46T-U{HWk&?U6%V~A`*ocv?sR&S_<$ePLaY6+R0)xYS
zg&dQC@92Zf_|3fS`(@lf?!t$vk|27iLlR@hK^pTxeY+W^Mu#$tm?Q;mZI)yH^lQ%S
z{l;bhdVi+W7`a>#kQzEP{9Z}@{g6-G*$i9mWQkMTIXAUXua!7Xs8-O#V<yvZm0xbD
zUPWX93GeUIujG?1nCU&WQzPb{#wXAXru~f?z}BAZE)cAmaO*o>OtzkFpFQKZ!;iJ+
ziD(@nk+_PleUQyU^c^%>zsged?+dnNx|;HGtilAfzX)Yu140>{@Ie8kgX$f5OqSlO
zF!6Uzb;@S!e*G&Su&Cw_iplDJB2p+eJ3i;U3qwo&y9_r!tWwpz46Pk3*7>YkJxd}w
zgDZew%o&X9<5<e?X6sbwnmyOYtNUQgYxzGit&Uv8wrFDIshzqRudq*=Pc~bQ8v@r)
zH@-MNFRB$g0xa>ON$cBJir9Bz8(~m{#HR2eTJJU_9LI1WN6!g2-U&nFmD+OK%wFZ+
z2sW8z`!$}m+n?=PWQ5Tukq2gD)y+jnkWK5|=z->#V8$|&HMcyd3@vIBR)0Z+<ItQO
z7E;4$3a2rO$%)<5n|EeBk3P3dqN6jjJAt9uf?a_VV7zKYN3t(xj=-Y;(zMU!)&Xm`
zWqu#CZj@;YoR=wq7dbYQY>lX9vx%^&;9z1vq{%s^oKc7tsu#C|u}-FmX6QB_xgc5@
zn`dS<A(`O%{c1%;qRF7++viR_jXFhZwU)Rq1V49Kcj<x0Y~ne7m)O{%SQO^R7#x(T
zqFSR}pf_;ap&dl*khfOoZ!n!2{!M7P>9<#&ggaX=i*opipL6Cl>L^@15#_l;$lO#y
za_zX{%V&#Ky-f)-$b7<ke<aA$ZsD1NabhEJ81(?O8lk6p>I_kPeHw9r9mQWeUB;y$
zn8H4E$EX)Ysvn+bdmvB61j#$48@4RC?ndu@K#P0g=;t&>*>p3BZ#@xwdBfO5BZ-HH
ze6w9X4PO+mHJl_4Dd)<U$Yp<(MOAtT39TM9E|4T&drg|p&IrCAE*%>YL`@xwa`PTU
zt_n`a#AA?B4!P>80PoVrDAA<f-r$3%e^JDzOxwHNh~c(^Ah~_KAR9(VZT|H&{`u_`
z;61rVV#@41K^<Yjsq&nAG2+*Hw=SVeWN2;Fw!dm!Q*idKZ1odpau&G8_ezu}km%5D
zdppHn@6E_Yj}-s8sRqL7aXO|AiKeptIg{~i>h?Y7BPWaQEoU*fBlV}bewm>4*a1?-
z^noc|9QwyXNjoWQk&!7qoh<I>Oh$k;08`HCl`W81N-{W4i??DM7UNN-_ngOrPXfoM
z_Ze9a{x7ng?xyt(&mg1ya^iV^3>P#WVH6<<f9AsN6&LMWTn|sDqP>5BsI#p&b?=Fv
z94Y8iU)<*()0>5_<4Z@b0zK@!sxUYF{XP+z5DR5)xNd-8oxGmCD~W`XpwS>;gnPIb
zAW9?Lc_UrF2Hx&$@7kg#Xj6`RZK_a@3nR}UPcB~woiH*u#r_-h^C+RM^rOK<{hbq;
z1}$mIx@e9M&D4KPB@DklKMxOe0-K<X<y`9|v=^qIyf_|+vBXLRr16W1inklhTp~L#
z)L*6N|KtiSCV>g>--Z4}dIfCFu$!<m4Us#l_dpWHPsuL5+yXXm3%4YV_t76J^fcvK
z6!<E{E8MVuJx({jq0+GDc|<wUE2P%hY^*j~>GdGtoq$#CjLkjV@k2Rt+Gw;seeve4
zs|yO9*qGp)H8sTyuUjA1SaxhgOMk+2J&rVz=4JD{uB;=LT@BET=mJ`;!#_6zon(!D
zpB%|!M*UNgjG(A?9$4Ph4vDMLTFZX6DSn*q)#OyP6w%+wPUIKkl5|E(e{%U#SHF<0
z*{ixTZd)8)oY5dzo_5bj_Ym?PC&UO|Caws>{KT=9Q<0v>d~fgf+IS-!Hw;pyp(9<n
z&*7JJdf@{tvzf{Xo_V0qM)R=h5xcUyL9m2yP|;&Q9i^z)ItEJK<{*Fm(M4!H^6px2
zaAl}5r`j0t)F2t8@G5NI&J^*01Cm#(`2HwF{-5)qHe!0b*PMz$3cUtJErdg1G%3+S
z)H#qd7miboj}PW)Whj9S&xvzW_cBSFq?gZKaOcVO+JZVQ5kT#R9RGd$H(!M$DoK1y
zZH?Gq`QS*kQQ1hi*RjY;65^m(gg9+|2&=Lqx|M79rZ}3{ssosRyexfVl*}5pZWH19
zQH>vSuA{Ri6VM+x_i3{fzcivWUhAcyX?!MKz5Q)GFPK_tAUvf*-V*u&ehAMtA<l6?
z_p)ak8%?sT58#<wkR;E3`|I`}?}L6;(Z+O?y8z-1s%Vn~NAfDTP-jR!2wd3bdTHFg
zK%#q_c)bguo9k}b0dFg;P%{?jGt?ao6!Kcha6hIEM=p&Db#v!tZhh+abv=6B9pnH?
z5pO4O6Xn(c_1~Z7>@@Np1F~xH9V##6RHX?G1qcYhmqYLFr&Oh`7P=5}FHkzJ-`TZo
z^cC}A5H$nsXw7Y3I`bv6cPXf_6$Q<pBB^RQu!y@Yqb@wea7fW1YVf|tUNl=qk70Tl
zF8AM$&#|P5aCwc+kjyCi#>E^B8xjAAeY?Lp`AvvuB?5{e(wpllQmeK3Lp%%N4D+k|
z6pOMYfQq?H-CgFjfjzkUxw!<Y(PCv!)=@gRWOl0HrFK$QR7j{Kp|n6#gtFDcI25^d
zX8T>EX{%-xJJcCJpEnI{c}A4{Q<snMs=>d(p!ymLZ+Oe7*haMreul>$f$)tPloz9F
z<VCCp9(T6hYT#Rwl6%pcXkGXOSFycz=dP;*bMfMYWB2GFVN3sh);WuE^<ty3J)e{0
zd)WfjDqA#|3Yp1{c^x3~-3!Lq8X@LN8J}<{O)DKW1#_X<9lyZM)E;x#wpH4Z@kluG
zFXDpII|yf5i2!t@JhVffYVj9n?|0-}xuAqEH$%a%sDxQ=d!~NTzC*4>JNAPsavT5{
zf(V?|^@g4_{H@2JzctT~Ed2*+jo=5i+d~&3uZq1k^w$uc!Qtk-fVZC(nzACg_w%>w
z+@?asuE1&h<2*1{F9?A_w2$$bPbMUXjKkW^*p6o!g)U2_SdKZGJ1l+x`VO|@iEdz-
zU>!b*(_>!o>Wxy16w936xHdGx`2%1LXLQKlbw|PF(W}z%_D9<DGhPz>OxVqXTDFTS
zGl&7lw_chY&5$*oDT%dsw;9b|8A`kk#{UuJMjZQzEFy<^segYFv|e6A-0$7j_79G*
zwk3ZyeW>9b>@W0+xwM__zaeqoU8qUVdRHwYcF^Y{W~=-%rQyI6c$Cv)OA1tedtjv>
z&Z`swQ}_AG>5}aT4Vs0DseT0Hg}WWn<JYB;vy2tm`Z>p*ch#5@axL(yyT4kuXj`IZ
zID*Mazv#*DrnycZqYV>86+h-<J6@(=S4$<kJC-_@$CM7ZRitVzNPm?wvb|FX9%?CF
zea!TDi;Ls_(?c;hE*gO6hX)Yj;T@y-W>o(g+D9Z%mWO%tR(F>Gn;v1qcG@$%z=4f4
z$WAG|%idTX?^9<H$)S+_(b7^BAl^Evg?PtRS`r=A;wiw#nbn?`EufrHTFe#T_$S`z
z`WP;%Vv$H%Ry=*=V0vEh_EfW(<0loP_^)UL=?#wYVsO8DXYU<WaU_5puP)(6NrN_N
zOv{KUOgsA500Iup)UTSlj;}1gcA4>9#C7$G%r@e96pEg*<7gD)IP&Y%RpLCxh3A)F
zGSg4MYeLJl?zHDwpV3K6>f1?KqJ-^Y3R5PNTpp-vWvv}ahw-{c2fN%F*_Zzyw3fvt
z3?ddYe6@b=FjRnCYzBa%S<LP+aitIi<zb=Ge#VoBk$X6dFcsBhh|hZJ{5htT(P_n@
z?>ezk4Qy@Ge8PpMzi2A;t@-uH1I*##etAE{i~RtOzzV7KgccsML8hf8Gj*^ckm76o
zk{i*fj-AkcPgBm6OK%~$obD_JTtrR5#wn=^5!Lig4w@@Pjv|o6t2adF6H~R|TTjMd
zdA4-eGF($v-^mwkxROvxCF9t!=Mr)CM|tfH&EWtar(Gj^WkRuJ89o;8gGSnEO^T=%
zOu}}jb~fqVb@p7!@>ay+O{hH@(i(Ct<5h{e-uHCGItg^CD7hZ(D6n&2gv6&?jWkCY
zp2rn@E{o*uat*lE?7MKZ_U3v!rI~B}CYd*OVdFv-@l0|#0EQI9iZS<J#jupXracl)
zIjD`%b^hi+8AH-)KW8;UH_K%Cn%%+0;sj;N*PmKaiZW&tcF>mT?B6M(yF_ku$<(Iq
z`YjNq>-wQ_LT}UWYPfP548ksw(P}Slm0Dw?TjB!&p%T7FYSz(BJMq582$`@~)mPEY
zp(gDI#EruIx(<ZDwjpJv^bftZnf98c87;pqetl;r3;#*kl<L?Ds#W^@^ZO8WvEWEo
zp+I}H5bpL6?o5^^ob6Au4J|DPcT8&x<E#Zy;%y`<0)NJ~ZTr#kJ06@ZH8rKgq5hcK
zE4YgyV1C@6Q4fDCE=z+^?tLoP_13h5(Lu7ouBW8n2KPHXoyrYdRcwwsXK^>2@Z|AS
z0Pa)TL@uJmMqX2Y)Zn9&lib^88XmK767EP}kXnC9!Yb1<N@{{z+h_*(;}W)s*Y-^i
zBDG+}R{L8}1Hcu@_)Q86EaXsWxz<*%uVlC#0CA@O>57qtL{Z7JhpsWp{OC%3y8F6x
zWzC?GouEjKAthMUi(J48f1D5h1KZN5PPYqov-7%4ozo$9B$Ho-dfyg|{49#ti84#I
zV3t@MB<nv#H#Ocd7be6)fAm3n&yXyBbky^wEIiQ`geZUDdmV)tImBMLJP#TlR|Wn@
z<-H`SX)Ev2%kcfLXmPzjtPpDJ@+*9XOQRi?_zrAg1K4*t^L<IS$`pSldDKQNk3d+O
zvVkB;<?h}qj6;qLShkuKgqxmsH3W-wT55K(U3Ze+*}=FH0Xw(VijHRVaP*&D!Iz+h
zP%>K8kdkljjs%YGKroh^?p?@Ua>U}~^&2xG{YTRr^f;#WDt)<&=UXv1zuJ0F_G(+U
z3nHIXaa8_B)%v?*RHj(`|IgyPqhO(dnjxguR3lX?3>B&k(_s52C|nuAZy>thB-obi
z6d#pj(=7R&GHPhJS8gfaWJYf+_OgbJfY;gS-uUgt2mwSA$|y(((e%SaY&>T;8UElp
zA(e!dwKYsg5m}YMgBW?3Aal=<xArz8*H*KXRmu@5f^pjBsW0O98GA}RH-tLF`v%`D
z=%6fk&OLA#QESC0k5*cN0XUzW?>1i2CJN?g6t^GJVgTsG%9q-Mwz4|K9>{MaEF*TO
z<`#q#ZaBy^VjiLBbYh_G6qKlxO8%q$kb*w!|BgtujC9P^cR6je)d;(I$%a~?Z6{HY
zHXq|emi#}s6HRpC7JYwS5Al7leX*^|0uPr?ZHA_lztR!grtAY>6kI!Y7G63^(lmq_
zXkGcq{<@8{W$EN_1W`4|YPb$~nkO$PS<T7ODAH7d|4_j4Vzv6{51pXg%*&99Kd^9f
zFY@#VuM<|GfjQULGT_*d8(0uvRdASeS}8yMtzOjw6}fE2{|TX&zg-DKJvUEyK3n+8
zsBjXoYsb;GO6>#v%YS7mkL)&nF!^XXO|~ptd%llwDv}!_A@RNblHr=XsAd7W5Z1*b
zMsQfL{A^cyn*SC2)GxUsuxcJ&hR-TP88X*!X6Fd1d^zqPra~<&d$7u1lnE24yN&s2
zgFjG1#=hF;RJ_gqr8@x|2`;p^h8hJKjgj21=4bjm0Fx^E?lBU2pQV8*rWxCNDyZkL
z0S$Xy`8@v5XMmrO;%!-eOz1e@Is80k-ohodT3O;-5cjN--Lwpc*%q_3_)CHj<uW#7
z0FU5hAt~APE4@(F>VX|8Bxy2}G$n@WUFYQ2vnJdh7t)`S0L~%3f|m+2lg-YK)yjDF
zO$!DI$TYMTUAm|QvuQ<2u`iKmySVl3z`-FUEpX@fP7!Qn`}30K7x$=-0LHu%TDAI+
zX$^k|qHae-3Z75kB27T8`nGJJCw!DrpZ5Ee0EFjq?MuI&0cup&@I$ky{!>f`lW_bT
zH&@g$w{j0vd+tXp#cV!8V^}8Cs`H&K<i%hm&>1jaQoeX`?gNXu$ZqG-1C4v{r-(bt
zV{7?l(?CZltvO-fcz;$+^+|bzf+BLt(t42C$VhW(Y?m(IBP+R&q&}X<95*?$J^q|G
zkh15WeR)Ew;q<uEvcGqLfyyVs)`6lY!tdO>WK6YGB82+8f`D$3yogbdvDVklKAf2d
zB+y6mdB_WpZJif*cJs&z;V`w6u8Qyx&X1n5$gD==CWad2Xl{u+#qr^BBT(0JHwiTa
zq|PCX>mat`QHu`1?;&#JMrV2!wW$@m^o^Iv)2-wN(3|fIeEQ-C4<=0!C|z6Q(~D-^
zKrvEpF(aq2Zz3QhlJH>p*vPhP<>#gWh0(Qv7ge$ii<_{46VqIB0<QB7GW<Ww_*zJ7
z96~4qS?-8ZX;)DYHC1BB2nA48-V=m+`LmZ%_928Cy?Ju{o9&Z{F08%UPaGrX_F^?(
znu<EuB`U`SBz})**e0P}w00G0FwCbRM+~kndL4a-$WdZg05ASZ{`ZQg@Cu3)u-0-#
zE9*72XoKD9HmzBSdM`Wy@|X(rH2W7mpw`n|N|7ndSC{hZ$Ubi;sFnGqD-5g1R4@A<
zk=zkKONT0u_EMKzphH)XD&fS_w--MPSET^bm=KBa_JDBiEW_B-TGQF<^Pd)&k=5>8
z*W=!4;e`6GLH1>DhpN^BO`C+kmGQQ3%FgVQx9aIzs8B!i$0y*jkgB*ya0$^|OcKWS
zO3)O#>FSX`Lp-KhoB)vFsP0z?Mnn!HGhN8O{t<alD-qMZTXXk~<R;#6w&JmYSZ{-N
zBtYm6d_Vc^M_#-`TB7T<y-lCst$NZ|TLPt+?-zq|dWRdhg-a^H2p$A<QOh|fF_+40
z*H3Qdxbv#MB)*#kC<=t{|M@XVm9ujVqOk{pmd<I)#=C)JTFa|5tp2jZj#kLHFvKXu
zzl}MzKZ6fRG1wghH;*3^oSDh+JHo<%weCuDbWBmCoOGvkr_kfLcyc;JFY*W^^S`xl
z9nbQ<2hibX69GVfoWUHcrAKvSjTY{QXOgwtxMkLx?OmhTrrL&u8>Qv_Tvkj#zAlSw
z%K^2YBaWH4Ac0?lD=FKJjf6ekU3gK8C?Yzf^kn{M)xB2#^@Dh|!=u=mnv|WCjK_?m
zBPmz<zm+i7GGb`}tICi#`7!&5$}l>Q?vCg;b^t|<BU<gAkR#6jGa2$^T}!9-4!($L
z?+$V-@jPXKSFR(E;K3Jvk0RXaI5*J!k@JZ>7E)N=;CJKLyqk#^GHjlNxQTE<J!0?z
z)FS3wV-k2%EOcyl0jcy6@=7SvptUH$LRaFeZ3#(m5f58$pauSJpm%if$Z<rT6^h}r
zJP@viyE7&mc5?s;b~FUSg~xD;Yj8ryJ1C1cl)A3jXf|abP2mqGC9!iqEiF+hVHs>J
zn2}DYTAI*&FizIIRVA;?*me4T$f{kBrqSns`Q^>1^jo70GY|25;WNA{<Af^X(dU|P
za%90(V%_9f5S`BqNqqVs2Tp6*b}+PLZa^5~mLT_Is)u7O4kxjS++$L(h87J49ox#E
ztTq|T37YN#NxukLtA+b>imy=^k?>Kay*&nUZ}f^o;sm(>+1-*fdh_Sb#ILeAX#G8v
z+v}T~9Izp$oiA#TVcZOu$d%PkWP)kXX}W}=@Y5_yhp|>JHOAZ8Vl!FnVTyI58Gw#`
zQ!FOb6-b}S>sBK5eyRg@lvr2*apKv`)tiY!(YL_kImZmv#_IS%`nbuj2}JKNe8YKO
z+Ix2Y03;}1bYN5q{iB*?^Gy34dcu;pM7e4wtF-Yzb+eA2GqoewF;<i*n|1?b?6*{b
ztuEu*#AT^5z4EyYeUU;LlT9XU0JZlkPKv+sJgG*ngBkYJ5lCE?wF!F~TmI4tTtmx!
z-?)BCKma^oZTTx2BtVD~7a!C%1|6ZTBgcX0t^;((dVTxwG>wQF_vdjmx(dqel03IG
zHc)ASIfp><^S8RLyBd<R8MY>OYj3vo|Jihg@!D$n|&{A0c}leWDYkp6N7+8)zW
zwqf)s*YW=#nusqw`R};Zs9K7$zR*UnAbegdBp3#F7YJ$UP3zM=k!e}SC%P=@F%_?<
zsutE6MvL#RNZ2|Pfubl!pM+jma8OrVXOXq$b$A1doKs4zt2JnT3`dpNbusOX<Ih*d
z;v5LHqEWPRrm@hT3WR(>s!xbokB^8<X_&}}y~a2$JfuigIYH=OeF_yR2P(&wdZ=;B
z0^?5P55jUj+{YD$`<l9FhxSN4j6RGWN8>EE;KDwyEYf)WU0COU`!;sN65^0~;32yz
zkSs$673Y~`g+rIVjn0;`p~DKgh9&;_$e2HB&$d+V+s)(_%QOIn)!RY`J@@)HD$tKs
znVyizi^QRMqYQGwU~pwt_)5`BveIQ6JAn$kFp%C?24#iF*p(x*D2<U4@Op9+fMDQT
z^&<ZYf@ODLKJVu^GW1P|<T(CzYA8qwz?JMh&3ivUgn?)1>?O1lOtzbu+XzEa-HD3Y
z?StqM`0`D4#shAceNzs9doXaX=nG~o3ZQQCO2cu#minYnnmKdhmB?{SIea!M#0sXM
z)KS4IAT6A!DCv}(9u4`|k34zu;djZ+0$Ua0op4V?M^2J-U7GC!-Y3wmSHgk`@Yrwh
zGNppCcDVb-i|0~dzEV2jJJqmk!)y%9DnckTaeDjZeFa)q4`^1++#8Rv6y+!U1TV^-
z&sSUN6>qqP4yC&1s%~Z9G!ZKZF*H?T2!1~p7{?09Fo|b|Tu2-AiZYYpknlHWmXArh
znDDY@=40z%wc8wBanUy^l_tQ+bn}#xAKe5I2*_VDUsye+6dGN?EA{MhIM5EzlbX`v
z+Eb7sJ^~t(aY-giqpPADKVtukP!g0Eg>$F$F-x!Fjn>pc`4Q*`4Bph)M`r$%xjVjZ
z!mYJCqT1g6y|s_HJ_go~+cQR8`&WZ~{L0Od*X%QJX+!0l=ot<pk^l3h6|fZsDg+B?
zMGX~Zu-;l;w<X1oZoT70M8bMIrLj4b{n7M43V{>gCtc4CLkM>|3~p11q{MoPeNXa#
zB)rvCo!u{5@j)j4K6bChzm=r<V^;ymt{Y-^uAbd}Qz$1|pH{sOQL)<3De<Zou-jFW
z5{79%i|57jw8S|ls7BgW%enS%tq&ndo)Q)xaGu9fV66cKt3tO*%O)`oIw5^t56*MI
zQu8H<Z^{ovWUgiidCu>vEQm~96&6Cv4Xs~73`TQnoincCV$@3wZs+5c8<D?O!>p}>
zck@heF7=W}#zWX&P@8Qje(xG`%cAH=01G7^**Pu0vZ&m|lpIjxu4I4u4*l0D7o2)e
z&C$r1lrO&EFx_b>EoP_VoA9i(Momo@P>+1XpPO!;c&o#2g)3&L;*f0K_<b=vF=hnr
zXY{4^&RmYwsrW1yPtW4F#xv>m6^0y%?2uNlxi<XV&b`VwaZBolQ!a={v2Sd}?_6WN
zL}JjD@zsmAB$FvjF`75dY!CRI1_$(YVbH2rIcNp8TgLt<%Cac;+1!;nJ=v&U0HY4e
z(cX(+DNps)ufNT44rhm&g)#PM5(mtiUt^db71z{2^L`@)<5N0cs9yLNN9`70PF(e7
zKJ514I`QKBNUDXA<qVfI{JP*VBm^Fj!3~V%Og35*3pMX0<-qysEjHHGE|jVDeMnrE
zQgNDs{r#g%exCJ7Ln(TtHnSjD0AGh0C3IxI!1+W_n6*2dvRqQM5?{3nL~3n%b>TYy
z>eWiaYv<;>N?poKaT~loY_sA(DbUyU(4KIM`Xb-n(({o!^5cDk+(Ih{r+mFTHv+c0
zV;d=7zhx5sUDhM*e{)_`Ot=fwmZr7;fq6Zg1c3`8zUl7MGecI^c+C>cQWgK$F_S{9
z9;PFxhpE=yXHAR%ubXvgf|&27{fU!8UA??XbZ~M<sm~rv+Kn<QJ>XRHUi^dcTKU?}
zQkVBjgiI3lDO=32`QOm+)zMhOUfF$zUSH+~TZ{;*dXnhR4^6-RZfJOrm(XDvJ&Y3G
zxB&*WdB?qPBJs}$Uyn(*>C7aI?G+dd1}mFu%2P>d+!JW~M=Gj2{FGoQ%tGi&0RHXk
zY}+7Ol%3ofuAaT=uj_(M48h4w+<oat#Yfhm7Tb%-Oj+S?k?>;A<#>I<rB=(emA@7Y
z&G>%Z3cu7)f*HyfU}9x}jwTN}gS;j2qyS#IWTVT%n6<a;Ze!`z%rRE2jX5jzDURFg
zw|}D}ySSTp^UBr&8E9)6XfGZE2C^CY8)JbwFYXrAIz>yh<jTR~TYws06L!JU!p`hG
zVwJB(aCYFj&cm|wfmOQ{<9Ih*cMWcz*^<xAOdatCZIwYO{I@Eb+o$RvX;bAS0=#h!
zmJ~l^SQ{yo;R0}99>Bi|N5P;T(~0^W-Qcmylp}xjAa-7%_)WcvXhIREJKCkpdJ{$X
zs_6V)!n2F&P-KQpzUkoXOS{U7pOLj3SFJmp1vw8Diy1d1uMM4vt$b{fYEhGg{4T}E
zt#BD#c+zY+y6MQOl2zj*;q@ki2gLEjk_G+cu2=?tQzWFWc5C5(e<yyVmy<CF_o9kj
z4Fg$}3a;sS61+D`vw9Ane}UpY@);(9Xm7#3nP|A%9OwNz*D)mbyZfaNUy_p6t5=L~
zm#P-r{?#Il#VjU5EO57jDoQ1nUON7WEln~||A-4&kXqA(g8V7OD|`A6Jn(;!_7<>h
zcI&oo!_3UgObs(LGcz;e7f!>>+%PjYoQ63~gAFrN!*KJjy*3@4D_!a8lWbYCCEKz!
z=QGB5$J|HZpPL~z;nNtaY2=`&V?|TgJn=i%qd#Qrq|UEYqDH-T6?hr^KHHgeo7-4M
zf<FFaVr~QY82$Ph_0uo&RVIJBz>}pxmFk9{aH2IvPD6{iYgg4LqGt6hFYK^z2JolR
z=UwG6+hmk?xPAopXNi4rcsbkS{Lc2&S~y902iP$M9jyoUB=_tA9<;5bMLDZbCbn)j
zdec-**<qjmh&oKTvXbN_s3IsD+}OCN7Uv0U)6-X^Q^5aDzo2)TC~*a11hbeqFF0WZ
zCNmY%wcQ#a8^i+2q9sU*J^T)bg=@?s4|F*)<jrz0whiR6sPCm=n?gG9fo1FMG8~Q_
zkWcEC<I$ciY|UgpGWpm!kU2qsdJpm{*w7_8Zy2y&slXa1m4p(ASQPY6suwXY)2~R4
zdR3D4O<Yk~^=X_3F$mc`k3=u>>>wmc*buK*F)Hy&c;{zG8xfWPGhPfJHgvsNfEo9a
zHM#MsR_ScycDuP9aJTiaU8t$LqcBSVIN$uYf5Is1yH{4gno9e;l#<|@&siDbZru6u
z;=Tu~olBwEe*+ur1-zGkU)_>zxe=_|GGcfAOwCy^lbI-Q6nK|{FVHkgrfS|5S~5s@
z?t~UF+zv>{Ke|EtZEu3l@%$bi?YLeodTvqi3w|B-0tl;!_N9^2?^B_4r1|g$7D_B7
z{ZVv3keQ16vq2sp!O(U3=Wf<YzQ}k;%LX|d=~g#DfEp%-sx*L|%yc|^I{4HWKXh_B
zg%7nlgwG!}^5PrF6upo7vS?c-3@?u@VmgSC`!}H3r)m<`-(;+i!o~dA396D^-C=o+
z9dlE|XM$@WI#6hs$K1h%B7m|7)}Yx_IK0yIk3FwXCicIw@X$A0$276UFUp^>$yf{;
zr6#E*ukvYE<_NuGjsdB9{3DLk&RL9F!KU4I|0FC>oars*sN%o`zsvtDQm)Kv|AyCA
z5O$W%2id)=y@&}(Ssg*k!B3n6hu45=r$mgv=kiMovuWE>xPRcW(8LS4Q9XJx&*5na
zRn?JlPYT0QAy62V(ynaESiFLhiJL}pN_)`%@TWnDPyw#gtx|gT6T<EBvN4}(9vSY2
zs?_!?6AD~13xZ*9h_bK2@lr6_)xJ{GU6m>;X-J7{Z(k~D1U7<7#(|E*<|KfaLm;~1
zQY@gkej|5MQswfZFDwNT1l~fjK!2elMoE>^2+4!kz@mtF!gGY>7nXkEfQZ9~_=V0z
z0%ilNx5qAXAcn5sawJCI2*{po-biCUV}AD4*uY!QN|v+WzZZ$d25Ew6gMy7=Sr`$k
znh^^&=zl03>~>C$yfX+l#E+N|oyi6(<4t4?QWyWHNDh3e<YTaGYPHE>IJ8Tj@9|cQ
zYvRN>58ARI>>$Db6Ju6CP@o1%gBB~>Jkqm8jLurhX?>bElV~>Ac;4j;FZGw$>+`d`
za1uwvAU~aR{m}MrWMGKSC`01Dg-{+aeX#Gs#xQh`#|wg51v}KV{E%oJHpFjg1LJE!
z(56bT@IbV~dXUsiZ#4Z{*dkp3`2hD`pKuN&v?Ju&oOPjgMO)RnbYOgqbFE?L>vl+#
zlIXUyOr8wvGt<=jovTQs96b3yvj)Rq2n+&0{*lt@rwa(Ry+`R9OU<ut)hG$=>_YP-
z<Y_QIcL}Q7-&C?D+*HEzytoq@SLzJI95){U^=f>cF*C)2Q;Pq)lK!vw3PDEN{}R&w
z$n3e5BLQ=I0bU5F5BH)L7FAh`R_iFjtH_YQSNARK)LXcy<*XP4;w1BlIy1gXxmcOR
zu++5vr;<MF4hlPMU87e43%mxRX<+k;hRGTb!)iT0UQP)-s+C7BdL$s-_qrI#(<nF4
z;B4Zd<bBPPLr7X!b2$H5V-$Lkt`&S%5DE8N=yPcLqR0UWW%syBAQ8Xm+K2dW01k5c
zhb~W1>OT<{->Wz{$v+^SCy0=LHXwfbK5C2jlm87kzpwMvRB_BG0RDf4SR7;<Tv;%C
zDCSLNgXz^JW2&^x@QyIao-_vUB)$10Fariks|#UT>7NL7_uOB2hk3j=eed9C;UfsU
z-;w&VExZbWKyZj+FG2!XDZK<Rh+;4@SXilJ1%39Urn^rIm68x3DRldng-Vu^7FXVE
zCRbXtgB>NSq~~H7v@^fG1^csv`mf;JP7+fUv=R@sbaCAdIDwtC385j8X{f%P!(oHk
zUIQymkren%1QmDU)XNI^>AkP`ZnK(o+6ghPG4TD<puwA4_V`V72wQLP+6_EH!24^S
zLw$YWEPvWT=oeBhZ`E0^Xp8@+SwzhM2M#XGqEO{(xf|@uJC9iTMpw*k&JU$0(&uE$
zq<xKLAI|i2JUg+^{0xo8XMW}?uRyLq4uv*3Hs5}h^7AUCZjPiuB|NN%b}0oXC{h7x
zVtU(imFFA^GQ&oW?jbUZp5$WwiOoRRK-;t@<K~$<`hf<N+E}4PPS8Vr8$ZQp+xXc1
z#qPktH%97XJO)Oaxg@*sb_d(dE!UqqIZk(>JJ_7(QWT(*sN+m=6ghBJ<orUVVJGoK
zc&_)IooPyIeb}y_bU2CqmyAf4R~8V;K7=QRCVpraGtqg@ddl(G$FkNa!;xl>lWSke
zfLU2%qKEd|gqUOd*vna|o|T0=b|3;>zV&FCz#%I{GYA%yt*2q9Or2NA-fCb!295XD
zb9;_EDdPuA2n#qpFF1X~b6*vr9}7H;=fV2u9m~@>zq~~32r7=M42Zj<%mJ`R<umX)
zbC=)CTQ<4yCa_0ALK(f!=-&p{f(yio3GnSyFEA{({g)W;{ErxaHSsCNN7Bjaazwo0
z(d7QB1MQyL$oeOs6NdsycFnk=eTKT()-xn$>%`Sk%kjBF{zpKk8r&}zPs?XAuF(4C
zmkF;_?l}nX-*P-Z@n=G3&&OdoF5$n`c*LN9O9a%U`2fdhnD1*Qzh7Xjj0KrTzuOo?
zMZ4!<!u@)=T3pLmL9z1cR-ul@xL{u&x#Hqj{|_&H%PLE!FY5=JynP6FaA{gG&1j?g
z%3_Ek_Yy{iO-rIYT=+FmPn5Egaj>F(I-+JTY&_qIYy0KS!bf$NeLk*wxwpMAxqWtV
z%IMo?_51^$Hv=gjQ&(utZ+Ay(4_~+E3@)+sB&k}Axm2z*Pu+Rq{zu|U_0*#U#0)8q
za-%`9M(T6?m@Y+)-q(Ra{Yp@_E2N213HZxgV+*yQRj@)3@dPlS2J>vTd_-2!AV%gW
zU3Zz;o4L3@d`w-XznW5^f2pqW>KxB5drxvp?k~5#mGcj0R;^isnTR<{G`q_2VTU81
z5}oSd6!{U*OvXcuo504(GI-wRsSB(NcKYp_E+sTMOoQ4Pr`uRH*fsxQButn(8?5NX
z1u^&Y<&z61(&P_7jPh8;1TJ(>T%A~R)*b41h%tXg)88$^!pcp1?&#iLsw4aFGMP#I
zp0v(PU{ycr$6o?Yf6mY{naPbk-|BOshz!277$aV}fY<R1cjj`&UiQq5)BObR<4<j;
z3x{kc!nmlD=M}eF<^E+t&3=CNJv5*rtTZ<S<;Z9rGv2if@Mi>N?{o@YU#e?<f>54~
z`m(yMyyZsKfASYmAe1OUwce+D!wDTOm^uaVw^KE$d7>Dtpa#|Y0P63xUCe<ej&HO2
z^@;Lh%YEOjLN7o8&!EpE>82pol(Q-ig72bUEw43U5Cmndhz+F6|4^_abuM-DtKv5x
zTYj77QH_`00+4kgd1XrWBd9qyZ{o`(niGvGw#n6>rQKA|SRma0#0$NlePfO%#SdoI
zDpS)ctBF;O5fc}Kce}EmVjRbFVC6a5ik^26g)+3ci@tPV4CK`ITI_;<2C2|6$hXF=
zM#9Hbw)5{hZQg~*e)B97f=~N_0x|)TM&`mQ61+&H0<?O1tFM3if$=kAXo`X&v3Ni{
ziW0?Ct`W+^NLEqqH(pNKvIOcCq_lTP9|C9<0thmsP$hH)X&>AbVfae8OkdE*?W@Um
zY@P8hK}qmKCQ3;lXo!^fa*qatL^h@@$lxt3P7~Li`+X#^N_evTS4w;!UPU2T76te`
zKNRXk)&bilwL`?)^1%X#NwB9uD2zMvi3?Nz7<s;T(#_q{$bHHC;RNPOQ4umO)M#Cp
z>+DiaDRIT9j)Gggua}l+sHr3#Qpn0RDZ<03aHbXu`UMxlQeAnLI`*(-e=#=?Vum&#
zF|>z&>J+LW!x4^n_Vwkbja;mu7KKKQK$(4}?0|pA?25;py+};YP1lq%2DFF<=|RE_
z>C<If+FNhP$N%vb_)$*OMcX7=S+caf81~I8NHtKMZJbpFhTEEaEAv9_M4SGJ+p$WT
zzTVv?C)D{D3qCs-rQg2&MI7KDeQlyzUeEgs0*8)evwk)nI}H!d+z%&8*5G;rpJn|P
z|08twTDj{`-iO~JW!Cfp4fY^%Qic$m=TaQ6duhm#7oJmeH}-m}{e1-V6lKN1q%{7g
z3{hL+hO9EB!}dRALrL~AXHfsj+mR=`W$)4M!R9Fjir}<Pmg9}m`6gsbi^Q41{dHhu
zTKR2gv5o0aZwIX8X`kKx;(c~zIuZ^TI7++tFy}OVmM2H}RxmtkmtOUE`js?AE;(WP
zn~XIPat%@~Q@K48i~|J<d62Ah#waZd;_&jOaTCIGJu4>Ktg@g}c?5^Wz1u;ojvkS1
zU;W__jnS^ZNAs9U$vVoT?ku=^$rv5}p7Ry@*?jtCpgKUECY$>$$13=b=(@P<^zOx4
z<gk>kSdw47eViw8LbG`e=f%x1dN!+kPI8VPR)f4WYWO1tDHO<lq4Jho)BfkVd$~7l
zdy4)FUoq#Cmfh;(wiK##qHI4DK%0+;8ww-;Qn`g!W6{-ZX5N5XHgcn`3mU1Gz{T4s
z_CIfr=X?PI6j(8F<;~_8G|Y77e#7sgB)T26RLeUQnmdRmGsvo&A!x%o{}DJ56gj3z
znMmQhmYhQAA~@<$<oIFgK5#E(p3}~BY9Zp53LB`%c6Fj};T`%bvq=+Gfm@#Q2K<UR
zHQx|1+Jg9Y4Gm&i8pJ4$WD?ZN`=PZid%k~-5aI^-e2OD)FZ>4P4~7iWtJW`$ZvWDn
zIxU?yV_Hmo#4EJ043UWcEj9^mPqn^QacUkx+H%%g6@l70a)DQ9)v|^3qHUIfy@(HP
z+`_A`4T-l#Z>wWCBJ4VUPHTecezkWg;}PJcX{^s1PO+paDlGgZv#3%-nxZej3q)P_
zzYTgICn(t$pU~mOV;X5sWLz4{T4d6jY@*!t=h<7FO6^TFSR0$Ili+9rQ)XynH%g_?
zKkhW*4I(2`WNRi+kacJ6jFczopL=rBF+*({^exbDQBdtHuMo4pB_KkS6>KrgTSBKw
zFQpi5D!L#|u5+PwX^x4&PmM>406)(2bdc7-t6fag7{$JwjxpqGjEh7Nxy-a7dZ|Xz
zJ--LKo8hdYQPKpHo_20Cp@rnj<maCluIov6Zx?`EoOH9aX)&dUUsyIe-Wcb4lxWD9
zeb;xRGdQJ@fpS0yh^Wq+$*mE{Hd0mBE91>yk?taJEvGtA((Dxtk333201(iK#B<y&
zsuk)L4pB)-B0}l^Za`e=9^dQwTYD{BcH@|H>LphaL?XOHNqT%%z!HIcj@ieDuqFs;
zJ95gFSg=V|ztu4}crb-Ua%eU0F|7Wtp@glez^-fUidoWeyhwB*Wg2UmSp@&+H=TdS
zu<&sJ$dj4G|DtN4-1om2Hh<tt=FC00exyjeB|exmP7}wn#8ok@f0KGP$$5a64z-J=
z+ogPCH1O|Or!0n<T%t|Qnyr-p3!t^n93;vk58h)3<#AQbj!8msxpPc>)f`)NE$i_4
z8u8f{Gs;illhP;R`$b>KbA#4w4hs0Xwy59q<r?%F7LzHU+l-{`&NCBzy1^$oZ=n%d
z*v<Ot^*+m>ZgG4E1^xYNW%I;0FyHmRwS`ri-Ba9W)Z`Z?0?=c|?i_Yh<WM$8SVRHG
zPPTE&7v)4y^wW#n=FX#I(yp!}AeXW~WIR*yz;lMTx4X{Rxn?O&mZA!8aXNrzbG}Sg
z5ILL-R%Y`=zMJq)2rZU+6QiGfQ3UOb82wQ}7`|&EL@^k$BaLlD9!ERN56|-d8SmOk
z+GLzupQtgN{GDxakw&hGi2<68+OED*ugdZ&3G<mrKg&D8<f=eF{mY%rzDKMeiS{`d
zA^9`RqJ&T#N}k8!8UpxNJWD>5ipEaB;fAVFTz2P4t<_F%nSCRbZEJe}V`ig`xh&Sq
z{v(x;l?`GOHMl)HrJF*?qkL&tJI2#vQ_direzo<hhza@jgu)c?@!UKlz8)J^jY-v&
zyetyfneTe(>g?sIdBxfskyDB*ktKkn-jshjgK;}kjiEscU`5Sae<r<nnEy~2ANALN
zm$PtYtx<|erKoVD44bweIETz|CJH}Ss#H(W0jUs@qpd$CBKuRyR@^7clXS)j-l__|
zHyB{lr{0U^Rb3-^@4I&ru3ERrBC=`(si9jUF+nE4qLG1$l_RaKHBUU26tq}2X(JzN
zzMqpwjxJ{llnwu`EO8M>h04JoY1lPQLQ+_XW)YK>p=&*2)k&^&DQS^|f-;QMmJ_J9
zkCJ!@4=t005{-Vkt0nTiCc}nn48YvP88zu{-^VX^>q^(3zd)ws1Dm5dF#TOdaeLVX
z@-h#&{*5uIv0YP*a25rd?;*uHud4DmgulA*SC&c#sB?kVu5^MU#ZP+xF;e%lY%{X=
z4r3Y&mhR$6XYfwz?%m<tfpF$6$X3okzL+Cfoz8K>3}(p6bys@dR@)%4M8}%vYPc!n
zE!uabO-g|vGqEH)^wK809kZiEteb=Swfz2Ns4)JQ1_EnAT+xPVME8Y&vQ6VdcUUj;
z0J0}=`u8teFmoO@#lcH7Z3MRJ-AqslC?Pwzl*#e?HxVjG)gFBb>Xf-r%~CYnuj;@R
zCn!Y4{hO7L{6meHDqR}hPcrf>b+O&ccxtgG(>B|FtwDZm#rNIbA&8iV&;yvou6xAW
zsnB3G>#t#^mwgVlX~rPkHw=@9cu%bPqF6EjwF$-M3m(<ujmA9IZa#h)z2GQh%?M{K
z_sU@$I5yonfwl=i!bZXtHY&&H?~$0D@OcevQBF%W=YwHDL=aC4vo4Bc4xB?vbs-|`
zJn!9yEpsNXhT$zW!b7bfuQ*YC<tyA}_k4!8gJ!dszlHq%wR;EZ4Io_oT_X&I0}5ar
zuotA|8Q^XrX>J{;!`GHz8C-96Mx7dS#@U#cbTC2jt6>I(&$7q}RO??CTZVPLp#6
z@zlq^z$>dqt5nt3F{Xbu6aAK8H+AbDxBp$jQGK|TVcKe*Qd!q&W9E->Ur$>eq<>VB
zDOT{!q8u>=6(*Ub3%x29%|={QHV1$|_$irB`TV>K-+jhWXLrMuF$(!dnEIuhy`)Q`
z(YD5~g?1yz)9CQwn4^{*=`|M3&s&I1wvr{O;g>kC2!~-Qq9$ZxgwG#aip;7_h7#XP
zk)3Xzx`G5RwDU<y#vabzGw+%S`)>D@EGSZ;G)H)<sLb|XZIOX*c@$@n@b18olkf0^
zmOR&WbzyRdGTT}yg=GB|wIY@V1<i=i@pp4Za;o#JpD`;Dc+}*DD-pp=(DQ9er?7J6
zF*s(%6l&jxc;q6(&Io|}RtX1a`~bD-^u0zWe8S}`f0y=NL6d4wYM}y3foLS8Sg*n<
z<M3BEIk`2To!;tng#nUK20q}c!vZ|-@QX0W2zlF;)@F@qG={`fm9KhNok3kn9Xnj&
z5T8hJcI28UnEFMYz9=WEi?jbC8A90T<+?8=Id$E;GlM6F3&}(0IUsI@8j4YmwE6w#
zJDoEG-b-C5&0PJ5O!F%SIpOweqJBIZvs(hQ2r`e`7j84=jw7r;pS5;Y$F1J!6hu#(
z1m3k5B7Vipq4mz$p^d=jy+xSvEp0gR1~_LpGAf$Ll?l{do5$tEzI(i!8?Eu{dLmgf
z26HJg+&!jDnTZ^U97+UBnAH=V6d$&#kVs#K$Q}|>GA3t66H9(nV~rP<KP~K&Isv@k
zt?;z1szee*X7iU7FCZ6tX`uZ#uORhs3v;czmotO%b#`cdmP?uIiZ6ej$%3QELcp>h
zLzrG(<IjHdQrED4UkEG%gt~QFjQce&hq}God_$+#pdq7u5F8ooYWU<e7A+hG03!0x
zC^%J%pd$~tK94sIZotQ3n;5QPt7$&LMsmkp2+7X9pkI>15SX0sJq~1ilnLY$K#t%K
zlvt>+cw68oqC8y4l!;Yjkllsf_0zG$8!u70Ch+-kZcH@o_7I-wH+%M{!?`9qQFN*H
zM{&x&3wfpOCS87Zwv(&n3MaGt_VsR2N>zXkc6A@qVdp7ED3jRGW*&z8J|%%(MCG}c
z{;xE*0K^OFCSaV93ALodG09o)<v_#T?L7QgUGdMCetdRmGuBIE$esxmK}X`dP%{5z
zKf!%3W^TtM$=iiOhzQVAkg<y=7`n)UWzD>H!q9|(^n%HrNX<pPH)VGkXCk)<>`v*(
z@pb!ys(Ly%9wLURuCC1AX7pdXGP(D}+#i>@2(RvkRRMWNC%r`i?$_Hp-p6*Vn*sq9
z7dT}{02Skf@&Gav+0?38Cbei`OdLK&LO7QsY+eBQ8&Wuj)1P>9QVM8k7%+9qf&LRH
zsAl{4cV!wwTD4G7!TJ(kewy2R$r$x75J#TY<xM?p$Fov`Y)hjphVyaed@4arKbFWn
zU|>}#WdZy@U#`#PRmS{Y@+d?X-7D;)_%Tm)2NHC!UC5@37A|212R>(~Qpe7i<0MnY
zc#SVtH>R@q`+ltF>v6`jN&S9N4293E&K(IzN^NrTIy+H&fg>;!%Ov9=sn>MLs_A+r
zD!P`sDTC7YMb6M&nr6a6FU@}$*lAGR)8YyYbO(^XYl)KYe_Wc{ANqtN7fDI0`3-RE
zFz9#BAXN9%-94WzHM?oy)8kvft+;&OZ3?R(#g`I%Fk~`S&WVb9&NVM{r}p&8ai%7X
zGr@pO{Tvh0LqFx@k3M1FfU@57-=)&Z_<@l9T{%Z;<4=bvE~-c)bbsl^=+|TdaP*nR
z0CJ%BwbyWT$TiGRZ;09eJNBRv*p&Qc@?8bqYMaT13mxgk^NRs~jKBDCNxtxc<X*sD
zfC5fkTghfHw@^$1*`A4ToG$t}ru;a*jHKSxWwYZT-NVm<-NWPKT6wr1%~s?2ZE1f^
z3swbd^%gx=a(#$`Yp*(Ho$t8Z7n+)F8dQMlw35c&8A4OPtB<Qj!cq#y3%5aN=7~q9
z%jg3SiwU1<x9}?_?fd=<7(WUoV-tf2xG>3K0-OP!YToathvv)$wKcXwPo~AIDfMN|
ze-At2aUBB(tS1e7dc~ziMo-$seK)?BdE*rlYsQZ=^q8>Ng)&eSBpi}$SP!&|p|1ml
z)l~MbxEwJ#gXFX?>!wLFNmPqpT}AJCQzE<&NXhO~8j_egp$@)&NnFF46TvrT1}8Dl
zr`RlGTwGUgUHXBHPj)%>prUNEc+Wva*2u4v{JY25hLH-cC)xdX?@>V_`yvk7>QGBK
zF-g-^NX@2TeZB+3QVZmdkP!2i-s^NADSI-EG`st(UEk)uLjJ{Y(Ra1J;yI+(`818n
z<o4x8{YELMei9hqZJV^!sjGoKGK!uEtN=%{GHo}GyjtVkEaD(!i1j1x9s8ru<?0c3
z!@t5+WoD(t=v~|zP-aW(xWRtj;06G<Lh7Cc12LeVjqjD(AwlD6BK14)5JJGsx-+mk
zukNIP9TiA0kZ_~iLq6d*2D-+dUEQ^y(V-JljNvII*3;o6h|<0Y@l`|JBn}e&OzFVW
z2C1^fhVf-SkZ76;$!WpOrm9q3zUfSW(}v-FwHqN8qA+2GRh>+<FRIM{vYQ}NXo30$
zj2wl{wR>jDOqw>I^!Rw`c?sCoz9HuiXuMW`uEvpb7f{&^>mE(1lz}`uUOwE1@#B2K
z{F4*u+8^@$BWA#zcj>8Ft@g~;WKMv~bK49`yoUJuS6(u!%wn)@E1kXg#C??rZxD<8
z7nmz+q{`C?*1i=I;$+LclPdFUz)F`Me{9D+F;Z^p>4gl@+&_bRrUpP8?6d?o+Z)vk
z=z(KmvBJK7v=qvNJrER7wFf6W-jp68-EgiCfYJ|p<CG8I6|WMX!->rb=#;taTocx%
zS)r~Pnk7iLI%82Mc;sE36-|NI`Nk#rYI(rB*gbunw^F~m9bAU3+vr9<CIuHf4@&JG
zkA1M{8>h#c!d%5R^$K`cBq7?{j^>ru@rsK2k6kB%3QriZLA2ohG%0F`(F(3}GP!2@
z3_!6%Z}byCj<|PKHIHg`JPF=pHvZ&Z#NP^-Y|L;RmlYpAt@wfvUS84HJjw9<&IHj@
zE%kF?{)uSx9kk)&*u8AU*3hGN)cWBjn7z(-e9F6PiC5F1lM~p*DpH4(hUb1<CxwqB
zcaXbxBVR{Ei)UUYNpVtrr!e&6-yAO1K7rVm{rT!L6~4WY`GyinnxRRbXu8R*U~5gY
zmiYFO?AOZtp7^FYnf4gJZ%3g!^_x}O#3rqOLpRvF|J79hYMj0he&of<+M7c`5>wM^
z8WVM@j#LU=z5<x%!FYch%P(mPs2G_0@SU7kr;{PFxII2@;=#tPBQgwk;g59D*VCpm
zrznkV=b^Q@?O`1sR5X8AtdBMWEk>=>P^&4|F)h=fnUoT(iPFJnsq{KxI6OgCt7%6^
zJ_--Y!sM{GV?UR};X2X%)h0dPM#J%wgE3TP+ww8q_78B)^a^^i5qfGMgws;@Sez?%
z&g(GU*mUx9VIVVCp-})+^PQ5xg1+7u{5KMk*j~3#=tI2cub3cccrox<g=QLZL=&7D
zIj;p4HXRtY>?jRT9hetG#hq$!z8WDiV-$OKR1;J3AG?{}tn(8gDY1*Fm-4?DN}T;Y
zeqZE>%T5AXAlM9(>rwi3=HvU7zgCx0>Zx1gK;i%Dk7Fi<Iou$U3OF9=Qsz9Q){Ff~
ztuM^X`+_Za<$=AMC{wKHDEVOIQ?TN?WueF_LW7+@^hlm(5<aCK^u>ID-DG}pZof)S
zYZAL=PV8z!#+E(wb1_wfF!{U*3x+#icoYh{rvX;Wtn!DWX?R(2%Oj0?mT}9MO_^-K
z+(p1jd6qe7G;C7hbuhN6iR){&Rr==NP)srC!iF<cT!d)$GzZ^aSxEI;BPjCV)kjL6
zAS}i0wiMK;ppwam>r{M4tt*X-k8hPYUOIYOR161Y_C=I=7~q;!BCHP<LJO(#4QVnb
z+W^P;hszySjkMzVO!WH$`Sw3w7Agkn31nKLROpB?zDYxTeKRdya|=iJvb`8ZV@fjS
zh`R0ztO~$V&UiRS0)A%zULjcz7otA@uiWYTQ7oK90am4$xtN7JG0sgUO`U;B=nsz+
zs9Kc<jAKaSJ86Wy$4S5Ip?||PcRQGw;%TQ;-<Pq?L`3XCZ5rIKZ>7QggNK^ONC3}p
zu+SB7_A}mye)L(}&yqO$?@pR)!Th;_hjf={#udC+enz}@ew={MCc<+1dp4dU+3(I)
z{s?TOQqprem${x!8_(%4mm64NZjtj(mR-IpK!*}fH*|@U)cSKPwA5=!ofDywmb2s%
zZt2MP2hUs2TdxZwJ<}T-(wfX_9dBJqSz?Lo5G_ii_Zm}g@iZF>a}kC-f-}SQi5_tm
z#IYFqKjeOzY(p)h3=}6d!rPK9n^cJY#jQ4SkOhrPf2d^S?|a&=uq3c|9J=5C+IX?#
z0snXK6r-y27sORs@Tar>6}2yM4J~boVg2|*25)V*)+g!&sr7Hf<U76qyF_m}E!Nz`
zrj|;zQ02NuJZrVU!JMVr9PsN&WI7yK`*bS%0|#Y#9yk@Q_G^lnh6TA5v;(MXT>S%j
zgaL^&cS+C~jEgw>xx7)iQ8^OUWZ8WEJ<89kl({{U4fXs`NW5P;!ZkszMJdmT7j(Wh
zIZ^cGKR7e%zc~}@E4$jrxk{6w{pv^M#zMIKBGGH?1q?x1$AYDiQpB8z&xYHtQq9+b
zj~UUmPL3mD(XD|ThhM<_g6!_gHthP3sJ|_D19b1Ynqy$gWkovL))y1*@hS>=uXcal
zBWxHApLk{Zq?U447WY%(&XjFT;F|p<nk&UC)XT+%Nnctza30_^LYf$tjQMp(@S7My
zbo>|9$>=`*3|1;YR(8!Q6GsZ6#Nt3G8IsH4o~u;by>V+>MJ5Bke>@&(n^bbg@_(<)
z4v^=bvA&I>iT5|O<TvQC!LpnSXeERP3eIZTOU4-B;Cw$caG6=kK-rCk2JI4E{dLU?
z{iQuB@rW{IEmkVjgXQ&<3DC!=2WMeKaVWpH9E~v>@6T3Fev^KoxcdoCA~x8$#IY!a
z?F(}r<4w^QQ(X*D7-(VBZYt*6?pLW_CnbXU_RKW(q(TPwItMc$Sy@Pz__3yST1l55
zOx!WkgQXHo=p0gD9NCZCKQNMSRQ0pLMsm(g+bOU3-cIB#b*@k9YD#+&gPJ`JJAKa>
z!+HFQQoq?0%AZ(>NU)(cP(>;CU!a>7BT;`f3aItjJUA`g$u>_S8602e7)NoK{zJMP
zrk|uMWBW<E^dZroq?_7b`2U4;A*((~S2Vf+!vFsu-E$1i{{gz1dYjVD{{mg${~%r8
zsb6Eo!L*(;wS}tW9iQhCC5DUad0xQbqNZqmw5N5!b|m=zx$h-qhZ%rfj&Phaahz-q
zj_qa{OtD(VW7BzuMU8iQAFUn1!RC>|Ci#ZXqeTJ>kkT!fwEG9BjZ3QD6QPQ=Rf?Xv
zXF!oUr9?mhUIBAchTqZ@GK)Z3++%4TQtk^@D-Tq7IhKZm{C*E)%naLSB^F%=D8)P(
zO+uLpVBbcP9C|EQcAirjpRK);dhS~&9-DwelL=izea4}U$2nuZh));_@AN|>qk>8V
zk^+yhPo(Rh-yTBe>>>EjE|c2PBsA$yG~JnB<tVuUp$OPruUzbt0i~r+2oCRL{^jmf
z$oL7tVlDp@f>qo83xZR2|AF9^S$wkh=mDn<5F7;TX*kt-Q!b`|KK@L9=xx;5N0~Qy
zsCwwuWjt3Xu^X3SZ&A4DLg(OX144IfD40bde`7r}!;%&r9smmyv8)K5^$TWFUfQgP
z;^GGvQZLC*WiDpS-=F;DX!VUA7oE<vP&kw&55$p_8+SB2!360Q2Xs`Odh5QKkZ*al
zS-zg?eemZ$5(X!}ow)qi%P%<F_&x(5JOP}Hb^F^)vEU08d;-Lq_<Na1*7(yC##7Co
z7ZD#r^zV9p!1-IYUPCs|KYPS^yZqKv4OT?$)m+gK{-+;5j20E7_(j-Gx@RUdBAl;(
zz|Tw=Az4U6&rEE18H8I>O=LQJigXRI1z+rLeh`WDQO-2zA%FG8L~KOv>eIkgIWMPa
zorQIKt78o#i{LD_xokBvqXW5l6vDs%f-7H^^;dBTh{?)Pw4~(_X!)FVh`F`S?Wazc
z`RR!b-x%S~weSbe0*a^8oQJd4oz3NFH8nk@Q7`V(y!PGq_u=h?x;L9>g+Dx)Zgf1$
z7Vv+4q~5R0-x}#dcy+1mG7x?g5qmw?X0f(m{q0=78po*<>8*A8P^@hpj}<w;&N;?a
z>+f0x@N{|%-c^=41R5b`k+Wt#FoSKHSLAad`wS!s*!ASQ@MYw1jukx})8C2){S7pG
z7&GDIzQ+1HPL>yfJb#Irvch>?nN=lN$w+Y#a)B($Nu!+EHRbnN&8MJnCY=@U$I8Hp
z=v$wiDXO3r;Z{2{<Bgsp+n$Msujwv{_`&z~d34Z47+-mO%DM=3&ZgXU_7?v6IZL=i
zl^QosuHI0e75Xn673y=apvSG}z}5SdElJcw)X?J4ph~uFqrb;hb0eY+{2&DPU|_Z3
z7wsKAwHq%h&hcul47|C)rxo-I3<vxRoNTIDqmraG{MHU%Z^BHX!RUuD?MzKbR;Xgm
zfJ^MyQ&e&!slRRst!-l;2xvZ&-QHA@59is8{vQ_;{JnzSJ{5fnOR_&slnd1{%LT{B
z-Efz2OCMiSv^@8Ef3+$>Q)Z$W<i7-8!GhNf%5xj`SXaRjzUuXfROnQHBO6lCv*qd~
zbtWX#0<|zEM?%x>1(4hIu!K5V^p<eR0r;$;tpRPzrK7!L<)`w=bNd)+7zxmc8V%+>
z*SONGyu5{ZylDwPP-HMlCe*`nFpGpW|9kaHQNEfyGW|;J_%l@X7N`L*F3(mDVq6Y=
z4-5qm`nBTwI6u<;Xb5i|M&hWDXg|>JF7g$-T(Z5tDkFcLoMfNuEmwXnKvMyzfM)bx
zb<L?iD=~$fcUD2eOFvbek#!hzNb(Z$;OZmc@(iRfEoOUY_^k&O1V~x0g-d$`cD!^L
z(I?<qYdn>6)GZ=4yH7`_D_}KZo6KFY`kOT=Z}s&yWbDtfnkCuqjg;!(10*pOS0mED
zl->((ku!>kE+UHD!F03S4bS5F0=@2&UZ%HsGL6W(%D<DW9Od}v61`1T{ydBf{iLuS
zU`&3xPuvcUK_wFFDRRsqzDHf+G*A}$v5f97NVhs;y--Q^3VjUUJ=X1$Y@8QV#EDbW
zF6PMEdb<2<SEpB<rueXEXof_m>)M*0HoA#=CQ1obA?NS%o|PM?UrXk34d~38JUx2}
zA1xMWqLsutQ&R=Z2{j{~Zcc=X3o^V*ZC7mSdZK)t1V-QZoVFT7gZ?NHw8>5hv{-*s
z`6%N3OOgK$MHLP;(w+U6xdk>gi~N<uXrTMcoKTkg?@)1b?h^Yv4BCukv?nE}-ET_$
z^~fK~e4+v;VX;kjVz(iHW(|}mF82I4*5d0pBO@_WywK8?ZYYq{cA<0h*}fhUP?!mQ
z+tYoK3iqg7K}%-C&VOmTTOphRIbSA{1>9*Ig`+3)%?=(UY$0`^gbj_s`Ns-2V~vtd
zPFa{_N_0$l5a!Ta7mG|PwDcsP``5Th9@CTLp=BQ=@lR&u`GJSxRZ3Cf!#2sgZ$r4j
z0#5Up`5{b>lsPJ*Je3Oa5RRs}WhymV+7eu!FWr7jM~rmJDhsQmSF$qeIAx>E;*Nry
zjZqr7&G=Pp^Zr-j)(3%`j6zAYM{rcA#tM-Dd6|0(H?tPWoA6qp`H6P{eqB}4dIo#g
zV2dP!Ayz_r16WFU>P`rm@H3+06vQ^Z;Z+gaG4BsAM%`~#qk|Tkqnjpf|B-E2!6h{n
zZ~ysIq?)tI6Dd>?<-QG&IauTDX3i9E;0kk_-^Y~Dm`r^BTD*NNRHGzd!w@_9Gt1!K
zONb1y`%?pai^Gi}ccZeb#Jg>~`KiiOO*?$^!xSo31i;TKTY#=g*Z8$%KOKM$*RVhF
zD!s*(XM+u6S_wfyGF{{=;!u&ao<T-~Ewgjroo`2vE*C6^rdeE*G-pkDl6H&9y}^D>
z$Pn#m#`(F8^W8ecvK*~DF?!5o=35f=a8t9(jY?5DhxsTGqpg=%LsmmYq)^nMuU|@$
zQry3#1I$?%$tGbnwQkO4^6^58FRd;uON_e)=`^2%n#V#`BeYp<dM^=50ZH7A!A@bU
zp04_pk<JbE<ug;^HP~!~Px}Qsd0j=GfI9^rdyx{XJGPjfF)#-vIm}q?!rN_N*8nI+
zJRTL}co}NeQ*=Xpi_@YEJMGM5zO}fnhcGt-V%UA>vCnqv9xg1WwI9+JsXWuBGUy0A
zf{(&)J9HW*J*9r$u6GMEZ*^*k6qZ=ln!k1AzP@78kc_1`!A<3T55rF8{2umLC$;?(
zUnp_Qei7C?Q&umOvo!IsivuUm)eENKJBG*6H499sRF6Q(ER$Xxrp}@hBqX_*I>R3#
zppff}NW?84@ieFerQ>WniLoTRUfOK~4?Zr_t?>puclU)DCfzzzs3%!+Te6g(<vVy}
zjCL$(UfBEb_I=J1oQvv*-!OQR7GA6Hga-)THhN~?3dHRSAD#kZSP13LbE=tG2l$Ae
z^er(1zppbYBu2WuJB@Kn)b)`*mpZ;5fYpgJsrQHnqjF2vU-vb-#saP%Ytrt)Z!9y7
zQgqSTS|C@ZtdZ?kqP)4r>J4f5m_kZ)T{n<SSF+M|mXtPcDoD-XvO?0$1ZynoSu^J(
zb)IT)RQe^8i?=}p4lee>=4TH6I;IPSn1r2qiCGjoQx`!AoN(=}tBW`6<dcmPkc?%8
zoW<{XL5sh$4W>ghXVPP3GWn}M9zKK<#0*}LL#lBXuiH?iH@i2xAd`x7vdpU&iA3B#
ziozp>r22>giN#T(_mft~hv+gPcphc&DZ-hPNXg~+hkyTSaOAlcGqD;!UpP|k)IUgS
z<~2HsK^k@`<tdN@|83~_3~jjtOmiC@<&ht~KJ&I6@z?S4x_9l<4ZK~`TN@@$n_pgz
zMR7E*mX*n99^WCNyWc22z3Oy;X8wK|K)%1KW3#>J6MNCfvprk#7{GAFRZBt}dDPN6
z_~v6f+nleT;@85Pu9l*ww=_e79tyQdY#3ZGojPFML-NIrEW%@NPed{fke2CHol4(3
z88UMs1TW#3VE$Do3c{YG(?$P@w=d^j1Zx@e5?Y}Nd~vo23%goyLt62N?~l$)?rH|p
zSQy@hyJrK%H_|a<@l<tMr#G{1v>Y(fWf@FlZ93-$FO9cdPuc6|&AgTioYM!|LLl`+
z8-A(5ih=PJFPl3f^_hHtG6ET=wO(Tw7fj=pfVmubkxhJj;oKLhA^2wO&JysLOXoI{
z2muvc4(MA>{J(iTXYLP?D#_hUZiQjj0L#wDC6eD>ix2dxLf`IV?q~yImzuGpMcro?
z%KSV@a04YcaO(tp4NBsd!5JIseTPws#PL#9&Cf!j)1r_=K;Q{rv7M1kcQp$N*tZ^-
z);NAj65|-xNL?9h$w-J{=&)S)uIYBeF{&`o!9-Az?gbRY6P%(BUzxgUJPu>-7+Gfc
zBB9Ybd_DXN-F(20@JJMhhqOVR?=?kr%TwZUUry3-S9YK(c`2^pI+pA|qWu!<l2TP3
zdSxEwCvf<bQgDFc!?L}q^*@vqO89z*hztFr!C12|QV9<zA+*Hs6fk0JYrmfh^pLT3
zF}6hneND16)G|^HmPNUB?nJ9zWy0+r4q%!4m*QElWBQ4Qf<<FkGKxox7#X`@kNp#Q
z_Zo|gn8Tq=rJi$b)kKE9G|)ti@3l2Anx(23^7TFTZtVeBo}uBFVnm4y&8rM*qmW##
z!UBpXt;$FdrwqS7{476SnkUM6Y5dAP5rg?`k!kFND9#^{st^CImIu-}yvB;8&z{p+
zMU=Sde&4zT=4DQ<3dFI;4v0$o;dK`MG6`g4{=uwQ={?grVy<M+86#AAY%fUH0!x7T
z$M(jjJPHGwAxT5!zw(3f82S3RKAm}Bgv>8;Ag|3#bu`tl@X>w+70Jm&)Q4^LbxmjL
zod7HV`)V+7)3+jNZ}1z4AVGIN%h`?#ZTMg2>NW^i?BBWxhS<NOGi<4hyGyix0c7v{
zIBmqY3oJCoN=z$eLBjm-CsR;=y;cFjwQ$#coAs!VaoJrumX?WKz4t|*ow2<=rKB!3
zctWEqA$aWWEs*5VNRzQmd4XP3imGMiY3x~QE#-O>HO!($lQYb-*pGyPzG&qYm_k#v
z`xa^=H4otcW3{g)wf+&)6y-&v!(3+pm3daxsdQz83AgCiCzMrED9-gd^<%)vq=nE2
zBk;p@tH6c_|MBld#th5A*(413K3>#3$<#!2-63q~fp7X5?29^y=0{JvlFuIxRe|_+
z3c{#pPbK2-Z9}4itU6gM-mj^g3S#V&gZKOQaZiLfGs#G11)Z($a-W7yh9k+H4B|{1
ztxRDgKA|CEZ_`<`fmaJgaSo7X0}({pPA5X!DP3$oJv!9SX21GdHOtv4+>d-dCqRL;
zkqY}ksyT4}60`?zNJukp`2}5%T0I&gh<dpN^Yi3bZ!e>0wV}ZrQlc&YBRQ!R+2cYR
zHpjV=8AfhG-*Ep=BaJ)DCLp;-RPE<2Q)V)kf8is)4JV4Z+t41%7%U(uO|To$_V)z}
zeUZAd7egcR1U2ep5M!||?NkR09hI8gugN|vAwfwBzcWE7AIUt_lBqfY1tJ9{G94vx
zAo@_GA!14J$v&m{U~dRXG2kKWDL^5BX6^k-0Wgf`isOqRNau_$iDEamik~=StQ}eJ
z#bJquI+(0J^ezz9<plJZPYTZ&|7A=k=SvaeoY_+bG|>=n97y$dHAkU;*SLug2DOa!
zMy;R3_y7H;g`%nd)iM9_EqVu;H-Uw$M4CVoUK4*>CY`E1$AgI}?JLR3cMk?`@cOSX
zltJb3a0dpxXP=wD_9d0w1SNfSSnLm~O;XyVI*5W5BfXj>pxjDNoX(+@xm*_ExSVGy
zcEd~w%JDGcCEX<OX{~#<z?t+rM~fb0ygdCgANdF!MFUh1d3q&9qD)jZ2{sy{Cn|R2
z{*PJ!yfWF{0GqWNyW|_9cdE95+WC{@_iB^5-|_7tP#6l$Vm7LtX4%%Fo_eFt7JEE_
z2M9lwPnRzX0Aw17FIzikeO^?2ApUCO>9C~x72>{J$zMjDqeWIoH}jNXOUP;=JyzSc
zQ(sJGuW#?%mP*5Ze2Bd>{BLLRgG-%y;7}xR_dl)0PiHZzcP57=ZOw-0RS?c%6<(-I
zJIIsl)<WvMe^lNQ`Xj2)9{AIgPVkn9%Ou-7%ij^koq(u3R+R2#yy9;&^M_{xBn&OH
zHvouFNJw0Is26Wh=^<e{$4EM3K;0D2O%%OFwTvm<v}HClYj^$?s<j-|sK7*Kjemzo
z==vkK`7K*Bp;e1;R%R75n^EDpvWq!I)~Pb>AkH=+0f{6II#kan5AZvDe#M8``U3N1
zfD8IPDELwd7R74lL+JXDTPGlS#FTVD5gIW8y5KW=S2N;EPlkYxk6@LSi>LI^59*%a
z60X()bs_?ZH9t@|8(fIwh79$j2V!kBl@<jIbmc%=u(fyVagm3RM}vyGuQPJI{3?B`
zJK?wQLqP$Gd`*tpfR4fw0q`pz3u!y`cHiP>)c~^Aec~25SP-&a3-Q(YSJ#U|LPZJ=
z0fJBcLyNmGlDZ`2A0sqou`yBP2We{Pa!Q#k@<R}+%`8Y5hnDB7C-i~rZgt!mmW$=l
zAPHDB`nQ@ae_Nv;$PwMruGaZ>Qn(Ik<K7iNs1n--17e2(Ec{mD5Qj0UEOQ<;ypHGH
zfiuB=njm-v$wTY;&=Yt+-epVdpTaxjJt-uV&VNDNrJj$2vd%dFki?L(|0#TfZwThr
zlLCn-M25v-W!3+)g)7u#myJt18aC$kRgsSQX$GK85LnFcJY;4)(N&~`_q^Ramh_+h
z{Fa&}H_74vxi~%LPGQ|W*?pA`6e9bEFq5nFZ1xr>S%gy<p>x$mVnsO)<m8R20ualJ
zL1FDgJNrVus*pnAe)b%eU`ma_p|90w9dmYx@qP*xqIpjZi<77)HXTzbE@~5;S+XMy
z;D*97kz^%z6-ArblVt6t2BL<Z86HkF#+TU2RI(0$B%^nu-5^Qjw$>}5gkr5jTy^5^
zCPoc?zn17P^qrZdZjoH)!26P>xyLFFIyq-2eDT5~s3iiPzl!;Jd3i4PBf5i~y=AS*
z3)THpBQ5Ib8>hdXj&z7}NlM=4u8Q82o1X7f=c~mPa~^1}m!518iQ2Fd(9L-;@m36|
zuMsbRd@LezL|5!)C}v!F3o=GFe1fDr7Hr+v(#c3EP6AC1jx1vZid*_AbyU+~YIMtJ
zdI!)q;`m<ypJ$~qe><7%T!Tr5T^F1sxEYaJ3czL>>U96i*of6b7_W4WNd<YQ2NY*Y
zPbMD3*}*#E*6XhL2Fw=9&X<Zqk`+)vYg=jnp75}rOVQEyhb^!9+c8`mJ6|h;<sl}d
zU;~V^Hb<52t6aZ4@BZyLZ3WIh?Bs^o<g)m0y(w>6qo~=42#_OOr=)`$$5WzV@g^Vs
z4}xYs%33?T4acTgHAdzy^uiT8FIB13ouVSgvj5&B9U+~6x$~lX^_Z4?N}`7YhSrI^
zkoEO4>1wcAzi&6J2UgUbHY2+Rmg?hNd^<r6sf?3_O)h%J@4#+=YfH)amr8r&`QCxZ
z){}u%cI?S>USuXRMvlE*hmRVYM8lY5P47~Q8ui4Q@_7D~p*R_zrF@T{b*7EMgPHzY
z9$kV3%8I?cL=W5>N4cV<BMy)S%dEh+-po#$GLXcB-Pa=B$3}lFqK@X2=sceHy=Ul=
z!sAA^hT7gg_|oG>DH+h@o_?c3h^+pujiAey{FS)T?s-JY(uH1DEjdPTI=Fa?>Q5o}
zyi!1Nmkhy%txoOc1<5S$l{6>(&JQj|g7R9#^Ebn9OV6oLuN|3(01=?b9~S)zH>%XK
z63y5`a1%1Wedlri%9w1HVbvOfjS=BhxVu*zA(o4lkMQdOI#QVX#=ZN>T7BHb-Oyb<
z5r)WslCJhu`DvRnLt*!zFjQMD&PNeupDM3OU*!2Layd8L93i*DXs&(zdOj+Wt<&6P
zC=^YEQe22IFcOlcA`5_^GL%c|7r!^h^;?35<@qL)1JbowlugfaN0cJ<)Hml6r`M$D
znC(e?BNr_=xF<~#IVF+Ja6AC%lS(dIoa1GFzY`VDb3M#aJ{*dJQ#EJ>*4wy+F%O~j
zyhwjWghrQ?BPh~iupj-8<*)C*ML}brn&nea|FbKN47Sdf44^sE*wvb?q5K<?<zFhH
z5_%kKjhDS2Cy+7F#rq=V4pYiwfTLz9F-E`gw5Yt>SjiWZL`%~I;k+-EtSvycdA0Ig
zf=-riX3{sr@b@Xz-k^M(t#@9UXZC}(X^72E!MPIrfhLl-<Gto&ua~sC79Oqa8T-dj
z^@B4`k6Z6LH2{}yniRh<&Pq_t->Q7|w~3ONe#F}xV{Lie>=F+(+B#_kwly9OiMLfP
zzn!Y)Sy`x#avZ=?m83LkcT*OW>(xM1FnEJFkc#I&Ye#546s;m#%IcgY^XfT<yyDN6
zFUZy}N5sIilYYY@3d{UEDRBOXpmcT2;AauBFk#cp3J9ROHnS(sSFBz^&Tv7^<U?WK
zeXy3#>l$w(Kl9y*)NEkx8w#RBLfD8voHVDa<>Gif^x<;fr1DO^(=$PR_M?h|LFR8x
z1Pa^+cALNFy3x-_-R)@Za>h=LGJwjdH0V`1>10q<ZqenGzsg34sbQA^j*Kg{Bbc8D
z!QxS!1HvuprH5zI7rbrP210~Ezf>FvCcbnb^!ce6=1ALO{yA`*r74cdPEZmInBnr7
z*(iGZ67)50;k0M`erWMb65MV&m&Y%x&OSLY;)feBR5*iFTy<C5MbO-nK*;B#1T*Q7
zlgM3niH6Q&j}D8mL4j_)>u<el>9l#GHwIrV03fM9`XE??TMw0?ILC50p&#ZOIPn$q
z%?j%3Uxi3#;l$wP^ocxOBylBR;poY^P3%!tP$>Q7edwTuRYHIK9yo^z{&VDzR(7|!
zX?rR&%emshJkmFD5gsQZ@Y0QoD`%sfHC*5!&VC>%8NQrendAQRQ9Ch$v0rUyb8Is|
zP-vgm2af)Ais8VT@paOKt)%&1$q$YaaL2=x9YY;qnj@o;5S7kx!P+(b67in#sgJ~&
z6o$&c2L;?C+mT@Qn7Nu7wm(f_Q=#*uPvi{_<-;-fpQ;JF@A<?u2KQ$%cEW+XVh6Y5
zgU{i({DZ|SDJ)Jgb-V+yqoHrXYCLd<X?Z4v9wCbE<QK{w#Lgm%_Vw20c?}tY5Qz{{
znffD%sVC2Iw;BHtWc&0m7Fv{&H&#VE)m8LOVAq;PGzaQ*Y{EuK-=GuhI=MRiwj%pl
z{$<&|b+S3oCBPUa^n*NiOhoRJ2iVa|kt&>W3fQ(9EWrLwsPzgh$r@Z7wg44sxjCEc
zl>UErt~BRytZXW-eTXhg`Cf9Rqnq=fXh;@62L_&8?;$S_dr^+-q<JUpdj`7}lp>}d
zHD>U$Lak5SBW%)EJRS7lw6xovk2B*Re7pwYZ;2xA;#S%8&b5WkxeZi7w<XUbS6%a0
zlNVHcd{PTw#lK?(A0=y!lK_<_YX~$<`4RZZt|d#Votm4)K2KeSTw}F9HQ@zQ{|{H+
z7#v6&bs5{q#GZI!+s4GUGqE{I$C%i*ZQHh!2`9FV&HHU_ZSB_cr@OkVe|7hB@4-Et
z3Df3=NwP#fQ*aTNvfch7ASoir`;i<Sr2MWVi*KHC+=xojhN+Z)t`{2ujCh0L(<QB`
z85X~$3D(xoOWyBB#9(5zTj!J_Xv5KwBMf->WsFUkH1-8aU7}w?R`c{Q0i%a-RCOHZ
z{7}^gC5UAsSqoSrnS@;N6cqf5Upj9XarBta9s{tWhVn;hJG5qP4trJ<QL+kMM2OR+
z*tZe?>gW;k23%3650Jr9f}Qu^uB?(SLM=3^_5PT1sv$`9fA)8J8DkC)G6ayVCH@&|
zu5M`MRBMV2)f5aE=fZy;fW8U~h8znl^#-g%NLL(4cDS=e(Eu=QPQS{+Tg^`Z=Mtim
z7pRDu!z*-ipTmQ1FzP`fl9c(SlbzYfV_q^7OSVLHXX>H8=Oidnqc3C^T8>Yq#RXGH
zOfQ^0EFLOvck6*Q7|?@^mJaU6o|V_hI(N{`?DPlhcs(BJ`An%A@T!FIhHBGgxE)w=
zV*1q`*~VSPdV{MwIZyOERtsd7xC)zdpx$W#q6Ag}J%?hY1h%Nv)^E>h0nGn<gce96
zh$ZL(n6@S8NORgODDTeLGAQH%sQ>P`Uqy!$Y=R<e4jPOQ>~^Xif@HSa?wvS9`Ux>o
zBm*f~cuy0Y45nc<07hjXi@`+<*9~C_^X_-^YjL!9rs+=5r7;9QBJSt>VXozalebkN
zT##H0yx4IA?F<E{tXfuA{?C=+w7uIO<X`-R+{*`_jMA|zZ5mk)h5MnZdm0Fo4p9xe
z;GBjWZah;D`pdTN2M(D9Q_dg=M$T4YkXoX7<QEIcvU8vgkbeQ6Rkob>F9gO(1%~ly
z2)I(FhP<>vF@qZHh0H?vW@fE5D_V^zRg$~3u8Bl|?2QnI-fxn)XQz7ZQYqEeIjtft
z|LOr+B!VUONs9<~0jlkz#)X9ihkW_1AlXXeX_o`xGxd%P^+empI+-<szChKUdNX<r
zZ*Hp}!R=^3-HIL?Aq{Bw2KyU#R;u`#`d*&5%i=`_sOeo7Re?;xl)(DN_*Br`v6zKx
z;J}#|4Ecb5-sLb>sKCm9h@7#*NBeWT2k#?vRbkG^)-s6fLS+YCJUNQQgiuE$bjcDd
z^#90Iq5|LkGf2Qpneh9v760dufYM6cP|5=~aWQ;UUcGl`BrMB^oHJhBEAFlxj$}iS
zT4I`R^(A2>VX(t`C1o3A{>5S!%`l<__+cXq7!XrX{1_6Pn7E)t9`KzR;zSrSJcuRG
zAK;7tF)+rbkj?dVz;^E58jHma5fa`G^mRyCp(8$az~=u%182ex?G;3TGTN;*LKpxh
z&XAyt{(EeSD4Z+8Kp-{7*U87DHrGuL7}Ljr;qw|k3AQYSTj^>s`#E5mmuIMa4-Ncl
zIY13xkMUm$k3*gps<R`DR1zv<O*gW_nPjFFOt@6%-x^82{nxo+*c`w=E1h=zKiZcD
zcWm!{&R@?Cpqs8PiHlghQ4l~2)tAMhPw#(AU(hdYg8l}6EetRY2l2dj{1PgdS#Ic2
zmAxqMj{%{h`So?w7}v}Q{DZHM3zXMjUHpyy!T;B4k)(Zxvod3dk(|YW(s=w$pycUx
zrpXqOdjth&vb}mZF1&uXhqToj+}85k`Sw`}l(!JN{wq;a`w;Nb|CiAcd-q;F;gK!9
z18?|zwamTmE#teQ5RJ&+liY4zzmGexvQ!xpN_OT#|Gc%88Ve#N>8nY`DpFkaU@c{E
ze^)I~T-+Wjl|^O4{ua&lhCHkD^O^8or-)So4w7A6JBDAkGBa>Jb<h98%73u+ebJE^
z7)~NPb&p7rh=-k$-(1$DfB2FsjrSm(&b!~qvNf7|?L+7-kCoe-_A<TTpuVU=3JVSs
z6hwv7%_6k^v|f(2Ysn+Ja=&4@fhTM|!gtaQElXE1X|z-=bZNuO3Du2FWD|6ShzMk^
zEP0s_Ad%>pv)r;2Sk&?>pkW@h)}=oK+{z+UJhLw}3bEQ+PmvXL#o%6w*RY+vO4fxr
z&VRlhTtMlTXUYfYD>MfBD0<OYA}h25qaegM?IJ~#VNJf3qkQuMMcXbx6SN5&hrPBw
z+N<FFJz2le>*zg(eh2UdqB<tH4OPj12-PuDju-6;=1|iJ=U_T?%e3rdMcZovobS25
zAHmMWCO;7ayxbQ5MKZ$~NTNZ?5;_Y`>6$1Za~$&X*jfJ^9#~>?sJqI!cl4!t#wDk<
zp{X$tWaf*x=9g^rbd4=75f0=5XFdyn=|tW|bk^Hj;Bn)ytlmi3Us|UF25x!ZH!t!(
zD%UKU*DP6n4T8<lf!sR9cw*`Rz<%~<jOWyn+Z6sxj3<(>Xy^iqsJd4XDMP7wt}n6G
z2~J!%X*_PWPQ}<Dza+i?bHLs%ZYwet`R6%}Mxl-z?YDTYaVqTsa8#rMLUL(7n#`hX
zqYQEqEYnW!eeruM&wy{8XQ#rRRqg}b6u&j?ZLpkP&=NM-dl0DF|H{k&j!tNj*XQCH
z#DmaGQN%EQCv&`-dWwEEX990x2qDC{H&bx3laqgBc3R3-iolG>uZ_u_6n$qExHIXI
z6WpUKSKts`DSu^xG1GyJ@91o+Sz*m+J)1LQnEUGca2{Y!L0;gJ%yFNAK!AXW2tMg-
z2OHvH$M&R+#s2lHBUl{=TxDA|u(a#`b^Q#z@1Z{~k|Bg`KR92XwMrz#)tt{2l3t;H
zew0~ZtiWOx-c&3lWgayS`<SFQ)Su9<6EYX5Dc;rqMd=!^b%eq3vxDqPL`J%*+`Fkv
zaVG+MWWH*RONyn-b9B%Xs0^<UB_ikbLB3$bZkwD1yI^$Z#fRAju0<ukg}9)n<6gv>
z$xRMasz{Ou!Tp+kgHVO-HPm~4o}_air>TBf$f}8CbCc5c-;}{K%y|^N+>g&rGff^E
z@UBB5V>6RW@zhT2M@x(i%7^W3GEP-#C-Npg2yN>)lL*lzG`>A@&%GBqHun`)c*(iD
zEx$wbt{@UFGiVh8CI*@)Zh1kTdUMc($SFqRtKao%Y&>fn1bYxp6_*)r8JW>?)ha8>
zc(_BTjH!;Rqd7@2=)0Ki41f7=6raP}yTYR2z}p9ckFayIABtYgonu60v!Eu22mI)q
zG8j`uS*M!E@!qLvXd*Fu!8i|%dzwA=V7$;|r@xuamGQj;SkDQZAj#|D0Fji*tNC-Z
z=OUgebbI^n!ZCD}?cye+^0Pd9F|cgF#0<%@@YCv?#E3U3EqH&U|9y=_RdF}iH4Tbk
zS@<4U&)<4uQP%|glBaOoPtwq>I`!)XELX1qeAM&^+w1S8>kz;`kq$Y<_<LA~tw0!o
z!yYr%^^bYL`q7<X0m#=jy@J>MW7eLFut1@m(91wG5VBk6I$Kq`s`^?oPe6RHfBumJ
zJmanR>@yR)mVHm{W;JL+0GG!nBvtlYc0Tzw^*Anso!trvF+Loy>Ae0(SP@QkN+@YL
znFBGIe5Qr>CH}QR{zY(Bm4Z)WygIZ83d5X~gO329(5LO+-j=Nh=70tcJ`JlE>r^Gn
zkJl&4B9nad!Tu!dpIEb@P?U<iu9;X4l6RRvOE{f=9q&J?d|Y=eC1ZBR6CSluK-&r;
z7;dBp3N*XcoX=pcyyj}l-<sH(C?5gN))XD&PE>JG1{fleGj&YZ`KP~~CGY>;{8*;3
z5t0IiMtf6m#$2tj3s*H?*kCtQV5cV_$POR2_Ir-}|H~L9&|t{AWDp<e7dq8d?nSvm
z;<~%ED6n5>`3*?0h2MUlQw+Tp70L|<c0Iq65sZ(h{`_gVYOz8P`zqyjnN1ygW`EIX
zwINSX7uSP+ufp{$Ec-ii20yb$-hyVZ(<flC`uOl3%KLezL!H@eE)Yz(oh<X3;hHc+
zgFhLbNx9<6OBBbA7-hAET$^w8xDbDL^0qpMPbtvpej(<TeQ`Jp{ubRSqTzIJeulQk
ze~iTJ98av~F6J;OGh+rxUed!KEek(ucLxE_#1>l9Ura?%KgMPkVs`EO&&ciM>JpeI
zT3PPQqkNN{i9wn7N~!4<^}yk5wl|P~cGESZC^OS%zrIAt<xR~U?T6oo>H**0=?xe{
z*T4SGrC<`Mrlw(XmIfCLVkU#lKEQ9*v4=4LJ<I;w$%+oqi^pi4X3JV6qD`cjO-|8G
zg}}j4v6}#1z%&9Oey%CwpN$xW0S)*cniDm#Kpsho4`ZLQYBh^P5MT2ZiK<8r6zbsM
z#=HjVx{A({<Bs2>G!IaOU__w?!OIZv;Vz-zskHK*HksC86Mp^MZk!$hiLE&?+X|m*
zS;qwv9iJ)Fd6Q@9nraPEL_Yg@P-SKpenirq<K;;3-UV8$qI%wi_ya0$CBQ=e!3eI?
zph$pFm8C_6;>&<)v?YHvp9p>08Ce#|GLeEk+sp_l5;Gl^hn1>mTlf$}dQ@?<9Q^9r
zFB7>?;NcCHN6#`#&!gsy={-q(ml<^6;K#|nZ2Y)}34@tecU%Y=vfJi!1~CRQ{EwXG
z_!^_siop==o~V3*G?_6kSU_4QTcY9F@EIh>Ew1JciwPeLT@n-#ACYB}Ia<g=tS@a>
z+JjfXNw2>*)O-;`;Qi|8Bf@(3r_H94Wc?6&zzj6+y(QYPNBx=a9ERSns>4=CwQ>j~
zfvWlBoncTIe-Ohe%4f36{+m+bAIX7mc0@2r?Wwfg6^$HirO6t<f`M%4qi$|;b)tAl
z)nbdjx@$kofl$+T+|nzHg>hjK;ocD>uhi)Q*U?aEl>1?PVTOp!ivYhF$servoLMsl
zODEa#hSU=SumjLv>yXaCD2}BTU%!ARdGjt2EFD{cYtuZ_gj)<ZL1iO-MWQeSUCQ*Z
zw+?oF0!LogE@Zd{s2KGNo2&w*Y+Z5Jd<xp07&t<K2$tYuXfsrpHH@degU+-mn0w6K
zg31HWtJiYoEV9V4Mv44*_0sU=|HNPx)Zas=`pMbMtdc|H^lU^8-Ud|_75UsqY7_!n
zSvnpMTY<`+$cb3b60+jnTaA?VQ`oh<M<phCp-A9eLrCCj|1o*Eed9K%6XfN>7ID<T
z{b$!j`$BpYzQ>g3ldEHtf~{W5mLg-~2e$@;s2>}!+N2ORq!RS+S!(Ma+6lUUi74(8
zd>;MY9P^;`(sRg~R3&C?#mJzqNzkSJ=nccZV4Z8^sOTtM3~jIp)2yK}JvFrS2d|8L
z!<)xW(7#VuU`r1<!?RLNz(&;fcIWMGkF2R*=E(sG|1%DwXVu=RSD%Qf!~iB`=Q~)|
z4`)R#hVBjf>ZV!dUvZ*>q5WP%PWo~+eG&aPb0DDnVA5%9Z07Y=V<7%2_;5QqE+CM;
zMz9#{kb@O*_=2E0+c_nVD*J%9!qq`42m%O{h_3)0Fj6OE48rK&GeZQ&{iD#}*0eUf
ze74{B)h}3q5{fK1I2*zBdWQI;O4cY5G5>n^ZSSsv6e&_F)4fW89N#~DdXjGM#Y<Ep
zPDp?E%8l9^rRC}$MLN2j%E@7*R{bo5C`{2huLNU^O@^E#J^nZF7-~ZDWuOu~oRfL}
zI8VC~aC72NfI7DgdH;LzV7ow3R*tw|RsF)oaC_b6t=f5<FO<9~!6B_Iwt;EMah&NY
z(kgz^IF@yyUq>=;-PC~2p24-^Ijsbx{MQo;+Hl;zNT9~J)u}k0&z_fOWzy&RkS+N+
zlN8UZ&wPNEFcgfMIMvZ+<WgUU7-=*BVSgVVXiuoO5{r<|KatN!&R%Id_9R>)ba97E
zAV0F#bFnxeDI{A8b%f>34hZ0OyOCv7HYjSDzcC{QnGTw@t-$S;m_V%^J;D@-Ym>WD
z{}6i258jxB%1mMy+ebF_BkpANQ%X^fR$s-N(EAoC0XxjwnE;+$J2uI@zRihd!8i~M
zI5Wrnp4Fbw5|ucTcXs|Q3s|^#9ks1@ntJdg<r_b)TzW5DG(g*!=CB{vzlDiER<UfS
zb}Yg?&y=xS{Sj1}NJ~=sMoxR$X~{X7q9iKvUduZ4uw0(MJFDa)k1yOL=@_u}9R&ba
z)D6pg$Rxn6Cv-u?^XUX>6ZSi7+IT1d)W<4|z!+oAPMQ8>PW!k>-Mw665vc}jX0{Ez
zy%0;8n<}lGK0QS};vG7P>ca1s%Md9Z=Ms5Rq*<-5T70m=<WpFWX%)4faqfM0!10K<
zzo_5>F=u(-QKFBcC<quZ-5Uzy;o{sc%@f|AYb-8OW1RKc&ZO$#5eJ~n_G9XR8c6f}
z7^=v->bfQqbYby%sZfK1-5TVoM_4OLHnIIG{HMTQecsCqywU3cLj6tVC^VJz_2Sn3
zfg+V5aMjWHc$VQ0T7F3Q6!3J(xXBo?`Jbp9tLpeWoHZMBq=<+VP@Pn)^C>Rz;o7l&
z&7}^;fP+Txn-rxtiY3WGG-W#QcpM+$MXsU_sCu3hT5INL@KB^5(XB16LPYCnhSLrd
zQ4{kvWQ{USV;v)%YFjIb#-qX`ll{BxP(Ct`Y&<Nli{}&@dOqP5{lmYzVTMcyz*Y{u
zMiT777rOdP!6y%f$K9#q6<oa(*d=~R$-u-bhCLX4q*L%}xPFdNCMpALO@7$hWHtQM
zO0~+rE0!{)3muC*-^RI^Fy>ZM`dCF1;Y;u{G&GNJQ@U%?v}O}4$&?^naI{mvne<Yw
z?|bXuJtOa9t8{FE&E2!J2E|QOum_h-;d+E{SZe)pwQ#NL8*5!R*vOS2^bFo$9J;hZ
zWSy0+$R2v-2OO{;7-+x+q{c^|6bLQh;tOT2W(`k(YH*dFSwYnUPQlKvj(`?hYmu@&
zbBfxCz5x#;ZOc&+u8S76(|xOTvXPs2vO8DJ2jXF_m6dzcfUhz@6RFl~&?%g#9|S8~
z9*0A&JTt+A8?)~q;;CKG*2glg4x+c}u>+#^XuYjTA-<s)!wy(tld7nbXaX&f0)MUp
zwZLsZt=~n7=bi8V#_ifSh|Y(~(~;QFD!i3Dj;O|GJh(4r;P4t+%ZnLYBge6a83ekn
z1_kmWCeiIjIYrW|ufq|J?5J5EMz6DNEgRA_^)SQ0K#-?-Fw*K)`3kElzr<X*8{=Y-
zWU5)wXCmykpa+D;b_z;&u;#Yl(m~6OvfQC`UMWZ6{v!b07UHNymG!&m%(9bb*T#q<
zUpsk2N$VU%N%t5|Wy@S4*S<?z083Q03S%W3lHEo47haa7PgRO^yhg`n$GeI{%oHI}
zs8`;Yc0AJ=E$XetyYXu(>vy_)$h){Q-t8FmhN#aUVo_k*YsE@lN3x0C>xSFyEW=}y
z5^|6<+<X1J`LmGDjcPJ&F7tKym4Yf=PGx+lly}hjx4E2c)3ATwj`k3xpTR4o@q_TE
zhoUq2T{f0_nC+%z7Jm!90w(>|B#gH_b-XO`vmi*TMxx+Rw#7`BqT;8T9JV7d5+(<|
z8wx%9%>oeiQmx(C3#D9=NnyUoYckfh%rC1jp7hBm47Y+;f9t6xPz|@TF{<Bj7mfRL
z`W*pfSFT}b>>@zoYBqWo_t+h6lC4+1EaER4J`Txc44ccUl)+;daaaBa0b;dKGJL@}
z$uXEzat11#$Y)+EHm+dr@9sDBo=$?%6tLU(o>%~ND3<GGvOvwT5duyxuM^hRP47^L
zi{ZA0fad1|Y@p3NeLIA3p(~i+3J*#O&|y<7s{kLDD+FAHEMw~MvPv2T>gt5K^MKOe
zL}DFM7Mu4`;%aJWwCdTM58@fe+iX$TE3+J2!Z*oPvqWZ-{`d0B*wlS_`WMn7DTsv`
zfsKPjC`RiE?rG1ky!<{zCJ-Fl*lOA;eMn3PlcIUCX3NV1+Z=?o4W!rB#q*64ITdxh
zF<pJJzk4KwkH;6{=Lib=s&+3HGlh?{k#j1F%|UB&p2ZsqRk5Cxb;U2dL1w#54t>Ea
z(AP<1?diM3)|xYKWDQfE^Vs5keD*qb05VOT{z0ksB^JXi%5P%00$&p|;$UiPGzw0`
zOF<o1GJ2f8?}#`zIn|m?43h1Sro@8Z(6D)R{dW5nXaY-7ii8!w48IPHPS|0T!I|q7
zQ?xduQnZ2??-d%E|GR#ns>2*8fy)f>-(9Vns>uWO-|!HnB(%@sY)@|c0Xo+mytW%x
zqCZEApS4<|${~?PoM{FdKC=YEjfhq!mvM1y&L#H*=#Qhg`#G~CZk4OWo+|z;H|`Cv
zavHKt;0<6FL2N=-efcTB!ua5X;C$|f-69y%vGp!l9%0*$#5fRG$t<iR@1QrVBE@@0
z<zqTKDRD<4BA9Py$)Z7b01N76?{7JvGX~8B&HvwKs<(ixR$h{Ep9IbtIie=N2aJQF
z#8Q(Qj2l)?H1M1YX}6kbLsYS4%7$jT|It>-YldanH)Q02ioo5_lka?RO*}c-R#!cr
zkS64j1JsYP3gs_n@#I{#U3BxXxKAiwN`bQ90P-(e`D68@nQg2^o}Yh-N}9dztX^Nz
z&PC|hD=%KorzoAjKE+iIh%LX5+;-4Uf3<dzUaa4ptB(^Oj(Y<~KEFE=?%A!Cd)XJo
z!xR45&~W|DK-h^5MhMC`V_(FI0Ua@b2`$3Ap3xi_gEq6>>ge+CB{_MZXV1o7e*aO!
z1AwOaeJ6uP1j`vBM;6qW%3@YEJTl&hG*PTCWD^m<U0uP~G=4|+7jKheU3M2+VRR4Z
z58FT5q(~Zct!C|agn8v^>Ar`qt>g$t=TzSm5Qc%>O9w^8D?uCKfHpBpvQs;GzqsEJ
z6#G=Ex^&;+mVp+=)cV;(8SxQORHPa536y%+DLY_8k}5r{q}MIE+_$~{fL6I$uU_Xk
z2W?aJ97FVOYA*XM)FJ+7i#HusM}DnVQ~mMp=Q-=bAQ*caNAt3?Vby1(r?YoE$N6V|
zIoFY=)k^odD`xlej~BqQI(>y!o-h-V2PNcqy%!1MOwhN*nv5PfNq0v60p=1T6p;39
zw#^&cKfVU`%clr5DTIng7KaC}S@5P7DC$d*>RDiXj0VVlBGEV-CO8ldY_T|^k6+6b
z1>mSWttLHJk~(V8pTkY2xlh<C{IHA<6WWj1+*|Bq{Cj#u+;G42m4;ok&}$_L1{M*B
z7C4}c4cn-<V%V?AT$!ExeV)0@2LKg8F^@nD;Iy@%bP;7u-yVt3_8rZoiJasM{PUa1
z6LNMNFEuHAjFo?x(F(&Ck%ll|-P2CknjJhmVg@Ku2@M{o(G#>3v_vbUz^zMmo~KJ;
zMJncyMq&yT|9tOc9om}V^{IyVcVFTQTTe)oOp`}$>fosA*P;goPT_`V0XQpasu*{M
z`<?5^P<Tt4f94Aq8T(X*>N!7#i81gx_v$20IiJJ(xXd{BIiI_qcoItcjn2vo@U{um
zUq03P`{c~AzIyl&NSCYM>M>^G4earzK5PEdc#8ccHFU$!<r&G_V^>!k)eS<`!%s<b
z*^-Z&TcwzN*z*F0SYTpH3_uQn-$6+9(Tepo2>%joken__EzRvF!ZaeTOgdvkAB$&I
zvKkACcdQ3hkn;c2&EF?Tz6tA*teB7u^*}@iSG3sWi!YmI*EGpDR9;cX^6P4n!(97>
z91wjxBjx0ycJNtdnBa=u@Yk1W*RmbQ+P?{Lly*#qGL<R3>Fe<90LFIiCmpY;vJ|@S
z*2%a2&J8TdB_A*4WA8%t?ap@Sc?)6kSlm)_bDL4x`XBP9S>_khcz>g&bF~*2s;Hl5
zyS!(`Q|rSMt?lP2zux{~=CCx%6W)+UwId?g>W<0KPYQpU*5|nsXnBkuAz0mILx`lK
zWG~fAzOhjrAVZTX1K`RZY6&zhUfe?;E$go>`?(f;WW5&Cp9-#Z;^nASdj%{2f2o(x
zd%KU#pW`Dhbj^DRmKQ_+Px(D**;6uxwmfE8;VaKvgfXiC&_>q2-=CCVA6x$D%3zMw
znPlAQSP%&4L?SazUPch14<1y;E#Ddh-N_^RV=`9=1f}Q04klbpwLNuQaSw!!N0V}E
z&TEMz;bY!<k+X<NcQT|NqR1Z#cD5om%E$FE6M$`zZj2B|XApTJ{fI`OHuifOzg5vQ
zo<<XRy-@vyyOdvLn7@55Xn0tGISoo`A}-qZ+zunl|2pZAw|sagxl+nDiYqpv?mY4=
z{wZAryyyy2KYSz#{x;IiSUzmCXncq*7|$-Y8QfQSCT?kvAbm82>2=31yI#?ShIUxg
z>S5rAfW$g$=;Z}gy|~;nO%J`X``%5J!7a_oIs#8ls-@*Sb^3=q3;$XcsjkOB;ThjY
znoMY0Hs7I_<@k#n1?Aq$M$OuvkZuU2nWkX?vnB(50;5m($3}{okF6yXY;7f;&E4gT
zL7qN74kSmYu6&6V>Sr&ZLSMD>_p{(>g5sj@b2)|q3L30k6$nAE)jz{rf*Gu}LG<}c
zGs-&Dc_bb1wchTfV*g}HUs@jhNE}N&xn?*VP>__V3>4iNklqhioylaXb+M7d!A}wX
z3e<3$t*dlVUd8rLvR|SbG|~RkvFV_6yf4MEqL?Q5FQVlET_y*!TF-3q$c$*fNRV#T
ztm8HW>zP1t<WkxC=%Ty@Z|vF`DeP4^w3T9bra-n5ZfnDgB%_b{`aAnFD|teIYF*fB
zMQxT&+3oj-G(2S3d!~sLLe0GV8iUj!;74BeoWzmeNu;?^H)(&~*1at2a-EZXlz!Z*
zi%|z-`*{nWQB<g$c=J>2f{lOo4YIBCwn~)c!%-kYj6Zlh7||)-z?l*#Ki)0W*^8~K
zO?CX+#=@$YqVrt3ZXWhNdOHf2kTA;l?;k7KWmB%9hclb5R!_2gT%{Q!b|FhQz+e4P
z{2N{+lo(|({NMjhEXC5-3x6(^bijbC@pod$(*S;K@|T043$IF)+Ukx%E7*;@Mjcmu
zQ4+&{XnS$;hIH)15&v*Qh))I^hK*C(MJ!wDu`S>t9uMp+_8ns-GZ^AgmOi<&{2Rc`
zV*5y>jWCixZc05$R_|*j+JyGD1Rm~BE_>&<*5<MB`(FZj-UE}@DrkKYCMG1T-X6Z!
z5=}ih(1lL27`=c^`E29be%5foxVDBQk-%(rgfvyI1!sY>w4z9G=9u?<lx#K{P$<zk
zw%2GXgZ$S*+#9S$515jlC2-Pb9fhr||2%>$r>|0djI(f{#OUYcuT+}`R8^fj8-Fj@
zIqY$+FQtIob%Jd4#E%f)kQekHDA7NW?+d~ib2qrSqCDQrk))YMWq3L8unmDvw`D5~
zT28j5vTXcGa=<?L-ax(r&vesBE_r?LmMIMGCSvI((r<+Z+8249Ax9bHkmZl&8H#?8
zAJ2pAHw|_hKdH(Gf!WImti0izAx8I%vZJesUmIa=clKCbeO8*qk>Su7KBLPc<L$%O
zcR5x3MV6@!7adWO#4(E`dqFbu?ytCfFLzA2&+K*!(cJQ=mt;KE*2pthYHmq7MtOp&
z?;702S{LUGea{t`&GF99`>0Hd8AO3j!Nc-`ZH7;G*#L3w8A<8}fX>uTp6cG<((@jt
zZ4U}exRZ<es&?p{!^Lj_ZLd04@Fy385SwN$kJM<b-_?$b2(2&&LDw-c5#L&SaN&L(
zu%1o96{w&7++jvixTK@Xy;=04#F+1Dgu$<6HZE{YyMz7U!LZ#|s}$ZzGMH=eX%=lc
z0)=bNvA6mx!-s1IZXPcbS`s<t;exdL&>(}fceRQ{w(&4*!v-1jTF7gX-Z+nt60E8T
z!;{Sjz~=X;dDngL1iAS5tR@)x*ab0;HeW));oc1m*`7Wajg(<p$~(m<ZKw{>C5bu=
z*6-H9UV}jW;J27?coao4I4mc{(NRcJ7H=28`-iL@E4QqGQo+g{Fy_JhJ@Ii=HTd0~
zw>(lix1z=K-2Z6RTc}CTrO}{mgP8C^>*~sAAH|Q-ytXpenVfYoy26lzpXm3OC*Jo&
zZVTOfDr&dq60pzjU)3Y7%YOk)r;^j_(Xn#8SG#0)yM4i{)s-`vLiL6iE8xRo03*w*
zsb1{%JK%ViK05vr6B%E`y@iE%uqvAKXzk%C<YuSfF9tHq$8Mp3#PEL@prkXZC?U5n
z)c_$KTnQPLAcZ)I)d}fo@A5V6qEq6cXg;g^u#!q!sa<LMo1Z{^cuqQL43-TsnyScv
z!#$I$%lIA@Knf#(jwgVYG}#cgvw@2`NwGo#SP1KfA&2K<bzf)W2U2Sj(SCA2i#Jfe
zP8bT&4v-p>G+EKK3~46J5INX!)%tqa3^GiUbeeOKV;Vdvw$BqbJnOwWj#le`FTaM+
zZeg8&9qakosZMw3ZLJs0CVx<Wf`0kEB7wvw9k{OT5x*IIm<5savsUVPMNu|A{hE0R
zIKs`Wpt24K^#<7T*oOW1v)EOSe@hl5z&0Li(4{FWL<+fkRcx#K>#Z7jA<1zA;OQ2k
z8xA0L^&Su!8jrVSWN^{jxqiz0M#wW9MurFF|1VJ6N#D<1z)l~Ol!}wMI}RdB!6!((
zX@{CDhYd;1>IfN(Xp66bQC#Pzt}bW*Fz|;l$b5FVB0Sp=ipBzvG$OzKL8*x-wWneo
zdi(x~IxCHCbXywA-)8>KgJ{IxCVvE%d0=n|hY}Axz?hIQVQ6q!X2Wj{8x=mJsk<s?
zlb*1BwJd>e&*&oT*DL+!xBUcqQI3Eh06`2EOwrKde<(c<G8SM8pMgF<op(tOu=Ak#
zlWd44j?YVmX4BPD?;WFudWW!%vxXgf&5xz)dd9lTe5`Ky`P0hWVCRAEC2O3eHp4eF
zU7v?G#@EFwJ~;R3>UwwkdVqK#+7+JS{ttEvzLA3IznCo<<2S@_UwD|%5uJ^il!hpK
zD)t@Tu~?G$PkvGQcqN~Zr?v?=V5!~#du|1n9l0-1iTL!H6A`4z@zZjqCu;rsLFkH!
zb2yRI2F25l&z!4^&u79xE=Yg1)=ww+gb>n}dFfWD&B?AE`NiKZt_hot`~mof+ELw-
z2Q=cHebP~HQMBI{(m}~mDB$XNF<gKpI339MP#UqJ*OTjH3t~d|39mpt;P;#9CkW^;
zEAk&;-IV|YV$_#$s|Ug4_G-rP@aw~!`G#wVrOqF+FqTZ}d4H*}!R(?TeqiD6@+#&A
z^G=hr<{&kS%Pk?sN^ud-wi|}uzV{v%V%!JP#p^pAk~xCZVDf4})E^Z%ozowu9_qD>
z7oL*Gy(w&@Mt}}F!{pfwjO<-18r5yc>dq=a(Gp8yG!r3ryp!r>bRdQ;F}<L_BhFG1
zmMFCP-L~!&vyaN^sZ1r-dnlh1JNLPXMqejmzAN*AavOEn3tsM~2Jl-kCKO05A0s<Z
zse5(WA~*8*ER$s2<n1z3hℓcgU3+p)K0`m_ru@I(UnCw4`Eb$Q$o2aC-3D-)zo
z@hPbQ4U*q5$!C}o*D855t#6qM{ZAD?iw$pK5NoX|Q_f8rR~;)fi|ogAk8jV@c0!$S
z@5TFRWepH>W=^LU%_~21xVp66a>BP!+2W<^SIul*d|1Aa%}m$B`Q3nCYWj*+nINB*
z=~tAx>ejo@KU+&c`Wl8cHo5b@3#@d81dalS;jFAGv^Rj=OhWhz+s$Qg41|+0T2u2Z
zhM1QX<`p$se1xO|DDfK_aJc&JaOC|Mgn}+Mm3JvzDilFkKpxVG2?7Q!%xW9b;pPLQ
zHRW}9dIp|x0j`Qv{ODLXhJ|-f?>&)Bsq1BQi1nl=NhJp0RdtNIg4X@LHwfWKv3j#U
ze#2PJAN=s{NzmKy)?%GPZb-vN>-W6=R=sS4R)A1IJ7st?NcXen&&U9|7cUP!#2nNd
z)YRm9flcdu7{@;n5W?vNkD49r7t|lg;8a50)A)kN3`q`wL+;v)diaE=$dKKcZ;5u!
z^`@MA4UT32YaV|xqiDrh;L&Oi-%ET?BZtE|Uzr*22sT1A$`qt^*#Hawe?Oc4GRr+l
z@|zAgoHq;@+aLtn-USnD%cgyge+3`8b$D4K!XhTbp{c8J#xkTqvb=Z>q8^Sj#!(>n
zqubkrRI*)Q{Y(J^Z@%h1ppjW8aGPUKRLL5I<oLionmC%PeL}e-%<ylpx@qKR2%nIA
zow%zNd9lgplU^o14|g7UqM+y_P=Gv6Hw!I8hT#+O=?5whpt;GFjz0eReH<jjhU8?V
zovhz*y`(d^zX^w8%<=ci!>-7^h-#R!Wk{O;WH}Y(f?rO`?2RF&h{F<Ls3cA4zqXi+
zigci4KirgiuPEkOUd3YbS9mt(22}5~N+KWX-A#dV;R6A*%nzce@k)YHI-F^}F90y2
z1!B#MM>PGyq6{-flqR79m$F}XlGybLvIlXSBB4NMi5@qC{wFMX-gD+PKSE>zIG4Ye
z?`fQ!?JuKK)}BHkJopIk-s@e8&WI0D30h#N5S1k|DbZ>{9U2^$_EAyS#>I=~#O0D(
zF3as7dFrM_Rr`kaLXI2M+f;yQ^l`d<rP}{xIY^_6k8k@t$A|=6P<g}}*GQ7UWit{<
zouv82y%9pS4=;+U9%9yr8tgEVR^a~VSr~JRgW&xK4jmAQe8ogIF*F0px0^p2kgPK^
zo`@7ryjd0OY}<R0CHOH{6nVHdjT|9T$o(K9(h(SJaCyfKHKN;sBmIRYBp$;i!*zL^
z&uzs!m2BC1hrgjx!1f1yUwJ;uEkx<$CrTWQqOHb3I1C`mOq3!)yuT*+B9L3<+v+Xg
zjWq5=rbV5;XZcvt$_D_9mX1aAlu>E<61&Vd*r&Rzc|ev=N`2~wJvEjDI$HQHGc$2i
zNDsCAw{JAbReWrX{$ZQom!dlxOfNi)KSC0M=s58jM1B<jiD?k#!Xp-K=T*|@1w|!T
z+vtC=x$RLP0)lUnh)Wu#rf`2__$m-f){+u;@H?91Q6&BQH_!`M>zRhp$XI4puax{n
zBdU`9e9vsW5f$4z0%~xQS6BSMrluKd6GJ<Nv$CdO-{sB78ZXJw+loocEB_n%*sVc$
zj{o^W%&NXQ^hKjjhm41eqYNDC!7eW9CuVt4YhzL^MU7H;isTuxd*R}-_vWkr?>11Q
zx8$o0^yrcV+$`)3(lBNNn}-u4rf*+zrnUS&w8aGKxQ!p?{|~7m&K=jiwIbA|V0n1R
zb7B+cl2A~LMjo_G>0RC-LW9YOYnTl2ulS=}+3}u<+Pb}8>3>>5Qe1nq5Oly>slcfC
zNu!b{QF>qY{wdkXh^nBTDd<<|$=JMw5#V5CI<Go0wGk0kY30RmA+NqFO{zMBC$FM!
zG>rNYV9Fp>!D+{^u2K&{P@Z1mh-1y&V8gOOV4EU2N3M!rVRefbznJpqLCbx>*Db8G
z=xlKIf~xv(sbo?bZ}3SE`^;;&&mmKf?`K|QOViiqa0LIy8eiqW3o3+sz)nx67bsvG
zxqBzeMux<O&fhRz;|RUIkzqRwC+argMLca#IuMgV5iqr#?5^KF`;XtjM26U|p&$l_
zn?h>aj*u5ls9>bZI&lM2AD*TGWS10ae;0*@Q+krRx@X%v1$BuKsiKwBH%+|eBCT`j
zZD#2tNbMOQBW%ZMqwgT=45OO^P^6c?+nET~xYv=oI-EnzQokQ=HeOI-XfK^z@j!1n
zUT~kPuC+3rr{|nmu?+lJ;ioM;XvKEKfW3eLMQ#g=oH39aCC(!9PB;l(qs4>&qt1I!
z`>R0l#Zg`w>CZ2vZzcQnJO1Ibf13JN6`O`xv?Ewm`MMP_&GC3W6Fn=J0LwqpmySsk
z!3ILVzHd=%u4c_HJQO+9U0my7H%eUpy-wxbaJGpkF|k~7_qVYdB8w(tK#fN1EV<#E
z(_;(Zv@l5xETjD7NROzsRT}1MAiN~uBZ;8wK1I{Lm{p$DB&)Y=e}Cv^b-S#3pE6GJ
zMo*YL2rxZW$ueS(sH1Sp2F%F^#w&e=p@Jq*YWgw9D{ZH(_FsyZRk?~v9czlMz6MNd
zvaJ5Jtbq!d&1$>+*M9kDyn!k*#2~ZvFX9>p&B<1o56f+jUJPaONw@9dkf06eqFo^W
z<Lph0Z;0?HFU++5D>N8>Ee0;k{(PDe77|SAB2L!gbon85VP9Q62av|$<j4dL^WO4U
zV?PW=eEqv)F+=)W?k|<qPLZVrQ)vH{8J?`Bw<y9&gH{Nvzp6osX{OV2u_7UpaMZs=
z)GfZ~k7r&1@w>uMz4!7(QlrUk3k}}hw)Z@qK=Wb5-T#5_;rafnj>d$P&e|~@Jbanb
zzvy9#RIQL5`gNcL!te=?;r*rGs4tp>fStMgLS6-ixA!+L5HZ8Fj8zih)AMh1TE~Wr
zZ@5d1+0bEVg|P&D0LxO8brnCidwR^ca#*^RacB#!t&{~%L6m}i_1~%x{DXcI^m&e=
zR1t_K+ucI?FeWzCl^L=795Bn7uMcq~>B#zij|$1`R}UP|1|0n@vJQ*8jW5$u1SOZE
zy8bGk6D(#`U~|A{P-jWPTUgJM%}yKzOX8R=KaE~5x7DO4|Ha&{C{?}MJrW8PiNzgT
zuX-sRR8ISKKC7+PNDxjxeesddh&;5V@Ium5u@;RyV{9+?$q$WVKR)vkLMQ(P4&(}q
zb<K9(h69yvVBC?A?&D&veU2Oec<G$)RdzybXO?H>f22>F?N)<#(a%XQ1|i<*TdY~w
z>a;g>>U)0N5{~AL?U3(8=@L$7#@bk`x*3_-x@Qy)Fdqg(GixnkiPI*ARN73TYtgR%
z=?H5S8AG3s_zA@GnPyvB?ypgVT;jUje<eJB;s1~Fsmp|m%AAg@zxlB9Q3;#7B<kEM
zNwG<IG6<2U3{<YisL=4~bDO%U<_tk)*@&UIKM6yFep?;tgCg#Y#cqe=tX+q0nn^n?
z#gq3aNeI$dAPY>^1udqZ#<Q2-{~@D|-*g+34GtPdIr|OAB`Ay}$@l<Gf@;x4*o$BQ
zn7~Jbf`=qiZWw4k<^B^6cj8y5reu3;ObIx*%SMHx#fsc#%)?$ulyS8Y$ZK(Rco6zE
zRvmAYAb%#jcjn@O%fjpI4tj0_oqWNl_bvVK`H!$aFBbJB4Bx7w6?RVUXqi}GFvU#!
z^~_yhhe(xa{E7`S$}`C%JsBpHhxx=E0GZM46!VCr`x|>N6Wm0Q3mM^Nv$(#~s#mrE
z7s{jfBBZ|{lGlO-dk@k6S<e(Upd&f(*A_%C@#USP;$r<xSYBw+!W^>M&mV`A^(q`p
z^D0K;%AF#(G)nG+2gnR3wjI%sxOoKm-O#onUY8VMsM^BVO=-$&YSXIT&01_l05vv_
zEcDM{uD+o!ko~P5c^bm7Y#q~0LbM*%b)>DgzDsyg0EVLJUUGRbw*a?##G&?FsWOt#
z+6rH@kWLQ%Y25L7oKaIvW2x6fvoyn7VI(I82P_AX(EBjJ+Lw~H9j)YKfqZ_}F~V?(
zo%T6tkmyfCG)A(6Z0#*!DzTae@a@MaCwQlN?xUsbrK&>^D^#zM6Yp&9g!6Ac6&`pp
za2#Y+U-U=KI=986&dQLPOZq{G@9YJb4<};hDqs&oefnlumpW6RFdiWOjD_(a$@o2x
z(v<3O=;@}JA$_B|mpuMY0Z&e3XZHVM?Y27YT)ckkmrpw)*b&@_W?wES0?k35ci@x!
zpAgHuo1L_CI-A2#8-9-Grb_Ng<Y_-ggFsZLpM=y<azq~6L1DN+gd5UBt0m=bMi}%w
z4l%-yihbKORm?7l#1e0r8Lpu)2wJ#HRV*q0lN*DOa1e64OqM!jL4OC08384`3+aPz
zniTsz1nu+gI~<=qyA?A9+;|6**sCIbr+<0yuRnWwZPVkxO+>!e@p1~OU2gh|G0iNq
zm!9?8Hn{erL(z&I)5e4k<`rwuk1<{vuF#{6Dg>iA72WQ>jl*X%6j_rd)qP^jlD^eO
zkDe(J9Yn^|AkL{Xl=Vy9N!Rk4>#9_~pM4gsUQ(#@xR@AWvt(%wG%ACJQj_u;WvZu&
zB;<b9|DfENqzB`C@*_uhLasgi4)qNW>hoT!(x>gn@CX&R6&pnODau(8=G7FGIP|(N
z?7;j$w=RF}bn2nV&cQXp(W;M9`9>p2xMB~HpxC)i_wh<Bw|x51D9#oC!1ou}{D5yv
zPEpZ?33vLl$x)OxV32uZynYn^W8}uP5N#?p^e>cp--2Om<FS$2buOdK`3)Dw@bIVU
zJWWb%K_1(>?18`(r2;+|y8k&7dvw&N(faIX<|XInU{?~NdHc-=5e6@nI}%5Fi}E_H
zI}IN)qV&|$JjkC6=z`I7tdNnL74ov|%-<!qPSG+u9!E&DfTD&Djph9wI0qu$S2sXi
zV;kS~Y8kWoir~CVu~_#h{_$O8{ACPI@wLK(&|8T}mq7Jd-$e7HJpa|NFmrjz@~}8@
zmD{OT(t`f@vw19fjh?9Ij~ampQEVTR^cy@2K@_7=d}&eDGLB<}1cc_<3Cpa`iIObA
z5(UclhoG}!01J-sl>8>LK2eeD>CNSO83_c5n>dKvvd+Ym&o*po(eiuT;2p!hNf?p~
z*6*t1ycdfhQTbT)akf7eQ?w*Up7U>8jHRF~3HlY!75`pWci@6Av1C6DKgh@ZQ#~q@
zi+C6isx>!>VVh91_S0etIz4@B?ob{w7SPg{Atj;60Qd-#Ve+s^Y|y?C$ep+RM0xZU
z;mr0^D-b$gHH4WqOhHWIA6?~w#?FF_#3`LRAH9FhB`9GljYf5iyb)3`P$YZh#$p<~
zdLMrLKr{NS;gN9~b6vq+zMljKMIkVgX@G|Zb%RU~YAb{c{d|!ep)$PyZ>P<<W#F1^
z?Jq&@1Yi%==Hx(L!W1a|F5Syylh*p{j3s0!vtOq~AE4o{0SCGpolsi{doT>zn+0P3
z?|d|5NmKV*PV<0-2$3V}xUsAtl@QfnEyeTzp%#)9XE0*RH^}Dt8@(?d&JcSSZ>3u5
z2Z#0;9VSPu*MM~-ua#4gnl@#r@OStdwAy?*U4YrCmAOF~XZFou5<&1+%4@B|b9&&J
zcAk{l!-k@Rvu^+x`U@;NI5rhyuAEGI6s)()c3w}Q&)mjqn9X~LMUP}ItZCG7XKs6y
z(1q7y{jH@8n>cgL-HeCUiPXrV{oXY{Jq5DFCVz8$^q<3L{D$t5mxmuz+NFyWm0RCI
zo&Z8U^V9^X*wrtt<KHPyL4#eLsYCF_J4?LzzR0tvqEg%;##7eJt7*13=$zo=l;`LG
z+@R*^Ud5NJ9cgPvct>9v4Q|Fu^R!bxQLlL8c>#K#k|P^h-h}_r{|~oA%&QT}aAU0u
zq027DY^ZGLD0w#j&fM_zN?O~9r=T%&6L9vG-Mm%YCP#&r?R-88Y^L%};woc!7b+T=
zzoN2Blz%(cJTK5v7URI8y1-~TVl{l79nVhW-8*U$$u&Sj=7!3Wjacq;Ks}xsYsHoB
zT;s&b>W>~EtI&IA%a?0pW$ghkLcmw@cqmL^p3SS~Nw$ri-jsZT%xAixxepV91Uy9r
zjO03?8!2U$VQ|sD;1qe#z<XE`sI0c76i>W0nou38L^YO5u$(gpao{T1lAhZd^65I5
zbUyH`8G3x_h^B1`zDdF_@gBeK$xegDHMDWED`S!#(%R_>Y2~0}?SSf(OAVi112qT3
z`&Fh_DCj;lh1=_6e-uk(vxzSh$jSZCDDW>cq?vsSOMTFL0>7gk#zj{rnin|6UV<lI
zD+CFBy*=%9chIZM${g}z`VHMfB`a8g&G58n$nGcNpE3;}>abVodPhGsB|nC>d*r@}
zps1rQV8u}G6s0&XqHo?{|JlDzUrUP6DjRzqHuvi6mhTFMvLG{MVRp&@yuaz9Dypw9
zb$^Qu4}lU&l2dJn)K3k!Q<Aee#m(DdE5z(#Y%1Z#me&*&G~<YtDY2Q8pu<vhlGpWB
zgmuBdol&k1<TstC+DSc$BgxHzg*8J1HM#Mh$Wb(oPWf&WKTGFB=90c${T2`s$D==6
zp@dILk2-#i=cM_F0eTM)h_{KGrud|JxG=nocpIJcy#~$lZ6Wle*>FW+>0mfBFxp?G
z<dl{w2kmU_=(qRwpen%3iNVmqevM6AiN75{?PZO-uWdq7Ctv8HsFxSP>|cu>Kvwz`
z@Y?Fj;EbzyRku6}gXKKU1qJtUUg2QG+~6OMd^PZLJNP;60z*{;ED0Nt8#X^)Z`qkj
z89<=HL<E*4L~Z<{g-}!TNli>mjE=C#po-gz(`lz_qzBIza+l?6=I<NNXu(5kAl0R7
za=bcHF(7OmTW5h5#Ea4Fr@qYdqS>3qGSD0FRZn6DL|L9v+BJ!dK4?ehpZ|8&q;AbQ
z-KpP{@Yg@)%F*2ddYk-HO`uy_65!x$IY+oNOp2edk}>ksT5DGhnm_!fJA|pr1TYhK
zxBhai7L)1IyL&DwxIB6tERv;kL7gkE2X_y-6dHfLp2Mf6-3g$II*)ugj_sz|KD~80
zQjkMBC!$=pSw|Jw`X7MP9tak%vZkHTBVP+Kt-Y6bWQ8T*(`SAG5PH;4K|k$A33y9B
zjeKWz=Ffay`8cGEtHvmL;@XC>zcl)Z%dBTK#Bks;xv+lL!*J;XhanR6BLm+L0=nfy
zZ*Ni(9ziHXHaj(|2+W&nmRj90CWLwM2j4@YHOWC@ar*=Gg+u=HLx`$^+`wl?MA~yk
zeRP5U5{VWN()T*Nv#~qND~lNZcMqp%*fY9puu1BHqc_o%XrkL<>buFb>*_t_$x6^5
z6%|5Z=JV`)HJB-MD_70vl$dz9=yPt=5I0|tcJ9V)Xqos0$VY=0sLz(hrC|-@_}GlR
zX`vo9^*4u-Kvc-xlPI?SZwjrK=|VDanZGO>x0VF~nQ}YLG<F?m#EfR@0x9ko;_DMk
ztw<Gp*E<1<B9Zex*~EUgm7FoeMvGnzdda&<hw>-6LKbKrixdUN+}3OGh3k@k%}dvn
z@zS?f3d96N*q0JEsoeaEW@mGyTgc+2SeU*~-#I*U@dwyk-W?Rp4j=BMvOC|;%!+0K
z*%VK})U4*S1GW$T4!Yarcq)k4dNZ8A$N1{#z}<Q+O=K1CLPShb{+CX|>9EeY*j(<%
zdMnaB{k5Gq!5^Uj(SaGdg%}yltZcaP=$NB&TclL!h42E%Z~QW<2A|M!yejhR>F$eK
zr%DCF=`*Wb%#QLVcy^}JlwS2Roh<m<lmf+o-#nh#{jB-Jlu~6m-zf#78#@wY%$3VY
zec5qGsNsvt{H-M?BeDqGcEN8YKJJL}b}jtzliRCK+?ih+S3_UAKW{&X6Y$h@1?Mrc
zv0UE5RCtZp-`bTZ45y7lv4V%ky>O_Qf(Dr-aai-7VC0-l6D0X#6yHataFnX<^$xiJ
z*%x#|$v{jv*H|4Yz{)v6c}^M8-aj?Tcx^7r*(L-xYRF5@mm<&^Z>CVeworMPPuv_p
zi1XdVWselJvj2;ZqU)7cptI{II$1-oc*MlIQEwM=LSLkX5HHYwCYAb#Xo)@fla;6N
z?hl<E+&A15e}Y!pl@!Zm_WLk!!ung_DX&X-3v^sL6n=@^sdiLwJ~wod4Ns`MZ*^i~
z=teWo7V}^NPWL6o%XglnGt5*7S1J2grle(%_s*?cLK9E?&v#ct<4Kdfc>I%DtL0zO
zzvS^cf=eZQSP|`H8^7J{-7t3)XQ#5=8N=TXSPS_`gKHjSjh(iiSbqH21V|$RZWI@T
zF_UpTx)2)1eeK1&y*;J+VeWbmKCGq9hlzpoLB>w10<*RAy?{`fA4?>dVX=qeJ7z4m
zKa=O()mw?@-xGnZ{`W}I=k#jrsC8UK+Z3*EGYavSjfP{v(2rvwY~-k_=#Mv;^Fn47
z&~3k9w~ezKO`7wO3-bSNxUG!>jyQDN@AZ28@ER_!X9%<<(jK>H=s~W9N&@1LNd;i9
zspgtlZ9&Q5i2VNCEZ0S{^DTY9vv*9qd`OhgB>tqEo(h|`p7+vJ5dW+TM6XU;haA)X
z#Q_EZx(l|KdKm5Tv@K_ndb+1~*2aab@Nu`)y{Qg^2TG8P6%vAF$3(CU^i(93h1@p>
z{x7z^u{qN(d@{C^iJgh9CpIUX*tTsvv2EM7ZQGn=Vw;m>^V@%I)z;R&xxc`D>Rf$w
z_v!xR8hf!jj=N<D78FP0ptbjb7(l)6{(Szd_-5nnqL04V_-Ro+bw-g8aeP?k^GNn!
zFN>@!tP`W=5S>ByU`!zzVIX9k9EpQEBr}kA6d|IziEn9yt(G;r_cf~xPs+->DFmLu
z%pAiR18DcuxoUKiAaUnibAZoh5yuJ7?hH*eoxgyM-dN_MBb2%_l*_kMR@=(!W*lyS
zxmpwngd?4>6`xKIH<!g8^6HsswS}%;CBmB8tE-yi(3!yp*`}TyD<1=fe?82|M$?|L
zP0va4>g@&&8Z`0esk4PrVq6U2WV!e=B4@nr9{PA9o;tAT`@A2SoIrh$sdg${dM4Pq
z3T4JKo(la$7QjVKtYq46Un}nqR|4{%_&lUVQ3c_lFtA7MH9MWbe}DP1^L<CL1d}R`
zfy&hfi>XV>oJ2j`9WFYa`XxyhFS2x%B};uqt9s7}4VN}`f6hwEa=Z2)fW5vP<Z2Ov
z=YnOdMb1)W;|{tkHvqxzEK{c4j+$?irnV#l-NJeV;r#r1K;FO{76W_y+VDu;z=Z4J
zeL$ICf_nToyE2|3gG@HKY84ZrI2^r9oE|{fu|}5@2U9qgg37{?gOH$_My0b!<7hdB
zSEH@t!|MIHYKa0#;r*)+W9hzsaC~q(`QR9sI3`mAN<2jr9k56**3ei84D<Vgz;9qI
z1OqU6Oj7u1*u~B7=6G(Y8h)Jye}977WgqJ>|Iyl)xF_E1=;6rK8RjtvpH@qKI2ggH
z+f%b47mDj+IseJwBFRWs!GDV^r4H>}uDu2P^j`k#Vt!%<H3U79NDVDs3FoZv>0BTt
zAvi3{DgM(1B+E56+Q}-#KhPxYRUBcoPY(2!zQtj$n?fx_fYOPRuWVo8E?=EgndboW
zkH#IHf(X=RE0B}n!p)B~P{)^WHD!;D)s*%BiMWBaj`M?V&z)Qz!NDj7v^UAKeG7x%
z!w{3_(23;${M5RN&+-Q|Hb9{K6=48!tUk$U>h9<bSP&H6UwfLTaFB}q?=FNiB+HJp
zQ*oqb9h;{vGP88Icb8T1@_JXGQu*uiW8vbqQLZhI3)yJ$AJ1jK-Az5TjuU=1zcXRl
z;zl4;`koyIkFVD|Sq@Hc1LVqxmt_C2$@cA>!Qn{(wc;`7wY#KR=y1ISeC5{6>QIr`
zU+J(0;8QO2c$`E!>WBxlh;XT*X^@KIP-Wu)?X$9D5st^BSqm^cb3P6=go~n1`U7>S
zCRPaZ;a9gJcTHd6Ed7YsYgqw=YSgRUv1zkgIMB&Yqq{arWVi3sv|P2~4nyiVKE@n<
zhi9^!FAYyEon>#^+8^>M3{G2~s~jrmHiGR6*orDbAa)W<hkoI0J%(`cB+E{04c9^H
z7xS-xC)bzUC>h7UxXTjDW!m+QZ26QA(_lk9=M%e9ymyU*fHUCmSSWaa2$=Ns@80j=
zE*RfZI>?5!PNkYeqUus8rFRBd#;WM-LeG|Hv)0uC?wa=>iz@vSv4?xDqP{@oG|svk
zFl%lf-c4s3)L%&tdyJL7=Vny*cZWiUQPLGGxbb{R%;4Gutj7W3>FQlB)uw*~;yDKa
zoIvL0d3A&An&SW!gZga`ZP8VeUgqpJuGUc_lv797dT$@e)(vYO+QJ^8!pJ?MPEns`
zpHy{CQ%pc!XtSTwf20Nj8{J!MxamBQ%loEZBZ_=|Aj~d0czCR)90Y<XXv!OOHsj>=
zeOeSWlq&S^wO;=$76~WHUw;wVye;br-K##JT+F}J$glG^MgU&&pp>>4jl*1{rC$z2
zK=?jYJEvUemrYJ$2tb4B9(+{-fd&K=)eu_cuuO%K9)yI~^-pg$cz<USfLI^M$=tu@
zFCt;I*!Y&Zu#V~$)|qC{!BpL)N|6By8j=34@uCKI)5>@beVa=B+0>tr(veWz*D!xt
zB6$Rj@GgUfJT`VAqRO&XppEGi493MnR$L5M8bv~m$X%{jPE;s2kwCvPdB2MkrYaZ6
z=;vh!b@NQb_K9i^!v>MTCO8S~N+wFUctD@OC}5s<S)EujYca#sPZ+NgX~CH1mwhvo
zclmb+&Q-kFei~N^7{BdDwljf({`tIpLiObZTvlL!Wn(0+es{gUIxCsBU4~L_wb*gm
z)>$X-MaY&9;<{<qS~#J)sKIaVnEvr+w^psSRqb;2CiLn%TSO%3Nv{^j&2d>P^?vox
z86SA`_0ONzEU-GECdQ*NO1tNVz!2obJ1=O6nxw@j=3c4(#nzi+2J_nn2Yx1#&=wT5
zH6n;yK0B<@bN>lwQ7+-k%@!KUX7m7!;Oa~&)5#u?{;t0P_>`nnpfg4fm|pWRCgEN<
znb(_%BIa3z{knGmx7h_MgAZ&Ky*?XDkRhRlcJ}4B3bBr4stO1bknAHlwDNp}u8CZb
z$MP*3x_v`o=BZ%EiAQW+hEyv&4B0okW9$cpVvs=ooV>z#Vkn7<d9`fP+%%N#6AMtU
z>j0dPW)_e}eCHDx^Cl6v6T}!^v2s13anD-gUv#2gak&J1p8G&qU8PD0QkD9A%r(pT
z#ym*j-;jJBt^O-l+fYN9#+@_zhN4tsqHY);LC=Bo!a2hsf;{D`mWSM6^qp|lJvn|2
zn|DUb4Hi530{Vb8w&Q|SB8!%3>>QvpbOwCPOJV>Vw6QCPG4>C*_F>cb1=?}zk?2Jb
zbDKD|Xf?)b8xo)e>-w^!5C&hD1oibpMX#`R_ck&3leJPhJ&MzQ6r~(g^`SMaWd?CO
z0NORyF{?`z{^YYE3#{w<$NH=k#Nct~d`Ejv&nw;mvl$#U>Tu=!t2grNR-{nelzFxj
zBdu?|=POJzWnokU=dYuZek&g05X~liPQ|&og(#+NwGH51@vWNfCTVw4=-MOBa9`@k
z<-kD_-&~JbG)S`cC7xq>6Yle>+d*ygxmPgTM(R){IvIz};rDTo4HD(Co+30)A@f*Y
z7ykO=b;5tFT_j@ja75a=i#5~0Qb;_bgqk~IED(+w!d?X%()zc`V7Sqnx5TrTJM8yh
z$E5A0BLZ*(^{>Bwt?l8G7ma#dgp@OTV!*v#YlP)M=~6VkO_65pHNj<(wra8>F$6gg
z849-uslCro9_dm@0Biz2m6Ofxs;IOfIWHzbG@2h83Jqu+Xeg;4X`BJ-pOaD_7=-tb
zU}<VtUTE8uy<+*!ZP-7z7$XmN@!c!not!uW!_dG{wnrEM?}Qwb@p0CI#peTZ96{zD
zWv{uw<4%n`Nl)%*%KGwYnc$y+0bTIy?b6he=fNLsTrxCXuf=X%u_56!RZrs%vKBco
zO&bE4+r5i9gRonLND+ZeRG}yVP{BLB+OW{U`+u&ztB0*)&iETrT${{<H^Cgm|KzsL
zGmQX^Md_<k|E~Yyd-_*l>P{`i8=4Hy(D_H&j6P1@_7S$23LI_H@BcI3DGB;1nkiW-
zIcg$39}fx$k4KuxAnSHVAUmb~_3N61-1t}r2bvs(CZ=pkr)!cGBKPDNPUVn%&=g6L
zYACD;QZR~BgmA)vC$wJYWHEot1OIL5qbfj9=%;m=)KE8~#IQ#WH`Lo{3d_cg%F^0o
zu82dqMk!yK!y_$F4cvf5pDba+81gdRxZGoKvcUy<5VrB}g9u?3&cX6o2qmJ(3bVof
zHSzPN*lXZN4M8_YDG+L#2e&grF`UzGnl7Q!Tgr-WklK6%5~@gFa$(enEx1uZE=UWB
zxl*T*as#G12~{p%>O9s^lu%r0LHy4Xu1cS^fo@u9CEr<+3<q%WdQ0;!iJ86X8s1Lu
zt<KNxzc$pwx@4M<9dpMZnvY{(9C)-C_*>JoVW|g^bF_gp`|DFf>eJ;?-HM7^5x%Hg
z<Po>^*0QodJS{1DNxf^k>#<;AT(Aha^PcUl&JK&jOeC+*qg3(I`7R~ygQptsX-zxL
zgR4m4jp$mqF59@%4@ag4?|eoAO`1$Cno~2!q&tE-4HWq6Vc|a=MMT83mHllZjpQse
z%Y_L8I1&DNYt6$ke<{*FZy3%Qm1*_g26Ike&WLD$oK{Sr!ZhSr>B2~MRI67rMK=ha
zIy_07VE$$|ij7$4J%;&~5*wXXa#1~&-4UH%_$W>Yolw?1-Q?P+@+NI7`sEt6fQ8Lm
zGU7B!20g*;iIHey=<bV`Q+k^hDguci6A2O|9(9Xl&E`a6k?}7ij-)Q>f6}5HNps9H
z1buwKO!KLK{5pDvxMxSd&&xgOVhl_ng~4Wp@cPW{p>kvs>RiX|^M#EZnY8=rocm5M
z3dNRv=T9Pp2T}}92qrYY_ar&x6;~@0nbbbkLdF=m7_BU2lD2L`!4|XFF!+OX@@C0^
zD7r~XqEIO*m$!8njmlLiiE<OQp|ZU@Hc5XV_9J)65iz_r7?zkezeR}1x6=-+a7A8?
zUWF02-nD(J8Qw;&gPXp>b>7G*-b3UF{t`ZN^f#(g=OeDOy7$rFiZBi{djff}z^hlS
zC=m<_aW;N$lFVfpE1E-b>Tdc!sp&_Ng0PdV_in>8-gkg*s-#)0@4@(eiTJkw)SZCl
zxB_CGEWeIJwFRf*?1*hOO0DsZrsVgL-tKfsIjLI1_^WBP*R=Tvi0-n?=(HuZ!|JyH
zIzvURTbP6L*MD|f#Y=Bh{5K>WxSoE{s3d=N+i8n*A0NjofOuZ@ux%{ZM3gq*Fw1y|
zVsIL?)Y@#*G-<JDct@(o&4+A@-P!<|SSCVMf+<N1ZpdLs25e4IE%h+{;<AHEd@fK7
z8?jhjl}2~^JUy>p$o!5qF7r{^_$`VJH<xVk!=g?GVN>jzraJ12ai*L(C`U3ju=i%@
z0u`^es|eZ-$T8@CZiiCfSMlb@7Kd!E_8NGi(gLx5hgVBHi9#>>4cDLX)~Et;)6E4M
zynZDQ%A?}LZDO-iQ?k&CM`KJ-XNaLM=|NqhE;%YN2lL(T6x$rEU;PUyuW@OI67zz&
zBta%NM-G=cV-YANjS)u!?1*@ogL_G~)9I}w6X(3|I>c6ONju*2RwHMnQ?Sl1on-*F
z)l$<-6uaBB*+9;vidDK{(9nTjm)bI2(iW(RIpvvo<1YiPn&eNcCJPLQ6!!gSAiK8X
zn+3T8nI~cm)uKA-VNWJ@`@d}?s`(ot=7imMC$aDEs%u1&Fj(Qan(rLgfAYL-*GnBX
zR3w32vNcT~uY5zlE|g_U1xYS3P;mUH)z+=Gfu`BaIZ>#SS{sqDP7L552$8UzGYhJ|
zNXa?^r!3@hbD4}L&gYsfx4)eZEjQq+D=p@z(Z&4_PMEVlg8wluIsjSFSC?E|ter}e
zswEA+q#MYbL!ER0!82)PXPyojpsgc<(aAkJw43%`4<c}`%xoe`AYaw7Q|HSgkc`}D
za4_NNtlzG2A+fTYmpI_>9c2WFY$d4r@agYWGH-L`y+$u7R(^;F=bpZj@y0=!n~vQT
zi!vJ>yq_A>-J2NT<wrPuoKZ@2oldnz^xt0OZt+AxHX-?jZAu36j<<Zar=UH!g{~7i
zn{j_LRTk>ciyZUHRDUWpQ7Kiur_ux~=aK75DS{#|-cr7a3<O{FtRsATcwIeqygec7
zdf4t`#9CzwyG;svnMB{GoQYW<sW9mzTzot7m$<myr@52GU}-Vpb5_wh=VSVs+qMN^
zN)HjuuS0$}U@Y?0bkip1C~c&FwnO_W;EF}QGvK#8_VHNDAlaT?a#n3}Zb^R7?RxG0
zUGao}so_H903>A3JjoCWpBMzS51wd6=&<A=uzV;Nd*59m>8>k#8xnX)F%`b6<3b$_
zy4|TEk|zEsq@3AwU0Vx3#2mhPdU6T1sgNB$jU|uQx6z#-@tB^#P#ogLx=k~os1l?3
zh*q_o^`es$OCzOkzYgSyZE#P0;pmKv>!FfvRHjv71nwnTABs1@US}UJ>Bd&u91>i_
zVh28~D9KfT8W6!iyq1`g8%g7puT(}#kGKSp3^3tfZ#0>)v}iH;Wu^t<fVJ-*XyjYl
z1-yplg{=-YXjHJp>^4NN_AdZCGXpMkHw5S!AxHTJ=kge3$oQ>NEc}G0Fy=be7reY6
zG$kR%fbOp7tv=N|>sY+CM28sf!m_R5e7{QIg2%`#ExOs$4R33MT6UA$e3~O;00r>;
z!cWBYZf{CL9GU`c)R!>o8^rQJ^A@AyQl5EG504_uBGXSzUF(myla7l;8idKOKpJVA
z8#o1m&3WhSH2FGhDCEE9BYjg?VW#NOsOzPo!0aE&C4%11%Yxv;oS@7|I5-DLe_#m|
z3!A<?m)k9XDjRNpK-xph|8^9zF#sti?9^J_TE*@zH7&MiiVB^tHV;a+wUshRyI~Wg
z2X(WWn9QWZo}=ns(b82+ezN8Z_|ZhfZ31MLQu)rXJL+?@>vb}3ijO6@by|GfT;Ld&
zfxfwGNtgdB<iwX+mPwk|-)G>nj<T{%CfH^RxYrVKUyzq=aPw^)MVyxzq^rrctcB;v
ziQ<$uS3*Xz0?p=9N#E&v<N5<S;9XX0oL5u&G~3#s)FqlN*%1r%7f@;wBAGy9GfVWX
zy*ZBy3_;bR7j8}dwBdYRo(?R-1*V@O0Y}1K*ppiA0}`<;U#<mO?VgizVwD5isd!sQ
z@#9>L;)i4$Ug&8VwJHW%Kkv$8hQ`3B+Wax0elm8zsWwe#KFMHDL;zgyM`qaLf2<!~
zfP1)_kw(+sYn(6gpZ44}8K{$OwJ1Ef!T{8WU7Tp*#;y0m&+xripgb>o%TxMqfR@KL
zF1#yx_Z`={>89M@s~mufOymsTuX{>&qt<mETb-sT8Xslay(k5W+|(A^{SEZwsdmO!
zDSGv=Aw@<g@?SyBNUr3A$<DLoFwzhgwh08z@!tMEKu5Ohjuvvk-c@{JRfEI&C;|vp
zjrYwhb27cX54mD~(4R(U4++u$d+Wj1)#s3$O?Rd?{KZdmad^s}Jdu7Pr#+c7*Gi%!
zI9ch{Y#>PnO*D1m<&%`Z0IC3fjSpRXINhf!;D%*lR-(L+(!My{xp~Q-X!t|N|6aGT
z4pm8_XO6wb9y?9&XSaRYJF<1gn;;;XbCXk4e?9(VImnE5eLsr%_xy1QBxwH)CMWbW
z@0N{rD(N>R^yKkp>wb~?8BU(Av8?c3QwVU%I@&Mg*zt}lPdXuXE;ij>n>;pTh|poj
z=+i5!vUQi(v<mhLdlo|~AaMh{u*iSEy_ah4mq-U;Q4oa@o#DhzPg|G5gl@5R#`jCL
zn%Rguh~3MYH+#lgJGsg?V4te7Lum#h%%6|PN1oqQaDTiJ#t{B|(=dJ(B!$lT{R+e*
z6VnT~3VaPE_q#Lx$2xG}x&S;;$;kRRa1is#HuwG3d^uW`SWQZJz>8&e3R_c<zoho<
z?T@Gj42f<}$Gx)v=ftN0KjR(@o6B73!JflE(r4vi&Wipt5Gs<Iz+a67+~V6LEmt>S
zM4$TkRB+_YF(29|&d17~OIK4uyV?S<B3ped5(Z1K;t^fi8!UQc;%ACzT;~fp-1VnT
zP+-DlW!nm!g|5$GPBKfE)Rc%WUQrzc-JePPWe2)n`i699l+Xf-@{rsgb$RBPJ;1N{
zH9F<$e1UpFZT@@%U>d%QId*|TSjb+d<oQ6`uVqi2qTfg*MLBT1a^fE(Fv_zAPgOOA
z1%e!C+mg;?N**0I<_gVvY#jza3mDVU766ar&%#U{spf<wi`{f;XA=&r6ie3$P2ZqI
zu6{I2=Bp0Q;w8b8X3jncQf%<{<2`ODge);~U)_apkV>=@$V^_8znxaK9VSJT+*2i`
zx7tzmpm18aFsculLryL;%-Y;hLa&SJBuV67y?uI<@&h#CSZ9AX3v7JSbTUFUlG$)n
z=Y?Zu?Ccv*xRgr5%X!)D&k54XaQ9pZQ@_}aLP^bU(VyZ`iCNG_35w(m8-38t<UY)r
zZux5J|Ep{VQaVKQi~jbUZuZ1D7F|_n&bO3?2v;bbChuf9g`vSIVOMnd_+cv0f_EV1
zL|A&)p4b0Gmgki{OxdkpGP|<o>6mM*X$=kjQN-S`5*0X6Qch^lli%d$zPdJO?QiBF
zN*Wpslq}$p84+l;oQ^96f6}u<A(+bR?CqIZVX@-?B1r)@sWWG$dq@$z@Ckn$7hEk8
zZOqD0M>&k;K|OU<OQZ{uqs}R2ex>@yp+=Q>?qa7`I4)JBl9J5>c_a$ls1qe0@G>=o
z_8Y=K{M$)<TA&3gkZ`n0M7*LnGr7ew7x*_z*JGv${SjOmC!b6$dB0~6h%jm%%$zX@
zTm>3XDB|6t;`ttHnFSk&gTXGLMA8%WALSCr_d703U}(4J#;ifWAbyAJl8+f6VXwoG
zv@q;I24nQ;#yQd@P2HDfH|-XQ!%T?1(~^y}$gx!LVK?RRKa+cmRn*xMA$ycJ3i05T
zq&WB>j>_DAxI~-ZvW3%Z7a6r)+7=ABO1>>mgbS&gnS)}_56qZ@x!#74bV3SX;z9dq
z7wWYt^?tvNrKsuDM+w(hh`$6W^=i<(Vy&H*G)aDVn^jSb`2+daDOPgdVE+l4(|Q@Z
zrX7GMLk{c`3N<D*CP}Js1;McPi(p3g8m)nY;Np>`=sN=9od-lBt0TppB;p6KtMAoC
z_2$UF?}M5ColLWT3eI+?=^8g`h1N*$bEB<C&6d>|0UFM$jX7!{X2gCZdQUJMmLp_d
zQ`iW5y;I-`>Ri~Agq4rebrMFklF0ySsEa8rol#8d-7I0-)0HmjB(F=04SbdS@kR{V
zBzSLxZF$P-)f0Fwer5Mr;yi}BojS6iD5b3be)LD18I$QABlXDzj8jr8lw7RbB<>j1
z3efdmwvC`4;6{$!LxN0!F4cs)AetF9hHo>M@mf*wnZ1ov4xPf6Y6#y}$H=}2|7=og
zK|Uy5;kC!~$7!moD;OovM>q6ad90C>_`THKn;Zs&Y&*WcPg-N0r6i`=#7Kyrz;@4u
zDwjbTh%&nz`IHTnD)!AvI<2!xrYt|L%S6vBQu`S$5<FQMv(VLf`cJNh{i>&~xUI_)
zaFryjp47=0!BncB;3<(wh!{VKcbL*-cNKQG`5jksNsW94NyMFs?*=Btoh((Z;&$#f
zEch75&G>PcFFSyOf_vt}On+h{@|Gdy6=Ts|(s$A49OiQc7`sPQDJU}??Qn+CId@FW
za|6Q#n<-80asv}oj+v;2vl|nTFfaC1W`ML!{oLe2`Q9Kv1MId0D^&~y7Joz`<Lm^$
zO0862@68n$@qGoZCq}*(v<>AvQ_Lt9f{$Z+UU5NO^ddDQ+(3Ws7XFQLO8xB)h6|Kt
zI3i(2;n;z;$=!W;);HuuD8K)M$RfuQ&1<WAFV`b8xPCGFY|hVmrbCDc^|!4r>v=hf
zw_HtX^hZ3-D?}#wV2yQ#nqv<nc*@>me1*>wo9@dY$cS%)UfTY2quLHNyf(ZRk$Sdi
z`2r`VCSNw{`!WMAc;tL0<Nu!Uz?37IR590lU351r;QKF@R2Ey!e{u^%a3AOoSWZda
zY`$C44-nau&N)vl!D_q$h8iXkSRoyX7Ala4<Z^}eu&}{>_OHlG(N#O&g?=zI40$Mj
z&DtQ1DST5J=s|)(gAI@&AiduC$*9RX^=_p8HN}3iI0U-qou=1YJgv1dQHBrdw!;0?
z=#+J|6I1=Q#%l(p(14W-pcXU_WIx?EuYonLNgnTAO^x*cBLaeo=9}zR1v>aybObD5
zXOz(9N3m=>?Y5gHWLW9w(0VuQSw2Y-vP;WX9(_nR98|e#v3pxN<LV{z6r*crMnz#|
ztk4HWna`Om=9#s;BbTmu6?Ie2x?dmeGEA$`8xi6P(~uS^TebO>ttW#!uC~W2A`yla
zYj(S;AH|=E(Fwlpe)rihg>)4TPz~Tns&H8@OT$yOQxU0Pki(FFp9!h&+et9ImM_)L
zCbL+xe8QyPqlj!H=y}G`vQ`4I6fM=p6O0&$?347)3PSpc0vx9n7$P>b-^FgSn1!Iq
zKwpc7X2n1|$()F+99^lZr4vTaPe))IwY>gCuz;9E>RP7>5si(E;2r#l(dR&vnySCc
zU9)Nw=!0*3NdFeQk1=jmlq%n_032;PNvtWZ|In9})MFzIUb@42awKh}RTZ;(rG*5R
zCQN$@ii?q0-S-r>X}||*ZdEDC!@pQ{$hmA)I>1XJm)h52I@XTFA%!^aAhU67w0WpQ
z&xf9DLcWPtn8xL-KaVnkkzfTn8CncVJe}7j7v&u>ym~!F%qgUQ({oH3cn7)V%j^ae
zGwGS&&6k2xj_uk4))!Fdy29JOOt(OWcar@VWxM(zP$_vAQP<}lyg!MX9V?{=y><+$
znPs=El&E^b+O|4AwtCw-+f8Em`Wo6jc)}vq>3P8kl|@xx2pSZmj{j|KM?SwJpL&_E
z{|*$Kp??v_h$m#+8Wh?fRKrlF$9K3<OGu24QaYS9xUI8HZcbiS>NvBRcLTebn{*;F
zrP>v^QIW3sIauY)*0qQXg8cpxpfD348e&C{hiPgu3M2=y^GzBRyRUk~bKnS|^uQf~
zP!v3u*(5TU)CG-~L*1hjY*1$GrRhp(2WdYD-g0qOBunvR<WLp9XfIV#V!8Mag}$oh
z97HLXKC2t|9<P6`jzvL(-+tRqiue^y$o8v2Wg0=w!cFfhFNj=LOoIe7QSS^bVnh&2
z%q6cCnwN4033puS#%N3iN|lUGc+tMxOHS?%_axX}H`rMY;~8z@u7#iKExU!&7@)hx
zZ3xOX$$5?Ob!qW={jhSb#{b7t1n!~`=TD460Ku!uF9P~*=7a*w^Mnx((I=)7W-EOG
zi%L_1tF&R>`;!CZTxgR7MtrkJV>~asAK*uURxl(oUJxFHiKZ#Jxi(%#7?4kb4HqC=
z#M%_knV;;~?Ygyr*k#fb|2-|BVYb90^x+9oV{u(_Y}&@$zjzi-n|+M$E;)tBs#aan
z<g_S1_J8{>_b36mFGtudcC=}XRqU>>p?VK1w!18Lb+>n@FTX>p6kPM&d&CrHxDjIm
za$HL`W4LDIi{CqMw>h<5{H@yCk<#&BNLqg&Ny;xj?mIuYpTrY@2w5b>R@$*c>OO^f
zG3PvbeNpc$yang%XJ?~M6MMUSZ|%QQ(V8oj+{>Tx;irJC&C1zPWj2)}aFl;%;ESQ&
z@iR2cO&wURjEIS6l`7p+*C)03rqP0OEM}Rjd9}A(u|{~)mf-|BL%onlS}eHm@e9;b
zQ5SVmXve{s>5*m;lA#0BK$mQ*C}cS~P07D1wuvzb(r24`i%CaBgxWD2m=uo&y7ST^
z!KDf2c?Ezt+(t(q0(an(_^{jlLU;%4k~%9ca?%ES|CGF=0SgE{<%B3DE%f>;j+?&u
zoJZoO{u;i=LW-!oAa<H)uud|thteM;h2C5x#tXv`<?nEvDI%k^BCge%hjp$^`d98)
zVg)PH(XDu{(vYp2PjS>ZYORPr^FC-&KX45z29|-~`z7^Rs4tghw%wn>y#T$4mZ#ad
zqsFbjmCG5)!yrB170<&$Hhz1?&f0u7AkleiO5s-d*{E{kT@>%>m2F_Jk|`w{vw;Fe
z8#4F_za4Dy+lCAn@BLMdzIuMqtjfj(#MaacT4%M7M0UYpsLLQ?i0L!4+edRf&Up93
zR8jzu*isPa90+~R4@Tc^4I0s~>4j8MTuY4Zj1b_M*AFC`9+hbAK*bB1E_`~O$)|R7
z1ek6f!P~ux4&tyMLPeaX^%D?fBV{5vFFM%gJs6pOgRR)J6oiQAvK>3%I!|maEu$=Q
zyf0m8Y+%cj*>D_NHao6y3bO@^v2{P9!Wsfm7@2gp-Tg!H%UXVt5W4(y!>CV0yqF5^
z**hdUpyUw72bF*Zl18D3GeiM6jBfP#;4nllijD&Z9t4DWzqe3&{;oe4C}}uzhfeva
zq6Xpk{p*9RnhqL6>rQ^qN&)qkm--X(3C7;&sUjo_qP9_tE=nlb()nj#V;1-Xs<^;?
zd4g!cg52n0smHn>iSTiv<7++N{Sh`H>Fl<Oe{mG|^wtTlO-Jf|L|6&Si-j`i1lGd?
zZk7^(H8@?Z-u})O^4V)iCnc{#h0TpucNdspTTf}SmR5W!l&-V*lst3w-y-n63fkuU
zkN;$@D~(cHq~jyxJhOltSWk*u4<Lm(Gd(&Tjg2{T{%{cHVoD`l%kTp<8-}~bl4tDe
zdUJ4@(w_Wm)xKf9XDEYCN~Q#hjQ{;ex5GHLdCDb<ZVsze{Wb45;Llj1uw3;nF_5T>
z!{4_=f7%aUU^alGK-QZ;my;Re@R)ZdSeWiYVxXq-PZq`h*2M8~ba-E}fGd^=s13PY
z_$nn1s<c0DJ<+Xm-2I`?3t0JtLxvsBPud9AuI9ev#P2K%u5D7k4J0(RtSD0IXws52
z1SyqRr_%#^m?_9I&#akJcqlpqv5#us97COdS}0n6@HQC!7CQ_U!`CG~S?FOQyq%^4
zE{0{++Re&0=AYGB7V@nEYSOTV%$6KdZ65X8HnjtgV>3o;Zi@2zeqpCNOIU}N4bH*T
z4$BA*zCLUrRX>X*D9%RdOY^nX99CmY-6ksArr;)vzcn9~@J0ByY&aa2&8z=T$2u&G
zK%+(WDzP1Kve(S`Z}`K@W=n7v$hu$Cl|B(yceCikWZz`q&Sn6`X56upx}Rt=$sqjT
zX=gbtQ;S#TgT2q2#R8&QG1mK(Zr;4Z8U*jgS7`Vm#_Y*Qo9)brW;B!y=dkAZWXNuD
zMUEg|SCHvAUK$=LuIe4txhkd@3c@oNpp(~zWZm1qGup0AVsd}7Im{1<p<~Hz<(<&w
z{zXqbY*4VbW$*-2dq3y1LEO8^LK6EC{jEvJc@df8DJeT?evO}`fT`5nr-lj)zwLX9
zg^qR1nQ}~Vk~h2Sop#&gC+=<3+cr*J@yYDYDM>g48eRYR9PaUCK*i%%9Q*PW^relH
zA1as`V!69W834jrA?&vY^7^m3Wt(X9l&zdO6w7qyP8<S}jOFwEXUEX)hw}Y2Q!K`Z
zlYh4!eM48c5fYA<64gB^t>Yrswj~vAUoEo^!tgcT$utm^<N=lST#0&rtoS^NFg2xE
z`f2+J@wuQ954t1<B<avlhXx-B>!^f!+KlG*1rvIrRlI010okJo^jLib3oRWFQZkNR
zEe}$-!BSu{<<`I0-nXG)@kRx1i<~KjnPAL+pjlNRn1<0Om;CZ56>dCB`cJf%KqI-+
zV6kP&&;IHd18@?uwFJ)N6?4qL^l`{T_;xZ4l;rzy2ob`|$flY@zHF49xex4G1hM6o
zsLS+A2EX^VHS<l)D)df}H!}MwA8E*WCpn1&mpXw%y*~xL!m@=juAm|LLt+{ax~NyB
zrq%yOcRDuaWpINr>3IHA?6P+ttnP8sZtAHX^L^&-ciQveW|+BYXNo22LX`(U(Z%N+
zHH$h}5fwK2vFToc?apd{tk8(uS8=J(^W3I!p}hK8en(&SF7;O9Y)gH%M`<cl5QC@W
z2`B*y5<=se<ZQtGVNT&3u37o3fTSPOQ#$pFBI{)Zr}c{?3*tKWo8$E*Wn$AepRdpP
zs;40B{{1f`9D+Juaz+%4lUjiwR1+!-+-bAsMX%j$s(P(haWE+l0|dgIc{R?ik}cuG
z=+^{4)qnvtTekVKoIkX`J4%{uM58N8+IRd4u`S4g4M<~_LK7n@8t+7iq~K5cpYA?A
zB07k*^K5UpVI|Q%h%p4t0AX0nWPmeZZe7p}+zrGTfP!{nO^ilLwl_VE-EJFlE_2|k
zAn{ke9ejt3&^sQS#Yq}z>~eY4ZdHx`Mi6YGxS;T~NTHTvSR@f2htwf&Jd5*h_+L<9
zM8kXDSGnvxQWO*?nbESVIdyK}Pcta?+4{pc%?AVcul>gj;J+s`P}7i^otax$=>tj@
zf`%eZk<Q;Qs}GgPFz-aYPee}*AQEi`s(HwPhQubpP9pe^<FYG8MA=NGv(k9SK8)Y!
z!2s&}kyN^BZaDOmEy(!ndB$a5_5~J@2_G3o#VfYRECzcw3zdC~qnz29kV^m#vVf)B
zNePI}v*j$LZr2vCv!;lyJn^{U0RNoKAxj)H$LjHbvL6BT|K}`@FzG!m(q@_c&m#*f
zbPFcF`R|=V;Y?))*8Bxff*Wf>FNgsoBen8q8<vpIr6W?7BZUd0R>@-EH(?jBUc31^
zNv_jDHiIJ-ZZBNnpaL=T!nyXT;>>9m(RWg%(_^uxyQ5oaV|up`qc$8odQi-A$m3zF
zg~Ib?B&ve$Ql|A=waPuKy&iG<YrO|u!~83!oR?vkq?MU03n$Ezxk+q<Hz`T{wXI`K
zbEp%0_N{w&A*Ku^z1YVU3RMl*)8UP;`F{Y}`%ua3lA_B#`NYPB&vCF~ppUII$<`h*
zTT=`=0BifT;oXc^u1^{q*BcbgzAW-;?}xh-bj|S$MwQnp4WyT1C3~tT$9Lb9Ml^Vd
zC_R*Tr$MW7*zdKWY!)2xjy&z%try`Ry?su%G3qVKuj+`DcqqUfBakK9(8I*S><UQV
z7^jr?WO+?V1YAwJ9`!$l;X=V!TqZTbt4-7|7~WluaDJr4#n=J>*DAoke%vl@_Q+$i
zqt7D{&rOZP#&Nl)qtkMWI<?5b4bFjUdM;wBp=eXwo5AYPk-F}0nzp1SJdPcS;!|6n
zQb<I5+l(RvWh>DBsA2UIsyQr7wJc9il8#km?sCYZsceH0S0^WG@9BYy?@}rBDXYXn
zfd`pH_v8jNr)^bla}x1J`<{KYy`DiP;S}Sl;06QEh~=!|_z&!lI$waL8V1d*Z|<aM
z2JGsS4%u8pDl$oA2G*2^d_0Ki0!Z>-ksWjB{DxJ6jZk2m9$%)fD?1-kKZqhtnWP=a
zX`QANCSI950ET$GsTO$bUjPZ|B<Of>B`sm0y^}Daho`n09cg89-jz=v9<PX4+T7Mp
zg_c24P?>-s8b6mUM67>6#wRE_o)4%9X3Haatf8erWOhTIa=Qf=jDnW4GbX*g1S?d<
zOImBb?gi>n5jZYp43k{hB};RtGDoJ`!IH)|Y63Jqo6W-o2<xZ+GK`YJyssO`>=&r=
zbRAfW7?HwY{wH4!dub~O`PZE<L0kBtWcAH2Q9f-xvB)~(l*=deid45QzJXJ<;+S(|
z?6{wW`p7y|amUIWd*jPQL2IqPJ@kL%KyTw?0<(#bl*Ft;fWsXLog?5VcLptJF{emK
zlqF*NxmWUeGG-PC>N{)?YAgU7t-v>lX<{1x*;z&l28lNcW|`2gv=wybe3P(w4!ssz
z@sJE{7TFyq@`f?5dAf+I=c7yY)hyoBZE!{Y?ZQ8o33sg&)Za<Qu-SC^Jx!n~pDj|K
zz#t*yo>K*vI|{oY#-FE%@fRpmjw|7jEtDa_lD54%dfT@!;f`JsTYls8YzOJdklS|+
zd5dg6ZS`MguZTFF1jmuQ94W@XhW`Cke-|6cm$Dw6n9W?%I1S~hZ=Zy-wI*Bv(NueS
zpAxaz)zeF4Sf0kRt*W1uDVVBH`K9R^4P^Bwf^?=`&3%UsN}7k=1F5&k!tv6QhJ?so
z*ZO~i(37ltbCr3UjV6|e<v$-ajrCvqArXdu#ofkn9Kc)G10XCHm?Zz{howDBvOpog
zf}kF+?48r{ai9;Jvqrze(Zv70xG}(5tWOKW5QQVw2k;#Ec>~iB=L6TtE7%_ljNgbz
z$};+#Q-()V6-tg{6xzRn45JMR;4t~aPoj{2P8>k_nVMj;9BJuIAAaTogly0vaD3gB
zFVfKi<!im(5gh{C_S(2YFZc|i$4~;#qe@Rq`&&f0#JP@cIH7gNC`W~q^YIJuCuv7{
zdo}olkbc6G;{eIIj0o48Sdc{xRf|Hygoc7h#aBC&hGg`_g=C6#JFB{i`^gL&0@E}U
zkZyHNWCveZ)I3XU!7^V-TmCZ&=-kk&qtvI+auE;z(xn>kL=TlqUbtYtGAc{vG$jT%
zWw;*BW$YkQA6pAniJ3l=pji@Mc<D2mkz995AA&lIAum(^C7Y3jkAez?7>9cNx1;Dq
z+`8~h3p94=XL$Z&66QIzPPjp8Y!sL|@IP0(=u&<KPHCLkv($fmU4z+xgsKlU&zlRn
zEfbjiMb1&*SdcyU)o(+q*-_80m{>a8s^~cNU!1qyJYOHKtq|hGpd$&tr4JL0@JR4g
zmX(Nfbaaa4%t9<CJGw9CiO)kmBy&taKrhd+bJ}P+QOfdQ($Yf$=qrn2wvd2ct+T2Y
z0hj6A6O{S8%zi&XiEmv;KiS$&C0kohDx@=wdqKkFI=>!M!#{Fd6nl&9K40j(i&H~6
zW@r)5(chWB3ctHci;G>k4fm>&LG{!Rde!WGi`UdxG16Pa>2xed0;_(#p{v))jEsA&
zL@ja|A8fRCg@5T3^p2qebftlr0%>QGEzI>Av}QWI>if~;cG4!r7$6xWv2i?<M`;kf
zK74f+gz|H6F3u4P;NmM=Lu0;4mnP(0!N!i^4nb9D%Gu4=qi>0pxZIZ^s=}o8CQpTy
z^O-p(vIp|>$wPTf%82V(bt@K4)!2*WX7BU%DzrwJ*ej)=M<IVu(bIu_xa@JK2Al4>
z+0QvSQp?L|kMmazk84+FvVl{8=HQwE8<&WF-vBhuusrwDF#B5uE%Izm9VL}Jc-gTD
z4AK!uG|RSh=zR_tx=2Z8=l$Q-yrc#Q;R?Fd9DgX?)lki}!zziLfBt|#Aps2}La0hY
zwc`ji?9{w;;SzTH#To@%=a${#_4Q+Sb8be6fWyop)wfRbXcoUm8CZ9$C5H9dTGr&+
zxt-A;?T>LX8x=CwhP||t1et`3NhK9zrKmxYugfk?pgzdfg?eaOo_!cNCdysjSgnP8
zh~&-gHhZpT#mK>1Pae(|0{A&fa@vF$dLgPFN>y)Ii4V4{L9+qh(5Gl^I^x+a3_oDA
z$<EhlPK}gG%GXF?Wr*s_7ses^Zr@blrnv8a-oT^TD48(Oap;HDzK2Ls^=y%ATO-h9
zSZmPGH`HMmh4g#)-LH*xQhPilY^t9QFp$dZOwSrG4tU^w=?6+}owY7Cd$jBCytCeZ
z={XJcMBkrK7A*j~A!R2QZ3&GZyYspwtzOoe1;%TGwe#bG?z)3v_2Wl9u2$VI`!icU
z(3BdmNPkS_N8L|`)Y`hpqA7A#L-Dw!(&+VX3uovD>aE6Y{0jlXOeGL12MZaF7hdQm
zo+);;y<jG4R;uAqGnIoJ2mMsvZ6kmglhdfv|9d(GQ|1ZuERW-HMg_C~g{2MlhRlLs
zaW|#uH}><$DM^p;j+3Xc&hUind<Sr42B9@V)eF|8Q(QdnMBT)et0b7#!bzodc5Un{
zQU%~Icnb~ktFM^cu>4n7KX8D*&$vQGP(r0Yc&OIIqyYK-kd)m_9jYbdlXe*_Lg6O~
zQK|Y$0)ae`pZP)^3RU7)V$o4K)pwzQ+$rCB=%1lQSAXDCop3CD)KN*-nvILd!X-Im
z56fZjm6~B1V_$;)CcvafAPK{WJ5jcym|w&i)$u4x&jIu2b=+N*`I<19$mmHgW~~fw
zaya7>C82FsV6qKBttG>H9lR>GE?gq_0M<5B{c{A!-v;9wj_5HS=iSlX{U%Z$O9&ZC
zxI>hg$GPxB%s7o^zA5{IqJWa)0NdI@hFM3ki<9kR&Pus++upUMUz}1kyF0tF=FNPb
zm$r4?fV{luYF%b{6Xrrdf<uqw{Pl5qnv$lu-azh9_ILuB#FE}5fw}<NMKNRIsP$f(
zSuGEE;^l__tC-OWHMjU7ejhk@b5<B+EAZ=MZym{NzeCCQw=_sc1N&hd?Z@q}GlXJ>
zgtu*0s68=fU>^hY(nPgTQ0KyyfFAw*oXDgym_LtJzsvkVje8)61;kH=zx0W`az*{w
z`}o>FVJq*G(&x;j$3Ym%gC?~qMpBv){HXhY7{g2aumkk#LN#FmR3?&`-NvV#v<_q0
z)Z;SyaXjP3d=6<){vJ+2c)yw%J%vl61WCzZLAcopq3CH_z)8T%Rm>S+S!VO5UQrn@
z1&(T#Piu3581rgz?iu7YG55Ab=g!15dAT$3wz#|Z7$9Dc(R@EL3<p~kSKj5m9UJ<H
z*EsxyFXR%XW05mTk*UuKV9Y=lOtFiUO5fuWpqC3z(p&b|j%SJ?<cqhU5t&l|Y~eAI
zPL<TORIvi2(nuk7qqR{>0@P<gZmmx}?+9tTkW5bI#!YwiZ_$V<i(;|ViE1Bpas|%n
z?kZ~?DlV+Tz1+!mrqFH3m<TQbv(R!XH~1?A-dCZ)2Q?2iVNBnFk}6<6bqk$G0R3Ih
z2Z2$xHxlV6n?wxIaC)G&7SXS3cEOc1q+q+Oc_TUuK@20{xrwGYn^9WMPOn3siuA6X
zD;wl|t9jbG<?K<;WZSv}>vc~&K21FFyHdG2r)n+4qWxYtQ1V#$j2UZopeoLN*vin$
z_AK&eU!@6wX9Urf#}@F47nG%zO-|`}hgFH0h!iErG`-mvg!tz078K@F@Q=h1YkJio
zdsTM!JS0#eQwspNzc3s5bMu5DjdmftrVK$50l<TG5J3s|p%0M!sfC%tZ$4W)!llj`
zL#O@wiFD`DhK6242SrnsnN%7R5_7F=$^AkVBUVoX=-e}mS_cHL=N?N5rX-gLmd=n0
zNN5Iw%sFuT{oJA{jk2i#%RXY>e|l;vUvEo>Zq-oDZxDG!ZJ^<Cz?q^BF?aS=#~)Pd
zN>g*zQd!M187!zs{zLOJoO=qIl$c|=<2(+{jo)QZ^!{8=0H{i<*ZmkvdzZg(x<XT+
zz9-4;b**|UvH;BY2uHE8rnygM0KR|<kD#r_<#{GUA8`o<!w&6a6x}7A-i18EL=zGr
z%LG@)+*dK}_Wl;2lG!WZ^-5fGpC!2YHpe<a)4rmVt6ndo!CP~>-B_A0^`77_Hd>5#
zjaII`dR~)LEMrN}T)BI+M`OjWq)GHgimY1Jq%blBlmm~)mIm9SVDQDej#*r(Nq)+3
z!iagJQG{myfgwOFuX;Z~Ld>CuJ03P0#5Y)sR`M+wput~20hyBEo|XXfBUWA~`qmg=
zf|wHHL4qJagb0p*&AsZYr-_$3Yu)}inm?@gZ6$*0K@8CTya`0Cid{Ql;plTe#X<xD
z^`isX-vw%UgA(|`2?h%d1FX4c?p(5mQ(rul^c>WW?~rnqZTARoYvaUW(dB7NA2B6@
zAg?p+Y07W4dv_`F<r)=fSGh>X7MA8?)~mZ<sfqbMo-+ZLV)9N;O|5L3i<)f?9K8Zv
zI1#II$EQ{f+0_5W9u})bwj<=f+mNcP09658Xu!|>bmZ1|_%UAMvOhIEGkBY&*3g32
z%H~;sKoG&h>F9sp&yC%8wMmshM?(`pr^NT)A|RaV!~TrB+7&+P2R@WGbax!dJ#;ki
z%cj%MRDsU~`t!bYj*QiHRHSD{B^&iz0m|jg-p{*FhQk*tzAHET9w8Mr`%>#Zq|YhP
zgTNPY%Bii528%?V<#~^qfenv&nvwghkGbC&%i_%%dU0d?CGje+epP7y*FC<2RB`Io
z+mifr&9eOoOlRQxVPt&rLt_RD<y{^_1UL{x=$f6pE~Ap|t+s@kp(U#2O%#YJIQ}LH
zkob-rwXTHw06(wngB<-?5&>lB)y@ftHu>$Qj~ZQ1Gwqm3-pHZ8RK0X&y`YGz(PRd5
z#^R4mgSp-iJg&(hs---b-@*X94EKEbzCG<{v(#C3ELC*@;^&sMLmQo=wwf+p%M|cl
zB@9<lWI<VA(`Ms>cPQ>CpJ9ilycMmMN<q}+xeMAwfl^z+zIbpfMgaV@HGmGqs|CoN
z!f?FAKk|!dlSFXH>A2$j7$))LCrq1oga}t7(xW$&|AQT|wK&h~PucDkAdJL;NMurn
z4=KjO%JzO>_!6(LKiU;vF$N%<2RGa#*S=EDM2zwD(&pWGn&S3)MmjIPKI3o+k*LsS
znHuMLJJV@p+(8H3MFx(^Lj`zc=7cvQvDwn+FPy)+VvkiQ-5(+goNx*CMh1SWeBC_>
ziat~v#!v-x{$1RDMBKKzh($jxn#T~$5R9TaA<*9ONqQR;u@RC$8@gkWMQZVLlZ1Fg
zEG%zl_3aFOO%y9MKqM0bm<&_I{%H!{+WA$YdJ1O}qr8Pp_70pM3Yw5em8lFvK<6{q
zctz{(`napX<5p5#i^Yz2U(k^`K-BRP=VDe9Z72C~O&9sOAli=fT5d7o#!Hi=7=pSK
zPcNfxXmDrWBy<mOZpxT8rLQo&pqf(<=1^KTtt?{JEMhfrVlm?@SFviOuc)<Bmo~h1
zMC4DuOkVO;)dmJm63sjsHC_!#n-qBg6v~1sAvepVLN8|GOl88#SEcZuAul^rBReXt
zW1()3BbzX#z5WSUu9y8F)4hhxEa4uYOIx>D(uYloCZ7|;-9db-9T{|;H=E0S%#ht{
z%GL#r{M5mkSf-J<A+#+F;n|gu>GET<UL4-=1`L)62LkU?5o4-{95FMY=_J+y5Xr1Q
z|3F>MEJV>A>f#eSdD}&v7|?ocuiX}sFBN~>Us)2n%v09iBi||p$N(`FZbD>$A{Fr4
zfHL6ks=s%klx%c1!q+vsIJ(;pb_#a(_o#>hl9K70?gq5`kBH?{udS2a3{!Aiq#NtZ
zeiBvQo50dW%&L2(<JTgAq1fCSJ-dmm1Kmavsh;YFzMk8(zg))oyEJN6^f|7Mu8**1
z082DAh)buaNFM4#i!+?;!Ys~<gNVt;_D#wML9b;@%S!ss_9nH6d>rMbD3@;CbZ`2#
zf2JeQxGbp-S&7GOyzoY%=fZl?BrrBC`#WtUC%^*Q=4%GV+P73_+#zxKvQ=*$X(5o*
zWRMz?6;JvcxMs|Fe`|TilMg9#`?d9f8+MXp58hCLe|ZPSTnNbX0G9QD;Lz%=0a9z4
zt>xXxAkk~iU+aeGC-+M0s)W+c_UeRRY2!`E<pzu6$1@1^1IM!itF>ZDlUR<S45ED-
zML-1A1}=lJ25rgu-Fu_KtO?v#u2op?DznO_i&-JPDP#0=*PI?TY7AGz!^3GX=$Hbh
z7vm@ka<<J#J?jo4&l{vB>fjAFN<V6iHPhwCNfP9DoD)6B>$UoaGql6ef&~+e&V9#e
z;UK;U0FeSx41TfTL2nxR<su^9w9MaM1i(F1PkT#Y>e2pup_eJ;n;?MxtRoEK985?^
z_!pm=Y*zqsOQ?)!0)3RbI|89MBpC$Br26XP&IBRf>A+N*6T6qiEFckfwBMAqY^$Mf
zh=&haM#NNhFrUP<RjnHEO3N032=m`)4Tvsf9dtqZ_|8S@gZvXvY>voU3)h6lvuugs
zAgI-Ua~nV65wo=?Lf}O<ju^4gy-}XDRc@ZCkPz}1{*gw6US$(Eg8ZW*j-?)BehJ&4
zW4Vhdbs-B(1e84$GaHNyh=ZJhvHq0ud05o{$JDRp%^A>IDjFwyteXR4Hrk6dIxt>+
zhHZs8(|PVD(K*M6{?3Os25lG58E5`1;p~!0JC{pTYisv?4ju#wO2?rk&%RvQ+fElp
z8qD;Z9c$kOZHuJV`b~ot+0A?Ej>GK3w9!vXg@x-(@EZStwK!Yz{d?WO!Xwucl`iVM
zKlHI9%PTTbE;hMby44n;J5)SAW~-i3SI<KyB4jco@H@E)orlqQfz50PDOqbcfK6>|
z2Kx<U!~#9D`t9~Qh`Mz83;sc;nJC`2ilAx?Q=FIg2$Re}NeGpxI1v024Rm5(^bhFa
z?4A)>bUNNuU3?)T0up^Flyp1`8-B(dB9w5YLM{Ac#Ono!KT&$EYO%hfmM=vn=FH|x
zX!J~~7n+oBg{JeJ>ZggG{&20k8#J6)Q&%R?mK_7G{Sp1}-KcWYHX=EdS`4M(&enLz
zL3bs08?mDKV_}Sqvgd#Cb<WX|b>W_lZQEw2V<#2cwr#Uw+wK^hq+{E5$F_}*I+=Xm
zy?18r%&hsRR;@a<&)%!*oU`Be`MqdbgM}TmDucYt6hH8D$`vqhy_bJj`Ib&ctfFS^
zJ6IwUggv~GH+2aS;_Ji=^4#2CAZb9J=M(UV%5i&=%-3g!KC!(54c^<+Aa2y*OtHcJ
zl7qU*&=xTQ%n6kY>7nR3>B>Go&)&8ggI)%#FW54hQ=mr@WpB)al`Vq(a2<1Gmr6x)
zdMUaqA)`MvAEMT9jTVULsUxc=K~(}}U!TgtSWHPBg&Zu<6^LjF1lye)baE-QQ(&9p
znMlDs`#?OC=D)r2j_^mXxu&3a>WF3$J1M=s^qNQpLNHHvpU^554USa#R4R<8SZEwQ
zMCakj5Pj%_*|Io?2AE)lON<q}VucUrU)3!#nQFLLI&q~$5Te4&r9{l#3Dcm1=UnuU
z)mz<9s+QHiyf58@XFk&UxX&Kgc?=3_7a-2#Dypv?6S{0-McM~I1wcVqgH)%2VS-?e
z*1~-QKJ25)S$Uby6f2`^wk_Nx9Ks0lP0|w0ocxC4=<DyK#y&>(zTM(LxO`;cD2F|^
zb`Hcw@47+$Nrgq&*hoZ2%dD}%w=%=J;5mEK7;qBG-U7%TW@KW#D$_Wb5{4^3;M{T&
zZ2y(|Ui)ORziXgbhs!S99BN#;Xv<-vgqRfv465}^I-<7?ywC$A)@rBr5a=Y-XAyr&
z_UD$5*XcWmBcznNj$H+L638{5<Tq@f*A0b9c{t};tl%ujR3zoy(PY;fo4DYs0V-GR
zA_<gI>COAH6!%YKH2D8uJz>rI9Jitm*Lbwo@*erhB+?%e9Q%06T`{Ea%I4}wRliCC
zr!vZ;s!*v<Qyh|Qz<8}ux}BUSJzDhB*WL-UV`hzC?!t+`UFNaPGcNNU4k|ZVWt`H^
zntc=r$>@g#cf`IbO%Gu$gD%75{Vj*Qo1?IAA(-Nivx=<)u-L|xjx_#Krz2W_7p-Vt
z|M_jf%>D1F1)2U>0Z`xjM-9H&O6YzAkfJKoKmYf_k+lk0t;SPdTmnO$bnbqt2gI%v
zN-$E4GM=*DN%KIYvNCt&$}Tdosv-n|_Uxu5d|5S;htQ2>h6RR1gDzvV1MU#2rgJYh
z<uu+18FK!P?=P39t2i9g^Uac&CJ#Lkqo%Hd66<aBWNCP`DxdY%A3D#Q+HK>1fh*sx
ztxzWxY7_t!FZM-wp?XY+&G?4cp6{&eBw70Ar+kD9&G*F%6!FWlpbDdfW1YYRMzuC-
zrN(5{vVbu*3b;@cQ;9gh*&CsdLR)b|^&e(Dg?Nn3GUR*giP-+C2zHTXpRP$J%n1H+
z52Nwe2PwF&+t8|&vt(NnQ^B8Dz<E_d@=S&~NeqQwJL&Zg?|Y5Md#c23nGAagwsnQR
zY1U3r(FXRlwySwM7uOXgug7_FMeiYV;8QKFZi5FvvWb9H=_nD8a-$14AFlCle@Oja
z2B0y==KE-13&Sf~bkwen4nzE|T4+@j#9-NYB^&fUZom`JO^t@~MIvIjfiLzk=`cp!
zL6|>K5lPQUGhg$-jlc;Zn+L)Mf}K^g(y236vB0vrpV4eg;mL`JT%r?eEAn9++Xd&Y
zDC9$#cngNd`z1wf(im*lMY$ww|6m-e|5#59T~Yqp=eRATtP~ZrK*cdJs8D#~u&jmw
z(MyY=_;*ZIFENDafAu)rfMDP0(Ji(Zyf_x}?z4_Z=D%YpUH^)oo`YBy3wbQd9W(dW
zh7i!9^s_}v?%`$+ao{$<Nc<^itWu%_E{{6{h@mUc((g`swD8X+x~6NB9>c9SI5@rF
z2fLgMa_1RwnO_Re!CHk~mcTI15q^-KUgt~_kE&Xr8vsr*zTK#R0Bh0AkR@=_qI%a@
z)8oSCC(xxFL#SDJdYnSSPwL1Px|p%l4&mkDsf^dUVvXX8mcH#Ue3YB5zv2vM+!Ckc
zIWbo7@3xObGT1{L$ul+_EnqKei?-{2FB^TVnaI?XfVLVX9l8m2r7swwXxE`y+PE25
zyu8bDnV)2rm|#CV22z&2Bb6pz6V`3_;1>35Zz3t_kohM#OuwX9n$7x^k<%g5zkn==
zAQ^xTJS0q3=#A#VmxepSgjiy2H161S#hGmSJBug;j7?f<!SxE@JlQ0a`Au#|p`DC1
z5mIir<!I9DD|ZgTjev^ts58DBlC*TRIq<X}SCtimY-Kk20CnDY%SvV}f0{xFOPi;%
z!&az{o)Qqlys07i$S4(z$C?CQmka0G;AKF4LqwpJsPT&`QIE)t@F)P@)Va~11c8xm
zVGzNEw6aerl0ZNL!m(lI*7CpXjdG-*vwTb}=j(w29!q2@VK<a=JyCo3pTS%Z0NX`j
zqZDb==Fs1H!2MbDEx>pUbDMFDWi%j@aJ;b&K(y5FKGyFk`i^od`QUpb%1RV*-S;m5
z7efJo0eLXC-h)If^C_r%p`RA7vNY}cMncn1MdU|DRB31Yg8xPmy6jQ5j{urzUFmRY
z0$nbJX^mA%&F1?XFV$WI!WUw9;1F5i`CG&~%U|GnDO0zT@Ywml;|0T!Ngf`?6F6_h
zQ?I+VthQT$Znb!^r_$^`+xgf6oNlv8*XmvvvG$Vtnn8sRp%O~Tu#GG7+2Fp9A)Uq&
z#p<eV-g%bwDlZRyJKABm&;&<yX6c$NVRXb_`Ichmic^Q<n&h~-hdHTh-Z=Sg&6NkL
zmKPxE^$*u$+!xQG;-uP5Nc;OsSWfD<>4uk60<{-REY96ACYe#}R*niUm8!#*=~esX
zmtXfXPr*=YQMr_U0_=_!fnsX9{TtVoWM%WZSfT~-_6tS((sUHCktgn4t!D;>*+h8$
zIFe{*z9`Vf7Bq<VsDTv`{>eNdchNI%Vh=#23+abkX6{>f;rn#mkryW^e_NB83dQrs
z!mKweehg*O7v{Fzbb|I%gHKKmJny>ZVwuEG*@523*`<3$6Oa7BsHS10?>9=?lL;CD
zKC07RDs2ujcR`-}m^D1pML$qjTy_3vX&+iwtHsYlhRCMbO<J(pI_FTWFBON%Bsv0<
zRkt}C{$Q(p!fT)BV=!tNK@O1dH!c)xGr0_HIPTXXO61dBG{LXhHHMTryzhFAI4Mq3
zK1sykt%g(GdlIPJv6@JQx(&}8CJ6HcM8A7wrn}AwjHNfRhv;Z5(|L$}w*Qvv_)2gL
zj*$g?1HsJ<F*h|FBQi0?0E$EXY()mv{_Iq9dc%)H_qQl{K^Il{Zrk#(k-BZs^1S6X
z)g;|84=#g&v(5d!3k^d(s-{7PGeGbnf`5}}<f}-JgaaCL(H{|t84MW8NWLESj6|bI
zUcL|oi$pN5$5ruWfCFqhqM!@?;F(Q!{DZI69n1}3`kXP4FiR?#I&6hLi|Ph!r^X|Z
zHB+<j6Gck#ua<2Qd@sJ1t7Sb+54CP^&q6xAUey{t{Mz@p5^8PHkjE&Z@(c;re>&sv
zc?t)GCu_;QX`>Hq{!Nop@!v9WPNx?lw({9Fn%7qfqDbL4)mwX$HHTN~#zZ3y@HI)f
zj)NM>m2aAhYt$w+_4Yo0aHIf6<?6M!B}0wqi7C^>(kF-qNxsHfSP3bD2!jefYy|Df
zph*1?!h<EVf|-{@ks^bDM8ObJM1kgaa*qcYXo)UHCW(be{L->1B8r**=Rv}6zF#9i
z{uz-VCW9uGj)GxILjyu2ypv+-oqt6pMIpQ?mLwU}9u1^OgmA<D_nh$jwZ^h&Qi>jl
z6iNU29ms#qI0!)gtw#|_QQjemCgu0f3cu18Vc<w;B0`D&{p{ZnB++UA3<Eo33XnkI
zq6{HgIsFpY+1{nTCi(x&0TkZ+nnWs<IoC5CkorcA(+iRKRm<XQlB8b3KwdBp3~IgF
zWrTSUtLJ10pQ*36e0$4gWs?MbS5%VV#k&gg692BK%ltoAv|=NVCRMwfLy=@y!!1S8
zr_TxjnXnO`%Jjb<z}-NtQZgwx_V@<;`k`Q-GKBF+5P?aaTetD-R47?ypvFPZgVFx7
zO{_Ll^v`^ptaYJN)HxkA7^sr`Q;9p0xrHvp2R34&pP2f@&%9x+8x<V*XofqG;N9=L
za5(LJLs-J$rP}TvLb|`TW3`Jg;h3JaDYrJSLgCW|eHBqpsD^)9txm=rgs}lF&u1OT
zz8DMdU?_-teJ~+BJyaq_0CaByA|Ljb&Ir|iAe6xNbJ05#5+65O`w1d)g4kaO4H-rI
z@RuemM37=NC<+%y5WP5a!LOt@kUk%u{(znD3j8^tKEF)VV+`-Uj0G;?jM2e*Gd~LS
zPnJHOW;ICKVVE8~vP#RF9SR_?d0;$rsk7S1OQQ7@+Y4^)kD}cjdAj#&A~fW$0+T;I
zhUzu;v?5@h6I)qYEL`ya7Oq|c#g;_9q8o*Kl9?_A>&$ok*mzdL-aK7rPbkJbDc0*N
zXPQ}lOx0$Vr9Z)5b!J^cFsDj4Y(VW#lViNQNDxI|gAV;k?@b-kuvG`jXZq;}2Kyuu
zyY(Jazsqtk=Cs@*XF2AMrNvIX5Zt=|_{^eb(pRCRO~TPwILHhg^K0wD`jk$#YU}Zz
z!i1|aAh=Xa#qb)&xH$d$Vtvc=I{QUVs@s1|c3lTdCR4khue+$s@Iu|rGOSbrO13Iy
zx$K}o1IF<WdJY5k7aoBcsTReOKR<-ynWKMJ*Uz3l*Nwg*)ml1@#pV}FS3jbu+Z|{7
zScN(?=)7qSL?7e+Y>uQi`xCjLLxixuuFBD~`9Cr0d_FK;ojT|EDm-kyisR>0`>gZJ
zB*uWCKA%%h^KjIob+xVmXYADwrtp>O)E7<Gh)FEi2FAk*)wX>AyZ0PkRFawVs|UFh
zy8EB}e)tv&7{RZ`XOv1I5}w|lf(z%#8Vay7EUJ&Os~NQQ;UxQDAmTT_I!8A>>B1{X
z%j)O~xA*&Ji`$dI$6(dc3^1&<&{Thw3r^MnuN^ABGU|CLpz9U%G*e3tu`U@wR{LP0
zCs7-~1cG6nZ9J8SBg{SbI`hEpPzFv399s4{J9uPu>Ej>d=|1}RhRm$zOqjDxktc0K
znS3*l3TIi8sDs0)_q!eU_#urY|F~@S8{fLuc<Tb;EwDmYA;I&|j;k#`f*(Fq@qlQb
zigPq~j6dr4`p1eDY^r_@JO??A(cTv}6Z*@91bo%_E8~DYGD--A6Q_AdP`h5|mR+=D
zBYDXqq1ipxWWmNEd%$?1OJ1p0`P)Z{urW4>!9<h|qS_(thVknv<zB+<Or_a^z)YBx
zP2IR_PXtY@Xt%SsSGZ&3b{JEjJ9H1b^YTu!@F^+d961}dyAtysY-2^vL`uY(-A+&x
z0$wX(${nY5ZDuUtHOZ?kw$F{CjJejC?@Qer5tUS}SHbWpVHv95Tf?R=eV@oFdNQ_-
zIm!zBfX>z-BzHjDIQFz{nxinS#+}!1+CUvT99m>AJa*S4r&2hgKDbse&8EyqizK74
z$+{(spB>l`|GK3~-FOyM(}I)iOiyt5fnT5_zCJlso2p<unl0rJ>dl9Q?#$H;Pend)
z$xk$;gXlj%@A7fZMrm9UNt&Ndizz$m5GTdcl>BY5lQCp)e<0gepo<AAztIqfY&-v+
z_|9O%DY+UcQi8AZVyadiKZZAc(%-JNs0=Szv}gFX5P|s;Y^`cO!#CdHS76+pXTc(k
z%R!v`ui^z#F;Nc~D|b;2>U&zM8RTi~oNM}=Yr1dqB?mi+$)ps~D<Fu>8%_zFrPCSR
z?%GlHO=Uh}P9Ec1&07#}P*1TPYzN-_)S8?EFxpw2)ul{oii@Tb@|hUqH-Lmk*>r3&
zBnuM6YpB1Qo+r&`YI7|UQNZ{Ud^<9I3#Y79km6-qxH5vvMPjsoR-Bz2aWDE<`q@K~
zP_2fxI$8ym*I#S#>k|lG$vdG1Q)RClQN{9Y>jyop(TI#vm`pS7Zz<|6N;XVB^<F$4
zURy>2LtTR(qooRwE-S(z0-?mWd9#nbxMZA@EbCE@=`a(rS2_T=Yap-eJAE4b;Ecrs
z30r&8LO>L91g~DglGNVu%VT)M6uFlDi%i)$ql;#;nsg_f)*0SHrtMtF@VrpT_{8@&
z_&koyyd@UV$?ad%do=m{(De5i3}B<IuwtshVUa(5e#-`tPp6RU&pG4~-qJ~(uDl)+
zxW&0it7z`ZN0>%FJOVG{r^4*Jn-?j)oUDeLy|;~;-9iNob^4l#W}5Hr-!+N3-NUbT
zx};PJ?h$5K?u1^CHy4G@3KfixiT6reEipgSiU70nWVbXY_^$;%>xdOht_JcuL9fhz
zv%&@W3XH7Ky*>U;-<*{KqrHQ(<1L?s0plCP@H(h94w7asdccm2bLSt-2QRkB=PZ9C
zQXO9Wkj4N`GXglr60g*fuOjcyr+Bif>t9V@Rb&Hbb%3vG5XqFyzxS+?k6ODvi}Ef1
z*k**F@7RUR@b}(5l38QSS)sxn#^y<H%+_fvUn4F`t&ho9Fh;n=(ZH?v2ueUCKUf}f
z_fhnO8OK{{(*T#{=@-pJ4~bT+IYh@eudllZ?<6<GRrLcqj1Qq#&x5z_#=}$&bWINp
zm;1^l-^X+x|7tm>StXI)CiSSNebhT!B9$@zmbdBF5C1XLWo!wR%Cc`*qd{c=Z{L&o
zNq;r^4+EEAaS39GnJcoJ(T^pSe5oz~a~2+P2hpqW1A2IQE%xxul5ltfx`v^y0nMy3
zFb0s~JXpUQ=l}S=fx=JEtADuG>#AoBdd*2D&oTwo$DA&3JrI45&oL*8Jn>bDAAu!M
z)<QbC2`7H|kw1iD>b}CGQ+rdJs-qbq2PxJIhCvAj=P~_9w%Jy9dO}@Fg^^b<)$Xva
zcr+_*1Sn`;FF+TIkPT(>PaF&#h5!(&+n=#ew>yEC%>jIg%B(hN3RWcL5D@h@j~M?*
zb!l{_O@7COU#H~I(EZhzDqFMV{-H}c&=>N>T7c0pphdvRQzZH^GORA>Zwp5-q@8pD
z#}urgG*+{gAdf`F6US^1oB?a0ObT(d3o|@B5Wuv|Ai7dy>9PcCe?T)-7#ME9Vrh=2
za2A2I=xP3=i~u?uYL``|54|~m-0{CNa`h<r|C^E998$dn1gBqSx47KY=Ry&9R^<0j
zGtN{T^s)t5ec2*xgBMRf7pZ{9sm9<z6U0FOfPdm-9CJlJuU_#SL0b)JwFf3E!R%a`
z@We1aI6<Fx=t?E&1%5m2SP`|f)@bmi7Sdh$@QVExBrkJYWmz<p5Duj)mSqrQV;9HC
zCC3~3PM)$ZJMSVQKYR7UP3NP>{E0;d+f6oorDt^ijlbKNeOUGGLJhS707P3NL6M;V
zAqG-`z|@|Akbs~hfkFIO6jLVl-$uEnF4m^>`(+gezM*81lxwMgVM7>LK>*y^aH)!^
zs{ShJCqLDOu%A+8#;2)lx~}t@o(vV74cdNIr0djHF$UKjMe|3@fFsRB?*MD_aP<F#
z@s}fc>2ytC-@jR4=>#9a8*nWBKxw<%|6hpR;DPz|3$b^J2Q-TEh&cAzL8LUs{A{dB
zk7)uU0*1(GiKy6W-9hLa_<cRgd0q%Y!cmCK9dl81Fjb3%B1+++utr0_nu=Ba=4nbW
zo@d(|JVa*ZD$qNF&S+lmD^eINX_R9Cnh;aKXnQno1PJ!_wD+Jw;2oXJ8v82-sL~3-
z8C|XXMakE<Qu9`8P5~=_6W%;TmQXCS_?A;Q32d)(KOob=xxLBIiTU?RS6z~**CP|s
z%E1ijKSAj4)j{e4AeT_h4<jfkva*7OX9oMmDNO0drukjQ$(-)ZM{VAAClhugqrph5
zO;N#F8|UQXW<wnzDfa0Zk)D|CJti@bAoC+aP6Vb51%_W8LxC48mM@1M2Dbgf_<@-<
zSS7~)4f0dM{sZ~3P$fS~7!1vBPZ*JCFy`*;wXGCs4wmr8FXF51J80GEHgGdt4@wF_
za(JdEUK@0N)T=XLW*!ZuC-4jN*fbXNFEr{H=$6f3Z+4>db4-?um)Swg#<3uEO5OBl
ztT(jf1|%?FDgFRD&=XJ&dSH>+YXpXLko*uw$#zS<#i>S#yh%U6?(%T)^SkG*K0na5
z^dZwA(A}EzUVJ=&Y=QIMt=^MMN6`3_Jx|I6RP9dDqu#7~rU<d0gTCcdrzN5wqsuq%
z&`K`M8YOBA;7XZ*zU5FP$;srhkE_1!@<^4*Z0I`K&i@7cjeh+r{DkxNRUTm66TOj4
z{l)YNAE%Q4H>4kK_xG-buj8)B;NiFC-HiuAen<p81i?LoW{RxzG3>XMI{xOh&<krM
zeo&~zEeQw!hL+zk1dggNxV-zWDR-uk`A?NqU;lH`zRA}V{&my)X5KyG>RDPE1R8oE
zzjJ(>eY7a25V#OH7<+)aXdqkwoT(-x&g*SNX(J!_k#YrWRk_uwlu-yV(Gf<xg)18-
zzKYpaa`ZiHm!|`M*wZTx{_hYM_O{;Wy4+`IHS}nC6<y@ilyrJ~qKo59D?Uq4ZJ~$$
zES-W}BkXjnKMV8^M<kG{clg(>#Cpp>X@AYPhT9wFs+E|muDi)t^OhH7nhTJh<gmtV
zhn0?b>x7{$OeKKbC*7X*FMfbod3ZMAB)~gviY9-M_p+C0J8{A{Jsz4M_7y0`<lx*+
z0;QGA$aBH^M^rY|Xt<3MpNH}39s%D(I_BrPXAoz_uD&k0iTpcL4}fk=eGC&QFSW$B
z&l@Q^Gw0|?joDxbRx#vZUh=~S-V^?$%f%DcBm#p(*{~ff5dcz=G8^)vbvvc>I?|^*
z6nYmt1Q|tO5LdvqAp)_0LA8x~j++Aza^~_B;PM!}!fef?5+-~x5)CM3lL>iLI@^PS
zxo<_o4)MxSgfk)v@nWzpAl?KPr7CtYHNDtSVsNC>4)_l6r3=0rn*F)Cv~SzmrJWMv
zC1<RJgnI~pVP9X5Jdn3_V)u*MS&jG3*(jkn_+G|%`7_Q$+k|fhn?qv3L|bF#k?=pH
zf@y}*iHp~T8PQH<rzfvdzixX1Z6<;rl%xf`nrBunO|+;h^b#WCKuIhG=-~f{R50HP
z;`$E+#{dl{0@^o*E?@!Oo{13eWy|c-D$Q;g7b?KvAGpf~we~d+KrwP=+@`lSQOS+F
zI#TMMe%a7mP%wB^to<1*p1<?^Cob|B%fH3Fb>H7M<IltB$bTvY`w!5-4qZH3&R(XQ
zXA5f~wyw!}ul5LsW2rJJqMA&wyacgiuTK)4pUw(}mKF<%U{nhUG#GJ-gxqiy*B&7J
zhe^}TTRoP1cD+tN1SS`9al`*hD@fL%L-A1*{7k_=O~&xvhu5lGq+T5!^;cyquHX^>
zu2q;wlc(Fq9Nr4NkKU4uCgM6Uua6L^ABwAOP>puM_;Uy~k#^ZO^pVF?3cSMQA6LNP
zpY@IQIZJz2FvYD^wsJo~VxPpf_=4U?0>#Y0=X^3P9vn*H#>W-+%JI$o=h_Za_RzBm
znm}`CtL86fAH+F`p$bQop_GMsJN6chbW)?1N4JpFlpjc;vh6}*oBz4xzT&u|ZccZ8
z2f?}y5L&=CFpjk~8XPdOXF=om`WNJDOxGqBxchov+iL5-q0MDc!mo3nPNQa%kF+fw
zKQVuqbDqi)^2d~Tzg5%&lpsT4L*k0Wvg&3F-Q0G=I^k^uOr}cw)8%plCZ$(ZX@Iv{
zcebVc$3XU(_!2XQ!Yw`G+#=ZKXqa;a-sGuww%@>J>DLtsDxgae)wXjt4VfBnuvEGQ
zD(8D%D2^JSq0^W0eC9n4W=+jG73os6XLMoJ2VyM~^3qivzr@ZOo3S4KS3J!iSMW({
zPKyIYEQVH@TG(fn&eLugh#}=TQQUV1%)llKbs#TdNji1X4vd1atN%U>@|=lSsgD7}
z&vERWu%W?5pLjN6iE5)0k9Vbs=0_HUzej>J{QKI<KK*X-=vo^vlexEcQ)SG*nj&1=
zOyxJl5t-48<B{eQM51<~Pt^g@ZI>;wRc*WBmFtJ>$Hkd8acMT=GcxiogtFw@yv991
zR-h=br-QJtS@fM`JLjHuZan`@UP6^f{SMUJj;I?fOe;H6r6YkKdvfgn1!j(+)v9mZ
zn{cH4??G4|NOI;}Jw|}vw~J*Gg);vxOv>408(!(wbFMn|&SKAKOy9sc8@3Kc0|d`)
zBOFI&NFxv|8rWWt5kz?pBw8>d0&djSG$15IbbY$mE>2Blklgb|lAN{gahl`(GpHWX
zKRc)m*zAO<h;TBak5F3I%3zU#De5do-FvKGf+a)S)uOh6sL=b5&}Tu|4vYjQegTvx
z`td*NULv5hGw`vOhnp^P;0qw7fux5gW}$P%63gQ^nhmhnuh1SL0c*VFUHfvoz=EeZ
zodtv@4@Dmj&r3qaj*gD*uzp^eP8}oCKLP}B*H4CM7+ZmBVh7FnH*q2)q78#jw4Lxp
zaI_7Zt6=LNZ+GFybOgmT!x8SEDXU%<`1A)xjJD|vF%pBwsPUrl0t#&VfzL!$f)%9&
zDiELnAVSQgRQf1R?bvwPJ$rWEKvd8<k1{j#z^E^o!3qL)MUs%CJP;O<xJkTZ9R-wF
z_}TGg09^oKPgnM{dm!3DSFv6~am!BiOwHio%g;J-W2OY@BFYPm))ydJkt&CNA}^ko
z{_1{besHW)*Cwr`gHf!lT4xJ^0qN$Ly>h+JLZ>8on$xbJPqbYoJ*{&BL}^nx@HrP;
zMaC;P=Jj~-YaW_;zEr{I!#A$$_-cGU#cdxRTseL7+B0`3gcd#zf+!f#Cx%SWnka{;
zV6ywEoWL@VnZ)JnW4m`}<+w15Oz>p2kLpxwF43Woz51v(X_7kNZ%DVieBIq2=VLtd
zc6R6^LIc03p3*EBH)Nj&6fejoCsBu$Gsq(&EEEBX06_m9GM7jB^GF=VhHXkwUe~e%
zGtw}Y-sku&2W>voB!7D&R_GZ3`ZuQ?neM<p-qo202)PBuMt+Ym^b)yTivO8hC|3A{
zizD`v*f=4BrGAR!$CFN<2RW>fR{m<`zU6xT%VG929g5RQ{w;z9IP{eV8;UWbJesg4
zWf?`?z#ULk=GpmyIyf=bzQ!Lc$jZ>HJSJox;u2W!NkYZo&5_C=2<Eq#1$IaJZ8Y||
z7S@^E=SVy$>VE2Gv8;=xtlyXRjO$lGVe#!hHcfkM0~xinHP-pZ!+0elA~3z!1RHHA
zCu2`lUGl<X6*WS%gwUxmme-7z;`)FJRp;;*&U_}3q&T7QT|!KdDTXP1TwY8Ny5pOs
z+LnCp%RdSOG&}V2h{Kn@K&Ur4+b3oqqgs(JUk*l(o8!wlER#Z#*-13k93yG^Em<zp
zpynlv3M)Oi9E=o5SIKkU*sSwf7W-PULvLB3hS}~o%TRzUd&r|MH?fZEFceh{la3!1
z^!k%4IQFJT=BNGfNU8AKxI+=IALV_g)Lo&Z2cf1+CYiS<Rgc=Al?9(v7HuO~oT7dB
zoH$kpofLJQ!H$)WFgip1N<CpTZU3XiH>5%DoZc)o;Z`c36@F)Sp`wjgQ(7$8)*6TP
zhGqb<b&gUF5QQ-pcgZ+tzmS3RGr05Fq{jME=<$-nTG{5ZH^PPWja=Xiz9?@$wD&b-
ziTL6-s4%Erf5E@{Y**==4Su33%rL=h0tMT}3CjLYR>22b{J+ieG@PoCqDg|e<VG&#
zbTfPKRX}P}Zc_`*U?IKql6h+Y-J<4{O36IeU%{um5KbqG%^u}=`|2G=5f_o~(&_|M
zu^e)*psS}!e`x6PK^*Y2>*V6JE`dNbGZ!3NmqCtSh1!+5h6vA<*WKip3X>5fvvv~`
zxJ}+NP@Jy3O+G@PlL5mbDqufs_-BD6j6?=RS-?wm9cytIL!kgdRoUwUb$aMQD)?yQ
z*CuVOp@%a19>wU>(97wn%PWS`?R+09MvwZom;An$Y)~kP6H*O(X-w!odjbH#R}XjU
zl>(0W-_*a#)d^R`PbXA~<d=UP3!{$_rr6Tb^GfUkwltnU#?NvmNFA2%cN30Q!xlC@
zwSY+Zw==MktGd-%hA?HbB(xfA!mp%#tVF2UzMHIc%5-VYLyz9(;+cOE4mDzeTU)lm
z;bpy#_WW3MX)y4(e~PRw3!UvKC%(gvmh^rGf#oKRcXf6@<`NF){N(SAbH^LzSZij2
z<MkGUin})oM|oPG^R3Y5m^;w<Ihtz7(+h+)Oj)ob9sOgMzvA_whH!*j02{v;f<H*d
zMyisPkbC?`tA;d%h&Yg$9(@2w*PoC)MjE!D7gu#hU`$R)gV5y*R;<aVOtD7m6K2Be
zJ*xRkJ~k(0vSrpGSxjH~mfrNIlJ40{F@Zl{q&2n=9Bx;S=b(<Y8XU!uyCpmgDFd6d
z%CUd?I^tBQx6)wkBX~h8o!uo#1`k3O|3#N*C0XDfz=iOFyW^wwg4-QKW*U3oC$-ZY
zEYoog$qoNwpwk*>Q3o?c-4c-Jf|W5PKBX*0$EJb(XOiV1{_d7#efZILg*4=o2mq_j
z(GPpNG$beO*y&P{(yI$+;W9oK@c^u#lm1?S{H-vptFjCZ(Lvd%V!kyZP}4R!s>7(1
zaq`8Od)ScBuPi$2FCvHFOFz;lfiBw!fhkjm!HdXtp19A^j+9IM_jpANHMJbl``==@
z8moZ0t|LGdddxj|atmTZG@*HM(VPaCup43DkB8t&*RB#r1<F|6CPg};_6%U5+EgMx
zBvG+2jj&?a3^+6ZwTvO76Q`<OXYH3n$`mHu(B+L`LDaUU!3G!S@g=seJ;(297n|K=
zo3<+^x2-ED`}@3?()`2w2Qw9kwyM9PPmJJlZ`{GJGkLIX=-agM_Gn(OU8Z<za(fju
zi1eY(1<K<K3f;E!K5wYB@Ls^Gqm9?&EE-OSYgeux<09ab`B9X22o>vYO%oQWG~6o+
z)*=npeDVD4@Dy#)++-evZd@8i^{M7{L<eoBccxJWe*x*;E=>XB0`wgDB=bk$-^%Ji
z*$CIe3}Mv1f44KrpL_H_C$TlwdW8wn-UBdhTo!f{GXE~a_A-@k5jF(kE47S)6L9zS
z1OdHv4D8}m`?`btAm6@^b-U;-J(RT-UY^pMB}Dc?=UY?Z*zjb?!h}rwsRva<>gWjI
z6d=knWyt5hLb16-0h4uEk+q>1B*iVknAif%GYIIw;@k3RN#aA#;;PT`0QE#56>NA^
z7;OpT>2#^#QRWw}T@Aq4a+P<RL?$$ehY#}l7qM~d8irlJ@wT>uwrbl)?WFdjHml7(
z`8-~WAFPK6pDE2hDDL^mW*qc~wj&!ZG8sHRFW^gwFNYeiJSEt?{1HFAz5!f{<G!5|
z-a}-057ulWVOn0KbZa}1&Lpjf6p+xz=_0VC<AF)f2jy~Fs{IC{blykkmQQ{h=5oR<
z;ZJMnKLQG--2;VZ+f8YqjNkHQ&Tj$^DqPiQBny6hAVCpr@1|5vZ6j&wnGGr(*lu44
zPg~sI_9x-+4l0cw!b95R`*ud=T3=6I*~4&5Zzp%QgyF&j`7;Zr*GOa1{uV1zN(ik&
zDjOZ0Ow7|l6Bq#61UYoJWZr}{M(-WMvo?3Gs7`CzI}SUWI0%VJJ}j(6;YZ4cdW5m^
zhN9?%>0>@Ho@AT7c9*0y-oMM|u>Cw#&MQF`N@3?t<_64n>de=Q^n%+J#F2cE%1ju_
z&}GwJjsR~!jff^s51nc;K|1CBQY(i@a6^y|79w&GW$XYS!ONR^2x5)HQ!!w+$TuWO
zZ{%;|++I=WU@}x77!<Vd!C@{i>MKC!Me7qGmQ=kf+cOYv7IuBCFc5HmSuM>TLJ&Xf
zjVmyq&WZalFhb^#75-c>EZ?QZC(T8P;|`@txcx)`K~&z2{PgbwqH3>Crw}sv1~K3_
zMoOh14o3j%jn{*_HcV*CDVctgdDm(M(j&>`1OXvG#rt{+p>=9&3=PazH69G%n*|Od
z+KkRu{ush1p=Obw$&q;1;)-OP^p%3ZS1=TRZ&B8IFzm@z6RYQUj&5J-@_8q7d+kDp
zbmP|Oq*F<J^}lfW@mP->jr+Qzpy=ZkaT`%)6Lt%1RQv7ZEESHToO4Q_UfOaVi$-#8
z7qATluf}mp36yuha#^LiFJ*-W;|9Mo49mCP`COBv(5DC*pvAGQ33)k%)MA${w@<Bu
zmvXd9+EJVsjQp#JzdEoJcey3({zA|YXFIN_mFA)inOASk`c`p4;M>PMvIXt4n9~?-
zyQvQJL>%2V{CRq3@I1!CK9wJbI>?~oS?*#xnlOG;$ES4yb7DzmLnrwwgK{rF#xN%1
zd)Q*Jr3;iFsip`GS|4*CPP9geZQXEEUF#y@DKa53mnrw4SyL@i!V93jOtf5!jKEk6
zFhwAiH}?Eu7umw#qBYLvyne?cY0gTy?w1Dygu2RU8L?WMMH5WJqQw*Pyn?hVf7Y#Y
zE8SzlA1dF|DNc=~);G+A<9c(+t+H5vElewj?!lCWqIX&VxVE{CsX`JwQKdUSeQZ5S
zsGhZL-W%Mh2@|@u0IgzA)qrLt7d>-7KFo-)+wuyb#oyp}AVCXlI}b}HWpfUKuHL}>
zMZae-ym-IU)9%gdZ5f5lvcqiZsKBHY`U4#Rg6FEx)5|vz&KxWp`MI02^3i?0J*jPL
zgYGS2<M{{@K~71*+8zM|rgVH+OcV*;B>r~3LBvtBm;TgcU$GXJeNKeRdE4RDlzxF^
zRD!mtjIHrJm(d^sviZzqOs;6bY2B}4cW|tKXGJJ1OazOQF+!HT?APq2A4fVm+>e>G
zjj_5V-y>JXeoji+<!z$f&>M{2fuxG`K)jz>j~So<Ufea9U=w|LSu#}gru=UL-Dw#)
zm4Z<57^nsvBbS2El-?}OI1zWWai}{w;>DZ!i3;`Gt<y4#Kt}4qPEpp52cUo$d3z>O
zw73A;r=vAX3R4)8t?2@z!YY@2ru&GFP;v1@L{{!#Ds?lFavG@zp71jo*1No2QIhII
z_%?M-v3(;UYs0R^Jp`Yox4aZ-zt6n4ePnlBdPKA_RX_+TcMCp>zz2zKxN#GM8o0GB
zwkirc=E=!SXjO75rHpd^kN}YH4KZ%t6!czL@LOLc-&S4D-s??A_6kO^{G^q$(Rh@T
zAOCPj;|K!9z;zbZZ5^mlklsNR9F&)c6y9VbYGgA-TD|jxq%oWvZ+y2Xvao!UGDGTq
z$HSlAzsEaAP^Q;f!gIf31Wb2L*)=Lsv+a3SUgRgUPJ(z!EnOW*o>hVQM(%ZUv9(Q)
znDUBx5;xR;^Sx{eS9J%rx~(M4E>?!)uvP}&p}y}UG($T`Z(fODH}Zl+t4Zj>`TSTu
z--^B`64K0as!YjaY(<(<W8gIr>VWQU^yzXxj@#+@)B@8Zar;=j5^x=TUxqpSs#)>k
zSC{xaVRqzOluwpa%Ao*c61zrz4EJt$jMp8wv!<-BtGerCOogKG&L&lo$HRIs@v2Cx
z#45pT(!{FwV@Yv|Cee5a%rK~jJE<rgC0GOna}-*-um*_}?-iPnq}~roFf#O$413W*
zr?t0_J7Nx3*zI>EBSG~r`O`zaryD|C)aT_Xd;%hp;Jzj~9^DBH38v7U38&-<d^Kee
z`T3$hPjlpQC?0Hw$nqN_)-WYcgF|N~t$w0`|10;!K8kv31$u<bGj9p5v<88aaUJxU
zZ)^Lyf)D9Ylt35bBC@qM-H@tJ7W8{^q`q9ATu?RyCuE;oO5!Yyh$;qb0v`65D(}!#
zFeK@vjm^vU4&n(=p)%^2T4f<c1-b*Cqtlh6Q*2^S^FexLwihFPg99b}oZemNW?y$1
z3NEVzq^-6DG<@u_<nZqST-$1NU-($A$3?ui6F+xsrL~lG2s`_C<*!gcs84Z`WPrao
zH#Y4gt^WiQ;jt&K(zM>Y@>A1`PcG6%sXYhEj#XVuKCB)Pk1UrD|4clGL<!4oz#c}j
zU2dJV%|U1n7hYA4n!Bondw%#JsVd7dXq3IODvm9)q)$@E&@3%beAr(Q98=7vuXzda
z#DlnmO>vGQ80ZpcNh&DiG$-4=sO!8hdsfz41G1Z)3<C>F4Y45x)2axq6HZE{ySDa{
z<TY_8=f(mQY3JKW@&x3;dQlx{<<Qr;*SX{|h640qy(kY-YkL5-J>M^+YaibUNGC!V
z(CcMKzq^}dvIEA6sbj6%QsOD*&`=an<p<&RZ$s5lwZ$w`^}1^@N0$^^aa^J)1uCa(
zOZxRUyTOBv)VpkN($4!>kcimP+$|JuK+k^72A%=e+YKqic#0&^Viit2Q2ivyeNF&-
zy61CWPt_k^5eq3^_WrQA>K@Q@HRz83h={j7#ln^H%C~N`TA<;%gbD^1XL*;2+rG0r
zpMa|~i}lfL+BpnJyey<=Jv3Jns<7q}c8#lo-7wR#%Od?#)Nxm4Qxc=8O##LXD-tYe
zJTkDi7x|m$nCSlN0y!4%a2anzih`V-WVTzZf<9PAevdi>)-;K~2v<1os;~F(>_nC|
zU*_6EzfgAqc8PK|hWY`Q@@R{;eco`^hbW#?9?)yrjP8fKxAV!N?6Lv#Iw0il{%!5C
zMqdN>Z^Vg!&1?{B)th~>`FdiGF`Uv5wg6xlpKt$BJEkusgv%U7qI9l_&#acvUpIr&
zWCvLTj_n%m4>681h4;Dlf&qsY;d98btIH!kCYmL3cgH!vD?_x(0F^!$+o6RenW4NH
zI8A*N$lw*nliCetqkXjk+8lza9lJB^b_}622qKSD0McCpPxdk=4iy_V)L#KF*>2z!
zqU<g!82I<88GGa5o{-Ie*qFD9NE4QuSOEg=Z*Qclo`&?sLQvh1?2|H+UM87n=8D7p
z!uYxE9;Hikw$rkRU~k6P_ubI$eC`>m@Pt4(GxM9r3%dg)q4Vx&W@}0CdbZ9i2`^EC
zXWjvPj23CAzqe-5>yG*G!ULZ2B~-vArbK$x{%@U4I`{?<{V?qrVoA$H>MjM>fFX`o
z`{VJYAZ6p2gqHqaIq@6TE;$9YYwpz?wkrlfcNTuZcZX*{_a=d(>ge9&+UX3Tua^)v
z@$Q4~RjkyAwdR8x&MdUY#UYe7U29HpXi%~#uFAdM?xU1DbCi~&;Aokcw7{iS=xN2h
zY57UMPfDlhyn<|b7;e<SPxz#BjBVz$4hb-cGWF1vk@`*=2939e=UQF25Wny2ZC1VF
zyd|NAxwv!@t;aMc&{UTEO`z2jsXyBcqW~FzHr{>sUn>)s;mPXq|Csbcj^k}pK1bO>
zX)_Dya)gIgp~qQ*&{<C|K&Zzb{OD&?-6}+z6b!lI=2sgOd9Q1=$~4$?CB9n%7UPUW
zYE@5yJ<l5g0uNCM6$)zv6=N7T@1XZ-8=xeuKDOX;(D!pVvj2KdhLN|`3XFm2G=8i3
z>m%frnzddq5~@|H4b?*NlEsBHMubd<bIk<WWJ^ig)c^q<Iu#&9{l!-$h~+tFzhlW9
z@`_F+28}FA=+t#iAH}xvRxavH6U;NQ-8^x4YSHUWsy523)!yRmixx&EQyd!|^7IhL
zJzXxdBS+4f(eFRHC2$y-SG!GqB~;h5_3+*no))vxvnILCychvgwF<wQiL(7t7p5lt
ziG!l%&ujL=S?UYq&f>O-3QKL3II2b&=r=x)Lg+{<urt=?YMk7uCo@g{%J@o=bFMu-
zroF3#;Vp86!;&Y>32|I2FeA;`yXxuZ{y@AK!gXOwDIg;NFYs+Eyq@DK;wS$rmOdpC
z#1^yA9NlgPt>HQolpy$VSEfv74xdg$te+QcM!PQ2k5m9OQN{s;cKE|FFPUM;aDoZm
z#u>#QYOtX}RXUD7(IdE*YfG@&)|k6ONZ`t;Z3TR~O9&qZjb|1+&j+W@VIBDjV_{!p
zef)a<1D`@9j)AS=b?L2sfie!euCHag>^tI_W{Lj*=A#(&8##aJ1C7!D@+**o%jGw+
z>=wr)ld=R*QlhBFG3ab%_(e)klBVzKhOAmI3n`{Re)uo-HPL!SwK}s92e{@d_~_4n
zEa^>_aGo4e^wZ(Uc8c_BwrDqDYiL^Zw_42tL4cX7L1?K79)w7F5GWZhz4-ERS(<OR
zF6M7EYU^9e5Hrsh5-YoV#uVd}YK5&2vS*DOW)t9hM-RK&KVAmXXpq8F?s<Kw;N_&k
zVc>syye8Ln-$_x9Aa&rI-EaF~><pdzT)`GO^D|Z6vk$Uf;tuVCOhc#)!=cNmW)YG_
z#CGBsCa>-^%db{`UeU5u{rySt-#lJi{OJ;)o;~ISqK3L#w}{pvuv@gpG+k}zX?>LO
zY}<NMzQbjiS6e>%9ji$4Pf|BBU2FQ{kWJQ=@gFo(h&Fd(cwW;h*1Acw$wz5YpqeQ(
z*$fg=i?^WrO$t?Mx6PhDN4XJfD)zHE$TNZ&Boamn6f1-p1AD-iBlg_fsi2OCP6x*!
znJi*vlu#wWE`90-bc;~Ny@`l?*I|;_fX8$VtN=yFrSb|`Y>}Wx550=k{<GNb5pPf|
z-{WD8)}uzN>VjB+#&6m8SerqbAkvj0N4KlMdKcJa6Xqu)mOUy`u89}@yEWF-<X3MC
z@EO&(oRE7Cm+i7nZ7a7Kx-}WDid%V362_LamRZg4ZJHAksLVSI(j}W1ck)Y{Y^3fR
z>kjL!k~rR@c5U|eWHn2&Z=T>kB-8FkkS{9U#gC9dMkrMGN>^&?cxLBje?M^Eh4Q?*
z1R5k?H=5#q`1%xDA0)}(g}SNl#0d0G@vdivc%z@$B}-pK_D=S;dQdTCmk(scfIv3h
zN<YV&K?aig11)LRZ|O~kI&&_m2}T<!Kp900#cQbg70N3_Rr=;&&MB%+M%O3U$^090
z7%_?g$)v>#mQ-`+<fg`ma%{EI1tpCEs%7W{0;Qo9q_T)c);IYzO#R6M`MFZ9b9F-V
z^Y-;Ffl`^S*GX=RW$r+jUCI#b_5KOqx5fm0r~V`qpaH-alDT_j7mSBmJAw`hw4TaW
z#IHVs)Qxh)@_8I-C7{@CMsvX;fdZdsbc1^hPUx)qrdrBZDaooVO%Tx(|Ij$LMAHlm
z3*9O~F~N##^VMeWJ7lgC-bz;W2<RGCSg+ECQ;7bYqtQ4H*Og-E5L6;or_n`SOGkdP
zlWrmc@ZGviEDVFZTK_go@DDi2&|;XH_7tvl1)&IIdLmy#vNCghVM3k0e#Rm7_!!q6
z2=4v}X4=L8cT5p2xc`8=pm-X8nbCk!euYw>1;UA7pLyK~8f}#SaF3j87U^db6Dujg
zv0XU~2f8iQ%7r5X<J3SVN7Bf5JWvys9PMEi;2}ro17T8oDKI%~0QHi_weEW|6kvzk
zzpDzW{s?s+nWo-ok65u0`%cRl-^=IGrLyLmXaS;(D><B<X#Fu)(r&Gp*)@?^B6Jjs
zP$ib=Ddsti@-JCup(l(7N^7~|4{(^N!4h+4`?!tDOkFHFd$wY@slh~ZV<?iEsZ72?
zU?eZ(Loy7!ove3nTQ!wodP93>0U09B@UZ5V3ih~!{X8Y)1X7%&aptdlOaKE@eD{+s
z+n0nVB9s*4N<!p$VkU%4dV^dUw2Vi|0ia4rRY?zxERU<r#Wjw*oa+BDmcf}j9$uz1
zG$EayW)&h^PvK0sFemz6i5jiAy5jc?L{>iheJ$WknlUyTo1)*c?9d0F-@T9eUB2Z!
zWFFo<r7<)$kBnwJfEWBVyP_<<p3dT?LmkQ5q^lq`vYu5}k-CF(7z7Nn%a-66hJQjt
z6#n4CN>5+n(wCzJS|%9#y^l8Ie64{G0F|ItX&Q1cjqyTVYW84lf4uNv#IHjKgs7c2
z42X9OQY}s%<3s5b%9XdxARP!m2YdR6D5J!#8DI6LtU?^$jrbd@w%%Dc_Wd=}$yjd1
zu|JHrKg<O%h8=z%A<R_O{;R2ftshTw%@$3~;z{i1t*Hq&!$U}9rf~+lNbxPGYsrU6
z3p!((UG-3E;ZHrY=q5a1R1~2$Fs!-l7h`x4p<o_=gr5??U5u(jAPJ3B3|}Ao`<i>h
zjAuFdg_~GqCj_Xd0!VX_f9$5;YD#r0=+pSJ)0gIvqta`T)$ZV<&8X8ouAEZ`ss@G5
z+F|WE75%h>cOf}S&A5j7m2gZScd+^>9<r!DvK~0M4~3h0U!vQ#BaSu=^r&JCT%try
zLYX!@pqNYjxUSnw&Dyc4zMWaib?t5MWeizoDF!qL=2i-o2Ou9OQ_?B6s5S1m`g3<Z
zu2?;+Bj*xXx)FB=44W8uw>$iK`!lcd$Li%M(l}prra9}}*_jl|b@|>X%>FrAcrJ8W
zh3(yI6v}k@!muuf{!qpq$mZ|N#(#FP!z=&ri>QPF;X;@Zm&UKs@2ek2L?U_6DZ@EW
z|AA@lMstRC;Q|ux0o<x}9HEhK_Crv((7<+>^wQ0~T3ml^^Ts9F0ftPDoC>#XCJ84(
znx!t#dyqL+!2^Z%N7m@Gm9@(cE&nve=|eb1{{-p%B%g%QeP=X#;5W@(1ds3w^Bq{P
zrI4a*k+7ogm6XW22FmoRmFiVd%sMGn5JZM~cY&>bDv>V<JaJiOrGKfE8z-=b?c<Tt
zqj>fibJB}-Nw<x2WwhcxI7?plH}yJPuo+ogGRMc$ZKn8ugPl^~??b9Dy?D)tmO7I)
z={QaMP@zOh9sHVyKsMVEWc6yZKX0?6COdKX1nScfeDFf6P=6?dJXAizMX&k?^>Ks<
z3HNp*C1FFaLa4nOq51ZcW5^R46pMKi*aXJbN9BQ3$v;!ckP{_&l+$3ad8TIM8oz<>
zC5m*9PuW)X@tv<SkvS+YWWTKZzT#y85+yzbPN(YHsS_3S0gVA*fk<3}krD03b4d+#
z#7TtXrDo+=!yX%-A_;cKk!K~9=9%1Zjs!LtMwV~BD*^o~q71a5P*J_x828HtWc8c?
zkXv^oM07L$%G6cS4Mw)_roK|nP*b~X!BZlXk^0Aqgl6Jd7M6iR+FwNys%sH`ej{P8
zC9-!4<KK%c0V;3z9Dh6q>kUf>i+)c<$@HAQgbIllgjp}pWRy``Q=uvkV5C#5g}OBU
zF2sZ97mBVr(*36JDhwe0+WAP6`x@!V3J7EQd@$mI%>$`|{0)ONygE%(h8}l4RKCF}
zQy;OleKQ+s?Itn-teKFgXAOBTkO^>*#8^s*<HUEL0i%u(DXr?wm!tOAzCe!ee+M}O
zQJvu+v40+}F8Jwb)>7vxJ};L07IjP1?b}Zck9k?>84=Dh3g@~&cNat(+QPd07UITN
z>S77CdLp+GQ-<)kJdje_H^Z4MJwUMX;#x7D1ZtK|Os9Q7j81#1D@bhn+<=w>#@q+c
zjXpav9m_TzS1{dgKPYD*m5&WBuSAZcf1gUwgo5ua{TUM9WA@9u4+x5q%v$4X!{`$H
zO?)=BB<PGdQJLB}Tt9x_tcsU#VizXLZKXLIe^{dZOM62<!PWrbaD!}2!kAn}9h83&
zLo1V22m}d3^dqDQ_1i4#d;ze2BgACq&lGlu>U&$!%fv2s>$hAO-oee{zt@pm*t;W^
zjzsz>JiS6iUv(Al$e4FYV88C>NcnsLt;q!I?sT@O7u`10+4*W0tB8V9ONgR}&iYmF
z<`liWOW3kYZtI!H9USPH1?|3v)PiC&-ksc$w>N$<$AY|w;OHQJRBYfA+9X8=t&lkx
zJqIKwtUcNW1_Y;)IG_kYNr^(pT9-oQ5`z;NE2Oce+OIR1GheWpF7Jn|oJyR*3;Bv-
zEh!~OBSV~6O$g!`R`Nk}AeXK|9e5D&aXW|NECb|dC?g`5>@Q}<Yif<rDSZS1*{e<y
z;P2@3F)Lj)8Rh+`uL^$uL2K><n%(^+nTH%!OZEwZF-2|pzV*?Y8D#e?pZ40F#nJOY
zhkZ!Qh8$mXB7!QmWO82c@h!2yu}Sv%_3MM>N`hCaA{k7~A#^Tcd{LL-f2qs092o^f
zGRjwG;E?jSwm27K=pe*Xwj2N<l3X$T3&lZlf&$YWE%0}rsnHlUCvqGiOo%Vqozy6`
z{w4Ktq|XK4N0_6xxQGgQlGSQVA(Odedbm!#v{*92M>k|4;z}`_(kL@&Psn?ew7=MQ
z*CgEkLDxG62hv8}gRyPfwrx*r+qOEkC$^nTY}=EGCY(%cYvRrOeY;h^t=g?q)%A2&
z_48ME-FweD_qh_Vpi~grB>n?gAbL}De1RN@Exe>QV%Nihh_h=_o!VxTq6!0Kme*k+
z3e!J^07dH{zD8qgzz@nN@IEF2UTrKJx<5z~Zqp3;Gm;s>_a3N)I&`&^INwMaAp&2a
z0skZa-nZ2H(N(jhW@?cE>Q@RXA7%~4Z(>3^vLtqmCN)FVOqMT61cifouz`OdRuy<a
z6_<OZY+`OyY(y58j3R|PrgzSJlw%mz58EnQW6$!&l2w=vENXMreJ0sl&cuPZq8ksn
zqb<aEEyCLWU_kM34MAs0L+8Sx$c;o~ST=I7q{n<5<K<CCw`5;}TI?Vq*!zo`?Sd;Z
z1zp*_08#C9CRB`po7oIha0^EXna7y?M<nTe4^Q&OzObJ=<>T%TL#XL?y5KH6|9^-i
z>vrJnf0(3UssETHYmZ>g(s;Mso1*WrD(sO;yRjhG6Y4?KenDzSYB1ApyJ?qI>@QOf
z`VTR^S#W=5bSJTzT9qy7SpJ7f`u}E<tU7QDdLURtfjVu5T+z1(XC$wt=li%KYT86<
zH9uVdX9D}SSiL%JXv$Xt?s<hGy`r&PIQg03XTQJ^TiQ4N65anFDye{`=I?l{d5YO8
zg<0myA_(5?HEhnArg|F6jO2F-y01=*c`Hzl?J<)%)_^tt{=<LRYq>%<(sz-LLc7p)
z*c9FER$MNkKqFCOXe`zOG0)OiyR-Gw&v}V~+$kt!b4<{vEr;`(FH4A^3#+UIyK?I?
zg`+>a=L9InF^OE)_8^{jd*2{SQ76ms)1=w0hm$c}0V+-sCGw*^|C{_O;;KwpK{gMT
zAIR}=GI%p_N<4HM8I(D)h%)=1(SJZXtqGI`13}9#;N~o9v@0VCh43C6z2^nPM_FDw
zZYKr@>`F50LyGt?*PgQvUeo}sm^oXiMc$I)?EH}`Usa%ajl}dO<2^xJY&=24WssR&
z#AgDhCi@XYBqIYSO#t1zFHXtXIm~1jW%~o07RoonKEVUAWS}tvXBg{3LfVb;+4m7u
ze5h>;2q8hdzRQu(lHD>oyebf)Ql-Ww<~~g`tvVm;kjB?|_V2{Q++)Kwj7mvCmgHkk
z%D1<%fQPJ_H<#hTjRk#L?Rkh#YlvX;T6bT}J|{UIf{gXT75vU-F|0jME{2EFj3ln`
z9sOUn)q#L8*tnNoz&%MslI``Tt`leY;&}xafQZgx#}&OME_Bs^Uh^zB*|g7vT$((4
zs9O{Jft3G_c%S2SG1AOgWl)QLCH;ZlVG53imJG5a_y9g#t`#xR6mU^?x#fD>_xbMz
zTFZLVLwAd{qh}vQd?yIEOn=2aWSRKi_Bcc!Z&-D8mT}OrhQCcvL{e(q+dmj7JT?#?
z0K0ceI|P}vhi_?HCVmR}LEYu@<dNS=SR@y|UD$G^^ZF+1@y5Y>b&SeLNXX*|*(IiP
zV5^Mx%Y?|c!h9Gi`*fcGPwgFU41mj0aF|K1jB|U&`g4jOZ}F?dcB*D4*IHG&S=ob+
zAt4lla9MWcVRY-Fn<7;rRvyx|-md2ZYwBJJUZ$E*C`-YT<j~+l420`hxMGW*2^XCv
zeUG-iniK6}b#(|ZJXIS3$04536vxJM4;*tGxGo;3Y?kPi*LJ4joV|8i-Q&!v70B7h
zs>tp{FNHVr7jnHnkn^aHg+WgAx_>))U5YesrQy9tUS0{HG`^brn`?@N>S0F#8q}vX
zVv8=3;YP7NHB$a4iNB>$xG9M%53a;dW(SH>Y00IX;wEk&<@dIWK2OW1+o4hMOQ6XO
zN|n`u`Bt~c`v!p!&_un#N|PHGT9t}J&Lo)(z_MMx9>Uus^|{nP@}~5MsWF){6(jH4
zA5)4hesMnkQ9hY<XM7>TC(yvRg5uSwbiG*yw<Q!)sP!+uNKlCK0Q7m(DXvaha6*Ln
z#ad1wf+#MQ#p(fLX(1#(;$&lOQr5EjlAN!`si`&l8^CUF9GombWM)!=#LWP6JBUHD
zr}bq>DDVl(w|EPLYWZ?QP(;AqT-c0DpQL7BE1;~$1`G3#ahu{PmIFcw7HLab$-+mg
zxuiz6{i?as<H!~3QbnsYo%KJu)vFPYf(a7|HKYcmnD|j1{va^KjO(&J2%7ZZQUs;1
zRuP*+5nS0IMQ<})DwTdTJ53m8(!?=j1*BIQn1C()I&NXIc8%!Un9+jF*4~@>qA=MG
zHwa81CEA)m!TWGY8Uf{l-w?|(>^%O*DwyufHL+SRzhi1o2p?nBW@%}SU~Kg}ud7Qi
z_r6%qEY?&^;IKhu20>qX*|#Z_(ex+LY$^Vm_GCs4ukEt|mM~xRnh>PtywI<lbb8l-
zi_vnuN^|h~v_1FYU0Tkm<#;?<s;6v&TItd$+;h~|nmk=DBVdwk$!)*JGwSn)t-Bu+
z=c4xI_MblxGuU`u)9E@DBEXw^Jzd>AoZe9MQ`R~gj6!UEBL}NRWl%KIky=x5hBJOH
zZ)9;q#5juxN3vrQ>->LN>6tY|3Pg@29r)>jQ)|7ugJL0(Z8Pr8QDJcv42s<!kR%DV
zU@j=HL<fEzK(<eIIi0ys4Cg8SPlD7xd~B*rGQzS!{qaAy^CQC2<T?A+llE&G5SVn#
zBb{NITz=01D^CP<BjJ62c85(__2?Uy_Q+~}HjyG@q!JVz#c>f)dquuV!<d@0QZ9U}
zv3{KK!t7m*Lt%uDLhxuAQKH67=>++}U?O{ViQiv52CQ}iLW3kF*sr05PDH)9-JE$y
z^TesnzZ4~yLAa>4M;y*CqIg<eul`CqRljJy5<_$7TWZ)>>qgjq*{w!%UwB$XYK?_~
zyjX6#nxKVI@pey5#VTIYZYT2Uga7VSX1&~u?V<t|n}Gu9I)fPrhg>P>P<?vB3AbB?
zHF%{AaN7LSy#%HZ880%hFI@mCr=pSb$Ak>{8pIcIH#~J$gExP&MZr=e-j!$v2djpj
zAjLPCwddp`PF2Mbd5V9mJ6=;|&LNKA-Kbk=mb})0>vJGkV=`P!5qnw7^Fc*xh*byE
zc_B%kEw_F)OTmHQQ+x??MW|YQ>o1>el#WOO>?dF|5_X}ATNlZvFhH?Y?aQb_TW;%9
zIx+}*v^uAOeA6uWxNVPqrk_vloY&}*)X15|40CE2sfmdLKgDVHmWEM03_py%y`>Sm
zPP~T`ivlOD-QRBuGlve#qx<&Q=xQ5o)SHxsSC1LkYTKr1$k`to40GfAxAjwDTAE=T
zSPPO+Tg2V(JR_+7x7ZPH?D<y_T>K+8EYs~*Jp)D<A#1TwIUAK(%C`oqQ^s+?8<hid
z%~ZtT9wF<WIX0}&Wc3?I<clv7JeK^w0033pIu)KV{Zx~kfjZs^ot95EsWgh5X~CF&
z$;Ftn$br8x^ph~uWN&C&REb6-Kkc*|7<(ny`rRh+gXvrOzXRrH85csT-)?_44<XV!
zhwxRup^4KR827&a=avd~Bf?KS7HC?FkmU~AP4P>4SeIdwln~7g5cL^&vk@KbCy*OQ
z8Tm@j-sqLlJa{D<8^C{2jwWPEE%6V@=D&?ghDF=drX)|h@kp1prGI2#=QY9ubp1B4
zITRxjh^#8Zumb$6=;)mPoNq5^ax*JY`B6qC;^pUmD+ytsbDEAv=C}5p%cVasppa21
z<wBa8HD}kz)+nSKTc<1V=%SjhxfGK{wWB0d!tf0*Qa{oO%j-vgtE5gjnFaIwB9%zZ
ztrX2Z&{siQ?BM2zI(u#yhM1cIaLWHZ4Jb)z@zq;QFD&~aL^p(q4f)CZf-uh<k^D%X
zP8MY_k0m}5a)<(pi-#`1tl_k>(!uT@&!kIZeZff?X;TpvJ)EY=A{#UJs<$o|jjM&|
zMxEQZkY%g4$K3~es+_LNG8rEb%Nu8QMizF%M{y2>-+TwLH#|7i@__;Y?zhE&a=s9j
z_kxcjPvjON{`kPxA~zXSwKtd5N)JBd0OgDUvv0PBI6|Ofy8+#I|Kj2GyM|XCgV%2!
zJZG{5?7)wDD&r_BHf*=$3q7~fY-hKF@9VbCo(8O{y4f4t(eJNY<pyvBWx5-{k+8Yq
zwVruDZ$}@?-ayi9J8fBkB~cL3P8B2?AwukpsDAVZrWwD4oM>TPM712|JD+IcQ{)HB
zGvdxRC2~>pSJM3UvIS!pOiFM^ks8nK??^@d)UlD1NXeyqj*b(n8ot%EsvaEw?+O#l
za)RZrxf#}7eDmd}%m-TIFb_OL$Zzo@qAt2V5Ws7`iBl!*y|A&nP@A#*L3{mF0_aq(
zWLbW&XE=n=+~cmS$()EQt?GZ)jOEeWbZGFgf3(CX+IHnlL1Qt8tSU80RbrJ5*w!iC
zgJ56kohtUeGLK^6M)TW8OOl3Q9B$1>D$g{qW#X08is06?OSG{FDm!zkg=Dl*)fN8D
z37liiEV#FVXT8a1@F&4lagGJs*V<kZ8r=&|GdIaS5mLgmumQ+Uk$6GKp~wjniU8b(
z+pbwQ7Cy4VHC^KOreqcDm=t}N0a4xnEiJl%y3=q|uBJ)3p)|5o5?>KXWNZCFbiu+K
zE>a0(8f3x3U@uVT;*q=~ay)Td7v#MLz}A3=0~P(%cDaj&%4#l3aO(!4;Yo)lKiSk^
zuw_&76h@E#e+4467g^!s^B63K87TNGRtBc7la+4dLvz~EZH!xgL}5YjhqH}5Hq=}D
z(&T8A9T;!RAHa82CZlSZqLmOR1SAK|-SFyl!Ug9C2{7~z%I=w>$$lC`I^d|@wn(ly
zXwZdAeXUmp*{BQ&(Xg(|ji{?I)X2YzlgRgP<KhRDZ~xQFLI$zzC#b-i{3#bvpG=Uy
z4xY@#hgB84uPU)o(g3&f@4A5ZO-JM6WBD!3I?~`stbpkkq}zk(oM_iNk1L{nnY@qN
zdQQ?hVvJ8Z+8oeUC3N?)OTcU#nYN86L6CVXKPjN~qSl{MtGZJMPZ~Q``UF|$lbAr%
z2?zH^x{U@4c}2piOj}<9ro_)U<hNH{ELtzfThVWoHcW`QPJ+PE&9^S=XMqg*`G3uS
z?<(i<zvmMxW#$o%{5ZuR?AT7$Dq69R1j%r!!lRW5OyuGBou<s7iv#W;GtLJdj=)Y2
zjr4*ChReKyEXGt8Iwu|tW7*79EN4_7vpwMM#}jGT6NsZsmuD!C?Wl=kVs=+=O#JC%
zBnxRxO@%oNH~&lF><f7kd)DdMP_2ov@Rhx%B9TE%YE957LJ_~072O`T3?We$stOj&
zLeg^-Qx#srjtiAK12EJOF1t!RfC+uV!VD>T><~0B$s})RIa5ECn@u4S=K&RVnV*ML
zd$qd;Lxva}d&G&7WJun1`1Lw{i@)fpN<1eazWom>oayhQe^7~LjFl6|77_2WsB?3B
zYpecR#Hh0rv--}+pu0|H(l<rRqBO5GjZBpDo<Z$N*bV%_1p0d>N_lPG0f;Q3*M_g$
z6KyfGJ3R5jehx1}fFT&)Pk)<K?@PRPb?`$lEBUW3an!%)$#ZQDhz@Lq*NRow+!x^g
z^-Ev2F9aXWx0QQxrYKqw3DfBALb{48Hcdcn@spgDbI8YiLGzi_WwnV;i>pp6=5p%E
znhO&ja*qV)$T;sBM{T&7<7v1Qg|rOn;=}^)n|JRvJ8!)E8uWUJmTBdc2FKrlRxYEj
zn9M<vPEOUD#d*iyXyqs1hUWN9h)3PKcm~=DA%vtv0f#8Ex4G(K5%iFi9?5Q=LTN67
zQ~uw*f@e^XH2#r5I8hcUh?ADguZUMJtOhKVcwZVoo?=I8&e`0n+fQ`VcAr1=;xf;k
zZm*JzeClzva%UK$uAp5uqRh9*1@lxp8)d?|KXwjo3g-FOFYjCzjg}40up190{nt`&
z#?#pYIGyqcZO6FPt%x{WlFc;H1+EyiX$a<j__?#dg9fUCE?w#U*_SXK#b-$-PBORw
zI^hbik+|WMmgHrQZeEC<X#dQr8voIo;ERFGp3Ow!@2DsNIc;JY7>5&h<tsAq%O%%5
z!GcVY3I$0~N9`g8*$5;@tMf&pax$%u4gJ*T1|!6C7%vxk{Z84X?y(ob_Q{}X<&98g
zd%ZU>ajw55(sPB9FY2z{v3om%uP{13^J5r*)L~&z`vE<s^<DYITitH|#K+>>{F+Li
zIf^DGbW{-wwL=8hN_R2V4fjSSygEdfCTkFvjLz%~h%PFoKm~)KOWcm1Wj7-v9ndY#
zLwx;}H}u+HP4rKmC@4eqi0nubWvTte*bykcn}CC)dY6xpzm#2}Cl>n38#-ocVzGb<
z(7Osq%uiPuLUEMgQY|C81M%P#|1(8~-Rxj&?AaZH&#CwaBSRYQs0mmr%#OHLGN^<7
zQIHFj#CUj-6{o+e5Bj_B_w-kRdby(?je;2GkTaFWhdaGBhQ;-;uu*D{_pz*2HNEtu
z<Jqdnb8fsP6N%*?LAW4HCE(%HOhv%=woz@ES}nL|`rnUAmgXr3rV^yz@xo0@+%c(I
zp$X83u=KBHCnip{<;(l1<G4dH8Yn+VJjHjp`KH%p>@_SdMI64v;CI|vrXZ8tX#^jB
zX_dspzPNV|2`RK9-0nE$D*>iOQ$Jw&cwC_(kp)G@F1Dvnchf#v`Ivko-**9t^5&C?
z@DrTr3&}zEouhwXmR;UQ|BC-|+R61Lx2TzsNspNFnWQ1bkk(=+tzqaipig3vJHudZ
zpZ=BVRIX+>A8-5)0*{l^*A$7)KqT_RpX6>yKeN4+VaO6cY^X9;j-85P`R_%gDz*KF
zqF`9yA8|SXKFxDw$a3gwVG%BX?Nm%9r3P^x8u1=61*RIZ#wt_KC;5j7jhMk2Zqi2U
zvRzjllNs_rFc$&-zC`|7J1kcQvqq`1UM^Tfgy-w6nED0MiFP2Fshiyq*O~S06?W=W
zXN_*<5QDMvOTmWVyze>3%SGf|Q{vN3%~BD)!A;q#=g|$m_i+eP;7u8@;(Z)EuJnq3
z{Ql=m-YBs0{p;)9@cu^zdTU`r()wCZR=>>hq64-)V9@^f;MN@+UCivE*BSvqq<hbJ
z#y^q4>_l~DuEkjKXkaI%D_pr1Etiwnt@_kB9E-s}lHYJ~u3x3^HJfwb8m?;6y0ZsE
zF5Y4sZZDQwv8}?vO*9Rd3^~NcH4ix-k~=!?|B>7i0G&e8-*)W;|La0_Rc&Y;t|5R5
zm-Oak?|V=yS)Z*E7Ng%fq0vlZYnFt9et4@hdvJ#yG!ZqHNT=jhEMar<tCUorCiDM}
zC~tQDJNmFkay(i?oqXhc;cf?zl`JI}p}Qk-3F#o`F8IX&sUAvDW;I?XBUR#eJ~r&G
z?Eg$9CQQHoN$4|AO-5`SsH_^gR^`q%E$umd4{H4`k&Ni}lZeSsO0glH>jy0N=7mTv
z!Tv@4DgtU@AxAWw)nQT`Rm-PbXg>$0#WlUZ!?P7$+&*rgpx94b@elZm^tC4Oq@$bE
zuowa2IZSIH)1WjVDM89;2V&U+0$4!KZcoQeYjHoW!j+@dqQ%C^GxI(vcSo?)Lc{pp
zy|$)>f=`?+yS+BXjn8nIa60@<p+zTKw|<b9(`_xVyIJagL->8_Q1DJsbo%IIx;)`l
z=Qa9V59-V5^nV(i;B;<#OL3p6ZkB-G5ouzb%CraYkGyJ*F5H2_tDkT7e;1o<N-3@C
zr?Wpev^dza!0xRO%)34?Ch>-!F^tu)<_18%;vVR&DEcCT3>hIF%z&2p{Geu&ZKg6|
zv4fwd4;LDx@T>Xx;zp#f=&lL@DtEB-?>k$L>E{<;;LaldH?Q0uFn?n*<y@f2PS23>
zm!0gJJ(HJ4r5XntXs$HGbc{HY7R;xx-*#~>PfE)x9q0Wy)LUeFr&)JmfEd&cX(0FI
zsJADVB_BaJO+#Dwh|CaIH|tekwH(?Z+1EHJh3NCB8)MRjroI{G=BB2mbhqU0<1T)j
zinL;V`*UlMKipwBB6t~eV9<a(0<E`d60g-HS#}XG-padTxXELqqI?*te?<=|4lE9=
zS|s)sfozWs*6O(4#h#3!$#{<fA6fibaX@W$1WS{_rR`csiHEur!;CvM8DZPRsy|su
z!m9+b+Cgb0^~$7Tu>?9QPc6%_*)ya!r1ybf&n?*ww!`y1C<)dV04$R9C)8<#D9A4y
zG(fA;m^S8tB5E=C;!n#3k%f}M*G{l2o6E3VwXKkifwLz;A2yS^U>5rJHV$tEQyZE*
zK&1L<bmBr4ZWeA<sb8KzsN<{ir&Z5N8^EdA{-7MSsnd;VrIZ{Ib4fEylg%7Qo~;0V
z0(qK&<|xDD&Y%L`@gvAP+J^6mTfQbg5L>sApO{M^sBCHkAuMT6=a9qfI44&R2n`77
ztw|GvdvZKLQ{tJ8ynanz=ojr@yfSz_jNPB1VsL8qj49_)tqq3?DQUV9Q7a@W%*UYd
zY0&+9Bk?=Jv9)VnPukcKiOPS8YNncdFk~>qB+!rKMPME{k4~^c^Kpq9WK(VEMosl6
z=GIV~I_MIy-Qi3zCsYrAkSe0jKa(P;{2LrF)8lQ%T_@b7Bo%B%nXo{;<s>wCSoudp
z(sR}S{8Hc)+Oi{1%=Oi-359x<8JFY`MHtmCLVeMj49e)vOJUzr)}08Ok5WltV<!Lu
zwZL7++;|7DT2XbeZ7%5fh1)xN_6vsZ&NJ^Oa_9l_R;P$;n)NvZ2ZOeh-dfD<j)%eQ
zPrkw{aw~YXQ|$pwGwdH@b+s+5k`dDM-gPVYpW2hd1omPwL=F>qa}J5sU}X|)BI_u)
zhx%GjZzO7A9<hH15wg%tlRzQ9(IV$}{eCh^Q@U~lz6n)g(pzx~%|^!|qZJ5H9<uZf
zqDcvRdSp(kYTAYClGZmT>@|d>X%^&^p(74Yx4Kn9Y6-Vs+=oH)d9(epVz{)rNfOOZ
zwn5=fY)~ZJQZA`Se6V^G{Ql8^(k8XYny0t=iA;;vKg5MW&R6IC%@+0;^;m@<wA)`6
z`w0>p;JBLfVPa9`K7{Fvm()27{k1x6mHq%@$qd^X`6l;gHfI&SpfJNwoy`S>nTdG+
z85^kr>k~^jRIJLCG9bmG7dWboo*^yyYwx&yR~1APOasZ*kBd^)Ai?X$uF_Sonn9z3
zZcQRbGSz#be2(x??~q|)`hJ6M7lFB}y+$()VE@vupJrV*RX7&N@$Y5K6g;s)0Ix)<
zbaWzA0A0RVfaN1JjwZnJ)jv`S&H(Rbaa5}kNw4ZwvwGWwK*<Z+uQE$$6u-~v#<J_$
zE=E+Da;5*#9NN-37#n1m^0#HcrfkjTV!?tScO{sIk(=<wvx3vFdr?j&A3j&JVX>M9
zC==_2RkEY5I6zG2k^6*YZ1ESCKIgdsZG)P)$aTProtCy~hF3Dz8NLtm;0`#zf@!*K
zKWL{|TAlG38ocra$(hwz5$HL-uJ>V2^)SewXR<pgvbZH$fyb_!>64``oWfAEc-i#}
zloIQLVZ@=>?v6&#NHS8w_0B-C{AUsYTxBdqT)TwhHW#uC_jdr9pf)4q3gmtHudWI1
zcUeMX=#2!5qg@@2jM^)#tu(W%P05AYwAeKk!I{kRk_!kA$;(-ZjZpME15kc?Q4`;h
z%58EK`3Z5J^$}$V`BcHg@2z#$fn)S?2Q%IEpb!ygusm%jfrsD(z<QlGfRhxk$-MWr
z1Tez3sOQ*8BR0Ip$;f9pQ0yd!JP%zl3aT`+e~JW0bKror@w>O7XDoF%5`tGm3;*E9
zt28!+772(FU}<djjJ5r`DN1nOJkMwJMd)F0PDN)(>R#K<OnAM&>gF+xRFrGxYXhI&
zhNrK)`JKeHTr;XGPJ%c7!VcuG_0Xo*qvvVdcjvkOe)5J%c*ItpA^*1dJLnXit_)={
zRrIKurjcp{$NE%#e5E))Q1qZByR2Y4Oc_qB3X`yiY3fAPj|2<Xf0*3W0`)BvVCk{8
z4rI9(*^h(tSRGkVm$=Go>bqPOwQt@~IhS>RUGBzNKnYKgn5rb2{TjAR!@tC^ku`25
z&OxLIc1r)MX0;<ex3?rkT}hH=`~h3q`SkYHv*cQ}V7ok$?0RE<X=W_hcvk)I`6Tc+
z1ss7z9~mCH{i+P-IC>t><q73XfSEYuork)oiwEN}a*W`5%iX!Y=hX&L<K+kDlhYxp
zi#lJ$!yitCkKe(Z!JI&1D(8F2RxleRM&;GaN~rkcl}97)!l_25em9zo8oMP`A{*bU
zij}bXquErT|Ies1xVsZETJw-$0+#o9U+F)PEB+?^$u34ux)asqlFSMUW&_9+dmfH-
z{DW~&U`UM=6n^YR)%&>Qq4vJ;??Kza>4Bsj&dxi(I<Ccf|I`9-?v6P?sxMs5#O3K&
zMwz_^+>{1A5v{=q$G4T<Qhnht4R-ZdmNs@q4w651V|HBp!=nY1$r-JfS|$<*KmB}c
z1Ph%CulCBX_AXdD7|N{A2^kKQQkfX?m^G4D<HmlsczGc?El*dBLOK6aBDh3M$5B0>
z-KlFf?bO7G8ioe41%KTrxNiKmM`gl9iRpR!TRgIYg=1a$&64VUS@>m5Pe&#zBEn+;
z6MHmcl1m&Np`_t7av9=BLsB@*KLyOiT)s;3ikZX5y0M@r7_wqa+8?Pg{*(({#ge&I
z7BahC*mA9U$1uAcq4SLo;`P5)39Q8x5AgBRb8IffhOTyj4b#pPuoz5fXrZF|h3+fQ
z6v?%_>L6EBnk)0OjkH3|YAm}05=!mBNcvzeS_RmFrSIGiz-D{59D2Za{7-r#dM_ua
zO}DKBG4b@0UtonHrL6-j@dvJ)4fprDzGr=l$E{BMUX{ze?=g5!f(JY0TW4D|$Xn2#
zvw}8-)fm%28P>eW>e4zsvB^DE0RF|A$;m0X3Q%kQ4(35n&Pv9ff1C66urbR-il|^N
zF_x8eKeC{37R<;~SJ(fq(sz_WPL{Uw7<OjBzllmmf}KNpn6JwjyWsZNBJl6uvsRq}
z&MO=x!=0$jlD9+bE$}j}A6?X*b3Yu34m7K1u)~CbU8^08?#|6O|B^JjlZB4y-)f|l
zVWR@3BW?e>w}qD9<9~!Q7&_|}cKpD8j9G-otja#>9_~>LzgG=5xb>Q8Kv#%UAYx44
z?%Hlv{>AyZ8-5!k*?TDKYD@gV>3oOM3P;?^l~(h5cIJ76XPSjE&gPeX6mw`K7J_Y3
z$6g8p>`k*m6oZ0N|D`ETJ^KzZpdvvFPUQ(wIv~QNI0h;koUAWb2$Di3*Qq1(E5-2c
zO>-*shLbb04`To9c+{QtOl}CNQc{98<9lsTQIskZ)flKKsd($bC<4gWPDa>;Ai}6+
z`g$A?IFhNZ4n^3;7+-C&qdo)560}<IbRd#Iu*!X`F^Gg{Xim@&<$&tU$cJsC&bsqk
zSVZ|KoV`Y1@mkP*1-T|%3&j!a(o#4QBR>8WY4|V5_R@99&4E2a&j`rJO&UXupV^!{
z9L6@<CCfnrH>lz<$}!iRlBU@qANBl3Le%P3xS>%Q==S`j4Hodh)&mBndJCKzlL?JL
ziI5!K$^(N>ma!QL;><j|^p~~0O5a`sG9?vqtS+@^Ja*Ns|I_g8!$5y{sziY+BUtw>
zbsW)N4qftGnYWI%0@b4kit%fXGsEqXdh<jG-8EsmO+GgRVg<~E^!osUmP4lU-Sm_2
zn=iPfR~Z?<2gjC>FtMuSIvUhUPFf9MsV^JHL!m(9Lp)yc)9a!FF-734ZREG6f4x3z
zN?3@^#jgNw9G;2?InFlA5>^@X6twORk#Kw!65XTYO`!et@1m(eJ!h!@^({YiTR2O#
ziQL$eR9J=h82b0Q6GG;vT?>|#jzH9H$O$Cf4r6%ruB{CwTQ3yKpJnIhd1m=Qe;e=1
zDrnU7XWdgZmQ1~Y7U3I}dSyUlU)$B?dY)Z`*A+ITNAkjR>sIfDzXo3~gbojxU4RkP
z{`Hzx2d=WZkMwx1dBExW@W0^>?dS1&mA>p|&V%at4OgEL*A`W0=vd6}TKl^;l38}K
z_|SgXX1MJVnuIn<Ww=p^wn)4H)hT+|E1wDTjd76utpr1D32UQ+0NLXk!&z)B_r~up
zPH^Cu<ih!w^uox`6iIf>oF$QYnZTs@mmYztT*FI4DO&Izc}o5avY$_dt-9Q(E_pGz
zgwT|DWw|4p#IrXhoL0E)An}Hvl%G#FS~dHh0azwO>lU#G`Q7RR#Q}cciJM!GD|3vG
zOkNY?A_^iC8wVXnEo^Kg!5qpC><ckN*&U?G!6W(@@=5E`HtKeiGw-x&7j7J_Co@ID
z;w!^F_0NN>iUHeZUV{jUTL{e2Mqu)WnYiofdrdW<#V@|2@-QC)jWc~pW4t3wwca9}
zufCK01qLSEk+>b%27urAt99mb^DiF$I=c8~j)CjuwNGvpLZ)ZvTm53TqXj+vPL70M
zXvsLpGtLKxXKPPqu79gNiK_+p%Riu32am-v5h2cK>h9YxvGV0l_1*28eP7}n6F4x>
za93deY|I+%%U@mQC&nIV{3Cb?fqDPKKsq2ZAVX_GVPvvJvkO!k(oMK{e8diS8Fh1k
zi-Udvw)mC^cUE;wO5_VpO%5a_;L1LUaaj0sp_oCK1EnV8vf{nEHR+~w3vr`=Gn6Pm
zv=AvEDJV;8{;&n0BxD37{)YuTj!_y7iDL!=bOZ(nhE!Fo;`JT(Xx9HTRY3U%oLoOo
z@`J(t`$92Aqou>wpm_~a|91?Jv9U=`O!IpQ_3;+0mwz~mtc@HV;duybDaTKWuvf~H
zo*v+4pk`&MjhS4aUyJ`Rc)L}k90c%DCgFTR8!zK&b*Bu)3)JF2lyP0s(Uz&om!4Wa
z@t78}{4ap=%y6kz56T5p33zH(@I=u60qZD+W69c3^8NUT8TxeYyjJg}Aj+!W4G0)G
z*elP*RSt6sdk#z<3OjY28xByVr?iTOlpiOX#XU^7$ma1Fit1wdl`LQ%Ue}RKu)QgR
zRkp=Orw+#Y_oNel#~$2^u1DNr+#p*-n>hl4CxX5(a=_B&hUty)h4TObDkM4Z+5BT~
zs$^_x#v0lNVtAO7$}Vdjwdo|P)vt5)oXuJnEsG4Xudcf&^>+ncOa@KCpp~3&GUu#^
z8K2cuBn>Li>}L$WDNZ!*iLRMy@|77u`)06jm{CVVB%(b=aDLpGBm_}leWUsSto@>O
zR{r7+m+xgJhKYngN6?QB?XdO#gni^kv{x_~N6(=q-6XG4s~5n(-?!eJsXt1fCh<09
zB3Tlc@2j{GH4nU3^4dn9w9^`0WWP~Lg6CfGEz(ai;}unMmd*6_7vl;5Ni%1Uzx(bd
z$NxCVN%bPTsQw(i{3we9%Kq;FkSluwy~iKs(YKo$Mo-JvfE4689A2Wrr~E;q(TVq)
z*CSf4&-HSDc)o>b)h6{@K|ze_yZME&T%gXW<Kr|@bd<0F>CKyf57%J&1yqBc78kz9
zi+6Oc9&!8BYQn?(cw4#9yz`b~6<av3ZQ8^Enl8$Vw@Sag%6%$%z^3OXV0-g3`|e+o
zP+os7GddxBt@OxN4fRsFn3=dMw6x+vf$DKQbK|M?w^h;!pC$WJJ1EgirS=p3a6{6&
z1A|nHr8oq1l_pOsmmWf*A+FQS?alopv#xPZ=1pV%)vWcm*A~*(E?nP{Zx9c#tZ(dn
zCq~7KLE#?v0oVz#52nrazzj27mRS@#{6Tyoj%tSugc$2h>*O+a%vj%PQK|$(N3@Ed
z#6lSB3XNH2=mDZ>cOfz4d=Kcd<RSaFK<e~nEYcr1%b{$_ttOhylxEduI(PzdU6%Z7
z`i>cR!T%c?NQ?rH2Btv68c%2(vao;3(s}lP@j1Z?yR*7NHUYy}N++I9-IZISxQ66U
zx;PoqS$BjK_spG4HJg=bdKHn(o)h{-Px7xwt(W}qhOcetd~PUk+Yy^5n~(EpA(vOF
zhmD}wa4;K;OP!Lx4W2icFQ9Dc`rGi~Qk=SCHs7$j-pk!nsdTS}qfVXo`Zm5oNDk1+
zp0{km<DbR;3d~jMmKE!xEH(Dk$mnLW*t(f)mg6;clFv2ERVHnCGUb`T6;6yUPMsla
zYLH%;GVEa;P>|PZM=b9zbV3tr<nJH<f5i{3fNdYk=lHJ>f=@W3fwVZWazyuvHPiyv
zs@aS}`?pF?)n|`90>jZ3Fb!dpW#=$QT`pB`@P}cv9{Zv!Vi1Sjz3Bqja4aLU=cW>T
z5V9&9#YWEwHwYGq%}&jKiQpqeOCop-#x@`yqD&X&%#jCJ;$_+12f(%}#J*5&4BkJ!
zElB$me`L{xJ=-t-L^_FAmu~JXh|ErYGE}ORH#0hYFK2GQ*)p&LzQKn#Yqsd-KtW^i
z$V9&18$u<}K@iDf6~!G|bJX5=Qd~5uNvA6E$|H~{a4`)*KPE{Flgf$dJubBDo4zZQ
zD&a9aTi!kQ@Rg}&?C%Te%8DIQV{g!fhlw)`Yk~*}Q#-qL=3z0BmCmSEdBUY;kqq=W
zhsZF@r^NNOXR}fRc+MF%V;{`L8xZHO|IWKXJo9Y31KE2Fw;VecD)V6@JDo=*U5xu$
zH=>~HdC<0I?Pb{`EV-Wi#`4$f;ZmRb=t@|2!@l{meY$_w&-zM895SNNhj3CNq16F`
z3K7EZzX{E#Ev^mZw%yMW6z$&bE(ove3T8<c=dC8j*zqlZ@`%Tjo#-`IO{XiA>5)Iy
zWB1|x*)^Q~!><PhF4AT5>QheFqO{SR8EF&vkgTP~#)-A%Ip@OLW$p?Dorm)1eo%VL
zQWq@*+NrMd-|1l5v>Pm}YriJub!EVp`I4@%bxM$7^!F`K;gy@C3j8G!$tC)6c6@_Y
z=x*Yb)rA=h%>G;Yd0J=5N#(z*Bju&g-xhKubEicmjg>FZ^rv)Y<QRh*mtYU&7isz|
z*M!dg&nnAg&?z>Kyxg$*RA9>BU~?;{B72WD7P-+858{hoT7_u_12sqb^$@V3cW1S2
zp83J7Z8gKb*Zv*!eXJ^!C|dI0jt!Cu;meWtAPBxL&^Bmkm>Kk7g;i_Xt~~!K^yK$=
zzZtLFs{-+}_ity$c!etMghukpDaW2{b8DZ{Ke!rR0;t*sDjHK$bgg^Xs;(1<WX_3Z
zD?-2Q@ODNLxMmsIF&=0C5mxv<0SO--{LJXRDte)<p96R|LhGanWY&($1rmlF8;$Mz
zS<}mSK=?Nj&L=M1xTrgUiNjkOw}rgpax-}jDxEb~G7MYUD_+LD^qT1GGL}Oe*g`&y
zMJ84#j#I+_ihW7)UE;UU27Hf_un)@Bc;C!_L{E<DF}sT0Zl8K|4G8Hs1s8vqyQA9U
zYFG{q>01xMc024gYWd3|ecP)f+2UNe2`o|oFy+N*{tK*>?K%FTStCO$dSZABvT3lo
z$631q@Pqb8gDa%>;)Z=!x{xj7s^M=>GKpAzi&u&vW1Y~#awu9@drE)Ok@(+yp1PAF
z(=p~##A?$qk{p$iP;lz~9OZ9L_fl1Pi`jKkC^Ttm;K<?H)R_?*%l$fMS@PX?exywU
z52lnvs&V3n@_u|KP3xZ{>7r>fb9^fd?&0BQ=c>51KOSgn;V?1gw__yRV$ALM?4HgZ
z2q)hA+T2VaMKaoUQQ0R$LNr(*Gop0KW2F;71*BSbhrFUuTYC!O(o=^h%Vw9>#SPPs
zUS8nahn4e}L6jK2E5VQX@SzXPNO$=G4%)K|^t&?6N29^=M)%)8B^~37Vi380Iy)Uu
zhUwY!uN;)|oc~2UK%jQ{nHx5xP+FoqekCZ>9u&amEzu`lrcX<;0s~*Js(@}B3HMcz
z&}11$PMM4&Hi%OS@~daJno34f!DgfV>=h7d5SL8iHBskRH^w4Q=nmNO0?R=H!eaGJ
z+I-@$;F}dQ9f)v{Vd0@?xW1mWyrUSm`QTg<Sms<^MJn?$PR00P!DK=LGA;~3KLr@s
z$}eFp^FyTkgI9181!UP$YtnLywUY)o9%dmhCHWKz!4noz(EYT8BYX3d@PZ!4hYvzh
z6bcN3824%d2+=5xB#xR`r;32yafILPRQ8TlDhwKDa<%wNWLS~yD>AW~t-oJQMrfwG
z{{;|6?<{#|YG>OBHW(}L$S`qVZDP!^e6WNaR2rXYU*hkz>YE&m++i&Cz)m_~dmJ+x
z9gJNNc3rgq;$EL!XS7D2EWYi>k6L9L9q-lGR2)9{7gEVr-izt+8{dI6-5Z63<u$3e
z-nc$az2h!=;I(Y4uy#SMi?OpYog`7E5LuQ69zwYyx=Ze<uUYn@M}$YJip+FQ(6*@;
z*P+U=cIqznYE3NPJn-Zl4@qyr1isT)Z=xuSMkS+t9M*cClw|q3%OCrI_sV&|pg0q{
z8-$y&tBi+i<-_d;LmIdSZ~ZfB;;g-LMgvo_UnR(J_LH_P_j5o;NBX#?hzrA399f>K
z5>fE}8v1Kl&*g6P8!^(t0ZFBZ=l*Z)opU=``My4bORC_S0NFqeM+Rc^G4<)=J)>kF
zW2d$-dUWa94G$N!wTCSvN()O$nDfo(ORk*r2wSZkI^Om9Pe3<0i^uf5|MTg*R=WlQ
zDyCJev8d(GKnV&B5k%)4_$Z4SCR^I1Oq$YileR|-HCZE#GTV#ICJ!l@K&0n{qQG-<
za`cP#``|5wJ=bpaZu*~!jVLXgXK|91hzSQww9e09&Is`emUNeGn><k-YF~z`H1ka>
z0{b{>^nkY1a-dvyA~S2QhhYWm!yqS$+?;ZPiNd}cmO3%U!mFA0d^+w~^l5lyS{}o;
z&Qb!?n&@v;_JQwzvAYO~(PP{grP?cp9Q9X_8Ec7zU|avZ0C>|Mfsmo|F70iWQp9G*
z-w-E~L+%NgaV|_%j+<-Oh2k;D#F>IAqbd>uvU*$mfVaGU4g87*ml`n0JlB?T;d=1J
zIBDfYad`U>CwFv|LMFzct``kXXL3wyL3c271mQ1M>oiWF;*K_1y@$Q2;rbX_@p2$E
z#Oq4dkCvz;$2??>3dxKI6jPshlCCNyaX$(OYK?TzAqBnPpv>qsbja~Ux=fM{x_6v4
zWa*>;dcwtUMd$7j%6|`44-P>#mP}u>f8>bvA>Dc;{g4OSy!PKhxb=XJgx^0SG0I4?
z93_MPZ-t%X7`yLDYV;)ZwQrEW0ob?M|Ms+-`2F?jrhbYeQiCF6+d7LKKlq7c({UcG
z$AJqV*5-(wciD~k+M**s&soem68a6;fz;OjYwgeSCw9PB(wz%Yi*5bmeJ@S5i0Ic!
zacIRx9gSa!iwMmXFKZmpW@rH;9CUqxLxp@CXye3Re4=Ff*uTVYYS6sw`b$$8%U1mE
z?6=+vVEJ(*a(vG48$^490u=9eNw_oO3#`bpqA2J)0{hqcHChB9cS-HY1tK)Su)w~I
z0*=`JQug)DZniJ9z!)B<p@%&5#=nAhM%L?M1n|4HB=45jUuZoylcB@+0;VsQaH-OZ
z(VsDAy|402i47*o^MPW9S?oV<iV8;JA|)XgN{uY->5+MO(5FgntjXFyZT?49iyM>`
zr7IhWiSbb{jJRdPg|SD&$Oisb{-WXA&63*As>4)(XL}@-P3?Oo<87AJ^^cb`%)CQc
zmALjUwiEviUJip{*hMcuCz|sEi6E7nmW+RjQ7+i7Wcr0>W(-@}0CWGkzz1G5mhF+Y
z`u)8JH@VrjS3B;(dJG1U|Ebw|kGNLWUF_NK<P~e18G+;TJDNHW#v8^tpk;lE%s0#}
zHD@N*vz^}d*xLM(=8Bm}pyK%Jb=C7R*^$d#oiM(Ide^bYJuN<uZ7R_0#fsd%8C-(8
z{J^?=@XrEgXrHuU?pG<8`IBR}%*!m2dg=IFn*CqS8<(D#*QE}n*-U-6ku_?VifHu<
z5NXlWP=xmBB@_*?mG24hGqwtggL5lqC@1#W*pY_{&VT2+Cdxkj`BrkyTv#hAl}Q}W
z#=&T=lsC#x)pbG=8&}EFaJ8>^Is_$?NgK*FQ=i5#_HPN5n6xm5d{VPfsnBYoeg#rc
z91Vh768*N&&>i}7>BJEUY*rTuJ~7CWF@!8-J3R~AV)`C{k4H&rC>rGIMk!H|KW1>Q
zqMwCtaa(hXS_6SaM-}DI@B%$gjMV1n*{OxkC>Y&4O!s$dKU5u&fPRT7&2j?$`eAAi
zo9%kvO#N*qw(3x+_b(Uh4E)c2S|T$mSc18FC3S-faP5$qUS6{G*Jb=*QA_&GL3khi
zQ7Q0sjQ}Q5D6qD{SSYZB@jOIdB`SvPp!tO+jstrbl70m&CPg0{C%|mLs3lAq6SyR#
z7Tg~yFp3u^%>Dd+xy4*tyJY)dEnX116jrwr&y528*nyxH8AnY8{d(1Pcfh`j4}e^@
z3|-)Pc_zH{DHqTkxYl9P?tqU|kK9bYxJ6KP;{#%?OX0ReEv7h*EY;oYhK;%!y0<s+
zdMrD=CY!0M2Q%@<Pq^)qdF)>wynx_rqfFIb!DKJ_Z4Mx=@nLUnFkSp=2?{%lEs+R8
zfC@V+GQVnNy4AH;x7vGerjC*)=X&zBowlgKDBqNalgRsBlq*u~GsMjXJ+BgpjU%mt
z7ubJJ=suD*CCTAI+E_a>+|N1scPO|hvig_*UbAn7m?+$?kelZZh+`;9i@!9ZlP27O
zF9&tYK!>D&<EQ33v}W`UihRp2f5SXO@6a}qaFf7h1vi}AynF5t2kACW0vt3Xh8ZsS
z#SC=5Vt;JbjpvOLfPUD>VAg1B#)wtn1mOM_ljde$nOMGDKBG#ft|T8$2Y<VYZ(hiI
z?kdW}v9B2{&dqx?`<tM&PdTnS+_X-tUplW*l$W2MUvQ6H{0a|=7Lo3<m+#X|ftf7|
zQ55Q7?c38_a8+P43OQFb2-89PxVJ=7?B~Azg@A#c_iberV{G4S*w|4?I6XI&8K5k;
z9uc4vdHl<T>WG@U?L}s6HGKw__1I5*5zj30!!};22NnN}x=gIdbLdfdQP_bi^A>e2
zf1bMHwxguy`Ktfv2a~@BIzNs|+2&piRkqww_>%Oy(aWhn8^G{#y0vX75U{9D#r_(O
z6dtKDIDVSr$4%F`b9?0?9J~hJ0sO5+zC5FxW~;Ir0pB3LJ8;_RXA+b&`c1{C_<POB
z_(!sbQYClUH_6rx?Z2UNt@mS|JzAFyAMNgR*l<>5H9zBRi*7nqE9nakn&sZzdJC86
z)gNNxP4}1|ZK_|@a;h*_a?oeTJG6E1XBx|O%N>viWjV*G51nOt`-F4dS%6r`Z|xnN
z-7d%g$>--A>NkDb)#ml|^|C#;A)`g@NfqL?X;RFU?!cyNDB`T&gU0=9sjkYO&o#V-
z^M9+%ZU?<_b<G}um#^{6X54o~<Ag*%(`(8Wo2;g-!cD7&5U=+i_Ryz2>!W$4&LIiy
zUnJhnZqU<)dzbse4Y(8@^?|p<<FSQs^d~qV@Nwji{0lMQG)RI4c-8%~;F$y5b~gh0
zATEi9N-4%KCXS9anIyf9xel_wU29T%PD?4%rPA6HaFFi9(Q>=}hOeBVI}`qHt|Bq=
z)dx{(w<qv2XnzD|acypz#f<CO%B<LMC&TP{>EyT&ZoBITi>7er2Y^2$6!(QKzNrYa
z3_qZ!Y#41LG}g>4%W2r;ONRP|(EG6N67dKP(TK}bvlkUtjX(?`>8-janK`aeDl=o@
zmu0@JjhfEb85E5A+=sijb-yyv<n`j#yJzCFjkI$AxpkF4yGboHk3m4UEIL;i7<tmJ
zEuep4WmV!c)oXkR10Z|0&R&1WRd{)U(1Ds<jn_zMuiE97fHrc*B4DNe8l`;qf!}sV
z{iI~Jy?#Z4hyND!Ib|CmE_I){*Y_-;Y_cskBi2lkQ7kymsge6dBJt>cMu$5e9Y@b)
zYkxl{rJJ=<z+g}u?RksjhWtI2y*DNjOV~x}ZilF^WR@Xk58OD**>hMfn!PaUtoZA(
z=SLkVgP9k|Si`#$lu#ByaVZxdMaV*jCxp*pl~?K4&l{^7tK)-#rr!!nOO$w;&Z2+P
zob}C*^=Jh!C9<`d*;I>fyBO`Gw31vEm!968$t=TixuV%3W2d8*bGdBnG@nNX1kSbI
z7dfH64bL4O0`xQtb_cCLrPa$|+pNyfdL8~!DpiX13*r8<nsD0uM#+v)fSLX;NEKpe
z#hQ;QVoxZYT8^kn&FbzwmHc>{d;U{YQsN`m_R<CfX@hKco|=Hfiyibcm-UF|d|q0M
zSY<mG58QrsbqB<~+!Be<v=EgwO0wYn0VBAok%k5s0cKU=jH@fmHcrq&Z(09G#2z(1
zB6erO=Ah{(^t}zyIKsRaI8_62a8n>~n1Na<J+(@p=G&=m%#pLN7&q<o)pg^}`@ynv
z%byF|D?dIG_CLmzUG}05)S!O3?kui_;Kj}Xo~p#vVg#wr1nl)DIUxt0J6X6V2ekq|
z2!NHP;3n=F;R*v!lV(h*xs7%h7b)*zE;|01k}JLf{2{c79x*vG7j`WyGhnH#X7ooP
zlcjm)Is5%etHt}4xP%O`Bb5~BVx&Qt*Y*WLkdVmbHOW0Fv%*lXw@0_@lH9Us9YHDd
z^p&&VHB<NrXzC2bH_*=xC$BWB7b@-#Euc@+Q(HC)?b;pZ%hSF{VqEssAgD4SU$$;@
z+R>-M#32FKRz(yib^bQEkke>{hJzs{gXE0v<|jr>^h8+Gk4!+Ed~ZSo(el6Kpj)Kw
z*2a%Ymu)p7$MxPn@wf8<*V|z*JFDsx0lXY-Zju2Mr-_V_(r{KY{I%8NbEg2x61C!A
zg*?W#4|kNhnT6wJvkPaeL{*ypNu&O)bG8!XrKsj+-4c~}DRQJ7)bbC=9HCq#h_dK*
z`5nr4E<cfPh!EI!tue+$Ag+c|9Zt(}s`a_cKN1Ed9?2D*{6~^Z#dG=b)g)!B_dHEq
z)7midOTl}YF$CXIgzl$Q8<hb8G>CAnzG=D_R|FQHaJ@SwSi>{nh`eOBjSWw4KRyr(
z4>xR?S+=W;BZh1}Y<(PDW-ZO$1kgc(mtCC$>_12T5k>@D5K<Al(MI-<x@G~wEPguX
z?ah*;3IBM(a|eY0TBdvEdb+#nc?Mz>EtYRJ0QcHsFk{w$Z>Oto>k<%pJrHilrti`?
z9Z*3A5_sZTh_K|2%5s3SY30AJu6z5gh`hzQj~{+h>(yKGw%9#p;~5<95E$+tKeb*<
z@&T51Yp*pJKE3s7=}roG5j^vIeH*p3laoYauAPM0zauGteioQGie|*#zYD{v&TjKw
z`AHxmgna7*o%i$Z(F-634@vszRut)<-?1wC-*Z!Eg#Uf5XXMTYT0R~XE^8;?JDVO4
zJ7HNyQbG9Jc}^VoZ8MoX?6b}dYS`QP;D4)$pE0Qg7J|fypb|2_x{k$TH1*F(B;jdw
zS$UvTTat?3gD@EFvkZmt6rFAE441|dqZ>HgeBfFEen^oR*X4EaacKlvu{;}Hwbb#{
zeL1V_TkVvY21<R9OcKc~HW3r~j$;_Hz=d$4+wZ%k(9g0H-(m(?q0<Y($o>BfH-g>~
zv3+q~zZ$-dZPMQ|GHU1l?N=>Pr0+o2YOPL|1`Tw5X@qhOC(F2H$imtcOM`cOajJ&o
zhAvxxaO^U-AKZ?JUXCx!-7zUTP#(~otLjaqg|8Sa)_~%|yn?)fZb&OG@gBBnvwp|!
z4_zuu$l3+{3I0K?Zl_;CQ9n(Kl`zz1WCj&nL{>oWqfXJ98V-ZfOIRy|(i_dlB;oWb
z<>XpqSl?D0{bwOl`F2bir_Lo2&DCRgz|0lUw{7I89jcaGQe{EBP7^bg+o?PH)^L#|
zhP9DWzoz6FNBE$k+E{NjISqjiZetvAX?hf@Ooy%Kq9HrR3S%$Y(z+%O16}LyR<ZB;
zKRUHwBHNn2<#rk5HAK}NM40KVkQ0b*LosBlZtp@!s4%i3N%=`mv;oIoCQflT4xpCY
zKEB+{lqagdvkNFS`cKiOY&h4N#*)SsB8`3HU1n$+IP$9jhqz-UC4K;R!@{9ToecQD
zh&spMK-!>PZ)|LAZEPDG+qP{x6KrhTw(Vq-jj^#dw(aEP{l0VR)KtySsqxc&_tkfQ
z4(jVy^LNQb#owl7b7$D)<|37o*j64|snujSX3ye;{(>Px-1WG723fTXB8?p;uWcK9
z(rIU=K!Gi37Ju9}+{#*{S<q6%0~`kawqlmmrF8SD<bz_keROU><zk#VRb%f3rEYT7
zsos=@P+@Q4e$PR~mY0znnsUmab#d>9HG!hjse_>2xUj`+<_SwUR4QSNWj-vceB15W
zP|e6^sWO2iqW?_Ykf?{EU#b}wtA3R$T<ST`d%%BL?di03ufhJ;t&cV@3P>F;xjnYr
zsvh5=m~59V+HK}sp+)!xXM%_(SGi9}E|&fih!Tj3Oa{(y_A9xxm1|(T=~BgQ426y0
zzVRCOjP9YgduS;htnQraIxq|~+lABO&JjtoAQLX&ImygZKtH+;=<W1>TiFZO&=5B^
zTJsRGdFi>82@@1}G<(&`0Jy-h@Qd&f`HJ4Hc@CnDWaGyLujqRRx9{mex<Wx1Op=Kz
z(WVO<%e)fptI?>PgfXDu#<OsOnvOAxMiA9-B;q|&+2#MXfTP8cL|zJF$GjGeU=Z<G
zY#T|(4+(tRn%LZ)2>xZ^d~?$`6tK5XKRSLD{;>TjDxoKIbMyZ26>Z%(4fV8;F%Wkq
z2kw!nNca&Qe1$?_Yjke0by9y-!wZxGou0^&q45Y6??g%o+NIHh74V<M@Q0E`uaG_j
z|Do|$NS!LxFIqsCvPuWlWOR(y9sJOw)IqO+0OlY23_bs$RLK5s)g4dxf2!_evdcO9
zbe$MR`0-?~K!lqJC^SwKOca%nvC~K-*dqAXh8E1`C`<wt-cFe^-xB+`Y*EYlu2iOZ
zdUn3VIpPCe6Y16u7b(W_<dq-@!OY{#%tP{x9)2%4otUSNvJ;1>qE@|2WqTTR{Or^u
zOZaRc0!pQ^Huddw1>Ma#EueCx*}W?^M3WCUx^WJu_#0-;Vq3uYX7lhVs}d^x5x|>T
zHqq{`DkFS-kY5YqbgRqRIJB`}SFvIA4H>5%zAgy$R011HDx}U|N6EdwI5U^^Mr3rZ
zX<2H~MreXqz~{HaNLZb<T`@UNFlj90K@vAT%v!=@&o9>vAMz@EqgwCUy?8+}l=n>K
zfweoJE|^}=a@&ozLxaA<pc@UQn@jq^q8Q19Jgz2OrvfjfT?2Y=d5nC?_0F9)$7h9m
zi8|TZ5kK!coELlJIjwd#IUrZwjpM1J5iamc=})rIFxxM=eW!ZTsX&RRh*vWvZKrZY
z`tTnI9VFz1fsSDVGL)uAq(ji7*vQ&pCAo$`Wq8bUd=)_?<c-r^li*@5x5dvoxnB7~
z#Zv_irexw%0=lpXLKvtFqtrSEYY)j1c>pIdM7(^3#`PR#n5w>_ML)c)J#6Er)8k7H
z`rm@uM|VD#Vj^r;JDv+_|8>g(`&RdJy;t>MKT-L1wT?_rI=}fJISIYB$$j<dPyM*S
z;0uo9&yXza*B(0O$YQBt1!OtQ0=IIZ<KM$<80_NE)irKR0$&0J+BR-qjog({DYrx>
z*<;_AXuR|jtP2W`f%ZDA{db|Yu^};x&faw=<o}Waj^U+fb4HU&S_{3<u1}y;N`0_X
z%xzNDDmBB&HBm|leI02*z?p#)uwf5Up+iUAER>}nTIF-bn)%h}VMc`+xlUoS)W%;Y
zAGU=x;IwV%kt^Fc+I-*KKD!?;O1=_ENZr-b|3MaH;{;}y7Nbrpqavf9XC7!18X^_L
zFASM){}!r2KEOWUqG5wG*u_PJLcNPM&1<_7ve#%;Py9Dtmw3|ow<(u9plkS*+J`8G
z1%y^-L&NFSWu2iN8}~W)Y?0Mmzmsg%n{u~^EG6?S!H%UUbV{09aqQvVRr)yBuG@}U
z!u`p!)Gy<CICJ%ZbGKIogL<^R{QN*=$u`K%NOR$(4)z-I_WD$t$wHdX;-J4An!>AN
zOrZS$zGqK$5bZzJ$G-OfNrjxNA<-)-BNdtqYBY`~n!izCc;QT)Q+h3jHz^NWqYVjs
zO9Y(Tb-9s*%i5%#Xtdz5F-Gb{=+R6xVnC+9_Bc+Z&TGQaYjW}>tD|qQMyB(3fSnha
zE~k*2DMKn1Vo#hd$mEK1kMrINx}ifXb!LJcXY6=Zegr447!8176|GeD+!vPq6x!k!
z7tSf4D}L~6@<PM4K4Y>?+vd26VRvz?t0zGv8y~1!mLMOd-c@VQqg?p0*MQ%7`9D=$
z8L%cpjQ3xx+RP;#ISb7_to>8F{9OG<=%mr%LfdF297vQ&af%(H+gEq8yt>eK>yd^0
zMsSsg_HE>0fW3^eHQcs~<>{)&8;WTASYu0>5A;;GN<}t#@pg{eZ(uQ!((C)#r3Pgh
zbL|zXlRJkdis!7&A3bP?)^nb4)x|P4W;EqHm=zPD4Z_Q%si{D3=4KEl0+<`kemp;-
zlk3=@5)6_u?UcH(jIn2Farn>QC_kDedZ7>Op@gpuflwv;3$JAQQe<A{NjG)NJG;(L
zq0PO23@(%z=OOFL86Q*~Lb3hz6QIIATU~txo|WtQQ?^J{z0H?i?uT#wns#7D6s!Z+
z#dAi=OV_^-Pz~Lb?8ZpCz|yKHE)M&8ugh(_vQJ~!k)=+h#Nm^%h0H|@%y#cA#(GMP
z3C~y60dC445j#0Q^vUE;Bl>zG99C)PPs!@hiKke|gxP-Ma`q1yb2qv&NvbG<k<c1H
z;3)5F;3NdA9w+e|kat{GXeE(eEAll2kp3&?X)%^-AlAN0NZ~Tm403i!|9=ao7I`n(
zeFhF0GyxU|&vVAgldoLe=905Fp@5nrfGgFOgBP8dU)M+w_yH5AjnYvMo0A3yJ$(*v
ze=CKHZ;wl9dMv8Z>88$wDdi4V75tw^o<Gsx*HgF!8efghAX>b4T7m4;ko4g)frz{Z
z@|`U&N$>!_k)yU_(>=JrO_i6=I&4xy`d^+NCFb5y0w$aac^jz=A3t#>zds<=kUso8
z%BM>9x`%`MUSwIZu635kh+}Xyo-FkY20Dy2yshO-kW+SNgSio8%r9fxpm-CZlw2A9
zRNY2K8291wdm=Hy_T=F|K&2`a#Pj7%CFz9XZ2vF9TSHT^XN)+>4wAeCq)`u;Ng&-?
zwv+wF`&`-FAL-x5!_{B;GQf9NUtNwgU*2&MLN%dSDMJ7FooOpHtm9lI@qY;wt83bS
zJ`%Z8`cuM3#sT+XZLc@NQxea}8QgXWA8}DK>tkPb_}&AEf>VWNKXh53jkFV&`#EhA
zQ+lz%Og*Xl7~^wq^El>UxOgZ+@onOay*t-WhL><{pm>xsfn0#yM4v-A_9;A%pE$&u
z#SjRW10xw>@)-T`Nn#@UUkA{@R|inltGd;OGy4|nFbZ@X-%?M%@Ezx3^y?Fov%Jfr
zTau`uwYEeQ>D%%jpjXw)KvIIpm^4vI_N=2GSwr)Azmd6-WB`V)eYmmUCw412%@~=N
zb9hKl0Fp(7Dp0W(&9oBtI#4$khhIYH*h|Nzo%O0|<7pTEn!#P2sUaWbMnVfFs1hiC
z_TaQ<oYY{TI=k%B@qo3Wx=N1Ae7jcr+P8ki)SuDmTzZ1^yQw@eI#=a%!7Hs5p--tj
zhrR)RP7fpSE~J&JDK_JZ-I_&2@a&~8AcVWTPP3?r0Qi@?+wNHX#NL0c*SurUA0J&P
z7d;~HBhDVX_!K4k(=Cu}73)tTFDd5m;Gmm!%Jxy4hkKb(g1{+%S2okIyF1Hd9j;^Y
ziGxO(lbaN=238~L820^?$$6mh=;Sbc27*J)b%t3heksL%kLfZjMr6=6nzFP#mBtSY
zQsLg?IH1YMA1yAQfwEA_{>U?dhyWU=<xILDHe7O8zwG;QI{2px|Hxk`8?ek$<bR0Y
zzbL>btBLcK_1vu!(fL1rE+PqAomFwwG1TEozUy|7ut~+GwJPcoa<co<ei>|1%g3aC
z>^_GI<ti;44zI(Cn`o}Z2*p4{g$8OLDAW}?uLF7KPs{A}cJB@hy#&Y?)FXWLjm7lO
zC8V5*IIzJe_P%|2q=tFC!mzjkg)uOy#1GCcdPq?4P%=kx*zW;@!GmPOu*Yba+>iWY
z2fn0D|LpQGInn%u1JGy7*_|nlB6U`#=dwEc{MszD_r;;6BJ^3`Y$ne2%`_-hk=*>(
z=Yf|O)wd6P`ba;`NfJq`(P(la48G4D$Rh);21WWNZ&J6{?x^La_io(MzB$sCcb#&Q
zDr|1SieQ&PF*&lU36-x2WZiRC%h7sTo1bx_A12LvDmMzz?`uzb!Z@>+XRqAo@&`=n
zE9)7l(*T0sLiT&5wam)lXGCe^R`!(s4e-xyk8Byc%7HiP#M*(>CD@FovZB_;q3A@9
z0Eg{#vZIc3cZ6}`nqm5$>Lb+fqYj?^1uosBWELH}aP9EQ)=}z5wEMoKzBtz1>iWKv
zQ~PSdj0K#3l*Gg_xB&rwzu%kbMiRUOZ^Oxp&K&h}i|oCdU7~~>0m^jO<mrAq9nk05
zbiryV?`S5hU?+y|`%}`ff1rT~;`AtGp<spsJu?lX!s02#0)xjgW0KWyGH!vJd+ykN
zx#Us1C`1K!!x*mHVBLF;@mN0>`(@dvI+<lqX@xV%CzA@!i5eHnG6G7!?5x<}#JI+~
z!W2_-5L33dw+AC}X<pF$r34Dx1K@(S_&LHT6*zbsm+S~UO9{gs1nou~c7-^T9Xs~(
zysFbcJ80f~D%7Ur82|!0jkbzQ!kWe{{Y+&pK^V+Qs*E~-^V7^3hxQ;bQCj1L!d%+v
zF6%1<+x*B%`hJTQ6eEh88S&_CxxFczSr+Ur>pXh3gCCeqPS?NP?Wr&SnTL+fcJl6M
zkEscbY#Hjva_3BT1SH6FKl?tWc$2H{`vAx!D!&^8A^mPsTH-(AOxO7*{QN5KbWawo
z+}%CO!U(X0yvONijPeVP9h2;-*T?~%-&H{m?ubw@52cS=0@6m*?*bxTJgUC85E7DM
z!P}PNMyotneP-ZWZk-)D1Iieo&i1|r3fCmF6?SvIdu3HV2H7A~FeJDxFGvc(mh_I%
z4S~bIqbvGZ$e7`(??I@X=S)HyIW3Ag3bc&`8o<)e#KBf764V6&jQItO-_T*%+^!|^
z>?*U_hT?cm_RM(&jh<WZ;9NB&oZ{t^fd&4)k_Az-uFLUSw%@zP06s3k0WJ-=jvDk@
zvgSRIqMj$V<e{9BiI5!LAe4oUp00Qrjk%M(X|BbtnThxg9BYem=np-!t7@xz-x$k(
zDDIjx%j-tb+#>yF^MAwx*LBg)v5$ZCT+18boTZ7&EcF$LIdqHiq48|E*;LTK$!$yJ
z4kikMU`_8UmgqHN01q_iX3GuX3Yx1w;BR>Pgs?=$Qd2Qgk8FNDSVf}BopR?S;H>h9
zE%lDxbz48*JYH~}j4lBqH_*DDjsZ2(FF>adrto1*rQc)LKlgn+KubKA0n&W%)AgUB
z`1MTflH%jd+@!gu3&Kze9fOJqA+y@*ZexA6oV1M-`Zq-gK%<4J^2zB__Om2c<imLB
z0pmP(xkM?^<EzIR%}i+)2FGo>b}w1Haa8wOA>oI@4ZT^TpFKyp<)JoU;8$Jo!`wA{
zH^ClepcUtfA@_))z4<AF6DuEXy3J!mXS9--h!8yE%x3&6?hsqBW2e?0dW&~cbEtDd
zZ?X+Fn7gMLIAJJ#W&;mk>3)(yTAZsI*K01`{`NO7?H#$M@OOW{@kR&~YY#0#V#w{$
zXqeO`zUSKwgHtgK{xm%$LD<f*qT&;e%P$_YfVEGOHK5d<++VBsZv}CANCw+{*<5Ng
z4$3LQ3rWLvj^6Bjv6^;>y4}DdbD|TBalyPTg=^fG|JanV0A4ub0Xq{7oUNWP+;`30
z08=T;9BoB<=aqacjfL_oE@SJg!S6FV=BQRQC12NQHvGd^gnc38$ifImxoHzwBt`y(
zk^q0<2-D)W-0#dM!!n_c+feyivUkEIWhSrpb%PSoF`K>f3^r@)o)J0yx&Yc|IQjrP
zX`|K@U>oUGuzOss<qx~@WP~$Jy3KQJ53cRDq~utY6)pQFks8&LHA}XjlATv?wX$?)
zf6i6w6{&t0sUB5+muX($66L!8k8ygK!uxAPH`Tqj?j-`Vnys7W$wQ`6<AP}{FAupq
zdu6Rsj@(0rQ0GeA(Ie$Fg}1oUY2BO)M#w)_ART!qWkZ|o(4I->jJO5V`{==>*4JhE
z#*P4Y3=_5djzN*O-nJsKyOD~=e)5yu^zLf(ccZr|cE+8p7(0ItoDLU9QZ9xvM~ABn
z=JCH4^FL5f80eo=EhocKM|a)fhH~|?0NW*E7h9$k6ao%3o=iDG0kC?+@)TdgR){1I
zten-{p7ee$_ToHMfV40b)U9t-npGH0%U0-qBKf-VYV%_F-XGY{5%!e97N&m<+Z)(T
z-nu=B^yaXt+y8%6ya`Rc^;Btc)yGHhSUO;GaC7Gk$16fxHXA`zwy<RGL%8{}_ToCK
z4yjRK^q<H9Q~E!Ymo&6xl6P$&!qj-JUbw=Jr&SRzSG(nLO!om`L+lK6m_Isy)D{Bt
zQMhmX9wkoB+`-#_!2Hc$#(e!W%q{uL)QWXh$DUTs4=G}4SzlUVgp<Nle;UEW$4HK<
zO8nt8H-rI!<Q>xmhj={&_Wa2m>#>D0l(WE`EZyNJQ&Kb)zXz?-=(`L!v7-2{vBNZb
zec-&D?$u<7-hc1#eY1%=7TRjJ<vhJ&Why<>f3Z0lH9T#6B)u4W3iO*#@e#C#6Suuo
ze3ze=8UGL|3pZmps+co(>A@FJfL+&q*IqW=Fo@lMLy$$%<-Dr>eO<a>Bb_90{%+?!
zvyxiWzwK8mu7Ce}6F(22ttm<CUH$3!W4wF4v#3PL`N!T|5!CuE_+IbtC(k962=`NG
zrMwQ^`O#JI&B4d$!#isEk{#kR=Z1*<Z*4*>6-I^+u!4vok31ms`RygE!h_=HOxd-8
zBY5S0!!o^#lBlKES(2%<C!K_wH6~XWBo5EwF{*<5a4rSAb4V(nR?*v*%3<QGIY0b*
z63S8k@38652Wk@oY+xNZ(4c)n|5MJW;MdC$?08h!87s$0Y8~9XCBRnWIqXv!pFMC>
zo(Q{`OXGgF6)*<BHebiI43=Yid%ngaWeo{F1b6+X8^^04N=w%IJCB<`td^^%8-G^l
zC#G}_RHo|+4BHM6Vi&ibAG7J!(CN9zdF$Ius=Ieq8YYy@JINc0t&ON~5vEyx&|1YH
z)?2wuyQxGT4UttHU&x!PvA~4_85(yGD6cZ7m5~IOC8?4}O^)7FsyPc!{wQYXS?J*T
z8(tZ^?e`!AXHq``21?x@^35t9{)tHV24c2pp`tCjfr3>a%xgr#Z(uHcc*@ho9vN{1
zlr-K44N*AKIgUGLb?28jf&4!0fpXPah3PYTcp5%7Cs|2Fm8BVWQ&Hg$31ylM#<~F|
zwftw~lxSHD^oX-!Cb+gn!OzPzjsrsv5EF+m@Kteyk?>?@>fb(X$sn$9mULiqDYoc3
zJ>mCag=m4m!3&B5wkjuow%y*+we3O_W9K=PT$0G+ulfaz1MeZ8xNL14F1SqbYl}Va
zp_zw>?FbcD%UX?H?;$eQsQ(`;U*R9uD<5Pd-RSPAA2TzGT<MzB$?SBFtjlkJVBP4N
zV09Gm4r8)|fn1XOSr`vV@8^i*fGO*W282IJkjlP=gKyG4a71obG_$R(Td(svWbJi6
zh}O$IJJ3YZarRgvCEEKe5kMC{)%g0W`FZ%?I@t7gg+}TA_x^!%YCmreb{Wm~$=gRT
z6MUHj^xVv4DtsCdl@f8#P9P{+aAZCEGA<JJf6I-2!7Qw7thDx=|J)?G<_*vT7_V6b
zG_BXHp_TY<6Pp`mK;KO0F7SGTVgGRXu_1F#yQ^K3pGS3|&g)KlFbRKJUwtF5kO1A+
zX1nthekYQY&4;?QNr!hYHClp#L=R6n1p7Ps!>I-Ih++_1szwO5L<o<4iLRjj_-_rO
zA;qUt31<2+>R?qFi8KAs?#%2Z5JOe2rd%|)bBhqFRQtLUb9(~8nHI(YK~J5Pu-Z{g
zEufq|v77&m`}nJoo;pWIro7AtT=%Ul?xMEhsXGF(%V51-h1M;n1MTDSCZLe$Sykk~
z-&97Z;sD<>$ymi;v+}Mps-4{|y{vtowBX_zOmI1Ov=)~=Ln*Pg0R_q!0B!3q<RE3_
zxL~g56@yh|YYiV~@mg4~A~@W(Zl-K@D7bq3sYfBQD>%%ATP2$^u=k>|W;TPW?=&L-
ziO+*Xj*fp`h%lwO|KuMO>(N{z_XbUlHMp_Ti_s(*mQm$1p=yU$RnWV0e5Srv&qVtP
zu}`9bkt26KN50a?LA5#vm_^d8K2Cl`O7w?7{4+34ZmQD8p=v>XI{-8FFAafSj|hF=
z^DAQn0Y!*YSuCxz^mm5~B#=yk-FlPUm%Uv?q|8Wbb-P-OQEGLgbu}6QVX{A`%=bE|
z#@Dm}`M28tBehlGIt5vc*JkV-l9R~B8<X!l^rTwlT%d}g=|md9W2t&HIGGlb7a>~R
z*;|GEYgZ5ia!}1u`DK+|W(R&Ce}2uX$+rI!Bw$SWMM0!>A-sZUOt@-g8aIO5eo2$=
zMAUK#1n%A=5PE3CYDg7+_}KoWVTWW&6fclX89Im)BFQ)#`#>4Z1S7~c6ZYP)(9`3$
z?w!Aa`E6hRpD7do+Rt&7Q}r{6(}s}k%K)AwZ^b^@=+0Lp+4ow*<eZ~lmuk@g?St>l
zkN<>@!>`t`j$(5Kc{KWOW)YeqT5Pkc`JmwLR=F~Bg+++y57QJAla}Q|G=cKPN(A(2
z!hu=E(xlmOKCru;rNEW^KVME6=4@N&KGt*@X~Pz|g7`PU&r=^7sdk@(2|d?_gJ0j;
ze~cD1H72xm*_LT7rE)k0+dh1Gvuacxv6&=br{PqZMx>ZAgB~g%cm@gFSxqR?cS?y%
z$=AHpx_)T4Qn5EwkdF=kul^|_)Py;vf>EXLi(zP-1au;>jOMVz32!diev+aKLMhy;
zY&H!)FYqS;rJ8RpTTA~1<A`jMP<PoxOuPow-^#kO*DPe}#CqkMFK0zq95^f8;~eJ)
zw2Pc#w4&PiGG9&C{33|eQ*M8<nWylW+Y%AfT=0up%fD@JpLk0kBdKq5@dm?1=xmT*
zIk+AkZEst}52$)l;=_RAStSfEKq9Gow!Jh|+1LVoL-dz~HBKp)fsNf*XzGNODY2mo
zW|Bv&AsZX#e0sw#1zX0wGpcS3#k7Vz6A<kpfl;}Efn9B35#;mNk#%t{bq&5R_3%DY
zi-VR)$xG?fO`*OXI5ul?d?E<LQe8%Pv5A<z`#MqZJY(t@H*h`tL}Ijy=nZMsdDyXX
zrp$n{X&UJLVx6t=$vn9?P4iv6_x%O`9rjup7hzdpBa9ch8bUn9+V2Yl+O5GB<;Sep
ziV2d9pH#G9Kdnace3%JUBl`~1BlpxRefUu@;@pt-G0+yF_c=|QMRP!W;adgv;h+d=
z_TZq*GqdDD7z#(kndupD{3CtpML!nKT=9XH7+Qjen7y9#!!Dc?zIootkcy+DZ)_Ue
ze=UEx`BkcAJFCShU@nQe4p~5zC}zvou?$`GEBv|W!O47QWGU0CDkM<6%A&`4M+@ds
zs*8cw^)}nR{4%e+8?5H;=*gX1w!Jo04s?aX-mP~Z!Ba1)#jIf5%8TZVgwh!S6(<AW
zzcNDNO+#73VspAJWK5MX2Thl*5@i@XZCLkj)*cxWwv&rUoPI+oG=1jUvrj6EbR9-l
zFgKptqAUu3@+B&k7p>;%UwB=&c=5CuFNwFeylqc@(rBI#5q_UG0%Gec&}3HCj6xf6
zz$-VJaCOY$p$bw1zwq0air`;I3|D~ELs+QF)pdPWUG_K|gUw;^IkJnYR&$n^doadu
z@5t#oh+)oB`=wlV=j#U3(EKd(&BR885gJ7Z6KqwO&|r$M<Q)^!=o4Je55B~$+$pGW
z6}|pb8X^cmqhxixGwxK6{V%iO>oGW@r`=>so~b+`b-lhXr%w<QVkJZm0ye<$etFdc
z2_|%-Yu8Sh;8nVx2h0s|zKg}+uzc(KIHnSmAou3U#k_p)nhKe0V0Uh#zRf|iUzbxO
z6A8i?)4p&U`|D5rKZk(ot|hD+z%3xE=P(7mR5){43*saQ-K5Hb4viH7?c3dF$y&!@
zgm_JF$8ND9QF%x>v-uSaFoq>DFfh>1u0WqJ%=~&0Z?QUZ#g~^vIL9yi82TwESObjS
zv*HxY*tu+12?n<F<K8ofshg`W-IRUyitc!VcFR(JEHZ~ZwrnO&EZo+xY&1k`4lRvt
z-r2uM?g_X>TQvvcl;Y4`w^$)iot<&G1cq$f(r_`Q8U2jw+{q^e#PjxaC(PcvlD&ly
z9JJ`J0<v?9PDV60sIBw5$)Go_s$em<udy`u0@5^um5nr?Vl*N<HWL%o;fUQm-PQiJ
ze4lL#WQxnoF4VyEDE0hsi<fQ>rTbczm)13h;+Z=^JV`<Gk-yCIC+lE9s&;=;#Euj|
z29Tx)kkr#paM3RVI(3mG%T?jvRFA)MZ}n`JM6y*%D)byr%epK@B3pVSZuPw0ho>2|
zsy98ZHxR@9ck(|^d!D*ON}eo<5h`PDG!DumJ+Ve~(1!xqw64+8O{{p4|8PXN>|sRa
zm(u>(x`*sler4d;<T|Yxa_Nuz=LC7e@N&gbx8u{Bg?pw8urCXyvU$PB+E1_*Y_2?h
z*lym^;cQ1R+KY~?Xq!CLz$?%8Zd$y1l*6EzTq&XISt*?;p~6J04NJ?mbcYZvurRQV
z3c+0|CE~rweY+Xn&AS;A)BtMAFz!Fz{7c7#vy9dh2zIWJOGSv~;8ewQ?<5IEQ#Vc<
zBfqEFi4oiYHr@z=71jhdS8-5LUCHeoelr)39^dW!54Z2PMcXcA;*@ni^@7`sl*d#f
z17`Ks{%Sq9_7675@H&JFk|M2pZe9FcoBH$D3}h!7?B1<(4kc5E@AhXl)X5B}>&1@Q
zV_qi2LEiG{Bgs-!dy=<Nsb^BW(0E`$J-1}ZsGxy1VAbNbq^MmYp<L>UdMYhk7{F5D
z07jo3XhwV}ho4hEh24_>ft$$tgpRi^hmn4|jzG=oEo(GADQYcM(`8n*UqxR*is|3!
z6V9$(AXmxeAPMJjT(BM&gSSDT+j@0A)!~^6ddl9}kju6sFx_mGSw!q1zqjFJjU`aJ
z{N`y27|)fsSL`Nsn+5qYj34Hny|=L^$u(}AHQ~_NXXE!vEPLL6Xg7m&YQ|VHF9Dpg
zy#mX1V3kUu;Xo*)t4YBLI8}&Yj4J-xZ=qpuO%bL$CidKHPZ+hm!VjSl#KkVts}ySd
zM#zMjvpWtmHA}&%CEN3kd#H0Uh8Fg^WoZ?_A0Miutw4*K6h*h*@hiEj#hCAg4juEi
zk|SAMJ4DZz?;6av5xp@3>L2mth<bDbyEC&HG*b*Twh4T`$AtQq$o)86cI?5FakU=C
z=N6=qcK4bWC4~P*WhIqJDMAH$`*dB<_Q)=nL`oecH!=wE9eK%KHW1KK`6<qEcxVrZ
zd_!o&6~_bVL(a1geb&vN)j~T!i+P3^>E}06uAP>_G0t;npD|v>B1Ggq-xiAM_8;to
z4q5%>zx#4SuA_n*c_ZN$d<Zk3;g8{yQ&Nt&Ofr2tZ}U&cbaoDWHvn5-An#)3P`z;(
z!dcpCGzio7k-P`9K6O%?kkLS5Z5#st55e>cfth9{pNzgr<I7T|*@kfJ3pYa@I~G5%
z4x<BKlV7qK?T#g<4T1Oy_c=%M%&z}e(DH32CjAvKWBa>B!D$yugHR%FX8xGTizvc2
z&4!~l;O_Y@B@LrsLJ8eQ`Cl6af0^=v82P)<z=d(@Rif3xY%mx3l=w<95K?Piy4<qT
zszQF}t@<92#*WRCJ5s*G$+4@%kOacVmr0b8_vc{4N6H1a9AmW*6URG@IiVXRxI1Io
zRMoug=bS3dOfvTdMiF|Nwp%~Z*voLl?*ecJc@`J~0ZeJUblO#wQ!SNPxzV;K9i3`9
zKAd-HJ<UIng&=8Hr?2g*fG~G^A^T$hqwVXLVraRO5*SvSE+yG-#;2-uc2re|N9gd`
znf1YuDd8gNaw3p;jp-xqWplakYycw8s3@I6`j0WF)B^)wI47YwLvX0Bkvdv=Lm~tx
zF~VHWl4x*1z3;E+V4diMn5-o>=P2lc*bGk%q|y9NLrQ$Nj*XfG014%pDZ)5XS93t5
z$}ylCH&JVN8H(5MfWM(7&RB9rU&p=v{a3d%4AizA{<amJs!A*aWly!BHDzF_W5hjE
z3xYFOCnQvSq-+ec|KH7FkntMu29i!C<%xBkmeF&vF+{VF;)Vm_7pTkdo~~3uTRZ&h
zjbq5@c2v7?JNKeGU^ZXs@SfePopH5u&xfUTc3teS;QPH(hWQqFdq(zoTdf&k+gS0!
zmorKxJ-|uPKlUEEJvJv(U9tDb0C{*3^3QQXCGV`4){J^iE);KaUb^!S(;&)yK@<1O
zP7~1AY``*DvXf^tNgV4L{6V|E&{2;iBf9L?W#HBktnM!g<b*Mse_VdjWxViaw(aIw
zOINui6mVw-N8$O@(kEWw%xxKTh1eakD>34NS6kuG959s^`xtbF__?j&XNv(#^Ah(;
zRm5Q%Iw{%B;Vj@?|2XRS9sF-WbjOxBQVZ`r!m52X$q6Ekuzn5>E^>d(W&m}YCn=Z>
zmff&}?Yl<-W>Z&%{&1Z$C5!yBckj&JkcL5tKh#XnAxU7{V=&k#x`$vJQc#G4vm(%t
zTXmll26<bTQU-sL>cs2Wm=y{jMP3r>tPBQyPy2(YK_bk)QY_&Gk6m**?PX;ZxtFk1
zv3nlUg=2^(1Ho!Z>_;fM^9_ZM>1W6^ss6C6)xMbuP^QT~NQP!O&WwiMXvekY;5@|8
zktdNQD{))J%T|JQeiU2bd+7m|63B*v-3YF$hYQFHB05~t6@-}LN*45`tAmrBI5tUS
z+^IApKReg_#npH+!6jo1j$w$k4{w16NqId)G8&H;YJ40BJki(TtEl~z!{6kaU*vMZ
zVB<XjxH1+pgJ!j)PW~g^4jq*=M=^E~Zj7;`#UU{0R1&rE8=hb_yCWX1V4J&g-bkMc
zpNIL205G_8%UN(}8Sy-L>|gQ3Jb9;PF=%1{{L}R2K6TMupY2?oNl-`9t)HzI6kJ2!
za-j!xXR+9W{aQQm(a?<Lkx*b=gpo?W{bk_?R$fw7-;9@_P;fA~4Lw*+^mBEiYG$r(
zc34g}b;bpzrfX`a_X$%4-Z<R9adbf&P&f?6OetQ3-!vwWBuQkza~}0Jk={t<LsNF$
z2Y|yqf4T|)T=Ki>@I@i(Hl<=BPi|1({}q#poV8@6y<*b7IN(sB$mq$hMliC1lYT-0
z8r1TZWz8eY&e+>pw?j#gvC}3zmlP7^XVPrtTF%I*&VRq-!QTI7rIAS)qNzL*6wlLW
z;;aoVs8JLaLSR#i8RJC5*|5WmOIkqYi7U~hCCVvr=pGWTr??{#BXwdj5E-~kYNlM%
zhT!}u!A?SwY`Q%ZaRj-b3}R_+Pq?T6bcT>p-(FP9C>803owOn8ixBcd=NrnS3|R#D
z5UT~Dt8pxj&s0E!#phq0$H0508jK-7UJLX0%td7j-KOBN>D<z9TYF{@LLsVijGvS!
zT8GR>Qx?r~@B;dX%kX>uCQIeb3*~D=z%Ngu9rP_)Af43yXda}@JIOsbvO+}z2$`;m
z$dyM+$ppUoHsiYPj~I|AOY&$8;1s(`BYDp)^p)_@wxtd%Z!lrn)s#hg(iVXXwTSpG
z@>CD8eEQ<Nzjyw!{_**P5sg_$>Ghmuf5lA^FoPu5%@|FL#tsG>J$n8Y<7q~`lYThZ
zL%N4f&|KUfzdOcI8v!tFQHR(A46&8acwdg!x~ba)+5yVPUOfiUH|0u(Yi;LVksW8I
zTdAoF{^N`-T<R!lbnE0>J3}q<xI&uXQCzJK*exYbg?ByLHpLPRml(=aD3qC7J-7;+
z2s+2oZ8H!xURuUNdDo`Ju`MoU=!f%Dj#j4eSOy2}`?l+t8pm)hzuk-hEPU*=Gq|sB
z|CaND>@c&Pq0cu!#}}mS#IhH@C3pq3j}|u7bs!bzg6*<{_)w4KXDxO^NFl73zD+%$
zbxas!>6yn7s8Gy5<c{A96>h<DjchtPd(9Q!=RHor#4)l^%rSAfb4n~q7*J@1#$hoq
z)=C12`=CU#13=$Z8-WerYr61?NSQ(9)RcX>OnvqG5_**-`VH&gOB+A?K&UQ6O26hW
zF=f1#Z*-u}0wI;v;GDw^ENc}DwfaMso7NV!N*is~0os8BnxJkD+ceevc?(vf45;wB
ze{ri@czyb%9h#($+}xWYyvt=~t`?zCSY_DH)3Y7g`a)S6X_J2dnxs*`7O`l>T4W)F
zzllotf*>v#hW)y*?5poEcS1mPT8Hy|h4nij+xqsOuy;XchFQXSo?#upY!QI=j}%G>
z+n<qXB7(XBEIueC|D_hc4~mjSQN^b$@poadPK1~6vBFfv((ySOowjiFk^6(MsZf2&
zT#6`$N|{mrvXU^MY%H@}lUw`ip`ud!#<fUI`<JkxQsxZ|%9Q#oN{)N{fDc|owjIv{
z^yC!}6|HCv)TxbiIw*!wD`C?0EcslzqIw<v6l3jKt^Pj28Sl__1GT=Mo=!pI?3mo4
zMn}KpZucrl(4Gz5yc&`+=NWX6nc6%5n#c`Dw@I2Vo_jnHXs2+cB5JO00=3evYn-vg
z{sfhj*GO~se0FkTa~Ez|a=K~a623QkX`YH%IDK=(-NDa;v^&Sjqgxxr=e}7uLjJJs
zYo`5Q74PCV$9tw9%5$u4;oqLCpJg)?|GtdyTPAhQ*ZQYEj;Z&F+=qlMkDJ6FjmS?^
zq`!ZUCRYLQxhl0xDGWDJe%Z5qk+T0c)zZdG7vF3pgqb})Y@v7I-3DWJd!0u77GNF%
z0w&09Gah;k)soay@!vBs>ue^|W*|Wxk^`QY$-;#}(Man&IdU6z2hB^khYoHJEViOT
zhdu7fRv*uP@70fUi(KZSTJs(9bjOg2hXU4jzvYDeuG@o}!}8ZQv_u~NBK8qC4M*f-
zPc-qsEn#W}?t!N~@TDj4dtaI$(ZHgohV(wH{gOZP!br3VsiRu_vs|N0q*Ys{I)>+c
ziC*S#wjb7=rG?}kpdy<s7EZ_3*e*KmF*;tU`nSok3(-RSTumCDFla0wCG)TwE)8%>
z*v$b%j%<^$+3A$(M-!vvFf#9@Oe?g`lCsD2CVR22YOTWfyF<)Vl!KJ3^M^L>?)93b
zK3jdyE}9bHAyjoF51=hvmHJ<thx%RFlw^`Ft-3fM3DnQ-k3zVHG9D%dZWWh~2_H>c
zBK{MuLf3dV#SLy_lE=u(E<q)u0i4+XOPl<spkotFL810#s0g&Q<W<DIT*xk{1%s-M
z;jGxdT<yHQ?I5Y`Dp3~|H9>5Q%X3sMQXGgp2n~==84}^#dlb%N#wu8UT%>2X7rgO#
zvHfGosU*f<l4bwr_|C)|Z0d#nen0hrUavGwbEp02?#VFBiV<@LfXDV0VLerQqf^27
zLalCU{!o}NoQ1#{8N8T$W_#)O_WMzIVd39(W9Pz@5SwVd78xDq;kW`~ev~cPn6*n2
z3-f-mJabvBjw?k}`YOvHu7bTQb|!R?fr)^IWq3LFC+@xA!eobMoE)lWe43-~Uk#$3
zKF_qY@dhE!XJ1q;V3+=;J!Wh?z87j+zNRgsv>O*zf$)XBY>}Q5vpV<-n0Job)Q;g(
zU|ROZaFjU;TNZ~bUpyKkP^WRt6VD0R;;_-%*Aa&sLmFcua(ae02&!6b2&S|u@^)|u
z=qC=3x`-x>3;gRMVNwf*L&YqigGH_U-2MBzI9z&;y=e41K+0_9=~628$6kUuRe>&B
z$(hG|hF{D%GHUjm<*S1*rsmlYHzcJ*H7=4Eb1<!fbBEAEC?^d|3(YaPEU2tY=Q^RX
zcS2@Y?Z+Io5he+Sf@8$a;A11r*#RktI<@qr2cc|&R|}4KC|x_C0?z(swEhu9*&_VR
zbJDJ3Dc5WWps8uU(ULv>LFB(;27>Te7*?VvI}90#W{t6paSIc(8e}ycs7mP2;iZ@V
zb%xNnu`c`rQ_T%&e`Ma)$dwN-r@o8eiOSHk<|Ny?VA7-5wz{HzS+G4NAr00_i22px
zf_7grfC<(iRbuZJ99MH==hov1<${idpIl7)ho%Y~fQbzltQnJyn!(*}XTHAsJ>Y%l
zDV-e1WbA|w7}&+hQKGt{MqIiWi@xLg@Lwn<v1`yVzpJBZ^bZTgN3JRD5rDo)_dAy4
zyW?f!`@?yBGMnb{*uXZdgBGpKOG3Oul&k6HTH|aLQmp^zP5gyR>0DqtoE~Yy6p}q)
z$cXd|Y@J{*>PTYWcLxP26oIRN#F)q^6H&759T?Tfm*LRf#b_fO=uxwKzZY+PY<B+5
z7bnE$)K_ZvTT2qVc-bE}xOfRrp)tThIlW(*W-*dkCR1%4#W!3Elq{%c>Rph)*LJ;p
z=U!ypjgRV%`Ly*6Uyr72O7);*?bujmPq^X%?4$;|RiGc7I-8uWW>^Gcm*-De5-XZj
zXY0(?%V8Huv>bW38CnXe$eJPFmz^cYv=mbs&WE)*xAr{0Yp8TY2%a?98Q!vB7CWjg
zGI#n#DL7(flNXHZNbX@Bc#dcIes_RL5~)l6?A|9c)<R{!RG(m=3<nL@$AUr&fZeGD
z;2>UoE&od0zb{c+3rY@4T24%yehCXw9(LqJio7!=bn5FY4BfLEcXg8d%!X`2Pb0tF
z$&Zp`VOI=nglJdS_^z)gTc{8bSA{c*ey|4Hk;^*vQ5jx6`P069gDpS1k43U8qtt-P
zWt>HX3}H4ubku^9ynr+Y)DtqY-?F0t_{Hr+t51)-`Noc8!j@}eyZGcJX{XAyv5E*z
z^|2q-O#<X0&p~X<{AU7T_f!2!wbjn;e(kGp_es!i`;uXLnB_Y&R1%d-MVtAtsSDfd
z?Lxy1lZ36yO38$*oUW4yU^3;ziq$=#pI7unA03cK5Qi@9`yZk^D;gbyLxXR?@jx&=
zoVIjPXQAw4@Xs?baV|6+$zEyJTsDt-qe`S@@{t~Ox}tJ!!R7F`74#cs=F%N`_~NnE
z=pAIZk?$>8l0s9w#_mR=;|M9y=k5DGx!i{Ok5T^>x>{|@>N&DZthpc+&+bC&bs>D{
z&ub1dFq|#%xAM}<<?5f$gnj|c2ErmPIOkIiBj4kbjPFSyQDkTfu#)IXm`Vjw53u_4
zO&a=I3h}c+S*<KA!2Q!(HjuA;lnvRoB0I{F!9(V4U2!#C>rIGZm8*7nc8+|F1I0EF
z>Vf$bNL!Bsfu<VdiO&|oG$wCbPvg!x3wEa!hJPrcRT1%Q6NY`NvrK?;mA`YbmguJ#
zgn?|A?o3o?+OPU)=0EHb6I77D>wwZcNOp468`@DKh{@>%R1Mg{`MKX-++jM>%b8l`
zmZ0w*5HqatYQIsea2(#|;ijh-q#$ty|59Ufo=MJAlBp05hOi<NmB*&4%@{c{wRVrm
z<nT=v@zl@R@`#g(&HV(n-TxK^+z_~flD{{S`O!}$-D{1$8(XNaHe;rovpe4G?+=l2
z;}#PQ_wH3BiB-}eHCBMEkObFOAxjjsawH+%aw)t1=ty;+*x>SDp!;BJLaA4PB+2#D
zK4&qf_t0*^D1@8uesj6iB+xDN<A}z);EhBPxoEw(WX5rwz1#q9ai6KhUwg>p&bxc|
zz>;*KTDu#`dU~c;Zs5N;sq-f3P4T%SM7uL3x{muTaqRBy2IQO|DEGA39D5+0w0ID7
z@-vo6sNnf~E1-IjQC_G~eKx4|IXKdnazX{na$3P>lIWamX5~8T7mxHra67iZEfYD)
zW;b|c4b@ip1e*gM?m?%#gX1h+|Fh$<m97C2YxhJG+^Mdyi)1-%T+1x|g|+)NnhR3W
z8qJ}W8(+?prg9$1N;mPN>1!0_^$_FU2zhMXpCIK=_KmE;hPw}OdDreLpCRQ9_ag1D
zx=dj_#hz3n3SDq{izS@I5hYzgyq9ewqUExFUf#_DC)ltq>_~WCXXmCpi9gSoTUSu_
zTjIV~HRgqRStr+Xr*o#ise*cc;cDkq;Ot;yD0r+VIWW?*m<|LOrC-+Im^lp-A$jXj
z#0YRIk@x{BhW+24onyd+)P>2@Rlz?g8&Wxz0f<}JIJ;#V+Lg2rK3KEcQ39CgWEs`^
zX7eHd^<`Kf+_EmfB#k~&6u}sl<~3vTHydQKJ^8_p$4nb)vtH3gQ<Ch-AvCn-RGs7{
zKCAF|VmxM4(s!F*W9<-72qiE<67$DnPt|=hx-MC0QrPc{-9ow)cA}{pvlQ5>I7c==
zM|n@(r@LDclR^Av&PNb_PVn)W!H)2;M;CAbJ`y+)Hs&H0lgx10Fx6D-<|HU^W3tqT
z2T@}6-v7V0!tVL!nP{_RMM2@8>6%&a-nr8-QpujJ_gaX>yiwVI3`k<pfU_SYem%tK
z4}ZQz<oDEyPfR6h^OvHKhvVs!2bx&V1_O$=$wUZV%{T$Tv%#MNZZT00m@o-YuUhWK
z(seJlVZM#f$OV5d9RZ{E2+Jx=tRw5W`H5vibXJXZ`hZ~+Y<F0=+W?~-K{EW8oX3eD
z!E5ac2d<FQfMQSv@$RS({rI9<^%Gh1)O}abOt>n!5U}IY@*dtk_VG3{=o;<rRov1h
zxc9NM+gw3y1l(8>8z<p`sSLdv4WzQZ?!EfE8@(HRL7RFf<s1ZfS$I=T_}YDxY=rN(
zb{O`#y7~6=)6Xrx$2ZLJ<jv6L_-{;x*%44T7pD994}R!tvma62U^P%m+>8^efjoq4
zIxe5;{?D@c4M(OT5%vWx_6|h+B`Adv*_dT$+r3N8;xOTsYzE-V2rR<b_#nsZRiyy)
z$HA6?(;0dcD?GpZdkZfQiPT{kLebZ-K~BHX)EAdP-QU!Q<rSH;4HC5MD%{jS{$m$@
z`Y^N7`tc|!L_ICAW9dle;_rBJ<mu+#mv&OKhl9Ri{fwfUd;b~Iqob3p`=7W1dYu0;
zbjZN5bd!f;0bo0RsvLKU84dApv9J0DO(;wAHs85Lrrl@Ns)KV|J0bDTj4mZg#A~4-
z;tCmn#?c_hM}p*pZmuxpct<`$#?y#7bV`A^aVvFhOo%O+3d#pgy-FM5Aqz1QwTPNG
zh#UtAN3gICGs~>W*OF?Pq;*AvnjlQf?u_FZ3j-<bz=P@=0RcFgS1X>5X|xaJb66Ih
z#72u^q$;}?0?Yz+L2ucevP}(<D}Fi<2JXVsI-6z?{eivEBaS-Rm(&ECGbytRgDo&;
z^Vv&oiK8K#!#epWh)u{LnPJ9o%0(_~%(nwdA!~e0=k*s2-bkH7)_r}Xhuo6-b9!=R
z>+GZhkgoMQo-<fX>e=8EFU*|g`?_23L7MY9KJTHlu`}k21S2(9E>;yTCp5yJ8ypL~
z*<*9%_N~b)CCbLZb~E3L)Oh5M+_*O%-12w#z4vK1Y0$ij^-7oAe$kKpyW&Cg?v@@h
ztL;VW;=a{AmVn2P&`E2TwcKlAb?=|cZkmGuP+Ih&+B24TYty!a>@w8h>ug%Xr4!Yo
zC`$$o(f>Pdz3*S=eTd#jVLX0J&zGQdHd=O?to|XV9|LLO!gQl$FTQj`;koZ2Y_U7E
zTYSqI?J;0@P2yXb%I}O@sim(9-Dy09JiUH1alY0aZ^P1Bj6l07+~a|9nXOh?b_PWA
zO}G6QTWKVYPO~6ZA;D8)f$|f%E!K*n1$`LfIKRxL5$)$t3Q#E{m-NQgM}KF-VX-m7
zSf~z_D_y`*Fe?d``;yqW|B)ePXNOFoDQ^lg4S$*8#$+fpVh4!q86l(OyX#1_rMi7X
z$tRmcK><l{&xNJHudciow>e1y16*4XSI2d6SW396r0!F%)=mTFtl7P*{lGcJvr|Zi
zP7<!p5godugTKj*A&C43D)%FA?fwAGK^6JoNG$5Z(7L8niwPBW#j`(4%1DrQoqWBn
zj7q!Fma`J=rhF!5E4gR%4P@KhZGvSI8KnL+kr@3;VbIObDeJu?`!YKPXpEB$wXIqP
z5Y4HOLB0wmjcl%39Q7Y8zhAzUEHUB37sG^eVLmIKY+u!)b!qblF-Ytfrh4Vf`Y>^#
zloO?>Kvr5qQtZomuoHLLa#ly!l@?S8AkUFN3zTnv53UzeKjT>K4o1IEA^Oq+T?Q|c
z77%XLK%#7(N+aHe_*6{*%TRbk;&L&SVQ5$80%N6rIp*oVj*QNX1jU<b7LS7_!w6k_
zNK4npaY02B!1EK`>`mPL6SM=@v<W$XUOR3F?1P_21|pA&?Ms_N^8Yk4?fcD$9(eQv
zbwiD!Vt`(zzB}Kc%WAjo*8m7TV;XksZ?s@;&kRp1G;P=a0(B;UX7O#YL3S)FLN>>S
zn(K=;BCV$qtpZOfz{w~p8N$+FqB71kGcygb%}kTQz3_c~xZd3CIp<kW%Nlo}(h|sk
z)-uSnn3pNQVjUKJ;1i62G+sa~P!P7to<EjX2lBDmZ6BuW^0^}6`(6vU;37``2euQ#
zn7}}s{moh;0gw^~4|2Z@G6Gr?nTAkOY;6Ky)48MM=<Hv${$|>novEW+Dkjw-XLD-#
zPK%v3B?)0g1Woq2<D`Hu)(-6SuY+G{ep$qWVo(R$_r%M{r-o?c)94U$vLhNZ$~HrP
zb1b1IlVk|5QNolz)+N+)q%qKnC${L;{lAfQVZJCLKpQjdOKlY`t&V_$Ckm)n-8c6&
zBPDQZL4+}&)Ya@{T3Qe|EE4J-fhtfqF2`~{F;T(!BiF=5oFSSO-HdwTTF3dT11}oK
zm7@ZvRRL{puHA9B8tZz^P9LaDVIv}7tW(P1COnCM`!DxbJWxh*={>r$T5ezn=O-t}
z2KoW_qCb!l+=NZ?Lss{k;Mc?Mck#kqb*8L>>yfle8g$~8lj|h3gQT!Ae``r@er|}n
z?wDl9#9RGXFxuyE|LrEnUaHTA#JUU{cCQz{?^MweEp?%SK3@Lvd8UitzQdXFPLrW>
ziI<j-*CVNvxe5kTKANsl!Jd3~mUEHAJ@DlKqC0G6lh`*g1?Wq*^h4spW7X{~nU`cj
z3(<DJUcAoCq4vrX+rj*7NZ*t_LrvxeW#`fKg2B)yFHL!zrX+Tg9v(0=!V!a`KL7PA
zRGo(#&M{FJwMXAwEJ2Ph@JA4CxBA8P!CE1Rx!6ql2WUc~G(J>S_=N%s>_Ts?78ob6
zNY;{9J{}QTi1f4TMpL<>$Kz`VhOEQ0wna=SGTk1x)`;&-V(^R-xq$xCaKJ$EH)06L
z`JsR-`;=X=Z4r%P_;ki^BWCc>>5NyJM5vGnJQn1%mxo+rsjNi<d-I!`)bMwhPKL+~
z5?BmB>7$kw#w^<OZok6{g}_5kLI?!{l@<f<8B45R&F+==C!SRk`H8myL*poUFb&yo
z_-}lS`W$xD@k{8To7;hpJ;ZWXN;wFw>Vi?4zgO?vX=1^uK|>f(4~y&G<RXP)@q-(?
z$BxGa#Qcj;#4h6uxkpGov8*wUzYrZJ9)e)k4y^Zb?@cuPKYX`r{iZ^=8xQqB2FBLT
z6xbRV=Es}Qufy@iREUMSNPX(w<hV!XD#l(iF7<CC$e7rxC&oX`m#Pn<PnW`(2>t4)
zzCDsy7feFEg9(AU&&f3eoJ-OQVGmxCDh@Yl?dFY9Mxm5QjrSX;Tu9ud^Eo==E(-AL
zyZJ<CYu_T)<2%?{9vguUVtJYaYH#n6-`&k1=;IRP1%E7&y(Z{)(9kem9g<FhkpxwG
zP*88yA@9RBA7M=p55|w$AKuk553~r61t$(jQu$~89X+|gdvzatXonVPz0B#@oH_qL
zT)ks(C2bqF8%}In6Wg{ku`{u4vtx5&TN5V}+xEnEGO?ZPJn#3_uG+iKs<r-Ht5<bb
z<L(RR(TQc8_t0NjI=%5K>GFAyfFSr$9M#n*=sueoc6tGBbvv!>&HTCTP1sWNvjFHE
zIt%+rio<kuoj*Rf+?fOCw9o-2Q05=lzRVXxF%~SCGeX=HN1o~&hksat%Pd8TweZ)K
zuO8+UG&;q;^_OG<$8>&(4-w$ds6gdPN7Tx&g7sFvcQkl|d+J<$`u6Xf^wj&jJ%Ug>
z`I(p8<)-4gq8HEjl}NLLgA`S6QXQyRZ;en-g=Py8L`=^oAXMEScyIfOG;(N8zyLGG
z<F<7(E$@qD^E?bf(yjdg!)CBECgomm{U{ZBns=kJL-Cb>HAJd2qpHa3l-b>tP1N}P
zz`mxyiv+{5$mxWBQL=nbQjNb`WY-;lVJ%gtzuDtxybJ4~`jp+Fz7zQ^APt}of@Ie(
zh5;q{A+lg-P0Ob?ET=}%Q=sx)gVvA?OZk=cw@s#WB&Qfk`geyE$Xaf`PAQ7mkYtjk
zvhS$(VP!7ACmU$<C#H-1y_-(iz0>O_-)kYWK?G$T0<#D%?`W?_1pB_(ilU3f=W(J3
zfsG^*YHrj1&e4U2+9w~1PWcDqU>#9TsH(F`2_HNq@``w6aF4o*MI~+2Q_{%@c0E3r
z1j#3QpE)%u3I}Q>Z08n!PgBvQ3ps#BMB?JxzVhz}N1RT%y{m8(@~GDwA#G>W&@rb}
zn6j#Th+qY$9;}~HQ1~7&uzWm1%v1`|stt{jz&$MvP0}LsSaOUie(45Cf*JI$@-8c&
zBp!2q0~^tCNoPFEpC!^*5#BK(5@^;xOw4MuPMwgPFaDLmako|#wtd3!FGbGDr5*-X
zeHhJN4{oF1DYZPHjwx~Eoq)Z);`dT`Iv^eVTLfyeUxvGvyJ&;g_DJL~;gk#!`Ogs;
zN6A|cr(P!kavQD)jT;)EA1QtZ93zFpUTd-bVV@Keitjr=N00iHiSOjA__I1WqB6JR
z&c*5d_;$sZu;_{jJ;kR`WAv$-ycK4fO!~v8aQyt;_;IJpH>|o#$6e`hXJ;JTdcjp$
zWZGENm2E!n&`L_?*u*u6mfNnO&7N#tq<M(-<!|Suz#pRarQ=Yb8v69dum&vurrtY-
z@B7X9!}Y`Xdox^<;%U3gK|gY>G&|nGi<)Sjbn9g4z9dJAj1}24WJ)F8&vqwucV9T+
z^`rZ#h0s}t;Nb-;jo+J=Y*#HT|J;{jxoMuduug5=TCd`kE^RPg9=G0Y^}9-!rs>V5
zW~+AOeKdj?u<vJp)up9%sON<z?bn@;hmZL;tRUa=XLioJ)b8wBdjoyvop+3yFno4@
z%_R&Oj*d?+)HaxkLp|;h?a!d~1ZU3(HgHD+%mm{CLgAKDv<KA3*Aa@Efj%Bw)U_dS
zqxr{dP7(MQZ@ffYlUuVE)W3Z8n+CS7pL`>$<}w|(aQ(Z$lVUxwd=^gSA4T&juR<*p
zNkZnK@Qi|+>>g4X2maZv8Zqvc%x;BSXpMp*SMo^JT7O!l!|jnSyLCXgxq?FZLiL~A
z66C~_*qhiBr_24p?Gj}Q`K1fd=i>Pdp3WD=G+MW-{oUtv<btKOtJ5eG+~GuW(R{~#
z`gj%jB3%>U?M1NCtA`#i;dls)qc0QoM&a@zxQtIZiRBp3F>RH<Y0o2+6-mBd~f
zj=m5g_vGF<-|7+b@RY@wr1O{6JmiLn`|KRR8f5j;cd`eu?qr6>8}_Sj8^-T`Q~R}K
zg=ejTHmUwtZ>2Y!Z-iblW5l$rp0M0k{bC(D3F!kkHWMg=UIF$4t$&r7?96$&7vOD}
z5ViC&Gsb?|l2^$B?lDJ?5-Da}HVr}d+7DCaH@h#MWxq^lNeGEFG?H)St<ii&94*z<
z^RLPqpdsfKN5q`fQ>L4$t(G)yoxJUT3=p70-N8?xD>oFB!cTeR+Ko>3r%orB9>BnM
z?t=n7V_AhW%I3*`Y2ile1%nonA(e8Z+f^*Rv|`TI!Za7eE;VB^#Z_;U1^jPCT5hFV
zt<+DOC@@{JZR%1F3+13+EJsh}cO8N$&XB4fl%3NM*Jz37av(*80?{o>)lb>XBINB0
zG-6~PZS=_TKt!JQ@S!*3(S@MB-Rp$7?Ad{T_h&;@*8()@(Pgfzk`umlo*p8m1ud?D
z;TTBPKK&*FRI;?Y2i&P1lMQ(3gO1&KyRz}2SI68UA^n&8VdH}y1{4}mT*l*6Ew7ki
z$?RzK;V+5>i<ZRiLeT<sQlGT<%Dk0o44;EyO?dl+NxXEkZa-X{)h%--Q%Be%lI?&-
zKf+d4!jG1<pOXZzm21bDtP5XGlW`fw305oKq?cs;^KZ-pi*fddP53X;Re<L%rhLPn
zv4hb^Nfq0Lj7Igs)w{(NwP^QEK{}*cUz63T#{@fu4S8JT2_9!$0%gp|@XzEjMzj2s
zpl_Ijop~rtSI9f&VKVXj;r#_uU^xJC=1`&USKbPoKNIu5$_5ILM(eKZ&Gp!+%(G&<
z>MxI$Elmo3A5O|K*N?mQe}b7Z@W1g2=ylRW9`1WWpBGprkFp7KzDAIUX20AfPfYIU
z-y}>$^<&G9T_%ac*<1XI`uKNhcp5a^8#Lf4?=mHmUm(e=u;$xL2V9)zE%kw|+N<17
z;g^j5FuZ%ain+sL^g=im$FiOT9>ZxJv(GGIR~f(}yLlPO2J=p3dRol2?ZVvkV;CDm
z0l7hdvTXR{+^^pLw?4J^T(`4Zpw7GlZ&4Sj6PAff{${tIk!)o-XHo=ACR^9``dQ3+
z`ah)1gbqXf8FDYjtl7bn@+zP!o}>xGSq%P#u=tq7?Kp06`>%FgzTZ#!)XdVe0{fTJ
zN!n1SxhBcG(G#C_HNE#o_xFl`nzkWC33OR)p9LuuOCe#~RbWm7t8m45R)ZJFynazr
zS;<;Ka-rtaO7_+Abrp>HXVdyN-O~M^;bh;}sM>S9X}Ucl#G@@R3G{FE0r33d3QH24
z8PyHlJ$huPf$-kJ0vO_sA%o9p`FB(Q8BW%JRCp~iq)u04KWfC>EWCQPF8FQ<)HH84
zk8P-T>Ihcos8xSV`ehmcHmsV4?_NdEEY*KgF`diGQ!F#5V+%6+Rc@cPR3f^wT$pK1
zk+gL(bw9%e<3II*kMPn9y8kREUxb#9lDbO-k_)?PQ>R_(H=iuG5xQ7z2Mq*QlGYS2
z&l^@s3wn)*_oJ9Sn?ldJ??7jR$L;>{$F}oTxIDXjoRW>Z37|NKFT=Gh-$gtv?P>5J
zJr;gQF4r2?AANC#IYq!rFQEUg%OvojCqxn`*#3p)W5GY_#<{M7&hOC%3A1`6Z)>wt
z=S9#Y1$DNi4U*E&+Z1psUay34so5Tr|Lcb$8o1VrHFK_Jvp{=3=CftNzTasd9?+46
zVSDisUI>jABEZM1p&mOQ=&FgV!~9F<1>$!+lG@B<fRV~$&Ix!8Bv>{`_DHLI*)qx5
z^51u0enZ(9OA*R*0rand7D5siB-h_R${C35;xrAo2pRC9=g7EKdxX}dbvIn#*zAfS
z_+?-@7H*Zc#?DsBr&UwCph!CuR@R?7&hbAH7fSx9E8+H7OM4<v*$5+u{qSsG%sQ6l
zlFgE?Dy5ZfqWCrUHM<C5Hb%}`*G3~w32syi7*m@H@eu1O7vb@X8;j?{o;vnx62v1)
zr6QlG49a(zBZ5HrM?qcLy7hsc_`6)kckO&|`18#ZZ&{tAHUY>O>oWM2NbFVU!F?39
zv(>JLrQj0K(KE6NoXb&|W<Nn3K%U^YrQOp8)6veRK$t<9BVs(3Zu+{Inb4Tfz|6(~
zOxE5mW@3IKsW4w-Auq!Ci};Ab8Ki67M(1U|DbBHjWmir4`CGVZv(Dnd2W?_&B(Avr
zRyoy+T8KJ5@XDYg()^=2JgVR2nV5g2LB<ZmpL^zBB0q-a=>469S|_MVZKhxmH}%-U
zgcbs>gs|3_6-G%k<i-u8#TMHio(9=~mHXcQu;+uY2E`FI_47@J!~_E#-0lRpG1w+b
zD)(NS=}Y4#t#NDV_0uC32a9h)2zi?@f5~j9Wp^jC!T7*)LXeQDq%%cB`gmyi(Wr0L
zHq}{(O-e4ivCK-{?(a~p<DDdmo@(~McjBXKCBKJX_sA|^D!xA44moW}4^*)L-PVLj
z@8M2y9G{mnXLAyM%8(lZh0l}F<ZPEuiXu~R>AHmhX<O3_SxqPkyo20Q7e96N*|oDH
zkF7%p_S49I;B|0~_#BS!=tV?OFY3A{{9G1sv}A_J<!qmvHxPMX$f+k!e2}E*PB86K
zE?OJu3|W<c%tRLJ|9MN_8}nBPNFiMlRd|3tZHka6sbUUoc&2aC!{PB2Z;g#J-Z14G
z3t;^Q;(ul(H{SSl&0}t5#M+WFytk(YX=PZFw-Cr8MkrZtkg3kXsF_XXv~yk{EUNC5
zw^@HH393o!-of#6G@V%I+t<agBuZ%^p5y&vYAxvF^|IaxyfxM3>5|(6>i>==t$v$D
zy-uf6UOqQRB0zr9y#DWL<LQ^qAJ*cHb(PwF@z}oh+I?f~lM4M;Tib9O-TY>tScG#%
z*Y<EfelFzrTh!p@-?InVVdk(hMQWDREQ!4`McNmR6;g1!w#MFNQrMTwVs41!R%%Y!
zB9>|xI9_g~95rorjRqh9icHpM-zXxe)}6`$d>68f6>7v5bTqF&p51d==t;k<Vm6oV
z?&<2U>67**F2r~nmRLYJUsw4)!F`XqnYQfT5iK^&V&Evb46&Fh>M8;U@c>NT^mnXn
z=5kqHqDH}5h*3g@l1W_nZ+?Z@H8{ulpA35>FWq1E&GLH&VWB{GO_2d_3WpIeNUdbd
zXJ{EXIbYc0U(ZSmnST3!AK`@YvHy9kMcp_Rvaoysptpx7BE}RhUP*$cL+SKNvn-;+
zzA&4)d?#TRK2oiTJV7J8Xa5v`3g-+&`Lrh~Av69q<Sj~K+G5j{<PW65x~Q2*#|%de
zXF;kihy}cbf>$t6YnC6hW?Uac8i`WNqL-UmytU7Wmdi=?K>AwZVia@mv?|?|h@=pt
z=#h%t)+M@L)8=2?((7;jj2K1b6rL`3q3*ZVj?EsPO{v|QbuDD2!Y$jQO;EsZWO}u(
zt*Ce&3Y1{etBV!~G{|Yzw<>fjO&GB*#L$YzBLVA)-<{#3K403L=AHPz1yD{oJ1DQ<
z%P8<04IWAdf2V|ekG^L{cv9LjX@!`-=D<rhOwuH3NxxVv;}mDq1Q_yZ=J*Z{gVY>#
z$silGHi~43=2-5L`S01ii}t8b$Z{B&Pv&j(oFJ}6*8epu1bL9OOUD~OEQR?SdT$H&
zH~=HuqCFs>qGQ{k{=&GKdjvg(l6FO;Wp_cLDrII6`ak9MdE{?Re<^M+W=5rFID4NT
z%4()B&iG(FAJ6W0>IAw>R?)KJ?Qrs8sB<vCpRff9UxzFV@mZ&IUSxGUByZ(dDGhsY
zVUyS(STLDf&A-2W@3z@Ji^*=4B1b!DS^(6x;cd6!BSIs&%q>v6yiWU?>r}683o;~#
zv{eQ+^Zhs+H=&tlkX0rf!({tw;uz9G1;kfbAGcgT*QFe}(@A3qlMi-7uqp6SZEDM`
zmDYtwg-CzUTu)gCx%7H}e>O+)+_29dQy^`V3xIXyih|Z$Be15OCY7HTjDimTMF1%I
zyuOp3>@Z>(IXuzQ#s{JA*SU_&6Lak*`zjT?42&l}1;ly;Kz?x@`a9v!d0h8?YB^%W
zSch&AEebLYP~!41?#K3ck~m=B958)-D>*_P0AX1Tyc$%WYYvf$A6VV6Dpa3|gNHqy
z`N2ma!6_{GB!S3AjKNfH(4dmxKrkfX0CHFs>i{ycXc4?!I7v|*w?i;=EHVL@waq=L
zuO$g*C2FqZ^>6WQ?BJH0P-En|aHlm`h-vqGsDK#_lotU4U#Zyd0ne&hRn1OP7OA2%
zzY>}(pJGFG0za{xgQav9R24nlI)<c9d=QE*NL(4?VQW=`L<Eca51epP0d%ZixeNP`
zJhv*yJGV%0ci>D|OiN7@7&)d@Kf7BkBU#7eCY)&Pig~ZJ7M(ZVpo1hW0vz%bud`Sd
z@MB$eTgv%T_WMI+%s?UVOpe;#z$?G-L@|yccae_k>-Dd;bmeE+p{77faCjk2@w2qX
z$M8-(wu3ohhXC$Bs-CwV;7!ZmV-O-C{0@oQQcC(#64X6y<c7n?KJN=)__G2=8Syz>
zHlDxo16%VjZ4N9oRUsL)OXZ1VODh%gff~}a4X_q;VLXnj8Q}3X944p>fHoELD`)t;
z7X%pu9UT4P2>#=g_C@>Yj6qT7kSg|)6i2TVtdeJy_6tivAzWk)d=rN3-hgvoTOP?=
zbl2;V2LG0r-CHJ%rP0g?#&h<<*pO|Un`1tPGzLFLSI1IOVp>~}h5@+|h#92Q;vF9w
z;n01_nc*^#uw*co-9psTq2$ZDbyM;0VyskCHJ4WnsleoOm}qXv&-C3<L@Vf*XvBqs
z^TerR-}{~m<E^w4Ad4KUQ|z(-d(Eba*0gq6IDLQ0=u(M}x_ks$8#dG5r21JYF~oa^
z1=qxZo20UxUgN*l2#ep7h`u`{j^U=$?t&h<ju?0rLS>JGbCG)Hm~8bEIr5=y5vh1}
z(ZM?ahH1ZsT=AdB5^hEdsS_EzI`zX2+OFoKdRu9(8Il=br;m;|#;JL2d74h;U}S6f
z%!dB*LS)Z}c+3WkJTWCRZs#Z~ec1!f*KM;E+ZP4=GG<&thAyH|WoueE1O`FlU9S}w
z=%0leu!nJPKe(|LgY9)|yO0D!aE(e5K6$*;bVZs*d4``@)5*ltJxj2>TPgYTI(Uco
zTm#+@);|X5`;{vnFonUo>iH`+&{OL+XqjcQklhpLQ}3+UkxWgT@7u##Ga0oZ+HBic
zPur28(spMJs!Y(dP=ut~wb`6;SU@unl;qqm$DTB7vQ83-+Poh-*)6`LQZ}~nma82a
zP#)H_*cw4>y`V|9Zsvzw?9KPT4@n3jydo+KRNn(b;m)+yMKZ>YG!c-j3G2@l3uBXB
zJlkdJndYeeWKO=p!tjv83TUPyu;OKkNdlf~>u=+b1wrr6B$|Fp2#qMJsvq&0gV#6g
z<SAQ9a`OcN-`vVh9t8F`!(lDJ|4hy_1+&G{6Yfa(SLNaS7C79Yef%aDwlN-juVz38
zh6@VB;5iUY3sw>fHcxz8M!ISO-*mCzMH)Cpi@NLH1wGM(_}uP3CUL|GVYmpKz!o7@
zBon$Pw%cjokh;JOZMp`r^>30}?<1a5Zw>LM?O7N?)Z>d0=l19^_HVYmYKm@Lu)`OE
z%WI5?FrggZvo+aX<^h58Z$vQ<iEots2|^F}FS=c5Kx-OK!wI<>C_a0?Z{gc;-`)B&
zclYKD^oTln-W*iFzF$0i)Tr58?HW6bF)y<cs6(OW?z;-s%B3uw3n79)E@qN2fY}mu
z1!A(px2n#!`b+niK)~@FT0lb9lO6N`{*4=TP2sGemBhD@@!$SHDe}eNV&#Iwx_khr
zrvC0C>#h|)Z5Kcs9m8)IC7A}q^CtVT@(^ZHI97d{8FJQBHh6k)4Q#cW8NRkb2ZG~I
z=NccvG6b4LE`<eN0HQRDs0FQR7zghF*|IuyziMl54OvEtXv(o32HtXjVpnOs3hC94
z$=lPsE=vC^Dq}&S7$+j=iWeVhQxSki+#a~-75|udzbst~yI}va$9=JWRe4++0&}1B
zQ=!#wTsz4Fk-sCvg$%X)-eQW*rVA_A?b+7t)$#81iLk-v^Bi5BL#V|*w6~}G_}cIO
zIZ1DsH(T}hcS2|Otqk~A-Vxqz|8(J&kE5%CQS4nw%0W#yBN|Hmd?tEW%od0ZSE8y<
zF~ywyYO{Z-XXx#KlCO1-dBC@E-)mAe=65)(%08w#jcy=hdyP%`UYGr~J=Bi2&-stH
zdxnT%{FrdJ`^x_Q%?%WinHWlApr-_>G%=fLuR5<s<<_6A_smQ+rDk?AVC^)axyn?q
zbp>xq+}POSq0v0%uG8xv3b4(0*X@LXKZNwsE|}1zV>YYnGsXK_+&YRrNNzk|={aD<
zP9O4)e|08_N}OP{lom|rU{8xu=ff0352`bp#(fQWSi{H#8D9+3v9Re!tUbCrIutkr
zroXpw^4*7RAU#fUitbE{htZJxQPDl!dZ53J@}Ht%Go6g#Ow<+B04odAk!$@ah)Ky3
z5&DZ~x)0|__Y)DjvXW<y3ekzEM(E}I(<O@OF&=Rze$3V~74Z>!evdX1Q_C!TM6q(E
z;R06!p7~Bn)TuwR%ujODb80yumf+^yzN<#XHGB+|{WPiy0Y}b7SyO<2Ca(c!PClAG
z>w|v{1~v&i&-}di0YTOoD#srKm{NbDYz3;j<D?L2Onz6Z;^intAS$D)N9u$}E#|3-
z>=53&+#!Jy42Ra86C@%#3J)|=?AA+d_1TAF>62i(wt_An=3|Pf(I!ABsbO|XhX2Ma
z1c!!U!9#s&&Zu2x96LdRghvLE+UE&;WNd?ORc3hkn1slA2cBJvH+Mf2Nv8$g951{)
zMc*==&sO00qR1&XXTN~82HNwbXI*r2M_B33tX&fMMd22xVw7-RiC7q;o-U*jc;oo4
zj6J_r3o$fAep})UC~J;#JvP`AUAC3^`fhXzB*;n`5{Z8&8RRm)j5k)QC|^6gnY1(1
zf}99s>QG9H2GX2QpdR4!LSWX|3zASbfGn?)k5?ti+#etOS#V5@B+0dm473PRilTvD
zYvrw7^w&Pxj}KOU`zvO{?I~@4g<xj%)`gh~(>e*fcO<hEo+M1oyH-LZ%6p-ivmWRn
z#u+9?yuz7c+mdCDtQ4^HFDJjl*SVErYNktXe-5Iu0AaY<oZh8lp~u}cyIv+fU@ppN
zd^=4t#-7tFr&_<->C$~HYss?19zYL>Um{it)5)I)7Y3ZDk$$0S3N=&}u$w^F{ku+g
zKy8>MOi^vtv(;~g{FU^xM=+v72yL{ex8Kikuy>bb<vYve-A|D`8XQ#kGMFz#>|=%;
zJqT$4LK;0q{yATglis;Db?FXV8_1(EP!^IClItnJ|94${*Y*E~e1#Q}#0s%`DxxbU
zsS@fAVME@19~XvAjjd&gsJ6SlAP^^-QQag>P0O51j}+IjJV#{QqB)C|j1kux#Cdrb
z+$kd?{AiYpp+^?|yC8VWn1TnC-rll5o<lV57VNm1SlwO2{(zuqc{4$qFSfa<8!4MH
z;OEd1o#^}1qB=y~(j#L5O9M&cD)lV;!=kAEus+!grkvvR-o+?EKoF{bI#`B6s8%b_
zz@R}H9ebHhu^fCnY?W?QtO#~}$3%_@ZO5c8DI7(E%}zvw1ynq-kA)lvC!=^~gZ=~c
zC3CFTz0xaTU!gB?C=)3Xd?{O(jHsy}4j1A}f8_STSN~H(^8598)PG@Lfw4Y9uK%e?
znV`;ES=e33J+_dym&aa_=u5dVL4=-hqFLxiSpIByelaIKq$->8T~wFX(4zy~A#E*y
z4V*>(Ul&sC@^Y6ZIxmrE8cGzkXMSJVL(}KHnEwPdFEN@VckE?s!H`luoZ6_}(t;Ww
z`we+T>F+lw`2VAhy1nw4%9w|MQU5P-DB~G0f2mHb0St4zYgNdN&KflURfU*^xISj#
z|NA4L5~~9w5%f#8MI1dw|Ea<Zk}k!!5fpPgiEA4ke+GMI`Ou53ny#pG7eN#LOEzSk
zD4*Z$C-QaIoQ_X#od&KqW>inQCC|Jiy1u95+fUI<HehB-2(>INEoAm=vwIF1Z)kly
zE%hLOnmr1$7b*z$w4_c|DOcPU7Xr@n(Y*)Kg)2_xneXK0aEJUR`!C_(S?<SeYo0H6
zCl4Pg--u^vG(b>?!v@6i;}o!C@{sPJHt?yMB3(J0=DS`O4+?4AM*MY?pIqEIi4-MD
z2=M<p#$^Zc?+9*x`BEj=ARAOUbuP4GT2$Gb=bAKfqszT94;wKmiCaFQk^?+aN)FUt
zHA^KI7UFO1U$my-=$rx^#5#0I6~Lo#o)|-xtlXeY3UX5@4Yov>CJhWD9Ep@cDGzL?
z7`r5}er~r=$)x9}Ce_`v<GrAHbkC;hx*p7%`wHi(*g-cse$DLlrcl0NxaB0bp~39P
z4`2_A0ZO7{3df!?Llj%P4ES15L*)Y-l*r{wSZTSxDBqO0KLpH{_t*ko;i1;c=otKu
zopqOJezYJP2GfVg#hnyhek)T=s>~+Np?>1MlWlzWv?+@<!!1+r+5?R`6^U}pAf5?k
z3*w>?>K{%YK;VN<)Iv3Er%^g#o2tp|BO(4%`8q-4aY{BsXOuoG2}JhA2Y!_C5^WMu
zv37+~E35w@P}Xz4Z?Xt{4Yw69B{VV&C%6xhaBI^ePDJ5&k=csg5HV)8)v!*^3>9KE
zo&6VMGG{#0s`+-1I%eM;QK3ohqD+-z(`63o*Pj#ILDU@pnP>Tgc;8_5SP4soj0TDM
zJgnIu1p@mIP6-&n2RMUMR3Ju6-+N@MS8oH_@Gd^Ca(SYVE*s$pcr)zub@)_>s~o|#
zm+q(TyF@;GCmhNdB(o$-#cr~l!)-{Z^YceJ&E<WnbsT;LqpH}H>QZDisCUj*hLdHr
zR34GjY(<b3n8q@ESa%`_{8!FArNYDaqcpxJwxpjow6T`~Zd^tjOzJm&GCuB_G8tJj
zVTq-jjEY+u{I8JN)+w39C%Qt3Ypj<Sb^!=(2)NW}Tri#kpbw^>u(Z@1d$%%eD^9mA
zz`e)o0pW(9Xs&&?Z+qwG!2bJFDub8M#%r!UQP(<fu_;IxAEGk1xWPiaVy?eKn)l}>
z_P-h>@b~qrB=6XIg@jk=!)|OnrdCH}Oy2WVZ3N8iipSj7k{i6-LosXo*9g|f+XxIA
z?g^ah*}$*W=t+mg_yZjjD9^~(Oqxb%>uO^|2b!+^m+yR!VvvJAHiup?BJoL+5a$rJ
z+$}#l;YtHNNGg(6`$g5IsdkuZR`F3pIU<ArW$K28fTCZV5Yolb0#x6VZ<17v4w0N>
z;#F=o1G1Fe_F=h;+3ty)WSnR>W}RgeTim45FKc|52Y1<Pwp+9@m>*jtrNpY^q_}z)
z4fwu?Hqf&DTadZ^Cu0M*LkNApSfO)~^o9AgZPq+@NJ5RmvEPBn)vIh4xh{6F@Vw^y
z;%{Ib)e#toi~LNwh!iWsqC%rmi5we1G7aJg_)>A6oKAw2z9B{Af#kUk&B9#W(`K6R
zD_D#zF85u@$cq3k?dJxw+{QO{2=;Xu1VPkKP4;vq_CM6Suu+HwJVe*1fr2q~4W)^O
zD0KmlaAlZ>q8~>q(60_zffXUe?j3-B%>37U%#nG_#X>8*(bMSJyS!@H^jkQe`wX}J
ztO=W-)|4hG>Rv#)PV0-;3qK>XzL>lEQG-qD>vHe$oVA~p$YFT8`68MSebSFIWsycB
zm`SgnYESAlYc0Dh9jDr2M9BF-sStD-NSQwu4f6RPZ?C#lo5tx3&ZTV8v??I!5C6=f
zy3SYuVrs1IkP@W}2X6MnC@9rfvPYr1U1PpzB1lJLI36s{Q2@5DQNo7?A2Gpy%tb8E
zPPP~DIh#T3)0YgZy`PPX9jh9ppYO+f_Oi6qU08{KWlByIq1CFv5ktffA>+qUvSIEa
zwKZMC#W!HUX=!}zDsk?zhk>wC1_*&3l{Iy)fO)Fl^U&3XO_V)$+8~dLPOFfQ$k2yr
z5NHl4`Vs%qCIs&kqHy;}>ev?e!5}XTt36>eX#8<d!&l$3yB8>iqjU$-Q%I}DT-Kwc
zC40WnF`;k`*iZ-<=(@F8Esa;P1}vW}F9lo}%PTdzN8H|xcDxdU;sGXkGnaU1ey-nm
zap8;p6CD$E)DQhPT;b7u&_k37vb14;%^p4(6T*=3vfv9jC<^fZ$+!o-_Kn}Aw_Qi)
zP*!rl$4t~FG^n`4aSQ*^rNAJw2E<ptx6Q5##SN5OwFR_Udc;;BMh=7bpNn>5@IQw(
z1XW@z8sc<nAS}pctsdp?_b73tQ>$M~7EKQ9K@K_y{}OsdN3y~|_?-AILF7w7$p;0P
z5^1w!*Q7Tf;<-HnjHIFDO9CjNbBG<Jf9imR&G`RX<iB#uAr8l)lOLi-#24XRFYG5_
zGbnIH%F?Gmtw|xA$}~Ma$%n^y!g}Hqe50)v+qS^b1Vn0s4Pbgv&1O*}Q_%ucOoV=(
zghNXGP@F@pnZQSk?hRnhtLvCqMhr9L<y?^0{SHKLnM(_PABDla8uA}%@-oO{uCt7O
zF^#O)@!YLTFpcd&`XNauWPpmY>{>e4Wd$u}z@g;n0Q0@6xqs^3t4gopAbzD!1l4~f
zMX7NV5|BNR=v<|ND|+n==wmFx(m2_PhV#HA1$Bd(23tNzV;B^sP3Y8o>i9~3pvE@w
zp;%bLkP2z<+_!SqG|`YQj4SGku<RC8&*J8mZB%A0OhKB^-w|b86NPvGf&1z#u`ozT
z(8D$i`vqd{=Wz4ro&@(67nZC;m7?wC(z{^O3y|w%#1B`zrYo+VVbyqDB0q+tbbK>B
zymwsTl`_J2uM1kRx;;~LW3sgvH->6en3fm!_Irn3L!^e&@pYQ>^?15*=sDV=ty|wD
zd!&0@&!o>!jB?#b6KB73((1-eZz2-l_x-y27ECK_5h|5NtwPn~y~eNWf4xC=UFdE2
z0d`yY6^!osgfr6{6_PKx_HpN7!)vb!Ww+gb`Q`$e9zT3vow0n|-dnw=H*d0eM&EK0
z`|awbjeZO!FxI3w`U))B{SJ9ngqf`T*Yx}i>B=7gv@W2k7mUq}jJ&HiIi!;|-|48H
zXKN;(+^;bY!}!A_r<wnsRi}12+<*r758x|1m)T?9WDxh$X)1)@${@9Hl2~g6FN+ss
zS7ur(I8Z_>OMD9WT^<w%CJ?E}^stUX)U8<F7@{+`Igdmo;o%GM<CoHR?*CuLgvuHr
zwMHUJ1ujbI5B+A^GX?D#YAWsl`k_;LMo%0eCZy^z!MxQV8YG~)eBgGZ;Q$!|+pe1n
z)C~Ok(<hJr{QZ_YFO9+bSCCPgMPGnX8~sEH*G-5KPmcsXM_$^6L0-5ju46pdpv!dy
zWb?{_O2>zC3(dwq)v?K+W@=v2hqDZT!y{cVO#U{0_a$ZFgb{4Y07|OW&u^!1vKz{+
z`;A>Sp)Ddo^)!G3ULK+93|tIYbVbjBgpjF_sb@(&{}c`R$(w+g25z1A4f=X&VW=)A
zR7Rw0%Oy@YbT}2uDzEiY+av>233dUV;SWO<ok*jQgvA7Cn(y`iGNLwgaHgF6A#CXq
z&Aq@xzQ}^M&S0oB<2GkF#4_JA0?XBps|-=y27+*F|6D)>WaR<+V#T#};ddmxbMth4
zGtYrj={nUbZEM2~WDjzyL!iR6Gs6;7SDuRQBC3t-5xb_i%Zc79+0hwU*566rzjxBR
znn%h3-`)BUUGpVE()NhLa&B{uaL@5_->lxObQs?Hhe)S9WMBi2Lw53@(fL1F$d<Qc
zDu3$_lVt*zSaZcRgh16p71cG@XiR#((QsS!(`&bNtjn3}E)*+h-M@fM)1Ko%bA}=c
zTfD#QyvYr>NiX5}$UN6$t29ev=XW9jATAKOFhRq-Ci75=(mVVAcJCtQVtrp=(BAJ>
zRfk1IN?mXzdt~3vy*k|>UN-u9eCDz)Yg`^OESG^Tg4@?+3R-i$OODkTU)O4`<B8WI
zPC};pqInM2*VhJ1-{!8z(bLzZ$T=QUMp2^e{;@e`3&2t2JL$*Gz|(ER`Z2V2(`irj
zq_2MC@m{MAHJ6Fa@XKa@Q;A>cMO%AjFG!-b6nfOK#jImext}Ju%P4uaeP^!qTjpVh
zPbz?Pdy6LHiDDhwk~!z1r^&PR{as$f`VFNb_92oWFbiu^qRk9B*s?{=7=uT|to^a9
zrwj*&qQK-_!Z93jAA0Th<K&_`9Dn|lhU3%!+`!!C_HsUFWl3%p&X2rIGJMA`t+rdN
z-(Lu3yy+k~S^c1mT#wsOYWa0S#Ph#3FtE-Bmd|oMyT2PEW327rQbD@->kNgEl*BNc
zeGJ=7(8q~B;%*-KPceetaonZn&8q>oHA+!-Ns{OG9wr-06xgp6wK>PNCiPw=SCZ+9
zPJJH}qQvTRZu>@>yY_lPB#NM2g}y3)EiW`^mT_TgF_^9|{7~VSl=yZRd2GptBCt+#
zR0&5`0BvOK{qK2FW8r6z#z;T{0`|&?_?Q6fpQAppuk4jkg{Hn0xZ|<(c~P$PxGuw>
zW%3`on)`*GJ|X*4qpcgY2B!t-umD6~8T!#x@!E01T&XpXJmKqWiRFs?Wy>bq^C5x!
zmh=9K5=(;mn1QD0a#;x79qbE!55Ty$3RJl)=`%-(6~#v$GcP;Z_O$5f$3gQPZmlYA
za%NBxK?yh}r~xV{s_^2vcq|CCd#~=h?V1O<oIT0yb!4}Ljb}$`xK2zo9|hWRcZ409
z10>qGbDRQ1b>8O{j*q+pWE^CCCC990{o${d0MBdH)nupQ%!_>NXO@ahQo!`}q*cFx
z9G18{2LzJzV5u}-AInUJh@=})7z09$SFnRgctLdrBlPX1ckegu{t^-*+DOgiJGZXG
z(pSv0s{`a?MZq@g3sP?H5roE-CI|{$H$yg28kRz_m!3ekU8(Uz(PX7pRML6rk>6wr
z)kt!88$E3O&Je|T4P}=*VDgzL!;Q0nNA>pSv#vB*UF1bdBOXMWNNyifDH{w)-vc{O
zR?`&>7G4(6Y6H_!bO}3Re=ct0QsNL=z|keuulZ3XxmBRx5^=FM-1-4F#c95xaiUI>
zUGWXthaBhYe|^zyb}P=l*4eK#S^w^ui)G{T>Q{#}r1i~<T@A#5USXG^V8c!W*_IUs
z65}yFzKxCNLtTl<64pe~hC=e6B5nNxn)!^IFTchVN!`8yR!6SrTe)C0UYV}ZofbmS
z=yaThD@8%#L!P{r!&zd@eK2G_eBP_Cmp9b#@Qe8NXhSsAqK8&Ws?@$Wzco#IG7j1@
zuaj@Kv=knnh3W4ALX1E%HJ<tw`1IUs_3Zo*6muy(ahFIP&ls2C#ME}*iw{{rB{2Xt
z&OaJ5sg9P(c^L_2<vrbZU%}dtsXyTJ$DMVBHABYQQDE^3z<EK)U4&y1_~aL6W)QYa
zz6jxCvWk0&9ZOl6&vk*4&HglmcuBd*N-&I1n%E332@VIu=_!I%(0s;%zP`ik=ICM?
zbxnA~1h#;av@<5uB?U3xntj|5H}Ye%kwmX>sKdd_-MC_m&D4%WGoJjk7RA_}?$H*}
z%oHK#(<$zTt{)$vTf`-;Gb=Y<^CXtY)hznNK$03JeS-R=KiW&7a7EMy<d#_?P^j&U
zb<l!?W4VDq52j%=Tv@g>i}(#;lT>+4Y{PzHuIQ{<L1^#fm4!5&3pD|2@Z}10-K@3c
z!x`u^rC!2?-b%ZnNulsda#CsTWhS*%aC3I2z7Jxqij{>R2t8qkEn|Wpm;_Z&PR49>
zcrNdF-){=`33TLN9i8_^Wk~p4BQUQ)+gGL<c&Grk58qFMlahu3LlVfb>}&JWT|D<D
zz7cIgrt3V}y0Lv8h|@pi8G^REaNz$S{{2bdsWTBPVxjk%2i&ro@|sioJ=FNfgNC4o
z)$6l=c7bn2!iDJM{0?OaSx=_+j~-0CCX~<XMf{djai;4Q_yk5$$QV_e6vS|A_WIvp
z)dX0VlOq4yS#=6$b+bWFt7Y0$(;9pe%7os4`60#2MgRiNbfCvS02F)$#-@kZ?tj5}
z9{(K~yKVYp_5N>q7aUgxd*3+h5KWy}?Z8D$7~w?6ghR@m2wTb%&>o<A<tD2MzgM5e
zd=Y4L7$EUGTv!;l4~2(td^*HSfTR%e-qE1F!B3c^BtZ?vLk~wOZzuDI9UEqgxuFbn
z_a-irC+AWbhgxhGOIDYvwltE9k?UXGo%`=<!nfOhu5;>WyD-MD+0TeS;ru{fRe!-P
zCn}eYVAV3hAu|}HQZ1wjZK_-#w+?=@u30Ikx8+<i?55~{Wp6C*AWzg_0)(}+SF^t1
zz)|lRh@b0>Tuh&l?L@9pVNVF|ePV9akN4(mu-(s9mSdq&OG2!xuYAUBz`$ZdbQx?T
zf7kDe=7lgEXG~m34SKmX8@?rObfI&8N`+{kiHKoL%h*4qNh&uK2r-UbT+Mh<u7ayo
zm$02Qs`}yk>_R>wZxGes0kr)Z)u?{}$^7PL9m#&Al#3<KJQRd%C7KRny3|t=AP2q@
zVKYyh^1s8*A=%^G;fj^gOBe@&x-sMzBH`q;3+rZom?mxiSfWnLd3xKMl?iIk9xID$
z*p(V_>P~)tHfnDEpx3|tn&n9Gu1s8X1}adeKQ{OMG+ZW}Z8QhSQZ^h&S64Rew9Y^X
z<gK*M;OeGJv%Pws)4km>0I$DDk`cXgVtOGg!+V?i@3eJrVryTGAJH|1Z@Tb1ddV<e
zR(@e%UHbk(Rd$)@PZK~h$*THw#-<3`!XD)b=L_YdmMW`1GL+HZF<cqhK4Y9wb;8a4
zWrR_vf-BVG-2f60rD0G^2`9-b6hv#xX3YxyRLWue-O3<PWDVLi0{x>yvexUj=c$9x
z<sXRPSjy`e8MIU#-pz8DZUW>+`4`7c-Skoa9W-f+H1XlYf(8DHy>tT^bP}<Xi#*(x
zp==Chuyz})ohqNm$*H{l2oZgA`c4`Ou^U0%Zy{J<oe40QsRw;HGz2=K3v9qw(rObL
zzRfJFQz?>B7LjF4v+ixjmRx?3^60l?19{u-3xDjzO57zOBR)&gqo5)Rww{VY3X3lW
z!9r6J=Ygyc7+@kRp4|qJ<oWx9fknn-{uNHa!=F1_(y3`nb#{H_DbvK1nF%}5g=+Lr
zjDlJeMh8xjlU#f*dCCUSSd38!QKCJ1G$x7g$OUW(R;Hy|I%rx7PfQGZsY9w=4M6st
zB@YZdxsQF2IuHmg_k-VsE#Q)LZKd|hB$_MX64VeoF^@H%y#AyU4n9=Ct>mwlqq0?i
z+pdaa8f7fd7kL!6EgTRJg}S*)4wXIOKG+NOVgjd6<zy_deG1b9HT;Zf&*#_q-WqNh
z$U{ReaxuGF?Tf#YAF0EVN=^1=rE)RtaxP}c``W7GzuK4dM1J=e{Z3MpV_+7~JT#J?
zK!`I{Pouia?|-(sR;vhSV4U6SSJ4QUnTAlNIczzwv4W%Sf-wi3G5OowVY%gdnlb~S
zPzDg6p^*r8RTmN(@a1Bc+Wj`89nxHaLlZ9+zRU~%2>+UkvM4w{6xwk-uS4lh?EIPG
zEc<l1QrnSP+tBc$cFu=!U;S089#*_0a1!VKGQ?#q-%QGE3<Y5|B%x*<8O$vsAbeMD
zEgPbE<Brz8muUF)?m3vK^em4$DvgUlCkG4|Pj$+EXP1czTjw26rDKq*&+{Tz&5`tl
z;;}mA2DQ9^@L`jMim{!O<S#IK0qaCD;Y}&wc|phdA7dpQdJUSaP4zD-p-Hu6<ZEFd
z&pv|oK3Jz#hUiQLWpm=3&7zHae>UNqjmaEQ6=zIr1><txZ&4xymG)SacibRbz!N^g
zb`Mk-oen~qLsvx4P3pSAR@HH%EGrf7q=mqo{V}R&4eT0|dQBl0rI~CaE#L^u=YzL?
z&#}zHVg+>cU(v6UgirG7;>euYc6QI}Ot<^r>Kka-kA`vVnOY@%6W3f6F=xhp5f);y
zE8=7>o<|b+hRNhA)uOI2+}JHMz-PlF`$WXv@+u%MdU+8-kg@EO=s8LcPLaLJcJkv_
z`o_Oew=aC5Jzi#=9OX*Y`(@4dHJ{Zq>=jK36?O#dz2jqUeI6d3BcaE48~evUTa|3D
zF6;9$YQ%)pDw5*6hDq*mRPksd>IiecEoOPfqWS~{)q`vKjj`tL6q~H@fnO+pr3&eC
z!}jru?2YCZXRMo*<?T352m`NOYR&i}`hHeE1kh1R^&Iad3q+I!ZFE3ik|Loz)%|!I
zB}>XmDAW3-T(y4b#)WmZ4@O9}n)~YsOni7Ve>Z@Xkm`RQMuDQ9c~ty&S3)RA=*&b^
zKjUCRq^RN`#c_*AR{sDcT&Q2(=$>q>Q%X0Q6mh0qlqu;WXOb0>m~+%Lu8Dv!CEa?>
z-{N}b=wK*_Ah5uT@5SDC*dJwvh8a2zMaFPhaEOOs<s@!mZ94JGfj^3|s&ozg{}`S-
z*Y2q`-aw;;FIv{L%ca5DzgW^3dPNFJQ)p)$&X8f0nS4v-7$yVMJmPf-hTC(r%~ls9
zypfSnK@7izL!p`*;6%hPAn{6}N6M4fSJ?JxNv*T^{igCF4_Kv3g^RLd4${+%=&%7v
z!D;!E>ZmPt^T5Ke^6&W7Ys2i4=z`F}J!UTe272XaD%5FBU@8kg`7ar0T0D}04GrUd
zbj>`4*g~P(gaIqiB@*EN0AE|I1J(p?tAf|HD(TL0g1pB#WoL4-{9T!$wON5cxXF<g
zY$Ah;LQ`ZspLF)8)_`N>PdDix1w|st<MZzc;&4;`iaam%Su6$~j&>vh>rNrqP!-S;
zSL~`?R=#!ma$Bu|wz*+c2xZBEBfSG`!B&Oy#{Sqp-XCWHHGb1>GOKj#eC_JTtV*e$
zh!(qR9e*11uFsym3NK6sM=+9-X=4S~$Pmi&t1#`PRWM7dUHi`FDSbqs-`Mjz9{aCv
zlw=*mcFjhB+1#@uV`GW-%6pU(X$DrmR}*AbbY1-H^It)uw!-~%>Y!21Tr78WMqMFK
z*_8nv!e8@G{zkj?UVG2b)iviW-DpwW{W4zK{&W-OPeJ#YD#o*8nYX1KRB!WXYcmsZ
zH_xUFPKc$XN3i;XVPH>Gm{ZN&)Pm?&0xYczIZ<FQ@vq!rALP*yi)aErI{!|u4GOJR
zQ7oJJlP=S$;|L~5hO4n8rsKK|kg)3QL>6F6wG3wi1(u-=QNc(O#K3^o(Dps9*9+gj
z6jX)v&Hno9WFH<iH#7Ac9}FTWejki(B$2qWV_p^}$^f)>Dme@__pHD>Y%f2m(_>7B
z`@1PSn8VYQzX4$iRNNRUJCb$k5KqPbI#jb#>u*VU7{NR(6M)S9t#8lG@*l2*+i#oC
zt*_R=8{;#zrzFF#%h9TRjeXF4t^K-v#`wIDM3`K#&4P|k7=7tVN%lLw>PsOCqV8Sl
z-Wm_TE7}~ow@7ou?p;f<8jrvkxE#8>KuW~UZD}#ai^6rvR-~I8QN64%uFGu3toE#2
zu5Ne2L%x+~)~T(5G83)u4h&rKmzvVBl0@RbFG3~Klf>vlsizZ8Sq18FT-hHoD&Lzd
zVH6S!DErW7yZ_w}%IofaEKposbPRBLjWz8wim?!V7OZsp4A8i9>JKXr4#q3m*`=Gf
zlaj`{zcx}7pEG^)p1v5p`npzg$v&cVc%gk{KuT?o*RJbDy$^@<^V047nOdIg1DBP6
zSkGTgn<P8JW+v2D?fQDmrYh+r2&V)XDy!)gsdI#_gNq*A;l^{<+w49Ejl{9AUmg6t
zOkN$B5GoF~i-8um^B2r2=@0W4F-~fBYFJq5cQKNoW(HED!S}CVFVmUpL8W-pMZ7Qf
z5oq3O<#VP&hZ=k*#c*Re-Cy1UzWBQUAyZHU<fo=R;n8z}odS7Lfpm`YXxiF&vOSGu
zMRX@rm0;9JKisu?Iw(Va2?_d_#cZLWcTnP`aY=g0mG8vU0;p*ioe9Q1k5=6bT38TH
zd7p-p*wD!(q%zT>E`#4qoJdp6<METxY&P~Vxu%?N2KJ$9-R-LY*P5b5q$v`RFAz$B
z4=Cjr3<r<7-}HmTiyS>G`O?T?+<LzPk)J9ft9Q2AX&A47Nq;(g=ME-2>)3OF|FkX6
zHq7ex50ZkTpakY@nx4S$XN)zbrIKBnPuo8ERy{4f{ZaALThzT{W9p}{Dhwr%*p-{A
zKW>$CCN3}nd2Be=GdFW49c+Ns+`ySqMl?HfWCozkpn*%!M#19U8u5Q_vwpdi14Y#N
zep%80U3nXVTzxmp$BoHny)7+vNCh(cr`%bxwMbm--?j~KG_=2uOj2bgZD<MgJE%%J
zS2mW_6?#T*60LYw2?N6Fw0l`qx+}|stGS{UDF!_1fb=R<Ab2Yfj`f?XiDsXi66a7t
zuafity>-W=2HzNW{LK+Bk)!;f>g>pNpN=2XU*VPHpkwEf3y+UBXKEU>nZJvKw&{c}
zJD?-nuWsjUZ`a=%vMbV#E<TmBP-l}4?$;!_mr6nFQaH6Vt=^g**(3H|UYK|EEAha1
zW}W|B@4+O^fmJu4Uut~Jiud78Nxe}xeD%r6uhSE-Wql`Igixtwwo)~dzT={oLQaCR
zZK@UfCvr3>S9$Byz<ab_%P=%GO&W|(&xk^ku>**DAIRG7nXfw`q?+YO>?n26OyKB-
z$10%taYp@`ZiqozRGX#A_KbZqsO2(RpeH@%O!;f9tHc063VEa~b;6r~ZHjW$P#=g=
z5{37(YA<DN4do3CT^mNSibN>C$ZTx?ki%_oI~huw<5!F>L?~qicP=~?m_Q#76%lA2
z&8T>)7D}<&eKW?qaqwV;&;XC84II@WTKS!y>S78TMeP2Cew7cs2xj94H@Lv@@AJDn
zdR^n4N$CBmeB-Rk<eECrBQabGfSzz9T9j;9*dXz_$Lr&-m^pNF!8p*br5*5BT(^I;
zZ*D#z>5uQIOjPF!TX9@9iJ(NI%h3jKEE8l6>Ja}hYM^1%9Ew<K!!k~97E`ud<@$jh
zuSLH-F)Rt4B#H3r`^Yu6<`onh27+w9DYe3k$krObi2U!^Jq#L0-=}2tJT~J>$nli(
zhySa*@=GfSC3we_19)Zo&U>ic{!I?KClddb4t*-^y?hqmj}=cf`iTZloAn(c=aBOl
zI@&+C->sj{w5<~fl0-82nJ9(U%mE{O-YSb7uUZGV;1X4oaYy=Iv<YBK&a`3`S=;g6
zKZ*=k^ZKFTom@6hl*Yt;rV6L|Pjvwut)|wk!mY19#;S@?!*geUM503Wn#WpGV9J&$
zG%_@&bamRG04K6jVsB*=j<S_v45JC@8zIvUy!RBQPlgCv?FLPHIysT|Bo?^SEo3gL
zqh;aX^ujQUPKITuF3XqS+A&UVA+(-2CJGF|^DLZnN+Tb^pt~#fFGg<Ela?R1$yS#2
z3%4daHl>%A2UTl9DO&$@N?-RgwlUl#3m!_n%`D-j+s+qHhsY=9qM`F$JA;yhRr@Ir
z<KpoMS6ODYhQ&Vm?n+kNeV>(#mKNFM92u!Ix#Zt6E74gmvr;aL>7_Dqb!qTKL7WUA
z<fhVeRVS-6n#-LOh#(8|t{jwJ0Xl6p&XVh)X#wLQ;rrIm>8tDhIoLkZKy0+clVtq?
zmBObvoPH#RABUcU)(a++91Vj!9pOPq!beF;Q*oUYZeXr?p`jhv!#Ccvme=P@>LEWM
zkbyt_^UaT~ik)~mg6ksQQyv0uhdT#A>;0xYaVWMt6iq3{1OffRJ{`&x8`fxSDb0`9
zmyLW$Ivr||{_4pgwu(Gz!2#jTMQojlf{&f$lN0ny-NXYCw&4d-PEa2RJ!N)KpK2mb
zWCWgv-MVl+vq3cNDBBJO!+ZcexvXc%IGb9FYynCVTSvzy9nwXtEDgH?$YnPWRLus$
z8FUKx$HI&GUv0=7O7k}pRv>@p%w^sivldeRt2esn?_yTDgvIOHMm<-K&O>x^lO3Lf
zklIpm62Q^vjYr+{*xAAm#q(m({%%;(8j?J;Sdu?7)LQNtzyGb+{A{C=&2OyW>MUPZ
znwrJNt^HRC_)F3^d6vInyzd~uE!9}LxKbCqJ<rmO`PJIO2IZV~kZ>7zA+;yO@7Ja@
z+HHq1nIpZ&wLx>AFf5g}jjWKr&hANNQO3S6^;GVuSe8`Qgpm<AHiVJ+WnghAk=)eX
zT)w1DG(VI?8Q4hBX{Umg5n7AvpW7~t$S*u_6K_s<@u<Pv>1##E6iMy?+Jc$?hpu;u
z4lL@nbz?gf+qPM;ZKo=>Q*l;o+qO}$ZQDu3cEzcifA4)xJLk50U*_|CnrnQ$kJbBk
zRLdCLGKc@XfZxSRR}Lg<H%v3&{|xSA=RE9RtBO+&3#$r(=AGD+a5!YB6f1|HQ~TnR
zYmeKMbRE7uf&U)WQqvSVM+*5#izUpV%lyI=3rA1s)pIFPqJny+jtQns?w-tIs?<Ai
zdAa>dH;Bd9_U~y*_nzHNG_wZ49^T=+Lz!Ry$1cWJD%z?m-8Tj?EU=5FHY~TQdMjsf
z0lm$_L=y!vUBUmsSlTq}vzt%gbmY4>kQ-xmizTu|d+IDd(ZdtE2)D~L?(!EtQ61~o
zxr~{+b-l2=jc0<x-+Jp5;#%du^!2Lgs5Ozhs_n9wX4N}J9XQXH%l4uGCxWDJU61y&
z>ga3JCe|4SCQ>VHZDZ}97YQ}rVEhKZ=*Bflaiai^k~(#z1m#u{jf2S`d69{0&dMzr
zIzkb&^5`ITaZ<AkN*1h_Jmwn%-dNj7esL{`e%8<AxJnGkVw0`9zRdvR8yHDAPjCa`
zIvZ7jwROB%lGi^wv+RC=5h!Gyv=AeV^ZOaH39QR*q*1@~jZp<B1P|7vo+ArJt3tY{
zGb@NHxsEEm7tA9I``&Zp9RwexT6ZGRpFgX;LPhCJv;~e2RcYg+5(7$cBV{1$6ueIG
zUEgT@$O&jmg54nuhhpRxvE`eVEI6}g$NP|M%_f>3t&C))azpC@%_6hx1eqRE4r2y1
zr`dvq??O>gSjsn&f1OWN{t^OQ)nb<&BQHcD^0&|{H@uiw8wT7lW!7&~GL6?|qoS_1
z#kAbYVT(}6-bq$bHt#v89#)Wa7zZ@5cLg^yu^2xUCD$NX#DD+TO3NE|zrGniCSh=%
z=PJ|<p%gdXuuDz=Hf_2MU)`!31VRnia>Pi|2}-0O+y!Bb@2}`1h{7a=F^7n1AEd=5
z4$b=9v0iBVV!{6?ViX+Ai%snuS`P`d`qg4Qm6A6%83p{!WJNTbzyS|tT340$X860~
z3~y%~>$3Z5Kf*OKJ{no)_x+G3{s6jm<_tU2Qly_`${&0}AeQd76)l#ILhxenx5)3k
zVEVjyu3Q%pmud;oZ`naAgjuj~)2)d~S98YCJ46K&zm(VGieKP%PYEs9Z3vE%ag8L4
zzWK`6B&8LiBhyk~hBaNk7v+LFXcq9}a@q7G-^ij;|H`xx<s<CdWbM%)96kH#k^ibh
z4OTj#u@Tf52dvt@I}4-E37IpdZU_?0P#+zlJn`cBnL;sMnr>pocXma|HuJvPH3h5s
z+4Gt5^3axVz{-uvWExkwNZ0O%Eo+{noC`E7`YIzgL1!4bH5h$^9yiCU&u)^$6_Tyn
zB1QRb>S%YyeKq)C`M2pE<a2ydd(pP9c{hyuC;{q`3V3^2O`EIPa}x73T8)8^TBH~N
z9Br6+ZJ?uI>lp^cK_bGDOQMboN4=sqhwF$Y))tl;j@aJ&UZ;50j1Q+?R0$cLquLOs
z-l=ol3Ra4qDYx$@F}fp=gJX^Myx$`4L_up2$6RmqzRiYi&a~dJ@5w&y#0?n?VS!t=
zV4C3l14w)607`$y0L(QIDwhp9w&q{rSP0}Qp85+2=9WQR6LA#IS)vz^g{3o!9@%2>
zu|gX|E0U5!6mREb-JB7Dgx#F$(D{<44+MB=<ppPh7a&9un<IHQ4D#QR>@_L~FyUV5
zTiVdn3LV=oNctSj9JH@12AB!o=*7~6A??AOQzuKoVSsG2*H_mI_<3rk{#c`Gm>bz8
zk8*A%(_?WL1BknM(tn&|J&1Sq+3)B4y)Rw&&QB}PhIguWF?!@XiMnI8qu`^BqsC)~
zD8VSuu_XS#8aP^&oX=Cm`Kj%f1~k!-H4#5nby8V>dU(SGooK&DG!gj@RiN}3b%hz$
zCk)nV83P5HRq%80%aJFL=TZ*=k6(YM6CDbU`NTcp-=QT`_2b$@h|i)HBXuF&Cf8k%
z9ce+M9_C#LUF}{o2M*vMWft|CsZjD4p+nFOQwlj-Go9lNZhzl|aV+)w(jZVYRG4_!
z_0+ft`vkw{wOwmiXk8vwJ(YwJ(W>O&hDi{@uM!w38R5krI=CWDbEhmq@Y4sQVQHN`
zPOi;QkX>zkY<n?A1Z#GN7qF7o4d@k|+Zvf+rE(ZGI7)lD9TYUK_da_j0I$GWCGxI$
zo#RhBNvg|EG$W(=(B5%xO+vWcN%0>gYm8vznbRLp*UxmLk>{F~$arm`(&gQQzqf>i
z^8o?2Q`{|?S{&;F8gW{%7a&5NVk}vTlPrzV*2@&i9K4Q{csSa&j{WeKy{>;m5NE{d
zzx59kng=%GB$0~Ex)A&|h87eFU0qF<gg7xAd9>`?y)M^QY#q_Wq7W4n9vodUHYy3_
z4Iz&6CA-4*)=^#8(r7Oq88P80xN{(5vzB8`APQ9Lf{$>~2%l#9gI9mV@Vj7n!fKV)
zoh`!K{9`qKf@)ucjlP(z>(oSa@fM{G+m(!I{nqA$90?_wvIBe(Lu7uvIi`pe23z1r
zpyp(TJa&b)zdshZ3ATU%LL5!2J4<g-xtteEuR+s%rBq~OX(Juh3EtbZVice@>z0<9
zL+ND~Z8`bq7>3mgM+y4auuEhxK~!3MSWpi4spSj9#%WYbtKLqJji*AFZ!+2P5~qx#
z+gF~@j705)@Xs9eWu0WhcJfF4p?it%=$yB#Sp$*B)i2tgdu<!1{rkV2TK2i9l2SF4
zTe1(~0;*eDM^}{bqIRc|`J#Zjl+hE0@=Z0K1il8TVI5gZ8Kn(MRNgfz(MTQ(MLtp4
zD>ZQ%pN%fxl)EC<-)ZQDWhD(uRZZkFA0Mo`nk(5wF0q6Ixi(hlhiFyy6d(0PXPoR~
zzN2=a(L@R4Q4r4)i60Zqr^ZQdaLUdB?bYq%zfP$07rAf{mpHF%icEksv1<CD2KE*8
z>zhM~Mwn~t^@eJQJ59}68!^i#bL5Pso!BA58rD6526sgFZ8Ex{&C&DT(kxOaPm?c7
zM3cYRL7LdOH}s?(bw!0NO5%FQr^5O%H1NW++dRS*_nMvUqb{0ads9N>6;es{?(LEg
z+ZJPnr?i<DlW*s;p#=oH78*+|@J1hRx^QiO&<}0Rs1OnF(HVB%$sdlV)_Q)CW3A_t
z@aU!QFEsA@$oammQ+)hNm1@oMS(YcuX+t=U(<`wen^x?Ff8BUG1_tb=LzpYH;dkq)
zBO&`*O-$J@7R$7cyxADb^|Tn5-!GLlwI;t&k%&iq?JYN!%BIRzfnx)?d4rkaY=yRz
z9*TQ)q_)0ZTw>}etEzw5257~kK=|vG{<0iIHl8GiZ!P7xUk#eCu6(=5@5O8Ev>s1u
z==(qE5!Iq*FD7sA?`>d)k!ZzRj<UhOqrv8f1py_5ivOTvW68Pf|3b&o${)46sbA;_
zfMfsvvZMKb*l|(tKkOLYuW<7}?D+qMjsgFLj)~rIP(e6UYhNQJr<uo$)qBmqnij6l
zffc9L?qF++xD3S3Y9$CV@dcSY#9xwDELfunt3cH_e2Qa0%hiz95kq`Y&M&+&UgAcb
zUY@jA6zW)9^3$|^mbzVJ3tHSXkqMQYyx{H@L*m7`Ya~?GZZ!EO<j>D{1WDYpFH9;~
zs6W#3BO-S3T1J*R^_|HfoGsv{1E+%WhO395afilj=4HMp3AW`N-~iSRnB5SAG~m{H
zfzZXYetAR`{MKqv#+O`gmncSF9APM;-G0}b;3eK?BdK<O>q6uFn2g)a!+J)z;lu*h
zE9s+Zr|LI82FA^8@vCB126XQHTUSnr@iRO3p8C9vP%7?sn-eGb`U{f9k=dYqCJWPb
zmi)Wy_t}=X6<#jQi9;EIWm4fJW5<eMpp7Ti7~>|kt^ZeoLhJQ7uHf6Us5x|ToJjd3
z?CjtCGwczCi7wq8m@Tk;8udXZJw7;CLi)}BVcDgvn0@Sxco$dvqM#sIt-#@2&~O16
z8O;%Ym9QE!1;2#(rLl7EC9wZAZw?K<(RK(BH&iaxHbeM(Ol{i*-nlLVb>3MUKM*hw
z1r(cd`^4D3c|{L5c99_`J2mapGBMuullz`UrjLT3HpC&cEa3k=|Mkeh3Ce9H08tTU
z0v;PVuSSr%n(14m)EEHb)%aaGSBBDC1yO(u`&;?|u~2+>Jd`lb#+ID$K}^c}9j7H*
z(+EybtSM_c;jSXmUwqjUAQg8~Z>Lc83}sYdyBC+h9QHDzVGiO3+uWKO340$`tH(53
z|Aq48R%$OZk3LUrDh-699acuW9ODyQ5QLm<!EJ-_wfvNS+0zC&WJD)c7B;r&PKz%3
zCptz32F%yZ2yNsmDG&RW)$VbSn_1~}_*ch47`~3c>Qf@-;bZ_pH&(Si0;@dYXCAhs
zK@qLIc*O}^)%TD2o4e0O{#HKL@<o2o&JK|Fd5}$nnt%S0>dBz}Zge@>o^}En3{-Z4
z#>2^}js<012@LxOSEUcZzS(x=M4TNqZT}^St+12+?&@AGmsQI*ZGBAnf-oFPC>QNN
zy^<RQ>H!|107nG#J(6|jz`BaoX)%*Y%YaT+adcVqr+_e723=-h#fhr^(T&u6xp>0=
zNsB6^SzokBCiA~&F>YRlOXPoOQL|&q7_(D0hLK5Dii%@D3wONwlT@q%UnW>b^yH1!
zL%C{^_5(&nkflq3MHon?>GI9G!H88CPpmT&FZ=CHK_58n{CZ%i?vU_RqVuh+(d>_f
zxaj$=>$oLMnj|+e#jVmjp#wfwlGp-ni9wG9B1PMO24zXMva9Bh1%e#sK1Rkbs{GK~
zwS%?IW|a9i3O#RkIfVeByyU2fX+1SWnj1k(xlBowQ+M_H4zRZ$YSa=@fs><Rl0#z-
zBu?p8M`(px+A1b*7>@a~72L|c<&Z`YNy`jkZ0&PhSgKB%t%3M2HHvJcMIoq$Qfm7X
z;Ngz9U5n>M{aYfW@Ic9Ag)~A9qhQgkdzM8)V!h_S6q7PN^q!57OYWb6W1-R_;RA7;
z7u~9dKajcs1>(w*j1rL)NXF|8Ye&D5gVOP*hND{LdJhemgV+$a&@lU)Qear-6{hGm
zA7J|S@!VYtV$Vzv?afJdpDpJGtV?h>Zyd>IV%k#1f5fL^w}r2tu)ODEOsdF?<v_>(
z<7l}+<B>Ax-^j5oHO6_4WX=9S`$gHNoIIq3zC}$X2{7xnPn8-WIC1yOwA6<npBHa5
zE2FVe3Q23&MS7i_E7cjrfMsb(>h~ClBl%hd>e&@quQ+C<PL{Zj%rMYIKs7SpDuiM4
zJznYb!d6n2jZ(klljzT;-Il@<#)jD9EhJ;;%qbg;l(4|o-~UM#Dc68F$(>D06YbQZ
zgFk$*1;(G-N~WDi!|+Q)yuFe@jG|~_Phlx=m%Si^kX3h~Z;7YtqZJn4$)_Q;|9}yW
z!;@x(-o;3c3+iLLQ^$thxl=P)eBIK7H0fzr)I;s~Eo>Sd`LJt8)$2ZXou`j~GQm2Z
zjFf()gkL?s@HffD{#Wo;WV!_w@~_(;|2Y6=NvHsQBT0Hq8uh1fdaN(#@@2e1+9y!U
zt|wK`XxM*oPJuCkbxW9^K@+@Hq<n4-V$hc|<D?^o&~kv9nG9h^HNi*0H;6&;*b~Oo
zSY4rs2>u<Z8)w40(u2cRc71f_)V5qcyl$I{=Hbt$KbTy-j3+=z3l~MT)=MUJ$p)l_
zN#KuHV}_;&3<dq&T$CN&%$d@*;B7%f<21yXGf`e)<dD8#r>WN0o5%`c`L!>;s4!TK
zefaoNt^rw&X|0vuwC}F|FJI@Z5bECexzI$8^5s(F4mLpwuk45G185ZcK)``}0g_=a
z{2%#!iUO_04w3530$NO*s=X^Ym3BaqXq4F^8`;-r)}mZPI@emqm%YW|7x&$i=s_k%
z8Y7y4bV(1oOo~QAOzYm+-7@WY!1v16+^va@wN`a>T|~@P_j<F>#EQ!2WB~%zRmpE2
znQF^_xYum#Q0GUC%55+ika{mmGvCe$pUFRH6Niz(PaD1A?OqQ=pN{UaEZqS~yUy<R
zt^uqXsMNggoiB2By_ph_cNks?1Z|K0mA>gozqVif+^)PwI*ar1z}#%KQ5(v_$vN8Q
z&VTXj;{A_g`v!600WP~kEtSp&C3=t({qV;}?~M}nS*9^cnGhgmoIF%|y@MrYB7Avt
zySu^M9CzEjf_kUb-ys7ed#Ijjp`4IiKKEL9d%G3HIMH4)l(~}!$n(cHvZIp_5?9YG
z<emC8gm4y7^LI<g@LMbcT8!+KDYO&Dqs_AMf~s(X@-H$VsL}%I16%-~W85%sdXOFN
zO;I=IMzSH@UB_OYT~v^3PP~Ru5$A7QUQ;aC8CGjQhZN_f(-Z-J7Ru9AHp~wGy{N>j
z&AY?=<8~ewn&6m4j!hBgbKGv*wQ-AmgC3?EWN-WJ?%P#W1s?d>7AT>VHCo8|84`Yk
zegwEZC{NkqW|cIu1(nksFFVnnwHh>IgEs$hdj4NVPw(qQx9V>J0miqgMQ`o9MIC?y
z`+6kP%})O@>g*_bqu5l#8m)G1!O#9i@!kHZ3*-9$r<rf8K~I$Z-!{iL${pPC%zb^P
z_<f+`x_Z0Uo3P7w4exq5I+N@2$UC1z%>)yvkSD0xVVY9Es-l4zI*=^M>7PG|pI-1l
zaaiCW<S9xNXD2qJI6euZ$TjF_Ee62ap%PdwJuq!&_T-bcXI<(pyuJ1*4J!L%mTe5x
z``is{o1VyjPgTq`0-pu-(5VTdtLorl*>@q2D~E51^QX3h<iYlx1@=y9#LLIBh8*5j
zdCW!|+?d0F<l%aoz1wh^a=9jMQv=tG81sAdDfCm1KJ>`|C+|mYl&oC53=;4H2J;U8
z7qZRtzVeYLx!tw?@S8oX6YP7g?bNPRb{W}?32meK!`Wg&9p}|wGc`)N^XXE)V?E~9
zs~<)Yth<iSR-7W6Dvj#8(PnLwDHB)J@4g}pE?j^LYM5NDb#szEKcm`--N$N`8sDQ<
zq~d%rT^UxGW_ld{yIIM|94BD2Q}5m^Y{!{%5MkiKkuY;X&?euycod&f4xM$V|HP6)
zOG?O?$Ek#H22HH*qky?%Op!LC)fF8L%ID7aIR4dq4K7Sb)vLa_i28e*s34EzB?|}5
z80z|A?~-Q)6uSRvDsunu0WY}(du8o#a2)SS&!RM|t=_Pa=at^bWg1|V_=NcIBimq(
z6iUZEIqo3&*8XTmQ6tml=x|7#>l{t*)a(pQ&WYUYJa1J}|4HrPV~SzQ@yePWrTT#F
zSv@mpa1@E7@uvq9Yvon3DLRnuQfs(lrtI5>W^@$YOh!8y?U5FJ!56~#n1|>~go%o)
zK&z^k>wQhElNlx_k^}M!bi=s5;ZbKuq(Q{PGG5p6Q`Y-U;D_nX62#8_jB`Zo%ro3m
zkvh=iYWWdU)4fry&4%zb^`5pix~b$$^ClCt(Yl@?;AA8VAw30PlUtZov}bdakpC1%
z7o<6IpMp)JDPaxfe0_s^4Px@r$yp7+vJhcbb~QoY*7KRE!2=$8t~q7+OZrhdiF%c~
zJBW}H#Ch=KnYBcCDAkW9Xda#9t2E!aO+WDZX|9Ti74}|<J$!C}Cr!~d=x(X49_iG(
z%MMUCf~JVsWO81;-be9yy*#|v*J19|hYK0cf4;L_9ql-qyoVbdPi}WD=an5D)-M}3
znNOx-97^XfAHa6QD2FC^$@M$)0N2`v2T8cljKhNf5yw!M4wkj4N_Fl!SXgP|pn#Gj
z^f-_z2()dFjP0EhrPyY7fF?5e87^<n45QygM3Tmhn>`QYl>19*1e@uV`#Z{sidX{S
z+5}u!4oOeHh;L(bNG`N~Dq?12)AH55EOh6*jMh9N0`^Q3S!wkhrv!B~FS21`$>&Ce
zBTCMTk;To5%U9;c(H%&>@|bjot#GUNkduU)n~8i+si~TQ6K-sckw4lPA4f$3y_&o6
zuO^ZC5N=7Vp)5{Ms>xM;<|2EYci5s^Z*PYfOUC>*L7YbZVU!j*AZ2m+$PM(L1>o&J
zRnq>#2HpnH9KC&+3+mDMnlxe+X5z_te%R{lvuj+v*9nW~MQEKXof8_!Nx%;a{R!UC
zcWKEtLTrT}a#~hW9U>G+b=X+o_s*tyZDr>KOMGT7R%s*sYg`S8rmMi?{fWw#j2Xx5
z_QfC~T*aV2fINanT}N^3A)$=hzr`cY@Wt&G1)8)|Gm|jev)~hsz|cQgH04|S;Y?OI
zC(c8fPE9~P*1?BcE6~lx)orest(Aa=+<Rx;rK<aTSfe5ouDZF(U!F+!^};_zO56Wn
zzmndQS;u|rTf`RCqa0dDmd+lG3e2j`_qzGvkP&UgF!bp0x9S*b-IiFd5iEWt<(N?1
z4fqc$KfT-BymHZOnB{TEkF6TWxbrduNJgwY%tZgNIek_4)+CnYaQsB0Mh@i?b>IMz
z@~9EdoyEQPPN((Sgbkbcw7m#y6kqA?Kr$d@Yi)<jt+lhSg-XS~?oedK2J^T%U%D~#
zi<}@Y3#O4i|A`v+U~PD$A@uuCOf$;_;8pRJE$P<PbxYJ8@~|YLo5;Bh+JX@QGGZYc
z$?MS}W{@&y=9=Dg{oME8h^#MpU{d~t$TAVG{33b$RwzB{!^S`MJP{(GN@g39s4ViY
zFR@@;A3EG7R@TS+Bv0H|NiG>L-(<7ziH#G|MzE|(uq|A;LJwGXe6);Vft{D1;fmkx
zEd_(yeW<O~CXBjd%YV(((+`f$B1BQAGYOx-aHy!~j<Zt4nQw+0XLSnML*8nJ*QIoT
z5N#S7hMKew@se4qh`C944@Q1<tx-nw2RYNUu{&D+nHLky2qW%ugw3SJCvH+uDf{|F
z5*?>C%}A3--aJE1+DCRC0rt&5EzYAITd|#3HZ>8L@>0Rz4ePOWS_4N)0=abz9Ki#C
zjP&6`s8UEjCL66@WnYi6j$C$MF)nD>XX>8E=(h#p@{z_ltK|c`s;<j~fIY?G{hr-X
z$|NG?844L^l8&)t>w76p)cXHegQz>EZp3bU4v8HLGBeHkgmldS*!n@Q7AiKIO_t-|
zztrUCGAGNGDthYffASL3t12(^Yw`EIB!ypW_OR3_vh~#*+W<ymOB_#s@_O#l()IHf
z`sY!4v7SlMQCq&2O9YlC>kUmLsluj+60OzVWA=S%0vg-~EQN87cFJLLh!|e$zd;g#
zT|V0!KmJ4|RE0$W*Mlqyghx2JiLh(LuDYqzuN$nrO1<gqd%>9v;4VQq1rN!C^)i(X
z6pkp@Rg|{`x&}riU-@N?bk4V8-=u_QzD<6kb6l<cDirg>#}i#hN*71E<zCCgG0!3J
zc49R%+xfNy0i)=|#X$X&3sNH0>qt&UW$|FkQ_`k+b;AY`z+4VdLM_tb)Ws=<0BqMp
z%VP6kM+}pwR{u+}K(W8pwXh&dn{l0)xr!HQm}bN%%YKYPm45p@9iYZ$hET3kjW5ii
znoIKbN2RIfLuhf|Hp#yE*4C~sJ4y2}LoL#}kxO};D}k$99^oEO;q>|THdM<h%huE_
zZlZY4E)uw`!1`NC)w^IAv}6`x$x~CU=y>+XbILh;gC_kMlW9u7@&30=>+A&2=^)#J
z&MYm_EL>y{JJ>Z=#7c`|K8nm|-Ki((8EeV+Gesb0r$tNV*AD@bt*-Q=WY{lDkv8jh
zf}l+LTWh)|&e|FBP)M(@218ysFGOoPjZS4;NC^;0e#x&fCB%gp`ZPObQI7O2@}MS>
zY)>KP;tpy;<DyMGr&Qr)f4jHw4m?28n{iV<R2+b10M}?<C3`~Mmkhn#y$NayG%@Z~
z7z80AC~>7v%EdNIM=yzp`1*~)7sysIAR(p9vqZ+;mr3C{f2@~cA*t0^+^K;Qa9Ds&
z6$MTwN-G&p#(E4~4|N^fz}r{2T3F*kSQ)tvFTi<eCMq|4iLprDwjcZHmh7sB9-$*`
zFf!B+0w$2eTkJKYm#8GVZ#c}mw7Ef<M(mbucj02%b~YFP_5=L^M2;fZfyxtl3-mw$
z6qL3&^{?UG4r-sdZ(c@e3*WakQ!ZL?CIKwJd~5q&m2LXC%=0trm@JEI+QLj-zf``J
z3PFJ48q-Gp%Hcct`T9+SbNB{v)~)ob-G}zlze!}JVvCcX&}DGJFfSFMVdgjt8g1^B
z?RQ!^7rU(EMN!?lLIenWuA7MIJmrV*9r^;3J!l@j4v?O6`>cbouG;>Hy$a@tKLJr^
zxNqoS=+o>E1T;|h0W04sAcrf08N2f%nP!ySxR4qk?}Pu`-8PJIJ_ozlHr&amHLavh
zVCZ<Hdo#pW(Rc%ljIjZY5KN2oFz&MnYk3r~4+m%ltWFGB%zXD@)8*nWWnK#2eQ8kH
zm+Wss(@nf^XHi|#ppP=eBpNtqCV;gdnJE&WDo1dep4>9k!jBf{oex8=K5)vR@ZvAB
zcb+2t_t2Nqoqbgf02xMLz)DnhxUGUk_3c%5=AJ?#i7=GoX7$bc9<%UpgE%@Pav?8M
zMtK(Po-w7Cn=>~OOAgJ%vLYrR<{hJ48urIee+A+lLanlBh9i3KRje|D2N0&EE{uOZ
zQf^e)Z*Zgk-Qv8O)vnXF^H{@<y}8*|Zk76^sq0}gf62O1P3BhY<+8I?TB+;c!_oSW
zuk~Z(GW#9p!EUztSva@I(A2@zO{<Z&W<A}A2@Rh&MsP1an0$(GC(4Yc^TrO;mHEO|
zT)F5sPIlhn-6|NEm)FkP8c?vo(8T!XBCc&S=>2RFa*C<;TLb3wj-Yi<4B-s66ooLq
zF>y!~sDv%Lt$w{x=JIi=4n~nqEQx&0*>^QrpGD%i-C>i29l_r@#NpCVK>Cs!7TJ`8
z;?MmXMkFJ?b=wZ>+0O*|TN^xNBz)^$a&DA(P<K|XfHWcjoTzr5V*pwwaoC#=<~iY{
zpH*77i(Bl;PU6@ziX`XgRE>Uw?*+6iDw1j^k$Xu1)>kM-yL{`-FKdFCe|@TO>+29D
zje4Vf_r?!R9hji)G;!UderT)tWO0Lg5ntX%te<n7|H(rYC;a;5cU<Hv+--PJ?Ur()
zJo`_en}mk{JW|gC&g>HZbngGHu9Mi)JE$*TFh=-SjHW`Ui*TpzBV{3u4gGJQdGqL`
zF${kI<>wQHE{eG2&gA{j9~15NaB^YClHNr5{OHj1eTVOC@p*7EV2VLmUUvuxSKr+Z
zNEowoSg<u4mT-n+g&w8bAqlhzxsRc}q;p^QX}Pth_05jW03~wAhT9Aa%0&h`B+nS<
z*Aynk*l>>nG2MtcDMDYP1XOp33}Zsb`~9pkZVx7#1g?AmuE&IhfE9+sAvatk%r}W@
zMi|$S!R<Zpl1iGc^~^$ZeDrQQZ%<mDJO5HMQLCl$A2B3HaYT4`k^|3z>XX;3!V&~6
z&KOuwx%Qji0gTk}X%NW^Ch6o7Vx{q~0wIs{!t`_5_+#5&);*T9cr)2nnM&k7J*ljh
zET7fYY0V?A2PRzGGZx3)TR4Rh#Ris(<BZ`Y*C^qukVT62@5dc1)`oegC&okgfmEvL
z0+of2e}8q8uHRYg1$Z(KWKVv3$z9(uq!e${v*m%e0c2RUw?!!{18U{aU~j}MTAnY*
zL&KU0xYN&fc3sSRe8@;MnfwU;$ZHHV?-4K&hq{c#5_lk|{yj5F;ihx8G<S*k<hUE=
z4EWkpT&A?sw;km8vGr_(JA9DWSE#5|Z!@NXvBMD%(v2_TgCNiSS-{^H1Dkw(plz(`
zAzYT<0<b|;l$@c^F}xx5<&a(TM!Ky;x}^oq?qr28Ig1ks;2mM8XdBgTmp`?yX7?NR
z)?4jQBpX!k*l>>0tgP|ON492to?xhw?f!H6@em@&uJh;=O)cQPVge@Fe6bk`YY%(E
z=QjX5&1aG+CR|nM7JH&!ymaHs!zp1X=f`2e07y=AH#uvniZPe9aYaFWXCz3iv@fW=
zeV{(BL6@k}!z;YT<OFSZjiJ0>Yc@YNH(x%%QNXASqJzu7b}%;{k|3HYWpjfnUDwt|
zIWt!kUz*G1Y}X>pGnqo6u!uq;?D+SY54avG3k+$17<ItJW|Wh8*^x#*{hRH$Uo+A=
z0k--S7XWW%$uzTfQ2F4<<lUc-Lv46Tw=Nz}+&xM+Aovc=RBj$)?9r*_Wl;?F(<`|m
zf;G{9lvT}r)E90Hd$0XHkV<1wqTw8yv`7*oTK}#$Y`IPU{w&>0g2-3D^1fIOx*DmN
zjr%TBZMAPK8i%Kzzy7Vm=&9#<MGEr>6ELF_Kc7-VHm+~yNLO_=!A)VQmnAdyG?Sw|
zgYjGlr&Rr>w@vDF{S0j)v`vTeCtVbhu+lUp#KL!a)a$@wt<e1aS3sC+ZEm%<u$dVT
z&U0e$Z(?`v7n>+l_VG>5QIk%ENg7{qaEMxn4lXzD>4%Gi%f`bQ^Ab}-5o5R;Jb-=8
zLhRsn99b;XxzY{3y9&=;?;zHK*WL2?4C6scBqP7v7+IJ#wXQOKntrVo3B78|HH2sn
zElS#Dr{Le18y(v3*ELi_NpZBL_Svi$tPz6v2VT-BMdcPvCRMl*^B=bvQJNAoY^c?^
zT$*8oQCF6CR&C0%_ONlb5gJ&19e^|)JjP(;MI5F>L%;N?giwsl((39izn<P!=9p;n
zP;&Ja+uuW1NizTs;i|Tli?x^Mvoru0Q~8I8iXN5)t`HBdraN%rj1lJn<xpo3=2eav
z!?~lk4ziI6%lXk;6l;I?+6(*Tq`ZPkY33#dbN78yM`H0`-SAX*aXWS`0dP3ol`qZn
zK>o~C@>+glRv$4o*^GvLc_>IosrTqKt~^VrJ|PQpb8;-x&hE_j`i6HBa{n9E^O+2S
z8?B7WYdZWq>cDmQzs-^fyZlRR|Gy>uw=hL1O37L5w3M?mdc|@LMFKnczL>&BZK+d@
zMzmyj9|RMe@fTi{R&W3VyuBsUePpY|AKWoRvT9Xrx|Q>2K|VD`GW*iJifNo>^;d5=
zdDkGftz&q@4EHtN6UUFiNAyTSbR>8uj=_=%v0s&aoI!e6j`~UR`h!`OBU}7XLU(TW
zqes+h%oCNoX*jf`{-L5|C)&TT_rS5o-JS+OTT0d>JZzuH@h<>_Qk7Ge_g;g4A%2$z
zaA28{E_b9K|Chc%$Vb2qI#W=q_azq?=$&?LYHp91vOEYZd+Rxxia0)ALvM9dX7(|h
zAlnNDUVUW0U%sAQUFm#sz90FFHQ}yURatZBMgMp}kleXfw)P%V5sjns<@c?qWcp$+
zAe2H7<mGz)m}%v$)wUa7*>1_xw4kb?TP4#A%b`ONk+7fgs)$7{V}LJ(H`@}WR}-?Q
zJ2UXPA;<ACUK+iP=NWPCG6j!RbnhTrl4&hJ;P()yZ5-aF-$eeyq3Xf%`R0%8x;{@2
z4O<wqada+?O|8r1kKwHD9crGDx^X5yAP^#@@~uvLy0cxAJ=%to?r6aq4i!}IyvOjo
zk!5!ju3OXArQNh<H#Tp#5pr^xmH49Ix1M7*n4Wk7VBmy&*BGhuBG&69Yr{-$Al5vn
zidwV8igOV_X*OBByWtNlPm;sz7T0}0fJEjEBGBdxczAHM%7Mz>l&#%rlV+9zt1G|I
zdYRzA_cspjrf%UHoF#P_#kzi5@!oRHY>w@9cAY_40kpo6QK&DFS-I7#T0fpVcukn>
z=)3u@=8-Bfxo4^LXZ=ZOkeApQ5&APES8qcd1d*u)ekD>b^Ck1XlpDq7$bNawcQ>vj
zC;Rj6`;$X()q^g{bVfTg_+wB&F(#|RM^rwQ{?2%C|4Qwj0q~9hb%-`^5)neECp;rC
z9>N~bg45DHzMCI?c02xkKm1|;rBzbu?I9-^#1M~Fsc2;?U?zcGf*R{B!Ll)Iw7@@F
z{l9dIo&T*<>@Ayt@rfGZE05uli!*Z!lMez1r{jn-;WTN@Ay<8YeU_XRO`dvdqAW`2
zDEQVb{ti=cVuD`0UW;V*UmM}u_l<q$J4N(biq;tk^<9ho%M0oLn6J%vagGDH{&*vJ
zam5+j0Blm|Kqg5mE(E0mg(#_yY#pYdXywU;h<i7`X*LZ?pu&8^msqhOwvFTN;0nMC
z8g6YHATri#ZTlqjp^OF^)`oC_>mBe58%s-<6JmDB^u(Lk<;&2S*?cV>n3Di2z{&>R
zzddW7tczpZrGGQqEe=#M<fos>ooX{*jc;{>8#PVWdnB9Nu5DIz0*%MNo^GYZz|PRn
z$Y8%WxBPpXl9KiV0Uef#Q&~Q(0RyCkBnGkK<Kt1|7vVw$Qzpp5YO#osrodV<)eJ00
z9~)RjBZ=03Vv+6XB5Vxi_F^p~47O$jV*dG03YYoTpkq(0&7B;5)tmgVt>=yh$zWct
zSBozsrn_~Lf>@}6&*ANH4lC12rA*go_T&&}UvQ7SL$7;XOMU6Lp=o6hG{E|00A1l-
zrz%#YP%o*n8gqv&$m)zzD(O}>AtS+S_d_^^CO_>KdxXa$e;aN{(_eTZGO_nGQ&*;!
z^iWDD5@{ly<BTFnntA~=Ih-O4&K7N*{P!uy`rx<l2Y6<2;gc1GVigi@3g>2_lSBi1
zxY=$Y)yx`vPy12N0<2QeR^a&{t>p8w!Y4mO1QBO7EVo<EKp4?BqV=L7uxz^49kef(
zGG%RHj61Z`X?J)3UiS0e0g@0&)q$hKY2?i!Z5f<v`W&O!H3x#_-Q=$?RmA}H_k!%m
z<Rd*NbYpODIqM`<Jd?S9!WsI^CGFETB8*pLyNUIg2aWw4AjKhzSAdiwwjXtiT$=51
zx06}-MAzEa-np62#qMf5nivUVo$wgiB8F?Zi;XCx(os4uN=Z|kBoAM6tbwz$XMRMo
z=}K#`b!wkj=2vJa%ttvr32bi0LeqvK=pUExIV#LkU_^BOH_T_BVCz4PB9#jDWh#jh
z`BzzQ%Za7}FPkA6QJ{Y!C2B57G6UFqMZI6+MrBj~`8F5bna5DIf$H|$0heprC5k_!
zhnH<O$aXBzfe<1{YeeAfm#$G2s0h7pL@=4$`iuKShdyZNAS+-@n;!U1drr|Ec4(h8
zpfGjkwhZZZPfxMj<9I$LT?Odxt!$`N4;ylb2^khlQowP{1BCnBR5QA|_M9l43GEpS
z{?RQSxC%&TQ-tlYGOXo>7!QP?i351y#;&@U5`RQjwdq|2n!Cm1&ON^ytNPf>AdGqr
zGb_=X(0#J9)%6^r62!+)ao*c*;vNSDnu9B-uuWa)j1q$H)ysqKosauW@CBLYyh^+6
z^y)OxtjxU70n#v<%Nc43xbaEk4hpcn`{)g)AZiIyzZs_q%h&2>YFOrG)=SO^B|R|B
z{Dj}5o!VR=?(sm0YB-Kh(^YEILp~1D6jd{AiaM*lazo1Ch>VK8%i~Thf9NPGZ9lxf
zXfrNOf8}DV9*Vf#cV6J2ey*YanVAVNN4Ht40q>yP1n>@|D<wj)WXXccNR0WGRgpI`
zUd6Utx&9#fNpp}pwHaq|xqMUUu>tk$r^vSZPl8P3QB+Ps=!^avAC+VkO~cj%MQpf`
zjnoD4lEf=@=|6EqgRi7>wqGz#5*g87bwDr+hn)JAvvelgb;R%`I5<aD@fbd7j9)G8
z+C<RH0O4FUR_^y7e5>i}7jM5oy|?gi_sj!4XUXR15KDW+QbFM@k>2d^BRM^79D14U
z<Gc}8qp^kXx<s3#Dy1k6^|D$U{eqg__h1Q)tCS;u{(TY%Dxa@dmVGz<>u8P-HXs=H
zz#&RETRdmBQb>Yfts+1wCn>Zvob_GCc9|I-6X^3<nB9rX#OR-Dqv=5F&@O1Y*C{1f
z@{gX=qB#T87URf0RrJ3W^FO!Hk`#qkl*o`VRPL*>gd9K)N2<4RI5b1FZzPtd(iD<E
zG>oos+VhOB;2{iL2z5<tPdp03UW64(`~*n{clmG+G$<41Fx<^76Fu$f<K6v;7EpM7
z0PsqYa8V-<<_1{_3V0um(~#k+g3JBd<7uJ3W1KTRBl`9+bli5&eB}5XM>cHMaJ6PK
zAys$mF$PT#eN}ih3H>my|Ebqd<;U+iRI6qX_x+yJi(?p*qX#qH<dC2C_^#~OPr&TS
zfXb)lzWy3FlY!Y@#DF)4?slDUC%~VJ7?5*W*gGm4UfNStC8Lt#xF7Y%Y{h`Aga#Rw
z(uFWkJ~<^>Y&o_5D>h=Ev(LKU+i|J(M`o^h9bY^pvs!A#8kk`g)~kiwHP2VfIUTow
zVNrH#f_nS4Ho4f$+GcpLgVI&!UMy8RXc4|Gy=Xw!xPUG%1jgfXG-gu3mjkhy0IYi{
z=8op~K-R*PX9q3np{DKo3fWwFxgXf;Lh>bET!>ybHyO;ERqN0@TKVm;x!H3J-nFS!
z4bR6sbv=8~aUYp93NZc)`6OJtaO%wHqO$d}d_I-2^8Z(i@rHn=t+t<Fo!jVi<CN~J
zG<x^Vw9TUFxZh=GfK0fuOJP&k19;mZ*O9*NM9{9=_C~-y@J!Q~-*#cZ+PirS+$9e8
zY2yF90Y~x^{kvq#1%j&i8XMy!7Rm|fz!a+n)?zMz5nM0zyWSl*jYnbXaqjE>_y#_`
zAnX@6IJWycYr==Dsu73OFM$ljix@*IoQW{AhCAp);c9~JWvGo%uCJN)1JJ2--@XzX
zgTJghLV5yv)b=;?d2gPJtNGM6!^x)~Dh7~zRs>CiK`91&TO<9j#obO!r~j5zcfy{@
z(2r~nvF;+eeZF>Q6C>f2N*fswxH;m6t$O0DdzBgz)EIyh$FWO^>C`yNGz;(kK!pyl
zr<!gV>$)#N2l$aQj8WGrfvJbqro!$gx%&}p-S1ZfUn%S!)j&E;BM+MAnSogjp<I(j
zMT%8&!oLq|*tN%0VV>^{AHoCqhNUR{Zk~et!Fz*vS#Pdjc_Q_y`K#h3?z(Cz7(}&Y
z^b5)Iw0Ih}SJp=fymm_;+gUxIW#(4Mrpp6+3lQgnj?GnkSfTF8z!b@hoRFO>>Z5*+
z$anb#k@Ft==soHyD{hEo$Vi1G@t(@hRW+mDTed6~JMlSF?V1*FYlVS#yYwUC1HMzf
z0Nc`^*6!7AsJ^V0JN-;a=?{lNcNrX+C)UmPASn@DR?`Ms(+2)OPVE&A?jSiA<1M$8
zWJxuPoa)!X$dI|D07K_pr(*4^-?)dgC~R%SwMf=fH!=*0<$6_Fsccd{9h2V<E5!aW
z=ggHtffw*bywS5|z@Q`w_huoSd6*>$u5m40V}glNs_I_ryk@c#ql`_~%}FkH99yqM
zp9z&yF+&)-$6$0gSyz0wse2Cb5nkhR${HJ5ORtU@#omCV26T#^ph`G~k$kKXXZUlX
z9~%qE)o1e1az2~az<wlz2;>`=?mJS{y*zz?t%NsR3Tc$0{_*r&5njTp8mZv$PuNgj
z&TOEdR`A_+_})z(GGh<XG?<EE3~Sl#$#Ssh@cX_m*Z_#$UKOjG?^wfRDr%q4#<9nZ
z?&?M_wP!o$DPYl#)B{o*?<zbjAR;O5NS}(MN~2(JJK}ZN@EB{<MIwDXbabpCZso%$
z%W>$u->LVemzboZO}lcKviXWGC_neT2~<PNUXk5!*vTU}-B`X!ZUdhiatS{-J<z`d
zEjK$Uz2V@eb)WMO@VkGeqWsR~e)a1q>&5Rp#<k_K2S6E_J5S!A=gJJG9rN^LI_R~H
zRgBX0VdZ~n$5cm5Cfy0NjGt4HP`X&xx;zFP*BYfWXO4(+wVIvfCs_iplGa?rF3xY+
ziU)ia0SHDs%BaYXFgpmwtTxTn=bwYyvajDeTJlgLuk~$RZ1EeT*%ppHC5=#tF%|ps
z`OzYc0K0DQ6Hq4l1J2rAlXOO5V@rO*L(JNi*3AYaMi~mY*~DU3^?=6xEiHh<t|D0z
zmgVjQIxRppsXqHNMU|ozU3D{*H}Xp57BLO7UwQw?x|39JIb8J=<93iWw@L6f;=KGs
z$We;VqSt+3y2$HTO7}+O1j%BdL!6mtCgU<EFwT5R05AMcd20a52eZ-qMBp!vg%hNF
zmG`K-qdjOkgeYaz+qK`Hft`PTvvHiWY6#Pr*^-<jSi1a+PQ@D4gY%njLK)z<WN&{i
zCJWnkEk3xoXM*a4@SE}Fm%=Z1J`w2a;mORa;A3g_2vwh6Z!4Xi9GE6|-yko)MnJe|
zz`1;Lw|i|)nYWM#W_sAHG8+N3dRiVI)`ItHj)(=?!XkYkT>gXQD5GAgqfV$F?AFeo
z*po79)^Z}u!lrtZG|PtM>*IZ5T!|c>Xs1OI%lgod{KVNA@Hr6Ka(Wn5gD4sE2E$jX
zmHQo~mImFxxfA3ysC|6E_~~j(<rjkx&_R+r0d1vB*AMF$Nw}G8UcgJ~Y%Fnr@f2PW
zUEke!){<GQ(%7c(*~cgMxn0cs0>2b%`1tSEeL;Qw#H;YpFxuKQH29tbGDobXX+DY9
zHaMPsbGfj0ojIByW*;?#4mG6m-hIWh_2rjaEbo@Rq4N2yut$5BhkL7fPP^hKApdhI
zNPT~+!sng1KSC{K=KP#FXIXt+=j+)ISCo3W9$vX_>1egJMxCR_j8tNAfDy}TUdnvM
z+LV8c?Hx469Vu#CWhtFDIqPk=zvWl=FOssEDEBvm_J)vUZG>tU8}cb7{An?7Wzmn%
z(yV-RSQSK}LKT0Pipql0y!Y)VfN5h`!sCz9T9Mz)iH#Du#DCRNyxSmZxk!E}1Wg_J
z!?_AwCXL_+Rlyim6Mse-nMJAoBc6rk9Y{*Gt(t(G!$4Ra9F8x{YPmGHn^AcAEJcDJ
zm3o3s{~)UB%EC!#_uB<!mdPGDG?irLNl^9phq~twj>jX&cCbmutfuiMK$WyO)y&X`
zA_VP_9h^Ejvq{ovvPS{Wr0qUNbCUp+U-&>&I3xuZ(ie;gwH*@$W}Y0o+BZ9l!BE31
zy1(dH@&D}J?b?#vyjrbY3$32w+o&6l5kM-!U_0geee_$H(ri9(x~K{*0^5jm%Zz`}
zt}0C*U9cmk>TpAbx6ud|KtKNUohFf2M^nvhgZK6ku`Paa_wjLd`>>M=3y*L<rK&TD
zR^Uu<7??i_QJxI>RTjOO5MD6UYI?b%RVPaHckx%7Hjlk#o2ZaqweqwnA@H0>>cV&*
zuL4p;+(PEY->3^gZcrb>IzPV5vJ=qJ|4VqqhKg@EhL^TXYo+l3a1Ds|&9@J!>GYeI
zs3&)$#aFgkfUQVmKIO2xPS(3h{hDEX_ciP9s1v756TAxAZt&7}gs9(;_xQr7N6mKp
z2dKluKLwWZu$~c!sw6=$q>Z9Q{IdD6gGG_C4=ND31=GZ-hG%h1OJhnlu&}`4iWs_o
zR_og}7oI5*LI$${w)kQ!58Ao$Md9SJ!)CccRYNVwvN`CQMmjj{OoS*3T^coN;LxQx
zoE#)W$f>DP?T6xGm7Jcb=ox2D)ylTDnSp_B3iR$?)SN?XPlQrANH~ge?Xzy3eo<J<
zFalaO(k4{$o1Tbj?OXQxF3wmPSSZNhp}DYNw@vS7f`Ahs=FVcDoJvg+(j>YcSwj%a
zeeM)ycQv<?`SzCCVXZU~nv1b0KECzczJ2FvcQ+B6Y62W*DOS8z0Z2Ho%l9lxt2b6?
z_(XX5uRQGs=o)2N{!F!U3?v`=R7n&=CL#O`NIS0b&iS}=Q_~R4@4p8aCHh-VRPko1
zZh|?$S%8|O-MrTSs_}IXfAcX(G&lN55d9U-LsEb0a~zBk=zF#GilMbzt;~x`!26gy
zlIu0zsaH$;2%ZccW*hJfh0y=2<xAg0{>%9{sbG}{PRNJ>aPNoxT195DroLsdNQ!i7
zTKeM^;*eSH{+>`&q89JHebsll%S!Q$?Th^YDyBu<{AXHb3lGPHj**QGTgf<yv3NgB
zf)q0AMOo@Zr4n%3L{S5Re0h<~AWj02%rw$?U(gb2n2+^xHQHr&_x}y{X#|w^0&|fI
zBzoRXa;iHmB5Gwn$Cs2wniq=CRYX#Yc5dU?83KShWM0idMH|Z6ygbe!BjDGXtJ<+<
zr_!Dx`ozVlNtUR?PfI(s#{UiaKK}#y0w99fD|9mF{Gh+?&Zrun_>>7GUQii+RIy(x
zOO7XBe(l2L|31{26L>#!kwos-uN<KIei4TVvZ67?CGeG+ZJi+~A+_!QWFI}%9D<cp
zjio85&?hDhgvn4jlzom`gouU=jvvHlcEV15(<nz6uT!Vvh)^a#BIG&sahCar0&9jR
zyo9VeKZ5tSlD{hQdYtu9yBj4Q0P``E(Go|#C~8o{ZW)hy>`yZvI+(>0I~mXx`Fq9a
zYQSrMcl<fh{1xoYw0!U@3C?NPM?9x-dZ6DL-WqNdFfrHh|BM((WZKaAidBu!6VyJv
z90i$zT8In2fV;b<QFkL-9y{+f5U&dbdm@dj?6I#1?dEXwf|PiakY;rHeT<i(_B(yj
zsa~r>;xJIMzNu?E#kL-%GyIOW<y%LXT>YPcd^0j#JZs)&;>^o1Onx{IVW@3d7!NFM
z`?EYjppW-MFq#nxx;;FQ0eQLrshol=mJ7!2Y8jjagXMkt9Y^C9ZU32UX`UmPuaq{T
ztHOmlc|zQ2pt)1K4rR#FrIsi^D`Y=M){y9$NJ&)~(dFq9>EnCxTXaSs^r6lv=Io4&
zrZ5-s$<Zg6GKaASp10|Ln&j)XquF&BVvclyMKAmbYW#08AU(3IAg#1-P>lke@HP%%
z(b1L^C~PP%AlGqTN=#ycLPY}#vVtHFeOng}VM%_i9K%_?XYhX`mgudOvYw{#3jQ_R
zL*_BNu@Dg_beSl7cs5vd!lnDrDW2PwYwXwFvG?{<-#3ev8pb+iB|@iZC+fu^sAy6F
zd^Vj^0<-uRD1PV}a%U}ewS*&5_~%97Kgk7yvWJ>UN0w8=EcLG??aqo5(7+pVSnR$z
z6MjYVnEe3YB{;WySz)1Ppt6e0GS-?bl@NkPI3a^15&kk1W8`?8QEmBPxh|FHeLv>^
z$Je{@-!cc_^nc16!2hLljIU3vn*W5t9^9c$m6c@fLA}}2an8x1c|_s$a+)kHm*-S5
z(4dW$Q7bO0SLs$`|72&brlO_*KHurFtjy`u4=<d4Qfv*IEA)ol%;cZ|zL<E~KPG+=
ztaA9j6D#U)@OEl7XcUG-U1})`*jS&TyRHCRT?im@6(sQ!1pDL@g>miG!<2go3N70+
z9E1sBx&P5QX6B`0v{X$xQZuE<!Rkrs1J&?O3$y45INa;cEM`_oMsX}T=Q#q2<x)3X
zszqw!Je<quE3cU+wHK6Lnf2qyY#MX4t1y}}wK;&^Uv+FvYfqNo&+T07{WB1cc>qz<
zicRPznm!agdEG^33Y89KV;|^1r|=!i4=y7RhDhY47!5{^N*PgNe?{`BdWq=3wVgLT
zn>IBDN5YaAR@}?mP-I#VP_h=;aQV;dWI7ME3Uy4UDTD_%@<_FY#eAc#P)?Oc`aCdV
zQQ>}N!i#l!YN_DhEaDWWwrVU;H^5WB`i3s+d$<C@fjDZt%jzak@%R}7ljz@APl{Gx
zxCiE5D%jgldi;ab5WDYwp^$UnhpRuNx9-mvVy2>wjf(ZuD@J;!$8OT(r27H<ksOff
zuxR^4uVZ<tPj^E05mIqXz4UYQ8FnD5+_cB7?r@f>>C?1i(igP?hmlPY87R>yl<3gN
z|Fv;=zfQu60T)i|0W=RTV|I_Co7zts0*h>>!4BgNtuzVUrwcubmp{;B1xCk@PztTM
zvrbA>_0GfEdWsO)F&eQG#{$G)dPK<Kp28GJvLt<2NhF8@#2!%Lh<)1l>Y2!cPyMM#
z6{AEz3zdcX3ij(x+?I$k0U3l)txguZp@gJq*(n9nxf+}!w+VJ=k{#(CR2x1--2rQY
z<nCZY>WPyH*hA__^CylC4-mEy0}k_AJM&s0?}tn(rx!51v(evnWvKCdDy2(R1E2{G
zfe#$ejHx;ai)s-B*mnm3oxJo$FL2&*yNXcd_7C!xonZEUaIcF$03kI=$6QlPEKo*P
z7_B&Aff=ZU|A(=6iq5MI+W=!UMq}Hy?KCzT+xCvx*tTsqwrx9UoThOm-~Z3dnwg_H
zdH3PF*0VP5`y%CQ$;+0A!N4@@XtuH4u!q!a`#*oLe?GQ*V^S)-7~KeB2T^<d$vP_U
zhC%Wif8fRo;t%8)rl)OO#fDcI(iT(Zy}=i7jKh96hnf?N48Le8B~+Lr1BHUNEg(Kc
zh6<;n)cg#ooSz?(fGe<9xvn@YBiufsA<wpyuI=)b>HBhS#Da_A`*+WtHK)g#W)|L_
zik44H`ghguP6bkq&!?HT++HRS4X(<`?4J#E_oiYFJCB|OZy1jpOp!S?@;n(e1B|`a
zCVFv>v9n4_=o5KbSc+cOyxX*?)d2bhf;0~!hhQs>@`bZj1vQY1dvdgXg43j)IPTg`
z-ue-&pz=q-*8ltR2HI3=y{^9Dv8D?B0tfUDVfGR8WB=)kL32ve5<1USZ3muJ|2JR_
zc}ADUpsJ<VjT)F<wLZ_{K8Mjqq^jTk+ZclTi`Chzmp3y5Z_)eN>apD7W|0}2RUdj8
z>!TQ4tX9s=ej9MmeYBZBYJ47{Ou5uWbN;mpY080*^6K>_cjiF&moQ)=6!ySKumCpz
z&iv{-haXn`zl4vOD|)KIeQyn1J-CrSJCipNkn832Y{B=5EzN=|!URxV8%DQ-qAD9J
zm1C>htEePYGUOE+l(L63H)h?knvv4;JmZau;vT-lQdPhfT`NsI!7oRH2>HofSO?h$
zHl86fmHqG>J|}leMX10ZIy5Bh3$^^`7vc?aVrG;XCh5adD<0xLRd%q{-yJnc&i`l8
z6t1xB>qmn;ICZEvp<vxQN4G}E@!ArV7vtL&wqb$a*!4$1b~RZy+0q};{#)ZVzNGG)
z?FD_!1pIwusAojMc!QvO>cQFS%<Q~)#h>)3Ubm~d=EZ3nIP_t_tc=u=rI~!99_6;k
z_YNlS|JU5-uQ}-Oz&Y!?O0CBR%}6RT6%fJ(9W$KPpFU}@UT>I&=vR2HS&h}*U-R24
z)5fn`g|-<z#^!e`Bs@ifzuhkZP_J^w#rlI;LV$2DIT?ntsr!rQD|}i`<Vrkcrm;i}
zmR(}A8R^du4z!h7S+|*ie@(@(al47BTia~g&TM+njlam&^wJQMK4=9kK$}dNRV!6v
z@QxuHDKc8jM2t6kvXrT$bfTNrwOgT?D(iPuTCFg@;fb@2i74z7{^y(6d({}EFGt&i
z4tVwz9)`{H&3`M3Gbac{hz<+-!1&J9+_@|M)q}idcdly~#<vjACz-QHr+aK@yD!tM
zSNt-CqVeq%w1Vdhu+fUJ-6vQWQcpQW&6f3KV_s{Yj^Q5rgN1OU^iECt!9GtlM|8(&
z4j}+zIstMQ)UP@JUk-Cm6m2BklBZJx09Uq0gpXQr7nM~Hi|subP%Q562XWNJ5VmLo
z<%4z#c6CeBo~+qVUeId$Wf?AU8Cw602`jQhc?<H53n}O>OmK@Z3HN_Mh44{t!<CQz
zZ=D~V;mUV|g>Hzd%wGxqMqNXCYD6l<ipNqUc<JS$FWx6n`lDhra<UUoi!2Vv=eTxV
zO!MqtRxQ;RY{5+7hKy0W@3gdzdHnPIhr?M>)%n}lO@&>Bpg!CAk#@-d-X|PcJ+wqg
z@jDVZ+=b`)UcKz%hU4j-zkb}sRr1+jD1Y<sXCtEhGDw+G<l48zA3e98N0I_z?=pAw
z*btdFFu{^_7=Pj^M8ypwILJf+@(pp#Bpmm42u9KlKS%dzHow-mnh!>0wifO-W}<u-
znc{m*?QewPbR$`0-)xa_>oq^a*hkcz>8UyQ6Uhui9!x_K-25&FC1xE<f9&p)t73Gu
z3ljhS-MUY~<LE8KQ_}kVJ``#>p_@ym75&C(^j#KCHeeUO1zl5$3sV*pK*9SZ8Qq$z
zNNW^SN<?ZFRhBF0p1g%X$CGQm>jiZ;zRA#<VajU*EN%&hURG;yDo4SG55cchagPNL
z<;ju8YTE1kleH<xc)<6amY;QggXpfpjk{NYXL#FK!-F?<^OTyrMe`x*TvYU`PDkIc
zm@rEqbnp|=JKC%F#;i>O#6L7Qm4~HPdTWFfHCd?^7>$xW>NTt3B1UL#bD3=VbldLx
znmCv7(jZsGlYGr|l+n;SdMz|pc1{=Pb1RtS(_Fqet(N*d>sH8|!W96Yp$4oIUZI!Z
z%2*h8Wpehu9~ZOFVEmZ(bdEz=JW{v|<r}w^l291|uGgHe8T@kqv6@d9cClW(l`97d
z9Vw$n81YOeS)6<S2W|+dt13sBy=l98L{$lU0toSalS@IM(X$^`plFkxeYE~M=!JZ-
zwtY2!9Fs?w;Z`SEoga%jMwj=@=!>GJ%M<1<Nb14g5=esmkpUPHIR>KR&YM1TgMffw
zWB;R$POO_ohV3Z8->p&As;0UKVcU_P=$|qS*kA^a8-z<Tz_;y^%<-~q@vd>8y+(}k
zHx5j}sH*)xi07&j(KCh}x1?Sa4#cnx)v%2x*1b9`)874G*~0td_htgZq|%MdurQsY
z^4G&1bai=6_U-C*&>yd~7&>>_ze-(Sjp>1>fI-TlH+SIR;@^heW=nIph1yX`6GsuL
zayaJG02)cxa8X@kR5`ba%`&d{QOVKxF^l&VOVcu$xHCfqLRh^sHF13Cc%?rV`{C=0
zhmiy(D-Xcx)uNjGp)LlLs(AmHyfm_lRTHrjL}UiSpGgDmCHMCYE?`;y!-|OhIRpon
zH)zg90a~9=1*o)M^#y_mpyd6evi`zP4W<Q`S*h!NDl!VXB4)7%+fWTBZx8iBaNt~K
zY;f<g<!dR7k;()J^v*cz6#fbUmwZaMq~cL>4OScr9k!c0f@8Cvy}nwLhj*iwlCqFA
z)vKsK<A{duWs21f%k#|$)?&mp!rcb043?Pb0iwJ7*D|t6qvNu%EDeSR%r(51S7ml%
zBpoe9qj7MnxI;digE0szVn)mg;)rbD5V*WW&j*R0U|deU(tm_Dl&X;+S*LZN{D=^&
z1P`KFNfsZ4$1Fkb*5Q6Be8|@hZqsA`)gXM!fOB8Fz)&m}9`zh-SWYBrWDAdirH$Q;
z0nSCtN+rH0P7t`>lPGufN*Sbaj(*X3P>3YTbR<Kiv>ZDcuUSosw_Z;9=`#IfRBq7<
zxl|~H=*;6V(2Pb&Bofp{NtDrYD206R`XU4IZ|CniGIWhF7*Phd5o2@+`>1_JSAj!(
z5#)oL_Rf4oZske`r)>G{2l*LaGJ2d-VBI7(&=F<oXMaHWajpZ~PAsjBx+%)EAb$Jv
zHd_`R9tAtfl%QdP7{?Q=3TZ>1LQW3ERa2X@qtfJ1*7T5(b+$&r;PT5E8hOjIf`A=m
z%kt3ji(@g>f7gyx?H%_1?5yg)D@mHiCic}6M)ui~`NhKeup7Dx8M9qme}e&TzP&lJ
zAH`%1Pr;RrI8b#|x(Eaz6uJn+O$!d}Zc?zcvdLccFhM#$VBxk=s15llE2(adrQFf#
z;qW&dLN8Zk$y(mHX6fpW+?9iN>W)$+1&6PC_@~Z;{N$U6nZLG3yRlWSb2hu{g*xY^
z3j!3O{w<Oh=ILrXPI<X0hgKjTwPc-@<yp;vf(S;3OpZ-jHZRk9$4uCBf+7Z_%|H&7
zqOt0W^C~3?f^`~5XO)heBqM*KHp;~t|2kY5ZP@z^y|G5u_<pZa-DF&*LCAOpJzW?=
zsOut1pHc1(4bl*6$Z3@Evvq181bhiP(q=8C3O==vpQFnC!6#u{!9bdai*;9>YC>>p
zjw-wN$|&jQXYkoBgOd1_%WSf$#Ly2j2Cjdf=!aXi2)Ry(DlobI)9KvA5|82+yreP~
z={42i9C!hR`eq7g>gtJcW<>0n0cE&JKZS-hC{ka8wvyg6k0GwKZY(|aQgjArTcVBx
z^<4;4@BVo3p1KgFo&s>~KYo6XdQhvE_?c|-QlXlBS_;OS8l~g=(XU7Q@@k3`{M092
z!bpzQUC8}_@~;rhhF5IdY&QT-Y@~Q@XDshCWTxZV^k4U9w?law>;``Es$J7_uA9F_
zi<Gg=YJkyn{51*9bHe+Wj%Se1T7lYVs>-u|yKor9w9q<8Nh%;~WxIBm@M>(-x2#k2
z>XjL%WVy`an)CL3X;D6E{$lDES{2#|u@-l#q-O3fe{1FZ1iHiEOK8SO@6qEAx>xV-
zFm#vP5R$UqM8A3KuuPF+g`B4}vpb~pj2<GyU;b~*sYAjwaer~fw0%@X5Rzu1c+|zi
z$30@X0noKc8P8HkBwmx-qejY4y4R=e1n<j|UDAWLjiDo#YYG5MC6j&{U;cnL(bBTz
z?UcfBf8<KS`O&>0_|-bHH~l-&x<5O5F<V|m#u!!Duu_#cM?Y7U$7L!lLxq=@*>-A9
z4m>aiyp9_=!4%Gllj!38AjB9F3^l;NH5-{J3K*Kv_sD-5Z0(Uw7@*XBDe8P6&bWGg
zrXCwW$db>cQ}Kf|w8MTEXZ>iXQ{%}@_>>CrN!;=hG5P6C_4N{w#UpZ^SGC=?k(%#6
znn7p(y|3d}eioX{R1cXwdBoo1z!H*3oh7$d2Sdqf%OYEp==s;2;UDFD(C<6}PlmLh
zK7at)tD}fEI-@Ab3{%i$y!6ko4%*WE)B%#<>{IZ%%>C`l%kk9<1X27T1JET*Laz7W
zBiX9EHk%B8NQp?#tCXPZYJnZMXrn4ae+8#q-r*TyZs)k!S&F~sa#PsuvW+E&kX)>b
z<$<RgcYS3%IyLUuIgSqoW)nR>a`Q9*4Sdf}>64TJjYFLWQgKmhyGYiVf~HG%YwGl7
z*>PE@@Lg3E>ub!%E+BB<MRqndC0D6Z4o68ql@!M#*U%{V%cf?I{GP=>E;;?a)5fx_
z>6(SXgeo(ta5`AhC}lKDbT`=1zzR&4pkAZC9-}!?Y);z8N-i=c(Im{wAUSin05P{%
znBYGO$YR%+tTof-xl0YuMgrIz+$;8ao5ZKm@on^-40f>xX6_>Cf5q>wKJfVPKf;3k
ziRTuY4kxO>Q&~k6%W+x_w+g6Gc+I2oUC}0aK~pgz9dMcbpS|R)a2xlPk81g~aKcC)
za}EM;DkDbmk<dtwvIVVWvi%LPrPMs!QJErvds37Xdlwc@&@4nYphS|`n+1nzw+c$@
zXaIa}R;G^nE(Wo93jZw5xxDTh2?aEnZs}X6W?q&ROeiU~p=utRpDON*f87RR36jQ*
zGwY?#El!qd6Uk{DsjDmF$)g!u5<>Et1c&4acAhA6718=KHoCmfD@TBKA<wmS%c?7{
z!^vRFJ<?E#$MH7!*aGoEgqC9c&fVPgMw>JmrkLCOVD*Yn@y7eIIH!B2&b|1L{V$nH
z@3tfXxn?kCS>@ZoZAoQQBKcr?W<!$h2&P^qRP_HfmYg0F_;$DV^t5+EHTyhi3F{SY
zzTcO;US551aRBmxi6>EE$HcVAlgQr(O6e<%im*sq-ea<hnYbRD9-LHJU>B!(uu(c$
zdvA8c$rAz>0x~?ck?W2kSp&zZyJ~WFKg?})Na_hg6m{l*G4-0`IOBr<pO8B6|HRar
zCBvg^!30`h1yGS5Q3-<L;*PW{S+5rxKq7Vz;8EeI6TwhRK_b>3_Xp>PSG``TaQYXx
zpM*v4=DGJbm)wLy^x4dV;him`ym%{Q{>s!hZ$6b%uSAY-8tVwD-8hEQT+hSi(e%E7
zZ0IDNnYPdw%J%@;4%aL}YRMHw%wBb#Z*mk%{&mb8FRRpwMC<x=KM<~nE(5!f`u=E6
zNP#wD8M^(_>mrShY+bpAbEFJitXqbnWCBVT>L@hJDUKP$&HEYA(H=+oS_bs{c>~<^
zt5pxRsIx${U*NGe@@~|zg9TE)@IVSJi=AKwb8ZkTw<=(b<GTk(;y*@)D&@QI82u5n
z%p=01JBvp)-|VrYQy^^QN*Z_UCM|?1VGc|i*QEY7#k6gE-G^S;4i6VRKZofh82$@u
z(F1~rR`kGsWM-e)O!LOlhHN`^p#M!G!cOsNVHc;3SHBdM){6-IiV11z!O6C7ZQc2R
zj1Vms83$s{mrnHsMZlIqjl&adU2D3)wNuI!2=X6YvXoH>3r$YbDqHwi9Qtq;$X7)F
zDM^>lI-k~xp1>uX1$)H$*}bU*Oix6t+#&Y}6k|QLJN9T_pmSbIw!onOJ-x!UqRk#Y
zxCDPOKX_CUj_z#9C5-*OMUB~mb0#+G<;o3eehQ4{X+<9vn}2%`nabDca4qV!DT9mO
z_!;d|MYGa|Qi`05p~VR+vFb=8QDA;99okw*M~UM7Y?OI^aBL8UgArfgCnZKsEsif^
zD3DN+qiI48E(AR=hR)$)cwk{m|Ez6MbzOz=Ejk!e7OFKj20U6+1?8Mmm{KJa%^vCT
z>K1r(lMN0Q%tpo`Vk1%oUrjF%feijuqydTbB25_BM>jWED<oS)QP<U~MeUjU;HA=g
z0QrD##vfiy>E*~19HkIrdCvQ^EaI-5$Ez_Ys2toCc(sZyxMun(o&EE+-%C*bzZz0U
z#J-T?b_rAImA;d`Z+v(q$yRPhG5_Y6)Bs4k-2Z5mEK<AA|J6^z@D(6L1~%z&5QtG8
z80*AR4J<`{$GR`}BINbBkUrbY$G5K?IMPX5RJ~XYO=M)r*oz~sXYoAb2rgMU<U7w)
z;Q>dl)4RFCtgFdngovpNVyod}8LhkdLlCk1v!Ko)wm$MA`~>`t+5=ew(HC%>_yx1Z
ziKSMls#DCSJ%H~^9V_UgSf8l&3*{cV#ogKa_u6T8+oT(tpUyFzghl<LW!4q3c_nIA
z8ZEbyPgKkgrDKRc%yNvfzNYHI<jooK+e&8yA(@icBU!zIDI9oxukuNfUfcHQB3rBb
zT4zZzC#9X+I>jaCJTchX!*Ku|sH2s*Q<NP&BiaLxaoesA(5qrgv*c9?X^aHZyjoVv
zZ^8A{)>(L^Z?r?lr?yxAO@ZA{A(_yzFaIBGrV__%@!MCRbZy4eoBZ8*xMeP3VO=%e
zXy)`Y%CezuW*m`jnl)()S=-=v1{!92fyI{oz=jVv2Q)e8#zav&+6KT7RwrK5%e+kQ
zxfvy}k;+R(w~UF8dl(;j&$0M}T(*5>VMeSgj^=J+a`?SPc`}_>Ko;>xHB5%z66)PI
z**AXP-NG0*(z40^M}>nZ8b{Qn4#rA(3hvYk4^jAjG33XMuy|Sa6S~SE16W{9p4qkH
zU(Np*iQH=r2hTXIJV2&_BDrW-vVu(19gnLN!fxY7YGBw`gB3oK^ukwz)hiKM-G++=
z&6t@srg^NY^q)p+xrfG<@*jJ;bgy)i-XI2S<44NXSYh(vDWcO$&BuKFD(A#rZ~re<
zCdZ)hmnxHQ?|)R8O*j9iD)SukOO*+@H>#YaV}8jp8v?V21i`zfIq!?q3J0QSlnH6S
zD*e394Rus<Psiu7!$FwJL&X@};b-=#iu~5V9)e|gw!8#aJS^~UllUfQ?y=@cmeJR)
zj*QC1h}>$`YhXF&CACf@|G#p~#xxLJ;(M0Es-2|~!7U(7KrZzhW}^xw^DnFOg~SQc
zDxzBn$E$!X$j^VF5>M%Q1Z?d0GGhNC1j1Px-6NZC|KlZg|Hc^gqMVq8mGxN2Jo*`|
ziLsXx(L2k0azL}nF;@<vGL#3D=5RvMw&Gh9d#TmT@X~nM>4;Xe&a8maKeerx#p|I8
zWwOQqdBEj46P-x;W*J*##B2jI$4wm5f;{`5n))?axwhGp4~32@Y5&i$MvD4)=rqSp
zdQnFCR{x-8MMKM=MaI@|9FBN2)AZ6<(kqP$`o>B;0hu$}JGRjf))>3-X=5Q*_{cUj
zBIhNwakwXcI~J)%=0C`+SWOfttD+Zs>P!`3HUKSL1&S9Pb)+{|R&~!s&o0qa3#lwV
zy;=T+)W=LcCMXQ{pKH1Vmc!wiIAX0^WGtWHIKD75+lB|W>;)+_P5C|hw})`+)5p<F
zbQ8(Pe6aN2f886lMlfDBsM0ZVj>qOO#$TKVmK0-RXwXVfkcUFBxac+RV)WY%3jIgz
z16!IJBvU6x<O%i*S9pxU=PO>LQbSN#LU*r=ZYe8&BHYuA!7cCbk9+4HcS&X<*WJp&
zqVQBD`K$Q34S4n|tgjDNJk{!#<k#JE8>f2&z2i|wZ29lp%Ww~Wet7~<9$oly`2tHN
z8Sx6UP`PyZ->Nog%jUZM-YPcvH(7~H00O_??YVl+v{d{$Tv6ZJpu@o6R=(@b_vl1-
zcG@GI3jPnRnNRRjM5}{BkT%3ls^52Gw2)eo4}(?a3g2BC=kht#$dyLDv@)Yh^sHtT
z_7IAvG;RU>D^LvsZ<zSb3zhn*>G&^96e@EahdC_9OjNeAsyF*c6Gdz*^-clc8OCo(
zW!la<W3pqW5Ez@Q>E-SJ*~#=2oVD~1-d$GJJ7=?1`W3g@sSqD%=+Al3r)MAKhe-j)
zGthkeO$<LV@~ycDWda*s(wu&zB4iVd@m~b&R8h#{v$F@1JR}hjc=S%tD=bQWfUzJ2
zyTz(uk2tkPW%H5Olmih2p}7!n!;p=GEY|cWI;%C^h(wVUg?@U;+9(x(K|O1!5D@B>
zP+oJrAx+{#z;pH+wEmU{@_shdnX;vw5jM}ym&e{}aum^WBK{vPisb}f8y<=6SIQKL
z?X!{SI?J23WCAWeb@{?DLHeYm$+%KN=FU`@6fZMA68bB-r{v79Hr-!9G1)a*u()m(
z%$U6b9I~0`TP9H*hgA+9!K5q}2eaX20<EQ7p0ikFz>SVqlRg6DY`}UJLc0=mudv9L
zfA_aUR7?vlEB^#<_m{E|X(cDd_7G`lEBP!agc$3Tjel}|Hv{D28W+*C5!Z_Ig=zJd
zA<~v~<eG)d<w67A6>z~oI*OTDc1k<OQ7^*7N)8?kW~EwjJ)+H97M_=;+}Fd7Np38c
zZNBs;k-vCH*`%1Xh|o%j@52E+g6T83rX+N5@2SoNeu$xQXy-&(?VB@i0vVUtX{zOx
zc;`_2KNQ2)|Hk;=zu2-(^nx4THEC}3Sn?66?@bo(Zb+RK*69H0G2eGBDzEvF-=dq+
z;he;7Symj*D#64GnZ#eW!C`+~)R9|8>ISHXSm>WsnDwU4c(T3-D(Q{_vVJPYI!<PZ
z^!1-yn&P+?G#glJhPXeTjNV?1bk4<fvW=?uWTN1_mNVNTEbS*5%f1(@H;y4CYXEb<
zdIni%hA9e7U@d_1T)Y$Q?>VdRi&WXlXIZ5jg8U1Pla5{xs<g>6GO7$x=!=I+!_1VK
zmitpARYg?JUq<!7I)dRh*{}PeY$fB;lk;o{$})hL5+gm@)^^u10EtG#{Av)=hKSUn
zkQepr7v@k#jik#xxuLR1Ml@7O^P2spLGn{PfnH(Xw`1UrrE5A%LDlZ>Kmu{kBtd5y
zL%DL}vgH)xT|+bW9t6&mKlp$^*0Kn11(Qc7SPiF4a{@X_N@3pkh4ooWWe_Ugmqudw
zUTR9Di!WAS!u5soJ@y@r_sMf`|NMTy^&ma0fXGie_8{5J2knAVETm#i!PAsR&W}g{
znfYZW<R73q8g>0C=)JR_#g%aQUc1=A$>5NM5YCz74D^Y5;GML**70{@vZBR^pyxHm
zX-Ypiy1(`S$>gxeQX<1Jm13@wF2=-l4R@(C<Kw}E1sbDj7MoRZos<yHYwtz<GVsS7
zIH*-~_G*`~IF?W`&Q|1>@;_;oaKh#vUQ-&r%L9PWE?GV;g$-Y~%9`?|M(aSqD*Z&Z
zosH&`J!akR2(5q0{V4F^6~d5Lp-at<b79BmUKE-zO36<=d^bp8??#98c@OsAaDqp)
z8_mL-#t88i1<5DVr1RA$JodmeBPL?dNRJwR@!|X%9aic$At^Iik1lme@FBpCs;|xm
z2uP=agMJh8?^*T?$@zl?Zh;xqlGbfky;xm4lRTBNW=zfaqT!R7ijO?V9Hf-wg;_Y2
z=N;kO$P|PZB3;?S{OUy_LH`%pG=DCsXnr4`0`969Sqpht5J4#1abCLUy#WaVLB}ok
zRMaS8Pl1oZHgVVdQTQ8*LoQ;2AYFe5&~$K^ILs=O?9d3YGh`MK7y5L}Dr?FL29qyz
z;{>lma-*Cru{-4t!iVr(Q>3&ARHpLvj=7-;41FY}BlZY@X(XMTp8vyEzLZ}gnT8Kq
z%w=Z2q7Da!-v0-t(OJyQyt$IPvipORFZ+ET3%ga4APxVk9{lgL|AvJWnOwX-kP6O-
z0t%}rmQjH2FM?I@fJ0*C1O|_?*j<HPVAkvFkjW$tW`NimOpTfs9uxWB1%^}Hfy#LS
zot=|TY1bE-2Udj)<QU|SGb~TvCut*(llXaxR<k2r@orCcKz}!xP`r^M-H!|XjuSrL
zUd{YiY0t{}3JP{XX3%pH#3q2pZQhy-eJLzfrDVANIU>rPqk{g~@P$l=2r*?VoJYxz
z>pLkD(XVTzV4sZuBBRfP==Y$83ETj{pay3hHN5v{-h^;!*(6FH@<u)wy0@bg818Z?
zcpuOS0!AsMX*ml29Ix4pM5Elk4cG7gi60uEtYNj36@bt$QiKUEPy=`aw>LM$={-Gp
z^jneb#`AuhaP3F`j&D$=pK&A)I%L@5l&|+AaUOTBTrwTmSYMAp*y#`_OF>|Ne<N{;
zS)J*}1Mr~V>i9x&KRnkk88lx4cH!pB|1{>d28j+@kZ8L5&(@4Klh3q*^||wc2YOS(
z!>nPMP4FUjZGpqA83GP8%1)HMd5NqpMimDUz6pbNLcPJA^cE;q2SKWBGe9`3yA2s$
zW&wuzA5C=CHm%6C!u-b3Yuoz**1jEsc}7N3>KTZI9^P}oS5&&=X;{ipVUfvYt*S)n
zO*@J(z2qRv-tb#<MfnDvkE*lRxlgoBTBzVl{K<5bF@u}x5`eYvk4m@Jr5ch|Zo;fg
zy;sg?w=(IkMG(uIohx*2p44Erq75P2+^A)%<jP`}>oQE|IYhc$qe2GAve*@I`<3l#
zX=ABh7}DUO0$vg-R4Jq&Lt$Q`*PF#ERB449N*LcH`~#`e&_RR;%Pc`eCYAU_=<%{y
zX;8JmFc9em%K=kR2xu7QgM?ATv;DLN|3q-gIT{uJ{d<9MCjS11L2ONiKQX-tLF97;
zqW&VQ_@Gf-b;v>15M8LjsW1T$R029^MDTOBI3&z%5+HQqN=n75=Tf$ykxYth7XwIC
zkA{(tFp+W@Bv(T9Y~li99)P1TIBU>F><s76{!qdySKu2O-+{k*rSB%h02c}ivGL~E
zYAI9+tUaf3O!R+$Kq8u_r_w|MF^)wg1~d2*rb#N`TZ-d;cEnQ{Ev$~61Wbw}iD7OA
zF1ha3FHiQ@G703oQ7kF}%+GMeFw%p6w@?eY=j8OO-kOg!lO9f)Vk32L-X?8Ww71ER
zd>>@-fLW=s*5k9jY5OPs@SRodDlG@gw39t`8{Oqxo$1%qO-6zJF=#x`0qbW*xVet}
zrMS^BiNKX{PXmKe9WU0dK9ie17Oi3BU-+{?9-VIFY5+y&Pa+7`2FKDso*OS+G|6N;
z;zio0Wyg)HmQ1!fPOnpFoeE*xZfdidq*0><;7B9bgH9;L)Vr!cZ5JoM1%BeE{B%%3
z{3Dz7GokOd<J%3V52t}V>P5EEko@I+m8fy-F1j?UpD_I+e@u4w>t^%haJXvUcCC4B
z=Z`TY;9@jk(^(>@G&fSw`2AAI)f}2MTyZnxS<{E*ikC~}TuQC38s}vMdIWZ8GS&zY
z7;a*2HFstVoK>B0Yjt5jdHqMn@~K)SjnJNoyosg>hSiBHt{_>F%KO7%!a2?*qGfgr
zcPH@d*Ki44vJUD$&|21d#6H2yLnqNpooA;r)vhdVX1@3M_X~ygp?aR0>T2mrBgr%p
z0%M58RcDnM?%|UIkSF-sfabs@Ed=OxVE;;uH@wp*!cC{ZtyY`n@A5CPX3dn$3C|kZ
zi{qZ`>P&A-(=D@qkG!s-TPL21wuwoRS7`06@bCwoky9t$xS1T?D){<2s%h+Vw@K@F
zrO=|~sn$nc&AM(Y+&3ucO~DZJSGiI^sX8_L#P8^V(c)l6oT<O1w@=i^t?N4@z@Xp!
zXj*8WAz_59(iOUIFwtXWJU{X1^Is-bU%nj6(CA-%g5eDucRpk2>3iFz5fd#zAe6=!
zl+zs%<S!ZIFJb%qnIbNkX<hW2p)XXN6E4*1?CExIYzW5aci9h=f%KiG<y^DS*cngB
z4A{Qhg|rprJi?+&4TPuHJchOFK$He<!<>%M@$5Q-G_fQQAtLSjLuB`7+ccqJ3Px
z)qp9EK$G_O2>!c2fn94!NK&3bu2lWGXhmHMgy&N<Zy>6(of4!KttKb>(eEt1L$HD5
zT*BZ_8|}^(RD$w@E~~IkHsgQMV_}VOqPO!lf9s}ua403`Nbf)u+zxph0prp6d^&mI
zt`~s}bAv7Bm?441a---vp2iX`FWaL`rK=@^83%!#Mk>{c1zv4TLlr-O<~h)t_!Ra+
zg&${*>fA5?fXCJvA2rMkEG5OO8#Pw7{9CAI-v692lNhn$KYd>|9#V5Fz}N<mT&Q*R
zzw_(AyIe_3K6(G+%I&=YKoFfdF#jjuVg<HObVA4Zv$c-USAltUQA=To?Fr!qCmqv-
zhv=c4DiT|cIqzY(VM}AW;lsXsjs_#M_|dae`%jier0a9nQq|(jg?PI@wK|Tn<oO$%
zfPB4RD~P)0#k6Ydh_b0Mb7_UqSGq;ReIzqIR$+VvvOV{ikqjV?sUw$xKN86w_m9zg
zFi;eduhNvpw?hE(F@wjrg(V=BRHn+iyY3E3C;iB%R{=wf+v4VB084p!x81gG$OAWB
z<tbHw$ji9P+nsPkD~C#pflg3>mHssnO9g`Q)M$ay=x6LtD*Wl{ZF&bS&fWG^#yOXr
zkU-}UenGza*972WW?Kg(d?^>-(uECg0e~CAMDd0t{U8q*nyb-~uaVyBy+5_oZ-sF8
zTNpoG-6J8CdY5{Wd2i1UOgeV@ij5pquHTHaDkpaW_v;hmc}+1)0?VF~sGm1}PW!&U
z74;^RI-Ucj^ALZ+S7g^td9RxN=g=Vs%319`)x|6!%p3@JVn$K|?GS5?<>ZVfYc0~=
zz=SBx;gQEbG+)gb%x6G@k5<v`k^JN#Ew=OFW4if^Yo+(eMHhvn8AYdw*W=)rycJPV
zPG$h_O<Dga4>qTlRQWY934i<K1>4#zd<8M%!e-}ZiKY%*EQD;A#HF1XhGUk@xE;TX
z`OHF#S^`=zij<o-=`?U4CBIG6vkXwQjJ1<~rPO^O*~4_^ICgILc)bzt2yDV#d^k#Y
z%kGs$sW)m`n%-3P(?FpjRG~3HGlN~w)nAjS5a=*PuR3oPNYSfBC7UT*1_i}`YeNvU
z8YoWOf7<BZgmG@%+Dwu?a;xm@_1F1+pnC{?s}D4&tmfLR63UuF4<x}xr=e#G1es<&
zpcc@CWfK-L=8GgAwML_OUtVhGar<zYZm_``60}IQ+Os)H>oGK}C=}B!MQ+Nn;9w`w
z*qnoC%2iV{%X^v1(80w@AC~i6J*7*o7;iC+A(t0U(Ps+%E1EBpiSE@G{%}w8-q$z!
zNdj!5)oh{p30+=_5R`88A?2p{4Z<P(DvJym*KmEfdPeauz+T^^fJu9f+@YT-$Er}&
zw49Y}iW89@q*++HNs<2G))3B}3-N+-MTnFyC2p+XC3Ia}I@kD<2QGf@t_OV=)L~!7
zD9HC{FF0wsLVT&#abza9{O`0_;-qXcw-XT9+$ElT+dFN+95Slbg$soeEC87j5Gtj?
zX(^+a%Kl}`H#<6p&4VqNHt1=H1O8~AsyQz0axibyzIy1VG(9`XroTw6iS9n5yOiBX
zggnCVIFr4|vmxF(gWRldz|$M-pT5>cZnf2>Lqh&aRW}DQkdB85wh=Afs>jd@UJ0zf
zUo&;?{&y6*fvRl~pMO2JQaex0EAMlwX}O1MZUnG%#3l3A86Gz|`gD20E;?ceHF+h<
z$Xr(9PiwXhY)|%Ht{w^gzPJl2p4<xib+r3=w8iJXC~Rkn)WZ%9cNzupW0;C0oity<
zfSycJa~iN0-R(Q@F#SXmcz@DV0bBzv2GA%Dds_Y8x*`q=Ww)G)hs4~npFiBO?zdzO
zCovlU3xS==g>(sI&w-okm^;F4Qe{<@>MA78w$)29(ORvm7u;{fc<;jog|q&-VX2#D
z^P)BfZ!kPyM>fgXp2-X{L#%8Z(a&@`@Nky{lZ=@1EECl2iPQ2IA0N_z06)ly>Dl1}
zJg6YOfyDz|qgnGhwIe#Mub{K_dR5BOK%FqWn7{*}sogWyuSpejehmlHlLTWL^BR(U
z4VW#w?V*r+68FfDtu5U^>%laM$E(BWs2%gr3VS?_y^~vKgWIzHyk#98Q5C+-K8F3H
z1miT#cFr~sj2rmwh15K0pwMG<YpjqJVrrF-<V#gvWO(qSL1@o@kpj?TOh0rvdemoG
ztFs0DeMA%Q)U6xecCNJB>hddk4&-R2Lp*inX1E@US}We)7zx_?K!8l}&wJ?MQV^1b
z5ISG$Xam)7Ww|TF<AA|xN-nQ@FhmU!XC2CO`-5QD=uE_0*62(zFdSVFpPpK_o^xf0
zouGFQUn)re=_ESzgZO8wYBgKrX1x{v;l{6or+kYQn(ysujN`9o<hf%s-`#yZKm$YR
zI~CYzna6G@qqEm3`eyzFA*;I{JsvDA{k>HC)|)UC7r|PG=NN=_^K*ZY<biVUQo^${
zJxJoI<}b0{QJLxVAK+1nBd=jd<M>L*NcSgqP}Z-M0h?x-C|2%#683SelM30Q+{!Mc
z-=|_|gG(ZDJ)imN7gywu<mu7i&xRblRr_q68x~$zVQYxP#cs`OdIY&Akf6U!<yqRC
zSH6VdBF(vB=#nRd4N}aYs68`vrin%tIOz(*z9Z@cI7NpgtAIL%C#1D@OxIMMz$DU&
zTc?Hvq>N5+L(=NDCTfz1>e>X>KW)x`cV}nUJ+i>ZSgk>Je&Xht_5%JMr^&RvzjlW2
z6Jqg~BV-BkTfYzV*Qo>!7v%1N5&qfbH63#_9BdulG&0#y1p%WHxr1}y$|>4+9}=)&
zmPs~}*H%3o0{Z$uIpYO-0zc{-B6NQF3VyxeP3Axj$R@3w^K@;J63-nnA@s~Q3@_%<
zA#`?l;ah0ohzv4OS4CKn1g%0Yd9*kF-i}Iu5GM*@VVnI(-WBSr;6*d<aN2*`9hWir
zPIzu1>sG&2$#z&I!z-XLSHip)F3>{R9*?64Odx%)0FsznvhvHBgY`5W)YDLTxT80X
z?H6g6GMvdG;Jyu1kTWPTyR4*ZT$6-8TS)y?w{4&4GNOLF9;`zzR*pvK;Z7^wB{u9(
z@LM<GVZHm-vc(qGd5Mk<F4+!UUdpAZe&qLA6I)|n2#pRgH<isHz^!x0zxLw{n}@^f
zC*8*DFp!r;J6(agu`NR^`t=?3?+TqfeIa;n``y#?yBjRu?@3L<mu%<5ZL6m<1oY_Q
zBGHQo*fAV%(pgRMGtQFx`Zd$}ijnn9c|v+edUv`+(k-0-&9tq0Dr(QmFRExBALa$o
z5<pmtcbKwRRwX*xoi^HmpGS3v%e7w-vs#e@kZcHnrv;UuctnJWn#5xo8QC&NC;YbW
zK`mIAxzsxqs|lW{JNZcs5-3&S1RpF@NWm$y`Bj{brspvv=3R>uU*u(O%F*f%p%K@6
z>>)tj5p7n&Gj#lVWr6(bB>MJLt(kJRaU#=4_Byv#o*C)Tr%|g`qte<dkA6wki3slv
zXl3PUOH~Ge-oc>_g#zEs4!J<D9Qp=FcL|%GfY{~k8TdUUQg%$|)wiG0fxl6joyaEg
zFLBE~diXMQ<O1r*wgw~(m^bv`wnWVnNYT~V+(FiofDFSomE@ZQ!rG4T_~s9)S*xXr
zCawu;|3^rzE5>2;39rr=r2N(+v(+XkV4A|7qeIumfc7hEQ7dV$J(2Zm-9K;%JzcQ&
zFstuKT3~Hz%ls|C&0qXU?!U!GEB&*HEa?ywv^vH~6PRJv&8_*qis#1u8wb1=C6lYe
z@JD?~x9Fzth6w3t!R7cLYL??i8-hYTx8uOP1Jp$}VU{W<D{E>cistpw^It5GKyKPn
zK5*LInVH$yBOEcVL`+`d(fpzjt5_jw=z?Ac>jW1r@xJbQ-Y8kdo=ZR-)w7?}uOVS;
zE;Zum+)=z|Hk|Gk`C)7G8JMShPw>9}A|E<jnV$N^2Y*O0-&kMxRTtdDS$rBr3n|6g
zT`Ou_H51qq%-UCf5roZlo%*f-M@MHe<Hu|)8gbu#0+{(}lyGDE7%u_Z3vXjSZgp!6
zJm|;dCE<&sSTd-c7|z{-(Tv~#%xi`*%ow8;&BBIsjuP_!(82fTy9dSB1r1Bjw11J{
zeOxOx1y#0iG>^6`)wFHhY!rn^q-?)da)clIi0b>$PtL2NW@va?u)yc~*_lugbPRH3
zQjs>L!?{z%$<#B(vbpxDit@o7EH3K$PezA5E=opKjxw1kR+DzzA*N;{Ssco_@34$_
zPY;n0n%{}#MA9Cq<P?u6i=?Fz<4Sy`(9p<Og86tV=WeIb@KY7*wfl$4w~4Xgk3lZs
z*jh)C5w_!>#TchY9Dv&~4tRPAI<i5<a6?5kyju#icur)|C=!O&D<YZ+T05t|wX6$R
z7A9ViyZsrEs+wn=pMGs%oMEtS2^$_G?-pr`16FLfYVC=O{v$yRd$q&6gktsBM?cF>
z5M`XR5DKlzwQVti*pfc+OY+B+yO*g-eu$Enho3KLmu7AIkpuPFr$?|Et`E+lIFsJa
ztQ>QwSxz}alXdN+sfCNq1q+F2L=WJUPK|v7Q|I`89>o@`f$44=C55UP7r4>?4ce6s
z&=d47C7$0VDWv;YhG*{WsqawE_bRauT|L6ia;C=!?H^jMk!Pu2hPWGLrQq!sU76my
zu9Bk~x)zyf0fvgtM`Kr|t-pIgUnnTgczzGj()fk5{DFsEP}sAD+7Q0?vH}^jjFmyh
z#L_ru0#4l3I!XCj8ZAbfz^0HeM4}|RoYTs?K+ellMd>KY1&PrFX2EKxXVJ943>|#Q
z9Nf1gSdMxlH0O(F@@<cc7`#F;JNV1}u+ZRsc9isD00TU8nN`BEHGDXh5ocvl*e~=_
zx{8R&(ci=o9m*X<EBfEQggiB+lb{k>>Dz-CpAKHIalf_FkQ4k_h2RTh&5V+&8~&5n
zVjoq+8C~`Fc`lKM{Bs-v#rcalMRt@V2nM{9L)#9GIi6IF3QpU}f*j^d2>S-lfAWbA
z_$YOjKtqE;`Q<c`%OE1pz6E;F$8S0}fe;Fq4cq}!DcK?+X*{R0%FZg#5uV_LCn`7f
z%toK)E;2c5DaNMo7&gfQp+L;g0HY(h(d(}S<`aqQs)gF0l|_)`c$AXHW&+Dz)JpwG
zW?&*i;Fn(E>K@0E_AcW4l0uZ+r{N#coquv8fDh@Pva{|MxGl8=GR-ZE7I*F<0-KfP
zm=i<lB$S8Tr4YNf2wD%I=n1@My66PuRGJ~Qa!n#DipNYk8#KY56}=zo#mY6Qh{Vs*
zwFAjkg58SeYND+<HV2k=@lxZoPR(la=Xf8&)ITpOZ`oKTDetkQRxIMSOEVJFOrFqU
z0qG60r*z-hj}JC7#uG;6-}<W-eEke3uZFaTDg|!?q-3{^s~P_g{C?>t{>T>4R}LwL
z{*FDtkqfn|oJ?sGgYOLI&ut}@mE#xk`9NOu(Exdhee&Qq@vgj|FL|x<$kf9Z=1OP7
zPqNj$9YBQ^zGYdXHk|yBYQmZT#i(xc7l7P=SVLP9$*4dJ);b2Si8KA|%~}!2rm?1s
z6pxD&S~ar$zUBa3>bvn3$2|PFX|z)svwN{U2hoGf<`0WxSK6ZIL>GW9E?wd7MaNFU
z5(zDOlH+6>Mez>L_0>h7V)L69IUjx%1?8`D^|nsQJ_O0~Vlyo>U*8}}r0|*%4&eOq
zPeMh~j|$rtW5ei^3}-`YOHd0^qFBu~If17u0ph$=bHun`wToG#wHm*Er%J7Z{e!OC
zUeB}c;nmN;;BlE`SG!x4fxP)W9ztC40uGwcfy!`l*{n5!+#QdXU;Hy{zrI?Vq19?@
zaUP>b4_;w2!YEvqj&ps}9XW#S0$_;9g45E4J&4Ya3K|L<e<3F^X8vKpR>V^#a6KQH
zdIV6$q`IDVDB5*Z^4~f$S1j8a?8>)Bqjnv7Ghaq<weR#?6kcp#4SJ;FWouO|AL<tf
ztfa1m3wdpH%N#HjYX3W|;`05LG6g$bu;FkYxN6|1Z&KO5b%}5Fx0V1H1zM4mT#7dX
zwhQU*20DJw#xTgo<%nthwZ7oQ-fH=u55==z_Uxj{-}l1<Tm#)_vrW_9$HA4Q#A>B5
zKFr^n*m$P!?evdpoES#UMG~f4S`Lg94R&Lc!*iTms<r3g)ifJE1jxbAI0301Tw)lW
z#1zZ3Ntzf0Xi;B7M*nZ%{^-d3P}U@E!ES3XqR97_&%yW0!LRD*;6p-P<dn3}AoEOk
zL6MJqa)5rKW{{Lotgp3?`T7MG^lOYY)E307B1)^y@a<}saKt|6lc=}y*2|N$NGl6x
ztx`lt+b_3D7s*s_9)i{br;^4G>aY`|DN<Qmgxf6=B2t%4cJ{zEfuhc~%UE}hGQEY7
zEOF+?uT|yM(xfQ*kjc^1RpqBi)JVL;F!Yf*TWUOU?Pl#tj`K^`0L1K<qn@p^<a1r4
zoyksAOyfNc5LC?)LPX@xOGgkVA;uR696?8avT`g?dc(+gIM7wqXh3-JWClTvRbg9o
zqD9@r;Ck)aq6;8&G7<&jQs6IgANP}_ZeLWvqZ`!E!Grt~fmk>7STG~T@1N`cJDMKQ
z`|O#USsti$aqU>iUJlv}arJsqo8D~j@zFOK2Cb7iDY(k^i6z>i{(OJ=*!u`Hjmz4@
z9SI7YiXe$^qD*^|37*7yHgLxr>l{or#e6o1NBZcJkOMdfyGj^6Bhdd<N2?!BAQ4>2
zNoEn#2y+)Zh~p-yXkXz+MS}`>TseFROm3!KJ~mW_?}LgXJTBxYIiR>qH~$&iP+?(J
zBu6!yaXW~~`wKSAHyvrJCUgqC5$)q5tS6hFr(VUt>kIAb3?AHwgR%AM0$?$cw5R%o
z>avcv)_|O766=`YJwX>^d;+gyB%`cw;?HNpy%s!^6iqCqeG#VfsIka+4D{avC{Kta
zS$j)ouWb{UT(!zHN>fYUP%Zq9)$VqoDGvTA)K`!WVZCJbLOJ!U)43jLBNu+H_7W?H
z8H-a=!a7iR9~x>i7vQaF#!~uUww_<C{6nUN?!Z|5lUboAY);O8&%1{|a+|<go-qgB
zkwgX59!lm6sROYCC`?@`tA=SC@%>PUw}Ye2!0~atTm4B)yd=dOG~6;OqzDV*Pn{2)
zN=8zS{3*)GQ!s4S0@7bl57tTM4rr<&c@efVRZ${+4VGO7UC7Qw*tMQRX&4{elo+aF
z+5lu34V7B4#GdP&p8#99D<azd^pvxPeO6GIrvMZ6!OTgcvkzj={<BY+HY>t{LfsBo
zRZJa^KZqKjME`vfj;mtWB1y+OmoPlK`E)Qp`^z@pxypuu^~*_rk);I?o?k}^LdfdG
zI|?G*E5EW=Wai+Q^|#XTfDK|l+hD8!1s0eMK`0&B@3G;>ySl_sGf=B^nC#Kexv)E~
z0Zp@lA!AeD_FCEre-D4Vrtr)0(UI>(QacWVf3TX3`#E`_cBDv+IZ1%*vT-u=y)`-S
znK_~NTH^&Q%?V?f_EwdL$1j_x%QuOA6F!YTNY3qjHV+g7o^O)gNeln2VGp^0qsf?G
zp?>*t0?$w`h^3(^L~2>vTtb91#W1($tFU9uqU_YVQItRom-)c+0WNG`T4@4;&PvQw
zIO?r_`{Mbn3l6`&HKO5IOkcu#lm*6G56K%BPB4iaD2@-0{T|Ne2XFVsNiJ_-5yg`A
zR_`e{2$|$u0{z!Z22}c2-1;>DZ3`*lzj3IKDkQ~>EN3($k1s-0!Dci#Y%T7PTka0-
zcRvI8Y2u3>3l47Si&n5_p+dVefAImG5?x^)|4RoL*XhjP%WhrYGPGw>bvqTeHhtOd
z_j`O-DD&yqf1OZs<X+p_tUvs4_3%_S)SD&JaP_=<FrfPDRIOVJC?B1|i0+c|;JqKe
z-QKz$pgDD_pK)t@q&gi1NFo&m2RL1h<J@q}eDr%}*7x+f9Z*MYIn^GhWa;!gTAG!0
zyqg$V%l3_re;a(i+N56BT|x2@9u)~g+FLucyU}PPhYY}jdxoJyR%5sY?%iekINo92
zN|4?NiuT-%YJ7VE@UBVp9yoUmvfX#8{I{|<)5*4Qa+zz?chTD%d19XL-?uUBL_cJc
zwOf^Sm*P8a7ubF&jk?e0!Yt9!%MJxxg-Y!FQAthspJC#EXNi^B!CDvQvk7cHYxyNu
zoQjlpkjLO|D@Oe-I(<;%v+FGBWVk9?w1KXYj3Q0I&gD>f$~y>N4VOXV(C&*i0gGLO
z^lZ$}tiFSyRSg!Tl=2KS5pVn^*vmTiMKlzwHaIh~+QJ|^2O2p3nC)eCYmCaze`7<T
zVu|>@r+IlnokbSkrIWOgN{co_Q8CYY_F6~uWYpz!G_06IuUOFGDx8XBHJLXkmmWvm
za;FFk3+>=yAER$PNIL5}tk_h1mth;G>aB-->5N7MNo=aZ_H}#VLxlcv)|mqrUzlip
zGz#P&#GmXc-*Ma&Gj9nWmWFI@H6NBrd*p%VavCAlFT}h5XgN?zFuN1wzP8IdOToYc
zB#b*Vs<X=^WB4L?H~*-*{ZnT(_ID|k#W_$_YK8L#<q<sz$AXXiqKZC(>cNlY;AOma
zMQg70*}7=)J!(q%wM&KGZJKtd!)Mz{{ru!_fo?52EkZe|>jz|R$!V?*2yK<CdBuov
zc|(1giW)P*1lC$r@ElgOYKX^#G^Ps8mlNs59r96VyXOxI9OCzzs+97FvlsL)yEJgU
zJ`SNDLsftEw|^Yh#xx}gR8UJ5Kfmt+$yiD&=%l3&<G|MB$KKBu|I_H(t1VwVbI*dJ
z_PV#H$)-QFMsi>EP*L)^;lBKS`7T_oUah^RTmC0TJz@*KnBS(qKfc`zUVTMydE1K<
zsRM3Bb90VzoD8bD2l%kyx%%KU&!~WFN3$|-U8h@1Dpivy4k1#PcT<*Afx3~aySb54
z`9a~sy@8uOBB(QkIjloiD6vP(v}Gq#R&9L(Tu0yniM6ddM1L&k-WR$!w}eTANi@SS
zN{_oS`=Eo5x}|>kZ}b@k>;&v2Yv0FAlcuc#>;mjUYs5<+7jqP2begEaH4ES@-42He
za16MGH`>g=s`YFINej)P2XzL$hws;2q8wDc<fFg{c^3UtI&}tppaa{Y-kRvC7??sF
z&pPmw<tK5kx<hI}eUruo#0c~?BW!dcV3kG^djoCrLxa<i-y((x4Jl%_<p!(LPL|~0
z%`0Yw`z7!a1doj3jk9qYSAbu$<(;Rhc4;4aOIn{wHUoJe)6Owq^RMcDctTzavmOI$
zf#c}l>AqEzj4Ztuvxs3?JkrcBdY35G1)tBLgji262+cJBUdj;kAAJhVNlm#A?oz~L
zDD#^9%U7duMCOv|fc#BroJrfF^HEzP=4kdx`Mf}L)k*q5Rb0z+RsgmS0(S?j5OnqJ
zx2Zn(-d>#Axe<b)GheQ<2`p>8ZPQ!zQcw5q?hZ_C;te15x)>-3OQ>$NUyySVu=QM-
z$PL?NQa6H$>Y$_-xz08`@RAhRDr&is?&x60l~(brP@(ytE@c^^6i*>U1(xTNEQ`bq
z6dn{=VZR0jS(-2<sex6B<!Ws4sG6?AbW3<P7Dxldl|Ggd2t9s%I3^P;c*AZ)Lynz-
zSCE3*utC+IqsY;6#^Z4ZK|>c_B}Ma9?y}PIk8&bZDK*XpwPMrQG6SHo>E=?*uzz7g
zEpBqyDLzt|Tgqoz>Z>Wq-Q`}S|2*FhdW>P!&SLj3#?{09A7s5_a3*ckH5!`}+nm_8
zlZiF4ZM$P<;)(5vZQHhOYr;8s-tSbM@2yk!zy5JmSNC1L_u6aiwRpyXA^+d_vBDXt
z;aeXFm)9c0B+g@U_IKsnY?B>U=tB$6Uxu9Bo^lK=@QO8ZMKg^4zxmCzp76f_5)ODT
z&a2~xznyBx$&RHYYr9${2W{7{b{ccH;ndCyMS%#XF&bKoBA(YtC8dq=hv#{|R}5`7
z4@w@yOls}0a40#5{F5@^Pr1q?Fd`BjW&-tt*HxV2ckEu^NM;BFO9f(Bo^m<x_*b;c
z7p1J8j+KWFL7C0w=byR+n&voWk#f%`R)}M1S4nNL@9z?2D`#An{j6r2C7R|<z7&CD
z3KwNhhIzO2NpXc2Qp!W*LFf3r?$-_EZ9>&J#t39le=K0SsNYM0H6dDSzU%_jv_Id`
zjFdE{7pbuplItVN`h*@pyDxmq7igX5gGc7mF66V+EUe|r-ddt)lWxYq?n{Hqx*iGK
z+lBC;3%IeIC4R@<Pmi{uSQ3hRaj7tZ+=X(OXRv~b5XZ&YlG7>t)FGb47WGe;TplmL
z@(-KUneL~H6+buwKq0kCR31zD>_^41&KK<{%I0w*3U68#FGzUdwtjE^6I~{AOC2iP
zG|<^JIC2U@ej*IF#HCFCHlBahBNE9RA~Ak-bUC(UN64=g?8~VvsU6r*taGO7Ii|rY
z3dYh|F8h<axm2V7Oemlp3&+@q=`eUJryx64+=(N)tu-1Km`Zg@3gsBZP^QwXH;2bD
zcB@sb;;@>?(|$(yY9FU(ItRMQ>$CH6PAko7=d5n4gvd%(i}44y0wyeWC!lgTCb<QD
zlqpvVNnk(t8N~u<K~uk#{@UtAgWSeI)<l4(LGZqd++JP6{VuO=2$LmDgvzI}IrQhs
zcPMyXk<!%z*y!|t-%#JwTv$gSaBY0WU}M2Ohw=&C^14CmFdF~tDSZolz3Vsr_f(^U
z$sH%^jN@HGu*to_Jl*WFe(1#Opwa&i$+*p%Ci+yQ0hw{_fAd!oYE!uV1jo4=3~r^)
zHCX5-v!kxdVN@0^VFr-I5X3q{sleGx7*w{)CbV}8Xp~HhjP{w+Kz`<*?`u9VzgqV@
zo=@S&htflT;(G7yiMoT|+#23JCUup}Xm7=v8N9Ay;~<jU43pdtRHTRAfW!`fU+{0a
z5msH>T`N@s&)}|vL<c1q=V4)ZFJ=D@I$H~z9GV55!Q$y?p{JTc*>S0#c7cIV=}^Hk
zamx${!FpsfCSEV{^*52!Sk+Wk+Pa+!GdTliW^#_2yB1|y*Ft_ZLi2u6cpW0YxEsV#
z!2aW}CmNMLRoxw5`!RAgR63%ho>+IOnp^Y!miq$Cp_B9(q{R6!Tx8GVE)<O+#|Ej}
z7O6>UqC9ugad$3BQ*AKAD$xCcihgpubrM@_mpY$`zIL3WVCJ`tuhVm1?>rSf&-&2i
zkK65oDw0raKAj#NJPW?SR=y}NeH5O~YO+e{YWg2Xwqt6W@jU*0d4_Q#Qt>ZrES0gu
z!;B6A)1Ng4_jEqc1SpV;6Q!jV*9iI7iX&>R8mmpM%3&3cE~TR>fV-eHU-L9#O7B2^
zw%eW?N99VYQ!AdV$@C;w90`nqQ!im^27TI0@z%T5gp3jsK5q;E`RM~am2QE`{Oq2y
zgc|P+58O69e<m0D@ef$?;ocf%<!A<0tBs1Q%bC*`)cL{(_nl9bA|!4S>UN7!PM(fR
z7qb<kAzuj1;0zBq8@n7XFF#)|Cm@-%QKKkpAd)g+mcIUC%HqNsX9=MRMYtVIw7clv
zq-_36IR~APsbC(0g&>ar**3n01wEq<W2bNC8GHWr_&+pJ8IXGr8tJ?`g}5e4n!Jd_
zYY|9)0>)1)UxK)Fj6plM`j83QU=U#CqV5EJIT&sR{Pm`#tRg|S`2LL8tf)=XSUzdz
zMO8TjF_%A%WW9QCpZ+oa642C`I0Mf*;)~gwlzcbTa=kKU{m1mUc;$<{;?hbWvuCY9
zzVC0mFZ|vBR#d6{^Nd0*HU*-nhJN9Ty}~blc)Om<0q(wQnxBxnd)?hiWnxuZZCei&
zX*@Cl`Mo1c#M^LBR?E8IciCa8Nzn8}i+QVHCXXzeu-D2-7PTAgP+~u>P2|0G8@cyu
z{{;k8bEHinoDr;72{;)C%s_cs(;Wm!{hvXMc2hMU;c)I``+=EY+o2v+=aG5hPWT#V
z6;2v&r%+bAodu{4HMFd9t-6dACo8CoC~&8-+?>_E{&xNb{nB*5{d3O-rt=%))hwl|
zJIot?shfABTRBMuP3+P_UChrZ)-SFAI7zVvcGgC5v@|za(f!s>aRMvU2<v$12)<F=
zILTV69<WolVkQ3Uo;JVSw+gUFkuB#jD)&A+y0{fG7RA4H2&cOfk*lCz&qogb0N}uh
z7#ZEsML2g5P()Ch3I%Lln5G9A%)XMW%!o2(4!4K(ny6j1B!zcgZy;jC@pd5vRzB%X
zrOlE9?YU#sj>`+c2_oL#`|Z7AaA${~$T^$UCfi!Dm56Ouqbl9MUot2-z6FexCVD)G
zuNPI+9w7XPAqm*?RnVvksnDsV0a&=K7QU;wr$b^b1ZUN^h>8FEEgSi>p{!uvCit!+
zxhLjw6HRcw)iv$_>dJ^{*IIg`o8snDE+-G6{>Et4f=xwECk<<aS=aI0qm!fN|KiSS
zef}@*?Bh$qfj^^pC0*uMrA4>T3xSU}PPT@_B?S=t>vf0N0AUn55p$EefPs1PBk>l$
zQy~ABAgta2HAVVN<pg*aNF01_$dW#%%t1p6L~d+j#Zx;0?e$D}t=+NSG&-az2h8Gx
z^S+V?%=7&k@xvP>?@@eaO(9p~_o+zJU(3ZCD=OfeN>@HD&`@T)hg04a)~?80BNT-9
zn*gi`se|vAVIjxxJq?2CD>)l2McLBnjV%efD64;gO=skwTdmf^l3Um!5`P+Ii$i82
z)>);Z&mm63MkX?w|H_;f|18F>^sJu-xll(i6RhD{0n_TiB<39X7Z)-wP&jHkmj;an
zX=!s#9Re3~^2`Vp{QvSGCEcex1VEa?!=4WUR*p^o3>`_R2!i{!{_6p21Bopm1p3I9
zjkcgty6P6#-tH+m<J|aDnm~AKOvO=pM$hMdj23RWp5%=o-Yd!eLPGU|)zYynGt8A6
z4dWZ+{FMM>(>l1$`Do4sz9xD|*CbIu_)qfkarzKt?!_GIylwV;=?DvO4;1>|C7Ni-
zUdcQvD}$##!A1%ZWWNq0tsXhrEPgFy@*W=hA6$Am6T*aZaCU&ury84w9sEcLrp9f`
z<^MbCWbiU79KkDw6Y}dG&ey8q?8-+6X%$+WPhNjV77)jrcp!Z+<`xOwuD2y=z4K=A
zqdjcd)O2^*p7pGQUeCqI1}?B#&{pqFbh&k@{1qw)qF8yCd&vT6`=4=`B^+QSKDY)i
zv^z`b@xVHrYUgE2)gHRO3xp^07Ewa44%Ce*{zV(UB)r;o-1$@Y{fIESUBE|iccj(S
z7|(DlHO;V0ep#j8Erbv;Jn$G(6#-9*xtR5bT1}_5?1ES05{>g8V&LfnCpd6e_dhdt
z_QDB@LmLn!)$sz0pzJRe$i=VaxxqD)vt+o3=GT3>EY@2Hb(3KzAr!{D470GFJ@h2D
ziay9*p-t`VWtxceftkPrCE~4-8j<f#a!*(Y9@Pfd?og3#pWwmJi}a!0Rwg^M-aRf3
z+GbAh7&Tf*7}eiLpa=pMITjpuN>Ru#(&06sNwUVE?-Wg5KXSp9BTGOn2SugvKK1W3
zs%cWh*Y)>T<-p+iVNgU^Pmg?MvLfU$guLoIR%7+8U_E=YPc?$IMDWPNY)!H=`)8`~
z&&PQ0WQn2-^0>r?B7Jo6OV~xn0;(?0q{&hmnVh>({<C|aVi;@u1?<&FMoukH<M-Ak
z&fF-MLbtvAj%dYW$xX(uN&Eb`+BY$xZ;1%N9`pZc6Cw+g(u^J)xkk|PztR?fuHc)v
zXR3(LM^wVPo4Am$4)c&Ps<+@MI=+%D2Qo?xzb%BQpuMnWIY;Ekx1WRMWw=ud^lmcv
z#0uaf06bpB83zs;Q(rkHXL?wcSF4)L*@nelA#%)%5SgdVq3+tf=6tjU3*SXr9e<fj
z$?q!zhJ?~;?VtY?MZce&&l#tR`4VVcD7(t3Esnpoeni*Fs?Nds`O*mIC;S4({q}h1
zb(l}55yFUdG~RmN4~;Pq_}YZc?i!b4=w0`A1S|=FCJajKZGf6AR>z(blw^+hUVU>M
zR!|tIAM728AvF#BflF1>fkD-DJM-r{Tusklby7oiSfE*=V^^4obsYyQ8*E?izl{o!
zzrC~?9=+l<o!}Zw#_GF_F#Q(zixY{Qi*=P4(C$BAkYvxhP;tgo9uR+;NK6u)`&r__
z0NvPUD643Je3b^#U2*55)j~CJVlx<Q+8~-9ZbApP-yvzcl+5(W_nH?<Jtz@8A=mAM
zID>ru8eNH9Jzu|`ykc?_-p$bqh*DwUu)_6zpYOaH&kAe*1PXF0!}L#;U8;F2p<Cgj
z?rzHI@`e0<MrnPN4-?&8f}d@TKA${Y0idHmcNfso{a3Wa!<?nZU4$&Nn~jcx4RM+d
z6Gp}BqnG{}|6g{X!N-o9{0b-LnjKLP{O>DlP*OJ@Fjb?2Bva~W8E|E!RE*VTjwvK%
zw=80(25<dUMVt%{N<^WNz1hG5P>}-die*_JaKmonQ)AKrU9PE7@ClnuxxM5C00dbU
z(0^4ei3^2jrp;J@7RztH`g5rw?X&3VkVh^M@|9F&#?6WIws@DoRN2ejmMHij_%Plm
z=wNFdV`(OY(!nI}=9ISj*Ei4L-^RqH?}ruzJ@qaZH!iE_>MwTsYl6+Sgqa^nH$iFs
zHyfYL=-9fuY0fm#?NF}&v3+hpOOSj^%~I5%ku{Ll1LyL!g8IJ+zqML8jZ}+}s=SGk
z-~?3|m7f10{64`RGkSP*%AV%oi}cuC&PO@2k1!(&J4iLO7tJuKi4uB!5cZ@;IoQrE
zO;Qz(=HogtyQN;dK=oej3{vhtEZQ$<Q9o0t266!X1(oR<OKzl#f9BV+NobOFZu%4%
zXKA79=w`Rd;$hZcK^LS$L+scZ?vffGNIzwSj7eq_ZszW%c{j~l6x{9$F(IU|*Xd+V
z7X})HEynq_SMM`@+~|Gp!#lT%#|E3=aMo+>>`b*)>{45=e#?OvupmZ=7|I^TwG*8&
zPJ;mahe=Z$+X~#D!p@<br~U5@(jw|3=_IR@11{=?E=?g1?dVqf3Vxg@dk9~y{^`Q7
zbPwiTaEzWeDIAG>;bumu!LZiLQDi%DXseW2=lAzusxmFsA;nZHzb@o#`~)kJo4rXz
zU$PUlWe}fD;1rDPgi7T(I&CS;PujNp6R`ot3n*yn3_l_=2Dp`h?$nd2fihKO<mE+d
z-1r(>awbr3D&&)MVgK=94zX+3?izb=@WGUIlt!&e<#NaT2<8g&x@qXcprhSloBZZj
zs{C2%kNN}VBe#hxL<G_yFN_lV^~hE}iX1$Ip|(@i0+!W113RrSY4z{*-{fTqJAMK@
zXB-Q*5qQXF9J8(&MP;QcinV!0qJyE4h#1;d=-MhvbZG3_Yit?3Gl_{Gq8hS>a$bq0
zTRLugla7{HFoy*}OI;r9l`s%U5I@z}M|{S-;vgFJ;UV1XX=eSISic;c@kofpred)R
zKp-0EUUs9?3mA*dxO7*aGJ;-&cD;c(;T-ayGtL=V2NTiLc*`<zqY(;gwoP%1Rm#gg
zxsh`9C>Y5{<3fe}Q}A;|olqH^yg+)Z*cYt~A&XLD+ff;%+}&Y`$3kvhA`)uD{bvDm
zbA)Mh0yxRU<=<D;^O!P{!$gi2RgcnJ+t1`S&2$X@;tt&y8%B@pvgLV15}v>e$jaEi
zh2`iRI6$?l07Wbl1WZpGJ4Q$f2ZdC%(7MgOOu_`s*vmD8QZ=k^&1*Frd&1!)->*8~
zPqKC@An=tSpeLZmWS4)TY0Kgd72`@u0!yu)QgbMZG?f^ifyejz^G9NphXSinWKQGN
zTrZzc7{_3)J+Ab47N0@I&MDyEK~@Pf|K}*Xo*(VE2?ENT)Kp${s+nh?_@$WC(HFds
z&<Shqc<~|aw%&Y2_8^nt4s(u*xnr=^(#T@!f>gW^>ke)1nBw8Z@_oeT&-$wPwiKla
zS>d$oNuo2|szE_R>F$9;*?@KlI^~47bZ-Et_~zr6Ul?+?)q|mpmOQY<?{W<cdZ^!(
zh3|2$K9m_8z`ekn-n$8;Yp+L#$_^o7{k6A5Sk>YE7d{cyhjDsu^%@QM&?Qije`7)Y
z@5XTc{Z%D5hQzIn226mcfu+yD5Bg!jzA(~sMvE>GFa*zod@up@QG<zYgduUKYrhV2
zBEtaOw>i-QETw)WU|;-v!V#rH)Pk6eiw&Jy^OWZ?dZibU_@G9|Ql{jm<jALAtK*Ce
zod{+IwdGK!nRPQyjBWE@e9VDeC~9z~w~vl?D~$sy!2`X6Ejp*_A6Y)79ne3qe>8Ao
zVlH^OYrtgxJhIJBGZzmuw2p2e@zqnvVS%cQ4~Y}7Hz8>*0p^&?CL~JZ)ujuPmkGeB
z#Z{Nd>25{9-Kfn+ehopzu*q>C{^PF=2$T_kKJu(eZD;>-Pdxlq+XktvMFm1qed$p5
zZ&&eE5d#Tk>tNRo`pgS=lWa32|4nKMAwTu(!cLg_la=<<b=9p+4pTAekNyEDe{HGH
zFr$lr_@y&S1JGnhL^scT-<noGgD>Z$<B-RWs!X>F8=32j&bp7)L?ry3aF8oZx@V>B
zU956-VAIYZz%v)32P}P(xwY4Osz_^wY=|}LHlB~Le$|}mS%72?^|VZ^?Q#TVB1ACU
zalnI?S-{{5vHx#$N|1qMP06<l<M|_059<C~0n#PZbD(UXjh~H3hKE4tC}Ndq-agzz
z%!d}W8b{R1QeL{a#)$F(5jA!PbIWB{TZ2jGJpGcKcq@9{j1m;%lkoYSF_K%%lU9jv
zi<NW{8!Jau6PXz4dtR52FKYVo!YTa4I4ww}q9M$HY$>y%aqtnlD=s+asEBL8!%t+g
zyegqaVt^-IwrK8(9%7=K92AbR`n`uV_F{Ge?_^fk`kxRNu}5N9(b^f_nA^69P44Ay
zix@eg;RnX+={Uy(NqMDmovJ0$%bZnxhe)&4B2u+yL8plBz7EeYk$a9Vr)-zAZUqHr
zIxTu{`-3be>vVUr)wXsglW{|Ht20NtoNcVvMS$qo&a2o-mA8U`U;(MaGuv1E^uXa?
z!)OlfK5ggyCp}j)$)8R6l7#LZxXjfYY|fzG$;ua)#+3W#U%t}UujckSj3^$b6xSV;
zSlhmt8p1LiyKy<Al@%L^bk2P>)#(VcS~;2z(-CG$u2+}>h!j(lgpmBSWZxOsB^)A`
zSO9T`l-&+~UVDr09WIeN76h3~PYMcp{Pxvq!S7x0k#X?LVV(s0ouu!3e|hXpE741M
za$sQ?&l;D%D{IFSglT0c6^>uyrnhmPF2UfDWlPL9-#pfE;^|;XT3wus1C(uMir%hu
zwJ&MW5ylCkMq4VfDLFPzlku(}mR)FG!9e;ker%u_eIQhc#ia`Q-;<&sr+ra+tB3k-
z7`E5ubPrx9DbG9WKVjBk&|z;x<69_X)qk<eT79cB2^R2lR#xyje-a><T|^(;nl0Qu
zeQ4QpO;o!sW<n{cy<M5=UE=qOcKbxD1=~NXVj4yh5>9C^9sOW-MJq(kNtmB!c>|6W
zw$~_Kgu6nDMRs-_!)EfDM1?Sar=A!U?jt7dfY<+%)k*&?@kf^&v!VilNO2OH?47^>
z<uiL8!Bpx}2ebnwgJ%FoIp2!R=;=i=3MqFO^x5Q-namiqJA6C&7CzmJocH&W6kEmv
zRT~n>Ta*^%Xg$?>%_M!hkBD>z0wu7}1V1QrS`Uxz%(ldSq&olY6!g0VQalb_scCJw
z<b{uCjffXU<6J#)bY}CNbdgp>bWV|Xq|yAGQOAw!M}Sr9(8i4INJdOmrAxc+o9?X!
zwLO5^(V1wxlQq_;p1&LDE}s?F#$~S`u|LQC7j~jx5Cc<)Oe3<~b>m2pA{@w7Gw3hF
z7vY(Y)0^Fmw~~Yo8{M?!20K0yjd+vqv&yhw(#VEYv!7<wO#~k^nW4XD5dSR-A0)!v
zK*LRr-yyl?^J#1+Minhz84D(@u2C_R-e)vNGh}GmcWPM5O`WC4Q>4s;*w`ylUb#w=
z=ONSNQCAw-2OFY-8<JQR!v;VHqJOn5&I4RVd?DE8SH0Lz`$DN?6B-I{Fy=|ZYbCS-
z1I#<oD6v6Y?z|xTt=hWkS4`D`P_np)m_Lb-Td5gSkY+T6!%I;1(j67(p_%;FxVjTi
z4lSw+XY?FT3RLPEyQ%v<iz1ic0{LQym+=WeevN0M?3_Y+z?yQ4@B>39Q6X&(X58EB
z3+g5eGv1rlY!j%*mIJM(JaypMlGkX$*;;Wb=0?JmY=QW~+3Ne5DM#@3<+(q8!NiKT
zRQyR@l=LrImiLUKFJNn81Bc*VqlI~3O9HWOE~@Ql8GfQGk1UQdN*Q6aD3KDH?sIcP
zQWy1<Y7)=oBiK|M2n9w;`T#xH;MO2}&_0gAEKcu?l`|?QzX5DW;vE9Z2F7n7q4ajI
zp$6g5{XU%Y(33cKl-|bkO^w<=mNTg=`i~p$C|T^CBL51%xBRW8p{TRVc-#~<%FvD^
zCC?|x8~+Ml2$iA#;};M8Fpz&o(2dfQuXP7rmYBC6eeUgzj0_-CtmFJ~l4h3lUcf#{
z=il(2xoY57pzZwUoKNF^jWJ(W${>2&@NCyV^qk$|W#8F6+12l(_NvFR@T_Qi`>9t?
zRaf;f$S9;qN@wg_=x&#&5Tz?;2q0v6oiyO^*-ENk30*#J{^BX<6r@^!l3VqDdkcex
zmRwxRe%uQC6AO%VJxMJ-znfiqnhv~apYbU^DTB)rA|G?i32mksZmx1x_I%b2tIod8
z$YX-2Lv-?l<hv`6&Mr_vNoD+&v#z1i{#AUkfK*7a(a(UG6`u_fomZa%6V2XczKo?@
z;vQlIO|tvba6;<OLIg98lB~)*=87C<FLI&1W3rU3gc%?~@wAEW7^cXa$1@N*b*@MB
z)XKXd&}%B{DOdjZ8h$447xGbr>aq%=+wx|XMJ>9TfDy*OY&K(v<sH5^gJ6VOC4QtA
zAtgfYDRuXz&Cna)?k=SFeO7t)ybcD8;kjz!Gb{TD?la=TY^H5ne2t&&O){5cMkhCh
zjU5E{84WPyg%EJ!b!>q|W-==I({?T5gq<waK8@xucmvD8z`l;*#vsuVl%0a@2X;XT
z>O0nPP2oiPv34TkN~#TB?1czzch0V4_{=63W&z&cP^v}#ki7}d(E<YnTGGPEmR|=t
zv4>!4Z;x^PV`X$}ow}}9Q)b@vNiX+>V0_{DsU8sdS~9c`Hh7_P?rEhqL6w?a%WUQ;
zPqf<k;o!eQoXDmi18KE}7Jf5}p+X|HY8df-TgknInsC*)ZHt|56Aje|rc0>5{VZ{m
zNdnsUY|g20!f1wnz!~*vg-z9O<Dt<zx1N7dhZGS^_GE|7o{?Iy=jsgj<Mn4%W$TZ$
z?jyk8>Fr=$Tn^f!Co6>Y<usX$;c3)Pj9edR4%PE}vTpG2q2<6t0QZzbW$XxE%1DTN
zDyv+)Md+pC&k(YDEcDiho@nqV%xIc2abwOdG0bs~nlde_=5!Kmg)68sy%uW01T}_?
zah)oS7KNc7uxiHm1K>X<N*yRT!CSscnkq0LN@k==ahL4uHW=oc%;&$$J07K%Sl)9t
z)xMW|kv;x5civ%|2K~ih=jF)5W-Yy+yWhU@+lId*ZIUDtRq{WP%}rxwqtMKK$6rw`
zZB$$QQg4AggnJZocwh{j_dXtN?x$=h`r(XN`gVf#mV`YR7M=_W<HUD55fiqM{wY9z
zw52zl_JU$Iig-9FmUy;aIV+ZJE;CxvYWn*KI_C?V=fCHG`>2xR@vUYV5>V}D(#V8>
z{M?xF4z4>0l<S*4-%w}UP~PQzVFZ#;J_KRW+`maE`o4$R!KuO%!Az{Sq!?tr8+ukc
zP|Wr*H@Pt}*|_;5qO%QxDkBJOM$kZi9(RKq3L~N3hJYkVh0ErUB;U`YO5YdFPvBn)
zbz9j6zXeAibT68Jm^n({jw2r24_o}Bf^o7jI$}J-pqg6)@b3enxi4UdUSJ~@V}iuu
z1MPe$4s)H9YWKZo`9NDmnBq`%ob9?Q8_wQ}-dYqvGc}I(QD<?V(j-ngS{Tp`1963#
z4)cQ%rR)!?UB*|yN>GY%h#7C(2*MwVS2<aEI;+GED;~G_V_w^?IV%XDtRM+){!sZv
zCLo%!25c~%ewf~I@*|V%U#PJY5#bpWj)bw<aleR#vlD#IP|~sezsyuQ-ebhX8#q!o
zAjn*7w=532kq&XLongL3fe~oCR#+5?KPcbYCrPtG?mWioINqKUh^D)PO#KY&cv#o-
zuQ%^fCstJtEfi!%r)GBCKEcFJqD9Akc#1gac&S`=CPyB3C;s@32Uo9FEJ#yqS;pzB
zi+2nUnXPrS+U)t-o6nr4I}K#%O-d)-%kQDHuBfcILM74&c)mz`J_D04$9UrAU=k`T
z%m9d6+}aGnlfh;7I3d(;?NpfTWHU0Ok+eO3W*)je=TO`cN`(eTE9$#o1-J~`kUsH?
zv(q5r`ASW6IW`Vj4kyUJG}K8=)V>Qg+0i=K(hg|o?$lR0j-0j?=u5*LvvaD>ZDXs(
z&Y!pOwD<-8hzSx~e1WtE0Yg$lyJT2W#P}nUNqcuys=G?tGc?NXopNLvpcaX@bOY2W
zk|7@-rm7$Ivu>xWtbAfh7Nn-8lnJ%xOs4bP4jK}1ueW|Ds0IU#xa0Y}<2YPXxZ<zo
zY+5Z}M)G`g*C_*03<9r#J&ssxSoRnAGUJC6OYbubRunGv02Jei%J`8nxI(#vGlL!X
zOzk`&rs(rQT$9|be?@js&s7fuj}>@#)o_5G0uYKKAP}h7Sc3Ppsoz?gXp~H#!BUDL
zSL2wy$<*qH+ul|8#JTGYGcgQYmCLi_@Dv5%%hTp{HCw}s3_|3et_&MHV)Qg#mtEVx
zH;$x8+~>rw0+z{J0vS+<UK+g6h(i72yu`#PD4$NK&1;B{aHK}^-kEyt>77re$B~(C
zjEqwEZB(0?&EB2Pr^Vj;|H2a>9QXH%1)1%VNYt*>t<FPpM-ZrqeZy0Q<bs(E)6l;1
z2mVPm@O%S>J8d2sOu0|iS^Pm}q+!%Y4|Q97`hc#53cRYMLzUV%d8qWVU~hP|9(Ana
z9ITvs$5k(U4-+@4SvDAcf)}s0IZr*9YKvjak)}vFOrVF@cI(BmxkrGtfgl94*Y1Jm
z)PmZ5ZgljcKO`+QbkuU4Mj}G*+>-3gqUV|V{vuP0@XwetG$Sl~pOfv516q85Y;QTl
zLoE@S5~#>`j}KQAp|+LqfZB`sV7yb&Jjjz0#jnB(bMA!Ehm;+l*t=V7Df~T{y+f}d
zOiU8jbWp^%Z;dZUu>lIlHE_7(h*O!G(l!vTGH|-)h#S+-y;2O)oLaPgORa<6RG4^P
zU+;EI*Po9-!VDg9h@1FchvG2RTYD-mH2nE#4Ri-Lj~Yl4g{Jo-hdq4@+{aXVq2KVO
z`L)S@_ikk09KACm3$}FO@(IJ0(-~K>(JQ(^xN@P`>eACSHQ2Bk?-1Uh90Kl#O~8<2
zO>~;WdI>GHVZp_F^yCA-$qke82JNwF;~>z`0Sm=ri^I&{%XsxXC4%z{0Z;RrT^P~M
z2w>A|WTq7oGT0^w3UedeJIMCZe)WEr9c{Xvci3T*k}LAfMm%KH@aOXVgCiYld{z1s
zf}hAdNa9TC5^V`I*pEQOISxCqz;BpIl6g17M9J6NQRFF%<VNxER)GYueehSf^9*J<
z3s5jI9u^RtIbc3FMIYrDx%q=}OwkqI=zu>+WloMpvb66@ed0^Xav8Hj=u1Swsaxlk
z`H2?x1fAzEtv9O5f`J5c$6#*92sHF>MKR>!EG%k-_%}sone3G4X_$JORLCgD{j;CF
zhT`~pal`R?$CJ*n&c;z|3vDH>3I~5vlgve_;F-|n9k4#{JK+8)a$Dgy<u^QDFaoeq
z^HtXAB3eAFjBqY4Cl)p=-5Y7H>saVQsa@l84Fj=U=^90L&vmkE1a*BaoLmEs=h101
z2=}|J63ed*#QbVWG2c_<p~RQVosAmGw?+LMsz=1%&i-t2gF{{U{+3FS%BsG+GCdYK
zduXyCtCXg+Cm#FlGuV#9$^A0WL;yHvJ~v)*?q@jiJtP0Rb;4CpVkIH;)&hO9o_S2j
z`KwKfUnp_d-}RM9bG;*RWSgZ_jGcFb6m7TCiDuTgW4l7!Z-=!dLB!Enf{6}z@Gq=f
zPk|RfwcB0Fh1JrK=w|*SML~h_2SHFaigRMEzT#|>#uI0fNL!gTi2i1r6bQ6QVHXTw
zQw&juL@P<ST+2cUGmE`Y#z&KkA)M+XC{?LY8-%lf%3*N}8byPl_>;&$fI)+WQZtLS
zrfDF*(v`%JV;9I>STIRMFv|%0l1D2U{dca&5OslDq7;-kv(yV^TD0&O0*wKJ@_&cL
z{eOqW)e;v0oSrKv2B0u@)i5@2{#>3&RH3l^l)})=YpK7j!W50Q25_w0K#0T;^M}M^
zY!E5QthnqksNWXH4C`s|+02UFic@S_eUt?}WEod~hH+!EZVkW+bZA<q?v@Ya9S%Nu
zY!Bc>U&ci(@sgolgiHiaG)S23eaOXU#C`pidVT<z@ugZ|7?5eOyoqvySwTIOs;8Hi
z5V^WkXc?YJ%?bl;hR{^tTps(W#dc(Pq<CGCh=NMpz4X2i91S>oclS{FEc|<rnE<}_
zQKh^j<eqIbxA*Hq93X%nkhqVnGs^hBr;9uM$b?5Eh%Yr6Yf0Q6dvw`dws#ZHnBsgI
zpa(2H&84R4AXGv#mi+&~N*p6h2;q9V?%fmp4lf-#TzLiO8kgIEUT}l$M8zo2o=opN
z?tGIH2)xt9Rd;U#LJ;}`<8*5-Y-K##*MqWz<NccDh9D{~z3ZmF<YPA<zTIBDVy_?v
z)p3rgM3TabV}|VlK$)Ei3x<M!hw`?Og)nLe=@}%c6<j9%g(NZs#D&8TY|CJnb<n=x
zs>HNgVnahD{X{lFLSdRjh%)_3Lcf)73Xxeoo9|=w?gMTW5qolUG82AABB|D3ZyQkQ
zkX0Kn8fIyGfsN8AyX-1D2^ZW!C$XeuK8y^y(TvHlPull@$1M`2pPqmttNl+Gg~HhZ
zjHY;<`@8qod{NqiyDw1`YZH{ZPx@3;H+o{z;tBhRWwRp*hSRMLZF<fVTd^Ds>xxyV
zDD;98d_jI^6;{8dO?(zL6`I<M{E9I~rS3;I^N<i31-*m}Mof}RJZ20!oATe0St(&M
z3Y)@(87WV|t;eE}d`_w?FFt``u6Rz$bNvptLF?~`yHTUTrytgPAfkdeWDn)<_{tiS
zod$K=OhV$}5++96SyRTQOC9Bak77b+%Ul!R<KG+UkZ2FG&YQm^o-c55i{qot%L%#T
zrGi{m0_@tQ8E`x8-Z=Ye)?ii}5ANEY58I%B!EX!#^_M~Hq<%q7rs_FVSKp9Bmmf^v
zN3VXCWw$+9*ABXjmV93*rw6y_d+}5v6x7Ir)}oHO8ZoO-c66w1SVDVutMK@VMNN7v
z8LecpXfD2ds3yyE%x<nMOxPGU?2k+PtV!$ms$Zk2Qo1CCGxocJi!Tg630JM=2DO!_
zHOILdcv0NbMUH!Bop?L1N|KI+P3@}_y>s_u`!rp?=+fgEEF=JkJ2VLf>xmy;GtAy&
zZf@6Jz+UuOYECq#V@<<g9Y_8ge;(@&z=4q;PYs$F=Dqa04L1ZRPp|l5Y^=nyCK2=n
z!>hKlJlcp>o+ou&l-h95wAh9C#iyRiMby>+Y^|-<zqYA04EEFy1}eD<$5Q#x<HJAc
zK9($`C!fUkGuarY8QTxa7cOcAH994lp|ydm_o*b+g84OL=uK9h82Ho3%*JjRg1?KL
zHG9QOg^Ayg9pxmer-VqvGkn^Nc)Td+jCs7yL55w}iVK<r;DkVm3&g>DUIS+ipz~J&
z3gd66lj!)1WIuETNXy(@99Cp5!t>equ~+Vcz8!zeC%_H(HT+CYwU*ko_ue^A`IEXl
z?ek_Zan8tGHlQ|>y6k@cszTCibmZX$N=Zjzh?Cw!KtWd<GQm(pOi34#^e?prJydnH
zw*RXif(f_hHEeZ-jDj7v@Tbh|5BZ@9L_#{qX&PbjvR*+x`G^pkV8Wz^_3}mETb3e*
zriS-|$;u5FkWz-+5;CHqB{+RIGZN7-poLOq-lZAXt3`klnHYpA1FJbNS+T+T-JaSh
zrS$|hUfCq3F`-uxY>J?2Pq6cILeF5vYxs)*GI#>}VoR8`{fex~Cd&+LBRpz=p3EVy
zzrBb4y2w@^r_tVjxF`=;FrS~`>hRR)ScMmFNf+ah1b&h}8r>0DCVJJ~9-R6y4#wN#
zeUN%if40Z<gdcdsms7^I>f_dimAZ-9w8--zK-LbK70kDq01UjHQ0-7R5bN;0In^=u
z)~B?;VntHFnNK-9-Y1-{iN8AlmwW+^1GZ`2zkIB{fe#Xpv)w<nqm&liJ`LB1!-5!C
z^4BObEe<?Jzh?7c^nShM$*17bHkQ7M6V9MTjk&*z=zI<6*_hz6n8a&sL2TII_~#aC
z%W$y6r>ZP^gj{+c+6kBKjAmQr2nWHnkSw@6CY(}JlBEfT^Zy~pRHU;27{7nqBd)Tg
z|H%6tdV4D6Mt2#HXS=xQB%2`t({#EP0;O*??(7mwCxOKlk6D1NzK6;DJCtW)vQlwn
zN49#ERI}}K2b#K8R=fEjBJYDcFWc${i!>=JS;WRO!oXwWNhT7R-H%MwMq9&%QIFek
z_QZ9)yLb9)BQtqPynrGe7*|maWua3hNy)n3mj29ld{nspM%U>zITvB;4E0-29Gq>J
zM`mNl(J0E#@rJms4cbM(jpX%H)-~f>ye^6kRws77(UjCqpd~Fk{*U;_jz2LS`cVV_
zbbget*DVIv6gkJp>XWjjo%)GEex}o4>YH_sx?49g7EY<X)zzO20*%_rCChdkQ_}mi
zPY*H?KEK;84hSvv<EvaA#jT#!^o#Dd&$quY*(Zx-)Qen^sZ6jbs#pHW-O7=pRe~`t
zNs8Z{nzC#cCnRM8<5RubJSv9iGl9p$DW~<f@F3WSQ}ZKi-&YWvllg0Zw~X7<UMreM
zdb9rNm&Z_30gtc|0C_n0M<njw=aczrV`A;w)79fVzGAgNJ>R9Gr}K_i^@3&MlFgwI
zw&Q8Q&L>&w`A%c5wTmmhP3(1<-TA^h<v{ki&UN!bxQ)Z%s(J@6g))0PN1RHKTv>%w
zp#Jc%L=`%}I<`yT_0`&#hJ-nw>vO0Vg`08I&-?29I{QWwXs<h0nt<@lCxg};OmJOx
zTBQ9=<ItBp#a>}396U%y7s5_=zmJ;45)t>hevdXF4{j@P^p51lmq9%>kY97WiX?n>
zd&3j<K8jqjd`|X<_XSMDF8F)BoMZ1AxfG^Nj$mZr9se)Xjaqf2r}5bysdosmr~o6n
z<6$ICTB6HlKrEMw)1a40*DyRV0r}$L=<wxH96IuHW8p{g0Mx@^$!Ujy0<&caPNxV%
z_7CyIJ_X~k2}CIENrw~vYYl>P1v5EJs&QWlGK#YQsd47e(GoJGF-f$zG^eEWxHJ}p
zH}M7Wff6z+|CPAA3;$Uu>B4wrQ8EhM!XHKPcR=7)f<^vK{PmDnuPF?#!7qr4(EfyB
zT@*s^QoG-Bco%tNSx`2Gk$SbMVt%@-de_eF>!jKHpvfMNrat2t<)L&NrbC1rnm=HM
z4T2WGzG^xFhCz#)r4Wuc{*YR%&e7S@{s=Zhb>M5ZtvK);KR1Vjn&#q6+YB;=WHEIF
zvjNwakE15K@9a94x(uBV8CAu?9!B$vF7e2CwG~$4p$q=lj2F=KGt=9lmc&SuGt)U~
z#t--4Cb>D?Uq?EUCFk-QxYq7kS*iTiCV=745uf&p($putYFZCr$H*5ozhC&@^WT{l
zHbirSlvT@=p9t14(%nE3tsjD2$C?Y0r3kYgb`Tz(`G0?t-k6vm5+}Wck0-vCTJJvN
zlvUkW(r_nR&!4PL?@!#H$y2uoOz;0-^4>@{USE0(BN84PLhQVAC>XEiPO^3NrLy#D
z#_~aXvC7u%UcJ{?DmeN`GRDuhQmhFh1KK@M`*K4LPF+h@tpR<vJxV(A|0;haW6gHc
z#SrAvEXWTo!Kn5gJ!=1g*vKjwH=Se%kUZse$uDf)gbeXunDScVs9SOLqqH#4M~RWP
z*iLJa3r(@ZB&i!Oh9!yqr=tgBmjP>fBz4BDP2x2_Uy-5oRyu}|RYuT7`sT81Nh-Ux
zAWByePNW$qU~$OtwfyqGYq|sYz{|gmsFQ)v!i_QyIyxJ~mQChLZO6g>djdyM=qW|F
z+{$kkAp;9_h`c1Ng{G1w<0Q=r2b=dx{kVxb+Q>M$0VsR^MV^(q&TkUjV9su&#}@-6
zYN0l{;qUwAr_B<wR!xk!$20fmb%1P_%p~EP^P;tj_vI80_e&N{i9|``FjAYj3oh0i
z2Ddek%yUz2uR8d{P>P6xaBF=^BoZTbn9js$;5YHlktFiWf!+|5pHgxHmmcW;AaVi`
zQ2qntIS`Uv6|rGXGAzoD+T9SAfBvVim0~uV!<bhpOYGH}{#xx3+ARtKrgE5S!&wMS
zz399xhWHyZ(HU50CP7`U2gISM;T16}hWPogzqC1nc1ToJY?`y_qPPBgpo5~B9y9p-
zA^D)oWchR+t;IThPnilb7I?X~{&yg@K{hK;PcuH$s}tMuc|JW-V3FYH(2RS8Q9Nkl
zLQ4MmAmYM9;=P)j#u{Jwd&DEb#66~BN)gR|gl=i|vO{9a_jP}!QZ6o_@K#PpL02^Y
z5MitShGc|5HaA*axN^X<tzZ&vbQz8#nw!1adTG#Hd%awrj*cQVxmSm7y=0Lwmol8t
zqd9EwyF1>Dg1*3SgCFwPM8XBhTOFg(!xQ(jeHKWxZ_Vkhbq=HlASpwij)R?8Ef9Ur
zQQc{Jv$zN@QHgm$V49HR2w0j7XQA-kOc$yM5QwQYDwkP}HuAkZCWG@g&(<p^$K|Vg
z<^ORUX4-SNd^U{x1u*O@7rdlpX8Ce=U#JB=?OAM*^38UL`*+|HTWl30;aF@{WyxYr
ztoOxYH&FOa&RmcHY9tO(a0hCvu~0SkAURFEm*vwl4QJX_{*Kzg1h`EwahxC!i0k34
zgUL*&Ql6gmMHdy(vc1`hEd|qNf8R^hE!70ce@|oi-@LP&;^e}F+@6AUy)cUzX(1f+
z3Uxe&BiETG4NaW8xU~wrHmLtDWOubw2*V7|BywCabMtKgsy2TQ)v*qU&t)j%^nsFl
z#k;hb*fi+rlqpJadA-tQv(UGmY5D5p5J@9Ggno@?nVi}d6wAaH8SnhFHC4Ozu_Wxm
zn=WgsSj{@)8e{_3w#_PL`hVayeXbW$LXa_=zV1P<4AE9wsYeJ{H5rG=ZBJaBn$9z;
zMCXD<7=L>KwiZ;gcOHrYSGP$PIR(L)J-yRL^IbrRYhDcEvLwL7-otQXNepHLnMprg
zxi4I3LQc{w8{2)R%eF^TB(7OABGubhPo(KnleWxZKSUj+d3Ax=wxlYl<2uR0?KO?g
z@>fnWP49T+*QqKp!~K{|k{^ojjIiT`O_j%r-HywETm8AwQYtv=IEsull31`Si?+US
z$Y>lpL!qYm8XJMPlLSj7wu2u$$`UndO*b}}Ve{82_xHDY6^5K7<grh=Vo;I{d_sex
zRg<|e^_P=U+AT#gI=X#JD-{f4gtXww>n8CNM3Y6NSR~rFm=u@@`gBBAl8Y=DSS4<A
zRNf}Q_}9H2liTu~QNMA_ui3*cobBlSR=|Pod7~<~3{?1Gpk;Q{(qunYsK-w~`ts7R
zier-cmTMG7MnKpyx8j^)kMSrR@2~T#8LM+GdnF3dm4Z+dF5A+eN!9YJXBbs7GEqjK
zFje-?*s(?@W8W=d7@ok`FNxi+l?zDp^l5hB+<RH>V#V8>W${_7m<30Pqp;JIrM0$O
ze?66U?Tjzea%z$vuhR_oDaLq9&Zf=f;B8KLf5#YCqH1n^6l5h4aed)3qswwrlasa+
z0Q6}HE_;X5(Ejc!ySK*d!J^)lPz?<rW9rA6j=R~enPY8n#;%+b6oIld8E<ZkH5CIJ
zOz>Nk)gV14xZgthe#@Y`3ohtqgL;k8WZ8D*IK2lQA=uA_y;$nn{f{c{!?Nl{kro9H
zjd!FW&M!s8_FY;X|5DG5+A%c!hs$Ix#L_3Leu#$f-|{SJLNQimxxV1W^FdUHkqDc_
zsmB*j&`7F{L(zFO8%Z}xj_#bLFkqcOKF>#Cl)fm`n<a!w7`fF_$##f?K7*vGMyb>_
zwhxg^Q38$V2ZJ6qI|?BTG8Unn!1gG3x{}Gngbh#Mvb6}*$+w;%*0Q~zJK=i=m8G%u
z;l|QG^snZl{qJrF@&Bpds^r$?8)o}e@3@~lsbekKH4vJH!pCHzz{f1r019bK6+||v
z*}WYh2gErWZG!M$9xl-)F|CcoJmpzrB#sZYP5R}N!bVpHFyHFC9i%+MzJWxf(QJ~F
zi-=;1I$MOQ2mT6)2BWBE&(qd6{Y6r2bPO*z_4&@{P-s=@)@Pk7f;P`rNyz>y-=Ec+
z2l|&gwQSGrktF`k-@mEg(*W~}A~$WdLF@!Ww4d3iOFH@J@07)gw~&A;A!h|SA%n-k
z0XGtT?ou^{7p-RulpO*Q2y$ppior~5KfDB|CYy=oX-h3KZ7o4$#CEoJwgduM?+H=&
zsWFlIs!d`OjYjtHI1~{sV!qnl;q<dd)I|d1@2o<rS00=@u7hr_&H&wN&0F!8afzd6
zhy~PxI~fzAW8zM_D0JnTdcq1`R-nu5i^_JOTO%K|uBkwiWm!52??NbU&gh6k;zV4H
zTWER_SS5|O;qmAj0S0xXq!5T3oe;4DF_Y{V3xVeIUb3NaH4q#Mk7!5rp+ug19YElE
zjjQl(Ff=C|j<!Gn0|XZd8MK+SxlTwq)3!Jw7#T+*TD<L*$F8~GaZdilQ@XP)&?moM
z1rkv^%dRxWw8f)VsDGmIV3Ky`feR*%tr<ElR5YenUi6UO!e}mq@+HCCy~Q0n52A=g
zwhq4bsP&SZObC@bat`3IRodyfdBwOLN5ZRs*{-er4<ZCk&svWTV6$VtklvF(b2~i;
z<nc0A*550RzMz1Z`dPB>rs-)<?>jWuQoltz=}k-MSH|Ou6uSav%1QQH3Ey-YU!LzQ
zjRK3dxGTpFBE8psi25C$Po|%;sB2=k*;?~xaL$XqyuxT}ckP7Y%J_Z|B4GYIUN3$w
zaalh^k@SFgY^PCNRL%+=@~?VCJb^ze>f6p*JNa-vksBc{dBopnSkIhYu>TvE{9=dE
zSM_8i08D9qCdUKED&nI@U<66YB!eX}#kcU;hG{qISuRR{ItTtuI0Ma`&++edcMozV
z(}g-ML2E?6e2^e%kWTba0L)a=PK^{F*vzWm>I6`6!ej3}f~RzZLw3pLGuO2@9zFWj
zjhv~k^Dj*&dkXre0B2R|GCi|G`>jHmN|^7>##FnawKemnqH&ZhDosAMltmWaS*n~<
znv3tOdY{nP>q)?;vy1XVcC;|wYIhrkJjZl`m!GRxs7o^XY1o+3K2uF|KU3O=Y4XWs
z(=&iGo=QR3K5OTdd|pQ%EPa2sH_m=h?7h`x5vHs)uc{uOraw9Kn0!#uB;THb<*dGj
z@kTD3)Gr&T>ox`C<jfo>EI`$9j{FnhqS)2d)2rWwoq5dFNyIGy8wnBu^ZJVH(JYv(
z%De_@BaBFkW~~~%FUO}i_m?G_=dUt<FaikBojtwfHBr^MF9<GKql=v-9#5;XO;d5W
z$;fYW`0!rPoGfqJ5!ysVXp_YF;T;9L^b~iM^1B5Ue+ELS8!8e}>vax58&|)^-fXOV
z0ut+(w%S4<Rd5pj5m_*w<EL^z+{^Nmzm$KeQ(XU}{LB2wCGbB(f)+RY002!Y4g@Vb
zkuSe)cTma<3&$W;ovqM7MjJ+xoDej1?&}qqkQEV7|B6=PWGMF~Vdsfkc)Ej6U7W(y
zWUB3~8kGwhOd1>DXnW-)gp7RwNrG`c-!~~U*tqK;MAtpV7#&WU`6J}Vb+k@hlxQ%&
zB1t07CX18Ig3~V64o9Yh2ZV|5q58*{rc5G8<j=`%T1LbAXO}sA!?v2YIr^sk!(&cV
zCjz!=smNt;U73xsUH9~C+_Xg8E2V=fgvcX!`o4Eq4s72OUVx&MT3F@xI76!y#}ozO
zFEbEsd-p;KPS{K78L6%0_~_nYlZax1d^oI35)=+lh{_x)K@|OBAs|KyLnE1lr}#v3
zL$o+Oj-gPyqWz~kZ{`DWFgU3navJRwNs#vljMOpu1xXN`%W#2IWY(8D7X;;7YuLn+
zYziu_Bc<vBA<LEUMQ|Y3F4t%(gK*9{cYxA8q=ddY{ydzCgsr@pjr)N1W1F0rJ>K(z
zkyo;>1Dmmtyp7TKIzV3cpX{Q|VE%=<9~~%uu+lVy14?K&Dhxi1os$jI&So;7WxBCG
zPEo_#oF-@1oJ~&If-&UcRaBB92r)%Nc5P;p1@>~o#Mq2W%v%j@XBGlTB(-EbCkSYr
ztu=VR{Rlgm+6M-VGarlljXBtNUnp>t)<v()kGNq-co-(68o*+=-EBe1*~PebQ^!0K
zzg-tC%7i?2IQs8}kVBo^CEj}N%IrHCFA{R&W+guy;34Z;j4HTARw?yI{Hf<!{I1~)
z@?uu4CAa$Y^__iU^7}V$3xOmTo?L|5;VaIoQ}jJ7Y54M>K4bRb8T7<bN@ls?b++JR
z<Bjd8PbiWkCV;0_@K-TPJrPgPf!aLHH(ge2_0obQ#f99`)$Gz;IWKS<U)CBEwSrKU
z$@3#{<2}!=A7>NvHoSi$ItB46?X+rT|IR~1km6Vc<;_3@fto}bks&S=wm=wP=dgR-
z?x9FDDB^HbBN&9ef0{MVP)(abqgPw(#Ew0mHhw}T3j%)(m{0HM>l>w7K3`tO;bz2a
zAqa#aeot*&?b3cXJ$VWCBgM66_~yX><!M7mpbdwus*lN3+Qnv2a{wl2KzzPW(lIq5
z&S$g}@~s6!DGQmnrE|k^6!vah;_u+2*GK02q!b5bT{bc)n?dp+my@%2-<FlgE{u!7
z5%;H+|2@zLB>Aj@dp7p)($tb&EtAq<eU4{y&#Kleqs;b^da)`ET%7oW-rEgD@ffsT
zZgZ8fk5feyo#bmL)-}rKfFbhhPO+;UcZEkStQ`<=$NSm36!C21WRy}S=AaM)OqIMV
zZ0+-@SN^`TU&0!dZ=v=92&Ne$2sfcJh=-%P`oLGDNR59abHKe=ORQZz*4`m98lD|n
z(lw1zqhF5J#S-5I1Kw2(!H<#`kq8|nnb2RXO<%$>@EBc>@l?x(V~aDi#0!qidoB}&
zdRxlJXRE_A&^8x~U=T2tCQW#(9h-67LsMN&-^=(v_umnNHOOjZ&3Z39L}nxL8@c~o
zodQFFoztn7k&Zvb(%+olha}GRd5$cQ>UmxORDT#DWV1h7%Au;Q_d*ctsTP`~9GHm`
z2sHOJ1(!&U8ArKG#~I019NXYUFz^QmW+cDSdT>|V?3(GAzozr%UaO9!kf|e(w7&yP
z#GJ*2<VofqFN{RZeke-&@Y-+W`PReqxB*+W9k;ekUQ9T~>qB#$KXq^EC9AwP-HxLh
z3bPk3r0M6bZsdo{D10YaoXo5tb+a}RkJqxY7cT!Fy52cBkT3ckjBVStZ5tEYm~euL
zla6iM=ER!V6Wf^Bb~4H4^Zo8_)vva;`j39q)&1&qRad`z&bjv<j;XQL)wH{DL(I|x
zF|xAd9LqZ2)poEXJBd5Ux#Om}ZAc-h)Lx7Y8ooieWS>mP;iF-#u}!OAo8MVLeC$~)
zX~p8M(IfN$Bi7+B(9e$amK}aMW9@SB=TOef-CAOvDX04d2iHjJm$!;Bg)uf6ly`By
zy2#g0y;Q{CVJ9F|Mr8$AYl#Y!@N8;t>eZ*ULDA9ff?0d+Vn(_QBF{;OJJ2y`Cu`H%
zmio9KsS8#_OW)@W)g~T%R3%4r_=m%eho9&-Oe-@$vjQ5OiOT~u-J?vj+w??7rcbs+
z+t}XSk4yFQI(8-2;;ZL0iJ>%zxTI+GZ*UlY2Srr`=keO#<Jghm=!3A_(8QelhS_Ag
zD2!3o5v)fj3^1uMvxV1w;^9~l<KLR1ErL!<^mN*TOzjM-@xwAU#i7J?z9_NOY7#Vk
z3gcY>@r#kqf4IwMXE_@Fxb4>Xu*=C!T@?l_=l#<ak*1?+6^BvWn}8I?Ak`@)kyP+L
z3aa$sR7)PuKT4x<7UfTm+19zCE=f+TwC!@fT%VdWV&Z%TMC&N^V+nb(v4nwBM(mQF
zDn*BGZ>HOydNPBFk`Q&}i71HnO?5mN`Uyays45v3+>*q<!DVJRq^Lo1kTYyEPjI?9
znXrGKr*{W+gmZ0V1}8?#a~jGvlm*W)kcYJpmUvNdGdSxHG%TkPA~62Fzkymy&*Z*1
z-^8Brq8sJ{2f^BVkM<j)e(6S?7Z2xT!saA(7=}kTt{61mh3r)n3z*fY!zO6At9*dR
z7#fG2=H+igi9NX+$F3sYdfDbUd_P0r$DoR_nKZLPdX;x-|5tI6Z8o}AtnX(^#1ln`
zqqffO{A2e|I2Jd3_vOx;wRX>|Y{q30#Qf@bR|=E8{X+nlCx<0R!D*DSl*p6&Erup0
z`0p(4<m6-v8HAxmR8OTgEkcP=eiiWVq+Zl*?szlL8){xXqTTq%Ee#NCvVaD2zTY^M
z=-NlwHR6i#FY1b)6@!v`c7lkxR-%Z=bg;6$8l0^0xec#CWJWT~uX@rRGZqP)Q(r`#
zK!l~6vU;P4xM<CDNe1t}nCJfS8CQ=?jG@Su9TOXpCp4PjxNr&XHS5EvOae?TF~rCV
zrbhz!W3lfoZ3YJVd>!p{hfjZ}#IR{T@)JYev?2h1!;JC^uy?_4Kw)6KZ$o-{-ZkEc
zNDquhtyu~k*gnE|R;{qDsNZz<9ve32$x<q?a8*B{D^VaKK*pFD%E?jOKw$prx0`49
zp}5l_X-H?xL|8Wy@>XK`?gG+1Z3~u5f_bE~37_;&s@-W3*4M+13omL?miaf07GhGR
zfB8xYtu4@UYDLOC6R7W1--AvBVfs;m`SWA^ufl^iNBV~9mw2$MGp#&tBaJ;>E87@G
zNZ0<LgzU}<R%ENM;m2jYF?#H}v3TX_v0rOTo-28!PUr5u<EUPZv;zKKpgr>(&+(Pl
z`ajl$i8i+Ryc*Zaxy}~D2US?+mbbPfKN%T1Ct0i;HM%}lx``1kCEcIi0;D%hPU0E<
zRgf@v<M0f!_@=$V7GHU&+Irkr0GE0<EUu7qg`(k5aI?{1%qJ-S)Z(@kt=xJAR-rDP
zO!~SvRp>Ku!Y<TMyMUexp9g}nPqn6N<>*(@aqjPxX7#`4fBl`6o|5o8b|#9F3)ByT
zY|VZJcUs6L^35ay&3|_6(7i}rd!4zdZG*C1pi4<YXYB4pVE3QbZQVL<S%k-tcuKC4
zjYTL__QzZK{l{@R&zR0<t~{6z<?{3?gUEHjJK!1}Ga6z-vmYRzB2Zk!doLe}tE8TM
z&I&C89$;wHFJD8c4r~TtsbY?y{=NQ1<7lWbx`rhP`dd&YCM|6x#H%tGi4VChd=SKo
z0QPi$Jh~iaC*eEflrdT7^dM9S`gx2PTZl@aBaNwHPY_f7{VvaqkA$fDpfeloBbHQH
z^Ydf+<8-2E$^;-A2W{;u(pNh|H+2C2L>A`zSynd`=lA;kaX?EAI`c0$ab&ob%|un*
z-_2sU0nBZ5zo35j;0YaqjU%<(BLO5_&UcTABxO`$bjR67g{S@|hG$3~sx4al2ONMt
z)&|j)1aiUeHt3j!U<uHQ)vNr|Rq~Y6vF2%{mbLj*lLYAA`Kn!T+1gxapIPU8_O`Y&
zMEm+<HcuI_AJ%K?;Y=h{c*kDdgG;(gNh|d@o-z<A38iP^naGd&9y5hwM}A_~d0>14
zy*l-NMI_E6cbf367a{I9$wSA_5zcHUkvjBb*-4sCAv+QDb3>TI<3l1~#YyVs4g_A{
z6C-vu)qs1xzt`1?7ez;Pow>yNY*rFG{>WSa`HjbG?+J`!jAOuj3-}eK-&d_SdD;~n
zP)sseuRwm7k0ZZS|NBRBpe#(_pn>#tc(^eME8qSn_Oc~~^`i?-cR?gZ<<Cl9r`ph3
zy;AYwmgdNomnMZZQr%28yf}&D;%3_v%PxAFEx?z~7_W_|Jv&dG3bwjjV@ll<Mt@FI
zP(LK<si9ic_(bC?EHMo9o=Dk(gpq)?P^q)dnL-}@1iLkg;N!B!4DJ;YBcT(H<1@|!
zMWGKc3l=FL{b89QI8<Ufjw16uKC!iYvXkjTxLOKkb!KkikZ|M)=fR<(!*Q3m7ZOl+
z1<0J8x^y#4vDzZ(eEDhA*{JawE+Dlcp{fFJv-__(k+J53O|mTJ@voLYxTjB|l;g5}
zr^sph3fsQFfj0kyiaJ2)=6+M2nPDJlK^Iay5}q@ID6DcxAy#q6P<G$Y-}1dQ-1ZYk
zy!DbA5l8icszGmFvVl!3+4)F}&ef<80QixtRQ${C>Wn0)WGBddWdtjs-`(GP_Zo<^
z%2DsWIyrkN{S}J|#-la(lDMk1@M20wzZ%;rMm&!Tl~<}o{7exfVx3+fO)?R9C<#ep
zEVQD$w2k^l={M!Tc3Np5*}1TVJ{Y1(Id&oxK@s{3mj9G&LDD-vs@Uai<{-&m7C2$b
z(`Ey@rpq+WMi2{#jvnl8BWD_CX>U*#8q2L*0=$3oMcc??wO$3B1ndpbbMJh0{Vx>U
zt{(;^@12l--5;qOXCbZI{#$p9o|7b_SdK%-<az$Y+GBHlbu4`Bzis>&aiNeFAFX0A
zDI`X!1->3|b*<;3Ox=4>kgfU@3v9SLUOCj>9kA6h9ZepbRI*inc2At&xizmTDMHH|
z!pa*4m!-1eOY#8;AM2Zoici@-_CG&5f36ys9k^#9Fc%W*IaXmG>0AJ!!fuG#(kCqC
zosSak1{((M4t3r70C}Of^3i3B_vgv!<D=Ts$DvSn_*ZRd)_EkN8{MfmQ2I$zqgH|R
zm!aMnTysF(hfryHvbAtmwWf5frnDX~DqE}ABS)v|__X%@c>PXDEQpDklJx3H$q8c6
z^!<fL=<i)x<H~d|1s+F)Ec(i-PF5?aAUp(SysdnLps&lmZuBqM5WuJgo&}G(opgnP
zxnZt?gEK<YeyRKqDKeuIaF!c%#=L)9`tVQ(o#pk@dSEv6ir`~5v}Y+)7)h~SfM4V-
z<`MtmX=>e(W7)cPR{yG|y)+e*O4BpwCfuoT#-?Y>X8fifAOHAHw{f%L9O>g}Tm(Cc
zn|_@0&v?nv^@z}~Dh94FI@Rx-p6zbGsV5@oNaYv|#afztQEONGfsk7^MYX3OgIj0o
zE(6w~9W`}qG#4d58xND-!ZZa!V$WZnvDxnH?rCXS%W3=|u+fG+f)tTD)C~1p@<J2>
z?loZ4WLGSY^}A{}iX-oXUZJ~f^VQbWkgOPWnsC-=6p084A#@!<;eN*Wx*#<M!o12C
z5x?hQWDsmqu}JWU0p4=$N)*##8vo?-Rs|t0QgF0s{ZnWKU(!k%{Jp92y050P$x<9M
zVUbFVRfAMa&4@$k>nFV8^eStW>djFd63MR{LIxPo-ZjeG9bsWTAC<nJOm)U7W1=TN
zg3h_`@!3qBx%|NQ26_;;LBZjwe4$~%;pWB2)qKJU?XxLE01dc^zsaJsc3+HVoh8th
zv$mL~e3CVAF3Q3X13rJNto%kG!~7gNd8hj(0Fg(4!BmYmc6T~&P%mjFn-NVQE|Ftu
zz<b0iJ5(B?6}L)WXgaAYSm{m;$vIwVN*s}i5P888h4u5EZ?hn_*|N$h*D@$c%Z(ww
zr^e*Z$fE-z&?M+gC^U;)=&yjWq@791GSZMk#bf~1$<1x#FuM3KXWr!_v(Zf{Gr0WB
z(9y;q;LCLWjfRSeev1^_;H(yT$ijVek^A=LbiK9eOde%?y~k|~@x<`+-|68vXH({%
zS#1Z|iVyvJo2Qs!SM){up=j_C#v!a*?l<nYy$#?8a0$}DM_%-&`Fd_0X&hqr3~sPM
za>p%Y)Ln1QsfJ%to?s1?njN>5U@2@nfZ1+MBKnk+!}(OyL|6XawJ=oF=doA7W~e6D
z$4}71eByB1yV-}5%7a!cY_`B4&-4Ar%9XAGI|$J8_C)xQ9m;M~(w*fAtnbMwfieST
zX`}xE7spXx#H^1`QWU1=jxJbOypL($CN-EESnk6LddSw~qV$IGE%7JJhPC*N?c-*N
zccXAHHdMZ`(1fcN<!QCnA;X0V)Dd!bj9{L4<eDs%6EGu8Ai0FKrc>Zqyt&h}LB+8L
zOPoHB1}C-QEKiFSY0l#1#BN}TxidiX&<)E28tXO&F(Yn4vr1D&sistTJY!>er!_x(
zXu)(<K_DMHyH&s$=22y3X5FMIB!8=<tE9KkD42>1%AtQ9^H{~QFlg);EUGBzk5Avq
zW%7gpK2a9!I&LWIyu|uxE2J*!-OanCJ<#mm$OI!^zN?xT;uzwfTv*E-#^@;GdE`JX
z#}&!xv{j-14s-3%Jp&gXg(~al{>60J?%Q94jo%8r4|Dw~$5rFSL{fYEZCe-S8qQQ&
zT<^R0A7l3-Dj%U<Oj?=9QJs21tg&x+2OMvdx6r1qk%BD@1?Wi2V?!%c<5Bff7|4?4
zAeD>{?_!hUpyuu7WUig}2d=rT-PM5R+R#phD0rrx;u^yWBwx99G~XUmili?Myz6x@
zA<RgY?5~Beg(9!s9sTq(c#ThqHAU^oQ!_U3qj(>Z=uh>C`oOHie}BMNeERD3?3d>6
z$ev<GT=tx6Iq6NbZZ<7_PBCnbjJ#!2C65^e);Gjf$RXisYPsYZ@-OV=iyeU^)^YpF
z|JlQ43oWT}U$mF(s5@1}aIHG^el_u|ZxTlD*=f<a!BVO-DE)QCj$u~<mhr^0KX&i6
zpuj`!jIuv#=lVvDY#M2|bHjpvn`f{#smn@(Q}cSxZg@2Z#-}U{Wy9DA3sSEW^w&15
zXLUNTRv4MzLxCn;&_}e|0K=hjyJf%FI}09a4k*oZ=l?}(bb35V1Qa7jgA6Gs?bfM5
zzI7=)<LfLuggByKnltfuXzrLhPE2&HFS?KjaHIe7F>F7fLb{)L+B$Z?)3$AD-a~)k
zB4WT^D%I7{%diuO&qnhf;orT(p3ynUckyx^S&Ri-Wix)dULKnk0gXVD0x0;KUbF54
z;({a8-jDB4PsrI`^Y#DUdfSQL>?jYK9lA88ONtzplBeypI<EM)1no?{JU;VjCyPS<
zDeN{mH?A>B%Sn9v)!Cg^4zgl;c%KPSgngnKDW2c2OOtO8swUlo-tvumE%lq;d|=r{
z^k21=At36O<iF)61s(?AMVNWYp7bvBC_iQqLZ5~P-aBeGBL|iQ=Qi)~@!<njDC>i|
zH0Q41tYfCjiPDidj45z3h3XZuQ2&&z`_%SN|2k@bpj4S0bSpime4g|XQz~b!<P0YZ
z|66-~g#YdjWo9HWeLk?4I~p&<G8ZHlJ`45ZL*56ECF$_j3DD1qHIB{<4UW}SOu$Sa
zg^;hk)dE-WJw>W$U5sKh0)mEF5gMWM`xQY?>%abbvv=q9B%9_+jYb>fJV+g82HE9B
zud)Pr)i_2^SfV$B2ncp!Vj6I=)Mow!?jmPd)tb3J;~iZk6<NKlRA`sf)xC6d+OQjw
zzNW#_S|&c82B5P;_+p;~WgogfJktm797i^(L+EKe=WxhLobQr4)U|U%nUI*b90VeG
zk{ilI^tbgVbn5}!QSwW^zpPfGtX??gVB}@=MH_c!2Pt1>rtaCF)*D7~{zN_@XBAXD
z>~&HuGpKrkk%P<7G=|`2Bn}nn(OUBIt1NK-sYGS%G@xyMJ{v59|HZ2ChSr=TOF$42
zx%k9Co9%CZ%r?GEfWZ5v671ZCw}NE#Lc`OViXkVD@zgO<G5MAn(kG2?T%5#rw99MX
zK;V!zJ0-(X(Y{IP(42~oc3LJmg4jc!;$JEtYI4SY9;mb)NQ*4nh*C0vjgS+J;7f-d
z^_V_{41_rBFJyfN)~MKGl_n|emz;3)#H_G*(V^A5&*yO9jEL6lhwx_7G?Lv=*vIf$
z-r^Q2roF0cscymZ!NZ2y4PFux%zKdPL2XVMxbce<Y3RndT(pn#*OE+_@Zb@9`v0O7
z<lLn$JX1@dgb=2^t78eTT-#jwwy2l#F@|Qn2q<;stP({t_*11etN99jx7Oxw{Gmj>
zV6!NtSrw`C%cf#Jq4=zDtvauXqA*Ga;gK+ybT+VnCGxuFFx^$o%#&7iY_+*<kAJLU
z^xZ=Ipd|(a<razTJXUZ=7~_XHz1U<O*FT<%6zJu)QNte}uVz|JNFT(ftG`?y`_H?+
zjRC(Z6|wHybH78E3`HWC81&u4Cb(s0{}UN&&yWl9VRQZAk-jW+qY(VM$VTYgqgr90
z-m3NjzpsnOJKf&B++u{|@$Z$gvg@xy2iE_fe#s=6W-p)mmnEzxy=(LOXJ3cOUb9S^
z>}!Td!I2l%!mT(*&<8NTvhUZVn4qLHzzH$(r6XaNj{WtR-2(;Q(SjDnTtl(x;0939
zc@x+uAJ-wSM!hj5B8@*^e{>mT)J#8v(q%1yK!y~;#$2$_k+K$X-O@^#obogHoBBgK
z_8(n!Xi1;RUBgwxM#MwbJ$k}lEhERsPA`yS`pv?Rl<*lJ<}0&A9*5<wGMeGP0h8BS
z$vha~IsK*hikDn$?8?jr+EN7kF5pH!SJqooI}S2ero5Iv7xds8SQt*V<JtttE04$3
z{oBVlT~Q#N+4${Dg}Pph+gclC-T8GJ<te;@e))Tk$8!r;ejNPIss7o7?B6WE7ugWV
z9?y6S2WP+RG+LeoK=yJ`rqq2V0a4hSkY~6X3w}`C*@$0Y<gpb(pIy1^0Vz0h_|Ot+
zr&@q~>Xag6`q~nmE?w%Q>PVDR;@^;fntFNGD>({LVAhVSQ<o@z+I7)&?NLb~c<Bo<
zVg4%0N(0fW;*hK*B;%(x9~e%uD9<%Z8%)ow>JOq}!iyXjBVfnqlopBvywYK8Ab#SN
z?-jr(KGthG>W?O`!ZqtMmI{b0h(#C+BmW$8B6N=sgO!3!U1#V^7R9YMKE@gEC72{o
zT7dshAN%`dD<|*0R<KaGAV<;s$|DnTf~{fqkXTv%6T<*wlVdgo|K6(7d9Iq-o;Pww
zs*?gbR*`=vx-@n7YyplM0GQLuOb3hR&-(J;3*VW<x;<-~JK2U@5RPUwAuxERG)f|r
zaCU!(^FSMtOrai0j=Zeoh<aj*#*#3Tj0(E2WG^0KE*!$#lYwpi?J*fx8c0deM%{t=
zp;MMuN=$==FCtIzx1Iy_80H=ftm{Z2d1nkWtB+B?SSlVHb>DR#(6uy*kdL-KPA8v(
zk8$N=EieKBBMdN9$$twdP{4@WhN^{U3MQL@N4Tek-~;PA7Fk}Ea3U=&r^5{95&IpE
z;?7WVB(hyq@Tu}Gh0p8gzl9iR#G<Z-pPQ25s~w`QNd`Yk)O$zDlxl&gefD)F<B<_2
z7>X1NPm=o*9VG%mndLNXG;QAl=@I9rMOZMSR)dVgJuP9ZUqCO|4Rk_HWiPD1Ni9O&
z?iWBjZRQ6i81x;-S_Mh-Z;!}%2?lBWRIxFlZT@oK%>0EzonQ3!1~!#tQ?r`OwJmwp
zG`}P(Ya7QqepI|fm$|5WpQL7;7FK`ESA?phrQ|&BsC@-Ynf0wSs>#=t7LsWEl!Odh
z3L1_ElL?Sw<s&FYiW3;+d4H-3xMmlaZcMP_P|-VU%5JY~xWI<$?k2CPnto|$tAP9G
z&oC7;WU|LL6?7kK_2@Auphk_XfJ_i~{TV~^`3J(}rt!3UYw^n@B6`6WN?X<)R*UO+
z-RU|COR*XF1)2`X`NZ045sMj#mk~#ZdR|sbUl2EHg&hjKG?L{n((5)-AxBN8VsSv(
z*dSg|uupTs+kk~Nx)snSaa{G7A0l+Dx&20M9^L+UOcBS?aH?l)GI@NP2ohr#=Argc
z5q7H4%2SCOO3lI@gw?}?iiLOo@?})?+=&-jRD=ic(%789n=2mnvvzeNa>GHG=zIcE
zkC<+BP_6+~6UZk}P<Ga7Gdon@<L3z~0!mr`Nq_q9%uAwD{ZFE8GO%g6=m_G!e@Fx$
z(xm$#36dA2_2;7sTdnC=XqSF^sWtnyVUUN%^9iw;l>c<`d}{aheX|#rv13iu9$mF4
z?HdFLtkP!uZ#gvSHWE->TF}_r$8ITdy-Fc9bEiv!wTrE;b)0)p<oM)XPkFv2{@Jss
z@l5!Nsu*^J)?J#k6SJp!)jXS>_(b(_pPreOgM6nONka4=B+*yd3Q{o@^UDmcChU!R
z<peW@=+W7;t(gTd4tJ1gORIa2(+IFFjEjL7gd}RyOseeQNAM3mqP%WC@?G|*2)D<&
zN7Wg^1Zbj-F`2VCA36PLPnH-K4!k<(fZ@<FXzOFt1MK#x(%C23v7ca;&{SHV%7bQ1
zMWO_+Jn!(@&<RunlR=6rqlmfvR#f8OvxJOD8DmL37HMdq!RQ;6LB}Ort@gh$<J<sc
zc6i0&Bi_P<j-LCTr2)4mHs1OOh6$fH6aEQ7;3`BYt`93$X7gt77e$n1YMy<abWq&H
z;+C;E&6wLL0wHF%M@3JA+!R>jq?iRAhW(qjR?C=soTXhTYl&tq^oA^`4>(UOQXAfu
zQ8J8V*l)4eRt6$!2-D{=Cf&VV^)dkUr%>sJN#jbu#N$GSCFo_<uRr6zi5><ZUvVNN
zh@r0v<h#Ub33tOW(U)(08=)&mlu<gJiAWV5VnW70owU4b+h0-!WeEPx2~FWea@dCF
zcO&yD?vB3eP$F%GysyEdL)3;2&ZT6^bLGG!!xp`=?9;Wf_mP`%sluA0GHwE!1{9Y}
zO7;gP7tQGLtUS2C(jN%CNQny|LY(>aJ2Ki;b}RVJmeB&1v^i6H<8J?$pLa~~C#Uxj
zp}S6nMMwRSRnZ9iDfSQM64O3&%VoGl^ri02sq#sk8#`*qYTUA-A*jiG?U&F2Uv4-z
zMzpQPEe}V#Fw3F<Sdm(5fSDV>wI=4r!c9`b+xl097H_%}z1(L&DXH-`Zey-IQkNNB
zgZKLq_E;)D{CKKBsCu;n7~FlY7a9%?-ay|0^p3Q{tszV5-MzQ9FZuMzGFt|#mR+Hq
z8F;^%`!5=%&?C<pihBW3v;{eXa!^uqcCY>759zt~$OZW7*b#a9Oy@a(<Wur}zi<_m
zDXJ&7p2c`pUJg+?>b39x*t~Csz(TuC2P&WWn)@Nr19>sZ<}ne3H?lCK2mWO!Cf#+e
zj2dw*KqA(-&6V3}V0#=Ij|z8VD>EjWhJQ6mVPM=$+!b&phd5F$I@tr?o8nC?FPtQj
zUf8?#HUUb@7b|tU=n^9RaZ|pZiI;D)6jJ?_h$U4<x(|@=n)_SIuc7jl1<w_~j0;j+
zgLPqgO|!H(T48XMJi+T2YxNP;^H=^m+bE5z>&&dEw6u`4)%kEJhdyl*Foz!T&yeQa
zV^U&4`FaIJQ;kFb3>2krNSY~SioCK{-|fAiWgYlwPhIy7ePs<ghwAgUgQIo`I=}fs
z)=UsOzoTFs6t&y&sZ3c`yUk(wHguPM+o>EGsEIC;>~%sUMo%K7UI;curr1LJr8s<c
zEytN|5qSb$*%G<$ZGec4&P1T#d?C->p#Er2YW6yDI8<{3MEf+-BJ?*}5@q<|4wS+P
zmkhu@{dJ{qi$O^^-bQtZE!MG+KBRV`qZWgrO8P?1B353(c0SU;3)I{=E*tu>zcxQ`
z!e6u|AwFya)l+|BI5nsDWIrMKwe$#@U6NKB+<o?mtjmezuEdz<aST;81#XH>mD53B
zT;ZEe%>omX`6=Uq3zgjUe)q2fOzJ#-Srq`eGH;@aH^g<m>JPP!uA^2}Of}z&WxxXX
zQN98a!}ENwFyA}wK;9MfZ3oSPsnhp}3ZyEZ*A2dYESg%E9d!H$V}5Y`dBV87AA~}j
zd&>jA5prIm-<U;v1wxw0h+~7^dxc1ZdP0D{xb7+9a4G`DXE%kmHM`tpJv3UW6*Iv0
z5i{y>#jyH<G4qBZ1)H;ZeX2Z4MJ5M#wgWXK?ZOl47B*r;WSy6YRofN8Ym)nuQa!ia
zy_f364tpwa@PNI|IxzASfAPt$>Z)lmk?gk4dW}ElPqTVQ)2TrHDJ*^>_VVk}^513$
zN&DO)BqRAS=^#pIOpq}5L1YArLJ#0w3aV?27Gszt-(<{x$k&hh3RHM(@=Rw;JhpM`
z3$8QKt}00|xbL)8xxmngHZ78rtPv-rVa(~&5R{k+1H9AqK5PX^g$H283@phIBa`v^
zw+RI4&<Ekm?9^RBir54y8d%yISj;tu9G~!<`H_9aD;;AKl^mV=EQdJ-SpaG$ZoL5$
zUf9$4q=%df`)*QJC$2WGL$ejB0c*XxgcbS_qEgvk=w;TwAukhzr>Y&V@m^6wi56+u
zs~41H>sHMTNwGDgYEi?@((?`ZM_BtTzkYR|C|i6l;G)mva`}pm6qNq<qQV02=;{z6
z;Up4b;R+a?W#Muea>l2^N`O_OxNj*x%c>{Y56BLQ5K2m^<n^d)jJ(fa*0J&vgN4#L
zXDEr_?soWppXo>8roV-H%%_Dh0>VWQ0-o<1MSQ&ANp}1Bn5w5$X$@@HcZ}-O1z|#~
zXa~*@TP-z4OTJeosrXVd&e_a>ZJCO;`62NqmDdPAZ3HIW4dw3|_yRX@oD0i!(#rVX
z=-*SWxyOScJH!?$pfG!4b<vv-iK?er3`BF{_yBCzkKfGMWsgVP;ZvLmw@<8d$uYwt
zseO(_tZoAMRE4zd`_7G#wqhi3a(pTSBpl~GP?M^`;E?#yJb&vOU~Pj7lZ%qEzx~-m
z6g9P#{JZbd>NdXL#sUn>%A1_<MZ*7F$UQFCLEu=_ZDdrtoQifS^k=}9cj<EDj*ezf
zD4YH1!_GoH=!bfk@oIYZ{M#Do`{u1-dKG7RMo_ICzB+ev+qXLnQUIG0n!}U_fo#AB
z3sfo|l^sSL?ne8$_=$oH-r`$E2lcM`!D1=i5e~audi}QZ?`r_6ln<_HDRU$xT3w7Q
zK=b(cqTcv?^)7i#e2^BF76uci%zCijvg%hn5#R^HxbqpJzwP^qu{k{*skBVr18VzV
zn%qw`y}u{z9M7PhREa#bzG%~#L-jgFFLqy~wo6J|Sf^o2^UAw>LfC%tadCNC>aEn>
z!uPq^aqj2MIRKD@t7!1oEV-%pLy<UQ{P=P$>#Cl9_R^|y7XO7gT=d+N@oe<}Y8lwv
zMnXwboAys1^NSZ3WtaaZeSKJs%=~Rg969bs9=n9hU*qOV=_ALR`_7I)hi0Ju(=KQn
zI$GViZNr5^e^_Ao$}!A`SxEEemrg?e<Isu)wCG~Y_C0C|J}oS3q4Cvr15*pfdnY*J
zAG5<av&A8T-i>EQ0_LyYDy+k6NeSSP!z@7ij0B_@DPb{C!@OK&rg6i@ejuCBz`8Z`
zH`m)o`tJM?uy^Ra2jS}ToM$KhlgCH8P`WhVgLD3bT~9BuN<@=`%ASnEL7GVh!n@}W
z++BFqLtO@c8#gsW>j_iI`+}Ha*66=N>1?)p?}FMz@thU(K&Yt~g6BI9+m|l6TEHj~
z0uPW%E`@dTdi->$?{ogVN9xurw-iZl*`UX8z>JdYK+2Kyy^Or%D^hIoE30(uX|h?1
zj7b-!HK{3Tj6;0zG7Kz1EEg(oysrrYez*d)Y&X0!3|N&KlpR4v7|mXP#ui1LLHIE+
zwt2#t1|u>2<TMue_6f2%&fd#RAF9F`dw1L4nP}h!jRriM210k=lPfu1mPN1*m0~tC
z&=;N4DFj_A`>$ZLbZ+A^NtiC7vHVJOLKEQ&T@ZfrxgMl5iX(5MA!pD!2nccmYACL+
zYlZ3?8|}Y-%RUPov>k$f^4t?Y`!5Xn6GRF7C>%b%-L2akymT#Jmp82JqS~XPP1iDW
zeXRTH1l8!>1*=Dqx<xUnQY>Q3mIiUU>5bN)ih@pz0@ajF2Z)@IlNm=r_OP+cB_)5x
zn`f{l%x@(spB}0-jtiz`o9y5LwV`zgmo5A4uiRAwqeJ*ZvddKKobu*W0Y*-SQ)uoD
zDAqV!qZ9vTK`9`d*&#h!=ev3bHuWvc9j=0Eamq%?Asz1?8FA!L6Qy|_Rl}w|R~6d>
zg>nWZHexxk=G5!UfoVO5FGgZP%p!(7JL_cG<gupF%M)ue_`H`1GodrEbCE3Bhw=Vr
zhSJ$_q`g(3A-a0~=d=yibscG^Ted-jB(C{#Z6kfXv8J@!1y3i|GJApr|5U@#fu;H4
zCA1UTDhQ2!=TyCUEmVW;$SP-^To6K1kcm^~eD{rf2Uh?)t@V)D?%YtLK1sU<5hnYo
zdT(x<>YE-01qQ`E36}*hr_jLO;mto?t<iow+C_L|_(v0Te>L`w!Gaq4)^5bJ<=1TK
zGr@Rp;W@4BreZR-C$7#hXX_S*r6HS^9tP&_Wq`9u*uuvT1#?`iPEt@Qo~>+?Uj;Ei
zKb)lTJ*nZio<>@TM6!5-9)`hJ@{{r(VB9+x2@P?wjwuiNg34t86z7f7BE_l%>x<X1
zcArHVw1Ex-h~n5d0(AxrAJ%UeB5*HBm_8nPNmLX5gY0Rqplb3}jH!0f{6*bh7M$|r
zc*o+BWvLpb=xJ-ArabgnaZPj@7lllyCBLG5R9P<j9?J#>jDW>%AWiGnpZ$DS4}boB
z_)G3y2z&f)@ryQKR&eW^w1<S~7R~j4I3vh`nH58V_lay8A8mjZ{OF5x?bT}R+2Or7
z9cO-Igg0v|aWdBZIQgCjeget;%hS*{FKj0%s6U%5N9Xrz6Q#tg_BSf>5tNsq`r2%l
z#A4I={;H94&_e=l+XpGl9BKA^OP%~Ha#a_)?&~`V##7*U0O+v`iu=;WZZ>>fhM}nI
z&b3YQ>=D(3THNK|5mrWjK4%Msn=-3Hn29b=-@1od@GI(fK)5+NbZ^90+4iSck-ZKc
zrI$_D6mPkKG&BG%j0yXNwwuk#ZL`?zuTnxrzHe?nePQ3b|JkN(#9uya*ExL2!e6TV
zVR=X^1yCc54Z7-kLBEM$t>-k->m*-)`fu`M@`LAzI6DY#sJ`W}!{)?3%mR#6m$KHk
z>DzuM<3-z2P9f!eP*GAwv>uXe#Z~VZn?0G-1BzGAJx+mOR39lP!fqq<p=wlg`qez>
zeYE9&c;|Dg>VTfrMFy1*f&aWi{w)vDGgfpM@bSnsDEEvC7`ApQ*XvshgP`~}B2AU8
z;>bcI8WEfqJnYoGRhvt~kt#}#NBliM;nMrJ`n)C|s5au~bKGwP@;FNt*Wo_ppNLz;
z?##Hl@x7aR><qS#^83ie5EYCalD<YwS-ioXgQ5#tS3B|V%(9(G>t;|B=*!qdzZCF6
z0H!Ja2i>DpvS7GECqw%jD~^Y?{vS+q;li1Z#1V#kFrCD=yr-ZK0<RP|Y!ZB8NMm`&
zqXs!PH8~3lN}BDXxGG*tMMC_7A(RvmT83&84=8E0-$H9aTz=};$}nmDX57@XbkO|_
z#t7V1E9m>pQ^C<(-O_3<a)8F&^hF4y0WDOd?BZ$h9c2t3K2H~^O`XLW3<X_>zq;B_
z<9posTlH0MK8LJ?S{)o~KN5<J0;23Z!AXdT!LU~i`5lJz&DG-0a`-M`#-)$l%o7j?
z4(|m1hp*SJRHV4{NLNCsW@2iDX+u-lNC8pTeXDL#>328kZzb7kj$Ar+Ek*~f2CdIx
z4?qluP%3ToYN~F2IQIINj0aZyry$AS3>`?Ljz$K~&sg4-pl#{I%831H;U}ZbV53o>
zQP9JnQvHA}`S17+KSu!(feV*Tqb+d%B?MwsCOiy-T&S@<W{7$z4|e9I(sX_afVZj$
zstDX@1MFB@ji&#n{$~+pq(6P4W9KTfGnfszApid_0CI{>ry%z5bOCIFD9Jqqx#KB^
z12NPs%{Dc<DSGx7XJGQ_eK!9|fYoSC@h!*b*xsG3nrs`lkQd91hLw=^FYK8E#ceX%
z!qTmW%|MQTN~i$KZNh|UIzG>6WvCc6vVIQ#mcl_2&sn2F>CXku1bXbH<o9%iEPe`N
zn7tkYQ#&<I&$;p(?qIidb{*Q#%<Q3*c)V|8I^vpiF?;&$xd*07B}3aE+VLHoj=~yA
z)9nMiry~m3oV(Gb%am|dG8b-?Tsfz~2=XIijR4PjngDf(SW2R@!Bo%srNvX|d)LEB
ztvr)sM~&Xc$zxrOZDGRF18@6>vo{<AmR;KFD0NMz*?)4JAhjn}<~kau{QT$bT!Nk(
z?cF9gZM)!xuSYGq9N4(_j~4l(cZZ07^n0;0<!194DhGwCXLhv4!R!S-h}UQA&Q4ix
zz*K|#Y@lQ2kIiasK8^hsglV-AX*?2!@mdcgTq=Ju%;7X>{*v9*4wo$!hK@}28a3M{
z_tt|yWXQg?xD2J&_`?bjR3SsVhpP(LYtB%mN;MMIx>zsr<O~Be{CH#>5)Cx091?Q_
zcM45)1phI|A&tVZ@R_&_@rHc<c#7nJl>HF*&h`Inh^D_vaAVN$C^_V;lV@ly5>@B^
zi@c~^65#&~H4IJ~8BM9sWgD?RVAEObdd1>D{F6EY&4G3$n(-7l`m5*(lM*55r*reu
zCZun%Uxy^Rx{z>9MPdf&Lg@S3_ptx?<Uz@7WLb+COYZy?JCs_Rew=VTAi+9zhM<5~
z%!MM07i?9`CHN3#5;c>YL)toaYlR@jfJ4eUdIqO}Rm?1S_&6pe!3rsqhgZxy_z~tJ
z!RIkt-uK6VU6AL*vmA8w1?#K04=SU^EJMlwu#Ge$Q;@I*`{%!tu9{I1EuF4k&<S+p
z5Y)z<$k+4M)HLGW!NAu%swNkzc3xgn60o`YT&>hg0Xe_o1$<_Lg;f-ujS+zE0pi(9
z?D<I5$%0IAPU^aKMd@0L|4?gsh+nHj3;aI^4jJqC84}C|lG%upB0(|lCiLWjuhZ8W
zpFVpaEA)1Ah;jZ%+&tSeHwh%qT<`b+P5k(gP-{7x)BxhelwN#?WanM58}tRzTF$DI
zRA=K`@kQC@>=)6^N{J)xXdPKdZyh;<Q9%1grr)e;Pg`y}W8!(?3ka$s;~YEQSE{BI
z(1~kMh0tZuxGAGrxTW+4e_3y*Kv6@xz%fo5eR+N2J~5Q47zu;}(gmpp0&GFXtUuL*
zm}PM*fxb=ZnroYD*hP{`QT3qp8O(Ahcq}u>kn-R^$lx_-87;*_{fAPh+_NYJq+;R}
zMWl&ZjxP#Yj{&O3boo{OwO}1^|2ycnY#F(+3m0pURpB8I8b7eMn_0=Pr%kVqs7f^E
z6`7Efj0(nrWkl00Zf%wsm%$9KY_@#U2goP|*~!28+!;ZxXQWL@-;+^*(RiUAOXgz-
z7zRgo&`;gmOeF07hl$$Me9$f2d}_I}FxyiyF13+%bPx(ETNXR;ke`&<jVn^4pp=|%
zLD(%nArDCCE22*0*OVzw5ittTkjF;uhjt6uUaz<L^C>bbQ_~W|Ll-3|Z|+uro3DJ7
z_66kh*TXfk|K#_+pzZVR^Ej8!zV@IYUPn>UFVPqbgA(-s#pYqTBJiJKDB%d!PhZxk
z4@C|an*UpjN3h~=W}rkOr!biJft{X&V8BGab%2g$@;dw`HwSQqDE+sG%0TA?|FVpX
z|KYQrZ(3pLnWm;s#|;GHjBMKI75t=U)hx-si+&T^Lu~{nBOqfEfi%PQ<Om{#C@G}b
zD>W&X(&HC88Lg>DTgtD{TwHn##AdHlaFzohx(c5*60s@?nmU;!3`_ifq<t~n_aX}$
zkd^Jq#Y)E=7E2bSG|L~Ky%k{&9!~)-!jG$Re{46ss({?6%|DsW|AmJr%1SL}FHgQB
zollpDkJlffYMIhM(%VF{H61kd!ek;wg$3lvQ09O1$O{T*JJ*e7!2GYmA2l?A$L6=Q
z`Bd(B`m|9VdI$y5xE#+T-~&&5ttj*F8f*rB>f~SQn^gofaM$gT4j7hiW4Bx;TprqZ
z839fAe)K2Y5u@gahLzBV;D|V|p{2?U^OnK4CtP_|*X(wg@nUjh7{>e*v7XmiFTvR=
zdH*3`R~azs=)j4M9cT`mw_Q(5VxlLB(EES{A~QD+_8_C(dH>}hoF=b}wzUw@Y-DPR
zy6_)ju5tnt6uMg&T_vC@yoNZ>HjnZPK!aRq-AWV0cf+uKV2l{%bA#KS-S}_&z|Tb7
z|4o>J3ffQkX&i61)o<8nv~NJihgi5VPIF{@a(Y5=wolUF?Q(Yzs&4>a1H~J?zkAu?
z;{9;&%B^S4)#AW(CFEY`D0?*iV1IV{ivQyL@8mP_3@hMj^`FoA1edJPGcC~E^>#~_
zv$ln1WQ882c`&MQ(1DOaBJZHb{}Tjc#{T<AvsF_tk)<nJlGC^e?*ZXHMJ5+&5uJW1
zk$bBbR*<+2&Y%_YL=hG)>vkFV;oVsuxraQlaG{>Md1u>qCXpud%zl>%$aLJV9D<da
z4C+I$-*SOkM_~C9SD9GTVBo;jXt?uTEx(jA|25ync3w>AP`D|Bno%!f^ZJagjnp4{
zcrOx)q_~K&-HV~AUX~wCk@QD7UNv{w+_*#4(^WMt8G-Ns$AvIjmn%zQfR)no!kNF<
zSDtp3|2WcUcB(p?=fm>VK%&L?VoR93+$J-bhsk~7tDLqXndD3U<TlVl`&vcDGP;#x
zTBnHFi9N<yZB5qtv=SgJ<fuu9*vBYK;PEDF3higf*`}gT+lJja2?>2~x;?e=vZpGW
zr#O~AZY_R@?G6VX))NZL!}`=ybf#v0*yC4$T2^uUi`%zL(L90fl}D))?-iL+<NMOM
z3f?!p4NEi^Av)saaxMT5chPdv?1p2a;ODDE)7*j4v=#5qDtURK#&jZ<y~X}4;{tQE
z9A7^X@tWL)Mme2rs!!g~6x1R&F@O33kg7j7@EY3tsm6&^{U%#H2xED1TW@ca*m~ki
z-`YpfSoD(N|G#9qim4y58b1pomh5Bzuy=D*pGGH${WVLY7whM{Rus3KbMT>8<JP66
zE;Fq>hcak>FcONL<R8f4b&3lG>%2udX$LOU9B=Z<ACl@n<er6h>v+&}Q2^uN0y^ql
zWe=Pj-Z6Id%oCYv%R9-UC2I|KQ})B6?&0EIB5Cl3o0GwNQcz%&A??2XU?B{MB}bq-
z-^A?$&Ew_&=hs(*QVZLy{-#;Q&19|%fT8FyoWB*|lwRrH(@G0wNt?sFErd@>>?^zO
z`-SU*B@ef}yzy))bVOGLmQCt*6^sqc9~9+Ns$mtQb;a1GwbI2X`O2ppMApNl@ChXj
ztd|&8U<*W8Dl%j3Pv!2%$QS?&T`UEKb8#dm-w!=@@#J4O#wXX?U12b8eKBv@mz@LA
z^6)!4theS-QuaYV;>6jFNx-l!`mExkpIMIXbs_$&gInla)yk@52HRisQdzULQtDTo
zNU=V<{kRw%FLzCZBKjFt`f*4!)^ZorU$MJk)2QzJkhN*a7s94Vw)!2AK!l<vOAdB1
zHBJ8~bZ6Xi2|f`k8<CaSy%%FWL#t}Un(Z`gS^^TuKFI52(cqJmQXFll4g2<a?u+QM
zCe!k0bVBJ;W*kc6W5#CT7|^%6pqO2~H?h>+sHO&nSKS;RPGOhc^|rlznHt;&6C6SD
z%3)P?xj`bdQI>bNSvCy>G`cS48w;TvNYyx(r|Exr*ggagi<KY0L%e~9k)LO98)({P
zgq|G?2~5B~T-p(l-s;gw0#cUod4@u+<qkFpSK-NAghG>GkHp*a6|EoVDqLUn;G4rb
zY<A(e*LJ*@JqxZUPPKJ>!_To@K8#Xiv`*Kx+1)yx%>{0nMnr(dzb5_?H#KWS^U0cO
zT)V`Ls-x?>+0iLGf@|}YXs+=r2ics#*?`o)l|Az8=lo)uI2*PfPFY;0-^OM2J+abC
z>z3#?A5&Zfa1u(lL6~gC5+(2YPi+fXRPsx3)ZmmPhQ)`+GjG@?MX7^Fq!?jTL`8W)
zYx57RSTQjD3%}}5@ze8<c1+8cbh;Lms2(&cAb151Ot+1b_N~CplH=|`=F2GmVxY^6
zx5$RJuoi%FG3iCh0R-MX@pJXEg1MUim>|70YlxKtx`PExIS(!B7OG_iCT;1QE99Fp
zEP^X+g=ARsvd-?l*Iw%#xZh@03(<TUk~l8qsy|tK1ZD%IDY1qYU{I9obi>a1@kz|@
zT*f;}VZV86xmjG^T$>_X%TFIpbt;@p%ZX35A+sQZBK@6i{c3&=a%5IF^-{sVd!y8L
zIC{HRG=wDRv!uqv%Fx*Ck6yXiPE|H5bu0&>kqzMsP)mGpX4#zoYpK!(nb%`Yo7bDP
zY-h+zB4z|uLWp}W01j%!VYSO{I<QIQ$b7KUuUA=meTDVa#{!Q=jHZQjXv)knq$cF+
zF5>L#eos|x456SkF^rHM9tQt8gz`Oa)8j=D7~-8da49N_6>KJ)oeSxrlv2K{5|ngy
zT#YOI*PnBVS_&EQ_f`n;UGauhi`>ZWM$dHP+cgk`M9Fu>j`opL@Ykz@6UE~5z1jNC
zc#_KZw5_;@M10$Od6dz6o^u$>XeNji0c-Nv>P6!NTV|`!Y|~64f(Z3-Lhm0$&#V*y
zpXDHde$Pc-S1__N$I1qthJ*ta0$t0W7XGYT>`P(S5!0DAPb;pe&ii#;sHzEe?E|!z
zF-PF&A^DQ7)}lVW?D{S>?%{>8{iwZxo5IL)d!>f2Uh`19MK=b|e<cZxb3c6>5nXfi
zImr_LlM-S1ryeUS-K*Psb~5*RFNjqK&Z0p_HAlyH&Sdyec@K}nNEn0GAsM^*CgKoa
zh}oG<Ydr3%-TkGP8J^(ydycuT@KQ-4JplCl!6kUuNj_3J!XpiXY-i^saBBOhtV1ww
zz?n5klpksZJyF$b)3&!Er3mxm>w~0ig;aAbCd@*r-X3@r<dlFtdulm1s2&Q_>7lN0
z(=qUiPDym%&P$WpH7x5|rq?KITmA0Hg>^q{w-{Lwk0QZvtmx*5Jc(f}dPVx~4+(TJ
z>--QK5zwCTUf?tnt+NX^7Oe};4u<Ud_QyTc_3JJ}R};}X^W?7)Xde~UJrRASjPB$m
z!@QuEZGLbCB3kN?8dHU?CQ}bd{0s@(ezlFLV3LtXedkCr&3f=TT<swwxLS+v{QCKH
zkrn*E&EiQCKs`N1K;Wm;44I&DfGeDAg8glm5fabAK79d@=;s10ShPW%j{-Kdov1dl
zT0P%yLg+HYNH5kEM=k-+`AYT}+@4Z~|C-oL<nPC}`JB#9HV-Ch0_@k20NO3R?%jd5
z-x+2RWrYl<D_M7Qio9)vUT{mi@Bw~K8!q~NLLTjY0!tS^ySu%gK3;Pm021QG6V0!N
zcf9P36i{rR4V4>%dmF*0sfS1|eR}Kagm0%<1|wTKs&1{^M$11pg%m<sC5l!~E0>Qs
zV;wLaN3ZM<a?P`M-e7!6_XFrkU03b&4auU;%@Z1RYu0PQj8V;!55BUm={J^L4;`eL
zd_@vV?B&5;nVt6D!A^pM027BjDD5Q(N{P|FSkE#YCZk=(Y;sp?ne-I!w5ir#wHF?u
z5PsG|50z>ai>yH(3ZxL7WUJ0W;z>u`4}R&&VbF!!)v1{=WntkTr?sOIA5R9Y@YSMK
zx^(dVN+f~83tdQDvX_f@tp-=2y5grZf318K_!&q#v5nNnZm5nr0YM&4{VSf4PPK}(
zY26>?zqP5-O^*XfZJeLC7j?(9Hc!9n17H8bpbOqQ$J7gj+cQT@*tEa{*^Cm~DGHfX
ztjIw^sh*?@4X=<4xNwJ6=ZgGkX4?Z|kd5#y40IzLe{T|bR;bp`b(?hLIjP+ye3gSc
zs!V!n?Q4mX8JRjwfR^)=lLM7sQR0{EtxJIX2)d2u^~$nuw|PE@`S*z#n5c(3{J-0O
z&U{BjMB*htK(s91`Flf%Eo<-a1<q}3@=-3}Ka~8)C~I`K8Mc#&afa@7A#)AodAe03
zHVYc~g$OH0=e`M=o%>2S^s8wqOcOc(g#`^G`|F{^ME|qQ8Bjt>``10`I?g`vP>L8n
z%KDDMpd>S|3i>S@rG~Xti{U(VU8+vq_G!P&t0Gjw3pDq9*nlx0aG)i7W}*I#=@`Y$
zomFJXno>bO!6S;+ab{sd3WvH%zp6De{xH3k8W(fme5Of4#QGVfM$p{1Q<o%^3Pq75
zvsXMq932L!5nvSjxfhFOLZ5hCM6%Q0DnJ}qMWDfI6h`cBu(;lAZs|yqEqTi=>&FBm
zr?~lh*N_rJ#PC$ZQqf{fBkwr351h3$5Hp%dzs9&ma8aSqG8u}<2)2?^NKdYx>r_gW
z{p-pkZN#^<H;CFDe}RBr%xJ=1zDWsitrt+?lEeuJAjA)HPe#;V6jC~OJ!A9FS>>wT
zBT2iUD8SuzNU`2cxe|Swu8m*xuJ&}bL$zEW8J;r!t+iTAM+_^IZ-s#;yT51&me2;-
zr+Z!PrU1acZcQmIJu)v+N3CUwcBv`o!y&x-r};tnLv`RY2TbBGML6*Yjfpe*z23po
z!+O<sz*~h5PoC$;f3X8cBy~Xq8KSV7e0hrblHA1LXSs4|4+COEM0f#VpNu|M8@o~n
z9fz9iWeA*I#mQV+nTAUGasbj3<>*iFOU|5vjLBZi_gGdZ*h7THBM;8qFvf^M1W7WJ
z@tE_tt@R2^P$dPM*#jj)ON6r6X!|zefHVW3t~GO$qgg}DU*0RR#$mhX{z6+U+hqFp
z$pkz?>i4!>;mr}pHmoUG<n=nt<&n}lWz{1`M+p~-T^^?Q9BN_i=Ms&PR+%rW;Kz?l
zWG6XEp3xFA`Q8-l-W!N+4?!bt6%3>r5u^^bP~_P(|NbuKCv<e^-x~}mi|XAcPjrF+
z8#YI+T!}8*K!zC8D2T@%2Bjch3~Q!Nrite{<lW9jfWtR?>ozqtBZv>Ox87d%^D-xx
zPDF#&^>X)b7m1lDBgl4Y@O3>lJK3>}#=~1m;IMII$28=8q)fGk-;&@i@?j(deg@Kr
zVi#8_wQJUbt2+BE%$&#|izPWh^aO|iXAK)C&0p=tayvumF`Z&A-(l!5chRC;ctGo9
zaH!h`Tf(iBBLvCJr1Q3Fv?ja9j;m!W5k%Y6vY(AH6RX3?(e=j=#rS5|cUwh;xN((T
zBonig!`l1D<?Vy+>Z>ffwD@R^$+<<Cn(h#EEu*1d5~(uq3=FI-T5XaYt<EHYK>W=6
z(7+$sn%D*u3TA949S-8_m*UFqf%)7%Fmhw~xEQ5kvTzdK<*yq*?xJzF0_?e^%apD9
zB^?&c=95qeB@4a#bA=x$_YfuX3^n|iF(N9SDo&Qo6z+Hr*EHZv<{O$_pTd3boyt9j
zcXELa*K<WV6^S0{@f4=4m~nX^U6)Dm0_!~L{agP~ANQoz1+tM>*(kJ*Hb2e#b=MM;
zFJEJ`+b<`0g`EbA`b!&R*PT_RA8ZC5rxZj^_oQ%W|EL-x1N=G!UHVo^G?wk)9i2a;
zZA)Ra82<$DKMu?6)|=d_p6>7cUrfCNkR)BSHQL6sZA{y?t!dlFv~5(|wvB1qwr$(C
zyZ`>~egAteDk3X0Bcq}+&WU}_T6^ua9qE$(rX6k~lWcccAlsVO!&|XGS5gZE14M&G
zgLz{OSpR0g7C0F?_m;RKBk3dZ4Qxav^0bAFcT4w66!$4o-2(@V4?z(P>@)1uO<wx#
z(q|9-xL?J|fyVd|`Cd4VMKA?21-f-aw$#vo;HqMviI%-JA2L3rRfM>TE9$v#^eR@b
z|Etf%qRHqLJ^**tJUs`Jl>H<;2KX{O-?RB~(MrDy2I0^9V0P=T*#fi~E<<R7i2%rM
z(j(IYr2KLJg?^wX!Ht655iTZuo9%(1!T7qB_9;gp<x@C6NxKFL^F2mGZ1rcl`)J<!
zV8;8^-rN8e@JlIG7tt?Lj8U<Auve_&+-P6~`i@*^ag=Oce7EG#q04r_3rtLfi5Pj8
z94lr}2M>^#a9A)C9FHMga+b7;V9{1bMd&Pwg(_O0=YT7TdJqZqp51M{Z|<MNJfJ0<
zB(ahNDBquE-C+XE4(ufUFEd!gq2(JbU*=lMEXi3X*+vsZU60>QP3vo;&F!LwjK?1z
z8o)PSb4pdLXHnFN*^QfkSSURuy%lEq=^%=AE{@p0#Vz?Ms8>AC+81H9TGfg4)Qv8r
zGF{wuuUh^fvs_&B9;d?1IhA!4w`_d^z&n$9WNp7)y#DfQrUMm7w8RE~XUp^hpbZ}N
zP{W6dIj~fjopsmY(RedHQ_+||l(<Mzm1-+RQ4+lgu&YYaoTSqLqMpK1n$pLPgnd2h
z;}OXxK1K-YER2FrHkkMedHvIGsV=x>+Yz?z7=lHBxQEU`u_IWhY3o@%Z_ldRsrYb)
z9p5QdrE5^t(KllYY+_k3Rx}XO6L@Icq<_SKQNdlE<Oj}ePa_eT6#vGwD)xn=^nM|p
zBgrFJzPI5~LC^p|+E5VFUz(nRTF>4}hoU7(^A-(c7@vZ_(TCg+-IZ!%w{XAEU5yc+
zDNgs(Qa3-`R4bjaOPl&-Ls4xDlA=B=qwSk=FFuzdF?mFEhyNa%)@vtMX9zdwWB33y
zf6r#JV95c>Kk$EQmB7<;Pz6Pol%M&U`hGARS>w=F7SR%beyH5)KXf4#(h%k@yUuy3
z!q$;YLOMafIJ)ANOLu!kahnzu^L<$T@x;<LK3pwr9kM;WYnv84xK-d-EpbKr9_QY1
zG2cecG#&g0_D=^T-;u>Z1g(!{{{8V60$1V*4*~WUUiRH$B`};`L|`au#zX0T{Fgth
zx3~^*^94X=8B@zk5MhWxy<uGZtAZx&fR<~CR}8iLO7tSF`7Nx}+;bq$NNggH{91Y=
zT`eE}0&L2{=-2&A6?jC%21%>h^e(#?Q1uj6mFc83^LP^Z#<xH8=fVPp(?1Pww|4@g
zksb!XE<gQ}pX9=HS;<w~rp4*S33M$lqN{sXwj=;ArxHCfN{^t$<%Pna7IcJ6ss>u3
zoTCOWf(zS~-}T&#g2C#KMplCM70_Z<M{c&tc^MQ{T{F$arB+etl}AJ3HV0<{LCI4H
z!ol)WNOdRu!8}^)r&f&9dh;cHr`DYJ_u+lW6Thbp>rMMeUcK*>zw51@k8aFQnnR(*
zCKce-5yR<k7L^!qnlIt-T{lA;o1*9hGef&$(|g-0uz8Q>1h4R$Vpk8VpaXD`2}lm1
zizOJ|qGx<$Ii1hmZN-pQN8l9co4FP#<lbu;1w3f4?Z0{Wp+6AWQ+P7w?i%B&kPv`>
zw?89wg@l*|-JR*VPyH%yjNcA~OQ;{&!UANWRo%H(FH*SvW{LS0e?NNreB@3t@chH4
z19%}1OI^ueO~Ty71JjQ?=+kz2FMdw!A-jypsoD-ZaEWei3yVBg>%t}J+igV<JkvGb
zT<fv{yc~a@I~PeD7i0jel2b_IBBG6x|Ls%q_c@`46Un^_YtH0}hw!CKIi1Yri339L
z@!l4i)2^}YLUcjsdh#O+9D&hoiK$;Ifc9)C>PBl<!KOV)pKfnD*ue!wpOS+*Z#&c+
zEVh0q*5ucK^J)U_PMmBEC?>F+b@t1cPOT>zH4<U2MZA=wO^xkGSlgq)#?L{p_`>*8
zGrzB;FCnhZE!N%TbQ^vHAw(yk8vp^o6Y$D@_XT6_%J;d3;@|muD)lB|MKc(h|4vi8
zNWImsMR_I_W+3(Ih6WJ=N`3(^d!JZZJP&I6mQntX3BgRQ9D_V)31vZ8L@4{>j=<@y
zn|Y7huU!lt#Aj$sS78^G%BQL}pbNbw(1IFPjpUWCjymuNb;R{~3qr7Balj|OZ}ZgI
zrUm=2n@3k!Si9F+Rx$TG3}o5*8mFUI^e0MPHwZUvF)n`?e{roApmSss{U}yGpx$Y9
zlf=f{e7&fPEDq0WJrgI@>c(-zzD+{+PxMULP3C40q{P2-*-WN47u(Jm8xMRwZ_b|7
zMlUZnAH~>f*FtcMsw6~q@&K>|-n)3rPYQz13$2)dFV?qZWF;qwxG5@6NH!?4TZ`1?
zzfk>lhKF9yqtEMyzkipVu#n14j%SfQ#0Y>!hX{p$RDm_cu_Lcl9;wV@OqluZ*IwVt
z#OJ_hPcs$=aBtoMWrGG~p}J(KF{s~cA&4^NZDsNDOn<K5@SGil4*<T0-OU4}Un$Ed
z{iWl0<cFCdE<g=yMB-?qjU4z!QW`ws5*t18B!;4SGc1z%^5}MtT|}AKANjt?c{Y3&
zC$Ut(+HoQOT<07E?a7l!v*)BF#pMHM1%U1bksI|eaEkN9|EAm3`X#N{a4P13Jn`1-
z`w*Pi5Ka`IU~R(PBM&&$F>@Y&fdG$4@XW(qn4Or*T;=FqnJiz_EL!i1k5*$pU4_D{
zE8cH!7$@UZ)nlfK$=SsW@T8Dp0!c=KL6EA`+L?4FSewXKpweOE`iGEn+CTE&>%TpD
zS*&Ha7dSigtlaDHa`gH)RG!adU19?2!0WF!x4bR~A7d<>0Oy;EWe3<{7E6EY9(V2N
zEj$<RRI0&3H0i=(oTjWtgYzKaHE7|>?^HT<q!yheF{zj5OUUq6iVY<guZ%>@BY*7P
z&7NRYllD7?x%)fC|Cenea|*Qzu}VR5-oP14-c*!H%Qkc8CpGBhrv>&yC(+@!0L3Zk
z&`$=;Z~_oHBI|f^#=&wF<DkSB9?+$m2XU<f0$`X6$Xyg!plvjy|Jh_okSY6sh%9z8
z8@vUTyx-(LiQ-VyzFeOEYxR>Jn`%KC?frd*DC!g%v#e3CU6v?7jlK_t7E#*11=kq-
z|64`rnXroXp-uiw$+<+76^aKcWWob#8$K!|P=WQgtpYw$2Sgq1Ncy~R`^qCbLn$l4
z8ooi-!CJmKP3a7=e>%?+76XylSt;fbGsYN|JCIh3^mJ7jZs)3%G^v$jUl&V>fBfar
zO=(mO_yA*!4l!$YL!B|?V|~RE753p;h=WaPfzh8pmkE^zw;*%|W=>QhPaFcsU#&Fg
zQ#A7?mZywgQr+0F2|^2F|K5Bv)yY!vG%MAOlz2gZT$wF#UE5&kGkPn%I)LGy&^~Uy
zTOYe#{Z~}DyGr_tc>TBQ4@C<jVpj2bnPI|>@{}(p^L{IUWUieg)o3gxA42309i6D$
z$v74fQG0=s&OIYjc55GE?@|F+E7KdZ9d1G!>2suGORhjA{OG{NCwq_7kgSoWu(AnD
z5~SQ6)#4EOC#0ko8jfoJ=u^)EZV+M%q_*aH%`ZqP85v-D6>2YqdA-@)xmDrnf#!!I
zB1|QOZkPO!>xq?QTDe$09<wmmWUeA$@cQ!DfB8O5@b&ujhSzGLx8(z9Qvm;WNMfx5
zM)6&{+(pLU?Jm>h1amOY`s%RyHS>O#e6XkZSQAKH4AQ3YgToQ4@}jNG@;y`BhqE>M
zJ90u_+2(1d8A{7r7YaeAYbu4}lz#<n>mHBW-v775&^ag>u~ejsWxuG$(V))Nk4OMk
z%4tJ2joHY@=CGP&;{?Q#zPAE20BRyUT}{=t+svP`iOj}-=8K>d`#|1{sf8R1(i5~M
z4Zg^2o?o)TA`)_Rp0iFuUhyZM^}EPL4N)1dw2H~8NLYM9#8m2RzUvsSgWcIj6LF6=
zf6jnUYa8K?Hlw9s|L4+z&RTb%L+j(}mKPeGoWC8PO9s&5ta0_Q7N%cu#j>lPuyE)<
zuG+X6=nZpBa*ZQm>FRkTW0rk1JG!tn7jBF9^}nRjj;*e81wzQ0K7_sYljf#l9LGWi
zXR;fjN*w}fe(HF%U$JCRw6u+zI~%BtK7S4Wh5Y6M&v2@pPlI^DWBVg{kvED_UL_lc
z%#b8N87NXgnvC9C3b|{LQhj4twIi2bSHrLO%r^r%EW!#CyL5*E*aURj0OVL6*Xa%h
z0(IM7=h0-g!{>RZFJ&<$qy>rt(c7iBU3j8I7#&fk8R7Cp$A*<ViORoQX)WL<6cNMP
ziM<*oVtm~NdbmEtegx_HZd%9n`ueFo7&QZ+6Wa#c*UpS08{2`xR-;h27G=E;^*jhX
zQkpyIT0cDA-{uz=G2*|y?oaPaGS;7Ss*IR65YO3}L@1&*jFsQ;=;+iriR*13v<O@B
z=H;VC3x-tZGx-OA95v38J%M~zy?K76?8@_xSLW2#G5)`Ucq>-xk!q+40w81zPqpn(
zpot*|ne`42Eo9&hT+O>MhA2RAw<a+Sk&n9lHgtii7yg%HvM`;B7rxHhMc%u2V76>E
zE*PUyer=l;m#H!eu@8nZA;>0nvp^Xg5Klmi3rR{qrZJ=dU0y1DotRX<=ui?Dne_{v
zA=;<}Inqb<V&ToSv05tS5ulnbl*XWz>LguhFtIozRladj%r(cRQI@YVl_|>CeoEsy
zc8MJLmu|Yc@W8Yxf%-%BE31R-b&(wH2`tls_rwVjR{zKKYRHdIS7{#R2ATM8!14tO
z_}%;)$^a(S(35dXaP7AT+WujJ(oCOdUFc%J*`-4{#BZ$@s<_7OQNZ*$lO{J77C7k2
zFRn#6-icey#Woc^vHQcvVb&*65p0UNt}1V#H9)7I(+i{9v8k54w=h*W5RMUPo~ZtM
zg>i6X3b>$8CYI%E@8EO*&9p6wo~MnWy6D}rL!4Kku<-GB75b=14k#^Vs{AH0@9Vf`
z`7V4v^OGW<>p3Qh7~mo`FzmVn<s}}#%*S!9!%smf>ccktb<g-=;wnMLJkMNw<o~@P
zLEzTZ??d(coV#i_(dfbYp!C~jM6g+ru2Jz4KemQ}=8M;{{-QAvTq-i)kEo#s%wVET
zatJ3heP`rW#DUl>+w?Bxqia^^qva27N+mOTTFl$XHj_}R11vhsIx#j_g|y?_Ke2EG
zfeKgc2wz20LZrrX1UWKA%V_qMWpb8!+uCQ>r*2`+xEHe$rnR`X-?mw>WppwTsxFW#
z&10JDvx>zY$jkg_>f8g0BHyj?w*TqJ+12pv6`3XhAv}Xy`Wv(7siTlE%c!FSvI<HW
zMzRX3qZGD90t{skiQquZE1>=1X_mkQNuYin)#hFTyHr7O`h^Wy?nRgzsKPf#i&ySt
z%<iauNK7g&SbgVCb>c^j6W+>(ajrsGh6P}Gi@oi|&Z>Y8t(pbyH~FX#LAfpC=B2kx
zYBRi}@AX>TPbMU-iSb(YAM6FjhotxuYOlD0pH~43IERWMn<&2WV4oOKM4#|p8gAH^
z|J3zD${-yRU26&?;)$%$$oERkRIV4P@PlWgQa@<>Fo0`-RW^WQSsES#$6OM~p&3#K
zjbfcC8>d(jAf9809m4*5(qjn6BH~{Ip*hq@tp!SZmQ$PdJDIaz7<xRjNQ*0(bKGXM
zX5zI400f3MiG}Pda|54%Vx<ZNR$@K*OQ{e9=E8L@0Q||frw_!lRb@kxrI*LA&WT1B
z?&lwn;8xT^n-ckky!(Ue*Ja4l041c6(E#;>Fpx41fiTbjDd~vS0>Q=w`>~+=3PPg8
z^bjHkg6lI?=^TR8D9NL!&w&Gz(wxTIoT|?ONZYOZ;;Z&S5EGpYkRVBf10W!n_|ymD
zSb<ka6)f&#OL0m4X!mhYY2e|9=xswWabd`op;6jNOdDsefjiX7vKJHLoRWZIHb`Du
zeRJxJ|GLgUoE<4;FLzl0UgjeH3>P8Dp+D0ey=_EI9_X>1HL8LGkz*6<LSZQZr|_`?
zpgo)Z7nbvhX^`07j8Dl*OX0#Amf?;G;R7KjjH?RIX8C6a^gC!54c6dsuzFKrdt|{g
z=$_>nVN0n+$+mWSK{2U9V{S@V@V{PxM@S>R1RmZ`*>z$r*KAtovQ7gr)c?>{iLPCv
z2+nfbRQc2&)ci6X>Vq{eEps#(0EEJG)F~Mw<Xx9S3AsKYW|xQ4tf5hP2ozD}o{fcf
zZS|gAhud<SDK?MaK%#!F9xxCCoC26_iJzM*LS!5Q6#2O?J2qs?Mj>=}q?L-(Ov3Rs
zw;XkP@uHdr<*Z>)cuRX>PA9T!v)fY7)O#~wp8_GahiiT$l-(!yl=3{*fP*>P*D!;_
zoYqZI!_)MBJRZKwV#2aYBYq=A%Zp9(rwOC!4~*&V{_2**KdafPpgiJQ)eHZPhwN2Q
z_)DcNHu~f?Bl=E>*?&2Kep|ZV2V+8fKlCXx&N0HIFzbKZDy_l}a}T!q7-j7T5ByF7
z0V!?|IRRNviYue0Fo4=CRpgd1IjRVZi9HDEZFT{p?|V^ff_81f@_HJka@?T=@?(Sy
zd?@t40Y01|5<UP)YVkdX_2yWBV7~TD0&@UC6-H9gXq)P8v}kk(|3|>0YG5h`r)qF@
z?aW^Sx)%TZ{L3G-^5*xNKgAHZ<JqxCG$Y=|>6EES%twS7pzVO~`Ys~H>EGoNi9XOS
zK9-XSi0gBNu}~+sCsZwN@8d)7=Mg2JG)dFJ0sg;zEYPQ-o#5F%7P8|2FLU9y!;tbt
zMXXZkYDbbl35{~9{Gnf>OfDCz%JmP>hkgol!+$9AmfANNxaBU8+;^O9tB5dZVYY!l
z6iFt`cw)|R0N}$XX6v71Dlnx`WH++Hq=Uq#=2lKsGW8ddsgiJeIEr(InjrnPpO#J?
ztK8M|9Ti!N9^)aY%VB%|KRNPa{~C2E7uGh7BQzyBy-Jm~OX^d^l|o@|HoWa3PWIE<
zH=j|zA&?tS3Wy)wgwKV>S`-{6CBO2ImDZ!;<>~KOfa1t!Go5No4y#4Vq-Hu)cOkQS
zMF7(>57+ZZO!4#hc}p5<snU~BxN3-;vQNqc{@@kQtNXKyo^Dcmk#fv*!fzvIfrt)F
z?E#M6xyz>@uX>w3ci=Vre*Mi<tJJ`XcdrAKvmlYjzOz)CgLRQt7e{!kJDYeM*r`I^
zoRddWz;}aMr^_lOErF1<XOX?Av;%&WX~5u@fntQH6m4lAiQPW)ei1S@SP3)YJ!;Lm
zj9|TDTbXl_5>ODq#b~S}Z2jMx35EvRZ)jHz>*8pIygLv?wRx&--C-JazAQXmQ%t%)
z5-t6VJQtAsr8I8e|44MtJBhNk_}AEZX=T490fN)`kgk9Nc+%iTjmX!?Y*c6I{0FNS
zy6~Mu>iE1{{gz<sxG#UV7`D@SKR!pw+bfP>uJ`U=NHTmndmeARc*sv0Ph=X1-Zm|w
z`C7t;tN5NBTlKe-Tb$uGof}k(DG0SJ<l!a$GVz{|28B#H@W4-=tv1)_mJEovy2NVc
z0Ox9Cw1J7c0O)9{b*g~gufh=^Xaw#3*S?R;OqA|Eq9KbSSsfu|w6sf3au~SmU<F7p
z=PFSd=jDQ;j~*8q-Opjf@CYhjb3beO(|$k>!kx)TLjSnt#BCL}I@$nQHXfV+SZR@F
zz2igRBtB^wb$oH_P!40(R`2Qtu|>=PfH$uaS5g<r9=+6)2>s!rG_)l&=5h_bV|D$r
zA}6W6^6KvQke|W_%74lby(rK@m8XD=H5aN)3rGz^@;_i1U;(5qmctK{am6|qm<vx9
z%iwS=!fVccg0LaQ_&MEne=P(k;bkTji*{#}(SnOX8d+?gp3gZ|$<LQTvlZ_E|KTN^
zdVJ#Eq#Vr0;!e}Q9q~+>XOGQ^Tc2Brfk~@vjCkmIP|w-m#dpH}oZ-h`uN}!mVYk;k
z|8$kE0|NGP(@hvcCG=_%VDas+Df;RUyUSQwuyT>q)*R#@UH>x97#8}8i@;5Ak!(7g
zR;UG0aeejN!1hWtm1wXiaE1Wj9qn8EYS~uu)YwF7gsU?lLCDwhFe0^lL7#OSOuPd8
z2S;NrpW!L58_SEvjf-#k+6)|ECs4E*AlESmGgBI*lBTdow3t}Lh-;D$h`z0hwXzN7
zKIUIzID9DLOO2k5Ai@MdEhBb$$h~+Ji*$`48Z@Tw4oD4HA_nGI&5!{kZ)hkDAlmv(
zH=BAyZB```N_Ayk<FpCi0|USOZe@ce24pT&ne<E&pd|xxVKTPI04&;?Bn$RdA(5q`
zc$)?Ntm!SDz1^|N(>kRr1Az+@8y8O6)y`Ko&E-joWf;`Rropz`(*5D(;}GioAo+nr
zW4_hd3&Xh}QOoez#TNh+Y(6m7u-%3I&Tsrs_x&>Evc+73Y_XJ8)LK~y^V(Eh=}v>3
zng1sZHorMmu29x=M%n5oiQG(U7-**XKbTeFNYO|7*Egp!@7v!X1i{zcl45-KHf|yQ
z^fw{5Qp1&{;vbtM`Ubl3EM0B|rP8g(qyJ)8atgOJ0Ll$@$E;DanO3h+rom$-MA(|)
z9%!pWeO|wv7nUK^*y~BV>dk&iL4=`~=UG+3c%V7Mw~ErNG9YFQsN^szt?hhM-ckaO
z>!TvZd^qo|^JNcTe16lk5zO$Q^cWLmGH+|5wnrn;8n`(JiI-4c=ot#?4ufy_=rWBx
zi}ySS(1I}26{}iyD9f6sq4F=4S>$_1y7pR!1eha)<xlABD_8hE>*pS=G&bGd(haUA
zCHOcZLEKZ!TWqptlIHnMecVbCgM{_FPd@jv9?$0O3VGCTlwhiFoj*AY{U5`Xynj`3
z7EL=<;Ymb%n~T(mUi2=KiX5E1cLXm^kt4Fc0V}(^1K#`|t#o*D{|-;H4YXcC!n9kz
zj|1e8vtX@bV5L3DAaG6w@J_PDvr0$t?`I)|xtM4e;r#$}q{PG7HS8j!VlSO(!%cnP
z6KB7&KMxVpvgNga>404%>njY)Zo$clwCR$-+V{Jx*PEphtd$PXD?9gX<Gv+=XxHlD
z0snkdDs@KJM{fqxmG2g><x{79OB2hab$5a4NpYWso0(=H`lKpnGn+bs4dN6*V4~<n
z_V6lt#2g>*aTZ2KJQ<nn7LG%D@KGULEDueEy>&wub$i7mnl>t9=@IXg0(eg1D_fQ3
zZ$^G9DBgBK%ii(U?lIFg0(BBwpEd+F0A{<IjSAJH>*8+PPI2NTYZAxJ@H<W7Vdd~D
zpj16CiS->6sXtXP%3q%Q2#T-6512v4^8WaDUEd(8QC6Mb(O2pds#H7C)>B+TwDD5y
zMqqG!Wf(8H?{E-PlF+l;T3>222+bhV6vmA7XimwpFaPrWYjOxLU=9r+_v;H!0p)g<
zDznAjQ5g4G2^$k_%@^Y5?ts6EuL<&xFxERORppzGR+#J`|7F4n=`)GwLz%W5iA-aO
zK{E+KBk{aFO0bgIn54rq?fNozPO8M+h588!gK-IyTvJ8G&J|o~tj{pf(*x(_!7z$v
zmy6q6F^qkfz_kY3k#^=%J$1MMA(T74V!fwYsSo~~Rn-#jpTkzar)uCRg*rOJni-M3
zBgO{oUP)!2<d_^Dph-d34G1#UmSTH_dDgqzMz>u{#V=;-xwBuz16CokCRQ;fR|klG
z=&1YUx6IEg&>XW-pX}c1guye&7hAdygjYEj(XWZ{R`PXg<@P8|sW)Z--0Pm}xXb2@
zzp!MNK=1+!S@7`dpc&d`tr_h4;im34*Y(6<WPi&53m6P9(`_>W`11pXaVK_q{gAd6
zt+qbw@i)h}%5-*k%!BoGJ-P8%wJ|gNfo)i~3hFgPcgM=hPY-ClqcI6DF{hT550izk
zl2Vk`WE_Ug<}#$T_;fD7QvhRO<gPp~**|4erI?O+zxP7;AU*>`zFI~5)pNobR=$8B
zdf4GdRlcaI=C!KlpM`t*x_aANP^*A#YGYBii4RG<a-iYTa}V2S91{z4^Um!WwF*R}
zE$fC)U+52wiiqEX*@T`wMQw}1?)PcN+Ya9czU|fU!?4?tya5#;b?LTJRqr|jX08O(
z`>5&zAK+?yHs1YC?rddhVHKjmx>@i+9}8*YzCvF4ScLdQdkb0IWNBBVGc$6zIYC=j
z6esT8YfV(23+hPLRrHQsxo6vp=iIU`p=N%pXU4Yj4Fs6ZBi<^E{f@oj!xM{`j@yaV
zbRA-z^);1nPVfWtRaS80Y=r3v`gybp1Zg3$R8%HwJ~kk$V0k$Cmw=)CDy}DAU?!Uo
zgOr}ssfN>7#|?)QUxhtQe9JZ`Tsaq9v3W+qKEV+6D?66w>y5ZEPn5P*&u(=aH`fCj
zGdAR($rI}eT<cXUSb*Y^Kdl-^MSAC3X6@1(LzA2`#K8n~5KR#4QHCL<r+P2dq^3K<
zp&M%IUJX)_OF-$dwVBbbc!Y{lupN{5%`d8iebNs+v@1YZ+=80A8#c#^V{O+E3bLok
z=+^pfIC-2StV9R%BhRNl*XdZZ&!x*JhevBtvVoSHH*1r_Y{vUT?Uzm|%OYMxcqM+z
zrN}z-In4mHp;CsBYk31QUKMd&85#PE@;i|b;M6undgV8s^Won?QHB&=i^vNxGb)nu
zdK*W;t=Z}Pz(;&@Ssou;!d`^Dm&2zXM;8v@@WG9`S9E_&v=hOtwmaV6YF)h}Mx#J|
z;hTwmFd^0An)&3lB;@CX4S!MX*PTri99+!b!4CksUE=#i$J(7!|J^ih05j?lt7uj&
zPbZF>;aB^=d|4S$>pvL1XkqTJmy~~Ld!XR1)MwatNEfP(Lt(Q3XUA&0U}a&?sd~Ib
zQ}1T}a^d6OPJmuHoL2ZXGgT-mdSzY3h#`=q6vZ*>o2olON}2dI%St`D7ep0oLRAmZ
z=rsT?Gh95=bxbk@5+fBRHpN(}e)q<npPDa&3OBRNQx3Fqq#waxeC=(|1BO2ZYv_h4
z)5_Zg0;3~lv={=yPmz6M7vJw8N?*=j)P1kD=-Xi5q9=NxTx0sKfoe#oeP`b6(+D5;
zz0EKOa0e$k1`+Bz>68=^aMd7~0H&v};dube5du7JS^I@i4DRNlkWc@wEf5%6mVL{E
zPg7oY)N~g(lx&S)iooZ?q}>Zle6(kAcT$pSC%cp8wB4Kh%qDT^F-r_<OUnBb{Jz?O
zFIt=dm|waHSP6|?*TSYg1vtW&kyskN@}7r61nR2;@;JiQ6olc}cict?NObOb5l4Xc
zTWtN%?HG9oMiRdGL!#qCnyNpm2n&pllvq;(d1GTia5X*F1_K?T=W32Xk+RquVKBDw
zND<qw9v+YPZ*-5P&k_N8xA-}HCshQ*lvLx`?)*aT$+(5izpGItJ0JKr{sz;)Vsr<9
zyB$(r%$|9U>p-)aA13U@VBxzjh`s@yPqo|(KXfGS;9$_9LmqVv!lC<D3f$?!-3_Z+
z`mTpI(77#HGH4~#{Fc|GC?QtslLg!?K=9;vSbLRFzVV#cXE!H&NmylNBu?6QO3{_K
z!C6cC(jB)WD9`)WC|Reg&U^z^V4uJ$#1{R9JNu87dWLIDnuwX&?EJz{$H4%y)nU>;
zVdOXL*n~jGP5y>t6zNi)B4TZ)b2|-z%i70qQ5Lc9Sh6{&;N5&JOQMaqtI`^bc{G0u
z3zN=_g`CGpAHNl(nU=5IJHS@iF}bZat-IKJiI!UlxS>rtzQ$~<1G567;pF?W0>g<m
zZ{5sbuac=Q%u##p+Bz^397+J~8_Q?+<8t`UWWj^<16JKM0cY0>3pxHe;Z!Rp!Zv|5
ze~|v0#&P+z0*6J3cT%-~kSwZEW_8b%{c&>hM5w=7P-vCM#fQv{upzSKWYGVj5Vy$Z
zauc1vo8g~$zx#!s)3qM*pwB)G5+vMQ4EMQX8)8QMmFWE=q8dUb5(${U1L!cCH(ANu
zp9YUpLkfJdl#=_tzYf#^U6w%esdD5^pj{04xSyO{AcTQ6G~<dD^AhOp#PhM{$S4T8
z0&V)y{wj1>`U}z1hsY-?R+Y<7$j;Y$^7`xgrRk!k=l<*>s_SxdR|i7CD%2IrcbC+3
z<V{EoHoA-rw*t*|>=R&VN0#&+_bSkRAaZu_8d~>QPZu!(u2ln|=|id9_<r4nEzZ3B
z`qid$;$m%PFv5B8BS<*O9*K%Gk4ZaXX1!ZA@R%YP#e+q{-<a+FRMLM{Q<kJxleEL!
zO^WF>RbLxnIMu2<h;-$XYw@}u+egYc`4W!;#+rjP5(u;5QU(yvh4L1m<9#KO6IQJe
zbUlP0Y%V|dExldyOhP|?qkr#QPA+FlC|3A5jg$-9T>B&4cD3STPl(Q9D6hy&B?YAI
z->{y7&URUO%Um=MCr|#==}#YphrVwcD$twm6L!(Eb6fB9N6fP{_kIZ&x{14{nKk#F
z>M+_RyBZ(a#sb243(F@yl$%fQ85+p}($-aMgTh}PoG5BFH1DO5W7n-%ZInJy2k%2}
z6FweXxX>6g4XadX65tb)Y0ieJWY_noOF-|?yn52`8oeUVxJEY(Ag+BKTS=tyn>rV7
z?_aPtc6ml*!(N%yu>y3_Q)h24!XeSE%hbkKai8e)Zvcnx$pQVp#oYK8QL@)|Zoz8(
zk7h<)l?7>!!)_PXUj&&3ceVYtev+XMT~qNgI_q9HoGV7X^u%$ialQW1c<r`x(EFO#
z(cY#)tuk{Xjg#fx$&xRH7m|&0LY{T~*)Km;6z#fU1-#Z{pC@P;d+&?Ar3ClJJp3!i
zyWM)#f`DTuhW*x{ZYEjcy>lPdi@SezLON;$`zKS^W>gc<xV&2|Z)?iWS1uV+9n^MK
znb~FIUI@u8N?Z(L5jBqGTV`TTLmvY0KlL?;SNRnD%(&Z!MOSBdpxew35$~(65YO6q
zK`P&Z<|6WbO>Zcc2Dsf6vHIH2I%~Dsm6W-7yMUX0YKQsuPY*}$+~!5=6s<!xY_Zgt
z_6I-kEYnmt%{Z^C=>n}nwl^+DIQv~97P51*U*^QRr7Tm3x6(-8f~ZB7?cM`o^$r>~
zzZUdo4*!SYOE==TP$j6cd|~5|Fmq}-kOkxxPU9>>9>ncRle|@}cgI?Own^xAH;V-V
zIAcG)Z^v`KIJq2Ok4(PYkhWCpB)?9{D86rRFzP$weD|XFvKw0ZvyMgrc@r&vUs2?f
z)8wNyjD%|)gx8GxoKRd@L~v^wqblqWwP|s9Mmcl>pYm<%7z=ZAoC9!Bo$p~ho0wT>
zHVeju<GL>0J6J+GuUD12v**wj?R)^tWZ6>>FN~799dz_Kl8(1`Aqscb;Xn5A?080A
z$K4TgRc#IY6l7Oaf7%f>zpChS-~YVF>H%$$U25CsR#4NO|Hqwoe69X}S}po3gZpVZ
zoG2TFSmc{~pRZ5q86D21MSxwaPG==SllVi%6&PH^X9$uG&YCg6YF&75j<Nc9Q8e*t
zYn9oNqKr%=w^}Qrg^yOs9{j9bDZ<$Ij$V6vsID?<5i~>u7zIT=;rLUpN$badh=Kmc
zd|>6#rAVWy??>CYeoz_yx!Gyagxv`7O}3r^-Fa}Gb^Orli?u2SHRCGWlpjg_tRzzg
z8g*4J(2u-Kral8Obk}~`Vp1D28yO)}|9d<;C0u`xhL)d_Dp0hSL#OUj@U$riGgh<5
zM1v{BC@7|<Ll!UFuU1WRQC@&A${X(Suu}wisi*=fvY6>%r=<7n)}10zNu_%cT)#!3
z<I;;Dr0qBm04gy%Ynz5fb(haKxFR6x1)dnbJ5Vm=yq69biY@WPPg^$N5C6iBGz!Qx
zS=k>Vz3O}P%0u_Y5^a$@2IWG7LrV6xb}87hQA?kMCs_cF`S8U4Cor|>Zvw+@>ogk@
zu&7H1&vEJpGZKy7pZ^?dyO+Fakw)Zf7Ql@z|I<S3S8iO)lU)&Uzh)W`TT*gdXue0w
zO*o5n;k^q0FV?=dcq<fUY}{3|sz=auRG>MT_X(Cga%uU*yWaIvuzBg<J1=|9*A6(x
zEgXDk_=X&7=@D*IM_2G8<mo?Ov?c%A9bf7PPUp`L1BNj*<(v0!mWnv<fBsTc&8QBd
zPujiCe$!~v@75Lv<$ux`@9)6vVFFfbDYX=4s5P_zSlR?WtV(G9yeCX%$r)SQ-Cf<M
z&ZEaqhu=-xk6p$vE#b5p=UH=Tg~F_lnsw2`A42K>?znWdco$5>=S|e~s_AXF0c~GG
zRW^LyBq@5fo<EKn0MGZl=iy0$-Hd-H^0R_!PF}_Vm==k{zn{aYo6VYFxr3>q*)goQ
zRCtU4qhf_I{wxGYE<kg=NUkMw6%UAC);>}-{4;L;(}r_0XYQz+pq%?yW-XvujKPi!
zW-L^XE?Pk0nMi|Jy<b2Fk*vW`-=O|rTxlK={9pW@+kKfF@fGiC78~v4JFxvX4EoQ3
z-a@`7_JKIhy>4dg9$Z|;6jqz)$i!hy-q%+@1u!Ze`e1tonpBBJnJZaM?hh2_<7(b%
zNryUNHB?}Bb_`+T!pQ7QG{-mUl~k4C=@ZHzk}bn4kA8)yXZ%S&QZR&nrC=uN$yu3=
z3A3R_NNlswVY)$|D0Jo?WymzRcDFouTtX~z8pXWy21%&1Zl|a9Y*|<zM-yqaPMz9-
znH^a-wPA;IIlG8jcCREo92>1@6rG+1W>_q!9rnN@se-vjA}YE}*gW@bA$jF7)ANT4
z1wZ1$BH_(C;nNm*+W2k%dUK4e9FuElpWU=ii;QGmS{AA#!>|q!AAX?|Nr@X|d+iL3
zlmZUkYQ|Lq|4L_|9!NLf^W&j+)vyIXSL`>_#<tIIZ&J5{Ch%)ygYWXn{3$(fn2+EE
zf0$Om-+gEm-c_pGYUiYiaLciJq<1(PsfgE+TJ@p--oO-y%F7?~I=3x8k1kD8rghPh
zb;KJR`ejVLu$C%M(jrXn!j`&n2d5Y5Nqr*L$I2V&KN{xe7*U`6G=sXEoWT46ovM{L
zO8C?{<@d`={NmQ@@2$xAH_sTYWLa=yMa;KCe2y==D#hp&b8F?=aXS!d-z0rLW`*UK
z#flgXOvEaO;9g>D4eCJmWNf2BiB7|XhcM;zCNZfLfc^Jvm=UpuYaWqmlo7$O=AlF2
zLj83REUM}*t)<C35%#64@i<h#kAdD};`usWLl`cFARDEY5%-_2sD~LUB;juQX%<&;
zj3Ab!#;M7V@WYrW>d|E~w-Ge&*u+6dI{C!GZJW6yk4D#Dr8%<&OC+Edw5-Y20}t^3
z)+?FOW^@K2Z~bQ;+}HogjcDwwTS~Nt)2XY+b=B1Qk$n4Ez-WW+Y0_Z;+O|&kX}m5Q
zJVxcWb@%-(b7#-A@h!Cl>woK~XS)|Jrh_s)k8GF?^u)G`AK|CHqsqs}weqJf4h26J
zp&h}X^Bj!55iUk4H|r~jm*JG!ggs!D+7!0Moo%)KcvPeJu(D86+84H&G@JA6*ZNc6
zZKt8rkmtE1y?oPKbG^d>!t9}vLk2jI_a!1>>_6=U`jVgeM*+Ab!vxXv>Nl~q=b*#r
zTX9uuUGF!$W#7A#RDMMSJ10h`W^h6@pT@)z*12P991b)eRv5?|3zTvh%Ve^n#arB2
zhzy(B%ETo1mE(g+8>fUUHG39%P}mZED#={N9d2+C`yQ{X<SSADC8Mp#N-G~{8I{>`
zA=G(Wv{)%cVaE+%*5QTXeYrJI>viS#Bk$SH4#IyWgoBAqlk`pF^rvLqQcca#?`*Gr
z=ngu>ODz-(G1(D5`zuq_k@y;N(ni?(2cWcbbj^5zc8)Ieg_n`*=GGS14;>(-^0=q|
z32?MN>ED3{P3`mm*=SrGCy&W_!{9|6?0v+Bf`Fq&&C9Y;sX0S%jH170xXg9rlWgP6
zP0X9S=-i8@%_S202KMZnf+=Usyxh1&_6e55-9Efh0R_0UHu?(&pF;EMfAh+ljklmP
z?ss0*GRI@(vRTZ)GZv1RjGaxM@Ox{BXT})+f;7Z3g6U2G2B21j=4F@Au>Y<sP)t%f
zwBl@AwBumzx=(DlcJ<z;SQ76Y23U3ePUvjG1S84AGIWFx5KgqP__iQ`(3Q!uzqzrn
z;K-JHia;|SpfLk2JOD);mJ<ABw!#YvOnQ+W*Y0^VGlgtLr17ZCK~5O~Vn7YS6BAL1
z8FGYH^N6JdV5~s{)%rOQmuEw1F(Nk5sIRRpvO{Sc{qC=V!3yOWCHKCyLe5~U!2$*O
z%bA(zhw+H}lzNRNn=z-9#t2v^_&7ATO&1;SSVnj5r+necWBi1TV1qyQ}C`eksR
z*k`w;ImLr{Wp}E;e#_c_-gl&qtQ`1fUI~Xsb^Qh`Br+{0?%?AvUeSXi(VOvjsr*_O
zMq}BbV=b{diPlQBxH_rjnS(75$T3UjJOTsJu5r6;C&yCev7|!h2UPhW4QLhp775El
z4JWtlc1KAFOcanMm&~<#O^`snDO<b@1L-MOwSHUHZa$6oiNrrW!3hx2EJ4(eg_8Y!
z3V8%*IZJ5Vpe#q|UW#)zm%s1zrkT*$wA<>=VGv!5gKBCkxwO&z%r}>y2n~|kpG_lv
zJ$pr5C8UfiWi9TIM}8Yint(=?#-X>~-N|f}#qq0$_&rG$eeJKD533&hTl0B#@SB_s
z0)}TfVSy<utkZmsi7so3rfJa?REms{O?Cm$yJIR)^`_5}t{#K+J4-5mh`kVg&67#~
z`}xWu*9Ex09%=!c!N6*pN#eR{8%uZdVRVu9)y>({bMd=J#{8vS3PR-ad*8xt{`=ok
z$j-v8B8vtw7n$#R3aMyZ+foJM+ZX>N|71iG;KgzxkFF2gu^5M6l?G`FFQxzNDjEjx
zrd;Hayd=mB%sTDLGd(Y}M{5w*g5Fvp(#}rtt$-5~Cfu#0#XOS875yZD_tf<+Qaal8
zY-$v4rHfq4V9XaN7CHWn8~S>xp~PG6*N;nS7jt}o>gKO)Cf}fFImzs1BY;VOUx7fv
zJ9`oGnwWkSQzmxa;iDO%MioY5on8h&sz89HY=Jl_);*$ifc2}J=7+>_u^iF~Q=<5y
zeS6A4gLEb-3XKouva0|Ek8PWT?}hK7i>YYEig%j5&#@i(ZMkS#0;&ZqmTX{ec15dJ
zG*`t!wL6Djv!bABVblVJ4&UO2^ISu7PT>hX9RFw4euH8WE$l#bv`RvHE@%i4IFy*n
z*!r+4YAN*dk?Kq55zgOdLQvRpztqM?ORMPy2;xMYetu?_KY22_o}_McBfe96+JE)v
zeVl0~@8)9VW8oVIr?Bv33Ffq7aY^t0Du>=Xc}&VBduUFx-o$azB>4zRg5sd+=tA|$
z?(XC780`@cYIOClwyE>z!Ui0h!8JP4fx-gnuf7rRTC)$t9~^~D=*B!z3G1TGnKGP<
z<}Ou4{tBn7s7%lF<{*1o+ppU6I_M9TOhRSg#zJV<g2ovTJCA}?<Z*t0;ihI^x#@o(
zwU8I_s3U9Rgr^KIj1fOAq}xC^B1myJ+3ZY2y1&&yFZaB}T}1qCrU4M3%B-^gN;Fh$
zy4JjFgvLjPe|alJT<yE{VFOS_u*XG!4QG8Ib?eHCS=8mgjkL>JaiQ02TD#&*ijlU5
zlr0TD$Z|@cTiwoxnJ%1nvisFbCH97JZ(YmrZEeC!YLWgrr4kN$fEW``+IR?MB3Ul$
z+bWNnnta?G{<;K?dIEI8u}z(2wSml^A1$x*Mc`J)&nR8zyScvz(0(b<M(d<pP!BJK
z`?(@bX0U>?qxovDt;cW@c2GA(OZ&JiV5ND<+wuO9-UpO?7)BM@6BdY1E{V|{R)dR!
z(n&0?i@)}x$E^yR(kHLJhoNSdA6&P+K98{Kie`TNF5$ZvKLeO}o?;B(nY9_d<Ug-_
z1C%N=8-c9yqV=#*Q6v38>%^{lg(aZX#pe)SB<SG@!QEHF!=d4z6=Nh0!e)Y2oJYI-
z3FA^6iDIzfN!wkJYuBMJkX{Vf^IXa+HF111LN6z#Ka6Z8PQB!w$6mp`QcIR{{z$Jb
zn_l&c(CU8ZOaj~&hs<GwPSz%@gBvh8ygPXeqS;0kG0fX=|LRmZKY^&!&`MWw<|t|r
zWL;PRDwh{5K+A$e+q`$vQnzy=4>(n8khTaWbZh;y@SKN!u~D?>*7!#WVg}scqg`NB
z=pTy`q=MQ*F5oI<MDv>?G*~NLW;_oH%T)-J^4d_i!2lv{C6iEUL~;a`T}*J)^ZD6?
z)br&WL<a}D2BfIf^DzowYR}`n!F!Q;lTTbzPK=%<)$_j%*!@V~j0&B&r5@<j^M{c?
z?I$h{!2djGJmdd&*%=*nJy%Cp#Lj@f^2ZBFH*MF18+(&7!?Axwp`NdDQEpK;eTw7s
ze$*rb;2oHpm6jKg`4L}(IjC=mmH$A5B^{p|I5?Aa4mQVcILfTkGB?o^z;$wd(45hw
zwzTUyaD9`%=ur;oXUnA;`=>c^<vL!+0REQ%(;jiyH4OFAj(%%L5~v^C*&~t_sNi?;
zwN82vEoeyPZ@qsH%mV{V!d6&_|M#^3$-u_4NV?{`i{w9=MgQdW?6}|V9=MD(uVj9{
z5#+G-1v74ab~4JXUes2Y&DjEhmd=9x;l}DFUxXC85=p=W9p#7a?FtPRKcHl!b2Rnk
zDsZk_t%{0+ylOGsOlXp$8b^^vr36Mzu|XeR?pQ474?jid^nwXzC(Lv<msB?Zu#W~?
zZTlq;Y5dYYN1n#f7!gNrasr&R<9jZ2(PpeCSf}dWk5}0OdDLPE4Zkbume+?<S+my*
zSy3Z}B2>^8fwpu(!J?`z*qc=H4@e8RmJJq)x&jVH5_E@RU}{ut9S%qv@26h|v=&YL
zu0IRLc+Y8H5-Mwya5{77*la%_^~M?uy(%UM3B4*oTY<5bs}?gV$da)Qv7K&J4UtCM
z$V7>-Lw!}mjTynV1y!ASJe|vE(n@B*vGGSeFz~{kf|oKih7zS_df<;lE$3T<tl)l^
zQmkY7L<JXFlR;!v!>6xI9(5nV)Napt!z;@Q!#?2u_*PSf&Vt78<4g@OaDOUuklYJp
zhzEkSQ~0A<KGDP4GqI3phhJJF@d|4@_dqJEh?Y0_-z+@rkt}OjAI6rtOErQc<vxpK
zMU}U^{EG_KyxxC?zyLmQHtkyRrO($wJk`<o20E;GhvQpg6XZ<IU=ig^9lOOk?M%%<
zVE_wO<)LohuL-WYQ3KcuMq{sN`Q7jT-<RLWdd+_~cI_0_lPg)!Vw9m7B$(IJ8EN;i
zgeVczO}wP=v(M4Xzh=W0Hx8HJv0^W@M$GG1&6g4sz_3clAkRUixP=zog3tFXMenXJ
ztMITHaJ1itdhdN};$bLu=Z}IA`1>4Ae;+E^d1Q_|l#&1pfa#>BS@R#-4~m-pwQtg)
zZN0fJFCUbNcT_w!GPycuz$m}7d)tEV<VOS$BWQ8w+9Uso{x^85iTmFG!Q;AzB}#&j
zitEkjmakbkD<H8^3f1u0q4q`~=qJ?E((ZKM!0F3}z*;!zLUd74TYa#pw4syNw@e&{
zO|r$4`vT7OZVl5l-3PKY=O=4env7IoInJeTGtoU9MtR)bNMI`B^lZ4KkCy#E>vD`C
zbRS}<7h?hZ1BSahUYOtTvRRA0KvnNyrMO@po<2;&YCdoKo0Z~v)P?1+XmdCe1S$^m
z#AC7A<ztmI5q|MiTk-!J`e_y0LuiTQF(LsAbQbm<=)+Zio{~UIbv7W$1TKk=^!rkK
zpWr|(qDy6la~dGmno6H0-}fvME;PxFl`E4h&O0s(evz1SckHsqvO+tj81KVsV`x(`
z#Qk|$)su6pTqMtSwPNhuZLk*FI8mr?jQjIy=`WU$M-V%KaALRF6a8w=AJ^v?^`Qe`
z|CVjJ<9?6b-r1r4Wy{T9?wJ71sDXz54<WC!m$I&+q1g(0)MW~XE^yqcQ_ecUM_}30
zo*0Fv`R{fFPcx#*6xKAHR)o<q>HAn5V>V7ZIaqtcw|sw|@$$rJDjbRp1F{{8Me|x&
zBXA4!?4@lfS<$*jDt}699Y7-qmCOS0K6Mr*rDmPkL>O2#$y^{5kD^=Va)-+{(_zDK
zAaL$r-GnR;zTVdk`J6G32@z_jTK%&GlqsRkYx|DO0=bbL5txo_Y*0&kiIR5-?GEa~
z2V(5Xz;FE(o{I1~<hB}-lq&Tr_b-IwoJ*r&JIE!FVBPC=25rZ2(5?9m`TFDkOG(Lg
zVEnz?%yKZ29IG!JO}_Mnh0?^NFD&~uK9Pfl;~?_a{=Y!P`SU!EgA4n?fK{qpK%8$z
zPjIF(!Fx?`CWh5WZ*b;*>gp)W&rfZ7>I=iO9h{`T(9xbNdVucxbq=Uebgb0=X76x|
zBI%2`A%n2ZZw@3|$sjz`2C!TxaJi&e{n?)@D^xh~Xq2jVg@@vrP8_d)zgy8PFYH;8
zdbOcyHucgvH~=QA<z5Z6$3~OI{p?*JCnt3<)U<hhY;Jv>Y{;HSB4Twzqz8XsHSB^K
z0kh*KWbOEJ(YI&)h|ZKKZAyYCG<(u+S{47hnugUj@)RF9NP(`01F&M(g+;)n(unQr
zbTZUsaq-?c0l6x|m^$g0<>U=QW!D`eUU$g6V>eY@H|d{bu_+Gf1@mUUG-335{;=I(
z-|u?$O$3VO+v$1*a_cXGN_S6`YF6{Rs_*=oBDD2C@6coJy7bY4pI|d?)JQrL_C%{n
zt1X8DgZo!Ge7C!n1Tf$|A)btdrAC@ZV$iPoD_O34xirO8PHHJDm2KC01$c%_!5X&O
ztos~+TH3Z*QRO#p(#T$IJFR-0>mRaIE9gA%L&<aph5!@xu|P608o6dmkCK9k^*K+#
z17`FNLoUCv;#;7e2~~nak<V24_(}=b!>h;eOxGOSqyAAC0B-Xn2K^jF*>dnr$HCF!
z?R{&;c(6i%?_V_ifl)wB#Xyb%Dkt&B+cdIIhBRUDtOgitK>B{|A0h3LF=-NF(yru=
zin$Jsql6Xw#zMy>@x!rv;x2XCs-*r^7=6$9smtaxOk43d*cWFY{)e7pna%^2e5w>V
z;<}jg`%J&a1kmx3L$<<kn_QVh!uXlMB+pkS><q+luvsR|)N$MI`felf#x8OSTm6+6
zj{tf&f31c$r}AyTsIW>P+{molo|c+Pzm{>h9P#65zp&@_zZ*vJA+qEBz)_k&2S8?*
zwj@lQ_z1qAVGN0a>cZH>xZ^9=&(1|&;t@{};uP?YuSH%$R?&ad55<3;JDjEN$aF-C
zuJeYuv<u5>o6z#Ou|7H+8%Pk;>notYtM3nntkj=1c`nMgk8RplmX<SwMY@uTknZwA
zM|RAAp9Ukp(1X9<3LWUy8;|O-8eIqGO{<+XLt+QO*^6uca;2{a<x5~>D=$#SCG16O
zg@ORa!3ih)swhXc`!Hr%O)&SJOA6&UXzNp@M)~DoDYd`;P=4fokB{aY<&e3GMj)W7
z!Lnb1L5%YBzMif_{(kkzQ>*@0=1Ole#jt&UVOGP(P`HDQ*y%nnb3M*H>CNM%Z{h!o
zNb(%q4qSuJy2f_eXDDbRp?KXkoQj1<7$pS&(_6{=@qc&CMRN@B+mDAUhj{YefAsQ|
z&jD#rS0L5w2o1t10`v2ct~_XoG*llVEIv8eHYum|S1nVD{_Tcfh9)wa#O#JrTSy^O
z>M#3#(Jkn+U0QFcEis9Ky1nVRNRJO{`ob`QLtb&(ziLJ<T4KEqU+B7gG)n<t4!4rr
z2$DJ<29u8t;6X1*r@C+pQ#YnJ^mme2UsK8Xo{I{}W9+%hg!Jt_aE$U;UM=q<mbQwL
z<4x^XzUf95^S9)eeofY8-uGQ6x-Pa0T65i4$d5m*N&DBI@jI`vxUG&%>xt?=yG=0H
zCT?B+4_ogTWm(W{i<WKMwr$&8W|wVam%41*wrzA7UADSx8@Kv9=iGPR8*k*Fz1N>L
zMy`yEh#50xyaO(~Ku~mPGJ(}aJcLF7LYEi2gQ{33c$1^HzbUPyDSrjw<elUeb=UJ{
zBm~efRgvW4F@UZEqv#Lt(SmF}?|#Wz@d}=)Ytu6}2SswP6KZepbuX5yttCs4>yD8n
z6T9vDM~=9fnJc4Jlrcj$z1(K7U=A9)Imr-;vf*jtE6hnl$w815W4WLR7V*Xatfydv
zm3LPj6TVyebsLvueCux&XZ5pDAyImn1YX`{bba-?Rv6C~E^3;sayX|mk1T5z>s`ot
zoWICSu4sQHRthM0Y*y|r%6}=|A)3zio&r2R_HKonA)hz}NFMf6wl5|zANGB(4WPB(
zR|l*^l(*_vs_$McyK_E<d*C_&nK2gT13$+k4oI$8kh+<NA+fs=H}N^v7aQ*oxE5D!
zs;XtW3&>haHC6C;5*!{pB=H`BNVU06y1pJ_+VPK;5pmf4F0>0@l33#XHuwFx-^WiZ
zzo!9l{&IcJ>on+PJ#n3I>FHQnZWnEp@vGoSlF*ipRQlE<_519bN({FJU{5cHn*y49
zV@{q7dNd}g4PIj75j_~cRezm$w!`}ja^R)6_x$_i8u0~Dw>;a5=6}YfE|uHo)BJ6F
zT=t&&BWa+ub(KrJ2a{->Ooags55m(^cGfuM{L7{1;&X8?*X8M^(ue2ED^*$Fa4P<)
z2aSDj|JjaaTj4c3_|LNn0Jr_seX4lKcLD$BF$MYiKJ0p>hq?cNK0sxsCOr+qXnJjX
z7`ca-GDGxVTqgcqhQ6Zq{_y{6b-8i#Dd0Qs)q#0i=3Ba-qcZ=+bnr#q0<X!#<!5oV
z`VSEal2jD$3z>C|y9a%X0ydo>q=~7;vDrmj0Q+J2J=4HyfC3=EqDjbf8c%GWRU(Ap
z9QPOl1q<IHoB7GUY<qZ!8L$r--C4>}xyixxH>S#=A<EIk*CeeN0TkqOl6BB&@R6|p
zNnuxvO>N-=eX$p%naSOC(DTc00L@bJO}^55?b2*Hwn`T2kYvllEzA^@ca+#8tP4RC
zrts@EUzTgn#s+}<-@sMvH*f{v7y7GVoE<F;<wS0{zZu(iHd>~`Sqxkk6O~n0G>^$L
z@G%E&;j6n7r%bQ!`r8p-qZ8r2`+WmbyYPkkrzt*N9o^lh0H~u3xUjXak;R>o8AucV
z`QH<u42whYO+XHM0#bF=242YYfz*duR+U?|gaEA{73QTCr4HeuBqlw0vNW!V9{!%D
z1>Q1E-a%f2hrf;YX0H~O;8==d3i*}3E2jcCFn@-JGvzrORAmNo;!TmW24DpYqOK!;
zVM?=+{h9N0$;r}TwHgnBO0Y;zhLCRBAt(F8@|uU6j3&+IpC>I^aKjZ%hJ$4*h9W;b
z>klB)CW#xBC?L)Up+qA#py2EpO*Qqd?Qhg-1LEiN$+m=W)uX{>7|k$M$1#}-OpRlo
zI1aK?Q!$wW3T*ks4oi=0k(5aOB1D4}6;6y!W!@!9!-Omq2P0N<teFp0F4Vp&PQ$d0
zTXu2Vq-b3&A)`!*?)iKeSbR*k(aC5)Xalfq`TO0|1QkyzCx^2S6_grBI{BX$>_Y(u
zws1;P8DWKz!rlfOPQrPLa2pfc28Rm;@)4s*kHCnJqK?Be-UhF3i3QZI@zPavE~LO?
z7J;(T!p{Zt%Su6--3xQlP|pPb2@j_9S5Xp5fd?`(7lGWNUHM-J%~p*X&LNZmxJx+j
z0doWv>m_TPSCNxFac5#Rsu<hBx|9Q*f?cDoX5OJt;90Iyw+v)#bdZV?q*LK+N3ohH
zb6LEp(sJkG8K{;D$9G`cNM*YgSPAKdaL{9N+=V!g>=Yo-PQFqQ`1GhqlMewM{5+}G
zub-FqRpmL0eMSUy`acQST_+O(eufB-ydtZ&I+vtx-$kNipXMc-3K$v)x#1m5=7#If
zLYZ0B%tbriP-Nl`cs-TC<5!peKu=(;Cgl0)M(upP_z0J3{aw^0m4WE4y*k@x4_?js
zkR||LtF&mD!ufL(^&Hnk6H7_%Fm}U_@sa@)-8D}A0_ENP-GH}Q96+h`P6)7QuQ|_=
z088)yA^9E3Pc5~LO4yL6*=`=Yu_42pEq(rcTdivK6ZKu&2_<Ac@ieOrX^ADS4Cag2
zPs{41iv?ugwBt;iKUg3dhrv<hvc5Gjy+7y6hkK-8j3<hQnsNe9qgh+2>nVnOP5cHM
zY#-70I3;g2Br{|QV*rVaE3bp3Zx5KU{J%pA&<K)(c$RdlwI7p$VkTN1)Htq!I0`;D
zU7POU<5O`wAsuI6FWH+s#0cmGDBI2w1I>CoK9ttJHNmAmjgFljoWrBUFtSj*9*?ZT
zyVS^m3!cFVfQCt4IHBc4I4L{IJJnDI!;TI`tphA*6IP%zpaB`uUP{ua3PKrU@g3q!
z-Wb5OfaAV}!0x#Xw(rOdYt6>2D;N+3({mO#1jSq~&?5VyLU-e=yC3M$*)nXe82fr$
zPn8xY=}x+_S?2=#0sor$_Zip2Ft`&{Ye&dE7t_7B`14e=MJFxu)xdci&o3<`q|`8d
z98;N<YzfX+1ppPIHhd!WM=(Q|d(IzIEufGqw9fFz9N_4sOri`So}?&*FC`60MpzK0
z&wske$PhN>_qdHNPd9+$(Qt!A9tG*Qd0wk{8Ikwd<VJU2wR&32qs=%)GVzl}Ytp9n
ziC5yYl(9Xm(cnKx=tbPd{bC_(R-C~ZE&1H?EBP$|)}H?ms8O)vROte4kwH@D2TyNe
z{}kbV$x$ie!LslV{Nn9i?1UDXOS{$5Roc@L#X&%P4U3wTeV&h+h6Z#L1RhHx4T$eE
z$*ZeS-d~|gvAmZXQk1ux-^2at)W0E_Dd4<%fBC?uSe1bkP&GqEKElpfaQER`vX$?c
z<oJXGC?XBt$X*Eo*~U<IpAgA@^h@kH-8o6i>Do~~@o;XkQ&0vAV(}|ivsuezEYX;>
z&Hw0ip6~7Mz_0k2?gi7XMcS=^RMnnRuT+*dYXZ~mGK`dzp+&Zco6}}Bl}k}o`0a2-
zik>x5Gb7uRqW_#Orjzg#JJo!*rX?v3@1F+@@QXUAt1a68MRfS`FRV?-c3u6KqDwLw
zJbj{Q9fIz}2a>{)-fC9|6Qmc1rx;nVj&$ZHdZh0c=mk)HFtDcd3w3|E%wOHgHbZa%
zHWC6}Zw?73lVDmaWIi$0r>A<`KVwuk{v4uum{=$o`w<KKAxG2RWHv8THLWzg^a4EI
zdWO}RR)WG8k+T@dl#D(Hn?-M7Q5_NkGNbq`{rmA=|35+AbIZjD{G|{^pvhOdBc#)Y
z<W>hLwX^je{z88WXe%NXr^B27dgyyHr4QtCP4ly@x-<wvVTKP}6>3kRL@wH>;+75R
zC308)5SO^T87Uyu72>9XY87yrK~{<A`s^HSTW``26e?IKDH4k|io6~A-7?oghXC_|
z`bv4PIQK$ni9&_v-s4>u18E{Y(D~JNhZ95i+t!3+z{F~Uy94*;(yZ3;=-58yc@pLA
z+YolzoL!!nHa}7(%vm`v@z_Pn6eB0iv=`%zr_6XNIRQ%GF^705BiYMcjz^VV#9j)C
z*Y%*D^T74y$;i{B-&)#^M+3b6B3Ci2VqkpNXJ&*W*?-%dca(a4H(n|>RxuqRS3x0Y
zvRHem-6#aj2|z0}XC0|DNM?7GYFt%+XDfFYop9mmku}6kh@m45n@SwZ+E4S{rXDXL
zD=7Utf4NbHZ^6!ov9X9Ozjabzk0wV}Xr|LcNp$H^oe%%)4QunMXom|0B`tMO8#iRx
zv4nUNmhViyLrE?8C$R$^O<w7e@J%?H7?EY(YOVvO-?a%peop%|g<~8|8oP{H`2qVX
zkxj>;mN7}DvBoM0`J?v9OXRQvm5yCC%CyaJ0UDSZE?CF8<ON7}JUg}RU-+G{SR9V(
zV)u5Yfg=W>dVn!&5l7itdIs&wt1mrS<UvSk5ua}cY7DCq|HpNFU%e}YK9Ec$3=bLp
zT^f9ks;p8*Ht9)eOR7rAZigJ?OTpiXNU=oARFBBJgze|4l=I18{z=drSG6m^w^Gvf
zmVm|0m#92#>Ry6Xeqxbt_~-ckw?KAx>0`dkoTbK$_t<`#3XQaUHJieWFm*V9^V&KW
z@WA>K9(BYjdUR=W&W@nbT?zBvx!Co`*Z}ISU>Gm)tdr){ldQx5;Z4~2TN6Ovsfn6=
zCP&bl_#F9gfts0;W9~5K=>-#TTZ9|(wOTKbu9TU<@Fhz!4n0<FZ^3kH8}q*RU~t3!
zH}ar*deX^dhGX^XIZ}11`2pEgs=q$WdWJe(lfzNJM0I>yvkQ>5UUcn542(GGsMj*~
zhd*ebbpYjL3$FKZeLqlbo3Z1ya*_bQ1miI1n+9ULuiY{zz>K$1eSZbW&yXsifZh~F
zht><H3hcv$VQ#WVy!b=vbM`$2Ce$tV7z1gI6&co7=uGCLDslG0yP$S|vh`@)t@*BD
zx;JE6+vm8-=y=ldm@mm$G{wX_DfoB7(ACNobceD;noyRFE;=-aL#lLwjwh$FF6s9w
zJg-vbTCQ1CMqP`aNCQ}d{i_7ixBr$r70@<jamQk@QRU}!O+7A|*jkFRSlBXfH=$aH
zq8r?c^M+lkSfG_5*L}ZFqS55q*?W1PQS;sv|BJj%(3#{@>I0ZK8&bpNzn$)QS=e_q
z-Q3T0pmH=zia}Isi*Q=A`7x^{ya6Ye+RY=f8fMsQhEw`L0azz1GQpu9ZO+Fzau%&D
z7!a-mxOa0LNy+53z+!V{xdfwOedMN7wwybzEZ&KH=k(qf3e~tvsBdGlr97zCa+Dp?
zBgDK65`g)YHt!w^#Fd12qp89hG!!M=;5~6Hg3|E6bJ(g1tX|=`W^kV4ST>VbKzYBg
zLCxUOm<@kB0EVLad8Ka$sYQ261~BXQ5H_L>dXjb}eh7jl(AMb3R8uyd4<okq5yk$}
zvP|(G7d^bS<9>i+Rjmt$Gs~*`B@#okn*5IjV#0%n=XpNlBr7QDT%Df#$57wj6ZjB(
zVlR9Y*2f9v4n(ufNZ(&<;WAl&9QNljWC)sH0}!VnfCM@D2MNqXA(G)hFlsoJ92fJh
z9QCeT$-P$fk7+ayNX0}qJD>i*(BPXsKd1I_F{7VAz?^b$ap40)K&U}UC_#iK1x(;T
zXfpTG!mI`ugJ`*x2BX`qcULBAZ7h_ZdPu{fX6^wQ+F%gx!9t2hR%z2(A8PUzl>Rtz
z6%a$bfG^Gs%trn`$P6GC?2z6{k|zkuh`pIJH3-5>fr<~}la2e(@gH9tPQx>TNoTwd
zDF)Y|!q@y)(|$D&%jn5F!eeMk*a?C)HMP=TW>RZ3BNvYql?d<NX<Tx-HLJX!#b3gk
zj-7C5s(kNm={&O7xLB@v`}ni=sr~Q~|7t1#0v;q}USf6TSUp0ni0l6Vggb&=x#C+J
zsS@_Vy}Mfkef_&?Bqi!CrTuN%ZBWUhV&A88`?^)Y-5tI}yLAhJPwI9ryzpyF{1?S%
zk!`3&$nMSD)^Fl2oG^jd<4N+@V9k=7fvshHW{1*#oz!%xPokUnL#38&Q^C2bbAVtc
z-D7Rah4f`L);ct*_2u+o&C{p-JNyiNeXo^V>Ati(p|rsb2Bb4ux-*(@!(SHN;`Is+
zmCaw|bn&dY-=on#4XHVDMG&~=tjrc(cY+$v4o&mY@io!ckz|cesWhcl+jD+`2#I8z
z-64_ER}}^azqo-`|Dc`9NGJ&?DF-Z{8uq0dUy(eIZQNgm&}FVA-SXCfuQ_C}xY-H#
zb^Kl4DlDC1lBs=&TIP`mZ(59Ta<2?=jo#WhG<HbM!&CRfPI2zb_wb}`Gyv1IP=TP?
z&V%^Xm-+e7tAId1ct3q=TO6td3$}v4oPj=|_De&JLH#paO-s`a6-}SRhYrva;_TE=
zwwYCDB7EXj8=}Ga@y%CA68`$;8a+f#Y*M}7Hq*S_UO<20J{#N#JlBE^=b&I)hA2~D
z7M1F5U;aG!<}bKal>Ai=u5-l16ypqs)MRe6()l0u!XP%y6p?Dnk7a_g!7cY(D)!Z<
zEt`>1u42m$$=V3wE`a}z-pRO6f6XLaP%hH(N^-h0d|%BZ{e-_>IyQ#SOwA<b`S?(c
zToe!XBySxpDkqiwpLEeqmQM2rH&Nk9ga1A^DD%nfO}4{keM7^8W_=6GpqaqTYy?%j
zD{yH>%5~UGXSar|hEPPePl5kFMdXw}w>5;Ea!!#uz%l_G+Qt~{O0h1Qxo`~%5z)q2
z&Wim-Bw(!d`w6sET{LogL*aoV*9Ix|r13Ij42h7rFT(`upmrxR7zdS3sQKUb<nt7o
z`u&VL4^1a|U;YMlr(w5bavzRE?7p=ha4qq;O3re&JfA}xIJTKq+g*$dI9f&1LV2#3
z7x4iA!&+nr0XUy<$eu>$5)FDdpG**g|3VS-6k+x$Q4%sYte1-Hh58hlhN<UpwojLO
z@V?N#(;hz53NX*g2$N4VYAc}*9RE_OR=OaQQYq{9uJyu@QYqo|pBM`9KNpbW+=T?i
zBXH3iznv3*(|ax%FEQldH~&Wx`Bv!@5DGDlJNjg<(8Kr#Zghx}fq><vyf##=)Q>qC
z`HV)rZ3%N<yAe*nsk#hU+5*U_S|<g#eF+Y>=W0k`UpwA<xGYGQ)=tUaUMbF69_H?w
zmRShpY2lKRX8nMZooWHGuZ@a{*uM1rMLKhsmrPksgeQTA;qaQ%jg%c0%$5!S@r7p~
zlGX%Gen0X2M1=lY$m3O~|DdGNQi@B!m$|u7J%HTe!fwjD?-by?7wnXo;G$rApd4Y&
z*P188L(Q19y8Q)`$h(&|OGebsoIiSpPhTpEqhaL{|63gADUFRGOLD}UkLB@QO6bTK
zH2{naWflvSiGm0?8jKBV02SH-K$ges&c|~uB}54fjWRo`L_lB4dw-3#MOYu<dH&j`
z9DIy*kr&W#J_)zUAAAQrjnP_p#D6tJS7w-VUmq8(jq+donXbD>p@Z;hdFuLe=4NaR
zzA&y3fs!rcRLYQy2Qi8kcJk>3-mAfZ=ue@QZJtsG#77&bYC-RUJHQVpFhbR<{Y1NT
zNF4bD4WK<OlY0FHQrMQ#1Qop$%HJ8t<sT-m=bk2AolHRqo}uY4^m@D0J1F51ZHhj9
zN<dPntPYIP*N*I1pV;GApUCBfqh#XGh!FI~&m-4}JZN})$rX-kXXLD;KV-(`VxC`g
z{pO|`3cn0+#8fV}r~y$AV`L`!k^35>l9CgiQh_PdxA^?T0Y6Gf9eBiQ?GoC{JBKpj
zjtrWPlH=vxc-e|r?{;SKQl~*QTK3mxLrtZK5R^R&qID+(W|Qi(S~rvOlkfKAU$Th`
z(J7Ejy;?U@`WKDrVKUv&AG?;1%jNG~Gkx^FE2MIVdU3xQ*Z{v234L!2SnkWG39^ej
zHr(E0uD%`k$UbIm;`J}C*PwF6`*^bsKiua4t7Mbyq$n%!YI~Y3LhKZF+RP_m5x#<s
zeju7Eo9k3o)1J+vm(?0ylZpvara{d$(swzz+wGid6BFOkmcOezje@|bGhi`abgdJx
zC|Z9|-VQf*4FJqRtxuXa$AMS&qv!`ss}*^~gMZDq;ROx3iJMQJ6CZ`vITm|74faX*
ztJ^1(3QZZIl15S{&OJKiWq$mbmroT?>a3eU20uj<m2TxKxJtST{qWpW<dIF3l!K8!
zLhYh<EbHGYIqQUE`f0GXlqfp+a-YM!#78vl%++Rb@duDZ0WFu)Z^=P^w;II{!YVxI
zRggoR;Sap8Ke(P^tRAnVj;u%MP;@4%6~$6|bKWt=O5C21E8FK`D3cL18b(cBQluq9
z3!tXftl}6o)wW~cqr4<Dw8bH@PF18yxI%odvXdF<z|i&W+|(o=yBN|m!-;!Q8Ush!
z|C8U&?F@L+I{xt>&D(LBs=|P`ep}q^x!Mq+Ji#_cXCLINIJRNreF`m0Q0OZb<Ni{?
z{*)<D(M7Et8zZ!@1^HFUM4CuODto+Px98I{I@m>h`{IB`yauZzy7%Les)g`I`=Vi)
zJ`)^zsu+R#*i@SVxdcxESzx~PuapZx#0;EG9TvbYZh1z}uZW?#>1;}yrBsDH-1yG}
zux9FGWv3}~M@|S7Rx&e<F;fV_wKdbBYelkumoIQ&9vI{TQ?C>m5vSFI$?qaUdZO5W
z0jd}5l65wxV)vc8O+ZQ0;C3h19|=;cjwy#J0M@T!pqv<gWae0+Jj`|X^ypHaxpBfP
zbHKRsU8W^}di>NouBcT7t-H;<6g8Ol#K-Brd5f$4Z2947;Zs94>!bK)jYzav@w}}N
z;u@YFN}cg;x%5|;y1z&BP!QK%rl&MN75QMehGo+t=<Xk$>5K*Yv9356Xbd`!(eF#q
z82BnozB{p0xGI9H)paI?u{oTMSF>4K8vs)S{92Tk&gd5~o2zwVnn=cF``-R~aNS^(
zFz%#oFO^5<aR2jTXojZ6$-b8%C^{vrhy)&NTj}c8VRH-11nKBgXyIc1Vt)hA=w=py
zI^gU-Nk?vLkr3~pN096*Ligf*)9`04gZ_JbaE+h#O-VV7>~R4?wT96b*<m~28RoYg
z<l2;W^<FKoP<@weG2pCR2v<>u#Lt=L<sU^kn(5bqWX9Q5EVr*+V=cY~ZX4^C9n4PF
zetzMVGBTNJFjquytSa1H5RM8NGz~0Xci$9_-IBlUQXGkJrD!Fpt*es?*&j40YLp_W
z??(|PnYKIA9-Y)g;(=3=CjeRzo+-(9HylHFDV43k6kI|l_3~Wz9lcS?$<TkP3Cwn=
z$5Zq-J<;vqWKTj_p#P_gt*=^}uVa@x%&=(Uo%UOkQ%owC-9oyduc1^wyW!7O@Xbtf
zRZjz`+RB%7HAKpNR5fI3XEh5Nl$lXE*G9g^tl=NIMum{Q0nytffV=Hr%O7GvLkwct
zpUJrrrcU0R<G(2N)fKRqbYA1%;mSUsU?&n*VjsIq1oPQHx_rixU6%e5o=ThvpLqtY
z@j(VTMvEWV-pEilWey_>--nyE?#P>qlHYDdEWDd7e-kH4VV+?_b8JL$6Ba_w*>M}v
zPoCo~ZbY|is7Tn?07z@Ixc8K7r2652A~&ga55`*>`^DN@eLB8o1C)ra8QcR=@+5|X
z`EDKE3x_#)9m>7xoReEId;SmFn|KttpQqa<mwG@G_GX{CoSDS82%o(|U}a+F7Ze`G
zoue^IMm3llQ3FS}6$Zg~=YFM2ZoLd6nghvLP<Sl8lC+3zfMGMLFaoUm#~H$)zO<Mj
zkXy-GQjoh57j|1}TOeZ`$W|e?wMBi*=le-E)!~0zyz3oR_v_TP>r@*z`T6X=h|VNv
zT@q{rDc_zl<+vnr$E!T(X&g8t{J0ITlrrF~?ONeAI9IR>a=zKjR5Z6nw*j0Ww>lK&
zE5o{^5*H0{WjL86m%9rY#fPPgg=S<BL!Ex%L+zQ^gYP6<v5zTYyWSxCr;?f3xqtG2
z-C0X83qcYD+n3(CQn4gA+$beb3q~8kj9SH<8vWKeK}#mV){w=FdPkdj`8mA!_kP-w
z44Is@`qtVrgcXjJw%EKrY8}=O9Y5zE#roUp?+8`^Tq_;1b$#Du>qd`B@8e~q`Gh;`
zvx=7Y5~<!L>!r*K4lY~Sj-(#tb)YSK{6E3h21Ggh2Aw}80Zb2LA(wy3W}9WYJ(p03
zUB;aU_B4W?1_*_%=X1haB(P*)PK(|B19GoCp{Jj8ZR8>X-PVseyZkny<83btbHHd@
z&g$|38;=B3{ijAO%v_Y&JT@s#<(Zfj79odi-{hNgxc`TM?^s(P5~n8u=@iRgbFuZR
z@UqzZv{09mlEKVRcv#hH*^9SQ5<R>_KKdMJv=(FOjU#fS^|w-sCk2l+ECB#9GePA?
zVj+nwHrkpGDymn#RF+HWLL-b20)RnyW<1!z8doXlJp+L(=gBfM$V||<&fD%$hPU62
zqf0S#2u0oy1}0rCe{r=!o-DKvD_XySo*Q%tmJ9f)&0TM;qd1hMc(qy!Y2fUCj!5L5
z_?F~hcM6yy=A1tQ?hj&Q9lZQj#z*TEh-=b&g2CKTz4Dz@iy_XIoiYu>2QZx%<rC=R
za;Kb8@p9<wbdiQg!zJdB<W++A5DEkRnX!9n^x)KhK-(2uaInCzvl=YR1h)0K+g-RG
zrp7<ZXn*B3b1u_B?)Dl~R$HK3s0!j+`qo%1(<svj^{mnhm20fkN^>+9YIIoe;E4dT
zUGU%wzePCzQvwg7T;g6y0_cahWNC)Ju{ic*xi@78vA!pe)=r%0<IL5T8yIB{l9lfK
z#<^mwMV?^x{2YeKcNPiv#N+CX{YK;iF){tu1)o=x+Gji9Mv_MjWw<+&?jB0=%A_w)
zKi|Qn9oV!#n)Z}`a^YO0sRMmHYx>b)Nfz!}u!Tz$-kf<FSlMGo16rOTVk9O~yCo$G
z7dc`iB5T6Qj9U=eKd>%0=;OJKz@#UkQ@h{~gSAj)ez)tW6p$@$&{HG7=C|u$ovMrC
z<6Rd@E_|?|9N7_jU_&Cl?k?j@Zw~DE4ecPnV6`9F-82RC#&}{JYSp3hDh_7tiy`T$
z{;Yv+Bavd+CrZWD1`HuQHMZr-i*ansn8}w({efiQYAYQca<qhg2{hUZ8yithK%^eY
z+=bvLIezgx=sL)5c{LJ@Y;`vmM((;X14Fbo<&=j<OubJ!(?Oqs6poDGoq?<#!+&0J
z*;iOYwM)Xi&Wf4%4-@<Kvzc}UdOhIf>qFhpNzgsqk{%Jz=0JZdByVNS@bDTjX5c*K
z;u^$bkI<Uu#`IGC+qTuDc(ZJAKI04tnaay0&vlE>_ST*SKWh#FaJ46f8(nt7A?k>#
z<^LZBl3bhPF}yfev7Y^sLVhSIAQ1wiF39d2B=xc1jzTOq4L3LJ2WwSEF(s~@LnFmP
z$Y8z$8KBP#dubviCGKCdO&>b&e9dzcELYVIsiEIQu0`y>l|CK#&p(1BIV(DAPX2D9
z(qzi}>4@)n6w85P@!L}NzG&4<3Ed2)JeF;_R^_B*0wmF_Yn9_x+iRe?|9c_#a`;(T
z-qTz-dkiY7y5Ha$JJuNr4d4EBWHMI*G#GZ##F*6CZhlO&jT9#jNZUKy#ou4PW`9!(
za<h^~P3&THH(4f1IEu-gZgb!O8e`^<F}71vjz$%k2k8e%Qr)9l<h?gnU{@A1vR6LD
z8Xx&Q&!JB89h|51C);q+M<s%4HLt>y(}K|6cQM_IJvv>kxGuMJTW^1#DyOdiMTPe-
zPwO<wFK6RQ7b#zCFG2NJt_#8j7Z3(`8}Gue?u6^7J1JD44<ES{-U5m%1M`IfUK`B%
zY`(L9-QUJ&+XN}+dhfs4ZEZ#|CM%B=?aI0sFFTThvt|gW=~+x4n(4_M05Y2=?NZmM
zVEdKiS6B#5?=55iX6w1=1GGJ0iXm8|@*0te{En33i~L#;{VE$pgL{U$+b_?f4uV1x
zKVjbq0}FAoU*P_6h>SV)21nk5k{3Y@tV_?nlVpcIBOOhu$C~Z$!H_KSHwX88s?<FB
z)e}-b4&yL0=sWO&yc#G>b&%JpwQDuDNVBQ?8wtBN+Y7UZ!c1qa&T$2hEmAH=Lu@}s
zvDH%1d1Ny#$TZc+oipUDz4>@cT6-;Xc-#9HbcqnZjq%8>r!X<oAwA^^CTA*>E|b<_
zd5k~A5DDydzWZ8zTRLosF52v6iFbP<-!8UP;gnI3E|m(GQ9F=g7~48eNt6-ZR}x7F
zYXY*6rlkD(n1ln0AOZ%&6*y$3n+uxQZbe;|Ud_*cOq4R_b{2GfzU-w6wDV9B*47Bz
zba4Afn?miGV_$Jou$I<81biiH{p#=D#R84xX}-B%V%8Fq6ZCP=a<PZs6y$r0g*0P5
zH<}Uz3)fC@|C~urq5hoCiYxoF)prU3SDwcFk%zEy`8?%X2ATp82M4S_)M!}<e@j@I
zZ2ZX)jke#5%+!FIL2&ne(+iVI2m^bM7Z)44eB4RW1FPOIKdGyot>x{tkK2@Tbt0Og
zR+f-@D-EKH(>-36LhQW7Jmv#E`h0j;WN&ZNRVJuuTl!qk8Dc4^lrY8mh{?~4#WcrE
z6^u7vWSFV(BNtS5|G>yQ^Lx5K`b>WwW!%i03Oy3@Q$$isSmFKhTmE<&j)v+hPWffp
zjWF3d{(Pr)YVdKE$;b(__q6AT30=vjjl115#uTMi^f&KSvoq&|DOG_4oE$Lq+U)H+
z7WFlO`GT1N>TTuu(~;Y<k#XDR#$ZSAj`&$4(5wss8|$xQVsIhayu4KJy1>TSk+BGj
zmaVVPr1=tb->DDFCO!G@i?mxooIzs86ZcarW0w4yiCdA>?kj#3tNN>M?c8|a!}1RF
z6WFzjb9e2|IU{||Xr`T5zCb`9r3<3w<@MaG?dB~GD=SQV<@sq92Yk62oriA5TM>sq
zL{#K+r{zaB6$`n6{EoVOWWv;5CFh6TViNiZ*>o$!EeHrFEPp_bNd=|VFWy@b46up_
zY7`Jfu$-6q#@RHA#57wrM*$80h8hcU6zV$|OyxBv`Q^eC%F7)eA~OJwS@wL88*i=7
z{x2gAl8~V??nl!4l0==;V}JCkPsTQ&BuI<1HzRRFRy~r>-*l0d(p9!M!x}2ARm3lL
z2vjc$_{*A(3Kb9~@ISpwMBpQ@E?Ck=3>63bLbsP)Mh@dX+WX;Xjgi_pb;F5!ix=iZ
zB=x*ShV`JgyEydqp{@XXJ*fDg2-z#!miLFra$yv4f==h*aa(a~%&@;m#Ci<gb~xk0
zuEp`*56ST3M;LB=<XZ-8x?sc<8X}eLfzrtR4aN{FE?noU;zf$bU3rRO9zLl`oE`#k
zDk5PoTE<jtt$(?UW6b3eQ6U10Db<)L)uhFXFT*5|QlX$p3yK3o5@5W6yyJfH3)SZh
zSQXpW6UoGpm>Z8z`AG<i-3EqT9c`%~#^1YH1^*?tIX!^;ZtOlN%}|S+wNf*#_vZ2x
znk3!!NwE)^(^z{Kvn{H%sb~+rY2Pj&{{`mV;M;1-m4s!yoa0)JwRFwz-IHgyvGe;@
z%jeXHCS&F>%KRQsDDU)8Eo^tvm)7y~(&t$#mWKgdw9pKdP~Xo`J)J{uk9*<znCUoC
z%c<VW66V}RD`QVF0lF!ePz3VF#L=qth8rd-_KyHU+7>e^MnvJy7xZWAXn;qqLk}&u
zNM1UU*aSP&vh!hAk1*7E)Yi0_5dp@Ku(c!9h|V-Q`2;2)3)Vy)90q5sJK|mBbfQ_|
zYuxxI-W%OYz|=|+{TMOY^8VEXy9I4d3EL70ld)Ya%&Wva6nzxR-D790_tz~6$y?Mg
z)9F|`zlbX`GC@UT#@}A1EOZwCJ(7q>DB)WghJi^7DDR)P$a8tMY;-9Y^03L`Y8cA0
zYgG-JZREcIOllHXDKL@D?&ThnW4%qk)z95Jy>q5qz2$>meW0S^w6x!AJkJAF(nHE9
z;V{yA@in)6fHgtBS6~66{3znLCKs@h8x?O-cM@uYih58=$(6M&(3}uZ-WAk$stLQ%
z&EXYmObsuLh}PX>#S+s}szlJMMab(U!Kx2t4Ex<y0DVl)L+PU%B-l4X=z+Q~k6hEW
zT+vOnp7;x7m|8^EEMR!Tr5OdU(kg~UX9SblWX8*7{};8_x|!sdbhQ)XsUN?96@3_k
z3;s_6qzc$9E9>w?b>MLu0klLeatNZqDG|jyO0AP{?cJEJxbK-b6xk%i*J3;Ag8UPy
zJVgM{M=)n~5+SWr4Ak)NaQO2!>>Y7PZbUBH6!JYGuB)tD*M1I|_6tTMA3TpwNM{?7
z7DVTMExD_27}G#HI@i>X6Y{&F7Gqu27F!NcviG<UjFTeU<sIY+_egf`-sLT`QO|7x
zuwxn=#}d1YR*C6h4SY{gk%h%S=!D>NMgiEtT6A?G<;=Y~f7W|(1`FPR2@EcFMr8FJ
z-Sz#)-r#8sfDOTchANa-50GdXXyGU`>MA5-^*bf*JRkRdo?jf08?6jdgF9)NNz%y4
z{NdzPnH(+fm1@{gNvA`P028NCjaQ=g=lACy9U#cFQZ;H8SuBa@C{|sh__j1a=`f0W
z$FOSfc+!4yxnb3JT~ifXKg-@-d+)C%e97&3Vy`~EMZ>6C%<yzs6khgK?h~Skd+4~Q
z==eUv=GPd+#@IaX-!YfLL$LuJj@*yb--kmch<35TguCS;nQ*@!U9eob_)(X;8Yknq
znLKRz?gqZK%cq4amoZCDE|&5D-JKc{p=0wn-_9#2Vv0{lQ{+7BF}4Ubcwj@M?|T-t
zzm5`on?aeoJ>uBHLC#)y>;p2b$V%pKu5tg3AOY|uTt1N4zQWUMC}JtNzXaq@CdPvg
zHR*s71bjc1ON*KO-<&tYLpx-92Sw3p3OTM}Qd>Ad&s`R-PY68p$M2DV>HqBW^f{HP
z8byXI`a9K71fImi*0$@*!Cy&we;G;^CaxExaK6urEoFHB(Y}^!BhCI~%qmLyz6=4r
zIzT$+&d4L_o2n9ka4z0bIi!G`J7PfbEZ^-tNlZI0fd|GW6z)PYvW>vkaKO^G0|o0(
ziwD;<lV#zB1>;x(0f2N(qAA*fSBC`IR%7ep-#<NcG-O*@d9>IT{<qz!ez%>SLCine
z$%!T=AK%(Vko_Fy0M!*rwRU884QML-_sB^{&Q?6_A)o=CaY8Rri$nnn(Qvgp6icjT
z?!!&~0ghl>sC60U-z=fF11=WG5JYx~a7;tC6cx6WTvONp3fm114ckHLR=7T58_^Z@
za4Mihj`nMj1q$DVlwoA*2-{mo+q+>7F<Ta*&@y;s9r6Mov~@C2I3wHvj>n&+i<X4r
zoVz3gmo|efnxCV>;Hf~)YNUcl^S0XaH=2j(4<{|sJh~YOZ(z0^h{p4k%U>HxdWzl`
zh0gWK#}}u7ioKy3e?cduesV*dC?A^`yr3%)w~z$`7u!hX1hpZoSUC2#Ke^)P3Q;#{
zy}s64$d8q(rxk~*pWK-R*hmZvsA)NtI=D3QDM!+#(fUixZ1oap5h`GD<KZQ>10=F|
z;?PPx9HY+O!-r5Od6q~M7Psvt`)NKspL4aGQb|Jq*LP6e0I>$VDVw39b9oR;E%iW<
zyFR~aO(n1=#G;)~RiWKW0OTd%rGcGB-3dqV;79kJvvQimm>kEX23kOIqxO{Kyj92H
zS8+fmr`_h-EMD|N^?W6A_RH`ak0+y}f5{rkjz`g<qLw;aG*}EVVIXW}sy*ub?UDOB
zsjE3aXDnnfT=TQpXejV5VmORE<$EvU%i0Q@MIWmS5U&Iy*LJ>W#gH`e4o9&(S{n27
z#W}@0@{_AGMcP}|JOb4j#}jMdA>^4nIs3tAT?a9Z&M?#<ItyrXKq`P|Z*wd>fpwVo
zG|BtaZEqmL79UHsE;I#sH!d#Rcf}LCEcyVsK4oFE38igOwr}~Cn-GiX-+}^gc^Lqe
zJG>n5F4CFC0U$x_os;8=1m~J_YNk<-{4DVbj2DA1oVmVF1$+(7QFD?uu}S3t^Y%qP
z3gj4$70yzIuL$(6ZitaWI0J+D#;%$e)m;(S$8SL_Mg9<Y_*C^Zg83yB_FASvG)(|m
zPx(U(54+yvl>2z3&m<of!9=LmrJ$i2mf5>gxF&yU7B~YwD~3{fBKU}(OE`P>1`B3B
zsEHHd$f+~0E*AP={*Zy`;AB+qSV*^>G#pWDR6$O}vsCXhm7e3lSq7~H{GEr({GKDt
zb=i1fZ7vg)YqiN{uy8WBOgS`Ypd|oo{#QY<0TVsFI<$*cOX;rzF$6M=4%FT3o&!U{
zip<&kvHc%nd?(`-$8vI>M0fcrutlh=Bf-actOmF!)>Eo)Pazl~Q!E+NWL+!Ke|pgo
z8^r>BWyii77uj=VJ7zmxExeu{7p^)j@7Ar%T$Gp)d7c@GiB!pmCM=W`4<P}_;tE&&
z!ogvaN4V^6O71^4!>cviZFB3FZ9`w61`8R=)8*G}kSOr%^gczlBe11-vNBlfiiep3
z^ab5KCjz`$TfOTNr?7CI>kB@<M<kvVoe;`$M*f-OqY~l^@BlFiJ_qUzMt=9VB~Fqu
z7ur!c0A`PhVu2=qba$IMKp+P|JgywZ!GvVH#~?foZJPYfgXTZD!nel`0&eaKiaWBG
zFk-k5J1`eeleAzAI2Akawo<6JPx-JAPq@?I$<Wo(XzAPEPXp^TklPM2-i|pLx}NNZ
zT%`E%TvQS{`zd^OZ)4!5@%s<Ca5HTe#koNMERsEEn>A=pa;ch7IhF$8ePLKiFjcDn
z=LSldnV{TX)%0^df3Lw|2bT1=^QzM+c?xnG5(Oqz%3+zA=FDe6u^^j)*dSS(|7Y{Z
z+u=x>7R=w>OMzfzS0g$38HncRR@@%pkM-5XBS8jl168t@C!Q)Auy!cnEf;%GF!;bn
zv!^L@=U)=`ke9dvRMl^Q7#WrgPO$dVOWY5JlR(l{;6sjG=h8^(4HF6;nakeMA1)0g
zd-NRpb<N%`G}|Oe12$6s+|spe7*@5Vw%I%@Wh1@Cdr4m~&tab1-Fbw&)q^}Jo-}0>
zh8dp;Zm<VSLv{xBC494hm<h>yESegf9zrr+vY}W_lRsHj;wK=nsW3#+uZW|@)5Jm=
zeKnJzm24>TeW)n|u?4qnFBKG|Cop(M<I`Xu%08SXUIS}DM-F@QXR*vmUE?fKzwWfo
zTw3;){ux!~@sBY$uANx6S7#RHoI^EowJgB2Jh{dce#x0{4bWlXlV2?NJ%QI3OuZ+<
z(d5kdQ=<k>J~kljgZMg=wm@oCjj~_BhEZuMwLw10-j6=6x6V?Q*hNYi|2D;C)VSg=
z@6sxv%X~i6V){v3(F?8%mA>EakTX5i|7a@z--Hx8f!$f9Wx30I8-`7>Yg93L{j>b`
zW-{b!FpcjW%Qu%Z`qy505QiPADjk#4fVUSQ%2TY%){~DxHwFXu{S?s{6rERg-jhtP
zU~<ft!UuHg%d;6-q<i!$7VVOIK3*QXCWF2~i$-#UG?u7Jb09(we3-vu8@mL9AUu`w
zh!ky%_>T9{rH`J&56o*Wuhq|`cHHe=%7yWDSs!)t?kN)aiZ9TU{;`r&%M+cD+r9`8
zsmy6hxpCDzGeg!}nqw|Zw-bw}<^b){F8Rr#X<1JGLZ6rdBFUIJv~NNF^Mw==T{t^v
z8=3ZZ_>nNs?cF$Ehk>33T&xz2A6}KgN*hdC@de%4IGN2+mW7AQt)XmH?QRa$NLS}m
zhOu))U8smxu!`i!r@NJA4tL3liW)T_qlW}dm?}_de^C<L>FP#B&e_jCjJkw@o5J=s
zMrouzTf!w$BY5u#iUe6~9AC^ST?CGr&@YUb_Te~;xUv{Kj2KxoMRZmdZk6Oq3@)8J
zk$W&OG)ZbQ)!J<;AA!TlO<v+BYO#PDC46877ZVLIDVL~k_O9#&a1YXK|KuS65{0OO
zPs}0sm1p2(Qlylp%Swy*yb6p8xbV~%%)G3YHA|L$2NoEAlCD9-mw0UfIfw9^;NEMF
zd%!>2E0)1iMUIhNU8>tlTK8_6MG|^Ywm|Hv$qf;KR~!jn57Tw)=NGKyOf0TXm>RT7
zCTn$N3MUolX?WWEaOf95_}~{HCTN9-VHxhQTdbE9VnD2y^(Z-&!i%P_@rSC+;?lDa
z6o$Ec$z9Q+-BV;bx)y@sTsh*;NszLc#8G*(6Rg%pcNr|ARu>z=X?R<trT!n5E7K8%
zYVg=A4l4*&G0u|*aH18z)U4+vmuVLp)Qfd)fkG=8XY*BDMsu5J1|C%a4X)2b$FXo?
zQ@ZlObx+oiZM?j1os!tmX+v5f{lAN`qmZ&Ht?BG}lXV8r1=N?(`O@v1V5&ad+#X(H
zC$==W+_|1mvO3lJs5pxJUWmjqB)4t6D+mb=MNJ6EFOHi`NBqc^T%0C8tztPgQX?{I
z#SSIj;ykR<zp}^q-lQO*!7#f<dDNjslv7)9!ASvg(8P>fRvL9%wb|bH(>m1*<0M%|
z{i45bB?*K*V1i;jF$w5Nhx|-bfqASi@bR-vNrh-GsbQZ<V}U-zM;JC;?o?9d@(Q@)
z6>xr1T28A>uejrq`FM(imKf20cEhAVH2ff5&B&E&CDXpM7il2Us!gaGEh<&3JP>Xn
z(y0X`RKJW-TdK>#jEU^)0gtV@w6_%nA~7+39qhHkh%9|u#tKD}QNmrIV^xu2D$c6<
z9OY(m4rXCkS@2i92zeXGIh{aHguiL=J`|)=!etd<+Fz@f=VzS%3J2xbS65BmxDz75
z6NxsN5m3Pfy;9neK{X<2?~ZA51Z&#>+CDyXj)af~Y^7<B!9g8{^M>;*o-W8Zmge(A
zFypVe%IlAQ7jlAwV-20&^nc1?)_1_6`CeZPy`9Ldj4G6QJlho%qvAB8Pr`3eX%6>l
zut+MoD*X6=DTZ&p=!pjp*XedYa5$>e1v=DXy~3i>=YrhfVV6}LdNfO@x#ms)LZ=3g
zek-hL{5i6^bUC8DxT@l|G>GT^i^Y*!fsMsMIGd^IJLS+J8Nav;6{+ki{(!!?JQ+Sv
z=Xs3eLdA&x?+4=j5jv$m^1v2p-y(5e^rY#%s&|w;s0TWv;-jM;FV-^nk$!=5DhyWC
zsYn+4h{!b4kXFO6R)jp02CAuW8U?<5;!l>U-`A!^drBl!$VpJ-$R|If-SAc55apOO
zXrD$M0^3R2CFubB$ibr?x3j-W<(Lo!8_I$3@tb3Kem+iRf=k7lt}1^@nM)6WIR`59
zKm|9z1t;kqpeUqqUHNsLr1B7eQv(bxyZ)1(?SjX$J=-6$kJnu@>CY4f)$e*OFW|7$
z6aL=>N}OP@R2I<M+S{19T(tyM{s-1MyD+F5^i!T97u9Qgn|F%^0@TVAl9-g`PfGd2
zR0AQsr|lDM40mVuW>7Z&s=UYhJML<@3LRELryeSYGHzy=XfZ$rpN5#a3WI@GF2AW6
z&C{sg5k-6TP;>N~lvBi)M!U^}R4lpdugtD?;OW9vE1Tgpz%;-#9$LW{8^cR18*s1S
z4!AnDcq~v0aLG}9<boXhf~4%1=H_1Iy*j@bZ&fW~QV?KhHG&C|mce)eTJIlSmK>}P
z?0fEa;9sk-%?214e27f<>gM?Mv~`f_z_Vm%gU&Qnmzc^~1ljl!<pM4fYtQjVa<m86
z{^O0;i8@^$8D8Q6lpZ4<8Qw?kPL40_fREZsNhi0(UgA`Z+_96zgQLq~(^>s%!hx;#
zfvr>D^ln$wCd$dJw}+$1`P?zfrvyOnGNtdQ*ON2gtLi0lm@vxsstMq`DhS)-;rl?_
z{(dW43-A@~SpGol)jEFxt8X6hc6R9f+2Z`EIWm&FYJ3^%V>=_!Gwn5uA~6v1!DxC%
z`4rIS6LRKZ!`BvN(lESlY$V1+M@?rLL9W-5e;DRYzdV3n_YNvIs!wNU2+-ckhQQ=D
zO40`iC~a$ku+n+bHi)aoKpq-?{Q#a^T*1JfAbbEPUddp{Bm$e9_<D@Q(K%>6z-J~@
zL^5?sB`~P~-td`eZbEuJP{!ZUz6>MerUTuY<sfWvi&Z#qp8K_W=r4GC-0xQ3igHKP
zZ&P|!Z<9uLd~?I>cmlA#fWTgV6rJy|jr#zoM;>SbL{jr7dKh5Gu+r7PLn*e=L-?Ft
z#WG_AE6$l3k5cv8L|YO8<Ji$4W5lvE6pTKh1R`l_<hQ<SE2Dsq`_oiK0|9q&&yTxd
zMqeja1p$3M-#u67=Y(;jp_y;_eK|7)<1^4N+D4cl??Gk=0#~cQC)LIcVQGHPI}L#2
zEu0R3$0H%)q}j?~1^^1BneOU6mz7@iJ)4_Y`D?vv9}c!4i?Koozi!hqJhl))69e78
z95e_2^6{|~DVR(c>igCus*_{olEuuY_Oy$bcrkGI+a?N)KlKsHXv~X>dz=B%YHOQK
zjwjGPV4U>~8TmLntF(4<Y%hS)bp%v@l(&YN%3Vc`Grjr>a}qL6GcW*1di8a4^gd+#
zE?=VsmOm~}W<FSN;te{!9y2{X4*$A;UaN(5-{rb*0t`07m~^bPG?&i$d@7>^71KT!
zdwX7-3W@v}3EboB#tWMrm)%2*ib#CH7f03}&O^0|!=KRc8Q&S_tcUiZ@&F%HyT&)!
zaymwk<Zqm5>z6s%de>G<_Di*!XR>DR20f#_vv@12k;AO+bPl%vFGN12^uG~#((}3h
zg~+dp1AHdkr(yd}hZ@KJe-L@=4dwqq<dbV>08+4(J4qbXi05vPyw!RqV8kOC`Fv@$
zZJ*dew21uj^Eh?y$HnZM9S4g}X3r!~u8XS|*edBUg)S|}mHN+APL!WlB@gcBs<jEs
zXiij&B&Sxh73DGwp85|ZL)AFrRM|r>OG5fWNLNDN0ATtl^xm|LMchvqGRHpbB0HcW
z08MNRoDhx|MuvhZN@KLd{t=AQ(&C9233<<#CrLzi&575-?a)WZKQyq)*=yGBh@H5A
zX;NY+4vYQ2oVqMNe%&6Y_x`YzxCITa7<j7KU7q2GzZCKZ^FOR!uFG<)N~Zouq{vKD
z^mY?Z={B3EroKR499|B0=zbLN-#}rz-9Mo4x!A+q24OW9ck=t7#(NJYW8Lb=(9^pt
zG*hbSO5IIa|2_ESu+z``r|zF}Z?;5j>v*sQ$jkR&52zFPa{R_LHj(|e=1xnHa3}*Y
zPI-knWY2S5d7lpOxQj9peQ*+aj^Q#+ph=Xz+I`4H4iKxQzkpwjSyn`;F$pd|Hj6@J
zhNHy0^D?=H!Hd^T%K17524pp+gn5ub!dyhgTtt{jBN5n&T$RYE68s{9?{OZ19*JPR
z{aHfcaDrJv=}cp&gd?&fX<8S6it)x|xqCwWWXUDxm|>A)QtQytS-FU4QgW?S=yli1
zemBH8ti)(G)dODQUg3&soBI3KqzqQooT2~1(u?6DL8~Xc*-bc5G5@aNKSN`w+Vy_)
z5ialSk3ML8UKmtkd!NFdIhM@aKh+7^m<wQb9TrKxZX|Hkr8y*<4#Dew@2e`Y>lkt_
z+x=rSo7uGg^I;Bc<s^ZPT)^p?vI4wfuoHNkb|A$>r~;7qGiAsh$>*~DA;5Q-h3P;U
zu>#Q)4$Qwm^iTeAamUBCx|G&|QJTLK7Sr}6G-_UYZS2zib|FO~;}LfdXu$$Hy{4L`
z8oV;%-_qOC50R_<r`fl_2Z0SSW(`ig92qqtYZ?cnL?7N@#dy#3VU@Wo6#OX@bSx#G
zsd5;#R7(H_tu^dGTKT;&Z5?<=uv{JZ!Mq@Bx}0KJk45YCd7UuHK=#wI-zX;1^p1ZS
z$Z(xa1*)D3TZH)*)D)suIQX<!vl#vqqLK?0PX^)adtM?ml-cb7NZu20H{y#^&i9_n
z==}*4{w2J)lvnToY=zjotZ;$JODG2$jcDQ1X9S?cAp~2r@Dch8YST{w3WJzt2s2(P
zSbjxp7P9r$JfZA;;r7Ir$vi<NNz6EmOo*a7cA^O1KAuAhFAMa&lpKbJ%?%6j;7v7f
z-KgGlQDgN-6dkkgcwm6;@s0eHi~~AHPFR}^^j<;5YirkN^Dc}UM~+ZN5YCsQF#BhZ
z0By>XUXjz#gl-qLg4fI7V4kYMv{%hmQOTDk#Hi9(xHazb)x@sH*NKjj-U#eH;fyrh
z_NBpw!t6)6)vU2-M-%6yQ#{O=do6|gOZ|x~HF7Coh%J3TK)IWhV4O8Qk(D)q6vii+
zU>l)|)g}$$=c)p_{E4;#a)cCF7Ey<?0gJPF%`w`=s$^g}P<(f#<}W|<*k*DJr)9vX
z&N^I4;!$dw^tbds0@6;>%Iib34A_z}n^S2Ucnp8cvF`*r!JItyV`_&_1o=OJl8E1z
z-s?2A6G+D4TxmZ4PhE~#R~TMy7=H|!oi5em3Bq~vTzcio<0+}l*m?95uwh6L!yLAO
zTg=vG0`@{*oPxnK{#5UMNVZY7<>5(uK78Yz|8_woYKbgQWYDzi@W1T*&E`LLp8LP;
zyjmn@Hcjt#SQhA60_cUj!-c1~vet4*`~NWZmQi&C&AKq|?(Po3-GaMYaCZU(mjniP
zcXzkoPJ+9;2Y1&HB;O|QIrpA>*Z1SDVfE|<o0*#K>Z*R8s_6=%40@mD7z7ORd7{Ul
ziC*<^6CL|42k2@&GtOjMDF4)|dt6~Vd@2TpdCQMY!*`syL~m(6<4>cbU8UfQEEzzN
zTrN}Q(oND$Vs3(};+P;BsN%@r4>O;+mv6ri1R}7Gs2y&jO5wP;M(k{Rx>I<P=Glp4
zXD=H2hw3m5_si9QF5PMK_;RALMpb-1{$C$2Nrx{`0E-`YBI#j5rO*6aBHmVl=?=?i
za-Nx0oq0mXkju+iX=fwn@mDv3@pg$l%Q+~4Br1N|=U@|DmrO3^C`crqZf&PSW>)@1
z5*%L7a~eT`ge2PNS^;Hcf-7kh2JXn5-oU^~o4+#rqI66}2){wlogqZy8tR&tabv3z
zd>vq!?`R1%Ke?fkH$3d5pwfb56y{c~N;U(y22~pu*@pitZ{~VCHRFN;3=uo=jvnn@
zuh|<etjzSk;omQ|zUqJ(igI@8>Tc$?hleQ#(|nGkICfQ5S*01n%B7bdx9iVW^C-MI
z)Zb>oEAiFQ8MdQJ9l@aI;6AnltW!G=h&;RU75Ig5tItQm%eLwZ+My@Lwq&ibWY_6j
z*0s1XbJzQIYK(>S;~NOqy}CS{7$I_KkyF}^43FIM@1UXkOCNWb<zz-ejoh9I3-%2p
z)7}KmVo5X2LL`3a1BcOYgZvZi0w5E%54bI&cC6|FTx3|^aU<fee*~YBHcxp2>|9yx
zB+B16YQ8!k39dP%&%cx6EE*VbpRqGP4|D{3_b7pvQ4WMWPW|yW$}|<Iqi+s`Q@w^`
zBOi!~5SM@=i8oFCeI`>^ELUeDqREptLhn8wGFiMVAXCSK0lJS&UHs>ft8UQAlyX38
z<vG<Yf_R-|W8;_vp_82V4R{B&&Ne&v-V!3_YH^7?B)`UBwntn<imX1$<jmAZ>3{gV
zvw#TSzRK+V%JHS9o)<#eH|FJ=0SU!N*c0}Gm=~fauNvvC;i+9{+LVt8@`=t*Pj-`U
z@wuteaxMehR(pqEW-!98S7DDDUWg74bv%vs4o9tQXLUmN73^&UfELahg3C!29c@Bi
z`Ys*nC@dO{B%30z@1|G2Ltq#_-BLE%v7Ii}vtP775Dd0Bux-6isJVo(2giWe$*&BC
z`j+3tq~=rN)6}#&B|gKC`5R)+*Q<=_riLDIfxpWa1OY$bEVxxIM<)ms$sZvvCWelk
zx>6a3-ef^VogW6s_3MlI_gzl}QJ6zxAOxkb3$kJ$P+`f?-9@4qxruG3c|1Du7l=@P
z?L->@_0))3NxaVh+c-rugK7A<iv6$JH`T;Th%bvpcDB6S`j;8H;wRJFw7W46M7G~B
zjU;A&dtX#cS7}!vgZ+>}MTY=Ik5;mHCQ3sIiD}4$qHy^NwBKZybw-Or;X`dBIV-}b
zXT_U*`u@O%x*U@$p){~Q@bYO0%f?L#Z|QzW&~1CguUTL~^~PDUO?6^|W<U_wFUbVi
zvfqzosl`mtx`0OtpRF_g%h2-kHSzX7X&9MAEiZcGXGBvtHYPo!NO-aek?%IhpY7uH
zCs;$v{mbb`=Fhs<M4`rTM4=4O`VhjlL8X?j7nb6B<c9Nhc1pgEcgXx~FQORm`WjI<
z<<trx<a1D_MfowVHlM}dZmQ$l@UanYW-6?`w4_g1EEBls(>^E`yyt%kW%K$SKQ4Y>
zg{M^^q8aPZqB~@234j?GGwSSFH^`N%bb~gA_f-(y!eNWDYbSYzg}>c*@Z5XtJT@Sb
zrRE4o;!e=vUJ%t_D_P9;Fv`?D07@IV%Y?`yr{$n}d%&BQ8XYnz=hz34Z;;5R4d6}$
z)4FbbsetNe$l|#<t;4e1Z~=y!=C~YxHSB&g*n9qr935==?!THZkB(NKa{soT#sAD)
z`~`1{X>S2p2OmVn%xx3*Yi<RjpX8~wxLy*vUU3|q&6CSNO{=lD`|V9)IZ-&#gQA8o
z3VDf3nuZvoFWR^9U^RcHI<Z_nG}CZU45}6VM>v3<ccS*8#z*1vgK(Yn&gzh9d{iVG
zuwnVv$r7Ylf{8w73u+AeV377u^`IOH)?;jEc4M1x-zEfwFnFGTJE#IM)oCe+o!znT
zd!g7ZN!w_~(8kaj<ig}*Sek9Fbt2US#dOe${svNG>az21)lh0Y-go=)h;PUK!YY!H
z`M{;>({NNrRw}saNyJnHj^@3%VD>t%*q`5Jpm1VgtZO$~HyT;Xk_3b&8!7^7aA@x-
zIXK+`bp&*T*B_G@Q^U=og0xUm!{XL4FMKrYdX&<|(HUE1A&K@Qvj8TtQ!fchOnr!P
zuyL>m(h*btolxJj!r$>c<#fJzzg6ikoB=}^iE>6^-l_RPHL?<AB)geWD9|b;k8tbA
zB4AK_H~+fqus8zx#H;L(-Wow=YAjV_{#IqG63c^+1+qYFaeGgfPd5<S_V@yMLLm;?
zT!<hHEF*mj!x#n%*BD0SMdTWC02S*a0vY!EiDAHCa?Fp;Z2%TFV^tfW|1KQK<U>Zf
zF}yLnD<siG>US1se85ojzBB9=#i*a4|6Y3%p9l?m45WEVQLxR344xV(;7P}#?RFXl
ztV%^IF)Q!h(g`m=?D83L2l9GF|J%4~^oxi>i;d#YF2U4^`^o)Bo~Y6QBsxS<@u5}}
zeCos>Y)u!P&$_jX-&E%8VksjK3VxKK?g^~+34M}rh_kW=1R4ztob%6JRt>)|jeNZ5
zB$c%b1iMwN+uOeQu|;Ak8*G7`BFQ)w&??F!jp2$bHsuEK#<;ZzfARB4ddW8u*;Xt=
z4!mAsJ}B>e+TFS%;fu7b7_wUMMKuJY1o-1jmuBmpJ*Lg7t4jU29$BR^!u_b7z5>i5
zoa2mKk-&NefI1jBDSp_r=w+Wb1SS#W$@=(y5M+M+JMVy8F5(d(Cq+t&*NT_RQ%)$v
z&pe1Lc%(K}(A2&YX!41m&1uhJc%v~uz@~~GCZiXkdk4}{``i82;Fhj6Ka;>!h()Ny
z5FgT{i3a;0-q0RCh0UTJBA7$g<Q?Wi3lWWBepeucUb<4~K5mw*cm~N~!Vy*EIRCE9
z%vyh_`yL$1tM>5(lzN2vHMX7WWU)-I&>@?n{QYDAQl|gt(h=V*NSD0kkiJkkwu*oy
zqW(K#y|YiIcVWO41i(I}YNWXn4Q$cDNVD-q^r$LK7VW%${*Mg6yAY}L(x)K1wi3&@
zPN+su#5r2RNnl9sH!FUuIyBuH<Ie~McVm3HK#l!C@<8HdH(#yVBDKM9rS^VmS6koj
zL<&>;^ljR;D7skWrUZXWX7fuh7!B!vj(h%v{2^?JG<UEX=5p5acM*_cOut|S7N%v%
z4gjhOqS9>WEyO?Q#-0?qKQ}O?xWK?l4BUVUi(u~hp9jdLC0l@5c*k83XyQg;hAMr2
zV6Kx6dR3Q@Km>u`Gg69iyL#OIQ|{+jaTv2cw+T=J$3HVNc&g}bd$?6gCxGo=sxMGX
z<FD`V#-Q<SH9O^Z>*CTO7V2HP+f5CSu2}j%voI8kKJ>D$NYvKHYt>L_Z_7l65=2fe
zP@@np|0=&FiWVcwj#JtLrzQzF0#(mSE9#rP5dEvcpNQ21Lk=i49sTd=kii>Pw)B^Y
z6O~RdI!TJVzE2x$?OW@<g@^Muh30=@2t7iP^EDu8V^XxED-rokvZYX3qk7?_X;RM%
zFIqcfUq-8ndL&JdnNfMd_{L-|wIWS;yxii=L(VQxch!n;uN?pPTbjqI8!~kR99&bJ
z%A6eTbdvHFPE{i*8tsEk0-=zouBnUD{i#0XoFoNxsTJS08akI#qX^eW<8Uj*45`F;
zGi3k^Gmew8E&nk24o<Mw)F!sEp@c$JHOC13t=OkI|9L$95Gzt0d(m8iA>b)a=n48Z
zL7jq=!r&}fRt1BZvqnxArwy_{SyIq}if~KoQ;(!7xZjvO-HB*09LtGlx>~vlDuZ7|
zy>m@Po^J8%QYE9&tc;J_)#T`e`KN|Hs5)S@ba`N9VE<`IRdVHD+9Mym95x=SBLpbh
zOEb@VpxA6*bx8BB4Q`W#-N5(h_w~XLs#{@k$w^>-A<K#vi5KZrCA4pN)+Eowc*aEH
z$z>RB^OCMg$(KNYqnIEXYQ(U3I>2ZKcAxisiAwDh{Ts)HE|X_h-h#b2vYG>6tm)S{
zpM@-cRVwvpH456FVqj%pq+sYKn%b}gi&KMWTes5&6Q$dy;E=Ygk+;l2$3IF!UJ0~w
zbZmJ+K*QE2)3(@1I;NCwtA}&NmRS1ffMBhup|6iEWmi_Eu~YGd!j7|(fvFnd`(eo5
zDda*ExW*DkkX2Gn<=e|?HaQp|BkMnZj0C>n!z(_G=9Wdig_O@9Z?=ie=yPG2YTu)3
z;6vFaOSr{G550<DpRA?FwhvQ*pqboPXDSwdHX31eL=n`&>dl%@XH+rD(4_^vz>wW)
zCKk_mhlPcCaD8Nrtx}<x2L(9@@>l=nz?UAl&_IHMm>E*+G<7f|pddvpe+KG@g!ufn
zJgXE#giJFE<tS-Xe1KeawHh9!=6lc4#Kt2889m{vtwQbLgI=ge{*O`j9&Va*Aa>J3
zcn)hLS31s8+0=t65p(ep0Y#+N;};4mPf%DPSRqd05h+=%pl}>Uupqxl1<n7*%2#e#
z10LV00S8!Y^z+n>$liB{x`44@ApT9(q`W`j%N)fh<NgPPmrx~Z0D7Ij=e4DwathFY
zcv<wAM~&l`Fq2(|m`62+dOsRDA@02cH4b((lGP;tEluE*F%;zsD+TOUBor+I;CnGm
zVs8~T!)F(aKeyKKu^-3bF`GlN_nHGm2dgF-nzZVvD2%K_Wy1+pP`PmI%fVN=LVYrB
zPa^}&P!rRj1_@hdPww$$>!|FvoE(Z+t!og@e5#wOrviR_90^kgo&P+8g@v{|c`Sli
z05k%m;|}+s`PE%rRAN>q7A0;&k_q{)!b?zY6V+d^0DVxX-?>r&HDDb>29S)?GnGl+
z`HrT$Ppm7e1dodv)tZt<qfuBV0P#NJx_Y1w{~7TDh!6xk24l?<*fFY`;)*2GrR6VO
zcDBgn;-R@w0;?`%*NWfJ?PzK(#cW>(Gr~u~6LhGb{8TrwU8q!LV|M>Zl%QUq-kZ7Q
zq;*e>Wp?1_sR@lHEh*0eEv*TqiLN18p@3RJ@ua4ybP>A>YfI`}y1H`jAj|b^bg|yX
z`~2{!&eX4%^hv19FOlVS+Pr#53RG6@Q%;1{RX5S6SS6CPgKoH?6JfkmYhGv~z=y2#
zlhv#1uV#>zoVmK(*z>VckbiG&73^y7Y-@d5|0(#Y%;qoS)PL?^&wqH3H@D(=DYk1d
z@43`KYMY<&*Yed8#01Cw9BlNp#7`S6{WebI#G=ZvQNl$WYZn`0wy?spEU^8vCrak=
z>}=<DU}66sFFY1GmTL$ihJklxc<festYszDvH3U4Rqhko3*(N}(^iK+8wt+{8{d1F
zi3NE!R?g-<VY|}dTUSkn!>y2Bi6qn)x1E)}6pUSWg(uSJ?YP!E0XOYt&itDHK@Wox
zWEjZSru9kP-BJIPrW>!nua5kk23S!n^F~8Jo3UDy^R0xv=K<65XYThl)FbicE9{Ns
z(irHdZkf3n>`ePl6ht1D=~p?XJm@}W8*-<M*;vvmb6~zx1FcY>-xCtUdi?4>kNrg?
zZ6!h&4RLm!w`<|hEv^pB+I!EuoRpy&pNSy;eklE1`t>UxycY?UkPr`S%z68SqXJMH
zaUI4N7jL}uw0Q8hrS-cPe1i!2EqQb6pr5_7GI0w1FZG+PWhR>(w7LE8+qn9ur<uEp
zV0aGcgYH_vUx&!Ykdeq~$7I~3B$J98E{WfM<hJhxSZs&vJ{jFzcsWtAetY{gV6yz%
z%aWk)m&NhdX&*-43zL&;>5QwTR}L;f?Q5qM{`1GEk-pgbkl!j_1eqsA0#TjV`+wfv
zeBFQE@11+YP;MK&-nCFC#(Of(eOwo-5P=!+|MJyFUGb*l{tqFJ>>s4d^Cpjn2K;AL
zb9jEfbr#zK-R{-`YOc^l?w?}BtD1jb7%;aL9;2QQJ3>RdC<M*uqYK9^*lC>q0!_78
zqAZdh6cU<L=gC5JX6M>U3M{x3zqyFYlgGvlwG8jR$iE0n>`@k&2=CYz7%ZH^k?wsq
z;2T@`i9>I7{kFhk;p;s7L6^*L$D;1Ag;?ir21z6x!FV#BAGX}x4#m&=F758XD@ko!
z?Nq)~(-j;hh0b7|A}#RV7dmKe7a(~;V#0;8gf>HM-!5myFbg9}(tiStu<6gUbM4wk
z<L{P*pLCN=AqTNPX!32;D<KQ8mpxXkduKBWFsi0;N%z-V(T2ZJ)rvA((#8#XGAY3Y
zRs{3X?ZJk>WZeMzrruYXSD6FP6K+t76)Ag29h?W|5;3qiS2xD~d8g9tH2~?1LH+WR
zx>D9B&_>Y@>sOF%k5R32vL~vYMgej{F|n6e^itGgChHmE4yH}&io`oXj*%iMEtlyj
z(!{9|@*{++>{$?}z6eC$&7FiKE!MJ$Ztls5m0RdelHTS7r<YpVN{cgsDl$}!wgD{1
zNXEaG`pmS_j$evyp2vH4{{jW@Oape@uaSs-P?z`-v9`iAsgq`!&@>KvtWb~vP5~u6
z!DNPb!ktPg{%0@A<iV>Is(-S<!#3LsLYwUy?MpGDaU>e;zwgv+E<)R{h2u!|Y~k6j
zy%0C%?i;yBy-zF`c=5FdUGL!Lb7X8d`!(w|Ur9+!`?ae~j6V}@02H$Qnq)tYPLEzM
ze`@{NpcGRdf2ss&v>WpRt;2VWQ~kVa$<f}l<43Nhkoo*R<w4?ezWLuYKf)wgyv)(s
z7-~5q4bj@MiQB;qlef8~xo#lgKQOWsoXzw6C{Ep`dYyGRwqb%h&O!08l_!TxyknF7
z8OC0j(?VhTR8F(t55z0y6d47CGle5ljLWuUjYf*?V1tGEH=C+dZ{UG41FprFqsf;>
z-ZDDaMw>bK?m4Si=}x-!bzmWkIo#ln&CMvIQ7-COpXLUIYFZ6o2rJTDplU{+r9Azs
zc&kfs%kD_%7WNgu<8}Mc7=4rL+FX-Gl!400g|6>lDxyLIKr4j7T?@u80$N8mjn+r^
z@Gb=S$I#!<U|Ncw*pMWty#`sa-cZ_+m7l&IagQs_vm#l1=o<Zk7jg1IT@q1+Zzhpl
z+%WFg8uDItypu$Pfv3*ebQ<iF!ibuw91aG9!bk)<r_VuoQaXn@G)oHMr!*u!P-{xS
z@FLPP7A>y=R3yj@WlseE_Hk0AlMvNY?QVCjV#HNbU2xlPt;$ouCp{g-4)S&|XbaFj
z`KoE}Fems$6K<(O@-G;V_?Ofei#=La;iR(faLd@F>~#hz>l#$Ww2q^Kd&=wxEdDr(
z#`HE)bEKtw3Nk!m*QyhY#@;EbF(h*hjp;emAA_$gfeh3tKfaK~ULr{Cv|R{#1zzwz
zYv^E3_c_9^yzO(ZM<{bNO;Lt^j3tV_Ta1P*17s*jFY3saNaLz%%4}E4xUU1<KRa6^
zmnG|#q0Lya8i6n)w4cUXMosX2-5=jj4%D-59>1{g38?mU|7^NLfczVG+z%i0(oZqh
z)BRB-0&p?k(lyX$yFvy1MMX3BWIxE3SQPYKicWQ{D~A+GIU62s(-mc*5!M)C)}e*(
z1t?Lh=U%M(vCX!6{}qIlZ6+}1Z0Z+U`R|DTKaWb<_0OY5*7~flQuJUO>E{?4Z0fIn
zyD-XmD-Gi1__XuT!o2~Y`Ai?5wNS@mxi0*_y3{&2-^eb^H^T8S*QrTnVIzHmH_O-6
zt{<JPFUR|lFNhQVLQR5zpnj&$lV#*zh`;HmJwGXYL7`SioEPr@_NW`F;qM+5cz*}V
zJdnamtOm4eLTGCS4tn=OxK*N_T+|xdKEN3wzN40c{qHrX|LIjl)cHjiu{PaHsfAHI
z84<rbKB9Q&Nk6X;ZDEB!t$nb?zPYI?m$^eMj@lndBzsTr&mV(XhySQa(aCAGG7jAS
zyn-yri+R)ss_q>wFM7Oj4TQE&F>Lo>F+zrpZ+tfGAKOd$Mm$J3g}I0y&{tha2N5-`
z%JV9g!k6P&1L-R=Zgz_h6f&C|&l>flH&WKoRg;ytA$%)j-Q@Z$bRjd*cR;Z7ZfZl!
zh$LD+DRmr9ZB@o{MdP-`;4%Yx^nV)?faV9?Vog-J+u_DiP9`<pQ^F)!upGD7PYHop
z>9~s?Y+TF(#@$r=B-1pxO2l4P?9R)Q>dft#_0jSBF@)bgBmPR&Q1;;3Tt;%Y?F5Sa
zUU-u-0H58<LflDQ6CQ2DsIsoEk)VfnszcdS-fh%>K1xd`@FDvpI<7bK2a(y#I`vKz
znwgnt_VTjD+d5hz+k``uG3BjZ6;N<M-f#IdO;XY1uBFW%8P34i?U(p3BW5gBMGTsf
zgw#P&^Uv?J&rfseE!@Iq<I^R>T*@U1gz@o$olJ$P@nX<Ka!mTr3H1~!bUylSPHclW
z<uV!+9s<wVDWmE<fk@IVa}X1%%0f)^erHj`be@W=_5MVpNxQt6&h<%{+6sm|Q;0=!
z;CINcf$%S>k7CdYxXBJHJM4#>VvENn_`8YUKK5o<0O5?bsk-9O41kOF7(2T=5^V#h
z*yLOmi96D#tX9Y~yhWAWIx}8OYY3HEGdyq|C-9Y!1Kk6=o^DT}sHyg^8^lknF=;al
z#8xH*k~swGIxttNY@(7pnej*rM;1BWG#i^o_JEG7duAadpUr!s8t7F@bm2pKT5w=5
zFF?Ng#UcEuV{PUhP<PCc%*|g<;`Tsu=Ah;I5%s&}c5!t(g-Jh?XuO+%o?er%6dMi7
zZDV$QK+KamY{K??Kg8zWsm&JF(s*-&w`5(np&+b6)N7dgpe}H@_5RlQPkHGZ_HMkR
zWmI)y!7!ut<&;1E{z$%$13_nS9ZfF|<*7dIgmtgmb*qCefYu{iq3%y`H`IWbaR9|<
zC>mcDQHnR2tY!(eeE77MVNrJ(eMf!b@Nu0ArAal#vXa12YKd70N2&lV__#(9bO}Ux
zbWWi#IhD-cWmtNO($Xc`ub?ZLhGV9wD5`4(xmSPN{Ibd-Ll3cvZHgtD$DPXrZE7!&
zNihR+L3zjwBnC3Isa48v&Dt<6UvCQgrvgM7RcnDt#cRjt5lSq@#jhWvZ0-efb)H%a
znEGoG1zwMzNLT;#xJ|Pki&Mj^JsVMPrT^@57vc_7YpNqaRgqe0Vb9QMobQ<%g%I=-
zW&cSvWcC!UGw;88U5maU6=B!ivxG{t8&#mZ##rJ6)W9<wc+Nkt&*(TtFsqtV3`^{}
z!VODsw@^D4@OGLor$g+^v%=oTU%mtyMe<b?h+}_n{#d*yob`h~q|KJ^`^lF;Xg!^?
zV17%@+b*sFPL%HsWXU27<n|2cy!@frl@Nj?>b+g~4#Q|W`ymA6B&E;{X1ULqI<|&v
z!<8b>0E(w+*0xGEs9LKg)(4}fNGTz1(Z8~0n5B5fDy#gFOOtIs>T5nLa;6NP<GyYl
zF(bduJc)K*l6CiW_O`PKPe$A{h{-5()LcF_>TdI~i{hS`oF%9V0WJ=V!r!K2^o1)6
zyI6Yo{$R<LO<FUh`M+Yw=A;t=gw<Lt8^d&OfF-&4-7z+(i1Z%hh2*Y>%R}zv;PoAz
zjGE+)M)7HZTD{bwZnPF3<0vgUcX#C%jSguQqWgZkB%=NUs=<+q*w{h>G}nBjA)d*4
z6-DFKKka~lNClJ%&wZ>ht<uBXFz&$k3pf=9K=o$o2!;V*t4i<0_UCHsT(;34H~3lr
z%$QFGMCCm2KdG&3)(|$VWr?~X9EZ?MaHnq-IjmNI*5^1Gt;*C&1Sxjlbp9%8=Dssi
zp$psFu`e&;Sy`i#AwOD2J9%r6=`4de&rqbx&s)^y%q16nz`iXlB97(!1Ijpuei<aK
z0PCkcnQz}eJkNia?B`=j+qT75o{F3WG#<wutb8o+_je2578drCxruC~OIJ5MTx=(B
zEJx;ZqtwrjaH~DMW_}UeWh^L>h{jGy^}A*xNq3FSc=G1lNq7BK7x7Lk|59DJ;mB*?
znpv$?s1`V&-uOGY$y4}Y;$^Ut?c?W{KH-Myw+W7AoZKt>1_2!EO%lJ?e3l7J;F<R9
z`|rOfYvHcU%pr-}`$vN>9*4G^L_XesMIRuGk9Rt3+N77TpF@YF&m4_`36{;D15b9!
zXUZjiz8TI6TL(@aI(;nrK~6e<sj+(RD<wP(x$jq+G?#I#d=}$vX^ZlOHlu0!2Y$wz
zy+iG<la`j%(cg30CobnB+SPYo%>WBv<Bthm$*YUZaFN{bgX>ZfY@WXpe-sP<y;3sC
zAUF@htbI<!H?}9Lwq6IY!m(>5SH5`V`O!mUX_9vL^y2J=v1l!xN(b#tHCE)`vSu7&
z84i$U%nFaFXRB+Ul6+o0+I)R|MU|8xp0kQ`JfFqI5lX->Cjxuu@3-W?!U8`ByHy<#
zb`66ztB#{9DKCgjBRpgilk9RM@jWT&n`|%ZAtJ~ewRH*eB{Tw7KgY_s_jbseM=Y>F
z(a_;QU}Bk2vR;3^!?~8;X{3(0RqXDRMA4Pr@!ma=F`ra(b55}p<095zCQrceq(pC&
z^DBamAk)>3AfplkU9Dfb7+ALD=Vu={!3AAPKC3&(mH>3O4zZGmIK>VKI~(*L7WrzG
zeKy@Tdlhh@nR;iuA~6x&dJ!I_+ikFfFcaRaM6qlJOernylQ#H_M+mM13)m%;9s-qt
z{M%vy`S~jwbq+uHTiBDqX31aeA4cyZOKL?~m@+;EILb-NfUU6surUZl65-6kXKB*x
zoK)=Tokq`vZh{5FLC;&>^Jcf@#j1xb!Zhp0I<8S)?9pk~O|HG9X%$)#b(z$1zesPS
zH_#uB(LsB<gnT+MgRXDGLeR=b)C!;S{8Hsd5s0wJtM!FPY``qog@}>P`kGg33j#R?
zySF5&ox@0dur6R8xGK;CPi8^^9vX>V+P`dGi;+i);UHZ(<JKp$GwAFq_&N~&PBz&j
zkl%fXG<a}L4S73yA+yz=Zw2M-??M<}Zx1Z+1)spDHtitz5qRWQQyw#^?zR)=n@8)3
zMpx#I6ejFLATzAg(ihH;EG4Oq$RV>Z=%A;PP?h#YS9d=EfrNS*4P}<t<(8G8meWh4
zK|fEFi;&wQN#H4iy+WsaTaqjCtBzm)<uLd{)gB7wLRAO4oK^#uMgu~IV1*9zbY2E>
zCi`giH%0t~s`)Et?<DM*bew}XcquNv_Iw*Y5AcE*)*^9G7w1i4u9)AT7&LB5^y{cq
z&FnDAegn1{+|VZzjBmU`X|t%enc#zIClkO}Cb1(hswF_I$pC}v0K+vo>Lj*`xk?Yg
zA;N_eomb6tcUwRJO{@zv^YoTJx%nuapbNJ}7NZO@%O7l&(VaW1pTz(V>BK-R#R_EH
zazV-mbIPPDi_9_$&;!obgC6h$A-e^ZK$!(N!1B@4Nm{8;uD7eI6(TQX!|GI~U4FhE
zbQ_(qW|grDtDBiok)rNm+W=XX!F+4!8I;`M0s|9ICve3Af_~`6ux52uFQFTcq_g19
zK?&4f38_0klAT(C`XvE5on4Mtquwu-w6oQ>M1^wuMn@l9W#pp4jVha(LB|6m6ktB3
zI>KOmQ)b>K5-7(FdK5R65eP%ytAYe<r(-=5Y*@8kCgQ-J(Kdzoh6|}O?J2FXGOPE;
zcpoRB_s0+{vjPV_h3eF7S%S*A8lSs}5M$ezgwUx&P{RFZ==&;=Jl-jYoMl$xB~bye
z_e@*+uq-D!>r)WrkY@m_3uyFJCm7Twv_LIL5KE9Cw(o*?s7(A5B%@e`GCyu$gXvj5
zZnYo-`@JtM?f$2-u!EGP%<`WQ6A`VhcLZ~+B>Lk8Ns{!HGW2@V^zUkz1X4q}l{oss
zMNxW0&!?V9H5}XquyA0`3sO#}yvt&iF_zBRpi-rN*gkJqlhsWuehiubtBqV;3Y+R%
z{3*WFPdCHk>z+=vN=O%oG#>A2e?)Bhaj%XQu5rr8&=9P;o}JRH&4MXKvJMCMa0$ZJ
zSohjZPXzid>`$=wqRHxwDKilQ8&XZ&e%QFB8&kKzO{LC8^OZpIWLv<fkC0OF`sUko
zT!2lt^6G3FbK1{_pEKUIl_|I#{3eGz8g~Lz>I%^hyP_@H*d=WDOE~fq>wO6GBYkGr
zIfDWCH4Pmt9H@r@l$YYJx&zHnejLA=j+?)c8T{3^S@`<uEcM0*8E^V6{VV2|$Bh?n
z;IA9k!iO<!ettt>M%emxWz_la7w3wu-=v)yH%>=7Tef4@Cu7$H34z?b3cQ3OyPLRQ
zTwk9!Ov<RlhhVJOS`wjAiKg~)?%Pb()}8M69#2TQ{T>H*M*gNyK7@XsicGhuyLIw;
zd&zq?VoW6%r#bG`OF-IuirRiq(_Q^#ouqAP-9gYz>3Zh^uzf`~ZZ^l&8JFAmm35$z
zeMr-w-r)Jk)?_=3aC^4)2aBR!=g5NF@7}l=qK~0BQyfW2bpnGAA^O4Bzs|&nG>hi(
z3L}uT;!7RATGysE{_PglVH`XASgS17G+rJOA?0Zxqk6z=5JH%%l@O7uZZA*oTG4i|
zas%_J@?KPY9T?oL?d?oDL(+tYt=Hj`h1shV6Di0_fHqB1rz<TJ>2xe5T1De#i(Mxr
zQ0w=$;TPftg-bT+rE37F4azV@y_RJ2&lwxU9X182GUvW0hZ%QyOJ%6M0~JKE5H69L
z|2?Ra60sNAt{Wc&b6I&V?En5KO-MT3*9y!563mb2DBcxF0djq&b4|dCVoAUzn}n1P
zw&Kiu{7LaR+voYUp#|ROZ{f+BBgK!Oy1Iloz~8G^mK6_QECTvsDW$_dw#j8l`M)|(
zjrtwkKe1RRJOflogE0<64Ge_#)yrsUkBfayU{`TEjgyRruQ;00UG#*+Y$9kBZPdW(
ze%QlD&Y!ujo3+MaViDKRRez+}+4pU`xQDUul?t+KL!L?>MD@B>bv;@dP56ZQn#_6v
z>$18Ga#h^ZuhN~L6$BQV5NUYjOo)rubTyoP5KRd{$ErWuxyDy>^25L><a@0P4kTrz
z+h~t-UGV@<-1$4-4(FL_^czveNMK{Kn!nOtxCv)0c?nNUd+1ivl*<087g5)`JJN1G
zdl#e0PNVS!4f=v|YR$3LL9FJhyGN%lOqXP}K(a$=YP_x$dG^QATJM~xfNFyx5q`dz
zw~~)Sf5r7e88~N^h!}P9Ry)e8gv(I*HJKe6YaxcRd2hL`r}-ThGtX$NVS(L|qq^@t
zytAe1eT^A!B&40Ebt{Q2xz2LuPZ!U{Z!ant>BSu@>&Euc%%UxD4BGRes&x!>-<#`X
zX9gNfXhMmPqnVQ^Mi@_S?c-QJ@zt8qGE(M<ieydUp}OsU`%zlTOElGD9_G$|p_!Yw
z>7@wT9?z+I)6w;o6~gye0vw1>n%ilKuN35MnNA>guXsHUYmy~!x$D&(GQuO2@#8-@
zO7?jh6-GrH2JI2DsusJv|1!fp2<~kCY|8g@wn?MJ=Yp?UfNa}kRo|5++X-;SI?EwL
z)l(Rh!~1E(@;&75CqjKU-&0b0e{2)elvwG%Cq{}`5Rt*$v1XCX*6_fYvHb4SZ<G7O
z>%b3Ncdhyrtlr-db{|b&-Om!Wb~P$%aUt-P*AH3*J{wOoT?zxw2Op3=LnRj4C)v1S
z@9DIVn2$O5A*8?X2Pmp&#H$H&V#{Wv#qh;v73*+BZYbE~fAfohtI}G&5rY<jmBESO
zt)DaZ+cPGjMoCE6E4TyL)_N%kwn~!56-}AR&mHQzu5r@01NeFfh&q)QR}9y*R;x(b
z_vw?qS|k-dZ#Oy|CY&CU{+S*3@f5fup<XzP^4HB^{*jsD5#ILD^Q*~NDI4Z93k&b}
z=C3~1VBp9y?dz89Tjc{B^3<b4(nHdvb_~wmuJ!hlJH+~R$R=L^x2Ls(N+Q~y>uGnL
z_bDzu3}I)VBq?x*Kj6y1^df@qAT~NCZaqm>&p8eIqwh)#@e&PNdYmcNwWL;LuRqKe
zg4X1`_djP|`(~oO%t)$%zL7y_9q)l-ZqO32bPsOd+85E%xNt_@*3*N}jm$%>S=Kkp
z?4sv+9CcUy?JW;zFxrOyq1$!#BAK?8D6z)cbNCsO=J!0f-V3~Nzn8+*HS^itmUU6x
zmNso5yl_13_d6(3o#ls)CUPJJgU3mB>E*M(`Ds}+AyAAoTNy-Vf}@krhK6sEead(`
z%IXLzb^I*(lc!x?9B}oW&Kxc|ReaHLRxmfC$i&gAS1<?GV0cM<36Nn|2lmaQ!0j3E
zaM<(g8OK`MgniaMk&`-ejMWGeS4H-XHHHiF2hr6OMj8*!G%_lNth>r6cnj}We{ns2
z;t?Qz;4chMEa<gKaL-6<!v&)nUr+2u;&{vvw6abK+xXDTvq{0B&q}M!=E5@F9+}xT
z)m5p@C4dR|9*_{kRl^t`Y;r+?{zpY&d>t_nkqXj3{p%%Tn3JhFpW5VlA<=|m57OuH
z^)X}?7lgFKhh{hz!?eRe!4{^~5HANSLe{Bjl;yNTsVv-X6t*d!+8Ec<1ue?CIjSE*
zR!WUsKmXba3h8|}%4YiR*V-qYV$I_H8J`8mj+PKGur(4kQgRq~__@?Z^604WjM*X{
z2`UMYmb#R9eO2&X(f7Qg5ZdgJY=3~E+7hUXY2@}k@5co1a_1yM-ppF2Y@MDN-*ACl
zf?}NTJ}r<ZycKD;JOta9TKOfLGm6&c)IVUxkNG*@S%(7Q$|XJ0MhVx_Sug=sJ@l8^
z<*q1@qa+$OimhaVTH5_Io?fUH?sSlqttOlI@mMK-zb>^mHj*RG=CrMr^UA^nEQ+<_
z8y<E(Q{YLQMmFo*q12L-U{BH1uxCYnMxrQo!ZHE=p$U;JS>Y<CUncYExf6VMl9J{(
zoW@5Z|Ar{uQ0)t<AU>J((pH@@>%rslH**$%b-&*unh0%?k+{82_0pM>B!<@KGR6oy
zZpnrekwSU>H&3$|Qv}A@&E>>L4dv@*Ray3vpJ^Za>uN{pg87`9Wot|v4Xk=LKfwEp
zcvq<j%(+4|@p2mIK$4EMmEi0Yus2#D`%t-M+D?DR?84q!;uH14lgyR~mKK@jSf_9X
zR`r?_H&1T0vd1rnAQZFhJi<R79--9ZI16+j`H=M#wImWpcBvTI^c0oW!uZKfaY3;j
z$EzrS^7lY>PmjGI0%TDYyvqVtQVKKOKl${3W74oPTHp&hL#>hadhMaJ@Ry_Ze=|->
z_@RJT+)qc{OZS9CQXwwV$w~D!v&t132qIFd51tycket~?hSJy(-{%k*4Z`NrRx7de
z_d=1x;`FBNBw2AARlSizTRdx?#Vfowie3aPd<<H@BzQ{2!cJ6oKvsKq>APufqJGDq
z{&^VRO?4%hn$~*T;k07hSbd=!5@98XqM0@Pnxt;4{CuUKYW(g%w%eGk2(JMBRDR>#
zHs*dW@H4b2w}DHoTGlBZG%Igk9106X_kn29YT@C1<KRopa-LXq5tV%^X*yE49kmZj
ztI7IDD7+(m5!&t6pJ%ZIsGsPuy=H01T(6yM3b}H5yc6wy7yp^$=^WvD8U1l^jqK<Q
z`k93Rke$+D!api+$Vbqjr~vE{y`El)%giIzGEBfe|H^Lo9gh89Y&+4SH9O?h9hmqe
zvGx@g@=qICQB3|(Y%ey{>Jfuz<dpW&Phof;nqfCbvG4~7W!+E@{*1QQ-$I+&tozPf
zMqkWMlV+1P3fI6}6YjV}$cz|nRwpdj@jITBxU%iAo;%Gp$S_OpkOA;xF~jf;G8htn
z(yHeOF62<?D(-?mE-0QDw}lZJDskhV$;rC&k}ZsQZd?}elPxEXPm0u*w!-)sgH!?S
z45SKZXYlU^13W<JzYIn%+L>GCf@rp~bC5IV5Hv^a%peU{Cwc?xh-wYURv<g+E%suh
zXf=FwaUTLmMkd)Fx0r&^Ze=4-LM1Lrv(&zlh;BU<-5#_BypNNK@W=|xIBhwkJNK<D
zWXEs?-w$*r5+}pf-Wt|ywZ__Aj$DbF<*z>{>B#bK_-U<h*(RwJD(di7!v>(lrRSSQ
zUp3^}0>yA|K45aniR5t7OW^8wq{u>p(XWxJK&Z%><~Lg4Bru2d&?}{7ZP7e(J5)-b
zh~br8U(H$bPOH^=^;RIjGTI#J$@Ni0FRmhWAQ;!}$am#m7CO54{PEP-y5ni<YgV@(
zdfF+o*}YFC*M-IcUQbHY`wXo(otzm);`%h2j%a(+Dh4&0@Y0CcHHjsXvNFd#<_SFZ
z)^Nt*wW4$p8!sTFt8X1>G$$teNt;4oKyNwXHMt{$7cNp^VHPg-?m>)*7|*XhBba94
z>BK}8F49T6=(9o+7A`vXjb%X$dUk#2^+X-?WE*6RY&MAkmR4Y*t}7t92R+dbp=-sX
zu1TsWH@uq`hVMATU0TvT5<dU`+G9q-{}*%oq5XfG<H^Q{W^n^&Mfjt(!f?Wk!;P!G
zZedX%B*yY0N4%9r+h$P@*I+dbc9vTJmyT!jP3cA>WH_3VnPe_~Pm2K_N<lQ79yNm+
z9y}|3F1Q@YuLaQ#s8NY|(~s(d{e3I+!gpu~nIlKt-z4h(Zb*tc;Npf1%@w?wNzUaV
zjkv9T!xfH(s)7Qr_;2;Dgk(!pWRSyLJ^*N2OT3b@qkR#C@lv^E$~0vgNH`_MJg6kk
ze5eEUv@AFp*aLM65#;}75xahnFQ**E`TtGA(B1k-4XfK|E|6G2)3;HvG`64*w+b<I
zU+rAIz75#5(fFM1Jwn7y=?<4LSigxbEd}^`rglNRUae{Q`D&Ml5DSs20<H?9m47Kz
zrf2T=@l4P+Cq)Pq!KkLCKrXp)=G`R|?SaZw({?RLJLre(><dUYEy_K(#N?3Mk+)z9
z%)(+&EefpEEhFA1<4`Rk*n!DZ8Hj_0bHNWlg}e2f!h|=`lF3w+_qVnYchn%=E@~6B
z7~Op>?BKkUK7@rXY>@(jnLY<+%_V~m+vf3qt(+uBTK^(6dWt&ChM>U!e+RpVcV_{?
z2Q$ouCRJ{Ar9zCbQ*82MvnYa@Z0}3Lhi?hYj1eU1Jb4%-6-Pwf+*~*526!5!BU&`s
zRH3q?Se$hqwduuOOAG=piDj(Q?2Fkp-E$e04OI3xr5EcZVQ^&u|0q7*-yK1aOx;ar
zCvZi?ZO4!9!pL4HYT%QcPEA#d=_NdFEU_nFat>oPiM@^x0t!ik;~OkL9C+N$=)bTg
znuQCNkvw1^O1F>=YQnDfk~sLgG{7qtV-&1V2^1%Y0JB_~!)7eOJelTXbBmUkHd|vQ
zD^MmHPplEC+WklXpwo`r*)Jvn18V+Kr)Y2qZJLq?&}^Fa=rPTM_cM};z|&oe!8_J<
z?@6%obq<I@m7UIflurFivW<^}VGU~J-W%t<ZTysM4gGK{ZGwtCP<w@hm}Anz!Q=|r
zlr~jszE#mxuNXf}Wv~*RY8m7~?P(JNCIe~!DoanB!NfzP==iLMl=Gs$jP6N?F(IFs
zpBveKfit6TUW|wgpMQI}^Y331`tOl;e59)a_v{$rg6iB57h}72SME6-TocOi^Xv%K
zN}(J_6e<f0t`59ySA@Vt`R8T#f=OCQ_U~bf=%E>sq5eS5V*O$F=Y%ez2jB|lIXufy
zps%mQVzcPv{))v4s;p=Qr`q)+=3AG?SeK{F!>E6pR1e>RRg<IKq>*G*%MW-h%>$34
zMMG3^9P>RF{iX82T)h_Ud}`V)EnhL=POdAagpbhog67FsU4MJ-$HU^eQe4d4Az<hT
zsT|||FJiQhe$P8{G$jdC=NxiW`ng6Y;?dHs{<<qY1Uh@}81R+qyBjav?0r;7U9$qs
z+D+Kw0+4hMC)yK<bx{WsK_$7KzZ5B#hS+eueh~JYTY7QW^Btw<nqz$<h7{CP3Hv7O
zft~beZT22l#?dvng?NwlWYGa?rDw847Tn2~MQ}d)_Gxj8*K`DEF;#gaPJ^JJFCF+k
zMIqG%x%;Du9N#pPB}787iF|GhbH@F6Zd5+R{%#Dc$dzOGlGJ>pg-e~Q*BYNUEb%DB
z!<0`mwl#_QAWBw~`;vi_jVa9-8mpJwCwansxEGvTv7XFdF_FsBc<yeGuMgkk?7xB2
zlfiX~!#Dumr#5)%Yo$G!Y0vgJ=DKjE;Mpu`Ez(Q?)-1V%*4->4#@5qJr@$PC1htIQ
z^_^@mJboXpuX)*C3)W3(1buYo1!uflN;+Zw>caU%<f3TYW`@4n^6}uT#<ion4F1o{
zBkRGx^~@Ayn1C$>%r6`%#a0_aDa9=N?y_)xz&P@wZ2&Jz7tZ9l?%)3#hOj;+e5bQl
zZ`|uci(o-_8lF`wk>UBdB}0p!WM_qiUdZ!6n>=w%I3gnCO}!pzCr+uySi@Op;Ha+O
zW8^_dQUCWph{dK1f=!(Y>7mmy;`7VTiTh(p^~_}(BJ4&8Oh6ifpLfKW{K+|!i-r7`
zE1p!p1a>bz`t5KfV@75-(%0;n{t?qHXxXo#<1yiJC?@Vv6qaekx1he9@%>+Y$=4IH
z!sfqduzsd}vB6fcl77`-HtqMB3#`GchTPaaD#+B-T_H`M3?FVh#>+@F3X^gy-q{@q
z1o;3E<}-6waPUl5@77Ofkv#e|sE=`wU_Qe=PR+OG*GA>m4f@4SCzyNKabciak>2-a
zg7h33r;87_N(s@WR;TxsiTl`S9>0e%&Lv|%><*Zon-|KTj~$MUM6aAb@-OYaMcV*7
z?$&iRRfM-U-s~iXcTFSTeOjKsoe>pvY6Ha!XDo~Hizzjt7+vm1O=BZ7ega;xTVm22
z9#h2nm{pvo)sf8h<B-Q0*^o(u767dmgZSPcI&SQ-LT&J<MGQmtlm&AMpF!HlwG4?u
zw%2M`EUXtlG&#d3-e93Dh(ivm7&+>}beI1)MFGi_Awz_Zpd7rC>cigI?o?m`oVB;?
z#so{c-8AGU(*_J&@vtb27`;s%6Q7&#?;<)11Y!Dou1M6}`7iSwMA4zUvcEbgAu6$K
zMt-X*QlI(@O7;v|HbvwOj_3~l$^@qGUC9I>SJJ=Ce5$VnujfGT2V23vhpDnYMR2zm
zNi~v?C#L&wB*nrx86Rti>+=i%xIthLvcy8;B5JS+{Og>hEB{~yB@eUs_09RFe#h4r
zKIy5rrnog5^CL+3(a!8M?X%!Na)_gX!CYWTUNz)??6E<n0qozm+ydFBFwv5{^2Wu9
z$D4_WmpV=;VNCM*pDQNGt9<{tHA@=Ae{PK<@y?=moXV00K$jYM_iGR!zc#m4l*{-Z
zzh?OE*F?$kFi074dsS(whI<m&x#Z7ck~Kul+6!0aP2%*DIx&K{I;uGbN24$YUGaYh
zBBw#ASk`b3ur>*%hr+=kwH~VO`GIQ;dr2VpF0m4RYhHLQSGf)w%8$ToduOd|NJ&mH
z00rp%1#ZE`{m+mjr)sdD1s8*PS^S6tR=9jy0v5b~`18(*HxZ{B5C-b{kX^huoxPUx
z6#ZZ5f`(jQ8(khZT9MnJlnNESJe)y!6)K2Xn7QMc#^6GJA%A?MXw1ig^k;o6WMj#5
zLg1gD)RoAjJa|gS6+ln;9gPhLMX2WeLlIK`AQW-TVxqM|1+j8fnrvWW(eh<cRXrd!
zEVGw)TsuZP_M_Xba{GNZgegQ<0n)c1@RJ?}gD*YWw5>U^quSD%1@B?-GlgIO$5O-0
z9GZ(%XkRcWLs{b(5nv#Ub+^}HrOiYF9)hU^h`wx?cV2kZ5MFJ{T3T&&sp~^_e?26C
z@Vu0vqw%~98i7vm2fNIE#O~*1t0yw?ylkI1$(y9*6XVO+4ZvTvrb;B(46{kvqm#av
ziI}x@cbvRfP8{PqAjt|wGZmAWxsLiT3kWAj?z$=XN#(uI72rVg8y+Bc>z=d}vo^-n
z;Tqzhn53LU{IrZ+q8@?c{*5MR82jtx5<WDPhT-Af8O30_q%2pO7-3zS@lja2R$H#^
zwK|z=d`zy1D#Tuz4sp9@ySgNUvGmG`ABGABVsV;{MEe%>A)SeBK<4UNcO`}mtV!Nr
zpK{^0h<r~Gd?f=>RS{5Uhu>rB)#*L=L;BajpS$a@%Vm}}msIr+1C8_s)iOC5$w*^y
z4iW8lSJoP{zv~71n6fuUxD#F<Dl<?2Sw~q<DsdFZ^#an#Q9jDZHcM|1a9AReNb+QN
z4$(VV-Z1^T`+>a=+9d$HZxW)qqub=LbB#~4f#W(fJ0Cn9umuarHW2Tw<$xdPuWev$
z6JbZgbAR{|fJZBpLb+SZF*-wpJKjB5+pvDcUg{^R>ln#u94M-bG&@>gk5}Z@3?A5c
z;ukHd`wtceHo~6Dv>4iRu?s2gtXm&sQxPgpEuMa&9b+WBxQeG<jU>?<j2u`Ul}cSr
zok~q4Yyj3S-D-wj3FqGW&!<fdba)Q-@681}kL(hSo*f!)m-%G8>f|a{jH)=QyUtZo
z_0=J~FRXW@p6(m;1Cf+UoEJfi4&47>bXF}7A^uNBCvhDF2mI(|==Vh%sL1^|T-sTa
z!LQs+&ZibfW^#c@r{7?}kI;{Y-|V~9%Y$0Ehn32yrgL3!!Ecr$qv)p=3w=OK#sehM
ze@jMnFSb6T6bHz=6ex4Bu3}|JS=>G>>@Ff1|F0Tqh)|DoI%nOs@JW6cEJ@u$2<nWp
z^W8#nP-WX?9()9-Ly13@h@jRmh}l`=X2n0~)@R$u0*TuW3#a?G9H_03L!JLY4r@&`
zK9z7cqg56!UF8TFz_eR485j(BEtPfvSQ?5JI~tgDso)GIf@J7pCqEg)0*=AYaB__8
zbEN(WC=m!J4%u;$^|KL|kVPS-uVe1BdOmxD7IcO`5*6$w^g)yj{ySyUMBzgwO4*eN
z{D-p<rK|>VHd3oxP}4FXW^Y;sx6@%-<|{;i6anY#J7;4SWdF$riXQ1doXs)8J7)t9
zFZqwN@e2z0p+J@o1`pkP1n$k%V%Nvm*U9yNq`{Cof2G5QzMlP}?3loXix$N@f+H-O
ziY*xnC=sf_qiVXk8)p(cWw9Vt%$YMLDkbz$SvY4blU5~>^B0!&;KEXIj)Mpt$BwDw
zL$dr~o0uw#NF?S22z-~KD-U>BZ9hBmP81md=ef&sbKINxVu&pFo!RuG;@g~x-D}+8
zLjJ6DbW=@flAX}4BWaVnGM&h$jc=T|AIJ9qZ3@(i(5~R#nGa!33*arap?X{EA-y0?
zuofTrk0#jD-7L#*+tc9>H#Q+l68x^W5f{bN6pq^Yfb}0`aLh-G19!}y1`w$=6=RoF
z`eD$&=WlUgWh`%JBmYB^K}#yewvrhHhT`N9^Y=%&g=0YM6&eD3;}w90AT)o4Hmby5
z(=SektT_#8C0^M`#Ny%P1%t$383lt9+TwIzg}#@lQ4Pv}!rZOK-8O*V^d<+x05#$G
zo8H%ZeByQ#rCk;R$JN5lyX;sjzx+wjH@)A1!^<HEIOwtbFW}Ji4jfJgK)~T%3<Mk&
z<pB_K&|?81hf`?Bt~?Y=y%_W^gM0LKdX;P+&BM4B!!7TmCDORhSY;X*N2oH~M`KV4
z$Z?;^qSU6g^7*m)`y!d03+P@WnQufa^so*2(OH5Dz<CS6wbclzZqwvpF097Pm=nyZ
z#uY3TD1$$i1bCrfuX`Z`fO*|8fX67C13w1}9O=I1CLvAgZIrllJGG{0WS)yP1_kEt
zS0ZHzivPfBOx9X?a#xu?s5OQ)bcd~;xY%5^N}$}#Q$Ke1_L2}(x_N@ssXTldXCTwu
zr5X%>;Q3GD+1Ws)xr%JSJZSTsc~D?C)H4mKh%9i?tKq^tbTRPS{BC(km<TwYD>u<q
z1Y%J)@40>wzB}d#&L{v9P8l1EB6rFF<SCk6&nadJ4Ym!ap2FLk+9cd-8;<!zvwk7|
z3mb9qR#W2q;c9K4>c&Eo<@@04)5y7lkJpwL5gq`9<Y4AU(td@beu)ahV3qs*EsUa@
zO$S~a+q;VV_a1oK9W*d>Y4lnNU$$c7pVHm?L-y^gc7l7UHFxS)c`jya!oruXn10s<
zKPTZa=p<sLvhMhdTYQBn2!!o8{xM4d^KCFaW}Ue16nPAe+I{bvtZh7D_m<^N&&csF
zJz2)c?5i}VOuWbJhK~ZY2AeO9ZI7VT4XfVUIWfrk1+c5R_7DKfR~I7!-dESe+25S3
zTlv%nnoTK+NgB_0)uDmE%tbCnCFY>T7=}CJttjOG<LoVi>gv8OUt9vgCAho0LvRgl
z!JXg^L3eO>cXx;2?(XjH4#Dk`=lQ?g@9lfLs;g^#JE!VY?Y++#XO7>PYfYC#C1`sJ
zV_8fsGqx5EQ<wY|$Rj)Yu<4)AzVtT!g-yMBl)bE&99BK7iM5}T_CLy*tHLIv1E#UQ
zmMR*lLNvNl#<4wV5!0)r{p-S}L4(E(bO2$@7ri;4-kJ0_KgZ#>jipRwcJj+b$CTG#
z%g+tdpI4iDe%N#v{FY9cqQ{y3hqOru2Zcy7=riuJ92w`nTUE#rB#RpE!jb42G*Q+3
zpo}T6aEE>DzhTT3A8t<T;)|^7$Z6slt6ISNHVF0jPjN!J@;{0b;4cWFJK#fTbe}%Q
zR|+BC^-NOC3&JWu<8vLiZ9{9(w6tjzHEGn7tul~O_!zLsk)PyFkYMgi{t`#bN~rm{
zGcLyhCP|SrJ1I#Kjr=2u0#gcES3}zA9kkC&QxxG!+4!hXk`&``boS90K@Y5uxp$aA
z17SJ8|Ne|>JI)U1fyv@keZ2Cw>N5a4*TXxSFD(}<D3nY~J#39N+BsTsdp$OCh3v)S
zuMu&dG939?t-i5(sQ7&SXm4-m{M<Hv+XxsgXEth^meHK_x3*@41!8(yU0ZBe)e!JN
zQ!%KTu|0>@t=HwBK*ak$2-E;bZO+78*j0`f$XbU#fc^6(UkjgjlQyTR{BT!7?OKOl
z_PM7Qq;keVtDM{0OzE)o@=dtLkPl6xeXNCWEL`|QzQrt3t@l@ia)pE<d!Guc0+lvB
z_D2<pdpRA(|CP+b!M~C@(Dxykw_55SlF0=k_K##rsEuAG=C6PhCl7`~7h_3`0Y83F
zJ1MdpbV8<?Y_i5~WTQZ_4Vtmz-qMTjZ2g?Rr@!p96w_eVaWoy$Ae3vXuU081K1{Aj
zxVqmbaC}9Fx&PWop#>jL`{S3b90b$X#zL^I439_A-?QG`KGuJP<VYS5A>Y0tPpe--
zEB3;_c>hIeV4MFFsR7%2E9x!IbT9v5YAi34&?uGx|0S4~0Uu0l5&nazr4oHGwN(8N
zrk0vbLAVItWfH`lWW|-T<uWHeN-6Agr;;uVgSqurEyHRt{!z>R??)qkcB)~u#9=Xi
z)e@ptTG+912wwf`;1;RoL#vi58Xs^j_LYoZnZKn#Mm^e>%-B3(5s#+Wr{7V$Ho1rJ
zjK-A=#==?CEG!<2BJ7Cy5*mjQHfvPNEJVTFl^GEk;fw7=?jR_*ddW=?+DnB#dr?5?
zbM5T5XS^^vg}uOD-0}>6Yc6@xr{PkopW<UnP_4NI2#Y_(Kilbm@IV|{zgPp1rMDj7
zb_IdvJTd3&P<2mVS08s4pOtJR3V&q{OGe1^BlYHMRMgRY6Kbp?PN2ueIUPt*ipg|?
z=t^xZXHW@jaf)rYjrV*R=0`wEKNrV5&Iqa9P+Xd|eSt-uik=GO9B*QanM7rt&i^JC
zi>KJIfc?AANrKAXk%g)IPk|&LQ?MmR({hz+O(d6F0Op80*-lPf!n*~6BdvOj>}vfY
z*?_MeB(wP#>-~kH(_1Ycj=pFHVu3}z(3|@k(kMc<EUL;=w{CL3yP77(=*7O{**gEy
z)~`^~9j6Y5=wFF4%sk)>$=lc<I?Nw5MOy5Grbv_jLsNbZwSLeP;2)eKJ^vS{@c+dr
zf2&=w+{oEWF&Q{gaW4DFPjUa|5BNi4|HcofxFPiu;(z0ZT87H9|4*qiWy&+2dhm(5
zu*@xJ3Bd-3aU~mJacR^U{`9(Z|3S<MWDvxU=JC@Em8dj7e=UHwJKT968UY4IU5V=G
zEV?6uEE0*v*-<vvV%|EBFCZa^dF020^bAOK9!IoH`ZM-reT)khLnNCLW#^R52MdTh
zv-Lc;;(cMoM)-{;qu*I>yE+fGt=t(h=3K%d==|reK1TkGh>528Cd&r+1k+z?npDn8
zIN}B$864`70m)WvRK)t~jwgcb%0l#7vvq+3HZkPlYgY<K-LaN4RW=<TUhl7XrsaHP
zXCVp2G%+3W@mu;H^I<c1aQ@9FvM}z`Kk|uE(!Q_doG0|QS?m}6Uq49Ji?_y)uLan@
zyfqe(80Dpfn3&{g{<-&1=KmpCg8-*|=)cj=$9o*Dqf~n8Gb_Yl=0z{77-GzO)9hhJ
z+$+DLWj-*B1o=N0CVKS0VHj)H{~d<0GX6ioFfIVG-aj1X9}JT`zwTTlT|ouP=fY}7
z;)@xySFG&rzlT%-tD15<FPg}{@BMph9CA3krX!X&p2&VDD0IM)yeVD^3{(k;VA$70
zQV7I!-%+0$eDkb(ITJZ1R{(S867<aj5g315{#B?5YiAYpZ)8jlwp)#frIZ1bi$8y0
zm@zE=e<)1h^}iG*2nBX9DuQyV7^^E5CPQSp%7+4(H`ss}E<;4>NBIpY!vBaPq&2cA
zxZU)mN&iVAlqku2=;Z3!h&~JiLiJw*@d&L=0D&SG^;0pDhVTlFsR=Z=FsSHPZ!Cg3
zbJR~U!2R@)eh18tb{B(1zIetd@&^RwM+ng_1wtBC8Pf~DTl!ZqJGTC(Vt#}W;Z^fd
zSN{>rncOL86y>V_5=`Z)55X+b`w&cY+7H1*xBL)H^fC&<B0WLK(7%G&FYi1jj>Plz
zAHiHSr~5}S;bSoWQB0_z55*L;hmRrouliED2c<=_RO3I2naJbxv86bU>W2Ylm14jl
zJPmYH7ui47pp@lWV|P^S-%#N{nSvp8=PgM<Bx%6v7$aJ62eIS9j`{a;mNC^foeJ(q
zbIBa;zkX{-(5dy`m}&D@p!B@Dqs-XX4<X5IdeA2yKyj7lJDmfIuoes}MQHTrTLn8X
z0Z>0;CXrgne~Fp}d-MlUvtb|lAZlaeACK}|E$rV%sr;)OzC>8bsXLOua+K40Z#9vT
zv{<EIG?457-QN!rDn|z^%E9+<6w5rS`M*T5{T<4i)eq2(mt?HyiQ776(o}v%xQ$?H
zo6Qsz+pWMzG}~$m?;@xJ`H%5msv18k#ouwY-`wN&>o7K@kv2Y#@E>$ouDa_V8IR0?
z;E$IlSbV1F;833O6L5Cd*nOpopkvSwdH)dB?$tYcOFDqu<H&bN6R|!?<OWMUHBpK(
zyd_mUxF<86Z9m%R(42^#+R1HxQry0z6zY{5+#LX=Pe+Xlv>T9C=7NCW>p|!EkN2>d
z8`3cDC5fN-FdibMf4m`zUu2zW?}R+_A&XnPu|X7F7m&VE%Ly?FQtMte3sPUe!YH~%
z4d;Hi4thDfv<2i({SE)py1eUuXx+ZYzqHPw{e#w}G3<LF{|%vlzbTZrvW_fTK1F;Y
zH1$f!*Lr##7*PsKh}7~tc8LHve(?KRe>oldFqQ0rAd5V80^3eKUliZZCA%u_8)<4Z
z2M6U1h<2f5x87In6lA_PstrfDwzw5!b0#fFbX$5q9;v}8CaQS$13NX%;?r#kL2aVz
zb&@0rAZ+sYe}4x+b@aDxHQt~;nnm&V*Z0TnHoDa#@m4cs8x+bv<3{%-60;AyQWb8K
zfqU~~*N6BAAZ>M!-ssly6k62&X-!MUII+>6N&YwNDEpS(^ev4lMF)FsUeso0;ns4L
zCz?&>as}`8l0Reyk!w0}@HIP^HPqCio>47AW05ir*bi{EVd9Ug51%-Xr@#KLk0-`S
zOFJ5K#S9nQ4W-gNSE~KTo``HSZ#sjT;JH>)G&FH<D3zt%X;)t9TQU)kNKE1K6obxA
zyWgbs4gzsFMXp80g!GCnPvx8i+m-RlmnGZ%_smI}39dgxYU2yhc}2ISW2zRH8OO~D
zYE3`tKE8i8@=zUXU8~O$pGJ=|$HjQG1fZF9GDs@6d^b%moNuJgP4`EW_ubza<uJ#-
z4=a7^XbUGIA<j*Q;k1Xwb@54<(su)IaupbeHl9D`9ahr7mJ1aKUzxVQ$~E$;{mgu9
zct4iWSy)<Rqx~59Eb>;KN+9l0``!R0K<hV<48(ro<7C)cG-&F$%IvQeumxgqTaVY=
zYBk8vT0OpfA3Gmg!gGf|{3f?9aNQA1Us90nEfJsv@sUnRcno&e=*}3|De_;$-F&5<
z)MovW&T%q+8GEb8i^X?WhbaE!Xr81xvjI*&D2AhHi%NeyCdQ>ioae)#Zz6k^Gay@e
z3Uqw8^Q0W?4~=GbK17$zPJ+?Od+oVY;drqc0%`g^R&F9Zbo(T3bi-A-3^zuN5?BUP
zIZ}NNdxCw53`;OH`HR3e-B!2xK88CjsdIA>3dRaiwR75^n^9tmWVezac5eVHN2ugi
zuAgJDubZr(j)GkPOpnVr(k-T8-5z1J7QnE;oxX~mt#zJ&<G}^T&$*2e&INQqxXq-a
zDvpmJ5TsEAPjV(X@BH<FdA`4&=L_zC5?WyIM!2+0TMjrEv-|{g@v?O|NQ%N+JEXJE
zk66y>RI3tw6St;jp<Gz@9bbRG<E_^@2<XTd8r8r?r-<rE-EQdHzLM1CkD&7e%$7E|
zTIw2RgDEZTuX?5yh76WV*euG$+K$g0c2V#wg7wW8A1FR#jUezN@wX;#e7k;%fchDF
z1Mw@N3+aZrSE@rpAkxd&N=~?}fJr1!ncgUf%_dW6QA<&(ty+`{`u7*B#@D5tt6&V9
z(J29LPsW1C<x^|v*LhksbL|Eos)1j}>{Pd4V?`yiot5=~0geEa{6(5=ZGI_vNAq{G
z869hj?VJ~FLS+5)M^o&9WfNm&H(>0yW7c_PTVG<5uSma8Wdr-H%2zp;YJ6KS0d^<2
zq2vZ`!cQhrb5oke0d=4aErhKIWx9{Bpns8S=%^k;bVj%zZlAkwirfu&n7tJ4`tSL@
z2w4&-rV9$C?_Cc~Wldmmr>eOWDP)1zL+?GFrk*vI1@2PHjFAjz$4c*~*G=6XOst$K
z&Dql$Hw)7Dk@onb=(wc}SU#H3&^2$T*#!P%NZNXeoyU5sXX&lLavmSK9e*=@7G-h*
zbAi5ZsFiU}Yjk5TsBHm~?Mg)n)XZ6C9@~$$j~-i$E3MCDVs5AM1d{xqqbGc|Je%Lj
z1yaIpAT<yYw`@c^^J4@B=xROXheN$;=SydznU5MJ#9q9JZkPqXlpOiC&lb{(yqd|B
z_bF<wJ|cfJypN(8Lr}Kf^=VNY)gHTzcQQ_D=$~$UHJ}l3F*^hz#}Us@2b2Dc*`Kv8
zk?-~|G7-Hb9)4MyD0phk^E5GSrp>);ThI(S%b$lDDX-hPph>WvFx^Oay1sqb-37>t
zEekg+@1JL{<^l@NHpAolNaau(mdn|nw)xTG6jzrtAsj529kyl-8?0HU&oLZUa?4qH
z=7%6&2Y56I3nc(!N|v_!i%c~gLLl;#W~|O9KAWKsali~a1$il+R^+e<Xm-bkIL)!M
zr<%HL!YJWg6|Pg;@I2a34SjS8g>*dG3Vb`}WW4iVS(<+IPCgqbs4sg8n<}y>?Tg(4
z;BaObtv6P+V=^!9j}O|)20vfJU9Ec?ev<Ts68!eUY6`%ibd@k%bkw%q#-;PR2ZtGg
z>oC`vz>!(~w4py$4EMA+zc1hvA74W7nyz??Jt>_Ur*t<BNgsBFOjM9K^i#22S>ane
z{;l1?&|Ig%aV?p(sfMd^#~-3({ImL{ZSlPsU+0$LH$(cm08i!<m31FG$US#s^In%|
zW+cYo;Uv)dN|cMu%+1P@RMVY0bTS6Dkh^a5lMCBo94@=<MDqRQV9V&b>HYS&^j)-(
zyOEnp8!y$*emv)fUo&7}WM${}-Q}QxA#*0rs!=#AtX5#P@hHxNlaIQ>q?CW~*ngZB
zV&(eo$+dR%%!Cay=H`kLi|EcnJRt2Hk8DKgNEH}9E5lcuDCm@&`y-A>p%q2aNc6F;
zkc5&qoq=~=U3C>sxxWi^*ZU^VcP{mYG<0@AY2NV6ad+u*Q1!?o_2e8l+OCB}u_yQS
z-RGMQ`}l2n<*SlyyvaST@8)c@z5Oxce2S}N&h-wBV1~{Y!Jvj$9uyiPyp6=Op8>4f
zR2Wb{g%G%Rq<Ma)KTmv$WltIlpIiD^z?@XL;@)VgTOJO9uR0Mu@jnaW#GjJr&s6{J
z5@Y$27zH&WGOzd%S<hTq!r42L&q-qKMSWIjt@~P1=F!B&snbO5o)D0d>==IXcKSHy
zrk=FUP`AHiZi4k*XZ0zb67b_XFgQFN#nd^_nb?Z7JtT34S`k>XNz>(=2#lHj)md+n
z9Pa+HeJ*5CZ_h$1*evA!NtEB37aUoD|Ac0Q(x&MxA8J+HbB|ZY^(9YpCU9Cgb0RW4
z(R8gf+|BssS_RLXhVdjJ=~0NnvvdHyd-3M(?)?T0@FKWSC0m!f1H`sIdq;8XrgMC`
zySo|0J3ce%cpX;*qa)0Rl-?28Fbxae{f6#g&un%kP?e4rb<n04aB@l?#}(ZC5Hw%_
zf3xn!`DQ(~U#jcXREusVjlI}h<ien}mrJBCdtmgP_SLFLnO#D*uCTAXE8@{Zw<(%5
z+K1}5BLO~?Y>SEwD?sDe8Qe1ckQMLyq5&q1b~i;qrBiICrjfocstk|dC0gaKtg{j)
zce>|SXw1vUaGJ7yJ+9M=8n!qd@OjV=gMJz8SxqfbZP_+&X`IFO$!H;|qqWu+tp)UZ
zU&m)fNDce^w)bl;#ot@%&sLoHh~k-F)<4pq<@kAmSBsxcT!2Ak(e9zYt9BZ$BN%b}
zJ((Kl6}QW8jERBvlQt%sQAC{gstHTrcE8+87ya3aq7a)Wm+o;Qjy%PqL&<Dm1J-+$
zck!7q)^lX*?U^FMXiR2oX%6-7(-22LEHB5R&FzY&mGO%i?QPSYQJE@2;Uh=SzxytE
zR@AcWgBSsBT%Mqe>9_lgq@5@xg3($Y^ZlQbhuR3YrH!62MIyUz^3$l9V$bq(MQSu-
zI+4zPN%Z@00Y$aCd-e^hl+W%h1J;EeK8q_Kc&*Qbrrs-l6<TNwEN!aN=E!Z#Q5;OQ
z*F~Io6s{k_aQlZ$!j(Z+ir%*>a{T$-FEMtS21LNHC$x-c?)2ix4;#qt@~g8ImxASv
z9~)}UdzWV)gXrr7Uc4swU(D$+ZuA=1N0CgD=HUgfz2M5Ox<_q|Emp5gd^kPddwGyv
z9$lLr*f}y}S?{{UAky$VfL%b~+0WA<&08)~Sog|{b^LhAI;{ccb-f9m@ul_Q?met{
zPXf@%${bPIM|Dl50e>#)r!xJbJK(|h?A3v=I%|y$(rZF%iD7v%`6GArknLT-eW5$8
zKBnw_7}vuYt^{`!=QK_mQtY9z!qAjyU7U<cs#EVbGKx6M;#FJA`|EfZ=~qI~{+O)3
zJZA@7BBR?s3z0`h;<U^-C?^ICm*Ky3)kcBkL+UNqJ1X}dJ3{BvmKnL8q&5paT5fU&
zEQc)Jr&nfS`2{g4)Y&PY5E5U=SADmJG^}<<{Y}+Q({!Ba2%>drJ}Z_cDaBnQQ<u`+
zyNo||Mt`<S66+ueHwrJVC|omm8D(#ZZ3h|Jb3ore4+M)c%KHNXJ?Kn6jtRl2e#-<P
z;^7K3q!J>A`##?;ZI!lDs^D|dKBBqweN&K#`)N?YJ2yFA!wa&bL#b?TN5p$-naW4Z
zC*?k%-R1Y)e*@_PglV#j_AMV-3IlhQj3_8^z6u#@P|48;CWG;YLKebR_A0daGZLQu
zt~sPbqj-gQ2iO}mQn3ss85{`9L<B0}sw)73u1galA((S#Q>Ee|54-jxRBMC<`5te!
zULSulF8r)}Xwh~9tr@a^=G2zUx@KN>YeDGZ4Lz!%Y_q|8Ma(_xA&)=xDhy5a-s*YE
zs}m$dMg)$7>s9q)ykelVFb9i_qpxEZ&xm*d!hTY{!ezG6JMZjpht%K*Z95Hsjvz}N
zOAajW{+4|<D}HbkC?V}Hl2}~k<dRsb8)hEuQII#))hvOx9q+673UfnB82tJx5(=iy
zq(Z}A&E@QSH)SaX81gnPuI)9a2|GTyAlckN5A*syze+$LQIlr8%(16)PD@6Ke=+GD
z+lVA}FrvO`b@6!n9+`p0tDz%s!VQZT40fgag@>Cf($XW9YLTF{y6v@6W!48JEH;;k
zWu<;J#QSnl=BuqGupnfAluXbzwdSOxX0~f_a2b;6>tB~bvws|oCIDXO*RunGd<W*4
zCabNR#usd7&Eb)zIdd9nBE3s10qStKAf3OBKWhVVVw*d44wtLL@_8D_H*t}=)EJhQ
zYSM`)>E%`(PShZG3@n)23rEx8Z`-Y%tBue2v~6GCwocC@1@U$b@o-b?aU5)pyyCr<
zBn@aAS7J(!)D;m47tNg&i=4P5w^%32*mmNweQYXDl7hLHJ3jdQRgSYTQuF)(<-KG|
zfx|Uv!&yYqe$h>&*}Vy9k|IXJLE_;wC-bytomW3oIup(RG8HWrRe|;8>J+H=Ms`99
z)`Zd~`yCz*T7S7&`V@wPp$ddBe44Z32$Xi%@q>4mXH6=H$5+flnPQRqozqa8zt4eG
zGEIdXTKM^Kxh*^T$wwjbKG$X0w%XVjC);0YOS`rtPU^Pxm?s&4mW?A@<-`rctT!ZY
zHFi}$D>5#$AC&k~imimO|8dY*kt`hTkJWoxD!(lAMTXJj^3OA}rqQ<*7>(?vQGtN9
z$|HCLNsCJvRm1gMz+lQ_qtBAs(^xfil93OWUW<7moe|cUF4hof8@AjnBs=^=`BLFH
z75k-yW%wa6p{*v+NbX~U<><3W?vm{K`UmWn+D$wu!@bNrLR)-(W_witoXM<<r=ENl
z@}+JW>Wf#Q)RSEBAO3_}&g8r44m}tDZ~J0F9IRPrnox3|_@7?|K0>DrrNY5()buN5
zh}}hs;yH&5vTmF`ty|lb4a*bO%H~KcCN&k-;cRm3#bYVp1*eP!g|Mb5`z3c>m7UGj
zyZ;DMJ5CTB71byyjw)}c_ctZCFg5xCH!p-0LiDi8oVp>gJ_}zccP^FP1Uo$ccO@98
z?E&)Bl~L2zxAe$eZ9%`vTZ=dvvizD;xFwon9WBP-^%Y|Y!4!W_?Ay7h?v^#aqnBrJ
ze0Ht+q3}w8wdBBQ!y0;BZntZ6tJ<Zym_y(kXjqhn4W!=^o&oa+$={3Y=4atx1@t5D
zBpAX}myfF~Bn51kXZfX8nn*KRbtjZV?6L@h9`WE7AISZm^kAO6-*T`3!GuR@C?&A@
z2Bngt-ECt1mnnE573~3VwV$U@Tg^-2iF&T+dQt#N>V35@-AmAt?_3T8m3#bc5#1~n
zs<}drGlfR|nLgp2KGSikMDUiCItMfSX7Q)ap@LE)kC|(M`k*V7k$exs=W^eB=WOA0
z>FuH5HSW$W_r<N&&pg6do8DbFkptIcNC@~aT|F6IaJ*l14CvrHTZCm|4*|=>_3^aL
zmwBMj_HJyO2wE>&_)-#_3}g6C&y{zAIRG)^*DpG6UBeL<u%z!X_SLrtAR|_8quA6{
zpgTTqpu*t-QJ95BeI?;!Nr>LB(!NuqU?}C|U=!R4NswPA4=rAwfMX6E?%x!K9XAT`
zXOAMEPC?J#0$O#*&Pb(U<hU7o(rYVK>EZzz1r?J%itfZj%8PFsV0~<+F9!7KPu68B
zjdk6wrLcjTYNflKuJ>^)ACSQ%K2eM7M{Po~^yE%Ar&1tL3J%r_B*58hSioD=^JGdG
z{``f~s$;Hi)nD1p%jYaFtj}{XAp3eQ3va9=PWXG!i4?z77o{C-TG)r}Mtvc>@jF1H
zmRb$30TY}7dB5_Foks%^mzm`C$f&}1f|h(!B!9A}%?Dqj=qB*GC~Lk)TggnI0$wqu
z<R0dg<X$l*M-Q_Ue9oh~Ee?d7WP#-u$QVJDi@#o-i+!roPnW}!*%${1guq5++PPik
znLk-cZ(bPy=?FN6yW#l(?tG}d7d?Pbq|H(4l+Z;=91hyh=HQpFII$ol#TvpeYO>&s
z7qp0=z!Ie{G^N%N7`;5h^d8+nO#A?;SrXBogrwu!&tF*Cn*WyH0CSe!{pXxSddYta
zNy~A{*~ZF%zVV@)UFM{K+ZMXpCU)3OXJXF&vw*t0fMi;@#4!}a&z9hBH&lRZ<y#q(
zh`)LJ)HOD-cu=3PDw%3Pk3j&(Mf!mJ*LF_)!H~QP{oiyTk6QUo2xbn)F)xU7%wPJQ
z39-^XIZM8oOW%^H&~c}a{;aUzldfCGfCTar2e)%*2N@m7GQQwHpWG4@<~;lUu!yFe
zuvb!N<mz@0c}p`-I<^^f&RGI5-10QcGYx1guxm%Uv^@o{ISRhVO?5O^j*xP2*=iS9
zm6Z03kSAs*gB#iW#AD*(AafyeLA(nSqF11l-aU<v7dp}^x=Ek6x|Y~*JZ+Z~&&8k*
zy6SRJI8EN4s5XB^)8ut7WZ#q?&K!}(WKdd*SgORJ=`(Lqwx9nFBi0E(&>u`R?^dp{
zlY2;A?uJLm+!4f2d=m5S2-I1^d#)p@>8SRIbqS|w@~GCIV1=oWYS4Tmn(LO(;y9cZ
zUUO20P)=8IRMiL(me`gbI&f&p(Q05jQRQSvpl-h*%`3G#JiymMKlr2NB0aP~Q~l!F
z<b*)gT`w9M=_0t)#03Hj*R;$MUGl<uBKtD0lck}SrMxCvpB&S^`9Blagwgl?f)DmL
zN<GR=d5UXX-&YihWyrST&$m<{voUE1o2+nR>#|}1$%N|X73UJ$+bvW25j-1BQ&^>&
zbaM+aXVcM|+E@Q&e!?N&jSsio7aJo5rhOu3+GCeCd9$5;_8|}WULAfJ&U8a7?d7n6
zD_0QLX0Xf{&A#7bA)w4lQrEkcyGL+)_j>j^X4vYHph0BKkN9mw1+>qSEwo+ZL6oHs
zqD`-#b%t=T+tD0Lf(@yj*4af*w0Wmg`=G?X$jw-!8Y(0!B$2NzdIJ_ON$KfXJ~Kvv
z+m{|GQ(hoLUI7$fOFpz2WDJ)R6tw#omyt|)>vG-NdN($Mxz|pD{(unXoY=i{=d<PY
z)mlW4zwVo{QZZVk^shg6DLZMT*B8VE5n>8e<_RF<ts(rz^YggPI0&R4Z0W|_hBmJ`
z=6G0FyUY9edMHg<4g{8&Vzo!@*7Y`|wDojx2uJA)F?SAd7gQ`qC6+c<d5p8Q91e6o
z3gqViKD9mA=FSZMR9%$iC<b?I@!Oysy&a69U5gDyHHXfJX<x1)7kUm`RZHe+Z1s9i
zIp&^}(&;+;{L7zPghy~)9+~@<f8pUIG1u}^k1v95SXG$Bp@ab_)kLQ5R=1*aj^22t
zB=na1GSxs@jXM-B`nETz{-S!~<>Xmvzr4FF&%3><Y260NA^N0>(8}T$JvhX6=J?5~
zswX86g@-n!aP5nTu`Xp(iOULz;G<KjPw`Qr;Z&lAXRadA`ikpb{SgA17)ot}tW`*e
zURKV%-1^I|VNGrseZOmqSW;q?@)Yu}9#l4PBa?x~#Smw`5b+ar&#nl2<9Cyoy8s<4
zyX}xQ1lQ2aKUDg16w^%c5mb%?jxDQ1M^Wdgas>B$(AWz}<nC(Sj>&>5;r9_=dvOi#
zF|<im@6))_$MrXa)FF%KZyZ}SESY=Y#HA~AsM&Ld$AS#(<Q}moIY1G*z%}I<=#yt}
zFBX7XTM%2HeJqU=HvL=;D%RX%A*5XLKSi)=Z(Ah;kK<Qg7CB}n1w$-;9Q!WfDZA!i
zwoEH>%bh*DI4p6k6(3C8Fgsaz->%<AyT^Khr%}}-`_7nnaNvlZ+dh3=dJg_o<3I&7
ztQgu?+4pwmxT2bJF0w#(jhgV~9IcVnzPtcHLHHC?G?{h_3Ld^o^&5mi<@opR@$dPW
zSEqx!tvBSlWklV{TP5+M^)+X17paCdCxI}`&x#Fa1jIu#Rxg)|v#5Culn_yOJa*V}
zxH_RqCu66H3ebaTs&C)luWzk!IMz7M<p&QcM`}SwB_nub*5F=V)CUNvZN_sz$;bKu
zgtz<%ie^sR8fo#gCZ?-ncr&~^1HAp>ZzI+bn5Rurb+gp?2#Xxkn2W&15NpXUc6brD
zjYzFrcoUtX2L(kJ->$cTjqq8zA`Mj&I5Vy$gl9v^=C6T5ZD=Ve;7I*7Ka0@&L<FzQ
z8X<aFEu8DK((x?@a^_7pi%ws5g9#E)a@<U9Uz{!jMbVPMXXNv_1gloYtrvBeWAWFr
zQ??FoR&&Ez+8J^_s=b}_-KeVV^2puT)skH7%SBwd@`YtP<Hfwx^bJ3*U~I;{n_J#N
z8#Hmm{7t~kG3c@GwNYf{x$2pvCaK1FJlw&h5*A0Xze!z`g^$nXZQqjGh(|pzYp0Zd
z{C!RSxqwMN)YRdv+uuyS;mH3j0q3DO&fFiH)a#C6oSa<%IxX`!EHiAMB}d_9=-U9_
z3Q`zop{((!!3@f9QXh(uY}D`#`w~N<DIY;xCuCmdPyMEyIG%A&BqEnI==Wp$0oNd&
zamY$Ovx6YMb!z{2M0#M?TF3-|qQ7b_+YR-{z|Ejwm}n|y&9Jt%zOdPbtSzhZ12<ua
zw->i%EQ{gs0e5q1BKbwkmbo_+N05AN^a<+c-c;Ph6q-kW#eoTpS{daXjL4T&OVcgv
zo3j*CK|{$b{UA@j+|_qHTHM_?x%2nC1JF)3EIuc~WV^-f?W^>ru~=DPfvBy;v3VC(
zaZ>2I&Q1pfAxrc;iVxLy;+GpotySj%&fbj)BrX{52zLI;Ss5ZcH5j^2XdJEhr>9a`
z0U+c^nF$P~lAJo5vw^4q9($)kIcAS;Z^khGPJF4o<iza7A_v>|v{P$Euw938U(@(&
z9EwS%2JDW`>$$>^QM)HF^8Ns-Rwb)IoxYTPr}Oj~nTZR#1GIJzvB!mSj#{dmPyqxL
z@*KB}3r$44yWC9zJRX>^7uxjr@FI^TpAAzG9<CBQ%xF7WBf?F&rR0hWVnaiTz^9=&
zs+MWlh*R~EA^n;9`f`5Wepj2xtKLNm9|PTY-KeP0vXA*%ZMFx@Y2VrI&z9udlOYCj
z34&Hh$KCXIAS8*Fa&zARsaCdhy5(vfBVuitxQpIq>LX~8XP<*J(IP3xS7wwb?zTl3
z!7MNm)SV}LAEmRVtp$Eho$(nA&!Bo5xQ~Xq4&1-ee^tJ-DK2NBZLa4tw0*wHT@nAY
zcXxGl_ZD!|auW#nEK4-KKU+O-y@lFu1S`M=I<Y=x!wha&=-TpRRc%MirsZkRJ+D1j
zYtg@MEhM$4m{}B*g4v9u5##5MU8@(syp^N6=4VvK@2@?jQwu&Dke*=lVC}%V)UbjF
ztBhSvP4e-}s|>t|J28HR=+Kj%Ta$(I^OV(PVqd&pr_%*8(tRT*hpKwATYkU42Fb>s
z#w+Ac>gb5Ye`Oj-kB?!E7|fdPQY7(}8Yjb2=AOii7tA;OL~VCNVJO91u<Jo=C>7YF
zhNbjL>CIS%YDuW!J}E{<D>U|CDxV=Bsdb+emj*JZdH{xKtA#(xHb-M(hKc=9JX?W!
zmu6w1icSRBnEf`cP_5IXMWQy|WYN-N?(Rb6on(iB!0GC}{;5Yz_3OwI5SlVU7)cj^
zekGyMNV83%qnMP(=uatw_MW4dlE*4(_DML!k8n#cA(axJfJ~f9`Nv>}zzQL}Z6(Lv
zWcyd8ZfArx4kM?fz9W{7#r%R|RH^_)co&o&n$rY8Wk**F9#&Y(ZAtKSqLLG|Yl9+T
zY37R4_raeGZc%Q6q3QmH=4GD{^rBFW->w~-^wxuI+`@S8i5J%LVeRx4=e{`0Bhrmt
zHHBq}Zt=Ek$4-Px?UAN;A^O5K<dL~qpOJ9VaEp$M!x@+6V0a;3-H4rZzHc3m9?VOd
z$s!jh_2aOTxD_wSzoV)@1^?NSKA3+`%kRriGMbUluTEWn$4kn2I3Hm42WMJhKIz!a
zt?A6kB-JCAafu<aHD5bqguZKQy1@F6-UF&p;OVZFGaEtlH+zK6pP6sCGT+dTy0Y5n
zZS{7Qijg=(i4?YBi7rR6c+3u8O*kC=Fh9iu8d@-s7cjUm^e>;WT;}h>pQFegIeJrf
zyV!mRf7?LH0~<iuH~ry<(kbJ*@dw`)ewfx~WSfl2(ipGkd=i%UXLqZJ-Ca+uxUP)V
zuR|D;1-+CWKI+~>wayKmuhc9vR}!WWt^A}%VTi+FF85A|`K#hv8$k=jrRXCD(R$G8
zfPtR`FY~z)Sbe+vrwTjkwRNf?=0Q;6_qz2*->XVjckK(xZw*kNKJBa%_)C$Zo}8sp
z=#&SfAP$o>1BZlrnGKQY18@q+q(<nrGk*8gzg0%5hPT+MbjdEzZog@(QHaDH-fWi{
z={pC4-dys1I+6^k+S(T(IafOUgd{#I2!N5MkZ5n*dJ96nCrz@mLoIguRvxohtE<0K
zwM(c;5%OYLr{B#eLY~}xF_>1S>IoGMEH8}p$zGFY6ILd#LV>7PCXdMK{B>jfHEJm2
zP8ZhL=D|KAjF<7B&lLZ~!8MxR-YBiO7>(nT=LWHoAL&iZJgQ}C`?6}^NzDi#(3_;U
zZSu3Z4kxcPGF?0Vcqq9DQdvhRHl0#J(}1U%ovmdQ7L%0W^YsbUR+{LPtVwrcpQPp3
zGCSY29<B_IQ<Z;YQ>f?_+*{~`>Kv_omM0#b%J_uqAA(4MOm>bQY*@R{YDKe<9(dPL
zKY9dG2Q^Rw2hY?+LiFNk;|2k6B3^uv>&W*+)znPJ|L_1yB=SP4!O>)bAmJf?<>9fu
zXvq+Ean!;YF@ddq1Mm>ZSo-gfLR;>!gv(2jd4hvs&|kyk)NOD%jI-1Ig}U#esW@eV
zYLQbv@?e=BS>=7u#<G?0Waz&a^OJmb6g<lxs@fR;Ygb=IGV~^?D?oMwQo)z{+ylEa
zwNDx)Aqf@_fwCtxNKRwB$IOoekd~6k+!F<Vr>j~Eq^!}%!m72c(Ljh_onsVKOpnPu
zRj8q3=}ISU=Xiz6fb;Z&`i^aYyb*oG!9*GTHEZw>jtkkRF@vdy@4lZt2mE_AUZ~;2
zwh)j|X)fSCqrZXbcPUDDE8K;YCi6Ow)B^<smCwo`<K3VXk%lHaTvW=6o-gD(R_9d4
zKhffNq=)j>O@XEpZ@VNZVlmTBRM|wSZQ(_t50Oq+zx<bJ<^~jB^t<fXA0c)maIYb~
z%EH6EVD?8Oj_;svn0&e$Fz>Z~w?MDUg4o&pF$xbDwv?QF-`%JB7&`dDIJ16c#ey?7
zPqRQ%Nw!!YQ-^z;y<p=y`uqzyx2zyBvcS_&5F$}U7UBNbaXWe1-Sy5QaJ48@d(E>x
z&oG&grbANcjla1SE3f)Kh3ny2vo3q;&sr1+HqV~Q!s<x7N>VM197tw){BoZ$rOyZh
zER`tMvdx^z&p|po+=fg`O<%+(KL0PgQ_=fh-np1{-8`%_F(2Rg|H7Te|G}M}q>ShO
zvlBSKGHhiD8dfLB-c)jPt*;pYHW~q5jC%a&a})C?_1}N&XpUkbG<Im}<mxbicmok}
zY{6y|y<RwqMIF(&sK(?u$}trdW|`i&y0%bZ58()9pYzP~bLQ6T-f@tWjju&gg(AFi
zIM)OJl1)Uzo_4I8jox}txZ9-`lTTFf$F6l;EilaeAVl8@PciFBppxi1;BaxUEw@}m
zgxckcDV>5Lk|gvr?hBv702gR)8Pt%V4gHR)lP?Ytl`ga)eFp97XfnQn^8qAB_7xu5
z!NUqqoP4D^KK}2c+&);5h-je5g7Ja2;&#vwD0DJ)Z_1{!>2lN{xT9k{8M#){L7hjz
z+UaZiuGxydE*Lc0Y00(E`WypSPVrzO8{%MsAo^wImL!%4tzu5@z<FZo+=)3d?meIM
zr>zIxQ_TI*54CTg*!s%?NJ4?FHw>|HeQsPSmrM>U^7dGTKj48ULr_kgCfyZYX*`>l
zn6V8NeRqw18t_hLHsgCb5TL49y9skPn~pi`8A%wTyWv?AD5xl?LWazAOU5|b^;^b`
z>MAKK!nvv=@gu<H>Ha|aX`Yawk_B?$7ULp{CBIzM85`k>mQ>}{VVhXPky?$5L`O@>
z`&v+<rq|ca^*Rrs0N2fH52Df9HLL*m3kl)RzfOiAYT-{#zk}%YhzB^4k{hX;x+)l}
zQ@~&|myxvf<z8dp$x&zX5MlL4>!&%PQ<p2gHbX`nR3`%$TE$6~tRO?r+X1#xwhQ(4
zN8<Z?Z4tlUU~FgJ?jcvZ?C2eaoo@hHHaNtnH3vIiiv2-rR<!=3;IY9wnUL*0uYDx_
z(8%cHmUHX_FJ7bw8%Q^s3{Kga#`9i+=D%imC$$zEl8k58TZ+OQwtW-#Fxw2YK3T)7
zja^cL5zu0h&yWnil2OA;BlAI3-iOkv*{ym`|3iffwZh_Bbi41Tu49|YY?U(2CiKCT
z6V&!kxP4B<^b*_>Ey3Zi!hIvqtxf8>r*LRE;bW6YvMEzUG6MG<;^$qJ-DfYGDoXw8
ztC>16@I8#ZRC~IQ<*xc5aqiA=w7#$9se50A0y?h`Ta9Q&+*>K}FT!>3<ou*#oONc<
zAm7NW2ynm#2KE-^m)Fla<5{cao(uMXORT1Phm*x28;z}jfM4{Scj?4AN3A@nUt6&e
z$aq|S8tyN%Z)Wj&FOE-i3TiC7n_c9$MZzaduS>gs3?29KXNmY88IS=JK@J`{zq2K8
z37~zJ0k`?_kOn_LrB)u<_K37t^D(M7`D%x+Ick}u5bg6V6&90l602BfW>OCFYX)0r
zRpxIdF+Q;x{pCd1^VCt#{e{|70?D4fxu9v-CQcp_<+FEI-72=(u?5Qw+{{<5K0l65
za??HYYAxe{UCVP8k6~aKQvkNgV>(Q37YIk<){{q;c+68g-<+14eaVw0e_EmFVdZc7
zT*n9Zv;G!;kjd~|)b$i>ZS*2VhT6Ld9)gkgexZArvHaLA1x6)0!zx@(;p8ae@z(@K
z#$oqW{vtCwLQrwtu)hXE6mDKS)dRFz9*v>o)?`#Pw8l_^);_fvR65)NM)asgJD^ed
z5RrfBMcjnf4R`S;r!@S#)!UuQv2$kiKo~TFjE(^wJSEN9oiNVYdhYCz<*@X;{J;ZR
zKA5n+_ql}7v}pT$<$Ix5?j@V+J<2Z!zJ@_(8TgJ$RoO3J;cRS6?Cd%3e5r{4>>XA6
zVt<-ZicEi6OP}7wF1-o#X{Y>Xv<3p7-JSDYS5zx_5YGT@4-UmuQIlov%QP_37LGlD
zEbyqX$mY12n<&Z3PH`h#q?)O5RTpCH_)$X=RAjv3;33|mX;Yw>Zi}9&(e+y?FU4QK
z06er@tu<;r^;9yE@j7J7HK9STCg@LB9J?h1yOny~f6?-7BE;9B6s2YX>@x7Qb*`0-
ztD6nt3=m1`3RenId_n(VyyqDbNOxSGdrD{w3@&Kh%dU0bQS=4muX?In!4zTIn06q!
zuBtHF<@M4xwcX(Yf-XeJL&?BNO9RhCJ1LcK!c}Q+f_iEyH05`UNjOo`(`pzZh*qNH
zpg!lT(QrAO`&T+=?E%px80oSmyQO|VCIVtpj43$jDq(Q9InclQ?(8G3lN)6a{xF@D
zG{K{+WlqjFpNCBOdOZAX4K#N9;EFc)yKV~pd8xru>urC%P>Hp^(`5jo`*9p>_D4{7
z5g?#p?GiDDXOyw=%wlTfn-p}aw8+rfJQSYNmX><ibW#B^xe0U_J@R~SZ+oSgx0e|r
z*l2zXTy4;8z|d?`^*nG&<(=lhYq^-2D)PVKItWkGbSrptDAN$=8;@Em{=66ya&T=F
zHlo~McvspdRjolEpYBwjzh`uW_%s|FJBg(@-x7`>c_-|3h4*>F%=GOGNmv$ZUkBw=
zDR+j4SX^M%Pz`{2++Ib@xYb%A922*IxFbtHRgX0(mF-CHQBx&STNVD}6InBHO1$V4
ze2_}BeKTza9Q?8s99tT;HeBeU2o&U7NarzP1>1Zne7=^kGB>ra{6?Yf)%^M3?|MM7
z)5|tX5k0vL*x?sFos%(=J@QN3`0-?)5fe#`q8@8sp9-YSO-T18*eEevP}m2z7V$Hn
zVb_5jI-R684;ib_N|yGuie|$LYfqwEZI<EKUi+}|I}FL}3x!XWiiuNhIPy|EdN+a+
z1*xS3KXKb>_jdVl_xsW|vpo!T^v~GKi4Ls|dZ6E9m_#bj@pf$FVmCD8#Cv3%A>^rQ
z9eD0ajRCEdjs!a>upWXx_305DCI%(<Tv*oH4kC!E*4nv#omKe+!fx{16Zzs!kvV%4
z4sNv5qEktrQNZ_&_aC6-rya^9yycvfMv+#KRUqOo$lhnnevlnhL#}^nA`Cx^6}dM0
z#N*+uyBJ-vn&2ThbJ>JqAK}jASymj@u}4>+!j4~>Ix1x!gpP7WuR{ZSt)LT`GtN^(
zsB<fp@;&NUjGMh`F5I|1q*LdPBEz;ryIZ518?>t4bSNm`3(5@2OmOFK{WmT)s!HaA
zH5IZ9vJCy5<W1O~ng+}!fmp!;LC(-yZ?tQOOU`pFudINufzgfs4#xr6nuZt#jLlo{
z@TiOS6L2W_m3f{X!3JyoP~7$<tQY_ERv+bAg5reb!c%k@r)ij_xp<MU=*aag2&2%e
zkyWI^lHC{>6<jMEa;+4LI}uTXiBeMhq%Fztudu<-S}IzN%|2$IG^>9ENF>$X_SJ0v
zF&6le0zMAnVV?(j#S{uy2Ohp1ou9`Q3WX@ZVGDX7^cANInk601S(g#!^P==2euGas
zT$#2qoh5c?B)lTHLL!w%fkw-4viEkZt+YQ-5)u{KGvuih)#FSiwCyEgWCBBgP2ImG
z>&ca_wW^(0yM<p(2PRu%KZ^!#%7ihJ!S2LZVI5fg^fL2{+1v*nBvINx@y(&g2ce55
zlez7W`_u3exRdn^e?e37Ri!7^)mM1wn5!Z*gwK~~J(<vgnCb@vO2iflnc{P4HS|s;
zdEUi%6cM8`SQd-0vg4#!cadYZ){%VbgiY4PfJ|%HsEcn~eGX3Cf!lA*yLjpZ?S{SH
zwe1%DQ+*bpWpf5N`6VM)bK;~EKSw_iKONpP)?6l8c|fl^Uu(Fp+W}?A&6;wVCIGTX
zG6XRm&V*x}u^iEhg7mkjJReR2$m*!iJieKnO`GJl&UscFeYs)mag(~4N2YP&O5Iih
z&-UCsm*SKws#mE`$9}SOX>~o(W_O*?2s>)wn@wugh(kbCPeN?4WI_+eN{^yRC@EaR
zMfmq|?jI_tfjkB7Di3LSYle(gm|nw>6+8w>gX0bL?(=#~wi#!Ee$1))@2AUa-%DzH
zzkm0kz#L49a997La3>dlTx~P&pIM0D+MG1=Iw*@S9a-pS&k?EX-alW#pE`=58A<+Y
zN#82}X0{zjTo&El!W&1y{l&ASn0mEli542(kJ2vPq*%SpgVW{S(%JAqus0-UKYDAv
zdiz$OEbkw1eUvjQnRfq_GoqTPZ)ZpK<*aK83!1wl^xpfz4^hR{(8b3XOWGp`)CUUT
zpM%7nu{3)HJydk7%mh^Dpaaj~n#*n-fw4kwKp;o;c?+)D>zHz_WkvQt_Ncz)JP;ZL
z`RMhGfMe+9?Pu)F8_YdWuot0bp!Ibdg6h2tcywzBcW>~6X=;oAbPj@^#<F3M48u?=
zRj0*=IOOEn(VehO)u$9<DiiluhYXFt>KJ^D+2?n4#G2KQ$zNH2{PH>0LeR71rA3$r
zIEz#2s)P3U9HSH%FS5=(HZR-8m0&H`HPbg*vQ&3_5;F>)&}+N0uRs0UzB{aNhKqi`
zJO~~YkGh@6Ml?lg=V*Wb?rncfFQFL!4Mq+(sSQ3+bNcmWa6KkxLVODiuTSh&;%Adr
zVG1ig0koWNP}~$Gh!l(?^0h+fnrf^PKuSX--{h(hv`b<nom4hR<4Dg%9KcV(N}J5;
zl^#?g-T=vsy}AA;c_5J2F=Ob^VsrXmOq(RPk9%M}-jaNFxGMJ8`W-Y7WTHk43vii-
zhC=tI?iQ6&b!2_}%V@!PmmG3XLL6{Ve6yI&rO3W(ld$rsngP99+S+o!jxOrl7b9pB
zy+T)6vo`tBhT&h7WUD<AU+zV4;zt{79ti|CKh0A~<+D(^ny1?SH#BQ?vgCyV8R!oN
z_s%s(+X@1ya2dPn>1)bGWXu0yV?5>pXfkH9+h4vJKJK0Mn;uckIojJX8GT*egi=k-
z7h@@5XH$>|zRY0hqE+(KV9Q`j%T6ugUN*H;sj1O(NoFnnzCNaR=X5ZiQ{<}YM?;oJ
zMf-qFO+1ve8go-CGREb5rMFTu6Hx8~Tve?#RS8u1D>x_uRQPf-b{@t9I%zOsdbfQj
zBM2#+0Ic%zSs;6Jo4AatTaKJ%q$|@!P1Ocu(C`~Thd$PmW3i$H-2e^M>?xvaO+a9C
zVZ+(M15g%)$31s!)DxUP{u%vJXDc!NRQ-PnW}6H1Ekt#S4=l)x129$)o$1ADtgF+w
znJGdWcn(Ta?3%N03J(4<S>RtNJM{%Mhu^UKBToN<bLxJPz%Vi1C4OkMxHYN0{C+;@
zfMpaGcrfLkA`H(ue(kq=<O_}R@n-=`QF6<Rck{MJ;3rlbRjH-hQ;+1S#K9%r=eB1b
zuEljwafVQOwpIXP(Eputa~IN$2443&D4i+zIE9bRuyv@tC@-Lhb{_H&l!!)7B}4QH
zwXdTnj-2Y7VoG-uIg!NAs6F!G40`7<p}Lsv)l~Z*x>?0Z4vzt4=d_hg(%+OeoGL=m
zdJHi?Kx#qf_Wu<}1^$0Q>bHtN{~e@0+HL*?slPgEzxX&EIVS0q$f&_itjmt|m*&f^
z>Bl8fGDNtYVf#v76x%>~iY2|Ku=Xp0p^$<ONio1LpVO-YUWdkN;%_Q=a;h|>Ta3dA
zX0meBb>G?*N&)qvhG8!us+(|xvpmgx9M_xhJ82hEpHwZ?9JlBFU{i$bCz7X(rwpMn
z=sn~$0msQ(;g)h<FV9AsoN@4y+>V6{>(nAA_rk2_r^NUIqzty|xrMIEA?Vu3Rt=5g
zZQ;qj4ius?wNx<t)%8<ux}xtDaSlYTFU;xm_`GGGlYy)H@Aic*>K+^JVu?l<T{Rq6
z*|dP}Ot8&WMe5C=YE)hQH<gbDM>L5u3`OVr-~QC~eTUkl$UGBL-f?fX`S_eZ9tLi9
zN@yl}Mq~*MDi1W~eiU|5*l1c0IM7h;omXocNH#>Q|LIY@qwgk~8L)CDuO)51ib1c(
zg#-e7Ws$m2O@50y!&g?iGSz|Rkc4Szn;_gDHi&e9fd(>C&twgOb;eiR);ME6NIU9o
zbTj(9&@26Ihg{04O0ev@!@Ua~Uqy7Kj~#WuqW?W)y!)a|zR4=~cyMMIi0RE?voh>-
zkp0S_AVWilf9NhDcA<n<9yWC&54zh)3xHI3{?ns)g`iV`cTt@3J1<&O(Y2%=swF$K
zqz~bTYl-uhLi1}@Hk|{Zj1tK^sk?5y3p;|83WpA2JPS;JEvSe(U*T>c*8ppN3s05A
zJ}gB(vhRFNDT6FWav|PmzB_2Xk8vVY;EGTYN>FLhVTvGmAh?AfNW9-c-f3@v*!FK#
zT)T;^zS{@!-DlohXGff+F=aQds&h^^=!j8#U_8?zQyMI4<=rP&M_UGQc=i~(BQ9IN
zVh5ONnflPEi%Ubq+1B@&Ug70#u2PvYmBlX7EN}{97S$@sl~D+~#|^;mJx8Q_e?N)q
zEI5@5pg?YSFghnpy<EPm$aK8{oZ6C*OYa(@bd+73T-YqNXFmhJz(kHgK!iXq*O?Xn
z8I4>&Jm+rAR;A?#IoQnFL7taDHmLlY!+ms9%AhM&Bll9MmNlSn_S5*WX=P6jE{t#0
zjelnEnA2iXi_CMPx0QMr@T*Uz{hQ&JOf>52-xfX)+sw->mfn5<MBpz$+zm)uD^eG*
zNb&S$vBx0&iu>hOvA7Xhjph$}jv5E{Sv%t~+YF($xpN^zWqxbk#}WGKtX{Km40YS*
z2;WLaa2)@G4r$t&`}Qo|+jN{};a>J&7h5L4_E6`D?@XKwYlBay-P_bHzs*Jw%^u$=
z=iIi8;w}Z!0l19fpMzyg2-atGB(%@s+?3P!t(!8XUq0-`+OEDA|I@wtznzk1z~4qm
z3SM)vwnTNY*-e#aWp@QCF~>aQJPRnXRx_xT(x5w{+GrlLij-*maWSTv)FoBCVi6f)
z6mYMz=HHK>{Yo6AdT)Lc0joIJ#SC9=AL4XPfkxQpZ<aV!QTnwOF%tRmqw(J21l^yR
zA1Xd>@%$j6_#XZ)BdHyY)nrz%9tv|K;)oNh_ee>WVD=ZYfKpcIB98nZHFU=Gd9|i#
z<z%V<2ZKO-zeH5y4NPv)SbH8(Tj!~>PE10Pifp_zG4&VrkolEMTbX82XT*LZDg`Ve
zY|w~B{P%y9P;N7R8s;k+vH!E52bkhMU%#<NgjH}&NfzCXC#FSJgRLgExCOCSo0%1{
zZp?HUCgn@~q7Cril>MZ!d%3RY^QM-jFhBOau1VX`&eYDU&5}qDl6J-qjn4YOL8F*1
zon>gUvbsjx)=Mf8|4NTl+8<?ku0R_9nlM);J1~FOc9y6YiuuU;qC`PIPXunQ*D&CC
zDpEQ!-#+Dkpc7=_J}=+o{SJi|m^-MhzTg8SJ>D)y17ZyC>(c>OM2lZ@)U_@51hJQx
zVz2sfA15%yRQZ&gB8_q-WnuOb_J;y%OdDu@+mOt^vU8fW)J#d!Ob*J#1jq&DG&HTT
zs*`^-!z)kee<N8JxoQ572{W(;O=mJ0hJNuiX8%^v?-<1y=moCX70q&guM6%pM>oIM
z`M2{;Atk>tZ{?OEg7KKJtIYplFP7o^!l7mn+YMa#RgR>Hi}uV*=Z^W?rSW(|-|RFk
zZqYIAXHOB^3$u5T5+WtUkq|B!9JuC`W?g@25h!&;g{YH>{+3=AB&Lu_mb{I7rk-&|
zwK;bNet~%ev&QHd_Wc2aDX$zm<pa{VpxK3Fi;nD~#b7VEB34v*`P?0L>Tgo^E2R*R
znb=i2DJ~!O1tAe=_7YMEq!8Fbz&laf{hWm|T>)=K<U9L{$q)1w&G?J0ghkoO%n^U)
z&663dDdrnBuGhW#*^LJpQ)XBkKyxJfIY9@2)c8s|KET^;Qv)MJXlQT$jxhQn5S*sO
z2T$9(d-cL%-G;`zf3M#?DX6Xd?di6yG*OG)+WA9yRAyaZAJIQ<JAa`22h>zr^`|>_
zL44=&f<JJ5(^sHs`2DK)8&lf$w?Ka{g-@;V(jp{`)7zTUgs1|KSCBMT{z#bUdCS>Q
zQq0k&`Ydp`^U!maaF;lAr1J%XxOrOzwzu+bfgZuYD0{3&K+QX|hB%4&e6d_?Euq}G
z2#1S{mFMe<uxFcwQ(kHuaNA04D%*u$ULwsv;@5~&n;zSW_jD{~{R{?(;1_>vsXAg_
z2T`BdPP~WdE6=jIygUF!BcnunroxUA?OLhXDi@k1ctySu+8h`!Y;&C#EQVF#=ae=|
zB!EQ1SY;-Jr^CFP`?7IeM2rsmg*jjC!vKwBd1OA#Q=XQmAV$t|ITRvxmU|fd>h-ak
z*X7c6RE4Ec(8Eu%#ObFEU+#bXrua|Up~VkF*V0-ml?p{bv}VwMlEk}!EubGqDc`}Q
z82xKEh|U7Imz<Q>X0hH>03%vNiTml)Sf!Tsi|q_ks%5Ef%X7#;X`$d*yPRdEW~q6-
z>`*dMZRKha@iD#t#|hhS4^0cBb}5_7=0vvNt4h#XVEZqKeG$xpYyp4l1;u?j6Z&1_
zpVzHowVDgXONn!28Mqxt1o|Oxh`MJy!+dtLot3r06>yWc4E7M`j^8d(ZuKfZq`yR^
z%>QkwthNCLM*vTFHmFnCz?X!?V?g3lu6Cjk<0NsJ_-@_tAvIAeipidGGZT{=F;fTT
zluo8JD=dc1`%ft4b^d=P^ApB=yXSm%CheaVbhzYvk~z<;SX$$UT=|Oy5UlPgqM<0x
zy@G`&I=d>eV_ju8tB?>i6D59~@C;hb=v!D7#gN8y++**|ZWIqCC_Sr!bW_p!*PMC^
zUa{ehXkd%F;2((UQjMT7rXzEJ`l`Y2?O}v}QQpyQzz-G2d^dk1UP9$n9Xzd<4}KUd
zJc!pbk&!m!S53pFM%25yuf_A(d)l*;w(e3IuL3WYa^mDlit!;>@BZA^_k8SRgysT1
z{-*snP4T=@H?`?9_0fUt(i|K(j^Z7JE7=>rnh@FK?zZ064*}xu@d&&Lo7(`x-q;tf
zbLSx?w8+R&yl8*%_oV)db<qK0ef((IeO$x|xU-`5r+hXwUetplas5Hj?#C#g`JUKA
zAx9~LA!mmK*Co5-F-FnQ9Zi%u;DpORO1~<XJD>YVw{8Zf7)~gnPyw!eP7mO!F^1|O
zAqUEpna_<U=@gStJI*RObU++}YxcLalhM!<js5crvMzrFScRqIQCZ%o!``>XYwBX$
zlZFkL1yctG;c*A3Bzxm!uQp6Ov*O(eyfG(H%aG(Dn5-$)baHXgS0{98VJ9x>BE#P7
zoT8zo<4flRW1`X#+wPn(kIKH8eXdR>yb33Hr~^ks{k+tyMAgMLw)OgLEsf{rxyCu5
z?fw<-8~J}~Y1CUX14@y_yyKR3DrPr?N5Q3Kl9tJ~Oxz+JsX=14N2+V7uH~*#7Gmy+
zz3Yae&Q7Z}l6}E_YFcjQgYN;#Z<i>=+ilGX9MX0Qgij{y>rw;7YN=d-f277X-T6DP
zYCM3b{~(+7qHUxbR*-RpWUKtWN0MX?>>&t06M28%xRIOsD*bd&QY{l160NlN5LS_D
z`cTyLLJ&o2XQ`c~c0M@m{BHI-ZDVTrVq=~nExYGJ;<+0@;h~as8aIDl%3VX!xILBN
zi;J;2SRIJ~h6sOSI#1!B4J}IXsWS198MfF2l6f+p$wW|ER3Mygn%Qb=BH)DEgRzDx
zk7<AL(QI~f%kJ0#md97~7~L`OaMooyUYAK8JjH8dWIo%maojgSV7*3bmP@5pJtB=3
z1&weWrYkoVoG^}Xmk%yeHjJ5p#tnaqFY|*|B3r<PPWo<%w&XI$Gx<C|0&mUV8p)JZ
zX_BN#0%wxm&0ZVFxq79QpO1T!CoqtBzUqI5tmr9+zc9EEXp_(hGU<r@ZnO7q+@~o0
zSQ1^fHG@#q=;KrLI@0<hG%!V<O<ntmk-MLlick7!UYU@7+Txl>KkbP9G+B^x@P3+P
zFCg|3Jjo6<L)D#TPHr3J|7Y)7m=ia$^<U{Qn~kVTFfW7c)Q|!jdtEo<nZ=$vTa|x@
zONa(^M@XzB?D5U!zu!KsmOvl@#z@fOq*4h8fz^G!^X&70o=$w=FuMW93D1p$&uu$;
z=eluPWkOpvoS^B>h|M*Z5};*Khb6d>WDhb@T*!_ovUzqfKdWDDbqLl>X7y=$T`<dB
z1)-I$+QJdFS++`mnn+!w?!i(wi7kIRTy_4MRR?3|Yv)`-Z?zSZd-uhl{_{NOu1zZ?
z@kHX5o@IR<tM=D>SG8Ssr`~5k;zMVl(c}Bv<AbF&s`YB)wC1CWV?8DxSp2gl2e@6X
zbk04k&Fq1x`8cHR<HlbR1n$jt-(OR}h-2zF>TpfOV)-W*sCUy|{9Pqa-YtLr=qtV-
zyJ?Z;T+Hn@b*HwuGRkz?Dpjt%{M?uyp|)+=C(o*FGSPA3g?uBQ!QV;;h8n3Rq2J&~
z9NjXB;*L}orYLzMlqQz#(*B~o=Hsiw1(!gVK;SwaK;gJ)bx*lmD($AaC$sxjs$G?N
zl__o!f5<R`mSy3gdyztG#7KX7X%-UW5PCr^wn)uoqBfi=y90cCH^<-`ox($afOqJG
zgg-jMe+81fCpm7S<q!f#2!Rn*`|K}!5l}?H5h8#Ooh_P{U!#a1I6@G(^qb27aQFF_
zUy%Nprhk)~PlD1L&{@hT37$q@EX@Q=n!IVUv2^rJ;B>e7j<jw+q)UI|Y$TNUTj52c
zPCMmlMfzOo=%~INK&I@Ysp?XqnmqU0p7KBxfpi*|Ld4&FrtJT@9D%W4;!MIFQ8LhN
z$A!ZJc*A|z0yIbwM-A{2GRga@-6x94ev#io;FY)~Pl6){eL(0FTPyvhJK<DgWUFjR
z&NvR5Ish8g9!|&9bk~0zE0*iEv#LaGX_mY^dF4gH7Xe_x0^hPl2{!giMs$a0mia_Y
z9Vn@r#IeSdXsWZxyukW(t$x{#V(suJoVCRz{v&7<9e`2vrPaN=y}P+b5&l-o=<K{&
zyOt>5krDr+X}ru?20|);IPgRU4UgW9gvD9CQ7iFsuSHmdKS6)ZE{$ey8PDpA!xTLA
zEJgEarskhVG(N#aFoH*Sc^F|Y2i8d}M9Bm1GP=R^Uq%ob98+(g$7_x1d7JmwSQcs#
z+#S9hxmz0UegJtdU_ts8^W~sinBDON{LSrNFzlJ3$X_pG3S`2(<LOezrTrp#u!3KK
z^s@pNAge`%0Fi$@9OlyT#1bhW3sdeqZ#l>gnU^TMOcf4};kq`W+E@R|59$vE!e=b8
zH)E=r1!B@+kv+ha!X<IS#3|vIUF6c5!JZl@iyV}~O|b?`+s7fu_DCO^9EPqXViSC>
zIr}Q#ANElYo=|P@UpRgGr1z3F7%o6hUX7LKBn*?5^H_h8+*4tGd{f7hZ6?=;0+#+o
zjM*<>U$%fQ*r(l_&mTH>53hGxMKG#X>y&2~(UV<(gxDWxAd$Toe^kTLEYj#gVj2bQ
z^$9bD!D0rh1PnKIp&thdCKOCr5loN)i-Jim!Biw2omY{Gyk!tn%)1gUl<zX}qu`u4
z=nd}i&xU`)cZqkwU9-btu;%9ujDzo?KtzF<NrC9)gQG}1Y$_SnP}fwWNLsv?he+br
zy|ftrtjiZxKclchVU<Z?<tMeExROL%ed2L{H8`1f)EDk)LUsQvjfY7vufiWC7fP;p
zj*wY$LA?!1Ey)66YQN8{Dmc+$R+e{E6^JYa4vK%n*@#{)u|vRz4AL)k$2;^al3SDg
zxc;QVIUoC;Z^Jne`I#vBQ1WFa-QRpd!FS-wiHSF8$rbcrVqy&BA0wS^TPEbq73Z?A
zU1qQSd#c-;WLM&mcJ`074b^eG)MVO>_CGN_*tV(qm46WqyStr-&)vHNKnhl~?Hwsu
zH0*zw?DXwF%Z{i~A&chVhUVT?yIO5WmZgM0;m{YC;-HUZDd~a&q~1S2-*!7!<WuYN
zzgst*4N`p;2<|cXb+vTaYV8IEn7^Dh3QWt~AeADL@bQ`t*<T{|zy5r8`EdL3?w(wC
zKYk#WpC3ApzuZ1N{L*>sKKdUX07TF@+CP78vS%F4$_2KakAkjZwlL1%-1%4Q!>3<1
zv4|EBf!^N&@vq9&`ZWeO_blL6R9=S3Rd1B5r&kj3YEbs|tQ(s3tq6s3OSL)+YphpF
z9k2JB9Bb5+a-m$P!aOhbic|`u#$HcU2ycplwQNx}nhEo{0$J=1JJ&)aqFJa^e7b-B
z_%$131oGY9k=7RT=uo@LgXc{ZX&B<y=jGFyObtHbs*q6~WrxV?)sdgle`E?N8-#p=
zLJIjf{zcSMsHIR#p_WQ}#q$bDUahlHm?9gB2h9aE7tmZ-f(eF<F=U)Nxgt#>j5r>9
z#1SR-+y<k>K3G;hl0!au3FJ)2N_l^&r=;9;iS=qwXQIwToryXVb>@MQ%U7hU*IFw#
zTGf_3Yi;_?vXmdxkvjSzzhoz5k(Q`tQO%;76%U7M_F!2V)hw!6Io5YH$ETSIG@!+U
z7SBqHr*l@mYI`#=(c(di=Kw7p*~rzDcoI(BF+!Lk(%CJ;;l!}q%Eg8E!nc1u;eW+}
zt&XYkFZQ;4Q7*+Td_h!7EH{=-fQU7Q=`GvBn_lZbr{!F$nrqoG!z_C>V9k7ijMd2m
zntZ({0mN16DBswHf93P!zh;+*)EYmIbjP)Z4Dbu3+)~M?HNq(y+C832B?T)wjt9z3
z1!*Y~<34THUpyOPA|iw)5A=W5tK_W8a(QU9KInpsw0{|Xdl&SmBIS(PM@>BVd(#%e
zhKYyFBIB;BMC_bZ8kcp6;O<bj=up~M%6ak=v3gEfjPst0Ry^oq$|RPi*rsKVm2!!>
zg27AX^;OvW8{3*r%Iq$iP83ba-Q)@+_m3R=gsR#I!X2g9Y)ja<^G1KQ&K_EFc&v_g
zh&3)E<`|+K((y&s#?3S4#xVF{hrxf~1R6f}tZxuSTTP#ACuYb)JJ6WNKq;BM$`^<P
zf)T@P5>0g}1h8|S!II1WkS#fg@m)z)fbm@^I0xgqvOo{xyN6FBjPFW7B#iGS9pBAj
zS9$4x@Qz${+MjQ3Zts6?4geAu^4~usvX}X~S%Rf>tcz_DGpmqgVy~Dom^HDMY)HkQ
z)<u-gOXW&OA{Z{Vj?Ne=$xr_Da8HP^_eU1l;4?JIGH&I8=n;9(FcUR<97xTln6+h+
zjPJ<6vf<cSb9KukecGE2!Mh19{$vAp3q7k$0z_iZDtbY_yXAj1mqSAk4M8*n56`wh
zLl6zYJ#*lZ8G>jDzIAV5Z^T1N4hGDi#|-)v@W2fEtnft!GwA(9?bVd3HtOYTna-TT
zdFMPQ8{`m(vI1F|Jex5R|9~DldhCZNc+g`<k6jjGp~sFMJEqD-Z#Zn2cu2_+vd)dq
zM%CMq_KlXDYUF>XW5DLZxCx}8jH}ET)fWAfY|*f}(B3?AEJwgi0g=KRf*+sN8(QQE
z8n@XuZkZjsSx~p!sg}-V3hIhhywb|3=9>1|IW@Mb1y92=e;OtoJ055EY&>Sk;4zE5
zDrK{tE3(lA&$}`uH}JemzVj|jaD3|y!`_I8ln3SLJfeT|c(}3&oyV-OBRY@hJcd_^
zIeLdhFZ5Z^XE`39<*IUeR=!-Y9qN_SYUe7rIOoWOZ|9DTxWs=svyM#U{>Z}*fp@cR
zK5K#t^A@Cw<?U!;bUdc2>0Dfh7mC&UU#h{uH@i@@hXvwHs0OyA#@1o0j$3?E?bzZg
zNy!W6`@Vlx7J*T?6sN}KM7+on;6h|RN8pf?MzQ7Q0YL8E3>zkM#kYh-xqW(mRhJGH
zp>0hP7Gi(oSVgRr;v=O9ND&pa>DZPDEU`IZN`#iyG^>=2a81=I8oJ{uW7XA0MNKsf
zg?^i;X1~~vpI-wjZB?q}>r;tV3dX>7#;6C{JfDC6_1E(6zst*51(H+p3x%kgs1h_L
z5+ZUkb^~SGzYE);c2@74RS*SoqQIvAp7KV>0Ti?m)xM^F&YBIYQH}=&y6S_Jru)79
z;cwcD4{uO&=#G@Bs~gV6#ee~6jTzXus<p3@a_f><VM@XLBpw}FUvL?cHIfv_*y=;i
zp`3p&E7R)Mk;HOVHa<FQ)K2SVd59Aim48o4DHYDYeNho-)PJtm!Dr=0s~&ko5dP%A
za-DXyP3h>~ot=h_Cl>SRT@iZ`5sa02Gjk+6=*2~IV5i&t*d@1jgbB}3^ojYH7o^+y
z=jYpQ=W1nAP4XL%l>S@J6Om|Ln;v`yQ_X*F@5zVT`}^Cw8=&{YKRTpKI(Pqb+x>X=
zp>y|;q7>fMVa}T(Acrq~rSpJx6?b&4iQ<di-^}04_Y42PH72hDY0YF+>}@bqNHvV2
zf0Vb9<vXqNS*;VGMdSy17ZcgH6cQiafsv(~gzsvexEASAqV^05FV%jZIP{6yszHCm
zd8oracb_RW+4o+}Um?#Sg6S~qaOfuOUt7`~gN*6WJh@Qw4g-ziaHvwLdz3^-gmq15
zf26_5rcJRVUyx}LlQK47>T&8&BCazmT6W*zO7qFm`=mD=It8XHH5KmCU7|b0u$Zk;
zASy#{oS7kPip$f^X-Rg{6Z%beTt|P&d6t-;=3HRr#)#TF<V!`{T+*z&SzuUT<`GAN
zJEFwVtjX3nJe>-j#;%fw-u}n}4ug3pU8XoG)0v-`;oP>V`jvkXG^%P3r(<fmZ&jja
z)zfzM96=KY8nHjJfM(0il9I88+guHQ|34C|zGvMXfu$T7N!*;hYn5{Sa#nv;Af>CU
zS%w98<y%-PwM)#h*`5I5cx^1-i;a_fd9nCn{8Qw}5+@Nn)qJ&#NOVU+jJ4YxiSS2}
znUI-w$V^bXf!u`Lgxs_XZsLqY`$bSua8m3XB~vX`ef6dQ?F)z$^hxi9*pERe*rRJ%
zhJ$Q~Y`7I0HmP3=M85$WIt+ht%A$d&&8~=((Z5Y>t7p)0p1_3#4X7!zOaA?E1#XUG
za7|hIM0_=l>tQ#}&rUm+Njd1LlHqS&$t+*_9_3uBYtCo4$|SUp$8p@dOkd8vg1=|c
zj74<IK>&E>xz<-qkQu52MQk~3q}r<HQae{5UTQkKTILyVUa95#aPof}3^k?fk0_KP
z{t^H3;U9V&6o@qeMDv~wikO&Em~|s6%vvDiku8udj({!vz8-U`!LWw9rW&XikOPi^
z1H4W*h=0WY8obC{ej0OR`yJ7pqB%7T9{Z4{GH{tTo0OgHlJ~Ii0d>y~bPsSHI^~G}
zwRm}b4FZx^BJyV1w4i_H({WiA;>4p`uQa%oAeFfhIx|LdBU9!E<HR{@gT2#j%Y>X%
zuhe~=jj_1?NW9B|O3vBj^?5+3RV70GwuPH#SEWWnzQWDW7ABePGY5vqjgx$NF$P&S
zmb$F#RP$w)D3gcM`1PZ6vP0;t&AVL+iG7)OC?wv5o%yaD`{sX_gx4DF<gIE&awGyI
zdSf^!=JE-D4Fv4(%mB9xL!CIhE1%bnCF9^(fs><AGn@9@Nnzosset(4(6$8_Xqg|q
z(%ssr!>aMs2+3&Gxa13hHbkp+1s1{|7|C{32w!iP&(7s5bBZ@F8l`Hfl}MSB<SPMq
zydw^K0F5fS%Zq=L&kRb=$&E^llPgQ(+~S<Xi2)#gt5&P-!C*{aA8vH_@rkV-(zG94
z>v8JPl`F4xPFts`=&htzzEB+bII!$TD4Bfv@u^&FoGevO^2^2UYV9C|i2U0V6M+bh
zrzWcHcrG{du~zTNWrIClDjG9s$T+V3*i#){dt^%0Q#XGAn4^Y7KEDqea4$<;J03jS
zls<00pK(0cn~&MU{H9Y`vP0G*lHH4C)R{nRj6ZnPLV7O|nWVw1YOemIy3ha>dfe(J
zCD7s4Z;2W-K~Z=pz_fzim+WTdkQ3G6y<vFod9Me>>;}hG-KoQWPanX~y^xXmL`e@C
z(2I5x>@R;YH(&=bhU?SHRSkQT!-k26l;eTDhyer#chyz3RIOi@C66{8FA~P8&V+;2
z-?kAmqyF{P8B#y)bVTe=GOGx!Wk-;h^asj7Hz3J?Yns(INTSP2FYx}3I3N-yhQ%IT
zH=!}T*MkjqVh?7>j%Zl`3xW}pFN0_{WtK<()r^0RsmlqGJOTEN#`iNLiS=AGd!jO7
zk35}($YRcKQJdOZ*s^ob?p)TcE5W=fg!bW)SvQ|_NADR_0fIZ!RRUaZ(OS*J`gyrt
zt42-~Me_xgT&YzquVkvtc}HND1QCsyh^CsR<-(XYua~G8G2=`py!+!9?>0xtlfFeA
z0{4H8t-d~Z;XmV#VIf?UGFOS&%HXE=RQ4M*D1wsKK-enmblIu2E+v|`(^Q1vLsXX;
zLiQXw7&%p9(r40T_8;#By2jj<jvWdF!dU^zvxd}n5C?;|F6T?1t|Q%XE&GLBtbwwS
z{*v#))HE#ZYn$mJwss{qFSv3AE~}{dr0jp^bTi3yuRFw)jk-;hiJ+mmL_u&w3s8cq
z+srAS(eu?N&+#~NDd&6b`hkJ2*3_L${o&~Vfmf4TcypC{rPP)V18?WASPVq&k37Ke
z1_bgJqyinT{i665?FTV=IJgj)xXVcoP~4xNNfv#e%y-_%|H*D<)@^ZMt7EF<|Du0}
zL;qx-{FTqMZ#lXF7O_|Kxh_hW<R0%DDXXlFAu}|*ofqw1A;LDZCui5UolsC`8`Y7#
zP9;fh75gI(w=$-JI%(cJLz!dUJo#QXzlUbRL*71kIE3JHj<MFZb;Y}cj#ORhvI?DT
z1-%EQ%Eg8E;;}c?jeZDw%kulz^6!5ru@H8EfNksK5D3`<LbMUY^jUaPwciVWaZa`0
z4<!HRx-_;V_ay=2u=#FrtzwEfM4%>??NUk%wnv(35Xi-HM^lG?)~C!0u*LNAEt#xQ
zYqZ-tO16=zH(ZA8!$tOx=HX)-II7Hr^eo;zvVIcOniF{$qZ*gjtEnYqk34_W$}xNS
zM^8UVOReUb(28^{wBX;Si?TFhMd?5;wH5U#vuAi#I!}xP%jUjDyfJ=Gk$g1zJdJeS
zkz1`*J3VhnPu}l{GE;J)unrTWb8;t7{;rNEKM{xl4Q)CxlyAIyd+>z<nW(NyZ4;V0
z{n3F-|M`P|@}ECC-R{ROgoJ-G1|`y)x<pLW^dJ?5q_wBF*^m^>093q|{b3yaedzu~
zN&-ILy%9j<KP2*xFtC>LB*I0LW>cFXERCo+zc)m8h)N9Ru5!q!{XPYMj9u+Be~m}(
zE&J=mQ7EMUgzWMqku{qjv31vx-Nk<@l{zme!1&LBI6H4YhQ`y0jG=!6!^8*#4+uOj
z0L22L0X}xzE;akqru|RqkhX2Ae&t_0V0D60Wbe2+I(32vB5^=q{xD43=1z$hLkWD~
z*~rPcRE9O-MA=RkHw!1Cf?d0_8x}YnIDbS`!3`8|@bK(_egDvq%VEzdH0I>uVm_B_
zi`!H#Zc@)I8y%yUI&^=gJs$hb7F|~zsY}j|nd8A`#90H;#c2b);O7nGHdi2UQo#Vs
z`wOvqZ;vgi{C3MQt4b<V*Cb{j<vieg=-l78ZaSpf`Sh{-(7A%F!U05=g1VGNev)U|
zP;zq&YARnKIjCnKQR9h_Tzgy+BgzFIuoX5F_nBX`Ybm+Hf8Kw;`a$>k5V+t_PAkwD
zIXGkFVBBXEizpV;7K;Y}glzOvlN|`4AE6aMa0s(ULHPCsZWu=l9W+H9Gl`F+-iu{o
zPi)<Ei>7r*M+vA84gndFQJTAt1Oqh~s6lciyQBjMm7^_o2tWwiV&5I3EvCe!{h&dH
z2H7#mf;fY0W_o`G4%@7AWbR9=dv|+x1269%T6b5i?iEq)KHdpCj$I+}PTZCUQy>>u
zckRw)>+^kwe7r`p56wO#7n*%2>QgoQ;-son>pP~5uWgz)V#(9Ns-0!%4yLBxn7gQ{
z6L)G;VUqci8x{I(Vz0@mxbD>2^_oN+n=Y#5IvP(I)w_Rw;8;WP9U0ixc#VVIZkIZ>
z>x7V;z*3uiI<^$9dKmw@+KAyRbKO<kmkI6rqiVWDvrNY_Xo0;LLz8Mw)}LMb<q|dk
zIr|7X+o6=GhGVVa?yGXCUXiHtNps}7lZ%s+vHE}N7@m-t8Z9!|pM0@wm~i_=cSl#M
ztDgKkd4GRiuDlxiA@z+65#w+rYTl&CB{pUD9ZZOlRzWm_u0F(ZZ4myJy<OEkcJoPy
z>brj66?g{EWTS$)^Gdx@OKQSf0t;?F>xO21E4tKii+vBNil&qcg#BC1Gs_2FR|=$-
zhYBVgdN4=LEy=tJNRT@s_eUNCeR+FJ62wHtF!6s}Wc%MYs4p~w5YO|vrW3Lm&aq7=
zf`N+OMr_F$wae;tt-c%1IB)RV(O}-cwDh>*d65;QS;kh=Zv~d+K-|g@0F0)vc#KHx
zSx*$kz!ivIq{RZisI$lJQ(_JHyL5=0gbNfAvOfk22ij^&nJs((|02q4*=4qFK3QLB
zm$`pPs_iPd@pLdSsF)eH(4Apvv57i_Y1%U*YEu%}b&5PAk(cS|5imSwGOL7B4P$=3
zw^JndloAo`wXLzLn*&=lHA{DjJ;NFn2R5bbKPOBK(Qn1cX!7&Ka@>;&+mVyfDf|D)
zmF}oLgZ}7DsHP8e?MMH&rA0*CX^=%cs1SdDslhf97~;TFPKm{~*WPw$DSEyc(-p6;
z21T07PV-QLF^o(Ir=4vs4a((Gr_$ODGR~@}X(0nG$5-hUcKejB?MByX*DkNm(^#u!
z85aC;zKrX}dEE;*F|{_Pm+;y9oT_QmbnE_3?1XEUrtz-c5BXxCx;wgL<`?6>e|UdJ
zv!nQ_BRthD9dju)SQo^2_xE!NefdA>U-?213w=LxAYhL2YJv-4n9Qyu;a0pO?ar50
zw{tODr8h;#-w*O^>#j>pKDdb~4Wh73F!ZkwG4jmF;8w>Q?sWb6j;Qu<3e5mfr=U*Z
zawTA<FWF+C*wDkowlt<BXrJ_6=0krFTXF>YTp8mC1Oftq1Hnr>lXogKbg18~a!FDj
zSzgNgO{m>`{!d7B@p>VuPBBOmM1mbA!7@gclu9m=c=W$sucxuXPj<X++a2+;RKIRq
zArRt#AofQdAb6$AAe3i0fs5HCYjkMlL!y*@q)b%Xp$c?qU@m@Nb=7<U5{!Rzh^)=o
zhwKuwMcB>kDT)JI9aAOmStaa~1yL(MdbV}(Bo!m6MZXdSLNY;CkQqg5!lR@`-PDCR
z5m%HvH~iAElF|!;t!vjM8y%yU>C=#FiT9+OXX{?l8&g*|oQn&gGK!v`qWsD46SpiK
zE$vXS=Km`2GGUm_GxmNz2hV@G@a$5PYq>qgBGY9Hjm<oHN8FL_5Swb&6Z>;Po{wp{
zM3C7_`hg4lpvxh>sZ076^=^8$?5}W+0|{3gRTdoDY?~K!EKTUvDyYM1WV3A03T&*F
zO<RIM*t@#`!QSH2d$6Qqv6OlLVWkfAFNnnD-73W+n^E^_!Lco5rR#rC#xSFrFN$WU
zj^nkR&6DqS^LwHxxgl??IUG(5%gyDt>G-vK(4?5sfS}n#Zh7#iTwHiB9(z;W=nEvF
zkU_lgpz>QLk?qo|+L!b|t7^|yRU$?diZ>K*T)gcc2$>9a%LjyrP-6%i=`Jz7a5r=i
z^8#@}?2X6R=Bw{!fw+Hd4S`y<UENJ)bVk@k9?qDbvgZ$N5OYlyr=Gya+xwc0>)P3_
ziutjHXX+p^s5x{;^GG0nAG$x0(sCanIS8>zX1W$ppDew<?(pfn(x}VSn{<|ASxPsB
zEF;k&IUYC`*^G^uP~W57lZ0%7AgQBHCdSL0a4PVEFQ(FlHl2SMIq10PGT6g;a!$#^
zhnEkxAMY;ymRoQZWg>F(5Oi_1R^pBl>f%(0Zm5pirDmVnwEu~j^KF}|U-=j369ZSX
zxS+l?IFC`&(OqPvSyht{P+OCzDI2MpQ&a!<ba_?sEMU&S%@}E5_<~`m6NiUE<fYJW
z=5p*NnWs<F<<Nfz<hu3CeWyUiR5cyqj#QWTn}UYYw9kX#UJ*QRp`;o?g?o|Q4tOK!
zrLWF?$5qX~YWL^A1hYBVjhgDv0$DT2E>W;2;pwo|i)J#`fGlJY<zY|UXG`>>cgTc2
zj$aJ!hsmTxSn~p~<j^N-GmtDUi~PvFcD+wcSJwbVy6u0sVkZjZd8BJ2*dCjXttW4S
ziwUPO!S{PFM96sbAUCB7?2Tl0<{OWO(ZP_H@aRTJMJ~^c$STpYYqg@n%$mYP@@AR5
z`al^`wa@;}doj0XTjS`obb498l+HLx5C!4=IlDfsoJnt3i2QAS0HJ(~C}fid`ZMF=
z*I&!O|Gt0P!jO%6&CubHyhQrL(Z`dZn?6}nSSTHpwkvaZlWaY&^W<hZWNWz6R7MRO
zEk+k)k7VxJQvGlAB8Sejmn|I;5~_a$LMc}upby#dp{5r;yG_|GlFMR}j;ogo-nY%c
zMk^VuWSJ#k#_X@GSjlJ~qkW9_G1|v_)(&VNqkVtu0VCaG1G!GqGeF+ZNGpC;TT*gg
zTHU+byBqSMbAR8u>5y*c)5q>Z=ZdkvK7hQ7pwbyiq4NcjgMJ?*$6mn&_WI&CGY4-r
zOjbNQuhy;)w^C(8H9+3InpeE#cgm-i?H$#gNp<cP%Ru4%Y1AvVYC9o#Bri0V)+6XG
z&w78}l54d?aHDqEr*_yhhv&$Jxe!*nz<VZzxpcyB0}HrQZIoNryP>&MomsNr;DB<A
zlW23qm1jk}j<d~eKz4<cJml2}3Li*bG_5z}dcB3@7=+oU9p-!KcOS_c9rX=Dck!C7
zT`Qf{g7$hSC=L!a(RS@$YZ@#FvL>_21gwAe<~zdVph;<;a4+>QGNLvmJ*ufwC#bT6
zDnrAjYX616R5zl!US-)^7$E$~CJcb4UAkg3-PpBU<EJ+D);gurn$)2;c^)0)7qk%#
z8lK6rlnu{<7y7qFbZ--%$V<Y+4fCT27|BEL{Tb$wu}AE5iQ4Qy9z9wGa1|%5S_^-5
zBlR2g3Y|s2>5l6txzMyAV0V-GBRb^D@ct?J7|pN*Rl&;1?9HwjE|%WVIVQXTqH`tm
zf0<aeODQqf)<LTe==JN4rqCdUl0MZ8)n-oJvt{?Ul!n;SRTjk>$!VdMu7aVm+|?D@
z;#gzqj(Dma^tf@Sw#ita_cGTaXd-`l1nsjkTLQb%v`pg#+WGM5eAIlBmcGqAdDh(#
zQHcj&GPZ0Ay{t@9ezhRO5`4L2){aW>kG4NiR$apTht}Oyt9wP1yN`FI)9rrj5_W~a
zJK-zlca(1FYs%I70W&T!<1)lx;r+pk%L9l7W5(r?YmbmXM|qCn@@RC2k&u7I{L0d4
zrCN#{<rw}1^D7UKyxXa=)qE|;9!Fz521}R5X{02sCOM`g?$c12JNZn~)7{`$G+t9g
z<SU{SIV4D7z9Qx;9=WNsW0i^din7sMD!67nU-1nY9$u;gW;0?bB--?Ex9RJRaxD^4
z5jOo%=sam4g*N>zZ2IbXwb6g*L@_}46V3qQQXB)6t2JbRLpYdB9O$XBb1XM6i)du<
zO^haQlS<Y4^%XaO-tOw9$n|)P3juNSNq6*~L6tszDq4eq^33RAJr~X1@<i`QvP;7o
zvH7fv%anXYl<`xxDA#mgO%uankH{Q7i>RL58t6ti+hB8<b=j(v%2$6!D6=;0ik@Ve
z#@hwia!#w4gsylRE~UA~39RFlhD!*{azM_rQu@qFv0MU1iKarCfN!EeO^x2CeBLtS
zbesAYqq2)O;rF+&HgIxEoc(qz;km*F)J<p~c1W<uRhKKZPCK$*HvCDpR()2kUVFNA
zan<2ZaMe<-x@TEm$Ett*wZkQ@-e>FaekHUW_0DOj#`7EIG~QLG)V_=~-@>=A-lJDN
ztwS~UQ=oV^e6oA?CAeBSFP&d`MY5COJ#6JO7>?&9p8Y&xuj^1p<E?)9L508LS`&7?
zR`x`iJ{*CqDc<%h*R{s8_rekt@4Vuk_qj%id*uVma_8O-r_eMoKiF)|GJuT2=74+n
z<-Voe*g98*x=BCjuEw6STIXLn@pps2>{<O6{(^mF3hnlr|1SWS0fGk=x7gbS0pJP%
z`?}o=wEzHVdbi`d2Iv6+iMPzX1`Yy$y&i(!@oqrde1f^-eVz17kM<1lesz0RdB0)4
z`QTyw_@MpUCb1@a9*EE=#2)N9>f;S_$HxUS(Wl(tds4n&OjAqnJ*j(Wb($X9mL^Dl
zpLVimyhwu?lEV$7kmG_D<hWrJ^6jAyil!ovFkxPoNb9F02dyq0<h3vN1^rfkMuvG^
zWhx+nSygMIx7`)`psq@9TQEMsIoPv;i^ZO!!KG|}PjC%1l;HZZ`z^SQzq7YJra=P1
zURlA@ej0p*(GI8MfV5!G#Nhj|XK&D8!`swx_Q-`08$FX@F9|^t4oCxyD};GNtD~>b
zJ5jv=BsG{jDQ@_l)E%@s%?h)BDy`2($e)w!V3!G*_5~4;JL>8PZOb*=+Ta-;d(Da5
z)asp~cS>;3qovS{F9W1<C^JF}s)L<2)B^fJs4c?mt+b<udM#T&2z3N|Hw$&b-Wc__
z{kG^5=$27-iQPNVt=n!vl=6b)wnriQblZ|b1hl89?aZF>@oZW^?oDlf+X-3_e1v`r
z5oDqWtgIMHqGzPu5`nc4Lk~o*27M5&o5;Okw4=*JGVH;Jz?a5+5or}hA+89&lU#-I
zku?RqlQQTB63P?4G+khQ90K2y_JHrnp1vnXnjv|BHzMW<`avwdz%wP$@05H&zm+S3
zbScAw3@PJ+FD+}*+qR>B@7WVpN$oUu>J2-3k|shVUiPfgbc3~!++FNGNM0r?5wL?r
zV3j1Fz<N(UFS|#RFW}=!zJZrOq}ij8{Eg3y8+ymjWXm0q%U)S2gaQ2~*$aIgBJ9p7
zq)Klq;7d!P!pcfvfp0^jbHVRK^1CE$TDC+=F@RJ~5{fW;<JIYZnH}JJQt2=XsW<r2
zG(f_Nfc25mq&~<4R&dI5+cG5O4H|JO1$+Ui)B-Ii7wikE+_1}}GQa{$Wd(R!4KzMA
zT(-PUHG!W(gyumd9Z4%;Lci4wZOcvy9FW!RR%f5VqnV{9+TWA4wxI<wr)=q;Lkjdv
z?$ClTz$YXEJ8ces3$(!;U4yp8T;WS&q3?-h+Os#uRe)&AAQC_)Qlqy3Cri=J6b
zV3*072=F`Ecjz5yLiW5CS<r9Ngs25;zziu4R&Xv9$H=9Cb(jnJ<8xV{1?1|$;?I@5
z1R}7{$=rq&T+Z|X5}ZAzA`s;%E`oh3D>%?I8T>9H+m_0I#WmOgi@U&sUfckCyLj??
z6dys4ml6${kjMpJ>MLk9A~1W4Z;M(mL5nWIgCs8@R(p<$fK)Ew0xw$$W5Sn84}sp1
z<g@2>i3P0P5*tJwN+hpgi6g-89Az0us?Z0C;us~7EmP72pF_zd!!x;oS4vNS23ay$
zMn!+YGo^xmuSNvqbtw(Br&2mt{H5Fh=1%nq+Eyv@Q0knp7V3g%l1Qd!Htc})c<q2h
zh(Jkm0o}j42J5FvN3vx!sOXUG_+UbBY?_$jKtk}WN(A0D0w9@cgZ5$w4d5@^vi
z)C|2726%!+ZfXJ9Sv6MBqH7%B3#f5A_@21Ht6OD%%s{_MhS)Pv+cw01YPtioAXDt4
zW)IR*GAzSr=LY(<2>66*zCd>o0iRIKGePg<4H3dx$V)BKU=&K+^h^W0q-y%Ym(~E8
zFtrwlMAlj%iY+o=<n##4>w1A#R|NE(W6lv-9%fbD!)|qR4x&6wx|qE$Xik9#y}7k)
zdC+`+f@nqa10rh8uiO4o5%7Ms;6OiUp+iJk1fqQ{7Lb-A%kJwYRm`68EnZ*)wm2al
zNo3g4vL(u0X{i_Jw{$~{rDX?xmzEQFi`p>)fne{*K&IbmG^=t0i?U_P25NbMm%2$J
zAptGA71|qHiQqwRQWRkxlti<)RItrj!8KQZZt#z`G9W77$^=cQX?S3c)*EsoTAk3g
zeuB6_BJ9E4jmwrH-ECNn-6zP>>7HU!-50PMML?2wUqQR=zC*TS_lv$KknHUtLZrBd
z1Xgtq1+3j3D#ToSpf%USykLC7hP_?{bl4spcz#5{LhkVbdEFyr5cL$mv(Qt+XFBMA
z2R&Vo5!KTK&ri<|yadt(VV@HL>C$t7U9(GC0x6Zov@Fmdd-Y&R^cp+Nc&X7aa;PKq
zdV#O=Ix#{b&2F{yRIns2x28SXmv@jkA_BQZmk;1qzI*}gR;0oAc#6u4fS2v^8+5)a
zgkjIBOBNXPTd0sLB?4>i3Jc%1z|s?cfmMG9^k~>H6R#M-4!&Xqzl#X0#w%W6n_Ve_
zJtuNQt5b%I!YdW<MqH^u#*J)uuotczVQpVIgME7C0&%k|Q`Gm$9rB2;e8Y)@t2j6f
zag}ky3@IC8;#UnIU9K9z3cYHh8B!a~#M)`iRq`KPbwONH1oED4M?3oMGDxw1+f|q$
zw<Bx!b~kuiZyzA0cl!kP!tERE&Ua`a#qMxJe*7(!51xsXU3<r{>6r-e63tN1gl_2+
zFh21H8{?J*27QnyDsrdX?HPZ|YO`1Hof=qrBHNZ#cOqR`1Y%%!rrhv52k`TWKur7)
zXm!qm-c~@u-nkVRx!l2azVifst>(@fVtRK8kX3n?O8r(k_=H4mTO!@nfxOl#g<R^p
z-Vk}XS-R8>*)zihXxmWJ>^=~IH8)(D_S_lLQEX{B+yZ$pyaP!-Jn6l|57-+;Aa`<v
z2)4`!39NUKY0vl(GT1*u78tZTP+Vq&2C=CTChUMC`T=c=2`tJHE9lpMBX*c?Ly`|h
zJ1*F1NAh5Ai-6@iQh<!8ks9P|57}lg<I@1Y^2n6iI&!9d%N2G_5t`Q`)Pi1Y88Y(h
z^vnWXY~&5QnFvHuM)4329wk7=fb<}U5ROtniyq|y%V?AvVpF3G(2+*j5R(`+lMJck
z#ThaRU^UhYB6p+SaOPouXv3gq$V)KKSwSEb@`$HwBYN8nF|euCV6VpMHk=)sK7f`c
z0-5F_AT6hFFs35emc7$=@LNtlAn!>8Y})A;cnw9sYd8aGsxu^zAv2IgIRizcXQUj!
z8A<NUc)|G}5zsoPBsu-g1RZI{1^VbrF2X3J1bH<xWlwLTy?&;D3OOYc_l21{c(!Mn
z5KWpnfz~Mk*;+GG^kwD=xu7BtNtt<p$7~i4a&(pkcH=B`23KT2Uzhu)77WO<n`MHp
zW|jr{A+tJ&e9s!erk(YM9J49e3o@iGl5cgvt~u+5EX3tHjL$m01t4@zLIk`KOL7^A
z58&@zzCxDH@;69-isdI*yDJ3n$gGe$ye%~9ov<JpyTXQ(Xp+e=M^~gXP%GYG9j>^j
zM-$~EtrRb4frve;R-%aKN*%mkOFAr!T)M%YTWNwGztY0r*|0OM93e8a@&Nz!$_uJ$
ztm0_K7Y7#DDjs&gRU&vaSE-=Yh=2@PWrHVcRR>aR)kL#@susvWU$sI^eANqbzgE3L
zKiF<VykvU^^LqOX7NrQ(DcQbX_V;Xm!>+mg2HCp<dC+&DY^@y{%&HwOh?eiL;LL*v
zocj;~AIgqoxr)&5!~l`I9gBxn$6Bcc1ET+1O94hswlfimeD9=12V^*Vor*y8Uj!_c
zohHr1v|wj{lB5F7ZfB~|u`|xZa?_Vq2f1In1~_B0YlG~WU9V|N@?Dp0muUSuTkrMf
zB)=&~{W;ZT_NYj+wSYtHoH?(*?Oo=S(y={EBE?>f{#2YYXQDqfqRe?M5^Y^f<gzn+
z4zY7VMTUJ+;}Bcp@-BMYdVeO+*RtBo87~6SK9O#JdnZM1`_zR9Xpr8eWM?yoG+V!Z
zCiFIlJc!(O>`-Lba?~H^F|<u<8@@CLd*2r6w&(PeZnCY=L+tYl{v@u`+lsx<i9ifl
zq}ka(huAViWY~E^B8z5x*><$z5L<USq?_{EyP9lyEppqs<so)1n?LoqVD8ju=Z1R?
zs_mVBJf*eln2|`gt!fc!o0pYgkEw_q#rEefyRW@(Fso#t+gh5)Wk=Y(55@NPih%b^
zWY`vq$h3DR@1uh?=VP|>o<xf6Q$57a5*4W+cSJz)iP*6=k<0c{iwyfzrpUJI33;C!
z?2SGJJCZLFZQrd(wso;n<jvN5z0aQdItQA6RkF?COLMlPaQ+-3fxgbmuBGF(q1!&y
zQ{Bb(*@*=EM4||34!&kPawh^O$VH%bpbrkt!ii)%ULw-$XrIVsN4`a-T~*@{I|IT8
z4>^q@-nMc?qMc<R0+mTcx}8fPGVJKsA)q~l;B1>#B-xb&L^7<#5a86TNQGGy0_3iL
zhzvXS?n6v=-8_+O$2CQYo#`i1?enK1H^{^gryXw;S=j1r+a<@#NBx$feTH2m*><By
z2F)%|jT@0ct3>u3{m!yas~iGSETp}lZ5d8!i$psoOQhJI1(9m|uSFWItUT~-+xBE>
zLnlijlww=_A~@g5wEajT%dXMkLrq|Rv53%Gs2y~%C>740z0oB*GeM++CKR<l6UnxB
zU*xhg^hK5(vGJz}pZ%pVK)hb0!}!Fop|4|vY~e$wN23Qh8e_04N{A%e!z7Yn{lo~T
z6htmqSusMzYLRVcOnb>A+vhEkY`?$A4XY%kY1hP*qUvoQmq@UELLvoZV$98d&cP7r
zcBZ+=MWc`h>}E%ahPhKB^>u;jDyQJCojr31=;TtoK^{o)hSiuH`8JX(b`HiNwm<$5
zJ7YlPf*nxA_B4t>b#ResXI}Z_X%s~4Gv6YRzaw(nXZsF;9WY59wC7F=a1Kl)!|Y9`
zU?-3Ye!di%U9C*yvNQcepti7o3?qos9%5&Kh(JvI5YUyS)YvBlMc}-QNP;~pr3`v+
zO4ZKw5V^r4lcaRg+n{Ql%tjBsjg%phFVby4xlaYE=ZbjHYEp?E#<bAhm<pUAKRM&z
z!ARASq^KUWTM;||K&0Eb;UWXP1gSzrWG~ec?78g7n<rw&7yLQwWw$zipG*{+c&Jb-
z((H=-(vTeZgtP&EzSidF#6Ha`5<uh1ln*OHo`oEmolh<@Ko`qWte^!V@H=sX&L;xi
zubd$6EJ87fjM62i3|{vv8Dg)6oEq%3IcK{Do=CN`Gejs>AOfC+oF~|gIWIex?GV^Y
za)uiQ4gm{Usu=iTbAgI~?;_F8Cl^T|uk$JA10=&<a*iUv+p4p(GekUnPbIHZu7O6o
zUUsz$e?+`J-->i3drXU);D;?I=Y|OQh>Q0i#YF7vS`j$kC^GEYo4$kynq84rWACRW
zTws?J$sKxzY#Ax*p!*j(XC#t9l9!n6Ofr#TpVt(*zz<s}yUCw_awzs3EtFO1&mw_V
zr)r;?6=|^M3f;DGh@A^6(m{7A*@M4V1pG)P2k_L3*im_r%dWV42<$%+6n1v4FPd=B
zHVrq-czqD`Xn{LVL@Ma8rEKscmFj4n)(b{X>Kpi7s-u{AbpbTUlW1Yj>*}hVQ7CeO
z|Dd`XEIoZF#2Y1luI*e85vY`M2*^Z<c#z~Zc!;^y5WtUALj;dm4F&X3Z6)+QVZ$ns
zjbm2;63JkJ)!6MkND;_fJmDX(>uV~oFVsXbz9y(aBr+gJ=7hIlUgrq=Le16At~~@+
zaLqSl_S6g&&O{(n$?I3BcMA4Sxj<gma?_ZW3Hwjga)d8`6@l-mP#sUC+qv!{H?7k;
zKtpLRz}uS3Fhkl&=>ReydfD8DmDN1dZ56per)YlKwbOl*u(ZcV1Y+VXRQo<7k!@E!
z@lARN67U+f*zJ?4A{W?ME#9E5%lZM2vj|A?mf)1BZ#mlMkVUec<17NV{D|D(>uebR
zX?ffCiuhK4;I0@EZ&%6}snmi(xdg3T_L*;y8~g{Y47+}a$b$SG5%5>F>Ol`_H5Pj2
z25qCG=FnZVPl$;W$b{*xZhMY)*I*Csr=9|k0a4QKCg@1rZCF{IZVvRPeigL9Qx4Oq
z9%avi9wK;K`w3qHDc~pXf$W7I`fbmH9wy|w^ytBV-tHP~Xj?4cVd}BLIu!v;s7qGa
z>r@2%J3TJQgA{@IPfw(I_u~!+QXzM;CyGrS0;7;F>bErQ$`#Lqo>?^C(gq8xXAkl~
z1T5E{Bjoe-T<kiHA{G3JJ$LZG^?XA<QqK$a?OrH{Np=KyTYG`p48B)KtFd~B{`Z>f
ziV`A!aQA=+oGlk2ee`k}e1n&Sm^~A(bLtz&f|&T_GtBGD!-+%@sO{*l0QKKQGI+8?
z?2}F+6ZH5i(#T!0f;M<1xJ^~0gB>ga`SDkpb~U*}z<+S%1ncz58LW@%Tz>!rZl?2B
zDPU_}1x_`KH0rl<)7z>8o#Ltie5zN?_8G!|L#PibRL~NETed|mSoMd%?5&h5b-M(5
z$nCm)j#A`;6?{856X|d7AYyZ$EKMK*e4Qe&wnadLlpSH$8x(=NE<`GLPVXbffNb!t
z+`)tY;12n+SMW`@N*w}PrwC-ih=6pt161S}xnMuN1KjNB?-(EkcE?O@i%@;q-zh+U
z<Us`NxjSXhh(*Bnbf;$5DD-!Ba3<x}wv7a=sXN0tFp&m6;yZVUNZ%Q1iHh8?qu&`$
z-ivJTyWGV=w!H{g?{_KGI|Zr%i(H^1-DSbPa96k)NF+f{)_vSXK(>9}*^g839EgW3
z_;GH?08$}ZK2Eh{B9P5J+y(ZW2<S+E!@)V%L!hsRhm+kR9ikP(-!R(4Pw-QW;9(CQ
zA%aKdRNb=I_6P+m$`Kmqup?ZcQ;0y6bc7qcQX(J^MwrQ#up?~HHb#up>KGx*ZA8jL
z5wTB@9|B&f5pT#j9dUqOHR1-VdZa+>Cl!#rBQ=m!Bf-6jB6NmOB%NnCoA2AlD^^e>
z<`1<(sofe?TZqw8d+)YX)hudriy71iRaLw8jM{32YNhtxMb#cfjM|?3kLN{l<jDQz
zI__(Izn}BGu9JlKJC40(lfd5$vC+yQxO*PoK50e?pXv){-lijJv42l?p5Y>Z%80Mw
zVs;z{vDoB`;<#Afh6CQRj;1G!rAg&-?5i$u%C@@WBE_^^^u;7Hb+d-N4gvU?fhDfk
z`^2lR{U4XCHr}y#!W}D?)-%wT3=#{ug#%5UhD!9SOal>rQg~LVfx%3yO_HeQRZ%Kw
za1#G1?Ms{7y#8dk1|893MXK-XFIjR+2e-IxTqh|x*3Ex<W~J&(ilpt7S;9z{<NKtl
z9hiPeuTIHeKm|xf!T7E!q66*!Ii!yZlpF<plu-4*My#J5H&!Mo8NJgO-Kw4LLgD<j
z=;ty*WJxi}Kkf2?IIK<!Z2TK@p4s3Ns3*0WvXYsZm|Y?0V(1BjqeYC4L#*<T+1`Qn
z-|7Xbf2VrskCIZEH^|HNLM{c_VT<MSNeZ>SVUKnN2}Y-U`5i`&_!9Ccu6n~nfUr{A
z+!KczxU6%D*I9=tIJeMTpH{N}%5IyZ-q_w%@O#ssYu5+3m<ykz{rxjWq7wM?!*@^{
zJ*XS&gUdDDRh}HDp?67VXP>5$3)pA<t6AJxD4>dP)hLT5_-q$5Db1CzUPcQFpSHJX
z(&7(33nv50A!)avPEcILOd+<H4teiHl<_3_X)2@I(g#j7RS*qZt+x-0-Db@2-0ROQ
zu+%$0>X?AxQJ*&D<h_9K94itR3+S8@Yk~gsCsn`pK34F2E#VsfJ8BIIf#LZ<4(O(V
z2h{v?-!jRgrya$<#puAxuUzN62xhnX_oZu9g;!tJ?i?iDXSX&NUhTNYebO$n>{`BS
zgsac@fXy^)t0@EYr40oT<Q#pWm+Zw+D5`kjF)(6d4%1VRZkpe9F+$JT%ylwOYay_@
z+@5m}zv8MT2GkmY$)q!d9Db)$HZnyT4v?RJrI9-kt#K85wNE3SJ`|%lN7yJ`T&)Y{
zW^=(SGg&Hk8fe8cd*vGt$+@A%K&-EFx!Oj$tHzga0!Gg<hPqN_-vk8U?tMNkrV*2w
zK-vIerJ;O3#$`_T4T`;wVtq;OT_&n6WDw~ccy$VXBLQtv&@Y&ilGT}ElyijJ&$7lI
zX_UM<)Xw!a_{@@=Yhb$1^0@$pHhG-wb2r1I3LUkW_w@o+H+H_Y;}WGx_;_)QoUq2u
zp9uU=-RUiW&+*|PuE>V^K$NxRqdNf9`QKX)J@+ffoFF9qJ|=wb{^B59=XBK2i3mKt
zF4vuK&N1^mu+-s-#7t>j%<@MxWOup>JpPle)JGv&^%VnXT#5nt)&I3cy~uqi?h_<0
z7*68d4v7voPff1;%?ZMbXwo8^Q$7SI*xDO%YJSBAX|WXmxRF5-S&~VgM}USaQZnh<
zqOf6VA96k_?~2L8MO_`agx)7q5s?t-qClu5evsPqZOBUZKxJv=Qta*221ie`I7f%-
zK+B3l%JulQc;||r)8tK<#qB0CBf-(&!SLBvKc@N=S9w6!jUKNdGgnOPl@*%)x-h-U
zolLX(d@)#EhFXUmM|aW=Fz1_1Q70=Y*YSj8%Rk!dH%fuFAa=$kQB~QX+i&gx4|E^S
zp<??f_sKH1kAngF)#4)nwL8JMY3Hn7rYA&a=l+i6FGk6DcsJl4Tu>NSX2!G=|Hb7~
zjKkcUcOIhZk$a`7>QyusJ7$_RgMzUP`=q>$c0G?&Uu)r4#D=YtfoQg$DwITPuP7%J
zjrl-~<!WR{ma%E<@r>p^R_d0)b>*~ojziC7;nbh=A7R}-coK`v34{<|eQ3<GWi&YO
zDbOlF27jU)|MZB&d)0`2N4qpYPQsGp^J5tMWGQ&sJSHfMI|l?+P7a3YlL-EfmHfy(
z^HfgdznbWb5(I4e7??5Lzx6W457x*oO`l%PM1x#7m>Z0ymSLz4A7acuc@>AfGmtl@
zZ-6EWnKn#~0?$X@t(Oj4H!7<Mo*E^TEPHfbnoc>Xks$5I>Sbl;n6)K6EJsoHAiEag
zbCQm$enASOYQ2yl--oVe?o>QMyX1I4OMN2wpf!M}K(6U3V6Z!817r8rf?C9rx=AW^
zD{|hc;R9>Ne%hgThcV+8toS#kzWcY#ewX-i@fCMkZE0;)bLfk3Yf(%F8~vtQ9cQ@<
zYRlgx57LD#QH!;Q#&bVkk470yOR1G;1UGJT-T~Pp(%spRyDL1MrKTJ$GJ<q$vFlzl
zs>y4VN)o*b$RxSLmi9ypoR_p5RtSUB<M8D@3ya|}tKI)PG#kBcskzB%Y1q#mV`#tB
zo~t!v{bqb3aHPGNtB?vZ{|O^@v51kDyU^yHHpfIy^UJ8+6Xyjj>>CK-{GdrOg%9@)
z(q?xvBlD$0Crs{x-qN6E@VQ?p+dp}+e^gj3Yk4LD7;)~FQNw-?HZnAGKe14liDxya
zKGof2Y0*$~3#@Z)+L05cB@!wo!_Cl01;}we<_(Rq<*`Q))P*u*=;C7!xk$}=T@@wf
zA5my6V*BS$*(x%<z7xr(v<1UfGOLOgelngk8$Khm$Vv-V+=fgu7r4VOjh3-hD=dSg
zen;9sxEWkzOp%AH22NXALlbkm`khXhXn&b(r<cWcP3<}DmqV^^rB7b;m94?vsd#-?
zOcfZ#_B#Cq$u%_-9e!=5$gWsp-zgq&9H#wbq}Ag4X)$-Ajz=xjJIPo?`4?swHebqZ
z@XSHVN$NyLNbZiLD75<CoPZZKugYa{q02feK$sGAb>nt9&YNVkx!~zR5qR+Ga82z{
zbFANM%c|TKlrlr!D$OU~K4QEzLOIUudw8pxhCk70=Le_NsMcZ=`klH;ll_kR*_}@_
zFzKC_l`8Sb;jG3X_E^M)e6A)LVC+l%`SGu}!w=7!i%0krl`FmmHnFL!fL!&?tF37P
z@uL)dILV47t)8k{0o+HO_Wj}5x2U!+8n{EI$M<Iye1hgeg}XT#vlhz-is#J#z|yQK
z{-tA23rHUt4kfI?2T%N2pe0lRs9^D9NOgL^UrFEbTb8i5GhV|}cf;p5#fx)k|Hsy+
z5bot^StB9fojPj^>UWt(kPBZ?W-LJLC)4y@n-{5Zn9Kp{=9eqCB^fk9VF#WLb$>A4
zlG^`f!`)fdale-b^yQn-e6E7^v#<egwMXUm=&%{S-)T6GgbWB0QW7WBb)^l{i`#<W
z0>35`x0#ZhL<-&Ae{WSf%}9mFO(+77V`NG2J5}D{u0DmNv%9HzpBgkn`+*&yZ%_AO
z6vEX)%t_U0-n{=cTkQr|!Kf^i+l#juYRQuA#F{DR!?;yAYRk2UeXGLq_3|EuqoP|w
z%e!_NVC*NZP}B>n<A$LVcF^@UTkv+H)>2-ZHL2yA4}s#`(Me@LW#8M4ogmB(D?&T8
zo5I6gD&<Jb=3q;2+KImdo_`5AW6uxHyQ<@HVnA~7G^$*L_U~<}ci*gYbv&BNi<w^W
zb)6a!4&l0q)(ypQf-D9+C<ecvuolB^l(w>W#uJ(I7==MXewl&rn9w86SB$r5eRt&P
z173z5KNw8c&meg;<pcIJRlip)m#Ej-DEd6-KgQ-~Cr|$5FO1l@14GVGlwZPo0xvFw
z+YD&3<Bc><xac+7JAWjJkTw$zC{rk+Nt?gslc4J34Y(u%-@nfBVsw(6Eu1sXLWmCU
zYozD)B(h?4AG{_}us7=c$Wh<b&P7mfB@8|P9T$yMja8ZWfN|TKux_&=m)jSCgs;G%
zPk(F5RZ&=1T^P_;70XmaTOCp~ltlC(IP&!5h<U}H{rOPRCV4*uiW_;a9rfz0z;cG>
z%XS{hW&hXUf1MBI>YwJ4K7eQcdbeVo{w2In`VX~Y3k&&5RsRZhPIgEtRC(!xFdF<I
zCY{c&5Mig0V>F~w0FUq}qVlOt-|+6~N3U5%$=-!g<57SbHeUdA9iC3!ZB@KO$6K8<
zn2dFl)+%}+;*rq^o)_a0-<2!ts(=o14}I!f&N&8I`}Zc_daPd@%;RN7BRC#ApCr;c
ze+&3bfC9;>K5C}WntN%!cG4Wr{~hBArxh-#1FOLZ#;MX`yA%l5Rtw0#Lno3Y(4o>}
z=Iwn%HQ>8k=j<Mp3c_Wj3Z5>Se9A4X+~INzbfC@VgyICu4IbL9cp_OkJjq>Oq}~qv
zDagQL5h?Dewq$iID2-b%rtfAm9jJD$3iu{-V9czk1+DJB`wtr^46^yPD{Fu5s&O*p
z!*(>X%YLfY5S_7!z{xTog%@_E?YW=7WrX&|0h7ICY#ti1`hPdLF)w>=c{=tDhK0mY
z6NK;Mf6QOpk0IPS|B5mMzfIcE+J55O*^qNZdF*%v!;)xNrv`}!EB-EX;j168wrGVm
zt%*|9QA79bU3UkE-KCIszc?d>ldOg8L)~{b5FaY{aHlaoyHBA_XWAq+&4o#Ffg&hB
zAkY>P<3pn^FRhcF?}1|f0!=^Ji7(ztZGO@1Rn}*vbdYI^Y>52PeNx#&pD5L1WIucU
zR-SHdo_RU_&3(!T)Af;RMeL8E=03r(8_BhYw9BQ2ppFz!yl$+Iq{)hk=bY56TG8&W
z#81%;H|LO+uLYWox_@GXS$9WWAB&3b0EgnRvGzGF61O>2g18y&m-1N-m3+N7e3N~H
zf9Q}FA|rB+>-S@nt@Y4EhqH1hn;(X>krvgiH<m{ngUA?Of)q~PeO00PrOiKj`dT0D
zW*GqUh>io>|4`#>Wsjss%0;V*_d8=4NkmX1ingA3dskQ0m6{>$IN>?bm|_=zgHA5`
zG4D)z2W+BhhG*2p+PvV~;EF`|5rRuhHoLDl598I~eipO7>;U5x2(mf(nY*FY5S#>q
z49Rv+?hF!n%IRtr44-Pkkqf1byCSmw6q1XA<4x$I+mja~d9E2I1+(e*VQIn31o*Zd
zX4&8<N0#HtW=-O@9>;C}8V(5ih$3-MZCTUUCFo{O%$xp8uH{iZxZ@W^pkaabh^3Hq
z0%sj-aL2}N-I-HKSP#ut&`|xsF~DMvqbM2K^OloVS&3r2gKyw@{ovJ^GL>M6Q)PpO
zI>~F6Zoe*^a`Ns^I^<lS8rLMB`0ki~aR}4&f3q7d+t$3(P#W~h3jm~}rZi{h3Err-
z=sNl72Boci(%pb1L=Bh*DXg{{ZylqT%@c;nxo3)vX};9fBe=@(khz|_;=8iu2X|ug
z8b#zd?ntS*=vITxT428vcNvFLhmYjGB~-);w`3=gplXF`P`t0<y*>)$&?Z;wsG@w9
zoZn<GpiSIckm3CoKvaK(m-x?w7$y2s0~XGcc6Ke2#~fOXJZQo@+Y<ZnX;-!rHc+*6
zZ?y9{l>4zhci<u9Q~s!`9m7_g^O%#t-C@1t?dr8;HlGR$`<b)+2WM3B;szMFmR9Gt
zS)F9f;jJWszpkNS4Waay`-%&_l8kh^*rz3O-h%uWd4Jas00<;p>J2zvU{+h;<-Sb8
zv|W$poA9-y<VN=$Xv~QW6WS#ZuJ&n{lyP&MoZ@f1LU+Q+W8N;Ad>-LlStWuK93Om4
z(z5nhvsxF+@jM4f7h~rfJIz82W0%)1O?sL38YDLNKsOH~)Egc!166_IqYJ*~O%%7s
zDJ%}2yyyT#<WR<O%G4fc*TpC7wb63#)s`NTqup}UAm(;yD~7eVa5Y|Uo+l6@SSdBP
z2xaobc}L-~4gx+WA793J2HLDMz!TPtcIX*k6N*Z{ZslbDBG0G3Ls^!*9=$cqB~2zD
z&X<`!a79mVq(8a(Ah#1LqdFLdYY-wNe9(3CdI|ts8eUEiHQf6VZqIPtduh~xZSUd3
z5uR<i8gX7eb$jzqX+G&4(D-1NiE-h}5~rzg7&|crApGWQ=EYg4_V8%z;(gjD`#!=(
z5p{EPpRI><>_erDyBZEre@X*DL*1kox-oNurXRkqC@VpTvRZj*<?J3OvG@v0?DwRR
zNb3Xaph*q1eCXoOkwICypAg%7XHi)VE2K7=u&Y}OCb^2unvwO^B`-4xuVL#&=~&eb
zQ+eMKeOZ-q`4E~RP|9Q~&lOhvq^xBpGtx_<Dfq`4q-C@b7g}$vU=*W7vg|dX@zFh5
zWQnt~6Zsuu%_bi&#oNMB0JcBK&Nv{%p#Z<RncIXR@Q=Z{w|edMYdiMra$^@3gW6&y
z`;u4Qs4QP017)w0*-v&scxsN@(4;olQdWikwEt<bU5WKuq32ggW^v?|>fMSH!XRnE
z7zQ~dM8Qn65X>Vy4g6hkFi-p=dHwWy*`LSXqNPmP%v@pP?#B(fMp?@hT!9&0b^xQ~
zI=A@vd{Lu6x&GN&9$AED=tt*6J$99X&O5HxAnOX!5>FA3bYzVEe%QkpM4-0iD+0eI
z5_eEzdetprg0iTe=9b92Tf)QxlG%t3z4`i#q_L1{`K8%i-iF(beSb<xL2W5)uT(yf
zO8Vk&Sywp3Ha}qv4RXF49?uPG-=H-S^i!~Kk}N7z4bma`yqcD|M#c$T{~%Of;g`5|
zZDEB}{y{txa2;g&4oBrX>26c4myOUb<?pl?LDR4}K{0RWy)3YO(1T#Z<Dd6;AFq(k
zt^Q;azAWaQj@K1{Bm%t=7I&a1OT|^v#8m+{ao<EYvW}{pudOC#siaid-D9i>OSKKl
zMp~VMVr$zo%N!c|X9z&hFE7!B_uQHO^Iu(HNtEIW;-_?V{I(31YsTI^1)XwaK#|J7
zP?_9fi8Hw@cyuNKC)XqK?m&q8$+$B(ft;2YTmKsgqCc97yMue)!LlYA$53s(aGPFB
zO<Bf-MDpD>TkU3Os|UW#0ba2t!giUQCRwB=YTvU!S^FtGbO<<_Wd8isJ>T-swN8DJ
ztX3{expI8Y2KBzx^-SmZJc>j*iEM(!eNXTE#R<_&T4PGj%?T~+<zBAbX;#YQaYV=f
zS9?c6)qm3973Gk?LMow<RZ2PMRno@8UN)VyR9eJGlXss%psP9<ZN+w>!>tbf8k{my
z_iK}1c5f;Ja-RU+6SR2kaz+#hhV0WDSmIt#ML9n%w!)^BCPkB2$(j1&5U)yhcdQ;y
z5{)sL^Y9=Ya-#s;4VwHo%>U&F)#N!V!?9UyxKiWVElyRE*O^niLvII|#c7>6nzSY_
zmn7t(5+KNc=`4=LT0#7Il@gk^?8FB)2RIH3Kuecm+yVc~5bM35-!Eq!IxVLiR977P
zLXQHU^}(Xthw=O6zKuCKvw}6)z_i7!mQb}S)2<1BBQA-$k*~k9mKj<)JkIypb7Nok
zEk-pC)6UNY^5UHT>T_o|^=uoDgoRqylW(-3f^1$WlSpE>v#FQe^PGO?3_hkPOV)ia
zOuRV^e*=utzScRG%LD0l`%Zq3CuKQFQ8eS`m&S3&VV}r1f7?<xi)<&1s+z1ze}}Vx
zCs84_hLqF2#YSev#a0*^eO*!!8#Ix%XY`tJoSrQbyt1In@PNo^=`q2~9o3~xz9Odn
z$z|7(<&*nIx!cz%E~uTqgJVW?zcLVMc-C!p2jC*U+|3_LRy<U`jXxqb{t2^F$Rdh;
z4^s-DB0fTgDc2ppqi*j{|0(o4)@U}~OO<S^pO|SXowGZ;orExglsk%0rtZvVNb-}-
zmBU4manU8V6&jyba78#qvYM>pqsQ9g_k+ZFy+Y8%4iP+Y$3Z!#(SJbYx3crkB#9_f
z;4K~UX9mp3`5uU4b?Tn#X+4$D_4fUB_)3@J6`4a+`$%5nUWz4X+fExb6B9b{FP;;j
z)A43=U_g5);9(P6#|-zy&9kR}q?ODLr0-Vyy7eXvjk;1Y+DN}`(sE;6YF<E`Y{c`a
za1rYiWWd$TNxVb0L@3118twPP-AnbXfxx#>!?uB)FQ7zP$0l3sDUM|6d6u@g6YDe1
zeQ!bSBnIxe)k&(e$nr7L1*;6muMXGFt+qB5k<MM(@rn8NC0=IalbJc`%55EMz?b0>
zMzy&s$K6-Yb^Ck8XYeZ)D}#<0R_PM^c8PFbzjf{Xz_sq1^}~NYTIY_7R*;+&0K7a^
z4=$04Q>M`-;d_e0j_-f)#9xWSCY<L{w>i;JnUE>+1O0C&O1g}i{~jze&9a=_p<Wgd
zMei>9X!{+0`2?se-U4(Z102v>L#JHrqq_E|Zt~EN!gsoJR-QhTY|9}_8A25OD^s({
zVW&q1iVErko3uOKoNK;ValeuWo}f3q=wM11F{bXoJP|tDBWjL-VxEh?88_(R{VDZ-
zZFrcHZ5~}MSsNe-D*Gieu=8RYHqrAM@d<NAu`Hx$c`ptt`;k!a6XlY#>LhcQla!ID
zk_U=oZ^*QDFC;zq@)k`k#Jm2da^r*6$r&lSXGnwDe4&eW`Hv{<zQjEMWoimNYpoac
zZ2S78ul`ivE=;eMOMEkCa>{v#1ZkYg)}PHJpqRr*qEj%BjbdaunlEv)Udgw8LQlUu
zHT-0J6V%}yI#>$6H-1g;r}QEqm>K$YphnxwPuA<yEEL?Wp{8j2Km_zl5FW%(oZ2*s
zG|F+NXYmkBvkLklftmtR=~xuQg;e;}eomRr*BR9OJ3<J~D;0)%R)9+$kR3cAhw2Se
zf*#sZ6b5khY|qpiz>O7HqpyTX(KPEiG1o*~x!eLZ!7D~9SQ=4ASvpH6rM;&y!qu``
z(ZW`7d0W+G=R+EWhz5)j<P2LCt3Mec{o%ng^0b5VeqSMpi?0A2O{8d;uCzxU*Q$u;
zsR@$Tw;!8L@9|Y^fYcqX>pV>yf}!p_@`cG$-snbstuvvj%(9?473UsYG7YK}3t2lB
zG_GtQ5o&);%X8rGi2hxEf(WtE;A6PL4$&RxCo;uX?FY)ve5n5>Y<Q+d+L|f*6{8K2
z3AuKPYq;CNyahyLPjGp3O;T(jV80ww5V?*eYPULXIAk{4;|d~WHh{!;%#TzNc4-kY
z%TurVl+30?Y^$HW6~0(_04+(UpJ~sy2R4(MJy{ASLAvW@U<9D%9`13c!Ya<Y>&EvG
zb~NvCCukY8YDeAJmMi;tI-(4tz;qTwQeCefR3D>x2v{@2=1i*_9@6hDJ!G2~cUAQ0
zL!G)r;!fD}Ve909&qCU>tya!^p=}XlCvaW*6DwV%>0($rr}fJ5$P4I4lYEC;0a+q-
zESG~@Eci~?eUZvz(3Uiets>qg1w+Fi(C%{=h0&s)Ii0%qy5yp3sw2i|(TgAS3qe@n
zlxB~CeVsW~_ZE@rhlp0X_h4=FSJqTS?HTQ$pl}^1R*?>&fOvH`;dcx9b2^Io<6vQ1
zIB!kxjT59v>%7TB@ZSu!Z?m9D2v0m`+im(OE>y7sN5%9er_VHlFdc~LWo$6&D~F#B
zU3q8Yc^oj=LZeGQ9J0$-6937ZZ545~Cu;)*SXnS5<+<~>bt^TywGTMjoo+Oz?FD1E
zX?$Gkeu%n_O^-ISEB;IveH)#F{1U34uqWWgGmb!FgWhqqdm;$4{RR5*-uty5u~z5W
z;pXpP<U#L7;)VocgX}63N0xs4_hQM>rkbp08dA#nC)^DCapqrN-K(K6X=%CcKRrNZ
zyN)r5tgu5tr?4yfU5@{yQ7Kooig!jBzoP0g3oc^)p+pH9Wrf=7PhL{YT0Jm!gXJcU
zzjU?B>z@rKufP#sH-Hhp9D});?mzx+TW~rHMkm;-Tn6;^9c%|QtO}*h?tB*-jxj=+
z#$X2|1H*;qcZLb{{q<3v>+QzFJHWtH;*W|~P220Yeg&hcCzmowm#+;I87QV#yi_{O
zFPz9Pd<8*!M`=Zv8m2S2<YjT+yX#phlTdI)w{6u@<h8fKDgSxjm}yXKYm(QML+qoQ
zCm^j2Q}J%!kiu`9H&ii>T(pQSHLS{6DyjcdXtfCN;EEo1U||Fgn#jhH1Nx=5B1xO)
zky~O()GR}`3hiYED~G91;2W0FySMBEU8D97gVHi_k+*+i9C5}s0mtksuwf<snx&$v
zw<Y~d%@FuaJehgK;wfBlTf_PREO$^4f5wVE$w=Jmftc5^trr$ch>hFH&tk7{{73hV
zQE)E$vx3b)!c?2JiKqcgbU*Lhq_9n2>yqvAp{IO7sNu)lpnsbw(%tkqmq^-spqT=G
z+iKrzkn!Y58f9^xAzVQk{68gmQ%b6{TKo`aqsS;3hzNna2@$-ffC_@}1D}6Tfwcpu
zL5vVxHoF;x-AMYAO_mC3Xmuc`!P*&r2)R<53VQ*{w1z?j2t>Yqr6!8_mV@T{c5wTn
zk$?f$hm&_?>&Fsg)4h0wS9eFpQCv$?N}ivVT7FIy3@N9nK=vPSq>t<))@jAdhpg;U
zln#f}A1O73FB%06C4u`l7?F=P@~DYsrlr{z^C~Jo2GzG|9vteb*@-9NsOq7xCH?k-
zbS_V9Re)9=F!Y|1qXTU3vM_4r*tUwy*c8F+wL6ulk1hn*?Y15VOnl|i`*qZCTTUy&
zB=w!78F{?hZGR;jLQ3dMCqiuAVc`VsFh92u6e#EY9i>OTq-9NcLz1ndlw{6`1Xl+6
z;`rL~!za!Xa;74n$dKFUNPya?Y*y~zs-LvIUrq_AJ~Lr3p(6@(ETj(Ee=;m0fQL(M
z^%@<2#ji^}n-!cXXO11FXz8V0I1!{>2{#p~UN#LzhcZ218C%-1{<uqzynPW{@F<Bw
zdN-C=JSjRmVG`CF*^v9yr=E6e(L}fXn{CAN@N(1h$zWy9zAdUULDNMKd?eqy#rWAm
zh@~y?Lbb^3M#*f%`Fg3&$_qaAJ>fwRf@`Y5Fgh~=Y|rRB+g?k^+v$v5U|EYKL%y!e
zVy+;DmZ;2|xpp|M1&_@kzs_0n*$esj=P5a-5&KLe{9HxYx$k|PqKFWuZSk3&YcaUv
zG({`h=ySvyK*tiu#W}swVvUm7vEja{t%m@=TaKx(=Q+MR+r->yVn%GVJjE~ki)i;n
zZ!;dxK4$rQo|5a{$Sy*BbCYRhmy@l%-ud%e`6x<zdal?AL&+uHO0b~#Upa8DYoFfe
zo1yi<7jvbgf0GTLC4}N4J_$pW4iPKl*O_3W`pmCnt<5;X6_Sj=K8+$?X(^`Rj~uY4
zrifzA_Ti56(6nsILUT*-f*dDUh;{I)hBFA|Gg}fqZ`7UY?}JTC^;wVK6i5hM`@DLd
ziHkt>rw^$Du7QjWXxfv_G)ix5te*%3O_ut^w(czyS<<p2i^_F&{N{Fvza_Tbm%Gi4
zKNodH?@F14D+i}ug3l*OkIE6^K<1RK)9}?<DiI42^xSPVREZySK!Cx1A&h({C<fG5
zAaC^VW36G}yq<C<79)k)3#wEVL4xhYlvoz3Y3u{XZfdzf_dsu@Y<Vu0oOADwwP!DV
zI130DY{bx_;2~F$V|X?Sm{*RD!?aSqs_+sUt(niFnavzxg;?l3!w6`>4}iXWayL*v
z&k$3{I_nfU9H4qj150|*9JAHxeh$Z6eitR+*|CMXOK)bVo}ZhvVOBVJ#nWve$-GVW
z`=wV)Ay%#}a;wQL-R{={KK#7Q^5J;LPsi|!^<L}!@_#?H?{X<M5xJMl5$Dx=Sw>N!
z=46fL7I9G^f9?-_TEK0XSEDa3azXL8ck>ldB0hnpc$%HI8?B;B+TLjv^7kW-s85H)
z?I{!kXCSFD@eNAqY?DPVZQ?Y;rDXZXT^1j|X_v$7;WzUpp5J@|8!a6Vzs?(KUC>~j
zQJkGMjM|)n&HO8gQQm!`D}HB30z$N?U2s+__|o9Qdem^>8c^SRusPRmh+YdML4;iI
z?B2w7D=WHJk{>9#ra^agr@0gj?@-IxTr_Ji(qAblxtkv2(s=TurxY>XQ-)%S=ZxJK
z@$gsH|F(ES;jPExJdY?kighiF*?p>Ucmhuwt{hdZjt|*9W;a%d&uNr>Ay|ru2dnUZ
zpFU;zSUNHW1O<yF;*O=X75d~5%eoHeO;-nWkH)Wj>V6HL&tf>o8?~x`lmUYi<zz9+
z9M$UvZC8AwLDJc{KNqzKF3(x7NC9!!0AcqDRm=g~k0TUqQzqVaU4#^IEdL-#i|u?8
zSl<PqWyU<7ZSmt?Vhc1-q`x7|`vuBR$!sN(9mu@EKhT7hSnc>Lhg4~Vi(h6x_&wPB
zXPsvbHW$dLFtpMVgs4f3Mlra@#b^h(9UkjjL-igr$S4XHX7FB|BhbHH@iyY-M0b5T
zsW#L?D1wUQ7or#mk0-RAE{IVq*M5QNjfjp0RYv1N8Epm4i3(LFoD*~3D+Ow8d$^|o
z4}%?j=Z_S7IJ<}bZg{c<i-qC-<om;z|5k{LPfs~W+FG-9J9?;*wdEXi#TflK3AStN
zR{>l{2Y5k^=IPXb)uEhsAu=|bIxFYx4K}I^B)B8=4=zR9(Q+Ewn_sU(g67p$jwQ+v
z56!xzQg#ir@Nd#yEs{}nypB0-v_6Ic9a^vbgE%{INGZYS@9Pc-vdn6!wM@_%GSy`J
z&y^v1YP|@7`<g1T`Xc<MJLf<u;4<cv5)2*YXFdy3h`dCt{9%B%j)rUg`)U9i(+CTc
zOdpkgWVbgy(z#qlJbqWLA@O6(9Sib5hsRS_fUpz<_aRQwmSRj*ge+t;;!zF&R&1n@
zQGgEbY>JS*BVtjs6jd=%zFqon`)eWb(9`ZIrenyRBWM1V>!p7Yzp3m)d+1<(*Uj~<
ztrkVN6GC6$c?nn1se0dplCt3au5DgPE((L^k1KCuX1VZu9m|&H1a|>SooRETx_HZ8
zqZ4A?c<#r}hh}#1oZ(j-n*dZ_gBMh`A!sM+^Ha5l3L`#2z8dZ|ZNIM~xG14H8nnwo
zMQUX8R4_Zst*+eE>9jrl3;unb3uIu+apGjBW_wskv?S=V{z2QzOz?P3Opr}qu7=#B
zin1B(nDgPOmF1oQ+*^(0GC3PxG-Quq;m0e#uPh$wxl+ru*|8P=0bshpe@@>HA8Oqs
zGRa>m)o}f0c~5qdVa4?5&+2_YS28rkiKSTGa-g|lA3aG0tE=534TrLTLU;6Ef`kdF
zrV9+VlitcDz?W!P_Z`{N)EH%Q+r;Qwz*C!ht-Fd#JB$bHy)fowhG5?6KNqJXvbh5B
zupUQPV9v}?GCg-6P&EYK^X5=8wNY}!K9w6KK1l%2*YUWXy79o*L<=&OUb4!B^FQ!J
zS}Wm0uO#q8{}M<K7;WH7>YS0DhWV2FTI!oN=|(G&fzRHnEmm4#bjCGAjw*!I;A+@%
zrUMC5HO1<O>dtng?|haNCqyG3jV958-CI;e>NN`|=71%@G5#7GFRDTN`FH-ef)9#h
z<H9K6bKpnO9v>&9y|dpVCt;>Div5rreg1$G=N;dS4Po<AwST7lv52+)wvChK3Mmy(
zM4G*=o!iIf+4?mWN4`GTPpw1QJlC#jZhuZ#yGtD6csPWs%tNZRUix$I$@d*!{TDoW
z`=F5}B@GE!EgnKh1j%0J|9Yefhk_?$`b#<w5o=`f;S1OkT76bOGImLN#6Dq{X4#w(
zn-BG}N7)XIa0^QHe0IE_x6sk^3T9@U(kcMOc@H#qsz8zQ=d}5cI*zb|=Y^yX_H1XX
zmjQ}8{R=TiT<?u0+BRax20E?O(TerGUZwt02iPjwm0qua&Ks{%F8-L}Fy4#$cf~ha
zQXjkLUAF3&KyW6Pln!FRYPkFu#st-Jxpma$K_CKNf;FvWhO|<gr-jsOy>Z%$DPs8W
z`3nqOqy&fG`T@^<zBJW<nPEcCjKS_ZH{1je?fsdu#f(h@;h!sEkVB>e+5cc;&KAIt
z-XW4-wqXxB>ST?TIDe`a%YdXHhS9s#F1FfLDp9=r9E0^{Mw2}}<8klBB=L_}-xQEa
zFQw4ik7vtshwQJA503DQ5pdb*U?rDbkJ6dnYT7kJnK=a6zh%hO{W-^z>p~3lWv-YB
zwMT-KjxL@ygpXORbj9$hxO2ah28aY8Y(W;sSI02ozsbc{8<u0d($J0enE8%9!mWOK
zw#P>~d*Y1#n^#A>fqxM+nbo>CODlouNV9bb*5F1W3k?|UwbbxGS`|5GgQM;AG2M>8
z?qLL-Nmd*OZrzoBiwo+A5!Z7Er6yf}PCa=!>s8|#j6s2qZjKs5v$McJNm<+_m-^)g
zYMW!rT2o<29wYvTy{)cA*!1q4;JA8g-VE8w@h@rkl49^`-uc|lZZ4Z>P44LO?woH^
z#7n{HpuG(bSlbJ1Ja-1!XQ91QY>(e6?AvU5xX;f!Hq9y#^1Z%_9y#q^aKu^`r&;)3
zYrAbN=6x=u4T&3;G5i;R3{9xTAqC4)tEL9Qpwxb?9-_<a`6ZSL`QV#U9I>3AL%^gE
zC*~HfXCCNzDq^vcw}jc!0WKfHxB(gKw8S=8F&1^iNTQ&ab7fkMSM$Tt)9unGE8(@K
z#Y8JwCV`x=IibU&;RvL$l(GmPgH|wkGuW64Gn4n}*}y7)27pq=e8xqx1^0w9bLB?f
z^Zb1?!!kanMm8;nm(26=Tql>$w-&4@vo0ypcU0;o^UafA5~V8ZYDjRtS@X$Cs!V!&
zYXS*NI{wAT=Xex<rPN|v9pU3V%u=U%@BzG&%>G(KS#*A|#ni$$t1Q_!6P$hfBQ^Kv
zZ-%)SDn)+LMt~wjZ-{ZTT|*X}sTp{^F)+~MEE=nK$7hK4HPcCJ60Kq>_LrZ*@s=Mv
zH{ldPMND_2qsA5DK{v6LjW~X;!B2OobVgKeNFr!N<0Kv9joZxzmHmU|{+QMsSdw6u
zsrny8Twus+PaLD^{;BI-2tk@*H^+ovgkC52IS&Ie7e**Fa`G%n$KXR%1QXUG3<|Dt
z^hTLxec8W3Qux8%r#uR;dEv;1;bkYOh?fvTunz6mXsFwF?FOxdZ52s0nx^g8ehx_&
z>snCe^EL}{LNZ<z(pDUc@e{Oab5}DD2r2>HcyWfOf?~x~bNp_egk2m7k+$kVQNtk)
zfO7nrsoe)%$OiIUbM-07niy*ME7L)5+Lbf$<r3!Va>oZv?)l#K9w|Mt?Mmun&}7HP
zTM+2Hnt%+cSV?x0FyYU2Zq$zjKftmk=l-Lg95l1(XapoiXtGIL!`vUDE`z+3m3~Ou
zR;P9}FRjoct7_ain!?@dVD!kA>mpJ>VNtqsoB(nP*-1QIrNaf|HAyi`Yc#$;2gZdw
z55s!V8DeHcu%!IfRpel$Wx|vLeV*u&<+HyZ!QZ*uui*2mlcWnhuB2Z@-JJE(gXcoQ
z=dNsm^lle+IkC!;%MDb)uFOMOb8=T46`a&+OlEDtDs>w^Taz6%`M;ZwiL-&zi8ESf
zN31;=n#SB8ii=r1P|n^=U_K`j%MOC+5t65z9bay&HjBUB!03~oWf9c7uUnJDUwOM3
zYZjlQ8|BKtcWp=h-cUVLXz6uxj5fvr(mmB5bJ;5ONCih0gNdW*TkPRdQ(P3{OdVf%
zL9r^)p&owv<Su&ag%3<`7r2K%4`4>`zPJbLnQ9?9u&Kwb>*Bc<XgR*zMuoDr;qVjl
z@K@~bS6`f}wv3x9qr7d=LU_MEOM9MXMMuf!A+XlnlS7{%Z~SRV6}TG{eoC1Nef_N3
zM<u>DWf%@yezAQ+Vxk#Mcxcp?AXzq4LVEDO)&`~R*#t4Rf`L<U*tZKMc+j1Ps76}#
zJI{}9S`|U}YOdlH#Rs?<knJG!);9Wibv1^Lk;TJgmE`<+-VS(Aa($WSF1(eRjhexu
zebu0i_TFglr1TH&Z?zvR_go7~{EvGJczRY8xe{}77`vy&<3c@0QEOJG+R^4(BM6l}
zM_vnK9{&rA0^r-PkK~9Dgk8WMj(+WF%<a8qvICiGy*DRlg7$NaQZKN&vK6BEdN=$u
zfEP{auG1})5=(%yxP70`xT9V$>^|J#eP(M|HTblX0R1GP)uT_YTXhMVwLn8)!chqE
zb#IPwrXBAaAxa~k^(hDw1gYlI8dx|X0BA!0$^inPF|P2_Sr|d|+lOJzjRm{uHJn&>
zl17U#|CrCtTyh91?;-Rfm&}vs&0ZLb4kfgS`H7e!VbiV{94+7LgF4yqGvUM4Y??;2
z7VTsZ7k4VMMuI{tNyN)xiI_tUhtW1YK5w!Z?L}_B8m#R-JlpTI!6A*fu+UV#<v-Ox
zes96PdsPu=`)lvLe~Zc)96C3jLaJ+Dhso}7+5MW-{^vCCYyF<utG0=tAW8QK<*$og
z;;)yEFx<kVD~Dfnpn4Pk)jzb|gN+97#^R1rLkN2_`#t<=z_h;QfnYxD*VF_`xk}mo
z;`9~7mK5e0eB`_VJ&k2UQx*fezoy;-WJtND6QW7=U8=uJvdKf5<U~}32|d0}<V1@V
zG^i+YqQt^M=4I};ZnK^AN2)9RjF`W7@&2T_T#Dt7EOgp+Gz<JiM!4GDRPE2pn9x1G
z8u7#558U%shyFc&XTsWzw(!TUIY%x(ue1svlY&<iP3*MvB3Kdq7P~h2s`h6<>&tt)
zXSd(ukYK~|oE#T%3oQ@9MiZ=@3$>sKf_1(arWhhJ+PUCV@Kas|AfJQ>b}GJd9l3VO
z^?PKI12KPfweGBdP@Z|>qnCI5kPk8AeSuZrUuG;uu%33h(x9SWIgZg}D=k7{_+hau
zDz~GDn7tFqW#eTH-Jd=CszCtO@rxRlt{m*+I-sjN=Ll1yF>_MzMbY7o{%zU7u|;5g
zb07+59U9oO{-+k7pSD&%rU4G`vQ62fCMYB>JP*$;;rBK_DL3lR%C5QWS|X-7uIh94
zY~s>|2<kAxlT=~t8%V}u>Yq<%*=E~FZNguG3xkROFW5VWJ<dV|$jPCXY3H$72@vUq
zy4vkCy)3pCXWnVnj5{@!T2{N2t_TR?iTeTzdZvUC8k8%Z>a!uqZy|)iq%txICOU6~
zkTzduB?NVD-MQO2!0=n_Q~{s9)RL+`l&hR*0YMpd+XZ&eAX*dBaUe^V_d~jLjLq#;
z>TWj2$HuUMsSqG0RIwSns-l5210e;xV6<jz`N*2fVW~gcst8uU=zRC;vpUV+*HEOT
zp>4us#$T3m=Q-cTIOVsM{j$9o@#dW-A7V&Ojz6yG)4cndcs6P?EI>vmqwpxuk#`Yk
zgo^GTU;2<<dhux&ntT{?Ja{`c0LJ)tH=VzG$KG3I#|ikb<D95gw#P>aYV3axzpYy=
z?*SVCQQo$YWfYD!mz(ab?VNW&Q=Bc$gm~m@faxaTmP?KU>gnWAlwtoz8lnQMTV@^<
z9O3i0WCzk5;uhI(=U&f`ml1~nG<t{789UB?3aLT((z_gqaCY&7+v$hIbDo1#`u}V{
z|936cc?7VdnFsx#n9L*HB?%e-T?zEoMYmcNCX@ENi>;HWK``(EZ0-Z|NVx7j`sEBG
zzdX0i6HKy=93(^E9>LhUCh1>HP!N7xbyl90=d#}eu}{IK{sq6Y)$&NbVKu&eIeSZK
zd&&kSb2%GjeG}(2kbQLfWDfJqsNsMu_gD}F$ST3HRp&p8_Aid4<Mv!2rFxtHGW3iE
z@T0OH+1ig2X`U)wmVtdPr{J{B7%MV)5j4pzwC>gtlk6G1n8nVN0X{N)|6G$Fmj
z4>Zr9cF`!T(;4D3g(k8$yZAcjX*JmYGV$Su#k;x+9^`SM#)ow)&Y5btmVkWQDXTXC
z-LRm`6U^fG)(yqpVMCKbl?f@fe&<DRhYzw&_K+QX(&Uqar^xuDcYlrr!Y=LFs6T#F
z<T$&#U-8zcQ!b14YV<6W{YE?1u$6+3ofK(yltI}|wy3E~bcPJRdcL}sL1k3mdPY9-
zS-Qg`xr2oP9j^RxV}RMMNV-=d)M^pflVA9R?E;*m#%?N87pz?MqVbdD_zidKq_IHD
z!o@K`qm$Oki_#Pr=V`oqKR>x=r$m6<_d7P9`b9uxVT20u9k`)#-8lKvDJH~y#8O0X
zzH=(rFTlMci5}3%Ta=Ks_g_F^p+te}?Bk}nmTqTbcy{vFmW%0~vxMU*J3tocXOg@c
z;=pHJM~z=bsn+LuDLxAcIqp-MR6$H_&d5Xddub6hCN>0IRKh^AFfYg$U|^C?Ah)ZL
zuX_qrS=xzAx+-QnYb=k6Y`rF^AAT=0YHr#pBwSCl5Um-0x%ryDtlp#ra%P_8f}aqZ
zA9U7631Q$_)Xq$(9<i$jKnF1CQVmS*R5wmV{6xZNyHlrHlm0A@43iWfJF4m5X77Ll
ziPMuF&%BD>Il?|}AH(-f!^j2Qe7+u0dxoEfXGNR3LWme|zEC+peSX>$BO(NAeM=#<
zGlk**+Fxw&Byk#5xU#S8>-4QNj;e{SB>6t|$k?^Wt%Z|_>$P+NL40WRQraX1({+`Y
z-zDtyrqw1MVGrMNn%ZU5>v8*4UKDFpo$8CRiPtYc)_;-P2tc`1)BBDyfHol%3xsTx
z(aE10Imn;42}{RmVA%NpSc>yzDnW5=s`If@iHPi`a^%WmX1Dbes!UBnT0<=E0%_5Z
z!oTw)MQob8m=|bD*J06n5KMV5G2W2MXfsnbPX#~D{JEz_cjMpqF28u39{+kNrx8PT
zK;Dxh$=H+}lU%ne$uhKzt4h?LVM%v1^;zT?jeV33aaQCTS>wp8k3Y2|CSPgS-wnH=
zHCE!3kvW>OQF^1lqI`LNMC(>7p^Vfac|WdRLf=`NO(71T9Uq<pFQdn9aD|bJ68r>=
zjHM>v+%i}luogA9GrXgVCd4dkUsuz0NVcjQL8?(5l6m64wUjor8If-LVhZ~e9Y-o8
z`&#!|jO|r_{Yay!C=&}#GQRXAxWHgg>o*MV#_HJ<qMIWj_RlG>Y99ob&qXS^c)2Z-
zLLc@XAl$*CyO)1v{-=E9X$-dXa)OeO<UW1ZmNZ_45Z2R}<Z9gw=qx(5h?KZ?zCUI{
z!pHty`1x$S>qxB|m269g50?Q9WKstnCjk`TE9=|o3qCmzt&8)-Nw&2I?bq<dasEWk
z!EofFD+7*?6q(|_w7_A+e-uv!9dHkA{|E+V$n$JTw_Upmc%WskI!cwvN8_U&!}K^k
z$H?Rkim3LCNC@$k<LU(j*0!k`N$T%Ko@hYw+SE<a<Av3@K=WA_Wpg3Ca8!hS*|_m`
z$(?4i1OI$;D2~d4zR?Ax@XM2%pBpr>2}_!4mBsuXm{8v<_9-c4M&c5|ZOyZJ`M{P*
z@AJhEdBxieL})`Os%rj3rT)s54IkfViV~i`R;k*JqAcepmITx^l~xi6D}%39lazy3
z*z?Yb<w!?`H*Evk&I&J|{G6&ng+BC!0EBC$81EZ$v^Qwl2S%o+KRFl*P>*4`R3h~*
zRpZDRqFml?QlagidJ>~du(#L_;N-CQBks?(03k%5nC3{v^aXd6q8xU*^*_%PLpWTx
zK(Q<_3cIghU$hW3&~=-gxn()rkJIB?NaPmFH7_BRAuRLF`7FKDkKQPENVq)G-d2j+
zASg-!l_-odTAWev3VZbZwsag=+n&k)cgt#KaM2@FqGT)sCi$L0FlQWq+c&0%<<_y1
z46+AHnN7)vVeWB*YVQ`C+*8;V|6SK#@9)eLJJ#m=U$eubXI`Xz&(_G!?4TdtijKC4
z@QJt9bBnjpsJ|;d_BDYJGBTS_)3>+JUjEH}bXR;y9v3`1#{E5!Yk?+RCd!I9FGGN~
z<P)S>@sCWcn+9y+LL8P0@H2Efsv$Ty|9r}8AaV9~vyG!C_=P-KQS{bCs2a!Oh!AMY
z50-;Tr*F17@c3XAL>{U6@8Q|)7@7`st?Qt~Pq@N8lbRvpIEK9p0)1Tf{y^z(f!>r@
zl<<zb929(uJZAW52Ai6Cq}%LOy{ttnWr!Z?Y4XnKlBYmZIM)G3e*$CGsuJz`#lI%<
z#AsKO^@Z-#;Qsu%hH=hQIha_G*N%E+B;93XX@9LtQj{1g-(F_Bayt8m>&TmLgJg)c
z=7}<2<~6LE`8DnV=<MucBkn*!wRUaIxJr%dn<)O&EFK4<DtuY<(^+w0ETg6@NIC-S
z9Zr5={)m8<8`Xe>A;*|db)&kHPIjwp)sEsWmwofO&|hW_dDuOt+;0pMnjW#n=2H7C
zqw4?PhJcH(Vd)LA-DH|sK0K4MoNB=cB};BIL38WOZM-L^s!7;B43#Y{+Ivi~>_z&&
zRq<_mdy#!qc9K${WtfF3nkX=@uxE-XMmUFx8U&1>_*7Bta|gX1wGK68YS;`)r1I&c
zKQd<mzb<6-GUXZRO=3TPhl+(0gcjP_F&V6Z!i5@7^irWKOcF3j3}@JU7sk31L_qYO
z3Hj+<Wr)r3n+8ZVNIbZ*$K~qyNjYq8>J=H0uDtdGGvf!v5#Y+x>J_B2wZ$u`z<3%c
zP$Lp3`H=YR%wGm3mf5C&Ob#VTaXVdm|9@6(T`%iIFaDOWSjZA`N&tQ^q9z~&+r@mq
zSOLQfO$}-4pBqmct_xwpLfgjr{?vjA|B}a9nYCH-F$ZlY!FZRwWF&q-MsI2_gDjwW
z;M*W5wu%auBiH6oU5<EV-AH<n1uQ)5whp)mdb_(bdwZWcfWVDNJ`?;TivQ%A7^vz~
z|LnfsCBg%;(b*%vj9gI4iO&k=T{PtaSvwHC&3N1;7WP*^g=!_g?Y_`yd@#qf`Gbfq
zK;5whLq^A9qiSG((Yy)|a?pj~f+K=j7L~-G{Vrn1r(Y8C2#=YbsO|&5o`hxiyx;QL
zg9p(@YaUK`-$W=9&z``Dl~f<>>2oadbK|?xAN5B0VVGJ8W**dTevfs;QL%|ATJ))z
zajHw2zL?p)e7LnT#CDS(e^}1}8<r8Sxzc7u6VLRGh_<=P2dYNVw!_6Bg*0bUX57Il
z-VR4oQ<<^Ciz##@Isks8aqqvZ7xw!OD#HDXgWOe-B!7QA5In8pjwZeEmPGxEs$)cR
zwv7=5S8+4_3H<sJ)rIw?+GK{cuUpz9P4Dx0V`7p9q{nV4`SHba_8$?bRuO@Z_ERhj
z^|Sa~^?#rG6)S8@B?uCfgH4%Ae@{87Pk~W?{#$#wB>%A<4p_NNa-P<SgSY}&){sCk
zd}*_8g)k$P^@E}o=!0LvkDMF+TwudOnFTedf>>l&k~Wk5ss360mv(e>$@Sr(+%9Lj
zRLE-t<k8Ri`rmGu#A}3Az|Pj$9PTibNvp8!&n0@IC1#HYoctBupY=`nihO@BhcOco
z$EseZlE|F`rmWL#9fbFm3!|O?1hdiK%Nf2=Z7$V&6~-?(#F(v<)qY5jBW_!!g!IY|
z^$7{sk(Om7ZWALlecbVon*;BBfBji+Y;?thE)688QRgj92fX;kUHP~ONBrRZq91q4
z4cmKihJ#;%hthB=ecp=@3H0u<>RJk&JTcUEP7pA#6O*SXQi~I|ZN8^+l7X7)&(@7%
zIe;-fr_puua8zIkfFiP$z7Fy>BO%~0+vm0kp;0$VKO=`yq3sTu91n(%7WNy8D^$(G
zn>ro&K^Fl(wk2vWw1<Bh7KEzHkI}AlZvP`0`asovPwlc|O+z_6v$@Olp@g<nP`4=|
zfx)FIOD!n16n8w+h1vT!R2Y&a0oz$H_+7<JWbN{SI5~(Z_IwZzw|Y4VI@zpP5lBBK
zpW!=vx_MmCnMn@q-T3Gle*A!KW&?K061zFT;Y4l!x+`IiMZUP!@SW#~Jr{HXHE-L&
zf76iGU2mI4C;}w|Y>E^RvtIX7hzFWP{hSYGEa{?)Y4?iMJUA9l4DE4ArIZmbzm1Z^
zc*46BK|&Vve*C5)b*dJ(Ts_+UmLRxHL_lADO@vTP=63dzp=t3T1U9;Z^=$Nmw&}F7
zcyX)u3^%rF7Aa39r?>N>gS$;xQ1c5r+36jhk9O>C@0|Yvx4hES6864K{?no%PsZxD
zzldtFe^E7_zzBON_i%ssFSzyqK{&c0r?MD~CYD`HP+I(7uzcWI@siw4+-(%!uJ^0A
z;@b3dd99&Yw3=yI0<0}mXBxhF5{-SwjK+z|FT1L0c4Cwpe)8-$u&uW2m~;wGyPn5N
zL-GE5*;omn(RPh1iI8E#Q$6{0vvp<es!L9RsBA+`e149f0d<mf9}ALu))^xAd)@gj
z$HCbDadhtSO#bg5&x|=U$0VjX<y;gw50fDjMTn^slK40uX16)zP|b8e#gIdiR&pqZ
zp;RkG4mq<NLez$_IsfkO@6SEVgYCZG_xpNZ*X#9sDQ;)LcOI#^rQ@;^KZRXMHXW2`
zVzjRDEnK<%TPP!7E)i7_WRY=2OU*GoLVnR?PwOevOaQ_4WMlTgd+cpv{&c_RSB>5$
zu{nO<ip@4*K0vRl-bL|>wpu56CQ5!F+Nq${`p<`+?0xzGq5}YhSDygSlx0@GUPr#u
zlJDQbwAPAUj7#%B)xu_`U^d@>^Zmt!y`A-?QV-`A$bZ!)KJ6D`Ptm>hQxP#{_?q$^
z2AmMC0ig!KX%(DL{2M*GE``0ieO_W*z_jkk;Uld8G@)!u_kP-b=hx`iB}%*`t{nPY
zf9s(=V_tZGThCIXvn2J#Puqq4!_RX!&*d7zzh&?IGTN<!UU0Ao6k!5~Sd}s(VivnC
z_nZq^(Bm1{o}%sjtpnPqZLzQet!ICI-}<mpFt8;9o9#+iLG8?ESv=3&oQ6d`;`77c
zSCXWG-7?<M&$+4{-PGD6xpW8EjbWQ2D-Wnf2leogkDSh4vU6$j;KtVjYbGn5a;EFp
zBcN^m_sXhc%FpT(Qxd=7!!6kSTHe$v3~Xv_F{pME!Dg`>rp`wy_D6H$JJ^4K``;5R
zbl`pOsJSi9(47qIKB31F%HmM4+0P82Q$YJUWzg68F-2}B4e@kKZUH)V@QC2-XNcj}
z-QIR|idKu945uU{#Yl8_y{g#1No>7E-E#7?BlpY7C(^@bKxJ8H$CDya-&Wb{m%(~v
zGNfF9>Ln21Y+OeFC_^}OPg8h`!SPZZ=$WW#10M+Raj#-c$JS~>Su|ERO7Ylnz-!iO
z_TL7dkp+7&o@<dF+|nmmjgvc~O#Y&hlQ0KrA`4>Cz6v!o*2bcexdqxU-Z3C05BCyP
z8q;E58tZ+gbo-k&3BIU*?4u1-YX$DY!UH69>xoDUu1nAxr_kj+-^D!4bA1=CZBJ-I
zj5n1gZ|mwcD<Jg}Uee*@d_^D(_b`bzZziaO;e7n_V0P{vW_?f|*=ehQzQ-6=g=~IP
z8d#n=gPhnlwvqnZv9)$mRH=4hrX88@bj;CSA>@X}@(;u4ob;ZK!~1K?=`jVkW{AVZ
zhWbFYdNNu1m=9U(VuR0)ys1x@#LprG)<|o0ZKEgsNTOs`U*UrOk1-&T$ceX9CX}Wd
zZ{S-qI9l8!fz}h$BR;4{J;y+4=z|D!0<9HMiSu!gAY>eySzjhd-92V$GrJbWOt0Mf
zmxJtIl0;fC+G9F`oI#y}FC2Sc?@FO!^cCu4@=7{e>23x3;}cEUQU|T<KpjXw30<T#
zmwu9Ssv5HfA|4iia&5>-60-9PfPO_xHJ`5fcKW7oT?^C=o29s#+;k2U4$8i<&P9{{
zXmZep>~dwU!XT5nq$QJI_ayZKqU>@$>?0<+CvQl2DB9ZK<g@s?u1Id;$~Q4@Vnow0
z>n(K(T)T>&xSQ;u$~N{9`eo;vznm%EqV>jnJiDOZ|K$f@qe$I7JsL;Z#u#z5G)T*T
zjYzy8{q~oc3Q{8Kibkh@1@Zimh6~E^f*w%4e48|e(P}!`Qb(@^u?Q!B(&SO!df8&A
zd{8-+Z&)wm_mh_P57aN+;8nH|@4hQf=@)BGb^j__8n-{DWG-V1MXD3qTC#o8(p5Zu
z-=4}aiU3~U3CqK8uq;-b3qGbrYfS5x?`K$m&`0xJ)NC^Qex)WW+ifmf39V?L5{6^V
z31@BKsU)nDz9N0>;AY(k$OrP*Nd{7{8JLG7Ht`*{rZ)BLtPC7A{F71<e7-M1x|zmX
z?Fucurqg>hZuU_7<oC#4lj<jvl{X!Sp9$*$7i}7Q3Jz1qV(?Y*1#hKIoXi~Q+Pe=T
zi`}-3;NAN*4!Zn#yn&6)!qIItjwH%og0Nbp^Qkir`24mqZccr2nVr8X{tR7{j93B3
z0sj8!J18X`7jAl`>w4q^r2Ngcq3$o_)RMnTQ_it+Tr^lt$*e=lpAOr~LEAZ-#|+><
zP-SgXLW9-3sQq2>CvyikzD#a2UpUrnS^A;cV`~*IM{+bN4Y9cP3;z0&!#<(j$49F#
z=g^Cu2^m78HGDxxP0!Mc!WDNfJ-hUI<I=9@5Z`@$1+GPxt2~06@U41*J$Y{x-=Lq8
z3C*+cY@L(W5JQri?H<qh8@Y&&1v$Xk8?Vg641H#Rt@^(~RGWL?LaIZ3xj_QmVcOha
zTm$p>V9i4JL+;1X+7?qiUb9JR9Qd$_RGdXtcDV+Kq7LRNL0)hZc}k9I_YvbFp84DB
z?*%nZ%Fvq1kEk7rL15M&)+6OhSC{ZJLyf6q2FlRSVCz&7-=|T}ab`UNSZR9k$zkdd
z<Z0@ooF0dGsFR(;yOO{}a5nc=5mG@d`+Ic|2`kCe_GT)b{G)5&$zMdQcWIv)_mk@N
zPotzr6LcOu8|3SAfhym;e=*I_U4Nma`ooFomAJY#f6ujkNLm+9Tt!P_M4<eQTpQz<
z32d)`)72@~?=8Z~1>HR-0D1*s<_8k1dVSd&DEyoPo=#f)1u|`I^NRE)nOv<?XdKwB
zKF|sfvx|{d|K%5oH<Jr}KP1v%8&=Lk{;79$KEmBjh?1u?A^JZ~>q_17guh;j2Qd-r
zVJPQiDOHR)L{_Xt<h63#gxD@aYU*&Lm;z-}_N7x5tA1=l0=R{~*YgxXSrNdM8%~+b
zz+Wr-cnX_FTKM*?<J;v{TqJh<LHdD0vDM*(lwbb4cZZ^R3Kc(q1-%eEiH}1<w~3le
zC&tI01y+yuv0@vDoO{#A&LP806#4z}uteHN+Qj@Ha|r1bsiKERzNhX0SzTgw{b=uy
z_iW<6BXk+)KZmE6Nw@oND-weq2%byv|M%h00N?vldSi-!{{Vu9R;Mab%EBybtzL+Y
zKXx%h-If<FCyIDGnrnKL_HHhI1pii>>Cka&XhKAqfGk@7Fs&;8*Ot)k@v8?k{ZB7+
zYv-Uo_vEdN2@I<ssbe`97`B2BFSkkYzl8VfRsyJB+Vc2++ocPgTGw}E8g;Rf<wDCv
zuW0GpzRtg{^FIn|TeXox>dk78qx-&(M3pyRb`Pm$&<~w|t|+sqIQE~xI4qcOOWmbK
zDBt-U)MX)<jR+8N`21?WUdmUYjFxEnN(!?@32k$&`H+gqOsDW7G<j*^F$wm~=Y9T5
zx(&ENwKAkc{e)h|B>e4E5~E<%QBMQQ{)AT|xa6Z$7I<`P+LEb#pZ&LSgP)XMQ*yhs
zhL7a?F>>s=rXI<SFK9+j&C%h^tZJdn737!bm*AG6okU9=pSj>GvaOV@-(X+X&#_WR
zw;8?@7`Q^bkWASYQ_sHJs4;S|N+f+n7qA0&tHySjmOnhPEuQQN9klBB-Kt&G6ucNa
zs&mn|3N|Hc>1R#t{PVz}g*(D12D-W#RTRU#2mhx9Pq$khM!k(Mu-TzbRuW!mL2^PW
zMBZzfBv92xhHD(?m1_{}W4>AbzdO(Bbb5o`Qoa+@DOV0R@zOp{o!s7gjr_s}+Ce9X
zk63ZTgvhM#@<#Hhty_o%Sd%4o;!S*!JQC9LQ(~5}>_a-!D`A5xn6x0Q^~=OaM_DLF
zNA0(LiG)P_l_n^ViHRE#AE1oc46cYyC9ltv^$Z^6PZ^!;?89p9UFS(cTTe>m{GOzP
zRj}Qlv!ApFeNc*v*kyNY*AbXa619OYrz$l;3l-JU9WxR|=lt@tv}%4mcq-)Tmlola
zeFa(cu`mo}acb^a!&%x?{IR#J0fC0KO5s%r=T3T-d9~$l8mSw)T@*a!?=)-W_-!%6
zDO5MR0Kb*^u4_@obfI;Kd6))uT$ib5IGynOVO9Ur*{~672ttcEEw+*giV}K1@dI`!
z(7T|Fd{Dja0GnLYg}wd(<iEh--=mh9xrP`jcCVd(Z;t48!nJ;v5|N&<{|!o>plEqE
ztR_F%a2Kr3J*o&h3S3AY{=!)Q@6D_r&&yujoi^_0yIh7vZo;?9b)A@O{D6b)Y+}^5
z%|eM*So{2L8T9=xtf!)=!b;GtO3nprNz87p&cJwrQ4)KpbEM}DJTR0SFWWepd6+3}
zHr?|f_dmu1z9{DHiakvW8fo*}x|z%ucflA-i^n@!eK@sCBtT^AyE|-r3>j*OZ!Gww
zUCZ-h-+4%Mr*DH>9b_5XV?~Q%TfZY&`CNK?OXn@v>%#DR&Yp=tg`slq)UD@Y{Y_k3
zuIXo+&v{Z9FOJ_j!^Gbz9d6?52n^XFtm9{3As}e*j<!piNo;@O@IS`~`3Y|{+Z<(X
zy?r7PQ+7+Whi{`Brgz6CKseX!2+U2k0c5dy>AwC`^%$`-aRuEruULa)x%l<}$^CQ$
zrU^vee<N55Wqit~<Qwc1>b1DL`~5g{w)~S4{zIG4$vCsw)TNd^)tVN6OQ`CV%$<b?
zC-Ru;F1Yek%(MH0jl_EyD6%UYS8u)(48DmRvd50#f#jPJdk<m8Uatm!J`mf1l=vPf
z^yQ13|D6ekizI#%<<E0wD#RF*B{=q)23UDI=F)nCF6oyqto5O5C{l1j<-nS{Gycn$
z+Fx4fG28R-Qz}HxTlxN#<L;!TnaBFSf7c0<ezn9RJEQGEPdr?3B9T<{9-PnlKYqiT
z{f%PHYZ_@*YwJaLD?b%*UU9<zT(^7X0h7|*W|{M#YiL_Gp820Xi7)!~g9Btk;3#C*
z7O?{I_EXFrLn4gC@+a5U_;&h+`jhs0dTcxW`z+TPCrH|B={;F{YCfxF!_2XA_Jd}E
z`=o`jtH7kmW;*4RSO)%kGA7AOwG85KiCDQXN{k6fp)ucHw|T@MAFWCF{*+ec{qPC9
z7<BKcOOH&($q<~TkVEr3UFQMFczdOCT#_EHL69^T{?x{Rwsc-;ynRu)1qM0(AIW;Y
z7C&i)@0g!9g?J|NEj&UNvC9SCK2liuLCpngXTT<tJ&<)G1l($PQuzjQ({Cz&nP;S5
zVmy4T7uNh&tEQQK5p0dV$-kDz!9M2xTkz6i|EAH-i(DL%I@F;Qdi9SDPI#c}c)SI8
z*k5^R+YDakxV`UBlZc&NLm3|#lTS6qJv=B+sKJT6`E{|2`ecf+48bEz-=DUePodn-
z#TYr0k1N)+?S1WArnApdzGbwaKnCv4d*v;IuRa{Jvi`ALXJwc_6twIF<u)qAPy&-t
z!R9qv7T2YI=0a6qI)LZWUVFg>>y-jHT2a+J=xGu0XG}t2uM+j=jh~XgCSR!&?kP`S
zeT5wKSJ{`?valdOQL|dD{=Kk9Lgx_mGbG<R+(^sFLmq3e^4sjvgs<ie;!u<qJX?9)
zo{J){PDq_sPCoxiN`p&*%Q>_f5#xR%`JWguvBz@sIWaXe`!3Q)Zs2n!>Zbebnydzq
zGn#9Qa;TF3O$X6P6F(;{$DYei<@T$<G>;AjQm)Bm&piXx)4%kd=$w!F<I)x~V!u{*
zbxoGRl6^jw0R4VR6VZ~ZstOANTA*77&#S<l=+ko=k7%&-h<|DZN_9<%KEGtNg_9wj
zW~kj?KVmJq5zvEJz_mcrGg;!6;bMe|F^3N!Ps|y1voO-24~QM`5nPyUF_^lF@=K~7
zSz5_ALn5l5zmnxxh|e*D2Q)X~Oa?3Za0W-`HIlR;azmuWI`}0925hQA>dOzb^qVU}
zR0er=zR3sY!v)6g&YW!Hs%aNd$ievd5Qjl4SWOsxzo_)!VxodY9}i-P)@$ajoXkSx
z@4Tis>h2<*V<d}BZ<9p#$lQ;XXvdEKf@9@OZ0)i&PulZTp`fPil<wyicPXx!Q!CF<
z1y!`vaAQBW?WX?#jQVY?#>YW2`JCEQ`ng=jgmReTH~l}N&Dduy_`#)DkXBZm0=LWJ
z4vT9q`G>{p7ZoRt>_hd1(sR4ix(L~KkAK_%ZwT{KWBRJSCD^bO?2@eRLWVw1x?>C7
z#MIJ0RJZFv`zyfG?Bcn{YC1JXhOU;2>QQ`#zd7v7I3?}piX4S012GlivwP4Uf$%vS
zgBa=S|C`8%oPrRjr;xDKPuI79_!l-_{wlz&yeS~D{5ciBFJ*qpL-WJ<HTY^#7UzWu
zNo23Uc1PjiF3+1@Dbd*jIv!ek-7QY)%$w>c-bwqrh1w@+x4i>Dj^Uwu#G68#J5JI&
z<h~;_;Zk{<N3He>fD7hWEHdoZBaf3wr7;Dk=-LA_^W2#@PVg~1@L6Qs^4_zgU|0{c
zIw-*5b+5`sw0b5VZd1`0jKX<kd)1C^JXAv>{1Pcqfst|0pc`L@Q?MHSe1(m!uLl0y
z_rpw4fk7pDx7Ti>u5Zi82C676;opI`6*d9^lP0H`am%>?PDt7M(f5132`Im<I8_pl
z0!#k#y|!_60#1Uk(@~)3C%Zh*ZZ*lj1$SP4r57zy6ujKLxLcG~qaV=~3-zP$4GT7D
zO^Mpf&d1Akd8;iPX}X0qRDn#W)bWcy-Lj?gd!_{}-%8*}G|Qdm>Y3o?lthR4CN#fN
zV?q-)6?_0BnEfze@j5EHmAv0oyr@G=%%!IYs5w9<7y0GMi9hMkyno0r#Yf?gPm`is
z-cBhRt|Ts#S<Jtf*wAUNMcru=&CKjn9Fi6G#8z&Sv1Ci{ZkY`QoiI`25%SU_R0EK)
z^PBdyHyNXBrNx>RYFzy)^=JOO!y`yd+tw*r08|R2CSK8C@E%tIztjSIxAsa|zlVG+
zf9)T2R3aAkDn7jIiuF9O|D12uC_VAI8pNPK@CW-We37b4;G64`Z@G!In94;#jj#6f
zU%vhVlX|Gm`t2UNXfk#Wawb`FQSl)_=}uG|aC9n7Ox?-iZZ5KT>}m>Bwq@>#N0>?i
z1Ro~y*e<wo=UKV>el2@pY>9ju)dfk5$`Dp?C)-$jgaa6Wbxp?E`>UE+lUBvJOwfAx
zkyQnL3bbN5yNbK{Uu#^(i5DANbs*S|d=qu?$4|woxglvkpWV@o1SPb{0;Y9?dvX7n
zEDqWG(ZMfK1+>bqMB|Snsj^ytUyi7vx3mW>3^kq1w(*N3Uh0^x$h+j$cb0;j>Gc-*
z`P@STIAB3ytM7oxl7k#F1#5hEXv41UgBJ!-?Iof}X|>wKU<vpa$bRQqG0?*YejQ_e
z_<=DzRYafZ6qI_g-TgKrPs!>+H`XD0W~WYo-7|D3hM#)+4Ur<>EAq8fp*?voa`hiz
z+|I=3zSD%oNjH;mmbQ&5Nfon?M5On%`=g~KJ!*M#F<LO&gP9v59uL{VmJmI~%+PIU
z>*`;N=bt53LUJp#U{%1+qJ5m>8@`y|ZVw^@=AL5+YGQ<4={i#ukjA*9Z56MUGjX)0
z<*?Ldd!QttL2PE;#MELsIA5wCVsuF2ycH#{T~0?lLHHGh{pcMZTJxtIny2-|elx!1
z%Hm|W#z{Vn4)o)p8|d;@XK>?IR$S!{taSM7E|YnQEUZi#4=`fa|6aR=&AihiF0~mO
zvXp5S|0JqK$0XE#i@7-qru)E0sO1jO#(DGU1ZSj0HcTB&7Hst4d}+-U(1BgxVc}Pc
zBXj+9U>?Bs;*k6CSB857`{cvMPd^_{>v579nPk+|Jdn}d^l3fRYWudMHTM~%MR1pl
zcp5Apd@7vsMR-RcICdW&W~}%X+j<E26aL*i#2b3$G*b9P@zZudG0a!(w@^zzXin04
zc?mqqUlPdS(mS5Z6BnZM#H|7)lUa6cO$vj&Us#lDstY!qGrt+IR>VpCkI)%ec;}SE
zqZh=_)Zn4Dk8>0icKC)5s+<H_;sci9M*|b3nVBu1Y}bITM9+`E%SV`9NO*zF-|*5Y
zy8Yl%`*jUjjV2odOsUl_1YIOrje)f~s7stEcB${lmLzzdA|dc&u$<<B1x>`%YbR{#
z`suJfd=>B|z8=9<RXu><#M=b8U!d9hlwBAds@}w8Vqi^gYPl5J`c%gcmRSxBtlxpe
zBq=1g4J#E52DlQOjzA8%FLv6_xeG`144!Q_@vvs^-aw~O?ro{z)qXMwxT-hp+m$Qx
zZ&lG4_D~vo<ca4Ynl+qY{+*bZdMSKmeT<m+9dKVu!Ps{n8#(8f98tNMh(GIaAo!X2
zIT0w>_&2`DbD>dXm&`MIS{t@=q;3bL`twl2>1`$Rr3=ti3WCeRruvxL9GtB5_J#xf
z<jY{rJtJiQ{2=H{JpLz$7XsaiG>7Dr*w3$p{EHM`UHW+?g_g|M+`n<S6TQH%EZU5|
zkt=f?CMA|>z&cj9(_R^}xiww_Ktgg&kHd5nnuF!zIQ0r`g-cvpH(Sq*n$Mh0!3T<R
z??_Tosn~|}<1i=}a9Q{p6tVaGzmJSc{+%nys<_OEY}_*uCO#u^&_JWBPTYvfJ)u-I
zaqlNe^&an9Iog2ZeR}(UlO-NumyT#*u&G8~IWMVrt<7tpbmZ=&WtFV_|32D#{WAB>
zhGW(%4=s_iqWN6q!sd`x|1?BQ*Y%X|2}r5tG<}`nkQ(4Xd#Ay*Z=8l>5?uswLfq)d
za4vq6wWN#RyReB<pw~dFef}3^sA!y=;~@}m30b7%c&DWY@)mTY`F3{Mvp>A--3)V-
zy~l)BqsDKVhYMhC=XKnCKrkFjx(JGFWG$7qGdpZ?6*}+**HymRR>#VyP<!K%b&3h8
zb^bb*e#zaZ*eY#SK(2IH05N;uRj4Jh`XZ|F<vY~UrIr6(j-hBqOvg~>%ZCAK)KZ>S
z(4-!m&Y>0|d#%H&52NHR8U6D2wQAy9yr3D~E8$+=<ah14T`aUDv!*_9BKNmGOdEL3
zyPjD8=8vf9Se0YCn<-@9R^a$cC+E<i&X>Fy%E=<83soUIUGwpnSMo)SQVc7y3CUBt
zF0T>r1Id93vO46Cnm2re{aBUPi3x~SHVLv_a9-g<$3|%>+D4b&k%KcQG7~?w&cH}J
z1(4Y<3@F%x7S>uRXr!E%A@GYK0T{~%e(Ek#?COXS=iUyIf7LqbyDuz6UF5ZGIQJqF
z0{>9Zz+fv|ty@sW#CQ6{7rP!#n-pbgy?zmGL`NjFJn@bE2BV&UdoBqKz9*va(~p|d
zCcM--z?`6&>q9)oQ{Zc?r|z&t)47{B8n|L7n}X9#7z&DVk!+~(@;^W}%q*&6VgIuh
zX9Dm|c#3Kq=6|rL1~K1Js{6*smN0xdX7|89+69ZZaW?ifEIe#kamvWr{F6EXbR2vS
zMS>$oieWY&SaG7)KqN9pT9)cxNlUpiZhhe`(t_50YA!iurc!dqv<@}#s}j7eDsD0V
zk);!eBBbf1WE+hI0EAwjzd;iD`HC-~02yjRHc$*X3$4+2+Ex^a8BHDjN0JxXukfdp
z<eU~U{Jbc|ItMo%D`>u7q5B+rDj_+S&a$p3-A55EcKgg1WJ=sx#S(fHz&-FW_KdIW
z?sVb}J2*3PZyFlEEn`UIU^D5#yJZ4WCKh%MtusJiQV1}UE4=_~jVOw9g!)bECWsNX
z=7gxV*fbwoB!cz=OQrv}_i#2IW)8v6VJqntoQFIsbgZMV6zHO;k}5$W>8`tj{O3Ml
zQ)P#ih9FOKuntu0UA|9pKy~u1%^>i`rPwl0@5B{L{rwm{2R4Lv*HZcZd#kJ5OKHtM
zqN4=l3(d3OTL$0acCSFam+#vwvfGuI4p(7KD_MRmfI7kdmjh3YEI2dUr+4;pVP>}q
zN0u~?{WlBmHK1DM=V3=`MZbLxF`-xPe}ftKE!9I?C9A%f%RXQRp9sGQjg!w%er;tD
zD95}ROpx08vExfZV5k(=6N^g^u<MaEG5{zR`~H-VbC`-hV!M_G@&CvB!kVf_*m*k<
zs<@hPHSuu>eBs4^S9!foN5RL#4Gms*gqTvve;U;Z)v>P0LX;=ZS7%l8V5+N@htGnX
z{U=99)gW~VhHA`_iA%WX1o(@R*4Q?FYMz6lRnXbze=eoqor@n&JG_OV&7Xg5%>$<5
zHCg}LkxsP8GmXZ#*5n_fkJ`t19bkM5mesoxQWdK2U3r8BsG_mAyr_bkO7yAse_?Uy
zeCMslqXJvk4>YbmR7IC=+8HB<zPxDI%VNi7p(?=m3NFcfb2SIQP1n_Q;%YjM%4;uR
z!-{9YLlV-J8S60}t?zAdr5rfGP=|9>?xZ(>oPz5Gw2kBTq@MY_zlN)mh2R?!GBE40
z$;i9-z(2Loeayie<k*)b#7a)c?Wf?OyL8$g&x+R@c-G0{o9t18xaOgYJg?n;jPpsP
z3A`57I71mFtlHW%QmVxWoTq$#L4SLnkri}v76)2?kx^+wC142xeAfuWRhRH}oQxhg
z&0JEg`?f8Le%$cFrm=O&hPRmqj)<2r@>!B3ykc)lk=i6WGW=pYF|%t%-%ZMA%@UEm
zhMDK5{<DXc{|1Fcu_|G{?w?qj)L3FnwQI^hBlRDzr>^)++Jara>zOKBFEaqhlv~lv
zE6n=H6J$|jCT<dtqh9TvOxAJPDdi{4RExrCN?<1foUzxk*_nP*x}K%+Sy;ah!P4=p
zjU>#fK{bt8r;8lhK8r50{{_!}rrztg&FinY(!w3x_i@wz{_p8$bI4x#kP7HYZ{&|?
ztc0C;@&&&*i#{i#ww8i0R_qJpnE!|L!;HIP1dncF8<6k7^Cur&!uHBUoqR`Q#!e!S
z-38l5AZspa2G1=)-yd*lt&L)oN#~otork~cM&sN;0_e?FmKY*@qLPPv7X;YDke`N9
zGuxVz07z?v<2#^qBd({%Nj{F8A%v)sHi~fC7LCX~tT+Cha<reU1;^U0r8Bw}^}r~l
zMBs2YSKQX$!^uQG)9I_X!DsvT!xfAjBm_^jMUBw{+x~UCr0keT6*punnOW+X=j^BV
zJe3Koy%n4E`Qh3I{y2%d;fzi@J3OWXjFwW)&e$q8e*xOr{8|RwVv;-)lx<3r(cdED
zTFYpwU?Dc6nT$5?EeUrhDzlqf&CfytgI|_x9<%RTOz+Mj&$g(pa(~U;PK!U{x=~Jp
zg*2rsFFD&3y-mlkWrL6({po*^N*~GZhXQGgZ1LiyM|=@VBPthqr)qz-zmQ(Ja0fZ$
zKfatJgL#u9%K3z*PwJFyU3^Ysw$%BQkb7!lpv!+<bQXr@PCh^2(n7pA%l)JhI15}P
zEmdUw?Snjh-<L@_C^bUgRR6n$nbk4dya>*SBWcL)FBj28f;lwi@gds);>P04LF8E_
z?J(3o$;a~=VV0h5BYGIzZGRG?*$7UKR9+=)C_^fIpp#L;L+qK+lh>Kz>6}A7wm6~{
zX}ZZFGlF^OCtpjd4K2h&;ta4RO;~m$2}gxGo~?WnBx3%#b?UWcOuf*5vLq3cH`n{V
z2J!p$XC{_=cEbFdq@N$0xR51mlA<vQnnd)FbP}m?M(ukR3f5?gQ`p1v=dkh@!w=a=
z7o=`+o0Zys`0ylX(BBipYnhTnZ@wK>H@FaA;@bCBPkg1o4J1U#*8yI9=CG@+k6QOe
zJd3s~K0)h8mJXoFgnO>U#QGmsQtNKPw|q@O63x2A>iqmkq0~r+AM>uJ)6e#iU+4KV
z{KzZnkdxlTHt)W|elA*Cd{dbqNYei5f7fC<DAYz{{GhgBx?#O(o8FDsB%DNXq$d34
zsGE38zYA)kz7H2g2b8FpKG5H<v$CZXAyKcP;^Wu1_KTWo1;C)Kht%pQFq_7>9WnI>
zrnj<wxg;d{Ma@KQwQr;{$2T&r1mvuMPeiKRBMkT0nGXLa%L<A^-hP`erAaf}k^dB)
z_2f<J+6_<wh!aCTx3ur|m`h~PE5x|EU#UG`kowAUP?=f)#?lz7YF69TYI**(@lO4W
z4jt{g=5KcfkR$}H!AH@;2YO@9MtZom<uFrt<4~YY)lDTuZsi;U(^(@{98P6ai5lfG
z+ourBzq9B4LB_4(B$=q(*IqR?RunyI5Kz#v+cvLl<zEbAR+>+%xG=s(*GYY)yZ${t
z0qTR_$EN@yaf3_j*{N~9-u@6`9@wmUV?7?*!G)iLQ+o2pl&+<}{L|FtaLkXeDJ8!N
zlkRN8<(?~H98vE>HOZ%j<)IZ}S21NpY8*u`2k1t7_KFahbaLHxC|}t~9Ae3ISWdtC
zW_RCCo@IpV`aN6jmXV~)X8Heeza00FHov5d21Ly2&YKLIm~BnUBEeY#qT59B2x1n5
zc&P4f@lMhCMuwhboao)PJw@gIc=$NMJsSOu>o8pPbH+_D^Dv2e9|>m#!LhMO7Zl~t
ziF0nVYwxQ;yT~&oYwzcn683#TKgT<!ZRPT^#-%;QW3trwlN~U<fZLJ_XbxY8`RVsS
z=+!%iz+aX6BZKChJ$;8)ejRN*t9EZA+4RAWG;oGRxF?R3C|zCPs*ycivGbz16CQP8
z9~6c~WXfis%9j^9hgy1cPxMlzm{$7^t;a66CR`?XucmUx(PUor{8xw-qITBNM0a$s
zPw9x=X+q?-OKP1oA%SV<b^9bnE&|Ba^+p=CZ^m+8r|-@R%+Ghqf8WS0ws=oViT~Bv
zyBoeIYIHykRkisIy8BmiDEj6TtYdca-sAR<?1fM5=Fh*dX;SZf4XFl0%>18$j9L-)
zhar5Cd&1pm3+y%u*dB#!vx4Nn7tYylnRagNlcdWK6RK$MC+c3i8GGFZ@b}6D%SU~v
z#h|w`;SRZY>y|Px@%J(7<TGoVhhh$R(8~%hTX#jKasT|X<b($9u3q*R6043;|BPpJ
z@$GK@b}&v=OFw4e;79npZY<m`QOesCt-7$Grq}+89GfJ)+qm5JXO?H9N2_<@B6Mq~
zO9q|Bc2ycb_*Q!K$a=B~5Y#7hoHV!1(=}3b=#rl`>sq~_2XBuzw5h>;Y9-YCHaIA7
zG6GTR$xn9^*+CmiO^_}rjSd+%4&k;Pn6Q({_3yKkC?_7VvK?2p(4?GsKHa}sD1mKb
zj1JAcDbzZX`^sp)-ra|4eIFFqe$gPf)%wv6U{gV^P;byQoCaL<s!P&EL6w(ZNK9K)
z<;-b76v_t;&L2+ief0;f^?Fb(AiFU^abvHskaNGzI>~?RvX#P$h~O2P*JvXk2u1`&
zX}uPWdRg>^!QuA0{zY(O(FqhGDt~-6TA3XAuIvl<S1z2DN&n=?Cdy#q=jMvZIVdaJ
zTO4x{&eU4&Ljjq8t=1EXZ6ZI6$|_X9szKJzG+m$idb-p~<nt?wbjQE;8$GWi{^fYf
z#CSzd(0`0eT`%dLK*l&lKg*lEC87MXRHWr^lw!{<Hr>Q5{cPIoKZ{=NhFkvoPQ(Y1
z`_TD1F)qCXDbUX<FKT1ZG3rzrFa;*oeZMc_S4)mq<VOQLjRivbMmBo1?E9BXHkEw2
zGZUv=z-V5zkYN*~n4|G#(&52d+~jONj)9YUwrIREX=l?YRqM6u^ZHBBmrq@bVI&Oi
zUCsbc<8b9W2^CR#K|PrFu#vpn)-MteRpH(H&WYmtEGuzpF5*Ry$*z^b3%-jxhS2A-
zIW~gk7vMlTa=dqU){`W<Ox9Fs+b@*kjv2YN{My*i1!@mnjStyTWOK@Ao8~jmoT@Mk
z?)}M-%Xdt1I}y#SoHQ*m{Me1~dpXg3q%U$UwUOv`C9$qty%x#6{L)J`K(4`sP4Ifc
zc(Bhmdt!3*i06OTTKd^HnGe#z4yi#NQXeZSg#?6O2r}gMdHqoTR)Qt|OolX_)eMVI
z%A5TAAUZytaAssoj>Z`?8L!Prr#5|40>S^gaZA>!0qgGvh%xa_P*Xu<Kf6_$a7#wB
zgT7{LQ_~vgj`=Zm9r@`?Ir>FIvJIZP0cxmncHp|v=tpd&>g;Dj5;iqNRkP{0K{lB~
zz<Ha;#E6VU|M*l0oQw3QrIE<FNWicUQE5!DxS3(Rm(pN;2aB5^v+3$2+Xd}`X_Zu3
zPiKd1^TbiT9mggsL7&@F1f3xy-AIb?=Mrf~-@)cpLkD8TCrK)9#W<CZmEzI3-+a2+
zXI~6C=s>l_otcK85<IRGq7SNi<^lgtlFb%ArAfFf{wp*GgC{-q(u7g-|K;4gSFE~1
z0I?wvHCKcuU3GJsg^%_v1UpnPS5dkmt@Cl(!JjrZU)sZIV08>uIqO^BGtFW|x0K8e
zHThsgW5(q5&ZG6J11R|G1$+Y-i|GU_<d^ve&5zcoR%c3Mu-!^j|0EjF!#)g`W!^+8
zKD@6b%_%%ce_DRAnc@nXY`sZ;!`IwO-rFI$C#}BKTD#m+jCXZ1r=0fVJVlUth8!~v
zgRLR(V*!V3nm-|P=vwKVOheMIrK2|Hr|!D+1qk2Rm_(AMpXDOSL+OYWsTR$;g%$;5
zwytW}f;MXb1b`2E0lq|z#(P_o=Wj7oQ@0CdSHiDSZm;0?fwhx1MKb9+{d{`nUxZp&
z>)#WnB;>~LdL#{yQx5~?-#|5!ce@<%*Y2#t?H^yl&_yPNn6*M*@Fe|!FTl!Z7kB5{
z+<0@e3l#Yd*()1Am$Y4Uf?+GZ7;5i_<A?|$d%Z6DC7*l<1Rdai5>z6O)ei`)k;c_J
z`)#VPYc~Aa2C+lEGvUdBmyln)b~K}zS!UB~259p*!QEt&Q|!g?EVMK&2u}V|7txa>
ze9>=UOG^9KT7fR;qn%dy@Iag-JnCx4Uwb#F?KXG-+saI|4Q~lm+5!6q@ft?kLZ+nY
z@H{j|^$~C;$GQFig%`y7xRvomq)8YlirU*1Sw4b9#x31VOjYYm6`689(r=}l9P(@c
zVh~qpK{h-JacxytF?zf|bH^7^yYMWN*n2PJ>rm+b*Lfle<$b=McWLv_R*@1BrAdW6
z-hxH<TO*F^lI#)YAv1Rgs}B!Vr7gVm0p0ltK=5D;D)WM@(B8<6EyBk6epm&yd{N&#
z_xweMyL5(m&@{Qt)pfm=25oxtn+LBfk9Z9fdUY}6y=&>#l{Lrd-F8fltL}9CVtRx{
zJRLqYbk?t}TX{Mt7H!U)PEV{q*U6JL2VOm?YGGEgCs4#c$XoEn2D4#5vGOB~1OeEU
z3WsVBz`>#MV(l&A#<onM-djQ2xX%?d-tF3`g~OUI?an@BgKhl(hN7h|X)90h*p(6_
zgLDhp?hhxVt;p^1y0tG|`UM2uuUupYWXWvY(~(6?NfV5!klZ^VNb?(YK7hc*f=dYs
zrI_1tXFL2oET)%U*zjTic=oLJec(c`Jqzv(U)WQ3Nih4XWD!#_H{WEOab^^Dt^bt<
zEMnNr_r{MK&X*vBbk+^ZUw*MZYfmq_df?wA5J+umQEgqG%6Dcu0aTWp_I>~c{JEMb
zr`xd%B@bX$JdBaE+TvWhMYlyBofGB8{Xm+7irg14H{#c5n|Lq)(OW<U*sX*7j3;Jc
zk}{;hz_$uyD2$yFR3QB{wa_4jAKf;ZpcMRzGACq^_R1(7y&=?p*`{cSM1#ocZVH&c
z@+oID)|JwerAEsxiK{DqjjtB_%8+W5qfFv!d4*Rm+muuE<De%7UnvQN6gxAY?y;uG
zsn`Jgy~liA-!4&LDXqW~0RN2yH`}dGzfEdPS2ypkXPs4Q@rmZ!mqxdNQ5JQ3DtTTy
zF7!ePu3%;^W_7o3157)e739oK2mM3)>b;=B>brD5O3ISB!{*(^!CsOJ?r-PTg#~3$
zl<@~S61^i9SoCH82D>E!ZSJ`gx6oZj*PlU)k5{z1C48y_XeqQ_iuYEbg5~ONT(Z^+
z?EVjt;>>fTIsLab2Ap^J-CmD{>^JqN6YY43;JPK!v6|*&=ahf_a6I+T4xD-D-fJ8=
z_2d<(LkN1oc`u!zq*T*>L$30J`N7KM^9qaHt^1hHo)WCme%mX=Cb#np?ZJOa5b?lS
z?*F>ase<X$=ZfE^(}-@hI!vDQME4s5f8%w`A%^_{y%ztqm>*>RK4drd!vOy~QBqF6
zk!c1oI%usS@boivA$NnDv||}ryd=mqT}!b}1L}2A{#f`b7+**oUV33UrGI2*O;Bcb
zL8e&Tv6Mh&WF70b^Gje2i39{<n?w8Y>WS5xWdI-8;F85PV3A;2klCygUhm{D?OCg-
z`^DCcd|i(uJwRnntaaD=`dDI*WM(O@ELoK$m3Oa}dI%hy4MM166qW2*tQ_5FcNBKc
z27YR+Z#v-?C(ipQePoMR*eJ>@0(QBdEG`LN(ZWt;cOSDp(Ct(&+41)JQAN`4pneem
zdOg0{P_>gxmjy!qAXX~;J`;eurNc_#5SR)_OD`|wL&o(&5+CJCB8!Cgh|Q3k_8vnC
zFpT_v;MBj{mYIh=3&s_B4c4Hdf4e%}-<evq!sfnr!FD9?#|dU-i)cdJp6~@kA?Yf4
zsbOAfzpJxQ1JQ1Ij<r5b;agpaz<7zrWs%!_EK8a9MXio@<{SmE9V|Rm&u8kTTLgDj
zUwUnv&^z|;3p?<V=Gl5Cz{M?=>GqPDyf7yqr}q`zmRtWJHLoz8A<*KMG8sWk)D13!
zdUiYTrk=#`toFGjiHsvH-p#umqKn2ZdhQi^y-zk|tV34f{jWVw1S@x-^QAaS#%pfv
zWFH6&sXaq;zLbhoulle9rrQ{S^=OE9v_6OFUQI9asoO4;MhezAx|J==-H=T6G9G1`
z2^%|UPjnZx{_SnsuCr!?;v*fuyNrWLz?jMDsS!|RphBGy)p!w=GT9SU!3QVY-3&3Y
z23rLJKfiqEuK62X2v9ZF;Mw1r(5Kd$vcZQNxE7zzv7htB9E-Y}y+qZV<5_;3uv<?{
zTzgyQ(k4>b&n}ZLmfvrTZu4w_Qfnuem0q`xez;}vdOj*SmPSOCfBFZnt(XwxSVu8J
zbC*TlJ|5ziy~&a=ZYE3czHD#5vmsFhws-W}MB}GnB!&c_HXN>|@SI?*Ze7YL71_){
z`X)SOJ*#j+EEn82NKlpO{pws;R18Yw*3jPxk(%-mxvu3DMHzx!vvKI`lu=*vugaix
zGhVbYt-uwaM_M$>{Thid6jwj19^+Xi75sqfzqmxH;7wce<;9IHpNetU&J#s6j^
zcZwK`EC;}3@L3j0g(ZAUe;v*=%V^tte`ND94h+#4&!g{*=N)nX^BJ*l!O%tc-(5Mq
zJG$LL674;Hd1l%!zkW8qPyQXM-_<BKQdE|Kx=!RnH{Rk8)t_)t6kL)1w$$7=vlwqa
zdPGV!#tibK(@TkC8={SOE9W}{YOPJz8*b4?o7RE<Rol(*hW7C=67L|rA@cQ|Tifm4
zx)6`cHA%m2;dTwcg>;|^4K>wk0K`brD?du*AuLxZJ0CZ*>Bws&@>C0vL(8Vkx18GM
zxxYXO4hhM+)N1^{N0mv(La-3XRcJMvo~a<lh~DQLa7#Y}sw`(?TB4WMV=QtoWeq@f
zV^8JO5o_s(w4V40wj7bGRSkjlP`k=?T{PW~yr$F#YNCIEJ;nZbV)XM*ora`;yH`M8
zYJ_+`b?BYWOL_~~>ttkNO68{-8H)|px-5L$$_m*Qw+F&HWz!EMHl+zqU&>pr>h0AT
zt;e239;*U@>lv9lbauyq-CH(*Vt<J3ZZac-%zm((=j&32Spil5<Q|dnbDtkzm-3zr
zRVl0_WJPIpFTs!#V=5*;=?Tu;cXOWgNFuJn39^{~m8ah3A!q&K6Cxbi;{K;5-<6Py
z2m-sPV9AuQguEXWGB&vt)A2<HeUI{U>J2RW*&Baw+6;M%Q&<@UK7)>n5_Zrc%>NZ4
zrXoXCDE)j0_nyRlyi4L!^3e0)ZR6<7W9lBtad$aJClcm6ckQ>EBkU-bg|mOJokwtW
zANll4lPB8#$5IP&#LEbo{YJH~1?w?84;GsD3yle&=JUpO)VE_FU6cv+D}2k@PjR)=
zaBR3*PGIEf`XmL=?i~3NXj4NO5DTpRSw?Uo30!2D44#>{b&>s9Bu$o4gzWAjR?I_4
zpf490x?gG>h#2_Wate9&&0E)GJQwmSJ5olymf}TxeM=bJzQH`xf&2d}|H19<1w#u7
zSrf4Vvp%g{z<|PmEXkcNf_`pfJ}0tsWo^wI{ip{>n)0yvL4K)qj%(_^BACH;!Q8}^
z7zkfDspEo*|Mq@M=NVEsBKIjk#e=ykX*v4m-)5cx8{Lp|n=&bDlYF4YsutA)uMFVw
zyYh(Ft~AN1P$kNjL4zcpe17(vN=?G;JGQugE8q1dFAICF)MX+kqFdpc1gXbz5quy`
zvl9sR4~+70ne_>DUkB4HL1RgdIAlRmMVYL<<Gk}#<4E%9OOOHGW$|Q2+-iEy#dzqy
zzv9n&4vMLV>SrX@YOQQNpH`wD*Hqui<0uvJFM226(OrbGo$a_M>u*A)1kcST!kFtM
zf?;kIQ8r_Tep=V%9k@!Sk^^#+n(Qt30JmCH#RZTGWklpGqu_bzo6ZaTkBF}sLl1e~
zDtNPJ%Y3pgAZus5AYN-3F~cz7ay2SRZ4$Acql!a2&T{V5dB>HyPJIpcJbK@wG&w~(
zVGHvOZwcQwLz-*oLc-I~jg)2?S@W@T(8iQ>6xoO*{P^tMV=L|{%|(4ii@CsO-p7CB
zzMY=ZZ=XLq!!@D_k-p~0)f{_L%3{f?bTdwf^snj0O<7m-4<Hof=R3ZxA~9J|ke;%G
zx!|t2{6pNbsaj?GrZx$(TJ-2&{NnP^V=lK^q2A3K9`uLeE3tB38VHQCY_Ylh79U2S
zMZS0WhIYhn4ZkxM5d5$wAZ7^kBe=i1SH3s{jq5jq9HR0q!0Id!`lN+$OVBUE6qjY;
zM@w9%ib`$_DJ1B8SBkKYgEH`cAL0Y2sLnrciLdOjYnf|<hD6R&BRDNwEI(Ag4i$GL
ztM{2V6kJPZm!H{fY?*05eEvKV=NKGvbe5Dg588ommGQr5K}*N*1D>PQtKXIA-hqo-
z&v$Og8pmedWNv3A9u8L7(rF<kGU}xkuBb!288)iJXL?9HA+J=kaY$b&+K0pihoE-%
z8BiRXGvOmBq1ZczDop~;claA-?6esUZ_Ni)qstj5_rTLwwYAh&ymwT(RyroyJ8@5&
zb+Jt&`!y8Q)+qxnMDEE;M8v`sGpBblK3@)V3MFw37Vzgx89SDP#Mc<m143;3W2$yM
zTci=leOaQG=s7hJo;ds<odS{mnk~~Co_sK@2f=}$Q7#d(i$2Kpq3e=-v&77s>AC5N
zDwCZ&E%vt+^d4@x!U)+Ruif)}RKnVnlw2iB!$N+r^8pDv-#OOlbA^*M*Qv;qUzEr^
zSUpgjD6%b>-L4TvkJ_`zyQ9&EyH<d?J-qE)^Q{x4B!dpaU;b6gSON1Ymz33a_(j{+
z6WYu}d*ZL08jBd|ZPlBdBmUnq&-dQIzfZ2%%e7$M-dTk@KjOp8Va>DcHZ<YQv0$0p
zGNNP-jIF@IpMIM+Ny<2!E!q0GZF>LfO{Kmgh(&ipS_8JFh`r|t<JX*Cx&-#?$trs7
z4J%6)?;3#JXS+WyLw+dH3w9z`X?IW~s!GbVii1b2c{{QT=E%V*T(UmMP$%0wvF(Gp
z>N2#n{!^1#Oy!(3-D-g=2&E_*N%p9O?GB*0Wq};1GXsqq@M7zJCB=B#61mV)=pHw`
zH{SLJIQ5&aSj#wD_>m^7+(+vb^nt2^&LhPqme5xJwuw`#B@Nzs3C@-HKeGyY4VP~Z
zAQm#7;r}-QKHt&VmHhPsQtIZNA7CHz_8GHblfUy7>`b94aNyg!iEh(47`zz=$if$x
z&9=_eTWmNj@nB0LBMP&z#n`k;HD&0kDs#`a%M`yjlhTYu7P;a`rgtgiYzqr)(&7>D
zskralco3%wOE<(DB^z+&xT1dMilSqw;$Yj=KFxHHf(zA8;E5wFQ?6q*kuEshAQEq5
zr@KjvI~o<g^#^9OUUR4&5ZLuhez`)41XVf7XLd$=*uqm-HH*F>^%7M`#mW*-$s6(+
zGfr8;rjs@DSPz&;Jq4uk;d5n;F~hR>>>tNJLN^Vj_#_l%{*h&2<L&zR?0K^pt{``4
z?O49vz}2qvMW=nsMHEKW?S8S_&v2KR>XVkG2(EhlyxsBT+*IIhzRvI&eb^{0e`CGt
zftkL_CcQGe_#{n!T2Eh;Nz6<X4ZB{`cAc6%WaN;PS>RDL(?k=+ao*zSw+pUW&U!(Y
z=^hdR9c#kV2l|fq=q;7W_mz;}tK=+HdX-BB_rAYWYK5P^DPel%t^B+0>mhXDNlWsc
z!8WYU^zBEX;TFK^K-HG8Sb0wb_N)5pM}|kyNQ;lHk0<rsX}pd5og(lb*kGOB+lp?S
z7Co`^P<#PHY<pWK6e;fTt-7_qeG5lEO<LM+S~g4<Gruo^$$VH$%;%f?nAywtIu=bF
ztva8M)M8;No+zKz6`j|&)?w0fFStuqb8N+IE<VIXRRPzg<ksQ6ic=Rwb)$a@MsnpB
z@w}{8oN<M&o^?wG5AiY@R%p~`N@3BXh;<S3;z6~GO;uQ0A4%i@d{G{`kYUos;wJGQ
zpViKob&qiH{UFi77rm<qT?=|Wle&{!fC-OLk=%Q*d?KN`nUCcx3F`Z{rKCA*C5ZWS
zfdr=+h`t5-8Qb<oVn4o$tIT7@+pi_4v0uTRzP^%2-e?jTy%3p9j2Ytgv4(pl^Xg|S
z6i@XV$M~V!I_u>kRHgE>x9i3rP51_3w}pSQt9@uxE4~<Wv0yTw+NA5d)F$DjIN{ER
z3Bz&0CIs_-R^qh?1)dx_zti)gHTmW4&<`oV(Z$&0F4Oi?^bGFVCw#MmPx(Miz3OaP
z!Gm@RPTbt9qzfMSjRnnEy#?w|2pNga>!pGlJwJK~t&aq0OZQDKPOAag!V#eV{p}oo
z9PMjX)I+|VLh&OXmn8oPgxV8MZK(@iF{;?X5l@o%T|7FX`zOJuUjw6)eUp`UCoci=
zbP3~{qradHe0nRjLhn!68Ym{DHDfFM<1BVzZexpb%{CPXX4iz+<<^O8%T}t(;RTA+
z^Yg8j+g>$tsdE37gPo6`IF-=q1veIS*ZofJ&+;5~l=T~Qw0Qj^<Z-Ay8q{V(6z)7R
zU_st5CZ^)b*4o*AFF-rp(T#44UUlGB+Sf%S0^GcMcd=6-8!5FBa_Osv?iA?a2%QSO
zswj`k1_3+trjv=k6M~X1^*XrnE3E9s+u+6eCAb7P-Cl&g@4NMDWHgb}6O0&Gnc@wv
z-bmojMd`(`84fDkH31CPT=;>qQYrGLCPXSk4NX&)AndBTv^-SmYi_sz%&v`~TojXg
z1Y=&5QQn&zfdNTa?(+bC&+FMJc$e%}p^L)Fv>tZ5nCxVUspTVumrf%-7GRXjqaG!5
zQS1v}aJme7I{1tz@&Y@3qqvT>=DiCnvyVd?87TG>i9=^bz#N}k&(a?fwZ9WC5o4S)
zFcoim6>Wc~hz#b;M*<&{%tisG^p$ifXf#-rzmCrA`GJFOcnS4tgBdDuw2b510iR-R
zy7&A(oofEA2vI<h<`THn$}z2KQXx+=ZZedl-@lenvKatbC1YZ;J`&=D3?*d$7uI*V
zTl`CIiqq;4*JFG9F9Q~R!0d7pPRAjfLMDba0Xj)3{&5{Z&{1T!|MPSqM!fhYPV}d3
zv$Qruf>&Wh!JqW^84~%$y@Na~Fz#_K4CYoL{I2`T*ksPBb8+dpm^X$LhUv7)f9a^|
zK_#kE86W3tVHNK`I>_oC(}0Y8P)ikgY%PG;DU_6c8PHd@IHa9Aqsil(gQ<C30O}l~
zkYu){IRvP>Kh2^(wnb6w`>N>dQr<)M_Pv1%a0X~9Z}bM8s5ZWCZ)xjMyLS1kjt9m1
zi%OH+M_riu-R_D7L-a0(%m)BfKZ(YIKfmv@B!E-|&0~8{FwYXs8b6JLBghh)n_Y$X
zH8{qI;t9=b$>#cPkL)71Zf>vfF1Gm}@<3DqmbW8FJX?`B#{3^!Knzdc=q>wG>SGx%
z&gFny=iLhU^KtAuvR7qg_JEH8Y5J--vX{&t{TVWD{alSoBNEJ}P59H#ybcUcF&9K4
z%)FWvABHF0G=?vjC%k6l(#ogq-|x9CwV8c}{8ll{a^ttS<%i#^+>7MS17M|@gB)4_
zc88GswFkbMf%SGedz(C7!hVqM_j=xktQ8Zk$MvgnY(6}|MjCV6`ygkiEhILKSoCXN
zL;^sp#AeUGSNFw6=LzeCUAA#JTJ$lqP22Uu{M#p>apt17=Zb7UvRBH}?_-O%&s%#%
zGSM{s4B}iov{^w#dpJZTn9LJqFJcZ*nUJ%ghq9oMmjSP+@vmsC5m_Q9;FHCQztas+
zAX_Fm_1+4qYLRsYfjvd>5CY}Tt`j!m!;j0P!9!GE0`ZA$^KB5%m6(<NX`gSoz!+>}
z)1bmNwFf*S0?!`VuzC;*mSOuvwTKvdmVLtQeAg}Jf2t%=Qx6y8MeT?MpcuGO1Rk$<
zLQZDm6(J$Tym|5}ddzrVYCd>T|F5g>j%Mo*|CbnvEfhrsL6xGaTC0TEr6@&Fvs%=s
z+G0eEqV}j5C8(C#ZK)ljYFF&NORcKdoA3Sn&hPxb=lt?VZpQmOC+Ft5@AEv*eU0f$
zvNFsRLQ!29>{pGT&wn0kCvPH5AUQjRS{Mv%r`r`M{-n{|;b=$bKjTNie`Vy}it8ks
zY5Jq(l*NAZ^CdQF;y;L_Ock6*nx)o_IFcH`2u=L^O&X}Jxvr$UaTa6Zz_7HxzAqVr
zz{Q#eekcTeKvC6?Fyat+B34WZpu_+976-Htm$2i2z{XBWr=m5f6X`MAJ%DY&?64tb
zjF&}#fv)OD2bZ2UsIuh~B)CAABOp}lZUrEaAKt_7HuPPpIqwPx6B$XQWilmv*INp~
znGxx}d-TvsjigFrO3fALJsb>d7W})nS+@&D=qBr+s{y6=fX1Bzz(G7fzbfvrC(jxM
z`=;4RVu|J=l)k+gDax2<z_7BY3NV=BK7pqrc~#J0mqsl{?A98)D;+Gd#Cxx7j}$6j
z6R~#66X_foaCmj$V(msEIxOfZf?L2Ue#~bKsUS!?#T-6v(I#1f@0+g^F0q{;o**Li
z-bOa9>diV;#C<PUMfS%Oay!v<b9YW*HR<TJ9O`{Z0E)+&-nu2Me;bV!oBy$8KyJ-5
z%(FR*UOVrXl24=!Vuz7j_Og}C-UeHa?1x%Mtg4WL&;Ve1i^UT+o9n)DhQg+`-BMD-
zcT!=pP;Dz6c&ie^E+7`>1F#aZwd_ts=6AJO52o}^K=<z&%|?6TL&d@SEh|FB&0Dh?
zXJyP-`3)Q<$NXX^_$<`QlF-z1{Ep10i~ootp}PqT*h>cOY4?OR#9EoU;olx{G9;4M
zU6!kp&d%(5`|X1gl=q5x#$Pn0{wME=SzRtaZ|2%|O4-;>Rt>S=NbmE$HLrZRPdrqV
zA4J=zK~Aoz^YUCoQkQ^Xq~!RN<}FEsgNbQCHoaYA9{1m^5@M(41NArpCQ=iu*VRAq
ziKy>7+o`YTug(Wt<13yiDth`kJQo~ygXuUV9T(1rX$JBiakVwS`RC$G>xEHL{8`AO
zIsQm!X@pLJvqU+%lEr*y#23LKbv6cq&gz9S(An=n2Lxaw9y&*^iadp(iGLgdbE^rz
zMNzLgV3YFPC(MyLlZ}BYG0U`|06nW<$1DJOdiSiA6v*GlZgV|DlahWp8X5T|;B_#-
z+?BkDW)E;g{8v$AD`q%j5MKi*0o$>eZq=NE@Xl~(1XS~j9`h)^{gy^Tqt!pR5CAIb
z`<=Nt`_!SRO!K3K5vOK;`B}5#c{eC)v6x!&c2fo)fn-$J1Is;2C`V~T?N-hE<D=~6
zdX$zRro2{M3BS$<&X^NvE#u2EL2@Nf&pLmKy|(gKkv+Zo%{&;Sph{%Gyce+BhXDD$
zln$$ti&C;;Ok_yih~ZJ~v6@3P`BZuYAjK2Oy}M$@h{m#a64~vq@7Vp4W=M+MvPKMY
zl{lPWH9S6nMu&A+{+*2RL}goJV{d!p74&iJFuo9<x5=nI@pC~)fb{H8;w`-pz(PAg
z)36+pog>wcu~Flll<ooFS`}sAVX+ru62pt3jC-7^EBhvL+&5qqlXlf10(h5T0Mm7`
zF>z1zRzaomxcoU}wBs*;BxOY`#h3L5QsO&<pR~}>dN0y&?Fg#^z-!naT%-VLu_P@}
zk2aAC{*6`oNs9Ikz2;lqIpYWV3Kh)D3Z)cTP&G^;%K3TPsIY;JiI#!#ndh?dmJlPU
zqns`<UX8(F;HFg#299MF11QM2JrQ7DKIQz5l5JT8=N%dM9obTR&un9okD+EgzE4k~
zQf?pRJ@m5Wu=>j99H4~{Z$N_`Ywft1h4&~LgWr5T3CVQyO0#qGzQI&Jw*V1xf)ou+
zVL*S9-U1Ftv;F>;{+46BO>&5yEpAw6$Q<l?sv`L7daOePvElWGs*>Kc@X(f4`Y3N8
zQ{aSSh<d}&8=z{dIP!5Nm;<^6^{-dr4|Lz8OWEU;whHuPo5kt#m&H>6DH4vPdL_(`
z25(YR4|aT^Y>DNB=Pear17tr&Dcz279q@t4zfw)X1=O>4u1rbY+=cbC_-ruYr=(Z_
z{bpAh1bg0iy`gv<RO^eb|K7VBr(P`B!(BQ0V^#$l6?pzZ#Ma^t)Hkh9TbIl|*LbOo
ztU3njP%WfBcT*?7xlHbB=sAOI;_QJIz>d^?nE>${3@cpJqV0O4Ii|(#`4*GRvLw?Z
zNj}%!#kw@r)p2OT##g=|46-{DPW+EhGQNu@AF!Ho1N7miQd6nwtEmbL(NRSeRPSCw
zt-6K&&dTws`*H5AGNlI@?;^1yjx6qJy5>0(QA+L2j$92n1>BwdiVEk}Gw2Y_?lnsH
z8NC9L%^^&JA?74lG;0Wnxt=5Yp3pMn``qc|-22(ijL}8ZrGhk@FA|Y3;(i<9W(eYN
z7c8DH|Hw1Fk021=L!TEQ7Y|n-(n{rKTHChP%7zd;tb{oe=xUP@^Cm&uu?+x|6c{En
z-<N=Wb$k`Yo2bfr@Y+h2f@x>Q(zL!}t{b2yR@2phcz|g>k0wT7;$P0k8FWi_y%_f4
zP4GqNO$C^7C2PO1tJAztQ)%eo(RccjOz1^6y&pq0gcLuUtz2H$cq;o7VE<fHGj=|y
zKTj%Q{zCLt(Nv_mo)lPxgDx2$7Yks~o|wz~{|$h9>_5yheK-t?^0Z190XGs=dBt}*
z*N<g}=szEGVJs-zXVgAb-%!hm*rrd4BBj3%aDG7X=T6jFfcUBt)2iTK;Au-XaeYfP
zae@CQ0KJV=iyONJ-)6hyS1eG4@cDigO=?71+96D5<r>P1SxW9sv;e~S!aKX^4-oy~
zC<;J}O6d<!qOy1La|P*H@s#M!UG&x6rq~i<|7?j-0gU0iY{S+WV5<v)D1VLh%)>Ik
zb!a;pV9(m2FEcxL>OMl-^j-~rj<E0fx#IQqF;96@IU`Hr@;9aYLUlB);BiB2!NEzK
zHGmUEsl5TvC7-IFtCkU8)O>M8ucrp~YUUqJ9m9|lmifo4X`Q4%HsBArh{WgfO9pCC
zV-kN?1#|h6ytWV%CSj4DH7lwOv9Apya|{KjVNF!R&RbJV*rziZ0aq5TQ&U{B8Hg%6
zZNYdtm_8Gz4q<qD&$-~IO~nRhtA}mGDP;Un;mAYbR_fIEq#zpDYT`}z2EcGVQ7DKf
ze|kPui{mR^tqj1ywoSc=_q8f0`#R5$@j<Za8V`aFgq|Q~Pv5Bp&E7erdtGze0Lx*E
zn9|69RYbX`3bXR!13Dr2w&4T#ws&ghj?rxdmlU5xup-#YkXDggj{>2Opp@zyj*F*L
z05jj)pZ5rYgMP{H6Ey&lE3QVU-PD}uv-YixYl=_EVXvGN7Zpqqv_RbcJj||#1sd^=
zR*~gfye&ZT#sdO@5E5()(21McM=)(S$Wr2OlOl=2Kmm5`{Lk63xDk+0I>@nPVj7`V
zwZ_o{%b<j?0%V>7P6_RViaC)zA3**ZvO;rwA>EYsiZ|Wc=rn-s9kiFKsg}K;u|gWK
z>Jui6-_=|qcab<;!!(v6XkD8Dnp$}rH0Te+y&#OJS9B#kh!Y3Z;R$Rz)ljpBFZ0i<
z2w$mXaoiiT4Ye?;l-WzC4$5b?QC}2WgGlD*YZWK6IXDWs$RH8cJt>}n@G*Wua<iE1
zkI%RKg4IlS_8YIe3Uf<!8n7MtJdoure$bdB^!7;f$ehy-(-(aWk3TZo553IUTRPuH
zAb9|^yw)fpA^ZqB9elTWSSyh()tCOKe?Oj_+qm-0+HXljzWJzIK=zMPNuTr6&#zCZ
z!wovgHFq*36{eh6sqIeuwh&x@OEUN!+KeJ7X7w)SQ-<%iBpdC5v+`c760W<qa6Qtd
zIIknT(Nm(rtVT)FIMCB*;mmHX%7ZKCM0)kvIOe^|lOPo;>y+|GUb}od$Jq+;M@R2F
zcd{s4HwOS<mE`n`svjvXi3>ONRISB{{PP$mkZtE>8xjjW`e}*eJE41=7v)*8Vs^E{
z$86l~kKL+rgK8@eP!}h;PB3eYtJe-A5_b7_$T65#+CwJiQO@HZ#JEW^P0TlZ3u=RZ
z4ZH6Gop36Vnow@fSYqCOp2~QnkvXjVBork{ct=CXFjge7RiQY8iaGKDD*>BCkTEur
z7+F_plyDLI$$EMeqgb#6t64FN5RQ!xHX;F*G#AHOEQtPME4$Fk(umh*_UNO4*`duZ
zm?OGTjppCe{5q0MrWJAu@W_(`if^w}dGF}vw`PqP?kdQsYMT6W3=Y_EY>gcl+CIJ#
zAN&&mhEHaT`;z%H1Pu-Z&IDL{sjl9s{Ou2CD%ecFucKhqFLugfu~IiB<4QSi!m;EN
z`u55#+euTUD?h!Xvo`bmz=Gb)&jLbA{$l>-tkxg)#|hmLYr-%fWm7fZ-oJ11dtv6_
z6S?dFSEr(wU7`O6ny#DzCIjOA)JB&Lcx`y|#6)x&m5+7qq7k3ht?H+PXUz^%s=Jea
z&?zW!$|p7iS;Kf6_Q4-=>lXRa<5*-#!eiodUOS3kpAYA;CQt2}X;vB`!Z}w<BPlQ>
zm2Jfq`1HH4R!XANUSnbCJ+pr)WgV();FAH}svM8il+PWvpjh-)c=O~<3cKfg6$r1Q
z(CYl5gkuBf(moTjQo&@@8P)F(x!)iBo1`kaZ$bMy1tqFED>%hGtvrR58n4RD-hH4+
zaVw!u$yk@K=pRyq;_J_)Bw1>7ibx5D8Y`Dw`;Ak#4ApEqWU5De#&;(q;mbcFPJS$y
z(Qrjd^R9(fDYWfJSCxlKqex}~Y_zS<DZ>ci*}B!)O764iKkg`xw`s@GJ>0KnpdaqO
zvS=esOK;aMPT{>5oKey*@I*r;Iaed}Wy>e_6iS{?IXTAPm^xpjhZ^(lQqoUg0>Wex
zeuo~ofoZ=Qw;bghJc0Jk_7|b;s+s6Vd-_oi9r()ry%P=7l|@;XGrELT->#%K?y<h6
zndRPBJl4*Vr0l~bTDslx2K~M`f$&e;0=ilp7VOL+y5<`CFsF+qF2FoIxWN$G%cNod
zNvTROoUTFyqBk4Y;zWv|hl_+`sQF@b=6rptPv7i)KsAR<Nq>RI3JGAt277piHOwQ!
z<@NRP#?U2eR+N)bIYQwx6twSEY4RDq_1=G`)pF3u<S>MtS^3UbkM+<Nowd`(cIA}e
zs6Su6@Y+Li0<7`2buA0n{?0LLLW4icdTmnp-n@+{g?G&CAX<!K>1Y;{BswZ@9)AnA
zxV{~X%E_os(AjNPnT}u@bUq}Ichw)-Ave#ViY=&S1QQn}8v~!j<d=b*ftZi^kFsvi
z6Y3su)NB+omwb2>Yuy-=p}X~{dY*&53VY19PnJlw^CEQ1#JQ_n9qRD@S+SBCJ6JB_
zNq02pHm|b!aA}7(h1OjRC*@adDefm~Va+k@U_$9Y!lndgVS2^~7vArP?qZ8M_PnZ`
zCN?mqJ>UK`9SYT0RK`^3z!#Rsm-o=xDCXYS?}TriWY*{LeS_wUyklVrnt_|V@`(wL
z<6rg-s3_*5lnj|0L4me=p!;2UTMJK_&!5C+{8AYTc{LCm1X7{Kr1OE{o1yE<D)NOO
zXLP8y1Wl+W!@lBGK!N$C!ztREBszH4;gA+PciXcY1|FQZM)X#<D2?3bqi#ro&;sX-
zu0%YvVSg`+Y->#^-HjGBb7IZ!pbUYe=ZBA@<GA>^-=|ZJh4$Hy=zZ7dEJIDkc?YQy
zLAv#ybC2-D`j}pomxw2INrSwv1=ZMyk6Yz~1!Oj_&jbrfI?~stGs#F_@5bL~3sF<D
z;~Zb1`An4NoBx>XS+w%2=d)4HYu(2@Dv{VmRp=8|S}go7+<4rMDso)wI`&do@|gmy
zTB;XVsuI;4dc)MHeD-F3=?XI|cwh+C*^U>D`z}aKw8P|Phb4T0D~JjFI7xj<KKf<J
zjHLdsuUS^6Hm-eGS)7}X8o`hx&6_L0qZGsfMlUb%|E2AUckAQiO3H4(9{gVBo`SDE
z6eMizY^(5=z?7ih&dm9MVz&0tU8WAn?w8boyby&W?&TI1kH@h5*hjGynX#NNSy;jL
z-wYH9ik~naHSg~^P=7~&M=u{G|MD8vf3>bNuVt<_m-vn2zN3@!Sky$+jsv)1|3lP#
z^4*N5M3ZD~j^-2B>qP3h$^qLZo3n#eQQwC#WuYwNt2r-<0I5*J51~&R!h^*;_vlyi
zv}MJ8s~gF=mRJ32sy$W}n09`4Jq)f{pHBEARLoY1kr6DcQFJ$bH2v7Pvh~H?=D5VF
z5HF$VP7Q(Ja0RErRslP*cx;k(0m1m4Ti!`YxBH8bIY&rEAvMUt)0IV47>a#&i;MP-
zqIscXX7td#z>cvERhq8h53a#O^-PhqJ+2fR+6QmZ!$ZeHr%*WlnDaGhg;wr-Z)|49
z)yy&olUa`6v1tkzw;<fssgLlx6L=)BxolovA93t4{kQzg=G-%FY%ck9EaGNEV`K1;
z>!j=MeFe*wV`U$o;$LfNqk1dEDZQbyvkS~zSz2*v@%|8!a9-ZAA7o0zWHOd_(4I@`
zb>I(+_fJ)Bc-)3>{=0X&jQjO(G6)@5(-h={LH@F8QYk+jL1OU4^SxOGtADAueOHT<
z6`bQ`z~)H7@x9B9!HM{(<BKm(cOB*6(kt5V?2(Sl%|FM_A4%`5ATXt$j(3maBka%H
z0rC;8{N-(hRH5YJImL$RHNZ1vtZMB0LF2EdQ@#F~jo&8@@A<fReB4~#c&+eC`jp6n
zc|`~m-PDWNJRm)9JUNxWZ2F`vRKm~%PrdMxrHhz(ZHd=ya)IMaj@>>L;MQtVAK@*l
zqjd;BKHBV$k`@l@_j!6cGWTD8n+}lSmM)&zi-?&wpTJiaR}YQ-*x30|75*+d9(fTl
zQmM1EzpHdXTHd84ArBGNgrMn8yT(h?I)54G^53TgJ<nXE3zmZd-~Vxx?KL^T-7!>$
zpKROv5toK)8#QIJ{N$%@oYGIxUQ~bk8lPV1YQAw+y8W^yw$$ZS6-QNM+(W$QDZYAa
zuhNn&UAq8(L0i4%24T8NdG|^?_X6%B5~7C*lVEzY6;4n|^S@Wd#cmN3Z5!U0%7*^c
zUI<FoWY*T6@Bs`&rk?5jkp9tocxU-oo49HyS6*UMF((+D5BqNN!j5z?x^OLlae6P{
zySk(VTi?j8;l!ySk^Hbl&X3yQ!?^wzOEW5@W3m_IBx%0#7pKm~V&raMU~|eJFOBFY
zd0p%OsBRQJy#Jd+hfh=p)sU9|C&-@NanY?pcb91(wv8c_`D>N!$RI`1^{&LoyX*TA
z*}0QDXRb}%EyzVS7q9iE)Z)6n?zp&HL(&OF0+`2K$cDT+@@m1ikZT~yOh3BQz-skq
z$aN5A=H+h*CiN}$69)CeI05<mQnxD(+Bj4z+xo|kKB$cAdEcG$wT2qbyLJCMV^*{N
zZuQ!nb^TmyEK8CmwcK;nJqk;tc<&nHwIO?KKKqGXfV%MgBctEex(7_1sO)nQ&zqjV
z7iOi!d@Grf9Nd*xy>;(JmgY#U(hrO1AQo(oFoQ>CHUWE1lOpFuqe^$$UsO^(R?p4L
zl5o-BmlcVHzuO!&X>4|&VI7{A&{tg8YeO8#1<3acT_#rj45XdQQ;`|mYKfb7t<L<r
z1CKrRIH@#K+{0&>jd^{<cD6cT*={CKDscbS^4A~t^<#e&ZLU*RJ&R(be>CRu5nQ%-
z7t`=W!o>gBT=|_qasBd+o@sx%pY-CwFYmD%$IB^){YNh(X8kvqVtTJOX-@cce@h~&
zDzd~V#7}}m;w<0?;N&)jw%jj#e!b<dy3LsbIkZhvznXRx*rJ0-ff-2Zjo*t%uHD%*
z+?j#SxA=Bt-qVjjitq4{k&#_ql9TCcSFguHVkxQPBUf<SATqLp>fmI^4G?A=d9_aZ
z>x_U0joQxy+f^HL@-`P@hK^9HvsS`z&cs;!S1ywY?+m?w_`M+(tuIq!gr2zh21)e8
zx2C$1{ew~uw#JseT(}!0_Cj*G1~}-Q+`)RT#~qB$8!?pZSB~feE3J}Mpa*XZZZY=L
zB7ao2>o`5V1x8gDd~$ggCeM$tLib}+{RsIaP9^F7rxRKUF>j?iNKJXMi34~{ksjN=
zOmS{?7PPsk16gnS3`JaR`#MQ6Fj52|WU!B=rAtf!1P4nAIjX-CHSN1aX7BJ<xj|Fc
z!dN7&xsDN5#v>sEVYc|v&6F!3|K29ywHRraPB_(X(-dndF&p1>ZcHb8L2J}X%%_Av
zBTkFn9{0Z0s~=Q^4s^%KCq;{lJmp}{LTV5$XR23fE7$>_keSfP&{-N*3;2&;FgpqH
z-F5c3I)Mqs1-|cY593;ebICb+)Fs-o_FJ~iC?h6W;H{OSu_O1td(er~LL_{fj|RE3
z`>_~@ZPx{!7tzCi=bzhDW{hiPWf6~5yq=+N?yUuC%5+GJ+`A~iR94kSbM21yYD-Rp
z+Osmq-xTPx|B`X5XTy^}NnnbPmOjomgQ!~J-w?a82s<jDRq-f0mWTQLbqr{4c+%&7
z_U5CUN6nR?)5Rtb?rkb$l4jrshv2;-N-|J0&J_pI0}IdKns5*l_yH%5JPo2uuFOk?
zyM$A*;VjZ1B4DmtxG3OLaNuoR?|;WdQygPDaHO}z8Kgr{<bY2gF6bGsx-=bFeJv0-
z@!z)sSlqP?$P+NS2<M#v#Qau4g?n8Ntm@AIR)u}Xon=6Tu92PA&CphnkvXC%$Y4N*
z{_jJi5r@o#D1hzUaY307c`zVnP@f4X4VXpZPBS4A|99(@cn@I(TffC=yoacQ@zJ>0
z_YlEr|7WiOGJ6<I_WuPfh{5%}2ZEZ%<LI-1V7-aBhgp#8*Z$wNKu$)+^uNCWcr~4Z
Oi_d}}sDSeVy!#(|sLmJw
delta 425738
zcmYIub8sNf`*dvEzSy>HT`sn5CmY+gZQJI>PA<Hd7hU|`=UZ>pueNGy_mA1Cotf_E
z>F%lOBWmv^YFLDYQ|nodTw8zysVPg|hQ(ERQ$3VjJo&<Jyiq=Eb{UFefy&kq19H%z
z<hkA!=BLdVSp-BdR5w;hs(q`b0#0UnmIQ|dH#9V4+K%9{!IK&m7Z%o=DAKF^U#e$n
zW@^4(lO^Hkk%hj-ft{PKZQKXVf`~qe<Ulw&z{78lkh1i%7aVU}MGMa1grEP$zOLZ;
z)wpAStZJ!Lv@!fn;GC&%^7#DcdHE#%V;OPmvsOXBt;<5OEP#=_LWJ2K{2A5fkWYIv
zb}zuO+^)M7Wk38Z3yyjj)~<=m0m-=0ssQ^X{v+b#y{`9&nE>fNx;p5k3z0uTuU_aJ
zSo&4uv1H(0oV_jZgyJC}#Cy|cu<J=kf^=08<M21<I9zPw-fo(>-^A|<mmI7eNAE6&
zk^*3}mWlTs{2Pvao~3!R%4OglJhqJ3M&Yx;z&Wf_GiBsJkN5oX^Fs0^dN26%ul=u!
zYj=X}xNAt?#~`0j%Sdv~zB(3cz0I{EVB5t#o(~lJVQh_I(M0Lt7L^SQqNL;@+I$KB
z8fo7lfBVLo(|zt)*2YFE|6C`(O>Uu3Y_$yBKmoWHaG%&d>JAXn-#bV5U0qGLwa;ws
zW<SNT51*@cOlmlZA}6A*oGVzodmtgW4ZO`DDZl)syM}P|pdv+yoQeNy{R_DSc%l0$
zo$D^{%hMIyxcS|WwC8~~?z>ZSiUY4v$GCTEz23310KW0%yuQcub_cb4Q)S?v<vqAI
zm6vj1A>;meY2kI5>Zyps7xE|-`bbZm@S_KXU>X4_gVVz-pVvF>0-iFn!=x031^P!i
zgR`rfx1*P@ySE7CBy(KAtP*}LFjk1WjQqZhbw`u*`6e_&UXRSzIqRiwN}Drw+=oIS
z0Wyv~<-1su=#Y2g@(4jO#smS)I=+n^1U|^}9J;4T=2#$Q`7Wy2kFdHf5$(3{dMtVG
zXrqDst)E2?gJ?<%n`irk%=@*tnBybL`+WA5k+95bjtW-|g-1|;*C)#W&<}|*hMU`b
zT@|@^{BM}ZESW>|9YH@`Bieeopjy-w+T^-!`-KN8{webPUjM*B&9<K2m1h-S2a8-&
zZQcpcd#|4TQm2hcSB9=Jq?k|<iJ^o*H16Bn*jW$k9~$|SWS)CBQEK?!jE&^V3!Xv~
z^N6<6XO@>lDI5GL4+T8l)J#q)IMDR@hIDK$Vuxe3&~o{Aq+LBHw!O4khf1vQ&>u5D
zo(73;P>n5f^Ye3W&=SZFd<NB;52dc+EuylGdj;!Xdc^#nR!%U<p#A=;euhQ~2kWJM
zyL$dG+E99L{GqZ?Ids6dK~7j!6$0t;T1zmcJ&)BgK9x#)7sShmp0_w2pGAdTE9)9q
z)({XUxL~MRnUHT)|CP!AmEJ_u@!aNiAfRQL@H2R!M=_9bgGZB(SJxnjF&RK21o(n1
z-5W^H{049M(>pm9z2|ig*cOQ(){=OFU%*;)H7gUzZ%DyonX?<$LI$ogaDLf18!BBy
z;`3ebNhx?1W%$lKqkdE3@;3h8|1$Jv?%cEF4ROOnf`bnjPpq#{Xi);pyyFAU)A8u$
zKl44kLuHl>X3tnpp2URK9;5*c+I}I1CAqzKK*#2Kc+Wl0&%J~yiLtm!>TT{$zwJER
zX6ubuT9;huXY6=y!R(LCA}r?CbC|J_sOclwzqihzbq|%wL<D+gFpQ<2O;V}11Rs*S
zIt89R^85|ad6z};BP3b$4<91q9wcW@YnE?*?S5Cc1EhAd<R`j8;p4zQ<J;rtknu~Y
zQ%9%3R{URM&*h7YW66kkTZ@cws+8g4vCp=ZJE7GOGE6u7&Q<R%!G5l4<9@^p_OpTN
z$E(@dF-1S0)^P|!hfo2}SGo_$>1o-m`^Om8fTPhAs?_*zXcBr1)1A^Y1`OT({q>Ql
z`1u~Z^3vq!v8}H^&o_awbD`F!R<gIgat+J9tOvhapYfypUut@PE4ZJ{oHG$e!A3=+
zM}KbauSzvlkT;^#p>V|8WCuer+dcR-ZbtuNRW7vFTe;5<dw;4)U3+m(;O`w!`BDEY
zj>WER=i~FXx3}?;#=rBRQ}GkKW_JkIw`g~@9l1bu*0uke()cNGVcNgq0pisBdZpE9
ze6#%RrV|E5(5X*H2*3jl9`Bl30iNzIejY(Pfx0~3OBHe@#I2=GSSU9}Nk^U9*sEjB
z>S6!v>-O2l*PpAK<a!tv@t>i1uk^3E(++NLhf%*KZXF(SZ+Z0~xgK%dGKt{Rv*Z2h
z@;xo=T<B~KT+e_R#h%NiytI-$Y{xsySLEw~3U!6Eg}>>wx9b(f7W!w>n0!4qo%-v$
zD#gR0p8)`NPW$YeI6DX-H4&nSYt2FiQOfE1L&3=Am$Q!xrzR!d7utGEX7yD!v>qHX
zUSBuClbh5IbL`{k4E^SBNYzNWC+2u9ra7dEc*d7IraypK;$mqzL^V(ca5~`S<9}l#
zjk8bMahW+Yn0!@N9zD?8UfG4&g|`=h11q0yz11~vw(#+@@$)emPfDP|#1{MDb64~g
z&tpKowq-|JO^QT?RS~??H1yXzvuB-+j27|j%ggDW*q9Z{EQUd>5K7T!6)|#OGOmq?
z-W2G5IR`o!;gkBM1Ce$w-RGcQZGcmnBYy{ftK9ZDX<tjwse6#OuBRT4uUW7^7vZ1j
zkL{96ImgWdgzdf@l)+4SY#w;AA4-Tz6+ABH++6&?P&Qkk!o=xA4!b}#zX(HqQ)$|`
zLf$yMeXrjzmR@pjs1Ev&dL2%$(3Si6=y*#oHz1hWB%9J#f;YCS0js81E9;r<D7GuT
z6}|LCRbp3ig<$0aZT0r8#SUi^YEdVs&t%93J-Ot>CH(a$4h~tCP)-z*?F{y~6HKBf
zz|l?&ZemkpN;ikqBY~AlSZi1Or+h3M*V+s?sYRduN@&L}fn^t0kXm|&JnztjSPo7_
z8^8|Js$+T%nW+sA$NEU228fm^CRSc-us&v_1YUK;bP<Q-xCl?p>kJHQn@~1{Np_H!
z>%k!X@+4g37JnYD#_ftj5D4ssdL>>oJ(bME(N<x?y?pt!dl^7L><vZuC?eJlD8!$b
zPezFS6~~V2Eue&ect*44re2)i9|gVT1i&e2mxpzol=w^QR#lkd8UE&*{i!=aYqfM6
z`>Bol^;n#0oX|r2?aGbfN@nqw=%Qj`i`}Yjo^*F$$p_q`{fLat?OTIn!LT3NJke6m
zmA3%afZ*!lm<5R+lj1JE^e%dYZaukQx%6&!XR(6;7Zj|o6Il4C7}vx>N>5}8KM>d$
z%TaQKL9;;tV=_N}QL(|jk*V%=tAAl``O)cd2)Vor#VSIq0;%WM4B8wn>DCIPeuOQL
zC@!3VqHFn`0LcyF%40$A51jL^sB=EW^B7tZAEj@f+RzEQ+7JEaPw(x=%{-G+><t&A
zr;(X{nWt^h6;Hudg~ZkB?<VpeJ^=X&j&A2#9L-i<HiWS3B_8SNP?uE@Zxn1%B$WFo
z3vFZJU7<f80y26%p^|rDpn^~1vq+Ia#wp7Wlt@9fqH)tfE;8?SKZcCsS%QWhE9m>c
zwn{A%3iEP?BJQ(Az6#uHWbZ&NGYWNd!52GWKP9VDJpi{$#t^IHo}kTn90)0yxP@oq
zFXL2X;b-J8o49qm7oNDSx9}W<7GXRk+yp%$Lh$C&hFz=boF=H3`jD28>e5fCyU5t(
zmRQ4=UZYub8+(bJ7F!djVmGs?<wTB4&u#sB1|kn#PM&@x$7Iyj5k*)Lc~41M&=0I-
zTyc-C;iTnK8d&69b?Exm0vwxRT8^_40|t)eO5_+1wv_x-#&%2~QgA*6iK}+$5m)Wf
zzv0z*$|Mw3UN}{~qSW)8w2z!scy3MthAUHQJc{d68qC_t&Mo7$!1xMK>uv(!{$6O}
z-;+^k`QZOxUNXJX{Vf4@?=`psy??G-^F3?NhMM~zu#X`#c0R2i5hz%vrCq5nv6%76
zDYB@6N~IJyp}j#p?Yw%?L_Ocwf$lY1&wPBfmhM`1KxarRBS|qoGpZjKE|%=`C+{08
znV{(#dmLwf!AmCF4d^hcTX?=il(b6p8{~VnIon|YGQJh=!bm!ho-##&VlxiIT#v%y
zqZ5|H;KLi2%dF#;17M8EMsm9@_b7B|UE<UwP0!GrqT+O`jN}OoM0ZcZn}6*sIeNOE
z!P0lh6Pe_XGF*6Mj`H0TCe$dLXF7e!$virygP}@M)1?v#j=4<-m2sp*3Fi>4*u~n?
zqo5cu8P6I5XyyVnSWE6gtq9m?sjPgVAMe=zdo&O~G#LZ>>}@d9Vnx4re-S2+J=%Kl
z&A11Sj0mRZEkp@)J6kI4+4;pCveot6(P$PpIMI$SkIK;2)-#w~lVLecl|-lwBmBsV
z;`R${^2j5XiHB+#F0dt9JSlqE7*HRwdL3!)dD-S1)vo_8fuAAwSCS~yXYTtBK}T3^
za(c!wO8O&^sQ|GyRevZ}(+qJVxqI>-^KqJEkl9(tZDpwbm28>i5IGKLEyN8;V)`-O
zkZ-(exAnT17aZ+g|IKCwH;2+@TaHDcX{_WQ3@-X^J363#a}?Q5s;9NsN#h0OTE%ZS
zaAMC?tirKhv!biBiW0X>)%KGL3*iwUv?FNdOam;*j#1MW8L}Ca=@vP%9pZ(qNRRg@
zvQ(MK-DnVb!x2gj8N_vD);(^migi>t!^iuJWS1AedD<68HqHQKqB<f*#ghy#ZiSOm
zDVgM-Y>EQ9M7+;nqa&^RV+09Z)0_GrBE-5K4NK`e%W|yYVkA5B&fI9vLnE;?5lbVn
ziNL%Sn#UMV#AME$VER<YH$fPq6VWS;)loukB};tH3|u~?m@RSUbeVeeg0G`kPkc7i
za?kF-R>8$1J9!CT^3bij&Y4kvqj@B<x0ygaLaH<dm2v%?YL*ShsOq)!MthDh4GVPb
zTqnlgu9lCZ)U!-6jgo(KGx%9X9kTcvRe=Xz^X!9~A9P(AhtHBxb(ORpcHAt`=ZrZZ
zsLOR!Z-Rt)b0M`p7TOUxz1$Wij7BWn_?W-sF@CwR>C((eHxpTa9jSRcmIveAR&Tfx
zs==3UV5FZOtL>!h^=W|X&vCahcZ#o_={{G^Eqhk?D?)2z<BhcT6520d<b{khGXr;q
z&ID*mZ6nD}&f6Gw?=B=#u<)oC11y@%jivacH7)R+V0kSf75ej)l+7Nkc)!UR7!>6)
zpg1Mu@TWB^Df?SOtPJCE!<wu8t)ApNjqb2qGrFF%q=r4jcja-*+N1*884A(;@0E_J
z*)_4LEN8B=C@p7}6p{;A+Aky^1OQ+XD#q>_@?3kq@i_Zie7GzZdpImv>uT*6-mj-K
zl;<{2NE;KlHFg^lmK}C2e6@~Wajg!2#d^9#k*N7>QD4wI3xC>4-3DrO>(vdOv_=~C
zBk~XG9^Hh#;Inm>{q;fLA+!3!bxrx~{!VD!#&dAfQTIVrq%uH3gZzpj+6qYg&4q#@
z6qS*c@B<<$^`g;v%G^m4E#!_eHXq%^f0?O2S_(;t6SYYvcbM7DeG{N{P(cLHpT<m?
zOmA#A@Y~aT_@2g6LWAZx%Q6)fo>n)}7CFo4O_Q4voX~c*ui_*6l$D{4pA!VI&6VW;
zW|_~8p*309r~X2`7p{2d(+AGJ4_*8&yf&#gIG=`6cGtTlDm-cjMNBEL_;a1{DeL(l
zxl7;<8)}Yz&Rr&NI!--V=3F9|6R5p^#YsC3zw`!WUO?JwA$QZ!xr4vZZA!R)wnR~V
zRODG=6<6kNXNNj|0=`x<L7{;kxI4@{fYNAZEO{A%;^-$VzJ{1$UjztHeP&3%!=nqC
zgD9eYaoWQn5--sGfznTm?K$F#GjBT7&xhYuc6AyXIU11}v#l}eY<h9!lu)nNNM+%<
zi&Z!CyW~VR^Xi^q8N@w-&=%*DM}ym15-|~RTitB5u*~)hJdph8JvqSICZX@Rhd>s_
zlvFgxSYpXX<q^<P90z>I+7BKI&;Puubla`U`%C@O&uC{UcFJl`XDLSQ#oq8_V*R<6
z^Q7;Ktqo3DUGfNW&P+X!Ma&*UL6be-`9GCMt;wE}QP{-YXk4|A(YVSo<|TNcaJNy7
zX)Mq2522dO1ycm4Lw>_MS&H(h{35zlmYUI%uN;4cAEBdw76JOaXSnmRBhiv;+e*H}
zqoK&GJbAt6jFipuPm+>c@G7kT-)+?VQbV<z(r-<~F*3+f$f^|l4T{HD$ZW0UR!J5o
z=sCRCk>{2XE&KOCJ}xqBNxm;pjZ)}UefQNgf{8N0M(X0RwOH#RkN?BxGNIE20yjOt
zq_{Q!h-$Z8%`d}Xh}jQ(VH?mt3qTGjhkQ!eiC+iKE69_>XAngO9=@QRjd?4=g>XDX
z)=iewLOz&;7<`_Dg?L)(aYttv__!p4Ve;=;!<1~lP!X^07l7N!#zwPM4y^NbkMkt1
z%P5?;b*j3X8V7s!!}sLjRUxbGpFRpS6rBL@(;%)+5t%W5PLb(&7H^u_{-`{ji&;sS
z;Z9h3%BB0y?~)#;7M1!Nlm6b8-MH0NAwmx9>vq_~9krm9S=xC|#?JlsHmT98&?3<@
zHh~|zaQ67h4eL_wgJX*8v(1dmS{R4$51sjV*=}!Y&GpQnRmnft-8qK}b2i{5!}7U+
z$eA4WWhRT&D=F}E<WjE8PR!P3vw3qqvP(LOpda9P-KUwX1`ZsY?pQwzEhc;qaakRT
z_{vQMeMRX6nRIg^V&@K~VJMT#@@%Gb@u9l8|JTpA)m4pHh)B6rhtQO3K&CHT_F|Vo
z`V&Ha#xC@fmIrcxGhVjye6;nhXac*qbW&}*NIKb4jlqrna(VWz4E!m6{h<QKnP2FG
z-)wB&Y_bbQ8<fK-WlZGFZ{e5zvohvz=)|4U(RY)hG~@W*ZXg0lGf?E`pqByz6FS^C
zW(&YgJZib0e^2T4wyE$**^;(ljjR&1GDGChd<@PIF(set5NJa`UbVh10E<tOCbf-a
zn1xrJehs$1YH2Ydp%~^Io?0%Q_sSw^UgeBTjMEL!n-?}dKiqNE@C;;N#R5nN^6JU7
z7_wgl1GlB{5U}^@#F12!OUouZ{k8I4f1kv?>f+3f-Qi*?L)Q(<G|&0v3n`#J7S{&b
zk#EjrIVF{oJJ5q#`;%$N0x1Q7wnBaRrqvskt-t6PNmk19yl1|7QXa=mlLiE$e_Ior
z1!Ma-eA9U}H#vrL_>=AXGQQh14!r<csdE($Iuu=r0ex{%8_fL2XEJXiX@Zmr^he3K
zUcJCF$^!+rrhKYZtgMaWTb`<5w_rb!Uopo95q3OOQM)@QS2myA7U06+JZK9lMBbE;
z^U%_x7-3Fb|7bV+b1r4=_8&~j{$$~VkH8R|e(rLI&?!9Kg8H`qD2f{f<CyyJ`#C^Q
z*=Xd!c%lYAAQ`=WWMSZ^j~i*j<MY09vG0E@2RmQ^HOGhxl8@U1E0eHd(J_S*1T*Ln
zQ~<k9>T~pPegsh%0VvO)M4_;cZAD0o7(Zz)ml29lkfAk86(JT$(34>`j~g4@_DLS7
z+#eN%DvMK_Tk+lx#?D!*ul5KxBB67fzrE*mMktL~Al**B?dgk8;|*qjGh18Nga$0T
zYO%GhB~YSfZjiGTa7YP#djx}^0AocCFwNb}-AQqc*Ud8^1&C1NH?p>sxiC9X8#4s3
zL>y{FoWNSFQR1wIP=`DY2BwMJJ7NBaFZVK?)AnOW??y$o)a2u#JfT|=0n?!CZENQ+
zZNH(jYhepWkoffLCm{rcf0}h1&uqDixjSO3F5t>vu|<+dB=q4+_p317hMSUP!h891
zdgI^Xh^S;G0mg9Wi&FdBlZ(bMN+BnFF4T|TkXbz8spliuU11?_edJjRCs&e3)WnI=
zw=Z?Zf(<4cd7xV4E@e48ajDV%Nt0JJ=bK8#wS2u}Pl5B{o#lEMyAXHYdBN2dJO%@)
zJfe?k&_da2C=9s5%nGL+-eALzwF;zHhYB8X3<Ym+!0!o>928k2DDZ|n0(GP^SQFhb
zq#tF1rFD(E2dWQMgJI9a+p(h>kpMboy07ea+xRv?0B$dMD9F7*)=Bwzxj_;iE{Px6
z!7pg1d)YHx>kFHdpGdzU9&?Shm=qeGht#4nA3T%>kxmL@HNIUUA+?v!Esqdz>sJ4K
z?ZM+&1m<9M^cDt;zN^*GW91?Rr*1B*=%!qxMoN?y#m44!Ec3wY?$}&=Y#O}<E$FzA
zE{`;J`hMHeDxEJbM&{PDi|5`Xp!we7wEsr2eK?ErZS(c?`|Ry$A==mo64!AEDQ75W
z*@y6V0dih-{1b2MSViO$nXQ-#)Wa8!RXEkQT>wm(s8e`LOx{zmQkC4oj=GbXmR7Ki
z66t0LzhUnuI)A`%Dv17af4P{<Y63?h$5Li_eC;@UO17ns4sLtl-WzH-o)&@;FPhA-
z58k&#TQXnNzO#X6QKgBP9Q2CBRw+(vo7UE<Z)?u+wEF@nbg3xiS@u_jXf!IMrwyhG
zlt4u^+iTb>|2$OMa<gbQyau_!i@*%a_uKQhT!>;Xx-R>pYr9~ZL4uf&O8&C9$lY)&
z+J%QGtNBX<I3}6*1Rl;qkK-MpS)r6^*P|w{&)?@d7Y}GxLLs`c3E*achbgpU@gm79
zir-nm+)PimsVOsBW%+A?8acyUO>3YwDj-Rv`3SL$R;eN}^#(3ca4z7BZLeP^6ps1V
z{x4y50CKzQcugbh4PS2G+#_2;(+`sdwbY$5Yg;W|Gru?u$00U3OqY0Hak3(53^58@
zx15v75CIQ*3>Xavn<_)d=bifkIE%4${7Yud>gERmR+c6PFm5vt9j1Fm@gwGa9^eQ`
z6)aLR=tLQk7SfOlm(r?Ho4i~XHT(jaQ_x%UMfjA=bt#xJB=ZU_--BKK1J4Ss8lEZv
zfr!U#PE&XlhI{h7gg;KP`W?>SHv0{a@eF|kX36uus(7yj=8y!@h#%7SAxc=Ppd94E
zOK#ca0&G;=IOIA2of3kcfA|ml008uCp1Edfg`Ba_ScQYe$8S%(x=z;UP}FF3O3Ciu
z;3l&^u#LwJfmNUAs9?mr=HU(58Lw@nDd`f`24vmNE5|5ASc=Z1*+1l~>HBBB2~WGG
z2<cw;Y00xnr~hL4!$iqlGj3Vyv*)6URU;(;LkaRN3~jCZo?-@s!x=Uk4|u>qWS3Of
zlhowywGX&-9p7W{Np<{@iW>6h!GaheS`eNH#^y1Dz0hzV$^hrF$)P_X^x241)ZlI`
zR>BiOoTCv|#d+Bd`lt{SAy_NBN{ji|iA_($aFSXYRNSHm&h$x?-A2#|zJ=e$Ww?za
zW}LA*M`YWu<$@`PaMVOf6tMJ$BEzvXP|8Ps<%5(h>NO);4R@rZ51GP->bkZ@jYBwj
zN_k{3GRNH)FrOMw^dTYkV=RT9Z*PxgmjDs;6+rz&&pO)t1se)FLlRL=*cgC{@X_PP
zR*gq)WoZ>*-MXZTG6gdw&CAMK47Ew0iV{M~dOCc2N6ok`U*`Ra1DJ_8Mx<pB5@3gk
zx{CB=FlD}kHiShV2_)0q#pHWG&GH3i_VtL*^Su%(4=kcLIMK9c=n~H3!a)6=Px0lm
zGuFUAU}Nx&qwQAk9Dl&pM(xUWPFx5)ci#Vq?(9$MUr#;KgEx=nyamO9lvS(E<sZ;*
z`e{^UwVA*b+F&^u1Jn=Vp}k%TqH9;5TMfmf(8>Lc)55}S9d84-Wm%f2VAR`2EX3r$
ziOVunlTPakhgq8EFfzv%;D89}s)J4zP5sAJe{+YLjK`)e>f&|%js+VjXd>E>lHbQR
z)(Un5{Y-;(6woevsRnzEz_o(sJJGzAAL_Yil>};LSK#az0ZSO17&dL&mB}+<9Xv|7
z3?8(fmcy$J&tp=!p(}Cxyec!bePU;}UqCw&<79hvv|Q6J*=|-igR{H1;+*27)rH*s
zpTBbqe@>1kzh4Ah8t5e*!Y?i>@O1F=F_DInhL(yMmEjo1^!8;R1c2OUr2hdzjSf*6
zMknAi_RYBufcgjuvh+AC!nx;^GIU2fW;O+QK~>dqhzMkLpnzTrVL%OAZmUw5HU1>d
zWaqc%z0{MikDr}Im=O=P5Vg?R#9T$dzN_uf(PmvGXS(J_Y!OK;jl~K7Exn|bop;>>
zFLP>0HX0SjFUve)1@1xw(kIlnGeaTT!oRGi@vNz2Kqx&zB7080;oJO$V85!8arRME
zVv)FpKaNS^9Km-V_A?c2sueg}-~JoRPh8mKyd`71$v}A{eMwT&^By-ZH~%PtI8Qlf
z2ZaM4Wxw&RIpNYdBA?i?RjQF+e@vKI=?<eQ$K*r4_Zi}Rqq8|wwKB`><hi{hV}}P*
z7$sW12G*gI)GatjDo?^`skw=VgRl!FBpoFC{sp63<c{gY9r2jX3kgWYleMGn*jkc_
zCWBd0<Ts+0DWiz(6Z+A*hfq5nd2Pj2Mx<;7_P1-ry&;}Tv-+<>KgLgL#s2hq-R{mZ
zG0a#<ot8rMsu3LxZbXUypfBR<;uYxN7v$?D0i;eWb|-$XVL3rXNkBxrhewEiinTD{
zZ)yW}{%`(44`Bgvu#OGP#OBnYspse27npy%x)SJfv|1C{iwX6m%=U+>m=A|4i{VOO
z%tD-bmBS7s^h#g3b;u7EMYfG>zRsu950{vvonbm%Nw5{iP?+nf&MZGU;?CxZOa7j^
z0Nl9Y`9oE7#dFFX#u(V2FAJ9kr#TDPoF(FNulF(A4eJ|+BWZsJC;bi-<i=znTTG3+
zO{-&%Y>0JA{u)ngQO6<Ahc}3=mx6Z(Y*i>H!(qaI5sZo{nZnR0o)SLm4h8WQ>0J5L
zg+U)+-en=_XA-%T$cE@c&7)to2SW3_1Ct3pB|;;HRS)6Z;soY%p=H*YM4pV>{Fi63
z9dsyPq)#Y5XC@9ucrj!iTY1c*1-78W3oQf<;rzBEK^49h>Xdl`<m%p6tS{hqrie+d
z)2O0!sk`VP^f@g*D;r+XdD?y>rrvpL#jgnv4q2qD>|!+se%yo|VTD+=^doo<14X}P
z18vJYpneEv{!~n>#5%yI-MFsXVW$zJ`3c@Vo6<1^V$2!&U4G?m-F+c-H@bJ%U6Z#3
z;mv%@4;f4wy|>|jqo0t*QlmsMy9`5;W+rh~#IAwe1-2QlOth)f%o!gsQ2k#Ry;k!P
zE})vPJl$-Rt#pw+!6_G3zfl!r4EzgUdDx@6uhlBowL6!2g@pL1#>?-L<h!+KQN2j0
zrQh+qJd3V!q*f%ekM?Ztd%PSsc&Pw;5fzmR0q;fL59gQEWe=H3Yab8OpdSj35La^O
zC5kOs8o~>6j?&TPj?$xVzokb22%(_fNNj>7Zo)><)ES{#F#PBje}<@XK)#4;MRB&2
zSZ2#mA4F&Ip>?OQgC-Cf-$mFn<pK>>km6j*NbPWT`{@bJ(g7nGswsEQocvj+0T&Wf
zN{yL5cr9bhckEO;z)^Z(sfR11V`8c75qis?_TWAC-d!Uos_G4X<mth-rlI}Qd;Znm
zs`*D&d3dpSQt+|WYHtc1fO%|lKgVFm-EAT|2>L+&*WbSlSbSaW;TOa)Is0pDih4}i
zi!;l`zB}*h1AZ-bk7z}5qqYb%L2I9@P{Op(2S70f^4G|fO4&%nsEHpw1y7yWtd>&N
zq`C4j%^4La)vTZndK0iBpz$row`KRA9ISRh*<G2-_};kppd(iS5fhb`H2wuvURn$i
z>U#^S^3awv$9SakD3%$^ZAa6$Myc5H?c%1#2$|}xLku^hX7hWa_Mi7L;t!9Hv#nt)
zGkKwdM`uvM`-KFC%q*1Lh~*K3M2Nao)_4)D5BWTkM@hB~;cb$j4L^c9vMfqwNS8=A
z_2b(tMz(uUEctYR?CeJdh)p$N9esJ+a!YOpRzK}drBG9|_575}SdF3_=ecB6WKSO1
zOD($b<DH`*ky2Sni;L${S&1(npUdMcifF5(BhYyx=KevBRlm%WvHXaOaA(HE84VcX
ziqAb<!I09W!su@mUO`96q$s#kPOmjjW<b))Lkm7bdAJ<#AX9Ckwa>PK-n@TL&dtlm
z!}B(Mca|PBK0b4uPpv39T!GS%S&AjByQ{vE_}fRB=M>X5Vu@gzU(;@-p&@yvue=~w
zpR}yzyPi@6Hf3ZryNR`K*LYs>8a4*FK{j@#r2BCoD*O5587_}4AxFI;qa9wc%wGhq
zCrXR*+8ts*gawsVN7uOk!3-hGN=e_TY`#C6=Ycr19Fgj{^8*!QxGJYm@@P;fR&Go#
zhB-!6X`%NcPvyKLEq+?Pz8oBZh!JGI$pG}>F~sF3SUT*n|C)L`6HYG&wxrQ$J?x%%
z2$`sXQYdAF=ls%h=*NVQI)W2;e+m?NXO9L3j3N_I|0!N#pFTWQ8oaYYS1UEO!j$q-
z9lp&nuuLrAR+}mbJGxYd*-s?hcHWdN0<Bz#>}a6h-r0e*#@MaXS-sZwW8dxo!llDq
zzbavl&A7G%ZH-Ia`)4Z_boMp!bCr6Fpd8`v9Tm*$`9q(;RSOPjX!E%|r!tPy_T!;M
zaTz~ArwVG~_rz}+C$9=Ml6-Q6=8!^Dq_ttjza$M?X`wLeJyumlPmoLRa`!5QU`AR9
znopJmwamk@cqvVD1o5x#RAoiDx{*dAyJ}I)?AqqD_D6r;V>&&?^`oYVaq{vppJdB_
zP*IgygV~2>!yMP?$lK1AcTd^>wBW6kWNfVkpdDaCGsO~p=AMo0z@MzYkE}rzQKu1D
z#GVdK-Q`4NS)xxyPT-A@9x#6=@D4O4q(lji7?yX!LGb=3nf4tc7{he;Z>dzBJ?n3p
zp-sh2M_TB6@A59C{iXf2k_`Td9?bYRln5_yGRz<()PF-`oM7M!TH69*w^3qX#%zJX
zQy*1Q1k^Cx;f7i}Onv@zQZN<=VfBQnkpJpIjMx(hG3f#BV1(fePK^1IRS4-2vkuOb
z(M=dOQy*wx2A|D0`4<stG7Pq`1Cjt!P6#qE7MtL2|2dr`-Czq5mNPi9xr)0duk9$2
z5hMI0cdzDqFVZnE<1JtVB<BgQfIm@DKXdI}?p@lW8u(D2%8*lLfQG#BH2A!2q$DkG
zD?V+eLSgPi0RzF7P}33$=x#54`)r!n>kUbOHJ6xHw6CVw-}!RU_DY0Z4%%19Q9E6o
zO!<6^i~X1=o~c0QQ3BGtg}h?ocy3)7>FAh^z|lpwCakz2)uEgR;^eVtfCf8aNM5BP
z%x`Q9Ml8grjQP^3-IMMw%#d&3{C^f2Y8cG0ywN&#Pdte@iaDHUI<md5MsX8X+H{4=
z?#c6vgee&-n*`-eunr7^#fVWGcL;;KD7|%DjmRp6F&axI4JEj?%Y%t(y=P%Q%xI2C
zqMNmcc{R>ISg@D`Z_)vV^HfF57FZl9E7YcCcw{Wqh8PnNO&Qam2h2a~Kq7{@`5?4u
zh2h&piEh+DQ-hd)+S}*og;zS`aD^K=bG!LswBiO?dU#x=VIP0NS<k^t<bW(f&5@F!
zCVGibj4Y7;^H;nEixDfu#L39+gQO$s#Q4dG?t{iV1KnX%A*=v>VZ1ROQVP^0ufU_2
zTfPvgAvUI40;vO1TGZ4|k>AwR$SRQ)Gx0KY3dH}pH(Z=v7**i~;>u+hb7<oR7_C1V
z3~3>VC9g>5(}b+?tP-Ncx0L`bXd`)qKR06%+BmvdeV(j-u^>0}O0^|PQBm0W`
z;u|E^%fNK%^?)*wh<VxH9?P)sVBPY2Wm<~4%>!mZox4`ySMZV!v!O2X=?UDGRpB#P
z0~R7wnJTcW{n_zf!kH48e*dy7Qri$w99k~w=tNa)IOCH3M~>OXR@w!`CAb39g6l+8
zBK5A8-x$ByI7nnym7*->J{KmclZgDNp%v5NFgB$NCx9;Ura$586L2EbBpB=%6ZsR~
z>=v-A^-0l(4!cTm@t<Wv#2iYD+T`&gHModW>^bwQx}$iJHr&%%(rI$F_xH^%&c1lN
zj3RTM`q?i@sRCN?P<=xB9u?Mm8dBI77mRVpI?NJrKLga<;q5Jd5(OQf?MmOjPp`DE
zo_(|WcmxU_KavK5$X?vtoMieqzqZK~ns$4*1xcW(W}~Bt2cPeoKayA5k4ptH^mvE}
zx|vPUX~>8x%MRgQI*SF9yNJT0(PFS;V65_p(ER3_gAb3*F6KW6<m^|v<{2ypKPK)%
zJAYpzSA4?~HS$$RjlrloWmp%K`LmRMp2;qj5CYszc$QZW*xDZ;lOn#gSh>`S99eco
zB!hc@Q-#P^?=gt;e3!Z<U(fa&Y*dvGswQc5QCxBb@o8I;j8bw>%AKZ#_wgBlk2w2d
z$wcOTm`w)1>=tS2A+MbOkde|dT=J>E!-jQw;VcZ^5%(R$=2~bJOnY?+;1r$?UmSkO
zcm^yBB9W{oP<~PVqf@2(d?zTj$bUX#8Tcp#LMxP_garsOeAwVf#$Hc$>lhAHs-~PH
zfwL~@7v<q0=P`{JJw0DCza=d|fgC!Dodm7$>GNm)9pmW8t6B9r$zr4_kFpPkDo>sh
z+W0GZd`z)y>gH5H_(nuGVZkN=Q*Ts^3t+>TBA9<+QshyY^=e7I@Z_4c6dS(Fb6#lE
zkH4;Aw@^zw>O-1bd@VjdKGg0{?dNN@NIQQRK@-B5rMok6*N!BPG*A7@!dE#G&(B&u
zqLj^3I-(RYM;&?C2>WdVjV*r!d9*B<Ofrzv#C$5F11BAgCO>ah-#z$SV}kL72yoAS
zp+A*6w^<`;UHWHkHo1CLq7jt6J33E0H<qbvEZvk?OR|;fP?lzs_a!xaXm+ndHHxnE
zv{547P8_;JRU4$>w_9mThn#xE@HTz2es0d5s8ynJ#(!igS=qg6c1*G4&;&}9<tO3x
zDQ?3FBMM>UPU34jh>{pNF@ux|H2<hZO^%qvV1AJ!qXiQQFOY+FZQR3$APi!}oZrBy
zCm2A?WrhsEPg@Z$FthlsiCr)!hP^jK)~N1UR7CghL}G(Zr+;>U9aL{-V#icr4Wylj
zk1d&Ondi*+Z~F6QLuzWRUPBeKC7dN~mlX{9k~lbZV1dl(CqEcwT(XGY0O-c~jI$6G
zz{-Xse&h<<I_;xLh^9Kh1Eq`sL;d$ALt7pPle+)nivfXf$Oee>JpSDEceMGe(?rK8
zVk&41U77y(%`VGWRT}xbuIc*m1Qt}v9~rXl(6dWR&HuDIYoat;)w~t{yhch`-{-h1
zuf(7X9y+D6o3}wbIHi0Ez}{#)1hFW5ZQ&-LK@BqhXP+i3ky&NW&`sRfUD_4*QGgph
zCv_9yt_3alRP3QNSa7uE&y}^_+{tqUuecs>Meqi;7t=?hbeCsT(16wWs?x}QU7K&W
zdSc&R6wm8Eg&N0XMs=-;9n}S8Lfo7~i^NP2pVBSJgD3<w7&~DA046~uozJ2hDKhNE
zuaH0)&~`cZr7LwIfC+lap}$8HFKUcFMgH$H(f9*{Zna>BX18O@e5O|UlLI8v@c(?-
z3Qygy4iZW@S#XexVhoTZEg=q)8JqM;Vo9UM#>HR#Cm?}1*NyKmIr!^!Mw#0Nt_X?|
z`W5*J!2c3N<yak`yh>pqCvDiL?Xo!IqL6N2fFFG4JBz=Dpt6zhepDpIQTtICk1I*$
zj!L!fW{sseN$EUscChP$q{)R^Rg5w{<pesoVeQ*YxZy~N1mX};>em+(pWV%{;1>2j
zms(N#|I|}$avBe$J0!P2H-x3tM$zS{<T4Kbr11G=@Cw^`ye!f|xmvy;i$AYKh|#;N
zwfzV(kx;C=8z|*o(5MpmKVbmVV)p!9>d8-&q6O_0q<GjJc`7n^pD_hpYF031pCzK=
zY&gYUJvS1)_=+a_6h*4{<HIQSI@yQ^7s%b;v}I*i{U%KCeHiC^QTWi}uxI(i#CO-b
zUN=Wy*KJ;k--p`O)~ZjFiG2gMcG+%##qu1w%!B1ek4rvwocsc8kRl2@?!2k+7Diwi
z?R=7xI{LBId-c5dhsUL|X~9^$*8RgS`Sh~7uP8po3T2B(dWy)vagLF7+(lNbJRnD{
zsKNmX9Yz>qt^I8rw%N@#fyf}nZh?4kM(uzr?%E^^OsJ`9e3O&<)im2|Hydl_y|yMC
zT|rl!XMRKi{aGtFijr}00D1-g-)n)HRH9MQm?_$rry5&!C~7H+IF=JxqshV1D=ek;
z%=P6DGY1SOe1&DH<{;Wvq{U@WV8D>L`LxBz0W06Ff9r2r^HOrZ*^ZHAZgE@4Dv$Qb
z+_`$o&qOG})0vo69x#Sl$kPlcZmP7G(z}I@y%=jWyjVy!ym5Rpz4zaL|LoaZB<SXa
zx;p!&=oWKU&!bN-&$@Ic<+wc;MMkO#W@T6!3s$QwjELj?Ex6~lPYD&41ExoE;7s!x
zP?oMzwi%bty^5v6&4Z|0J5LdPwaL`s<<Amm%XVcm)VpLKe_x+h8G^!xUA28KoiVGp
zbX9}?kgx0VcE0l>p88uaDh9<oT$5LC%{RtEnp16#0$1u#v~5B0jN)w=9Q1Vx2~a3@
zx{ec!LfW<yjN+WO>lb7|0W3X;Py&?v(7$+l)}A+lz4&M+Duo^@(DHcXLkq+_P2j>?
zLphgba|J5*9I_?*ONBpzxGq5<6s>ufg|PHV!h7-Uwb5i?dx#+}f3`8TaX|TAy1ojo
zeQ<dD_}zbPyggmFB--xQt16V8L~$S?`cq#2-VU|fxWmOiG2He$0VuWf%~aO2U?LMk
z(06(6cO0&&(4<M2-DC7y@1?k~M3ukyfB47RN%cTlJ9!U9ITuTG(oTS?@f1|Wr1fgw
zWX@5-(*6?uT`_UUb?%0?UN<W3(Y;{H-iT|m8Fn7qWbs?jDSGuc8isAFVm6L-4Poyi
zQAV8J-+7v3?1dsRK0pqOr5?csssBa1!3s#GOMYJv?B{&$@1V9nX}^wMwC{JMeM38I
z|ASn*KP8kBNB5<yUH3}9=<+*XbtSPrw2sa|u~qL;L3cN`zGRHf06zCeUygO}Idnz5
zwO_=vWVNpfW#Hy_GmMJa96azpL9EaRfON2MF3p-q-|6OH2LKbuaA~M(7Gh)w53Yl=
zE7@Oc3@rg;YxGA8@^_5Dyf3d*m$GJ#P*vsF8+_q)@pZ5+)~K)XLaye_$z(tr!tRW`
z*aEa_r(`tTv=?*7zW0b(4RRq1?nVN`?nJ;BfuZQ5Oo*7W@$R8Ko+!aDZh}~`MXQr~
zIwRfUC<MAg6d=r+k262of8)*4UFOE5Tda=+KG&j<1!yfR!Zb>Riq;GOb1Ecf6r=#x
zP#P<Am5{kM%FQuijbuBA4UjN0r1n}Oz*Xx<5hZ8n1nc_sQ$yAomtG>69^86_ga%56
z#l4i_#Y7BO4<u~CgPv{$)N-;(btM+2x(dsZLRW|U9q5+$N)1rXk6f(!Sfi#))u~TC
z2s+Y7FdW~Es2&ql84~ffvf9P3rk$5aEmhgKY?5`=CbZ_!XMQr9JdMz5dxuNBu#48Q
zfRm_i8&$*HUp}u4EmWc7xHM+``8u4v9w)-~C&Ec2W2^gvs(DC;p`xlimEih}S11&X
z{vle86Tma{x?`e$m`<uKp{C6`M5g=D7@Tbq6|7X@*Bj?znAcH{OsD5o^A7VCv&Q?k
z3vB2s#Je{&cFT{b(sIuzCL1pi$Z2s@2AkMieJ$pY8ZHrj&vX&&WYO(K)6Mut^7AI{
zI-KuEZ~JSqrVxvX{?NcoOl@4F#lsC4C6xPV3LHetCvPz{;Byl+_K?^1m>di*wsevE
z=`dU1yR@-b;A7SnQEQ(Xi6fKq`nP-oW;RwK4Ntqqa_k5L(q=XkvoW);@(jn?^sV7#
z(~Cg((>fWB2g;7@>Y1Ub1$2B~<nxyy?08zJICyj%Y2;@$2z*ERZJf$7%m~Xx*)#tf
z02~=+x!=M;7Uz86;Yz2B;vV5DGAA4Ly#%LR?yK@~j!;XL4o8r7M%c;}O$@=o@iTa1
ziGnUSK+6=1bF|X?wOo3=oz7bv7v7zjADgYJu8i~8o%5r3i(5)g&BO$3-jc427++Z9
zxF7pYY0$7VRWk@<HU|IQ&LvbZez+C^h3JTW_?t;W=Fk^l%nv`w=3POwS@|@}c4v>F
zFGz(^{W}Mhx;8_;$b$^EH7p;}Cg8tw37b;@L%DO3sJ?>|*kLm#Q6xY5`q`}?y-87G
zu}FDySl*lv<;9cyeq}4mKd*;Q0k+B2%|FjNCq<UolxFK@0!M0^ZIR{qo+6L{HrI^8
zzW7)F5v2M0Kw+^0(s^3&meSO=?E*DS&asJS$x21U7pSx@*(N`7-Nck?rsX@_D<L$a
zZ_R?ZxB1{crG<k*VY2$JG5$I9kw9DzhtzHiwv>x2K~~9HFK1zYjeKSQWzLtFQioxy
z<(N-hF1L6;;p^4>lms1@G4vMzI+}LTKBNoT6GXkrssK@{1m!4a{w=a_IAbW#QipD{
z;;q7GzbWUMuocqRP~$W*4QQ+j5W{<IvxJlO%NHDq7DTi%*{g;MAhl#{<y8!rGTMi{
z-D6yK(u?nic{GaXxip}5gV|q$T&z}l#esW&6sS89r4n->h#Mn4Yk>y_K6dS{Oh-Q4
zE|YW6N6jw}Q~BM!SX@tD-+jT$sY?}Al2l=N$E;|L8dUYI$(yVSmZgqt3KJ_(@6dC_
zKN(uk^gz{uS-p)q#;eOjUXOVynhUgLrI&NsA_Y|wDqP&Br0gwMlwO3n__{V0?o0l$
z0lFjv78rcs_tgpcO?H3=BOa!04SY*?d|9*r(XJ7$h`~ha7EXa>5E<bmu5p8)y309%
zF@bSjs?k+#K`KxdCrKzP35NdPH>yPkWM8R|_RJnsNw!y^xguHi%TIkw5<w#E;X-sU
zw=v*0t9fZD-WQYtLlSw63i81|Fe<{V<UV<`k~QJ0Y33VF@qoave~VtLGN~qTSuogV
zt4T$hj$pONVL7kt^#{>E{5khg0=+xy+~bGbm0SCJ^E!wFG2Y%_*xIBAU(fScVO6T{
z+gkUn$AMq=GQ~u51v>VF?Lo8SWw!`v6vaZRzQAwqNzWgb+1Q5!Ja;j_q>|}PYwwFe
zj=L+ahV!EA^nsZoE*OJcF{zvvM@}Vky7z@&qItoeJ*k|SA6}Lv1&M!b-2+yc!5D@g
z5<TzTQ3!rfw+LV_7eTROOSd|x-9*#tu+J(hOvx7&gQmTqmN0|s>8#S{NqYEudAlzQ
zrZ1gFusCEL|B_;~Fv4J)RRal?^y;H3$!!rSCWv)zNd@LLM2iUm(}bdRQHtEHh7Y34
z+qy;Ny^qZiL`_ie1PDiUuBttGnN{BxAdEE-ZZZq^&!87BXP&8B$^F>*7B0nYKx#qR
z`A{uinkOL5PbS)T93i7!yovF$Ep3Y$wR%>H8ei~25XZ{)+~!)?MlI$?W}ZP>$vxRw
z{`En#!~h4NpfJ<J5a=7Af$%1UKo@K)EBbOKRhg<g@2Is*rMqz?6zC!x2bw+%;uP62
z1C<_=-#OktMm1EjFGWc57sYoYkjXG@{OiQWMEJlEOl*)!LX_hNt<-XP0!H+_xTQI`
zg8iKDp~fUcTm-+kzPmi?Lip)4w4c}nZc~rHGmzl?n>3vp>|xLphR27QMkZ%{!dqjJ
z?A<wFcbSR2*w3$l1I!%aaEqj6n5AVp5bcI6q%48h8bn&&N3_5UtR?Yghu5~^Z(f8U
z*#nBVmN1j4ka?4hhK-`rYUFb%iXM%15ue0KlJB`bQC>Fo(fVbkl)!yPj_%MN+`^<W
z8F2Jcd-UhF|1Y8hj!K~68ptL5lHB2awfc0pcZwX|G+%17RtXyIAH`aqb2U0s|60XQ
zsiW_V9V#3izUOV08K`BIby;OqZ7BEZiKR!!Y+9iOZaHD$o-@$#x*4Dp5(7_b1~Tvo
z>-@XQ^bf+O;Mm}aT62(M;>Q6KB~I{rV!#pUpC;j7&9*u!YG`bZR2-D|SJ>FuP03Nw
z)zk3Fx_JY&%1(Ec9_|~RQJ8wvbS8C|q7<FP`q6i>xzj`}U(9?Z&+$r>Jg@85ZJiuD
zp`vlkigWxN-Jv+w5!RtlB_T_{P%Jk<5OLVG<s#Z@I6|}$|0(Ke5ZCD}0{JWVE#M<n
zUU%UXk+M!ocT8N{dplkP>itAo$>|6UOK0s_b3h0k?LV$EA|)8IbS>%zPYj=!R<BE0
zfd-rV;RSK^YXg<#1-^NyI|%v@)e$xFz0mo$d(v{56i1S)I^3<BMw>4idKYRTF|krF
zi9@;v6>42#l;x;};gEl@@3kHQ{OvX71L0z<v9M>ix|Z<5rw)E7tEW{{J4qKYZ_UyK
z4E8r9sz7do=xog@q_Cbs3ICfg)>&^92Vo6khQul4^w3F$KhJFUNV+ZYp&kW({-4gs
zgQFXm@|^OX(0kq#Dpa}5<Z|5q4vB1qaPTvQz*{30Ga=U+wS=Ks!mvXC+K~a2vpdrD
z3B4~Y@Pzy6gzso-q^7YynV`i<8$pD{?-2;W|J%4Lgy1LUCoRt^kblq_gC@8J>gC$#
zgVu9RRNw}ge!4iaPb7NID~kTBU;oKL3(!~{akattFzz8n3kLQ2FtrI2l?)3ocKYmp
zA_<(`wy+$0URWd96kxW{E|0iGQptdXi#`>XIx2h#5(w1~Pf<=4&c*ykjy{A1Dqo7X
z%fc-7(-m4sh+r@tQ)#ug)~AoS?*?JmiHKyp_tG)w1LAUU3+aTf5J%!TmeOLopDg$Y
zQ|@ZZ?ifw4%vB5_8yyW+Ve8D(6>Bnv-B{#1SCY*OGx_c;Kvx0ymF!UNDSMQ6C!E;7
ziiRZ`EP`7UOp5&yQl4ZgTswbnV;wf-DjH;(&HAFPzUj1G<-=nyVDQj_3a+dSQ5sb+
zPwpp4$bubO@j>Pc4c>GxkKqNA>z-?qyZ!6#+BTYZ{S&QIf=HDv^i{>46pbxW2NDFY
z%k$qfu|sSE)m|>HN>dX$UTfIC;B#D;3dY?ft@#zpStACHNJK#_N^d;4)}1E%7+LvO
zv-S8PoybQd&&NYN3l6svT;_ZCJIqOx8U8i4!7hzH$6KUQ!8qfBCy7(xZbl~a9KrP<
zb^XW=(uOZj)DQ;5Z-X^X6AwZ9zX*g4Fd0pU6W5f1+hoSBuT-I@27yqd_3XJb%6{^H
z0kf7N@;jy9r#ehXWgpcGW;vGV%C>j^8<CTNpI5#PhgfagC81HCQ4k#KcX1v6x^XYX
z>dorNPxAqUWP7Qju!uP~Cou>-uv*6$=7+Ij)ca&q+<8io#1yt}W@9$7Nt{NlIF20j
zE=xt=H0bZC{Jc$4k-!ga<@ztaF7C-sNjM2OKfzGO5#4`;=BkzD3d0q$HAVXVW!QWw
zuOy-~`Ygt3o^|yKpD?KW`z$vjs2DUrsAu0=oOsF6cs%eAxx*(<rXhvSY6Y1#b=#HC
zp-#BOXxpE9sL+LD(ry#~;AV{*K8tSak<Kn)PTKnCflL#FVvRL6rmcDdu$Tv~OGDud
zANP(dGnjim6fz*T*E`abJ&5Cu{MQRI=jOEMRpAnpA56ak0t3x+esB=##os~ad0rua
zDMnLQc9Kf+0p1TD3$~WDn9(=Z+89I#JmmXll|~v1tpLXd88f;T8pJxZWs(8#jUfQA
zxc4O&A)Qc7W2b1m&U}oe{=`QS!@2FZhgm#MA<qb!3tPqloYF*OcKsW`w_GmXC=BJd
z2lK62q=zWtu9~1uC>!ojzFRf`v6cJ<27sO2twn5%IXju+Y~W!QWCQKz$_Fg++fQkb
z|3}n22M5x<L8GyaO*XcZjcr>S+sVduGO?YF?Tu|`W81d9F>c=9z4!ZSs-~uD{u!O8
zd(P9{=e<?<NU5&#bI3b(+7A%W2R(vEyrl&G?MPfuuH;R1PlpWB=!{dTL5<gf@KXPU
zAlaAlxFH+dViupzi83R*X^Tor3mpz8kuS_-4WF2mi5|x7cuBBEEq=%|*HW*R@uq@z
zA!I@a$ETVaoKBF9g8pV|w7wCI7yF5$*j!XCzbb!+6FGX3A!HzDdA57EdlIHBdNKv%
z{nkG82q?my8I6O?W6XzsgRbJ-0$F$3BSyysoe+H<viZ=QI?CX?UFTmu4Xm{i!Yze5
zd9yzLVC(x-;0uiGI$`0olB!F2xexy)09dvM4^s&Wkgnca%%zMLOZ<eFqnFKgzYNKD
z<k^UTWU@B>B^OO=IQqjPOW&#W<!}v9LrkDa-rxw25^%_IQSh-0rd|L3FKS5fb+-lk
z&;Fc0%&AGeTA@B$RBRhF+$LQr>FMZ?UCuwWq$BxNQEhb;ZMS6K;=@Gg2v}U)IHC=5
zcVB{bX_~{`n?LtGt*XKcF#miPsl;qlo7YmZ8mOoYh_^<qYqH0I1*3=0-H{XlAhg&8
z|4KQ1)dCzan`qZSK1T5_j`R}?#($r5W$Xhcy~wnaFzP%S9mG4FGRn|5Xt*Xc;ZFdJ
ziCYAYt_6Wi`dh>syx&@%<pyR8g@;I&*GDr*$Cq!I)6}hMYU=r4(wTu+gw6M{=b7n6
zf4gzyc;PPqS&Z!<uS|Amndb$>a@};n^*X9{lp9IzZ^W{z0u#%LoC@OfPe*<vlIF*q
z5hJnecz#`;RuzI>@J-&(;mnHIYv%u^DXO26Hgk$sh~;`|wZ#mQvF#<y(h#Q^k;)?J
z!3HpHddyKXxrUFN{q#}s&%$`|<3ca*G}jAU3QJs6@E^GxZEaN4fc=y(cIO<@C6QPZ
zFhu`k*uUi=uXvs5!*u0Ptc@!gig8@ix$rXyAtC3`g(H7OY5t%}`!TO|j^Yw;mGgDA
z|FiXi5AQtf=ch*Zl(CpjZD32ni0#jkUS~YW1T1Cqup;q>uo#z&1TTVfb>^X!2t|ER
z{2we|h?WfVxrm_4fcjh-GIdK5S=qE-B-jf=cCzhucPYsncwFb9?p!B%hy`lrAlirw
z@m$SO<uOsq#Lx2J$FA9n%}nV+;JBmf&})x6saiiXPW~h#SN9k>%UMf2D)sA@;&5~j
zfx=69S<*VTPhqB;2!_gzJ7*)Ab(volg6TdCWBrzUf+KOH1aQ~=Fk&6f?Q#>8S3>6r
z9e>^_xu9bPmhZP3ctAJ?%C>+er;^?}CW@rOh6cxDANLD8iZYt@5qFC=B(b6+wPt_Q
zLH>ftv7{aPkszM7^ZQlvELwZg_vf4Qb+Rvt^h^JdU(a>wquL`1_HUS8ln2Rw0R4qL
zcYaX4F0R1RHb+z%Hh_m*r7*{6zIRsZznB>Bd%4$_DYs&XeaA+dRUM9_`0sh3?>jOJ
z{nk->#~xj%`9GVfoJ<bw3@^z^Y+{M8FN#q(8}^cEw*d5#7N1yRvNF&Pj&whaJt|JY
zAxuXBx+?UG?i`F>s~uxH&G5629KfFFxJnN3+5nG}AB<8*FMa3P_@z+H*mM#^#t7*C
zYHWcJFO?;t%(WNX`HK4L+$7a6!+$-Br|R+;$24Sc;c)9n17u3(NWLCJULt-gL0i;+
zv_lqqQzq+N$?f5cfU7-#k}B9KslxpO-^Ov4{}h((_nI1?(TAby@W8=Mzn6X#f#tD0
zAVv}O9}|O{{<eq~<-LYj(BxCrMdd$&*F;096_q(^$#T0HoXLd_<y5!f<YSsECPI1(
z?d#Cp0?qQdjkC1yd$#2oNo5tK=1)VV;n;$i)uJKSl`44O=AU$96YJsSx}J<O`;;!6
z<5@ddXQO;QDX#I)<lW`}rh|lcI(L92iflZL0W1FYIT3u+44tnYf%(p|8zQXEgkq>8
zmq6?IQ(A%bV(=U$w37(-n6ZaXvcNfrX;@pt*wRr(+ZeISZB^r|ZM8+5r&w5_t<ssv
zqlE&Ujca7W7fy+3^U$ZgV~|!Sx3<3P4L7y=$gDFw$mvF0tR2&MJ8Rr~c>=D-^%KP}
zVyDU4Q?KlWAI;@)FT~w*mxS~ESOJ($itQ&Ji<^s=i<=LxkCqvbx`m-XH~#=mpud-&
zdBq$~6bPa=GW;}+3*?o%iCS%VwAXkl%`#~_&$S6t?U47wh==O`h!Fg-`)Laa74#x|
zFPgZ=DAHFno4-Zo`(;g=xdCRTK1pKlcs>`@KmA*gd&fmc4N?Oh$mlQ}4{OT1m>`Qs
z+*m2D%x#p;AzW7T-PXLk@<;l3Q|BJ&9opA9JWAVMo4_iu+%bFfNyMwXt!Gp?jQ1h;
z{TgzUa;^(qzDl5acuy5BYwI(MbY%LHUU0+vsyS9QQiz$8-PJ~Y4-mJcF|`CKk~aAs
zTcov+aK|u3A~cezYwF?qZhU}jAw+0u_+_^WFkQRu8hMV|ry<v?r5|b<G^0=C$6aPP
zOI(w<`37@?`mYrNdBupfXt&3`*i?hv+S@O%xOr0q4R)}{3Ga$0C8AKKiq5Y86mhma
zl&fo=yg4cgJSK6P6u9rL|8i`AT>8u5Fy^#=3$)#jXU861>2jmfDdJ)(^GpBZw=sA2
z-ya4I2w7PkOfmtQ&h~ktUyk2j&K$%Q{7-pb`6e-uZBc286(idg=<E)%xw`*|L_K-n
zi<_~7*!rLtHn4&1N5o=QM`PadIC-OYXHqWewMYagnx6m>hrCIU`ZKBV-@Y@C7KI!!
zqXk(P?xW+mjXTikw$$F~HzBr5OdP6(Bf0t~SlUli{P@1%h?jCcKbNwmQTD5pG@3zG
z>Ifquxe(Ma*}L0(tZ~w^$4hv9)Juc%!BM0F4dsbQ<yXdM2>(()rF1>idPS#@HF+br
z5wg)@bLS{<TQ2!5JUk)AUV3hCG!2VKLLrmY6PdB;lx!JiR%Qv84fTu;DfSOCcpJ4O
z#x27H?WdU*If!?voXNB}no3R-=uq7@2?(sZ57;+m@A!ciT+goT=7Ji-Q;B3eEIZNk
z{{DZa8$WZjP@G_e3GdUt6Q&NADt=T;l163fkA)or@9h^^hlVSy;(?inqWmL+$}tY>
zyPu=`Pw8z<Wk5*$j0C<!E@#~vjZ3Z!TS`P`!LG0>ck_MHVC4->))_Rpw#j_X+Rr4C
zUPR&6zJJF1it$Ui)|53kl*IY`@N`=Q{s@=}1OL{Wrw`TducD0*;g?lWS}A&rlbkok
zM~<!l_Rnj27`N|*%7m$$tYAjo-OZI=A0Y=(&(&ylz~h@8(F1zzbLmCsj`KVb<n&ft
z)k^!Ns~$dbs*2`Yf!&Y271ha=43^&*b6J&_s6q|Ih+J;bL)iVDK3+u1$$f?DntW2^
z?3FdW2cY9nH{OPVLgNGf_6K<<IAf5Tn~;!!Id{=p{9e*|@|(E2v8Z5Ou+f`9(wpn8
zC3D@oRBubd_f8+nsyG)mr|kr%Ol!8-)zXDp>V#cbjNj7ADqxUO?*<|91M=l~XE*DZ
zt5Ajc?zLS%9rV;p-_@B~63-@;uFmV6#N^vju=bDJ%r`T0{&^l9$~{te)`PSNO?eRm
z_TXw=H6u(M_(G`ueQyF6kqSrUR8l8l7Phv8mz_u?)0va^*I3Sd?qw2zyw|sa-a<97
z#4#w<c>Sgc%M>K|@p^KFB1L|dJu_t8f#PfF7eh*)Zqw>ZTDRY%aLf3h5pmG<%TRf5
zyZ`J#h`-;Nhqk)y_`GSopDlU53)~U|8FyR89$jrMrM{7<8x#7avEf)B(|-DTly;Km
zH~W*mioMlR=dy3oI0@6Z@kE7-&BJ_F$aX%MotYRJw*5Ix2B1#ASU>Tg^CAYskRuk_
zBk1<E2X!<bxVL382QTwqALra=EGlP*3GX`M?taJK(Mg_}2rYx;@tA@O^yU=;_c7UA
zzo<-*qcBD(fKGTIXH)m{u^t=u8Ta$FIZ?rhpOUf^D(YpC?-PYd>viU6zR~mj&ZN^$
zil+W~3ckg($iB{c^hvoEB+6B5jF#eb<1SRNNq~Iogp`%2rKF6Nb#ih?lCvPjTuw3a
z%0<6W(KC^h;*By^-JYaiO9C0_o~re5_wn>x<Pr;yj*E*^Bp1(Q8qivy-egr$d_8yd
zh8_y4fN)?5r<IMy>NJR7UYzaQZB)-|!2hKUr<agQJkF-dlqYVoCpi}Xj;++Zb?hah
zt=+)UtVMY?Dxft)0V2TXA>2#9*4s2yEo%$sZR_*iCS)n@lT`c2v$h6E93CQ!WQ2NM
z{Q4#ru#es>laP|qp9oI|i`My`F(G1GEvo$*Frnul8M)emKvX|@0BY7UFo*4=-G*AD
zIPWdRe?teQ<Zq+lFQ;-ej```n_L9CA!9uzlTt!-?4n9cA;%oFuD-8n^%Bvy)qWXtl
zayAHAC06{Z_fi2@0}K-|RFi;pss8t8pV3~@xDhX}78{5Eb!qv**pN)Kq$o=)XW5u_
z!w5SK3U4s)0^;=*+*e%OCV^jLjYy;#8HuEYf{~S{kEaQ?2J-5m$<0;!U?l*O<aOJR
zbt=V*zKG{sRa~{h>Y~wYV7g8dPNC~3qy7~RbqDVeZr#rC9=i&_imUN#U^zDY@BQ#0
zuT_@>C1c~#G$qqTQf{7E0bc>P@L@o=Y2FmAlJ_h}1)^WMgK`0M|2#q`Q&%_3Mcy#d
zlTdD!-0Uc}{?Zh>O?k>z!6Ugd;q1hN2A;X0?^Q*_&O}t_Jo0sbFZCIE$&&G0hl#%z
z{M)+P7SXnzr5`vT&PkDc16RM{*=#c@D5a}2^<kPkEWGLGXxuJ(Zi|lwZA*MMW@vOP
z)U5#a(Bpj?M+^_<IRy6M`-FenkA$R`1v6tyRfM5sxNA*=vU$~%uH9Y)E^H<b)hKYF
zF);sm|1&0}!B45Y(akx5`)3m`Xu(o6e9|7YjHE2@I_VZbp!|7bKI6CZmm~za9-%@~
zG5*Tc9eIK;K?zM1Gyk;c57<fYu=L7{CCKrDJ#z>{(@XLKWz#qMJZx5?YB51!E?rX+
z`h4&iV%=K-;UMTfV^Cg|wRtE;)G$NQ1ylIC8@nNUwbwemyNCEEjzo<2jOn{f)qBkT
zB&Zf6B^-9(6*OiGWc1byGKsA-pZ6H}{=8ZFH^c=D%M3fkRQ8A1<6q1c57elycnkc0
zGZ2c#uZZpM9~&_q_g$jx?^s-yB6RjMeWU5jex3|<7=nHI`4;kHb1(X)xc`1hj`Zp@
z+Ic8;7aDp?ouuAbghIu6>v?TgjNTDjzHl0Cf-sZ<rTuSOTewcJ)44(-x>kzVB0~$&
z;In6X8MtNGcoEQIAHfyRT$05`g1szz0)qxDKhNR-i}HkJ@W40W^ITA47}3z=7n)}a
zD`*KeM{a4Pm?B3}3OD*a{X{ctwtWSd-V~whqVI?>L8q`Zr~LBKd)$RG{+EHfaKt3U
zb(VTSehc{_$1^YJf;z)A?@(TW;Jn)KeZToWLq7cXTB_QXh;`{mrJQryX5_6RFi_+9
zxp(Rzns3||hhBF{W<rESY<rD;m#V`&a%6wc*_zWYa->42b%am3Q=n5lM8H14>CEP1
z9)R}4d(<~q>9o7GTJvy|f$e5Z<?w>C&dyGtk&??`<Z)*n$&TRZG_U+z?sy{#yvg}s
z?NHv!Tz4l`R-GT6uJgP52V#&S;U<3rj#u}D*w29jakrU6m1sdr^ycBP7yo(f*1qpS
z9~1P5)@LDB)d`4yvlp<o_&Sm6PHidHwSCSJ^VBQ^rct<cRzV%CQ&4GAJI_71U_EdF
zB{jX@isN>Nyx4;|!a+LoRm0c^_aQ`uJDey}^<v%85-YMlS-E2de|L#i7^Hh5@Fhw9
zRBz2mE$fP!F;Kp_^T%Wy3oOI78~;2*q)0{6k#cWr4Vhe_fMj?S&c$!cxQf{#oz6yN
zZFD$*UUh1(g*UD0oZGB`NJ1=?xk@zvHZuFT;Y%Q`9FBj!OZ>1+SpGb9NE}b*2=5in
z&yM&T;mn5$72a&4qrp}Uoz)qRb63Q6AR-mh@EBo`2yx<$D#Va<c9ni{;WF}1c14kq
zApG}%K<Sku479kQ1nJR?J=$jmV>#lnj|kE*K0Jp3GgJ!^L0|~aHP<WiDQ>?jaFu2k
zXeBH@nU5?;ftK|qwlOqdGtnT3`%H@;<4aj?=d{`0=t?wbPzcM+Pe*~3rluM<2|S|=
z_k#Z7n?b92IaL_{o>2xKKH}JHdA}Z|s^*bI#S*8cPaZK_Lm@qVH(RraSXMDtLjzZg
z_wV$XJuPuEVd3`!HE8!UQo2?-&_iNgeQH_B=XVk-&PX8dK|7vf{UoBgM|vbD{P8wB
zS$jBhc<5CRmtXui)4?p&?&XQdr(DemW7!#NuAWG%g$}u&Z3!9(Iad9LHMJz5z_r9_
zh4j<KXMr^IFd<C3F$~a8xmyTKEBSdNO;=F{U45^8Fg6!h_TD~)enC+LEas=P=FD~x
z*G_LWFecGn0-Eh>JSo{{XIq`#8^u-^EinZ1%1ZWX%-2OphN-cNqQ@A-^(bh?DJQY1
zx*8*aOKj~ZT%+m!jlAj}ckhQIi*2{CxBC=JN9iS~Z*&4&{42ccRw*$#WDh`=waA46
zml`Nc>I7d?l)A*=tJ@mDGx7I`s6eGstt=vHbr?P&`)srvKh2+*=jWqp&-OIYxg&be
zbH!))*qlhtjNNYFfW32mMj<z7!OHrp2^x2?n6cJ<LxjOO=df?wdk2bkb<WXCIz0^I
zR3n1y_d6Zq_CSv+1yGSK-lc;`JuCQxx0u(~kxoq;aW%AWvtC$0C1#g>b*s<pYE_@K
zes;f3)?C#Bma<0?k{4*DKvK0%R-oGRXk1wbS*|iY7(k-gDy-7c=b`3uQD*l<X8c&<
zs5#a~U!5|hW#+`GPAGG-u%qLoxzy^Feo&X(P*RM9k(ngx{fpsjbJFZp*K^BloL3tP
zztMqQe^Z$K!I=Rl?^-d(6nM#*<QK`#=TT5P-)mgsRd=x;p~d!tXG+?53RY(}-ijkW
zIITbUmssDY9vakI@6nOS-wI2t6gVQ6G1PwM_*e;#mgll%e@+y!FiKn^j>s?&nvQdu
z1ggnt5Dg{XDZiQU?Kz`u2nPFKrB}-vN4}8(b&h87b44Q%yM^;`b#U<_9r_#bwy9Al
z-Gxbv914er=y#a94#ExTw~MFLSA*AVT8!>ENa8c6CwDwh`=1ynDT&H$_D+hie(cCU
zY}a0`kNJi{>5({Y27b;rB;zu`4vG8RPM7RfB={%c8`R2?xIF}-yjh-!U7cG5p6^uJ
zX=6M9bziky$&72ymo)^6-QIOhx=TQz2-~`;d8>Pk7WIahqjK3_e7(Ua2+Md8lc(nA
zQ2W6yJX+;|^$&&z-W31)#9R7WooVD9W7uFVskbyA&P~rP*~Z$0=CcCD>LX}M-v?SD
z;SgKFO`3@nb7ZK4is$UL4Ygr!<MRNc-*i+Nq`{3q8QGbJo9vs(->${luouS}E|C;i
zBi$*I`n?&uTr-n0%7}mnKWx`Pw%mk%#}2=<$jrG)ln1>gg_!~DpE6yfsan=Q<4X_Z
zvBqXwAr-T+Ve)y6wI?<zndO+FClRgXW`!fPbZ^1ErE$yQjK>a_7xVNTlPd<kcXl-}
zHFLxQMxHLRt7Ka<603!m3LcU>h$k$keqPe!21vY2+<fTn7T9%E#JnV;J@YC2WI_Jr
zEM~dd`~~YYM=3soBrzg(q*%B2lZdzPn@ybkmz}k9q1MbN9eyd_4iqS>zVye30=ITM
zQ25RF!AtXZBjEpR;yo7$WAW^31FT$7p+kLu&9%)nYultOfJ3sqo7Y3Ek_EOd@oy!5
zy{PRYvdVF*@RDv7w%Sz~xzI|lcc!~p>qO_Gl9evC?+QIufMP<yb;>4qXYCV-H%^ek
zm7o|B@4aM?AIGXt;2SJ+^9#f!1l0SR$LraWw9-8^!%S*wAO-U8?n43fqAjLTF<n{U
zR4JzyuTh&}kO0<6G(Gzorc+}t!pn12J2XQ~wF>REy8LyOfOjWVO+q6ZQ-}qAf!^HJ
z^gvT|d`Yc#FC1s7KS6}sJIUp?{Nh8}as#(CHRJNHx2dLwTC4QcZ!#zE0Gq+d*855@
z+8*{R2uWPhABt7>#Iogs((%#XTur9{g97ry{3)~i@VO|4<luU<!;3G`$I{+KBsn0x
zN$c$`zMoeRGi5=D=S(&#ut{VvCjlFuegAfYoM%u~XOJy3l&1Iew#Z!U=%rtzG$<<C
z`GMK1<YV8@f^DR(2Tc$OE=PSWs4tnfT`L^@IDXSb)Pfz^U}#?}i*$>2lvD`(u+Rkc
zlq%@^A^TlYZZx|!b$fr!t9|5cAi(if^!x3IRN`l(>to3ykM_6j$s{`B=R6-b!UV5T
zIcTcl2U~o09ULresG?gt^dNY&2-M?+XH~+3SQIK5ONNVgbqB$Uc{04nVXVs?;i6yW
zR$e8l`R6CjFB+{?Z?Iy8)(cEPgj7`z_Go=E^-%S)p(?}Fr=3FF&lPtWQTceQ=7u9S
z<i-TTC^K^A1q-}d_tR3H1HJafIUOvxCBp}W#whFU)D@{tJ^iEuO1(DWH%J>J=iJSs
zroLWwrm;u#wf>4Nr9^xOgN#&W=DG;{<*DXyDVMay=0{hg5;m-3W^Hr;lFawPiki;<
zI+KnLsUl&{PRK3n7H@I<{&AT}b=*Nco6TDG_9geuMA$A%ix-gTLiHiBWvrZa;)#B|
zWXou8x7qI`-|sw;jbJfZ!;da#>38hXX;Wh#<4Tn_pqg@2o80@cCRAyekbLyVV)_p@
zefXsh`bk0BkoHLDKRQPMzX?lL9A(y5Yb4V@LGhc<cB3KY9CSiUX$U!KoRREb_<wa7
z?C>tcQs`qI{<i`Z=CvMT-EnPfxtHUlL22=q(ueQ_>PS|8p81Pipyay%X{F~otq035
zDyR!h12yy;og*c7a&6~;Sv^{Hms_wux0Qm6uilH=1j9~z!Uv>)WEIA<20x&ta_tff
z#|<G^3X0-ZCN~KrWAfud!g7U7|B29IzkWk=!>Y|0<|vb7opO~qkEvKj^$9zjHRcb$
zC76ti4AuLIqEY}W(rTDB>2dB4I`Nj-gYwRM7#TDgcwb6kJw?-t$=87>$*gKS@IP{0
z#-S<}!$O`|4%}0wxmx@qVLX|b&{d*|kmXizmUW4lIw05O`~6TI4~bN3g^`PinwxV&
zri3Z1O7%n4!c1C_Asr>X{3}QqE^Bw%KUz?b$O0U~%OFl)&_uMJvUpH=zHAK%=W^L0
z!ceIIj~c{sc0tSzPhR8?>8MM2u~K_m2jd9g#ii!!HL%`6(ukNp>Wx_zM54Jx)HsGd
z`)ZL;5PLYYHDPSz`oVdo=gKfH9}7@;;&g|9lJkS*5MNYDtsbeT?EhTq*lkXzM_T#Z
zl^0;jfO$>}e3!}vg3YhgX*ouAUS~{gbN=HFo3;Nhci4gM(slBxTNM|>+PAes8FRwp
z2H<*_|HNzYdHCmy$$sM=#rMHPF#w}dKtTjUgq@jb<yUYO)_}wwMwOmIRb=fzLm72C
z<cA;XvXZ==66Z)w5Q78R?id3;O4oi8C|2~rd?WPO#dp}JywYkfLzH2Q<@Jf&Y<`HD
z&8Q;V)T|LJD8HUDoKb%~pEl3P2fv&B0jN$%fz_DW+IB10Nal3B7j*|+l^*%vDu1r2
zO1{T!{d#=n>ND<r<N333&ue98<z|K420mJkDXFb!Ui3j9Y11(-7$R=c#B&|RE%hGo
zB!AiDno-PEpLm*2mZ-f|;4^FPf3QYMg^F4E@C4X>j#R9?BwB>Z`g^wuSi<LU1N}$Y
z>c7IzUMx*_`{+F#BayVdGMqs8+1=zf35zeZZ(VEcPL)4ii(O@t-6?S9?g&@bHODL>
z=6z_NZ~aI+iSw2oO{^3(rCpyNOAhz~V0(YV6_bJze-=;K4!PO8a^xLx^{$1g8Mn`f
zeWfnxg;k2NJ)zm>jtD)#EbS*-0O3hmFJ?~lE0%1J*)WkRp{>}U#{D><=i<YqxigOy
zgf1r3R;!LansBc|p`^AuP>)f;c>1*W&w^i(OUnO<OO-`3oS8&deY|12h4+c9q;r~;
z7LvN&&vc*K+0&yB7*9p8?s#GwB(nG1sT>x^IcWUSIYgDEj(>7sGj5st07mV2V$;zm
z_a=R%r?{Omt`J)IvZ3vj4`!AnM!G$$F$@rI>OAKjFfz7>GmFU{zeiM}#+T@7-Oo{%
z;76v<)&eEC%7`qJV513xHV&5!k83h6C-(8HP!RcmUsE)9@00ScK3}a!p{i889#Y9g
z^MVxSPe+>Qa3{&k9@MQ*;KKNcY%7bHC*Tpua=tDV*S1Q8qR`fyO&t*=GCzlAj5dG7
zgfL5ITRn>DD6G42_lNGy)m+@9Gra^12}FHRt_+gsC~Y5#VQMtlhrCZS({1ZbF-=h8
zEC|-w_8GVTeMMT?9z&}o$TH2pK-FD)PZIhoRhdu0*YkW3gRz<~{c~?J;WGd8X5fZ-
zP=Yzx0m-X~Xcb~S(A1vwCE3Kh6%^HB4?A<ej3gBp-a@gNQn_j?S(?TL`;I5fKsxMS
z?ygV$mGCU2d&&zXw(S8UNDp#+`In4@r-2ZgOs|IL1NZse_^I4Zn{_|*2T41__hnxc
zG&kOkgGYP#ATG-?fNaQS3rQTwhm7<i`2bO39(kwJqN>nxfC$bRp}iLecRzckYsTVQ
z-L`X+2zYTF4jRKDpp<D58zuJUeJO_Gde9&;y3db9_<k)Hi2LEVaHdE2ud3bbc5PsZ
zgo&8RR7CbC;S;|@2ai0zeBj#6HaejP@uzSG_bvrefZwzt5Rd^b77zQS3}M24`pn2g
z9F}^08^u^!kQk1G1nu0z5cCnZ=>CyVbO;l5`>5-_79w$D7nK;)>n5(4#<%T^!B@K}
zNRJy%hD&mio~hs#M@_m}*Q>3)0C{Jsq+O#ViG3T2(Xt11QjdDnW51GEG4E&rg4dwX
zH-rklh#>6=1l~V+HP64ep0(<5!42a(rA@K@9(sSP%{-8V>38(i_>d{_F}#e_OG000
zBN74$%B4Wz<+jt;y`W`J10iPW7b{?(AI8+OBc?+AT@5z6Dex7QjSYLwNw$;DAt*~i
zQ9rofy25qgvrg$Kc|gaoa4g+VzRT={UpK?G9?Rtj@Yon$I$MNdXXCc2`ll@Z#wu?T
zvK-|>`0={=v}kw@rz3;(O02i})j=Tc4!n+C>NzeWu0UPX?0qR%;ornE4AxYm6-FGk
zK#kX=ExM?8s`{<GmiANnnN5_8gxN}zAV)VftGs?}lK*5)WDw{+8Z_w&IC9G<FHd{~
z|1ul_th9c5y81c|?VqWI{SbRQaWAp`ia+9hkx;M<nf%sf7BfALt+YR9i5Z)g_}}dm
zN)K%1C?Lkqt|z~kRZt~<HF@Geji8z93qkA$P&x?g`4X9&{Wjwsf+IH%4DM}cmx&7g
zOpd_vA{3EQ)EniQXC;0YI8K8%b@91P?gj*jNKjb7jG8yN4%&*Q><)D7QEs=6>xs=H
z-U?93cx?W&n34)Ql6n);*+Gy7Fu5`6rCv<(bJ!6Tb+*`x%82vGgd|KDAABkBp>|S{
z-OlwsJixhn)pdaTuE_pvke$%$F&JNolq1x)r&w&{jeH`MgW8Z=jSieAJb$V<05!JN
zq5mv$QjVE0OX-Z?C~iPTQbMnkxH{-wXFnmVzoTV;pv1KqdLo(RU5P42DVwY?X)FZN
zVuFa~PSCUrB(`yeDjQ?y;k#CDgVo35Kg}gxikwvmz%D0!xdJIK1@s;dCrJ13A#Lg<
z=NO;o{`)%xf0z$9YA9E2&sTo}_`eN`!pE84;KLYJqLX1^<52O8A+oYk?9(}cxnn5K
zZgd?=rCY&%1kwidx#k}^4+AAm7&S002p;$i7c4>0uYL^=&5E-mFU-RFJ=m?te^KEy
z{wo|j?8wy$3RB!A!}}<XD}uI?f0J<vCiGt(kimoKo^_$30c02_Lan!Ii#j@hi{m!2
z1J#BspL)IUKlA{>G06<uwN~WqCFoE57ofX~-QcggSq>E>mTMI@IGzw1V}VjvEEEV;
zxQd-wr1gjx$sN<{8}(<+?w+p;9J=gV_s6r-+1V1JGRvu$MSAGEGW4?Yr#un6VjN~Q
z;tcNA&L<%d_y?IFP#nSz`~zDl=Mx2dbFJbIfUU8lbQq4!t0#>5Gk|1Cngclw7&Id5
zAg33yh_OWFf8dHq#z>ql2DJ_0#YOuC8%3m$-6lcKt}Ut;UZ|{oyAxF@-Yu+*Sv~N(
zzG%p-O{JPm5v6hJ-(P1=a#*8mj@`Yp7pA&qGk^sK7-J2VQwJ%~gntMeKcDLxubDHw
zg-5|&bclqr2t`wgAUHzETYoP?)D#EpNy!W3do||dn@9B?QT?GpVU^fly9y=rm5;|F
zm=B>#=TH%Hi*<0-rj7_RH$EwrdjQ!GhbWOfdrhPk-6FDX>fi2vgzE}#RyQOU{nD*e
zFY*SM`U`l)v6`26f#0=Nrk8PUFCYDG5em`Ei3bOxsm>kH>Xb)9l4PMqyBCp0byrUI
z!TbuD7s9GXxLJa#)lN6PBK^4S+X57yCH^OUOae`R(E14I&(sgYXI-z|lcXn|8t*6h
zk|#C?>P1m0By}0iFs7aAv9xcDF>rnT|E~^`jfjw8+r_P7xy?3;HCTjeHD7CAfI5Mh
zQO)VjGCY4FInNH1_eDfQJH>bkrMY?|_uJ?0`RQqe6P!K-zJ6Q(`J2OH=egik)B6vG
zZuNH}Q|oJ8$73oKUyy<iNN*4eZ=&FeMB#vG-0!HG-v7UH0s1s30V0?VXo(Ian03ZY
zTwi~K&?cU9Kt}4>V`%D-7mi(L7PiBO`tj@koA{TdG})}X<}qgY#Irn4uK{I6WEykb
z5Tq4PrYCbZE~ejQ{Is?s&VBCn@+Gv|L~%>I73e20`r&o1l(S_R`q=-^O!{+r`VT+U
zogN#Jwd4jrA}(H-5+En=t%c~Ju2&E7MbwZUG@m}>c(EM(;{9?`-U`eA0b{x>vedl{
zXRrA?=_&L$%nGuFJyz=-y|D!}sDQmO&aa4(4U>-2E+Gmxk?S(O>WPPOp+qvC^Wrc2
zD5woG!YO}Q+^H<TU{kvi2)K<yImOKNs0T!S<=i)XI@0t?bl^X6Z(ay@gI<spnIwOj
zUOXwj`P72)Y06Q(%<X2pw<`%gzy5mS*xlWj0I`@jZDEC^-o@n!TOr(ptIl-DnYSXW
zdZZe%)fV68D(`C5_G;jb;@r=v=Z;8&H5-%^_P4xKwT~|!+Tvk<eBoKQvdEORAf1TQ
zP34B6K(E*X@T80T2$sEqr~pkYYG|b`SwVL0dVUE2v@*LfjIM!&WuqUA#$yIQlPFmf
z%uLOwBLah0N4Ltl?{=jzZG=PxbE)Wj%bQXt6ew#Rf}$Z*usH0bMB4vs>bebIEu8U0
zt7JrvL?`Yym1ceXDi4nfgH;=;{}UuspAM$7I<x5nC}iv(Z#!$C88g7Yl~jrcwS`z4
zl#5k(A>raNNTUWBmWxG|I$AdhvczqEsak3lgh6wQO6IX;_)$ZtsdxzPXeGs~xaK#o
zGRp}Q8SQgfL$^JfV3v>_h$bft<(O4}!|94ei$=@JlIF#rs-k^u%xp9xjqwjV^9DN<
z^7DYKZ%Tc=dy2aq!ixi(+u_aY-TLKyz)jT^=eMSnJcgmXu_I^45tetPw)5VNmND<8
zR6f|mDS=p>67w0un4Um#eS`uB%2<l(P|E9FDq2Gmbzf$W^0`+mHt!GFMptl^$in$P
zJd>jk#?>)==n*1h<WCI2P4A<{dug`D;zFPbjp}%a+`QimsehKQWtHo$KQmyj@7M8I
z7&o}P$tu53?lm2D&v16qZQ<3|Di$=ydc<!YLd|qXX*X55t_dmIPLF*AbSisC!$-Lu
zG0K?uv09cnD0QFE|IRGAuo~snX4T(7Yyw!8`ZSu35+zP&d6|5~J%+mnzf6o3u-rh_
zQ^A9(LI+y>S3KPe^trCg_q-eJ`AZhWb6|$hVxs7U<6}@xE{PU_sZcTkeSi%?X`oIn
zYFT-YQiaJ_DJj{?+>KZgh0KXasp5u$<b);XL=7EZTCY4Bg9plBC)%tG1Yz#+Bmj(#
zH%czYWr-n5XYS!BpY_YEhq$37y$Tph6R2|29}kod8{Td<Km=`ZXNTmGj`SJ%kR`q%
zFSgMm5zG1~#W1}Vo82kG%{%0IaQ_<nO%d5p8jnZ+bn5E>Goi}3ser$=lq7}CWqNce
zRv>L7OK(~;|7Ftra-lCjHiWz1{08Ix-k3*ikc9n8YcS~mHI&`Vi--vQiwa<W736(r
zMM$5RR}k1_;yLo^m@oZTyMt|<8rIXr_ResJHl4TU;#0!&a6D90TzlE$&^|5waDVo*
z)c(z(s60x5p@Dgh^|Y^%DErFQtIZrochDnnob9bj=f2#$I&g$|<5wS#_ZoCQZr5|+
z>-Y>`Izg>$#xq}t&BMwHIyYbyJ|<5QD@!3*gT-=>zOjcGv~CV{5K)ii-gOzv_whOo
z`qqYZ`$s_Okmr}X99?jC|2Kh@@F92^jBC7;;lN-K7fvlqLW=-KS)>pGHarT~rSJ8h
z+=0|!tvmaaPEi^xV^x8~;u<7$d;SNC&6<T{Hl*OD_1sS9*lQ!Py@kLRRKXAa>055k
z)d@^7y>*foSJ()U@X3J{7zxsgKxf>s5Ip3`q^u`Is09&6)((}2z~p4IfYP$zLa%%n
zB;{Y;Mc!!qj?qz35-Ywav#9m7HUDjYWI#7jtQmQJiUsQPcR+G_(0IY)F6O!aSk$ia
zbd(@`<f7bbd>h5SZ~(v(3+rlN9mM}G->mP*b&KW0`67$r4AW-QOoF2?lt)(mmqwSA
zPg6$WQ28s!7LLgoSANI6-ifB{H{+q=LgM(A4Z^somSOlp5Uo1S<DWd1cSLV%b}WSH
zz!OO7X@m3Iq}cfF&cQ3L36+0(d#d2BW$cmBZA+TZxRISV_J_bV)(@ww_5~H2>r6M_
zxmBFBb;pgvnh|H;c1wjENU@f@=d8iG-T|MQ`ZG({72s^a?5;`VV@tRF%MQ7RX?by9
z);rjK5hs4ZQ5)#CHjwhx(|$3HQ0TVc$5>IW3Sx0W{r9gCdcWZ)T=upW2Azj+GcGM$
zigb2z@hv$xF9Pto53MI^#jQg4!yK##n)qdV-`gHa8G*1kG|UhY<jIIN$O2}s$H0so
z!iMQ6ssfp*nR0xmkQYMBk<PA&2-nI$pNpm!XL&@;x~i5y`&&9Yt4%GDgGjVl0C)N*
znjHVAp`#1MW>)|tq=OS;#TWuR;u&*6<+nvsBi+M<z#*V*+~(18%Oy=Q!NbibS}g6<
zfs$@Cie=7EERD%Sjc!zwCC;-sMl1~ugud}>kJUf$10AoOM<{>i-tb%H<dEpoyg}sM
zAN^~p3l@P1UiMd*{a#1UagLv_BhL7^C=noHeuLnOj7c!c_eZ4<M~Vv&ET(d6_i~Vw
zBMie}fj~0nT~m`L%#1pW9SY*4ZzgKb^w^=V-WwAby8m}*M#?8`u%qu>)c9Dnwql?k
zsrO}|hJIbsM8$>UeEiF$0}wxZK;T1InkXBw#Q1|}C>h=#mz{qR2BJQ{+j=HMmO+z+
zb?;vv)d=j!UupIp+7^1&D)KqSkqgMd+g+?Hn*xr^!+&2;A%%5AI*5jeidg-sbojZ$
zF`X^_>A%Ny)^B^s4cwLPK5pjxP&OPuWu@07=h$JLrg?T;>pInWpX{<`m%MBGjzOuz
z#<N<qI9f>LhE)*!^G2^g^!gXm+=5Xat;?^^$PYm)hK24M4x>D?s6)YPVY&UaTke{>
zPaz-@&d@po(gS?hdggbK$8ju%)aex86D>70ByWS;$@Mho(9Ti1)DRQwkV7P~-E2nu
zG)04*9jkJ8=d-eni#rENsd@&=po^B<glA&`o8)3Pq=&kVB<yk&;gQPKt_p38975sg
z%zSHl;xuRQ#6n^{th<mcT4f){oZ8MdE&<Rk|5x+l6b}hasNs_Gk2Gk_wYM5K(C*0}
zd65LCQFpH*NUa+juq}h+ZCE%7an<rp4T1`Ubw5QB8lf<yvO@7$a)OANBW$`eiV@GO
zJlg+*{NCzZ*shTkH64YTPP*|_ETG655!L|_MUsy^lY;OP()R042uE57X^h+`C>4;)
zS-|hkPnMHGu{awe!M=3&e^n;t^~wxLn}p4AKC{2%EUapE-Z2H^F{Efmzrd>A&Ty|E
zSH1~kZW5_&@2T-1W;aL+r2Bf?T`5`a`Me&!g$(0vpN$PVMR;djg>_@w;jIO|@J!JC
zcXDGSlN`$oxNQq<#eDV#RXGx}O~5ce`Ab7MJ>+R5Jrr8`%_|N~<KdNh<r+nGO(xpR
zsf$cG`-fBmpwq~60ru=zsk(;q>sMtvMXSvs+W@sB4i0Q>p`2OJ$SQkkNEFHDo!_A#
zO3vgOlqw!n;Bf?~Zrf2B@`KL2`P>oMD?=ptD@<#rFax+4YgAXy=Tp{BF@UQ>#1B2V
zO>y%~+o*o^dy=Vo;>;mnOXhXE=ikaNr0+FiFY|xF@%!(5@fwtS`bbYwt!#@!WLJ!T
zWx}O`JoOY!5=uShEhp*MuHAXh5x%m*|B!-|6BVQZku*P#9Zc+RGWGP`T>u$zC96^}
zG)-T?E#@&V!-|%jz#>R1&;?dYa6-&&yIz%ZtH$^R4RW`MQU~fb7?s9-$k&r>HRxm!
zc_|}(x<O@^e}2e6ypt}xV)Pjhtjh1|hlMAq9R<xl*_47b;s+Ql2!-3Vwo&{@H5?A@
zrPCC^(<ZEsSK%Po9{y`@&_<Z{;?lSQbt*8#gb@p-!uwa)7(-xsaR}Ij{|--z>s_wF
zBO%%m<O-X{y;}4RjyOeuYCt{e`}~9S6oK)*vGT>}<qT0#v;^lYx!mfG@jZAPX-=Q|
z=cw#aj{Z~cBkz!+r2^x7(seSeRn*tnlOHqZXvN+CJA%;2Joyp1iu^%Z+})(AFAH>n
zMjWdcZ#cTx5EKL5aVv%|w5HN&t&8a?SzAP)>{HCYS21P%$U6>3(@ACh48des+k;>k
zW&K8=8tm(tP`~j^YbgE^5#)4xlJ;e~&`;eOOKnmhq%vt?-XU^(s=JEv`Zlrexc}qD
z{yyzZ)cKX1XQ>^fLx5!(aqCORHc2FprxA}A=uK2}#dQl<zr;ovqG<nU5xG6nCwQA;
zOlpxp^f{jyDRr+Og+@20JL+Y1rUkb%Z~r)51Fxm)#$92<N#TrwhkBLi{ZPHjpRN$9
zxrz>ri)Oj-|CKS9Mu+{FZ)UWf!B8~Nf_3L$krKy5_7_`g;jjoQsMqudqD3Dce5=}@
zzf0cOV43lN{F!{MhzqL*tQ{8`HJP$x;WR^@t~C5~oFqMJ>BGhZ*;N1;Tr?(5`8^1t
z6h;D7gG}j$Bbkz9e!%<iy2S!T%^%_gn?favs%ahKXknaSSJ+6+83r+M#7Iq43nmR0
zna8(pjj!`X?>}C^D_d1Q2e(?9##L?32$hn!so%2zm59R;Z%liQm$5G9%$?!<t>={3
zwsKtGYq+okW*JlKvYogu81mTk8F7Nkl=N1@;Er8RdaxlC%1zW83MX1q7R-dX{q~sm
zyHv(|Y=O;#ItNCXfiP<rUU(b{f~a34dEkARwLFqf7(F{a<dR55%G3$mJ<bmr*1ad(
zm0AeMrFM<;`%>T=)+y<&*0|K6&W5L2t*d!3Ej>XI8KFv6XjfGY{5|<Qe8#wx%3L)5
zR;-E}XPL=_P~-~X^c~i&wXPV(slpB+d77crB&8<g8?~}Dm2!!v%fuDsf*d^cYy@?L
z+xssD+eZ9d<qaq6U4c_yzZAR^o_S+|+mU);&zJO&D0h|XqcYS4J=)VKPOL86gr(C2
zeM%D(1Sh+mq(|9#a+M=+$H09&VyL2hfY@_hqgh8cFm$Lw@(iuS1tDn1v^XN;ojOm^
zl5i2?cMyjMR&6=mW~cfR#17NouCED>I)_%hkIUH6W1cRYD<se9G#)Wk{L=Y{~#
z-TlMoJ)P9tywE!fPTX0UOB9`q=afkO2vW))#V;|jTejZi6<YSna?!4u<rp0`t_<&k
zXc_|qM@@i;&nr^5?;A*ijK=$aj}R8_YkbtMQ@G{^8#f`!Ja!QEB51}Sc&ly~5vSoj
z|HuL3{Y)5{KFPamr^unMY_IxA7y&B1nBMsHLz(8spjU5f-DZg*$(lILH2uKJAml_H
zi;2hm6p(ZJe(oq=3Lw;XiQH2TohK>>+P}*agw>iuql7VmI2@R`&+L=epu!HlRjvuN
zO`JwA%6k%`fr+{kRdK=v?19>*Jm2``8z|n~gKx+q^h6AEIN1f|icAwPt?0jpRCu<&
zT>RTWtiw0qxw9^bvJS1r#Yg5QU|U8i#x{k(&n_tgW8_sBsD)&nn`@4yzaIRXciFXl
z9ANVgO9HC{G?KH8Xpvrh>3HWDtk5<5C(k*q-g>O;>*Znb>Pw=8zZ0;UwW!<av^2Y2
zV?<5Vzn%+ZhxXa&J&!+xk`5LXSce1)8k*ZGW{+_Cn*D;y^fVYH$DyUPQOV&zzycA-
zAthaE@X3$$38ltHK&sKhj_MG;{@HR>#qnUCvc;FG@ViDq77xqF#=NIHZ6a9kBZ8lW
zNpbHNPImFmFPzmP2!9|1y$!{p7Q`kJR+m%lOk&Uyk|*+KBgW6OhJ8Qgj=!BNs+k?G
zAJ1Ykp766KokXO?oMgN{taSmq2fVE<R>H7?jNnYqCMkV8a^AL5Vohm8V-5N$0&yDV
z4xD`BK`eGG;2YQY95;EqQfWD4q|qwk*oS_{u5qL#K4`h%yD-292`-ccuX2WUUy{sw
ztrdD!y-;=_;JR){5UT}qyuX+`PA-M5i~^MnY`>CQ_k;$z{5E_P;=_hE!iLX3ie`5F
zllmvstFXg3qr~u#RoOu`vnC>Lo@&%(Y0XC5FWw|vu5P!T8*>he$v;78!cHk8qx}C=
z8|*KS-5Bu*oaSI=<y0c77s~|tLodBgzu#CGv=no9>L_E19a=R=Mq4oB@&w9UHh(6s
z$Y_XVS4uv-rH9JEYcr}YQ?6C7YDEYnb{NT}2|c-pft1zs<yCE8&{aqqCdE-&wf+@m
z%15A#gxUiS{1q4nvVYFi&!rGQJ4e4_3f@Htz(@baz(^eKsySe4w}>p9(~%BGEF`Q)
z*uVz1qG9VLpi*VFb=g^q!@5Wr!~(TaCGPky1&wvS@G6Ol5xkvLmW|~5!=X&$LhDRA
ze$tinfLa-h_xMJ383pt2jt@{DC4`>7$e*ar)I@@J2!amgp_Ti?Kr+#e;a7_=*?^V{
zcq#5QT`}`vvvALUS*<C`%pkU=9!ho;3d3mfH3WP{Ys>LAdOlv*+JoNz{YjOgrFdA9
z_(Srr^q1WL=qx0Ie?22_I9L|@{9BCJejJ}#B$8GeNic(%Sl+|n*{`dkhUg#Mvzj5a
zr+cqqh;9qpG+R#RV_g&*1l=@i*Ms<#9$bP{Cy9=#0aK9$0<V37?#x~aO4u|zhS(sc
zRsc7x*7{}BOUMa3!91@uHN7~Xk!O3OA^8Ei4;k~1|JVYtYa~8L<9PbA;qo-AJvlk8
zUt7&y@frD>hV@;Ql;QjrCGh~;3Cp6)=EBoF+z$3}Q|h#}Ivzzms@=vb*`?|MzRxMy
zV=RlXT3amdGFnG*^Pg4eVz(Aw`~@lQsV4CkeIrlFj~(8mGf?JJeJ@zG)MNYc{Sh)+
z)skh=nI=C_6dW^j@#clMZkIwcw=BN<`0vb7>o|Oe=)`^`Gm|imb&fICvD@5{z47eC
zt@NSMt?}cmC^3UpGY45BWnirh@I<l3B*}gtb$GAvs=!l#ouUzp(yU_nHM|rMHtOK!
z@JN(l3(-<V*WZ)N(?n%Kr%3`%<JhLJNEgK8D)SUci6A$FgfZzX1)gH9^<~%S==jGE
z!Z)uH{cdR+ZLHdgW>oXbqrfeCrwd0`7+>U`-_5v@wx|+YBcScWT1}V$j(NEZZ26on
zbk|U=Qsu5?*e?ecQ{LACrpG_FT2JXk#v&7EXJD}|)DA6g4)8d33b@vzRzD)eD03=|
zUl`~yv8L;42w#@n7Ducyrd}6gDd-B9a0f5Mv}Q-4b6WP6lO)G<3Rw<Lxh9+QE`6<D
z+s?i+rQhqvQQx$NYqL#(HvA#ZDm>K~<j5lX4DdK#m`@?)CU_Sg_}r{F$|u~bNQZi^
zzR#XE{qvl^Ib@T5Q*R&!C?T>--tPB1hEv`|2T%^4pZOu(yqH#XM$v&Z8g?z@VQYkg
z^tLmn-|$spBd}V;SV9>Fp`prAO*bg&5XPO~b2X@S?bBduE5f({=`*ii!D!zE2zBtH
zKlW(8ccp$Ns3ytMg%|yiuY&&ZwGb~oK4ihZY8-DvB3|D9!NXr!I`8`(_CU@7lS24s
zTGfFLxy}uwfeZ~bHHhcS7jS6U+0zBN4w9H!Pc*Udui46u8J3JatipRaJCl<iR%mR7
zaOR*$Q?(&#UK)TXV9`iS8Z0i)PrB6B5%|kBGDWPQG*Nz<_mr2s?H#R*ox?+9sHjlk
zwK(sU(K%hS;|m}UA9(@r?=3HJ0r>a-vKt4eiH$@}qLxdxQRTcgslmwf{Kc}D_Y7lw
z_|1E~kK^i0IoWUU(HbgEWW<;Y+b&Qag3oxp5Q2PC(F6#9)wL)vaoO-FYM4+3Shnw_
zj>J?@^Cs&EL-RB<yS+~CYV731WM|8ZV^&n6WKuDO=?@Z(MOkQ}B2sRIgX8lx3gg>o
ztF&u3Bg+Mm5=4?(MIyp7XX}mWf39w&S7!OT#cX^rN3q`h7j?>jJv)f!KsODpQ=LdZ
z@k-^=g+*e(;D!<pbf<s{A~$=e`UvQ5xBP{8Tcq`={ZDy~L{b;cJua>F560QTsjZ@)
zE0Hb6a!jY#et0@VaIyY6+gVomnpwuR+Ubj~#F?JH3-^*%^Fr1eqaJJS;7~Lal_@hp
z=?y)ulVWHDVe_9xK*g?0$8K*Wems)SyyPfpSThyq9!Zv`9pJ4+95R|hx{G7m|Eam+
z_EyCh*=LHf>2q>_Y?ZBJl@+sSS$~DpedxIFbUnteaMoIX?USuCt<eE4npmk}af}M3
zc{atkq`n)OM}t`D>gM6G&><R0)hxCT?Rdo$exRE8M9f(X8yO*$^Du}^B*7Xsi+&)u
z3z7tI!1GwIm*M+bUh~uBq14n=zB-5Re1z3TNT#$f#>qZCgxlY*--7jjTW`I@Y$-J-
z=yj*-X<$rEFIamQ-qOTFU~E>?_|>RGt&0cOpnPqKl_2#4`_vWCe`#~K%TkO;qD-^4
zJ=qoCN1VNZN`0?^zqGe^EX?|i3Tc$zQ}hWCWH9`Gm&@%*tYRKQgkPf$*mryP7pgim
z2<9sK_qRF*54L#Ebg1i>M@b1dRH3@Xw!>cTB1{kEAe=fBP2}B72{cDtIGI9-{PZG6
zZBlu&c4U3#FzPr_X5wDp-+~0rzHkd;E*{bWD<#ne%Yx=#KX|F})nlJBOH2tno)Uip
z2Y_3n+|AP_{z#%e=(y3#$Uo+4#3_9=%`&WJJ@7`gvnH{sbgQAADLDRw4L_gf?xH@(
z+Qr5qzmi7o{W3Y*s#Tk>0YB1ymm#htYW3l}&}JiSrx~T@F0uMz{5|eoTYrZnSN8zl
z$DFmZ8KP>RQ~L!=Z;q*Paa>KTi>?{)F8E84v{Y7<7<h1#w)401eC<PFWdxc?vsmt_
z>M(bwTmrUXw<m?MW4%COeu193FuO!RG4_2OD4sXF@~Zcx7e%uAjWmb!tEAxy|C^_?
z5;rf(l^=6cdX_!VuK`I!G%Rq%_!`C;k3pG8At3hukoAtym37~?cPdUQwyjE1v01TM
zF)FN7Y-`08+qP}nwr$(Id7l5d_dVylt$)~Uf7q?<w)S3g{pRRn^qEyvS;q~n5bfPx
zu8J%njs<AI;!){^uC=zbd+zmPU)h7Qg0K=Pg1JAhiJ|{Mmsf+}csA((87HF3j8j8Z
z8TNED=-m01;q$|lzV}Sl%)6z1N6jkZz`vO^=R9>r0-5U5i~vJ*s~&kB`ZObc>i>5$
z!ocv)|MeouSgL9GVK#8`fa1!P3|WM_=Yl&VbTM##s2PjLVpJCC7qw-!gq7?x2~D^T
zL^5k_gg6gJE((@a)i;I=Q3p|TqXXJC-|hWT)EV;u{~W&rWCzI0!CH^!A6x1h*vvE<
zlWl^JrJ0)XH`0^~9{WSL`jaU9Jn+T$6BrW;S*X#~h;exri5~vPK)^9YV{;YyBo9o*
z?C%ZSrO>`unw{<;JZdEBNw><nd^VLr^s;2|HP580M?AcxcF&wzYkju?miIN9N<7|K
zA5y*cUQ(0Di<)QspHIPB!)=~#TD^AIe_x8iUXuTAb;t8vJ9jzC7m-4tv59N5`L<iw
z#3dH36b2WgIWy*F0pGknTFmlH^J3&2w!A7c^}af>6JpZznL0PzSfW1qRDf?@c{`ti
zVX>qXWq7t4i1xlX7yUdj=wb9(S|rRVr3mbBwU+0k8_;8ZXhTfM-|)~c`78Wm!Kv9G
z!cyzW+fj}8m8@qtLH3~;Y~ZXrX=A2BF0cD)AZ~~My>ZgD4@iO(6@EPn(jA{^oe+0w
z^Aci8u5yt(9yr}OEBQ^TGU(jCa(0w%VvlWrfZ_*&l}8*kLe9uY7nSGf`5i4<bK3Oe
zO?oFo=A(*+aKGW~M}k->?9%yOHRv}a+$b3NWpPTT4u_4+Irt6tX%ER)u^G~Mlb@gu
znN@238-JL}wt*2J6N-9fryk@EQmF@Lc+!iSufyd=^q1~ygisH~Nw3$X|2hoQzD6yT
z-NA3<7?7`%=64|O<Yhb`NH6B`vCON4j4Dxd_^Ss#nFr=I%IGrIlb6*o_f)fozr!$~
zzz7HB*^{E2o-YWtRm-)|<m8`U;@`Um)(i5Op_ak`1Lbg>I8y@`41q`a#Jr3T;cEHi
zZW_88_e9t5SN-LVANW}Anwty0f&y3_1S~NlQ6DIYMci%!`7-*%`Z@DR#{81kA?_5L
zy2gCas!rN({xGVEHs%!=^xuQZwYcd#lE>7vE0<o>4MGprXa%u5(Ra&y1*_yg|G1PB
zw3EaDf0X!IdXQ&Ze8<w)V8xp{v%|ifOUGPNy1Rk$w7rEl>mR#8_zk5ty3(zVSHSG`
z;mTKTg$#)3s+k*D4?C=S+R6$+y9mVu_2cpzJP{XAPdgihq5@~=v-(%jTQ?QS0^6V<
z|NhONo83PuxgPQ=w1g*{*xkTKLB}FS_XqF*!z4qh+it_sQEpc;PvPagYh=0ltXhNU
z`Yp9XbMVNR)A9V)WAjaFF?%-Kl#GUi>l|kYlnb5&>nJIrgvdZCVQvO*^#DUMgLG+V
znmA~*cS-YWRg#omaK570uq?5{@%A;gXqMg0qNTD#T*H6Y_N}eTM#s1em$jC0W`Ize
zIFwLlCJVVDdx%j|1@EfgOiUn|N;}rlG%rMxr9tGxWSdQs#X^wdv3KmL<=$fUUWb<J
zi3fki$s~#0dQ7S6fMX+|`K9daq<rF+$^8V=>q0lfV2yOc4%R?i&hV996`rMCCMIDw
z(CRe2Fcyb4nJWJ4XZ2l_C7akIA)xjU@@DrK)4S!NEkKjFgYw_N4#r4=KHMpzOl>C<
zb{^LfX8ZoHO^9Y~vk25&&NhUl!t>gPDH6)RP{)0M7#ueZk&uKm5`A0gK26t(b^G2Z
z89=|^VW>=A1<nP(34gKy2w&fA3*>1C3>Bz$S=fq(XG;TD(Kl#VBm9sCEV?Fts|L(S
zGe@r-I71{VdfpV95ubQ(7OE!P<*wK6p!UBZDYi8SsrK4R&tOXQHj5FxkL-(EoyN2u
zh0(tKmYIi1W(Z;qB%x&N4f+S|QX7fe0{)BaTBefI*pj)LAjX6ae`c@HXPMf?u=^>1
zdjQMq@M{Ggo=v6<7DN*pKORfJRQkHlgXjxe{y(C-esN^)A(1=1svO|UVD|DfGVM{b
z#ad>!m06)&QjCHtvUKzF*y5V#n&s=M>K9sxX+Q^O>|=!3y8M<!L8?p#-shDZnzEIs
zL9W5$jDYx_Pv&QczvAiw?;l783`Hr;ycgEj{@k{v8WZ&f)dF?<&CS;Vg(p$6DsF6&
z)}n%Xr)~Mo7g3EfO`GQ6KiD?paz&{%{z+||Icx(JWB-fV&_1bcF+}0pe97EeCP*Fc
zf1u6%e?ps9t)-RtTRD*9P*K5^W>&UxEbECyk0ZacBia)&x+m667yI9Ut@{4~Ti*ZV
zHC(@Dy9kdzN$DANg`XSi7Wy;-kFglDvY9Cji~F^$1w1Xa0buFC1cr-cFX1{j`rs$!
zq<$;DL`)g<;dPcuF%av|Uyk-T9exv_($dn}YWgc%J$8lmx~9RUMQwSSuDiCByycVK
zqLNnASlZ6e;InVdIU_lK?2UCJ;EtRA&A@Rsam|&2dk_m%7H;l^QH1qE+gjlhZ9bUU
z9?k`;f^aH#gRneK5@R6v3aK?M*b!mhqI1aYIjf<UH)oq?57<ymN_LfKi&L~6J7|vV
zonJ`l^x^+-7~d{5M*#b_<N~Gz+W240HcGcAfClXkn(*#s-Ymbx?%mx&FE2b#6&b}2
z)jb-hkv6Wc@X`g0hI`Ch%gRB)iC0W^?(_?kMXf`2&7srP2f~o2ey>Qwlp}LtviCt~
z4=U3l0VD*2O!ji*dcnFmIr%uGuWpAetit*KKVUl|`xn@7PTViI<rL#T*5=sycHErE
z<T4P%5JHaL7#>iJ7*x#PTHOzX<B$8w7Jig{B3M)jg<iiM@k_7%=#2WDST$}X{YQgR
z8_|4oJ=$fh?q99Cp)utakGWO=+~w0+6%R2U+o;fJkTE6=nS#9#*92(6qCHP&w7_di
zT5~Q!t5IiO3f>Al>TeO#8Qoa_3*HhaBXhq0po^jWU@K}>`JRSHyTOCcw>r{(_7R)r
z4if!)+3&&>!L`LT_QwLed(p#7@Bbg+fd9pCRv9J168#*%+dGtQ5KzNrh>M(`D?x(l
zUPw`7DazdVVZ(fFj|zWbt_Wd;?|BJ7d7_8ZZ4(o2wZt%xqcCsv!?d@G*LPnkZ%)F3
zUGA^lIT&7I_k`0LPgyz{hPh@jWig$}C;bC=7J{Ez_6OK(M6%yN8w)e{RYmeNzUiHZ
zeOWGWcLqCZLu%kIGdq6{H)!*K!!9N+CWgXIi|w4g$P8derfQK!3R{X{Abz|Vba2JM
z--qT=yY>=wyf?mfbq1vmSF|Q=xsJIkQ_w(cxjNdFF9D(j;P|0(+#iQgYwfDaIsZEO
zTAX0)!*MBS@Lv|YH+`OJ$;*wPCE9=m8fZxCK0S}U>SeCh3^li!M43Mr{`C*#KW@ur
zl*!AjcX!R$uyzT4j6^))KNi7AOlEtyhvxRcy+Fs{(l!hq%eLY@do|1X7c=9r$>g`l
zeaL{2tg!tcW>8_@JEm%E8r0j!fy^)7bKE9q5fOdd^BR%Ee7@bB9_lSNKJ<tTHTg)I
z;d$jaD4)>R_mh(f55$S|;CMrItjUQqN-2aNCy20n+<i=O`ng*8pspN|jSb!OoBoe^
zD>$e=t>q(6Q{9g1ue0AhcCXU?Gu32lFU>oHm@JJ}f_t$r2oAOclsxodQ|XxOh*^)?
z>Vn1UVrUZZf#;(EZX?Imhy<=WJ~lj3ETIp^VG!TP5-6xR@X-r``>pBQ7CG~~NSUK&
zS*O#fFdF_zgp$86TwTEqds39sYu-DKTdB`S3k8Z>k2}LrtbmrW#r;SYGBV&s^7zW)
z!Zxi<PgEINWxgKs%=qG5rBr*4%k+QSLPQq;Uq5OPQ4{f9TM21%X~UunoiEM1(JGo3
ze{eoNZ6NI@hK@6EXqVpN$g^MN+EJiY$W35*VfG|9F%R&+)-J~=*9v+87GC`%HgfAd
zq7dcBJi(da8QMf6eQvvipsSNZ`v}b4YrcidORIa(iG?>CYyYgFiy}ATh-XxCUSmgL
zc(d0<9euq{L~Uy%FT|9M$G^VtU-@^J-;fQ^kEXjk=%8@~*fo{6nlYT_Y65ZEN~gAU
zH^H!vDVVQiKiY&?u(2&?V}cWL!7Y`F!(xgO6_L^8?rZxOsVJX1Nh-Z}6sNgj;F36i
z1~qv}?d6Yb{>fyiJFC~Q_ot7-y{U$U)9n1Oq#hruM=t*DoNy_ci>g_qKZ-_Z9Tr@|
zNrb-KG&L5^{^WUg)!Zk2hL_t1k!H}kfRS`1#EuvwtF6#7om8eNUyVhI9R!V5${Fu#
zv#06y`S=eDCHgh78r<b%6lZY}0#VnZ=NPA9*0BqC|DxBb(I<LoC;<PU*A$E-x2Oqj
zIC+9gG)w2VG|7FDc#`^S+T<i&pImA#+@!II72E^)R?r})bg9Nc@<=K1Uh-`G#7<fv
zrf5bv3com53isZ}!)nG`=((ka718JbywAK(^c#*~Z9q5DOs$v<DA(R|z<4OX87)wP
z*%?^473wcB7M%Mlk!gIjtPn{diJ6z8add7FB5Lu`mmR~u#^{j{@3L~C`F>a%biRqo
z9{KV_W-f<+up;l&QRlICw?s4ct{y;{`x&vSp1)!fx-+b>_gJY0^Bb4wOz4sR@XaA0
z$I@g5{AnSwh0lEy*oXeWJ@I5Tv<)C#W91#7825UdK={TA{iqkgH2d3k>+M%M-0kJp
zdxKw>=idjcVq_*c8ht-gU|{b4eBWCtc;Cdxa|r1kQ)uC}9p4<IL*6vNppe11?64g@
zfL<iW-GSOih*YYu+(x%i>=}2W*&b2>hdX%O-Yi5O=^RWO0(dv^S#<GmIpmr#tJ7`9
zkPG+X#&Vk1If`dPKqbB#`u5u7#FJCtk%oV!y^N@E@3p4nx+rn+nc~5ajkz~gXO^+P
z20U<<5UmOCOP(J$2FOK|5NPqEY1Q=m{o(fXp7p)@Il9qgzgwv`TN@-xJ9E=>^)r{J
zMus8K^oq=E0~lu?=Yro=odV0{u-%=$wkwAYG=i8KE!>@~`rX4^M@2{+7%3_~0%6j9
zQG*fhDI)w;{dp~u6o{sZcmT(`V>6<XQ=ZM1o4poiL*uqFKM`4<0b)HOYt6OS*Nn$x
zxqk_aL^Y#{PAZHi5z40y@23_t@s7G>vC-s@{hl?N1&qk_^rI@{=bOw}r;J9v?_J(@
z>k|HCx2QjI)+cW21p;P*d<@^6@9YK1ib@Y;3wmt{q=gz_bB~8z(uGI)>O9nPDyz9a
z_*oL8aiweHKo=#^6p#ITxrs2$46ChcaBUWUJ}li*r5_JkO8q9wz^Eg96+5CW9zC5Q
zKy*8e02q8(iogV@gZEL3s5Hj4)nwJf@(g{VrDc&DI%GdnO@<m8*Wr60;c$0tJ>{1v
z(r#in&7n(>^S+|*lY!Y5;iHH2f!XY4SGpa9PvV)I`=QIEcBL|bS~#}42L2$eIg>W)
zm|E#rTL}sC$OMUQmLg*Xb&j6%{c2=eK%2Pkx1{2<@#eQ52|3IPvD9FW%6!tVGP
zHTryfFV(fmvQ4*_(c!f&4?fNWe?^YE5C>(M?n0k0R;`eTtto&nHFpp(+AA)C2^Rdy
zr7k>NclN7iJy5)fA&x$9978l*HCR^OE0fQpFIojzT;5!QLR#SNy0`YX-PM-i0)u*%
z51<J%VbJUF&a7gz>@B3^3vbRu?Qqtpg*jAcp755UnH<aHptVHu^|O^$J*qlbRGl3z
z<;QR`TR|5*wW3Paw-Xx70+Qb%!@;qvvQ6iH)#Y(+N1GgYy)~l-m^z>HeW{AMK?69S
zXps0s@e2>&;^OdeV$#N%=6XG)&B2AN0c^xjIbSOcb$xq8%GTsKRq>0QP(|?@ySwEd
z!Ukr^aAKAU$7m4+Y6Q`RaQ66*`9>+22@W!310~WA30xR53T0gBajzU~mH57~lJ#S5
z<n*<4ApJfuUMFF86-vT=R=O>!D!k@;Q2&LuQg65xPn2KlLptId_M1qb4Zp-v2;dnO
z=UErxL2|Y_(urB0IExG33R-0fLD>NlGyQ5H(iKNcOwXm=ERZYq_D8a((;9(GUFeT!
zGD@2BPP&*!<&9p2y7QS%IYVFC8SR_OeK7o3f)539)tjRB-tsIOYQ@6Rl&f<g*-n_6
zt(?Z9bHWl2Q3L`F<(%Ah>1o|P9H6Jr?2_w8roXHoq))T{QXm~jv4x%t_P`3(qshuC
zZo3}!7u}}c&k*fPr2Fs{oI+SCoWPYjYT~i;-z*fUU5@M@(0O}b@^L(_V!W4Z*O_>0
zYB;1cm3x9s%9WB-P6aVW{DxQkd0@B<W0JHKgAoGejM1Tz=E4yoBBVe~RP^-TBOU5h
zr2E@`lj>PT1{0&NL_qYNYgFx$J2;iqKIt^|=0anJoO-K7hFDN<rSYVa8i(+2fm@94
z86<>6YtzM_KYcfSD3W^7xrNfOD|!#>U@Wp*_6yEr4^RI>#6pKqgPqy)1trU|xAtdl
zcJ(J8`SNvze6oGac@`a@(WF=gTU*}Fi6x()7pFGsc}Y0rm9<*StJ4r>p@zq0n5B-#
z&YuFJ-A?r@mem`iO|gV>R4n4yZjCJ6bth9Pg8Y5`0EZKnea(!P48^;>d=fm#k5g5G
zOKdv}KV>F&Ax(^GdMJnUpz0CzoM(SGVw|*X3U8p3=0K|Zi-#$YIe&DkCAs5a>XpQh
z#W~9|OT*CyDD>zI>w_b<J-Kb<+1hP@0Mh6_4*Qs5`@1>u)HC7;`jr!{6eXm!YFmb@
zmHi$@A`~ZIqtQo7(EdnprfO!9ktbmvA`MDmxWaypua&JDC%b4PEgRWr<^{YU&y$G7
z8oOr<$}^>D8cG7-2yod4UxUd8%6QCjRElOzPk2tY<r@FM6JF_5AFE60tueXIC+@y@
z*VcAyXxG(3J}H)cl(`nwZO~r+FrlaJbFFK}MZ?3C3iMG^(B;10UK1@mx;B08)2>->
zYPP5!5;?gKn7!cAcz7<JX_{{?JKz0#mHma4FAb-orw|)p<#WU~`zTy@Z(8rqyHFf2
z9H|=Sq~_kfZAIaX`94dXMnG2~ZY>ctK~&5XED5bJM4oduvdbAlyBLA@6tV}r$Nk3)
z*kF$G{GGz!A;7r}(LMa;&QRYHz>Mc0<<So2Ql5I{$eRfzg4oROmjflDkOd{P==<pu
zLVarn=yH6~CInPuBPPK^EPdGhb)cvRF?gw5<-7d&9OD1A%BvOC<TE&}H2da8S4S1H
z6BTJG(B1KxIqYrpVeqw6K}zKN`+{h)6|ZX1mSkxQl_D*ORu)tI{(1RP?`;mY)X%MA
zZS5MW1=a*<8+Jx^Ygw*PPO!iiw_p<n^ksmP+W3gQCJ-%%<1aAm{>BC6X0kUhc{+57
zo?>cJzW(d#`>7fQE4sw>L2@KA??mfpFK~VM+{NhbX#`}IFR937G#aY4<@yq)M%q&T
z(?<$fey*jt1QSXs53ud)HHGG$Du!#k6EbsXeZHqqWfU4RN+S(wGWDUgg@pj#Of80<
zM`0ZG-d7u4eDu1&uXq1%-`x?lxFNj7)_8%`b2b|i%$c<*DjBQ%{F^+%Z=T+k2{lwR
zGVEuMPs#j3o1j+3lebR!ka5;lO1R8(ExRX(7)AwZ-&1jwln058*L{jV{vlB0_*i5^
zh6jpK?~uVL(8#Y}3Xn|m$wn%I<ofl~6Sy_}B~OpApD5^4wMEN@IoEoCjX!!?$~wo5
zfDmdZrG4X@KJ(1FFmDAd)hLgpLx9jI8gSn6H=Opx#kOmDd6`~F4DWcT3gj6ww8I`H
zlNiUPT6s^ihAIoCTI|L5c|(T5X?|h&ypOOWW*;f+;HoM^0Fp(Ye@L+8Ur*bzgUn^K
zCc+;({=CorVu_FMoh<4q#N<QpmzgGq2_8#$2`CID>UwmNo_o(frFHE`MJDLw(a7ff
zDD@^9#YA#pGBoD?ob%&=iCl`$AlV9bdh6U&nBXvhmd7SfiiDB31BV1pMXquJ*7>H}
zzKq;cz}!NW8#97tBkFU^4385+=%J@pHW4pKX^cE7&(-DxUD2%0QggQ$+oc+L0tFCP
zrZ=N0PR{5z<dJIVA`YhS{u%YVVGoD9!(=PSvChH8DTkGNcfWj%m^DnPLH&I<D)Y9$
zWrb|4<;3i|Mdf0({(3hX_9H>(+z*@bsISEnILZr9a`jj*nlG>0LqEIPxd^^{lZRw~
zLz8PP9h2QPWz=8DcFbt5{SW~%Srd*o9&z97CY(f<U~XP!-e>qUpP?qOdOjz+`{_t@
z;zv+9Jg7Ur!Sl~A=FVu;+1qA*SFgF;`sK(H5S)a#dm5U^w$+kM+Aj3Y@3zsMRv(=Q
zyr>>?q}5!8rd%>Da-OY1Kr*Sv3^au^&zR-av>HZQIDoj2aq>0I9CU0HM}tU+3-XZF
zN^PjG!9WYtp1j=GI4Z9lOXtdM=dlqmYkp*DF8j$itn7;~zS<qAFZ_9R!!`TW`7TyS
z8eAyyfCBV?9tSZfld;Ifd`_(osS(Bn6l;L=seM<c`4#J;sN}-LE^OKh(OZ{9b1j^)
z#*G+P#dvqSmH6g*@!EgPxd4hAc8|^TGkh_J&6x;U0av<jBF<ujBsWn<aafgd9`eK^
z-f2s3V@5UwlW1Oq(3(vIs!f<RkT$tuwp4@&g38MX@++E2pw}}h13mG+{ietcU?lK0
zJh+QEVPx7L48JyZ=r4kX{Um|b%%p(kVj`@vt<un2aWH}g$zootg3{1Wan6oh#lSLx
z#6urZ%v#gS?5|uQ1N^IGWX+i{o58;hJd#arxk}3HlVoY0Zo%5d6J0xJeJ`%2>^-p}
zyuxEZBZ0c3R`2|qXg}ya+JXypU{C~tH~UcQxqKh8IXP!DO~;Rw6r8I$xuz2B&{Q<E
zsl+I|2R*Ss`Q5R?)6r8R9Qund4-;j6r(C#f!F;KM*-g}{zPaxiYEsMoDjD*vbC>x$
z2x*(G{nPUt=DFResuiopb+J|W?tcG$30cRAMO|{5mTd2kE8aTJIJ{F14cmmwDege4
zOZA=bkf!g<`B4*WG<|sImVOHA%l>$+Wob&)`%VS_@!U8~OiK|t;P|-nJFf9S5__e}
z(vo!rYt4B6Eo}8&3n`y0R_(iLiR8@=Yp9CW2%;^Q@%-+bX!w<oo%8*p?xMN=s58@$
z5Zn31Bf5vnJcJxwo<4mzXQc~dF{fOYddICEIYOuSIDH|$mjY4TkRp-IyHqtYnUn3%
z`9b`ToE+9<1gKw?K#|et;vk-tTc$#Zx>cR*P!=K9H0@avJ0DAKAS+PMHdj);M7d8U
z%rED9HX~agE4IsStN-N{>H`Aol@@aZsq^clAO^P|j5NZRp)vSIc(EbDqOH!&dJ=`t
zOYh=CjrwyB-<?_ICzYt@Q3~ZVq6Z3Y(~}<>NxfXkpS99`;OzRRKc=}6m!&~$;cU2r
zMRz*<fwzv)IUU~0yh#@6FlQjy-H%;)4SnO93}u5-mEFtyWZDRvwViay1mZ)@jP|rS
z%`_bHHlgSG8-{)yJ*5O#q8PJuB|F-Pd~Gqbx&mZ>P$6%`%+9bdw8j++mKR$Xb!rHN
zbvpFL7BT>v7Zwg80Q4feQ_tm)=UK^46w#sXa#1LX021S0v5{($m01c^F(s)Y3&AEQ
zd7mMk^dRKHtBCH~n~ujMmrw!hwj<Yzo^WUL^M6RwlANSL5EIgZM)IAsekOjUA?e!i
zewI|!>8p=1zLrvrfV>qqKEOooXa<NpzbBR^O+5IXXS!7#hA&;g9m=L+2F0)oX1`a?
z70p3QkfAqIA3%27P9cNZ4FSHKKA)nleAeG*^<eVSh%aePbm8^70|aQl?0wgx-NMK=
z$bre;bMVLx8GDn=ow3P=6;RL+D~kd@XR5lg$DDIuAP7~ByrQc7)9Jk38dz#P<?`wP
zE3K?~MB-CtRGI%DC1DfL(hB+43eVu0YUJ_uqJ`;MEJ71<<5fU(t>j@iCk*F8@I}G&
zq2;sAm!W+7!IN63I4N*aeFWLfg5<|0lyaaI860}cA^i>y)(m}})B1}iojrQA@%bcF
z?}54I#q_cl>!!6I%Os}Q<t65|6kEbzRT)DG!LFfDv6Ns^8`~^gg<o&GLR404?2mW<
z)yIB`MVbcwa=erJC}6yQH<yqKPfIj`YKu^3mp9?8Q4|3zoZqGAnQiRF{3>Gw>Xs&g
z`K1l%!<M8Rchj{b6As0KvD0_H+dj6@SmaE_do;=~`-<Ip=NL8w<8~pCFCh~JyyP>;
z62g506>9CfuO!+W;QH9s!PNrJDEQ;N6owi~p?%OIiEi?06~Ik$iNz~6$`M_Q#w0Ht
zhj1=X)P($4zsQb$5D|+6#oJS;ew7FZxLvCvqkMQ3mk6G`FMiF_O?>-aWt|}0EnH-n
zt=OjkT^l1a>ryktqpT|@wM-kK^1A{D86yrs#h<oOFTM<V_7PPW1Q{GGv<3FXAKTI4
zo+U{s#-v;iXpicOdfq%kd+(~CI#nH1P_I|)swg~AC2@=hWKsH+Bapjz6Y_2`1=i5A
zl1;CP9@&bqluKSkIFL^uthqxZ&>?Lj@bzwF^0$tDZMmf}!L_rs<v+rP_BN8|Xz1Z;
zV1=VqU4xkk7c?whYpg98o7()be>Ei%y0|O`We*_xLi<2ZVXUsqc5xHZ_NUJfyHUI-
zn0q2|I@XP0-mf8$P;%}yRJV!3Rvn24vVkvBFH-+rMvZMyPRW>7_HjvSb0Q&%pbGu+
z^(Jk6r+^zaDLpMO1o~byTat@Kk!r7UE9D<O&|^4kgz_Oh?)N+!-FoZ3fA@0%)Ce5`
z=zH(pCk`ebvU;CV$mytEO9$MtUyX`rfRe-KkKufwn1pq~os%xlVya=_vG*fWd~<FR
zq&b{zZI@H20j`*)t{#t-^VpOoOLFZNQ*pykrf=5fDw0l&sXSaUU%fj_E23Nx#~^(S
zo76&))#{@tVxKpbBc_R`Gvd*XuH_GKBes6>2zt#){87MHb(hBW>%2~m8^5T0^4!Pt
zAx%Tu^sAYOT5&^LrII&kq1@uN_Twh%@#R_l#7jRN1b6&iXa#MY<g-5JZS;?Gu@$R+
zTM>L4k@cz_x=Rvs;l;!n9@BuZ_U{-Xt5}OR$lj#c&|Htv$B$Zb!m2lgJRirLm!xW{
z3S%nHI8KkbMD%Ow#l}U(GHpeF6eA%?;6uGITdsu-2srud13OVgE$*4+2aNu`4eSeI
zJ}NQ6j&`hKu&c<xr+~wl7=bEIP`Z^E_bTEqFEQy=tiV$jR-YB346AQJ$N-mW-8$0U
z^VxG|!m<K{)n5maZqyFq?2&sGT;!51jzMl%7<4=lVtf(je6zNgAMcv&+pK|Jf`Jo~
zrNRjgXo8)IoCfmHU(mIvbE*AC5mU2q<(m8>7nKu)=bSCqj<!(_isx}j&;;e7vq(ge
ze-g?=e*+MdsQl2z`(VV7sr+(nQH&RVjuM({-eBExt-WvK=^znHgIl>DB6XFy1d`Rm
zS<Xq##)y#r-RkqF<uccfUoKEA!8(f}^>fNaWJ#v>vb_CZVSYc5s0fYynq#0EGAVXW
zbCpA<Z9WT1zHB=<UTa0^XJ`H)9H&nKYV^g3*bd0rjU8oY@i}FS*6*B!PN{lQU=ya2
zaX77=GEJqYS0wgfcZLi^q5yU5$|jL({AMOXBFA16WxRodmQq`ZIm);+n964wKi&$K
z&Q`Aitqg5a%)-6yUTLH%hM3%EF##!Nc8=_RIuIo8XG|stVMk!fq55oIs`>orboT2)
zy&1UI_GBc%>#7e7EJxBNZuaqAi(i@!?iW$zAjO_ShxIgJh`=M{65jMifALp_YvFQB
zmZoO5D5h^FZF?(J<ww=r)@|8DNe+v)z&u?DK*iYoct6sya%i>dnVHrJi)ZE<{Y>_X
zh8#8%*xrUa4G0C%FGm%`=h1i^+v~smaSn*5!31d%XTZy9meCQLC|dfT4xz(ngV&L|
zw^|u)?i4*fk}VDmHcx?$LLVJ@CEFGpz{hbqb6?M8x8$<uLF3@%F#h`mFZU7Okc+vu
z;vEztN(0DrTH~_XcPV00G06#I-C+q3C@c){8Nux%9oX$Pc97@I+#jkPUFdiLcj_?2
zbvoAdf1b+knVm*NLGN#RB|l4$qs6CNso6_lMnuG*&EaLfkUNRx`VG~#4JZ#zi)*Oo
zo)GzvbxSt$ZR{{kX873SyW2eZ*z^8$oFoRFzVjQ4Ck^mIsNS;YohB_^aLzpuUx+Fv
z6o)p~mHv`v-Aq?Mwl+DO-5>=p`UY#;)e!p#w(L(w31HPII%gXaSFO+w^VPy2*R0Uv
ze?g%^XAtRwupDFS)7>Qgt!NMIak#k0nQ%>)($7x&NzCCVc<h1nl|r0?pZ0zM@4*-y
z{Qc_Q8BV)|B+#dLvXqHjGoW{gYUNSXn#;CtxEU6X2&R2#Y58<D^76oJR&7LB{gSX2
z<C>v4(dM!>PRh8F`Z_sqTG6r2-4Wz*&`*()gwQg#v{;!&c@hsezs$q{b)-PbcM9vs
z4Wt5_+PsJS=(`}kxjzV7*NX@;1Kf9*20}r!gpU^Vp*6H(F7|7!eTh4oC-<-=wI$ob
z%?_}PCJc1*a4rJ*EBdLY#-yxVvr4I}{BVDio#TZc5HatrROcMiyh(y#v4vF#weM&3
z>zJ?$wdk#*m%)fm@=<2a4E6ztOGPL0_H0Q5?31#4xTz>;k+|zAED>}xWl1%qEYc+N
zYhv1|`nwqC44k4?+&Q`V(vAQfE*>d)et|bK-(0+rfuSFJu3s8FePWUuu0ZTT#NLEW
zoKeT!h32(>wASm8hq=lJF1_h_EN0b8#5WLe(Rq@Sit8KKsyWd3euV%Mxbvj=2cwv)
zll;bafEyz)DPazc)n!vwd_e_4Dyw#uVdl6o#}W?$f0r!PKxf~E1`#MS#)G4Q>mU1%
z<)H`Y4~P1)Vv{Ui++Gz6RN9u#bS!@uvVu99X3`gG<6j-11szbtqzg<6KuJ2vK>o^S
z-T#;z=Z+tEX@Dj`qs#}qfshgV53~Zrk~+Xi5L;nz=FrjgAeU<pLm<dPV|-gHr7z{O
zif!sq<d`u{KpeHE7V_@ggn60+f=}QN!WYW8Yofg{P*mMtRHNjpH>iYrDTQyhj(^u$
zz!$%Qzr5~7lAo%*R>Gg&WDy<D+oexEEZb!rdtqLlcSZAaM#up-g44Z~_<hgptA+L~
zQ|2k!m3t|uY#sY;afC9rOKmp{9aA1o-59C|n5}ZaWv`n{g)`;(UHLe@a~Vm<iT8td
zhArX*!fSeKt?f^PyZqkvtMSaa(|e3(3ADjYhTO=6aO0t`o^jp``3j`sMj&6;XnFpu
zA>8yPm*8|3m!|-EYUGv8aqjYx#bQm3rEnvnc(msT-tnNt<e7?&U>?6o<TAz7?sj3$
zF|R8`($BPS+I)jiAYFc;e(|d%_CrAtlb^-=LJZM0AMZ;)VOB7<N6b(wB`e``|F=xG
zOjMjgVxX8-Iu|4|Tq_7&p8POTYZ+2PKygue(&mqEG|wK82`Qn+d^G5PgxU@9RH@Xf
zXFD;V=x{GAQaS%EKH3f9yXnk21Uij@sb!E_J*>l#ydU*6l)@NSU9sC}wh}&CY`C<E
zqpm6#Z=JQ~MwHgID-R2DGPKNUPU@xSWRswqKUxT<By$a@j334R!5f2zy`M+VSIC8B
zY_d%e;WltUE$l|LU@6B^Fn7h#Y5v;1(cNy_vW?V;a20bMLIKP{r;x2@e+6`|^{s#p
zWyWMnin5*N#zXPV$z@@%xGN=<U7e(kRO-^1jgf6#XxdrVyCjcWZr=PiE(X`DeM4;4
z^=k>pZi$Xy+F+q7l6mxT+=ZtKY-R}Y@(5DX7gY~n(m0KNJYn5=m4Kg{pYr+Dr5t`>
z8khAVs8!Qa=H^AIrX6r(8QF5e*$E+ST4^wDYUXHeViQX<YFXket83MByxaDFAZ&3<
zUb~Fly2@vit=u}frR9n&!()E35`L?cRw7^>sT{**zDT?=uNKuI*w8&m1wkKiH|j8*
z8Q1_82U7=?JMLj{ae0cW;O+W#PR0vy*2P@-@W6&RoETcKK#SNZ?~XFpOaD66!D!!x
zX<njYf$7g)Ll&(58Gby?YLFr6V6tr-f=OqHxj+tf$^X4;TzLT%{~qPA;W3D*pIRQp
z-fIb-UC~QGg6A^%`)F*1eYSRS)V}!iT7VGn_=EFzY?@Gpzy_N+e<(~TY0I;y>NN@K
z?C(35D#Bh9GZzOLXDt~_q{_O6v`ab3b@W^qnO8%x^Udjv2^ZMzZ!awzyP{3G8O=4&
z)!&zE>7HO$C`9kZ!9%)2DYSnA{o7YgRh-ftJTPFl1pZ!6b=5(wFlVn@YNbmM^@KtL
zo+Q{?+QE9@#e7Z7+Z`>cgc~02FD?@th|LAQ%M2KQVm)p9I9=;a9nyyp8xRG$nGj^8
zl9?R;1i9}!?9SrBYp1ch1%alk4&pMOjc<K0d>sC@3^EZ^7*5$R9>t3ZBVdujy-GQy
zZt-L=XPaZSJWYcdi;CxC`Sz`%QRl!4NFXiX`zf1fX?^uQ3E|3n2MF@)acQItP;A#Y
z<IEMxsw!H^E+4@=YdOCb!VMcONE-1k;>1whJE^j9D$w+S`sJR<2gU=j^-6dY&kg&^
ze7<AONcC+uL&vrttO$NMrNf!K!Fs<0{5}PBFDu04?jz(hZ{K?OjDJ|STPiab@RUYn
zPXNikWzg<0#wkP*L2oWrip*8=>lJ!M)v4{=C`QaxRJoTq+VtgA9+aRM;yg`9Xi+~i
zPKaVuCS(^bmKUVbBaUU5Lam=TS@TS3Mq!%rDvd64xxu_!YI2kU9R}&YoK2&LhJL0t
zrE7C>bX_{8N4g)&gW~Vf;zHg9b`GYs&+c@|N%vB6PUJGwF4i>JUhVrJKju6>^4#9e
zX9jXA6MJn8&K=R7IZT2hsQkK)2m${e()f)RL)}mPIt&R{0*&j#>4dk~DOX?51^X|j
zO|~!lFP}EzQP{&_8mM1Ct7R&%oyV@4tSQuW3XBFg8@K}a-=#84`$vrc!j4veF!BtY
zWMOd&&A@jMJ?cc@t{yz<WH2>wFwPQhC!sf_$>C}lvnifr6b7bUu{M`3Lz-<n5{)CB
z4L6N0?|F?jJ3#i}Xe93Dn;Z|ijZkzr*_$=Vx76GjWvZ&19iu0Xd#xBm%Km;gaW3Ny
z-L;Yi9X>d|Uj=HOA!!Q$71;NjsfEx3zWA%ZEqU23m5Q{C_O&Y#*8b#JnPgx%5c(jY
z9_w(8I9fo{Gc>O%?r$q6>8YFE9uB_4MdY70y=~L^izqj&DTxo&2Vo9<Z1A@9b9*Y>
zA5cp2sjviJcXgV1Hxzps#4JT}9Y)c(Wq5Gw(=YUBb$h)-dlnOc0O_7>U8uyrC{Y5}
zx-h(>V52?mp3aWA)kx=%X+EU3J)((0*OB_;f0xcc-<5yrtv+#lX;BbuAzuE5a@z0y
zM=miOg`9|hTx*8$4@fDL8VbI{B_{Dp6~{%F75lDvF436uc_>Fm*BUv!ed=icsm-4h
zvo_f>R*<~-*kC+DK#@mNQTjDWlL1VyNV0H`DYWv)m6GjqT<vYUp)<hS_T?e&i|nH?
zs?Xhj9w-bF?`l!T+30IW6c`=l=Pzea7w+;Uh{=APJ0#fKezGrqDHTk*hxZeyqhNoj
zZpP0V&8xij0zReEXUneJWMXiLv&Rz8-I%16yVq_G)Ia~(+F3d&3tt$)s9*gok?h^m
z4e8KD?hz!R{_LcOa$4lHYFOS*aqrE4&U-xF7RL>H^r|~}JPlPV&3owoDO1EdArJT0
zn&=6vufri-*nv%0^ShJHxq$3fE~Ia4*3x^q!h=z90oYSH#IMdwQd~VMEgOTr<ZhuQ
zh+d}g0M(Ki19bt#NX<m9|6dBBn5-fDChiG5xkRHHx1GOy1`YgDXtffg`{(vs<y2;+
zOPATXSu6Xf!-XuF)0Hir<7SgKlHW$)eO(Kj$}fq~3P6C&FV-Wp(<(h&6&eSDC>_mO
zKazf@RUnMt#s^@Fk4L5!^=ZedXAVso`VZ5>1F`DzMiQAY&-V$9hU#or{eJ_{e7@MK
zFMpQD;<ceFlHB?hknBsiCz{Wd?UNSM6^n67LRl=W?DH94j^^j_qi|mPv}N_Hl8*3H
z-@@LuG9aM31R{J#$ZJYM)<tcz46l=sJrU<ay-bfr*HVAqI9SnjnbUfJ6|f!53fH93
z03K}f-#uPi94j7ZwO{Bu4t3Bl^_D?Hwsm(p2WVGKyiob3{B+f{6+p5o>137Rglfnc
zF$kn~bX)aWm6_MoaK!L>CWCzpchdA7v>H5?8(ZP9FCNqDHqWM=-4b@`F$bLqGY9d_
zPDdKKWVQ9a?n5H7z#7X-)y^QRM}J-#z<1f`7x$Dk)%r3n?~h-2F{Ah_ExS**#*PyF
z)Mm@`nH!T<hwOUK`eQ`NOVfTXDHFXp?>^5X=g-&;M)KN03V^^H`4F0cH_>9GtM{gR
zU}OflE!2~Gx?IF7+#Tj2`z`KsSf5{EM{YKZiXg4cP<8rt7$W>PaIp9g7?Cw1pfPdM
zE)2Vr^7ex%5t_nPT8ltzPt)FD6myL><#D&@54tR!Y_59bk9HfRoXL~g6+A7joKF8*
z4yXhg%Y{Z}TL!2JX>tVB&>C$lp73nk6ucThv)v4@M&}}JLlI6G$xKAFpnSoM;hEvN
z3hpX{)$g}5H32(DU8?EP<jl(tz@bz>PWxy)23;pmv7e@uv-$lt=x)93caNN^`K8X<
zgHH3p1)m2GwScZyA_${^sDWmAB1p+y;7*fkr1mT|Z!dTt3=%34g?J!LtNO_%bZ<it
zq*Vf=x79kXMYdW#w<@!Pe;SnHkc}uOrQ}-sRA6*Wv;B{VU85>flSE`*0EbBlN=5|=
zlKa+|;u|C+&HiLnT>p{zdX>KtbkXGbvwAzaowz@WA~7u{7^NaTCe4yDQNIW^c(OS^
zgIcSP5><*4)$BK){79rI(_$8XTEr9aUrY?w2%+jGR2*J)OEGuWBG=8tC@aSkDX)Og
zv2lyVorLYQ=9$fa5uWG<u=??ezyHpD0p%A*Y1sB!K+!1kezWVg^j5bz%&Xnw+(}Yw
zNHs~Qa{kWn2>q$R{^AH@vX{U>Nd14`L+?$BA==C?5@ZG+*s(<I6y?vRH^7!8HFuA{
z*{14s^}67r9&CcU29k;y{0f8msWJd*cc42gI&*MH)_>~)WS<C7UJ=_MgJ~`ML8tZZ
zb>pS=5}-yR<w*oiN`x{voh=d(gZ^Zb44jk;^`n})%gk^dAx1aY;Ux=$x}S~|G5GN>
zn%cUzb!$ugITrywe)J`VQONB)g&X`!cs^xdT>+YrqQ9p@A9Jo3i07SENtbGzi$1Ej
zBc9s;qnd?~ZG}tJhRf|=!CMMzC7WJiO;?Aemq15r3(oD*#oEZnX*<6AI+pc<Dq>3A
zh@}%h?cwaPu0T0$Vp7efx7MV+1FOg5-cFt|@5^X_Y6f5B2>pU@%ELu@#Rj5fx#d3L
z{s^S8J0q)S1S1IxpHnClaN+s*=7E#YE|&^qjkxTXuP-||5(h`dUy0?tH~->}`d<IK
zeDsN=KOaJCzp^o!7cfoXi#429+%1GzQJ%hSKD%1qC3SGWIl7N}qVmsgHkdjp#~K$q
zVSHHrM&!pQ+On%0XxK2dZwZ5VTMe9U5r+yyHR8?=+Bjbpve-?G$4WRUq4%vkv=;`h
zp%pT2w%HvYSX!`?W(HFF8%g*$qX>63b9{n0YlDhxq$&|<YS?4yFav(E(WOLvvA9G{
z$T}xysHK8SGoEidM=^~(UX;z0K|gCeV^i5CYCBVp%iYXj<S5ggC#z>^BrqZAOGRGe
zFev*g9!h|Frtubw<i%6ZW)$vOp-BUL+5_El^m%9}j~hqbT~*Ew#j)|FhN1QZyx-!L
zB@JS4xDcE7jo{ze;jfHNYP)~7K9K`yJeNV5lscy0K*vaO(LdJOo*&P5Ru)d51U_wS
zkg=7*uC!kjfoIHe`?!rZifW^bpNB|}vTnFHryWY~fViIja^z(kOG|6$OGgBr)mVJP
z!dkfBlRs8<`fQyqQqX0)b?Ejt_AWYE9h%st%nu|pBqU$U*G_n)7TQPXMOB<JlVr@9
zD%o-!f&Isg-@BE46~A?<KXXFQMg}Q-HYuaq*U2XvsPeCURoz_1dYf^U6m~Tv_Mveb
z8>xos(&4DEeo_d{rp}(pedz$WDXSZnuL3_fKH$}(%dapk-WP110W&4e>2TS_sSi|O
zmEg)MoO%RMczqGuh=@nDd2a3_)Q=E3xHqX$*0HqOF>tv=9kTtb<o|MxIjCHr{2R2^
zhBCTt$MAB)$%1k{P+QSLd|G4Tk%2hkE*|QFd7PFNk%eOo0$~FJ&;od!6R-Fn-FOgc
z=iuCi_g!{`Q<xo|pYa3{T4J7JNhcl(t<i6~C0DYFS~KMfxE2e>$l+#q-G3NMB|3zh
z=f-7Md@1l-d+kSfCexD*($<d0Yv@Yqs;OEI#8LSgABUT*M`0{V-5jT*KBl|y>J^{A
zS4yJ<wj{iJO@3jrY6sFQXRN*vCF@+vijV|oN_%WB1l_FY_%~DaO-=b4nUGpi#s0mM
z(f<k-`ZZLFPM?B~N}*S&xeH<uW&!F`4NfMyTAR<oC(qr2C<#<*f~t~d<^##>K5<@2
z>9y4}VYFf!x(I(77}5S&GVmv34y?CbRqhw2zMRYNhf|XSci)$JQAo=zJJw3}s%1>-
zkrN~tB|<}ey3Jz1<>sIAu9b-fAu@}okt92lwY-U-v|Mf(IZaf1Tbt8y|Fzia86fM)
zp4AAzPH0G<sja-5m57_C<`yrR3(*Tc8$|@t#pt0i>0W%-6B?{O_f{$NNeY(3nEROL
zFj`ya!oW>;ID_pqbkZa4XW;Y<O1w3hESG`7+x^imU8wcgM0VT?Rw=TjfDTAepD1i*
z6f&@({aTJSBf<QnWLr1&(XaJBWnb&h6Ebwku#)ge30YUB%;Iv+r(kYsZm@2_c@gC%
z!r+YzuFT4AJ1AU9JH;88d`pc6P$_;4+r~^Uumky54Xf5yetRUFMRLM_4IE@YP2J=m
zyWbqeg@rQ!AL+JG2e()UrZ6E^OfQyWo4X^gi`lMuHxeEI<p$>F%fS?KMk7=-uE)jz
zSvB%`fG#SfG~By0nw6*}{>+2e!~;|TT8A8i%BpGt=3d76iPT2MtE<;Gbnh0{2SzpF
zIzTLjd{6oab1J5HCYCzh3@dqtR+9K@B`rcTt7LuJJ>jb%=H7R?o*{N)#hi-daJE*p
zX3j3Zur5dR{B>^Li-sXaB8JkJ-#>OACnHHd!!VLa4$xcsbe|r*;e_3HSUJhuKutS?
zcAWCa58{Dad3=)?36YN-g_N)ogPEWhWk7t{g5{*AG+cg6P;sj^AUJEm<HN<;{q@f5
zZAJ1}Tnsv$S~WZZiJgjo{m}jK0fie5M^`}cbTb^0-1O7yo>4N-VsV`m+i3XK>N;Xi
z&;9m`aCUI{2RMP794*6n^aH9x7CEV$ba0wrFHzpgn^6OQCh0?pj+#iyTWKgoGGL)<
z=zW^l#$rS-(&n!I1@q2E-2s#Q8el|rr0d8+L0l(0FDQrHmGWA9jsj7wp*=g8N*?PT
z^mcjez|C1QW%<kcKr4!J&e3n(rrYqGW^!~|tX!sSMmcJ|SYyO||20JUfgCp*9cWA7
zFL7!<6aP!_nyWiQ`r<=MDArE723X%7<*4JEgsUDU(ZfniV-XtqW6)`QgH%<`C|_)e
z2#w*AKiS3u(hiSn65J*NE}r=z*UeoXj?WSheIpi5dlDuVO4)U_@3Y)I?#?eLOE=oi
ztGARNdc3Y|n$GXn4_HcC<RqMEgwZx=fB$T!<14Q>;G@GRelxg_SQ0tp2W-M<uDe5R
z$iv%YSTL><>!>Azt05C|Pkk-lN?D=(A=dk(`@bhxZ*KcwGgFA&C2~?AA|_g&{LcoH
z`Ev(18-%UCKDuY+`6FT+L|W@NJKokP8(a4?ipQYd)dY@A$DflLe{EEQ!n~D=EOxQh
zPt{{t@xKFpW=NVCNg!)mSyBpoBxEY)WreB!S?-ae`aY&%-3nk#id*toE2bQMiH3yb
zS%QCIwU=SF!O*#s4#krxl@9N56S4#g0zLP1vO-SFO&^Z`93<>AAPP^C+Udj2CFIN$
zv6`(dyOT9wtu(Z04MnDq`4G@2juGnSs)p0)bUXvyWx;R9W%+Wy?#7}{=W2Aw)n_eB
z=1zC3=~z!}oXt4cSLK&^Y-_FXZa$BvLM69?mu)FHpcZde>mTl8qr8WoxVHLlJsDR_
z*1oM;ZNa#yUdUPeJX4kr;-wC=PYKRM21~AeQy<pO(9t{;3Ha~MP4=foGT1A?e{0(~
z-ADKP21BkQx26!cKX78?I6Hssi-#S<>QWbF@H&Z?PRN7BmV(8cPInm@_QLHcqC>Z~
zdE~2fDzeZ%W(POsQr!SYQ<YAN$158CiTJ^JOzN;?@yz770=8dm?(5*7aPa4ix$vC3
z@{<`2NbOpmo_Un$yqQ2<HxOxs94v}e_fAXMamntlL3XKy5Nf3gHVq75JEcH5D(MjC
z6`S_r-x^{v$8f%M#<Yy#f+xO};>0YE5q&X^c^OjyWDEAq6-Jhe;$YDfqDe+zRM5D)
zA{K2qus{P!Ylv*NVg+sJZ3^WP{z3}*2AMGoZ%uBB!z29s^l|N0auU<yyJDwZJV&KE
z+5e+c9UN(-C~w%L56^x@{EbZ+q_L8gL?TcTm2W=7{Do00?1o7bL{NjsL<9e9<2-=>
zwT&~2?Y0L`L`|r4ef?3z4dywSBM9R93|A4bmvwVUmDhdU`6)DE357x-X{~G}H$&M-
z25&rKTiWM_NO@mEvqx_~u75t|qyGhUtr7!<l@{?+atg6%Fs%C9OufDRY8jmU;wegF
zE5LEi6AW;Z2~AZk8HEhU+Ki%cfQI{V*s7{e`qfTIyF?oPF6GpC^9$<088}t7cz)D0
zuO#Z_vs$2ox{JiAeHpXNKp921xUV6RBejLv{C6>0DzqhASGk}iSS3iSBo#|l1?=5N
z-}`EZdLP2jgoOXF=+`JafarjZg6WSu3VD?%s>k2Q__o*@(|r?(h`a-o_5Vnp=@|^y
z<BaK@(ULat20geVZI^le(co$^RW_SjhGE9LFI@_K>((z@h+ds?ECtb%Eue@JNVF42
zt+Kt}zMG37nq&a_Lpyx$K=`}oNv-*aN}(1R|A={EF~3`~pG*}x1f;Q9cwP!HK$O}}
za%9)(S@3l(ArIhkY;Qy4#O-N3;nN~uDZ7gL5B8#xHHrNFQ`GKR2E0JpdTXsj(LBdW
zVbNNfy2+xSf~1!A$<%aIa&11xW6bHa46)Nf{Hwm0iOJh92{FA>0I=wK=_P3F#5ww0
z`SQ+u+Ppltq3m<8J0G#s*SGyDapArd#Mk0PKs`d5tuLpGj+Eh`QDE^56@RT)2o-@%
z>2H20za&1!`_Cfm8M_b3FOk(0Ci+6FI!2yT<<>njRS;q1kUJK3lVil~T4o!Y8^kvV
z`(3LR5}~}DPrJ$qAY7V)`7CpX;q{~rLB69p-f~^>Y@P0{*Hr)d82FKX`0LOAt*q5J
zWvD02y_`0$xbYWO(`P_Qul7N08yU(HS{)m(y<T5kKd%);aN-p44!PAn3GnBJqBd3P
z>yRrAxR;K8pW!$9NYwUwM_p*-q7TJZ-CJA<m%~B?Bq)^~w?wmp1O_rS`EQhW)-;lx
z7-MW!4yLvkkq5sb`(pjq`rI;?1U=d-Wx-=q|1Q)}3zKqOU$%C__Oa?Re~10CPWS9z
z*>)+xx%pjLWnMfgj`jc0^^U=veNXscl8LQ}?M!S;Y}*raVjDL$C$^1=ZB1<3ww=xQ
zSNnglyS3F-=Ytow?mboa^wZr>KOLL+olw)*#4;pIr8}Ua0$fw`n<jQGv$4t0NNggp
zmHZTov58b4Fj~j_0QYMw@!)9F6?M;=fmTAlH74CrFRjIJ{J1-wqG2Wfg}7k^M%@)Z
za_dOW8ph^qx{JqJCo8_LYa{f6u9a8<Tn^+5%ttv(6MjzgmHm%%gJ}fDw%nYM66;}1
zln*=Y0Na!xG?GK`zn4NO`I@03Z`=nGS5+4Ws7tm|;X<`7cg1Kz;|WW}l#E&K(x?tC
zL;b$dPd}#!bwl2*JYEyY58j++V$RRi9HvRh#VDLBjl=pY>Y&;sDK7G>OIdRG?*ID^
zK&{lAbQ%lS@BBqVP%sb=>VWRe?d?4Cd|5)8U*WGq(&N*s+W>qKHm+Z*mO?J>=i9on
zt9>3@M}yVZk-iQ7lT_R33sCaX%3r7|Y7wgR?%D>mI1fKs?6f?8W{~zr7{}gC9MQWd
zM*wx7h<%-Ypad8Xq;MU(%bD9xWHRX01+qeSg&ECZ0V$%u9hre}8#1ve3(*D~^SBdV
zh8kTlBF0^bFaU%>(QRHb%rxn_eJBjWHWs~5ozHDnI8gq=9eXGu7E`chUe&5VtjvhO
zcNH_TJB^eVtX*VVm5LW9HQvO7^=noLRmV+FJ^@L%gX^uhN8lZFcx9C#DX$X#Ey!A@
zKKKIOF^Z^Lhkafhs?V20u=DPry-hX81G7i3*1$0p2C`Kb%(9Sh|3@bZ#SnpcdLx6h
zWtkd^aZhUztv#s1<F>8G>0{?AmY?7=7ECt5&*-}H`uxXmss12)A5RqM>I=@A;~}SI
z^0CKgZFs!$ba)~mKs}Q9E<oL=7B1JeXeoBXe-$<gJfc47CFD#y5j=H~2#I5pREv+#
z6pel9aHTP=PW&JVB{K$DN2qb5YN1bhBj-yq{xozjH67JQSBBb$<GIb|0Xc(!uAJ#5
zh@@JBE@!c7ctvn95DV50{h~jBPn+_W4>sfEe4`GAOULl>*hvvk&n=faDYFxqQZ+tl
z2~-}Mi>TyHMI(Z9mS2r@TrOqf?u!*h)VTvd3Ev8M*iN>d;3rnbw@t(b`I~OeBNC$o
zhM~#)kL%+X0$l1AkHq4*o?yzr2I>q^;uoInTwD$E8b9K`PyE$8-RxA1GyMLi8H~<e
z8)wg6QW*<rx!0Lx1_kn_V&Qnfv*zp5z09|%IN$}*8pk*=GSy3t*pO_}yn0%8haJe)
z-K1K%rXMEsV^$8EdAEBVp${~1vi$!1w6PYbm83fdHAP1s2XQr?HDV}1B^iu*8smPO
zD?v)X7v}hf=iTM7eu0uqUb_p$fLI-EdI&y&)yi_U$5*c>0(N-a!#Bpb5^+Ra_&EVq
z^2TOuyvC#EavRl1n67@K)SlzCiV<LIM*gA(LHi?4?1VWjRWYKc+<NN!oksXO2j#vc
z*0qjGsb|3CW0Bc1aV*(1;99#vrbGTNo4+x9G$tcc6~PQs!WwYCR|uwBQG05ul|w~<
z5}*ytdTwo?Qm<~LiFKQ>^cf!Kw&;)>ceQF_Ac5PUCuPolrI9W=i$cPu5CIaEX3p!N
zeG>c#i7f&tOw?$$bEM2mcfPOd%XG&&Pf#$W;52eK;1HJ<2u6Mnqf+_eZ$7d%e{jQ8
z4Su;tHMluuKg)P!b^YZ3EZig+r&{_y3tuMZp5&j;m=@NdYr|n_fe?oKpZy|AC#$q2
zb$N)S0yX&E!3ESm7aFIB`>Y9p@-i1IJ2sZQt^Ab5$T^OnNAaiR3Od7Vu_@7cV2eOP
zA$?%Gr5~ulBHq~^M#hg68PAH3YT78U8r>LDPL5sCpo?n)(V(j-OFC6!|6)f~p>7A+
zt8bsQ&Vb5@gJhvom6X|+&mBv|HJ7APFf)msCl8*HZv#|x&9%BrI{ska;i6fib~K>~
z#?@p$kin41V17b>iRs5yjkEJPhp3?%cmEF&{l5T_ta({!<)6PlTy{%Q^k20?j&Cl%
zC|kUtIu+<n*b-`S6$jwxf9AnxrVezlht9orsUT&7sF*3#;HP}G>KlHG0>pS=#m6ja
z{iBL8800FU7!hY@a{lMPLq(Az-Vo5%28@KOgOEk2=ulLSoP4_kuc?qyr4)FSXi>iI
z7#`7yo9N1})K@NKCoE7dq}G!B1v|VruJG6_#g<DX8Hn(KEGP&qK`NG(kjxcGrAB_i
zcEJ&6(hvO(cl8uL81C=~$e{lb9wc4nYDhFeSC)?GY)BOT1B2)1`Fv1|%b$$(#$&^!
zQ>RxtWa4P(ji)JMp<YnmlyR;|kdyHPk&rBha7s-DhZ84z;^OkgtX~`X-m10Cg3-tu
za;f!|iaU}qQYL-jsiYh&KpOVvZGpPjl1P~VYNDwQJy+p#8P==<fX9$?@Go2>!PpQ9
zy5UT~DG=E{ngjo9<a-Ea(3hz$MtlL-Y2%WV5#Z-w-bN;*B7DkYY=3F>^PAF*A2#Ph
z5KensNI@X|g_DF-bc$zynQ_%&K%I+Y%6ypKYZ3{&xChEFvG2@{TAABdk<pR1Vh3c|
zjHDdCOBRCsQHX%D;-dF2o<9^T&E;7CJA`S|c5e)>FM8Y0gMz|J<6m3SN;QXtivH^1
zmP_*E`{38<HGEO_WSIW23DlCr(aMsF8W*^4{ALS2e{ZhBV;O-bmE|t^yPjFQW#QJ1
z%p4FBgGl@iMdb5#^YNZG0V9uJhr%6^^9j-B?U2G=12|c~DN8li5c`i0hwgDv)E3=#
ze;x-clN8jM>pRS9K9p##Bxt^<u$V$kX^=HPzs&Iun2$#mx6ruL*ADdhu}xwCn!}iJ
z&fk$Lj2;-&!unvtRYeB;wNMnw>B_KG{)BoXVGYNE2So_+ws(oe-N@VzX&HLCN@=Px
zJA4uMm`P2;s95jy({P|5wg-5BViU|EvKTU0gpftTTZCwupLLGw9N@HFLSZ!1TZH@%
zoyQxj0icWV+iTUp7ZtXNm$R@%zf9|oaT<>XlVkLM)W(cuEt%~c<sk9jr}GP*@bYPu
zX>Y`Vl>fO?Is!fn*cbO{HJU|z!x9z{+@l#rMle|NMU0ZT+b4|DY4Xk6045LQ*JYZc
zeFdeQ3SBDdoB}(E%skjZnmI(|CDLk!?!I!YrBPG+Rv@-Q6QoFDPlT5`j!aldNStv8
zbWXB5AGe3QISCEx?^KA&g8sWaGN=eLyh1K9!bE)aZ9$X*OxeG1lgg>m9n!#H)<8=7
z?^CgA)gduKU<l@nOk~);?ld@%Vgi1_^<dE7S7X^XL6hm9F%ZnDsHb8f$stGtatj2+
zaK`M3s{y2HEI)5SG9i6EEc1ll63VoVsvX1#k}zN;cm(`&vBCQI+M_U$>e!fT(+(=&
zV@>D0xL?FcA2^XvQh$ywLa6>)<y|I9F?&%t+{0CkLP^5*5Z9Tp;P7&A1RNJ(mow!Q
zL1xgOc^-$r?p$<HmP%$(+5m6W@y1q&n&Q-Jen8&CByjvU9GPOOYcz4OxZZbCLW;5Q
zef{_x0ZMI@K-9MCQj@tn(i~gLB{;Y`>}BZys@$g=A}N7{*i_d3NzNpltMKK&w*Qvc
zjEe=)za8+B+5fQObBxH>gu(uce)X`LIjhAFyK!OSsS)o+w{=fH(EaOj?!>oA;|g)|
z<Q%AzLvQbqCu`p*1nWMHiWh|3@aC`XzK38Ld>LaBTM5)WV43B_;;G&L%;tJ)t=$rC
z(ra7gl%+_V7O6X&g*w~u$LCr(7vYy`$Q6V3;Se#;UTk}JL8p8vE)F>8MJw16$-2nM
z5FGMFn@;2zj!H>-f@YtZ1{tq{UH+;H$OFVQ|HG!*vYZAi22c!pdRY#5*^KF%Ol?hV
z4+^+kMJBhndy17|-y352zEqGQ6OB@PbfML+g%ABxSxh&W%zvgZIm%E#Mh(ZAysc-l
zGgAcTbKNL^)X$t<*~%0Ck745_dwe#M{nCHn6ly0dc{2Au@EtN=x_R~hKn2RgXs7X(
z;kAez9tGl?Cl$W)Xw{H8iL4MCHdi>GC=t6ZSc$;5<V_?lw9J?pt<U`HvV8QIoN2hc
ziI5|*g7(mw9I!c*4bWCxNiV{JVkU_o4_li_{H_%!0v+c6f=u`xO6Fxw=Q`B?8YTD1
zT6=>tCb^{^CNjp5Fw*r1C|N@@cX}_EI8*I|c9#6k<K>f6Yf>$atcELkp~?ycalOmY
z<cXcHoEK7@fp)?jf4NGCt!9=}C!B3amDsKD@xY)LaiizJWkN~_BSR3P>Q^vFb^~K)
zrj(ye&Ey1=n%asQlgGfIZ5EO4m*1f;ASf#!@KrZ)Q+F5{%%X4sj7IBImdMjB2CTdB
zdLX%Rkbl;!YGf{1Tk+TY%2Q&caM`wW`TD7ExJb#3AIDhpYmjP&5U)aRlYNVY$vfis
z7dT%^XOvXK*LN?^`N#y57nguEq1fw`PC9+~629=1PF19@TMg}HNv261@*+U1^IDB!
z9Y6-J^Rno(Oa&${T*_`66*&$(=B&j3sz$epQ&xO<3|^)&=zYT<5e^x|;buiI`gSw9
zSPLa!(>lNlIJ?kTprZWBH=|I@ikqn9o5?b!KKQJAe3y+n9a_;79X+!N@?=R}dWOjG
zuLKAC*#OVN@y)gAB*T&VA4-3Ok(sp_)<fm{#b2`39v9%OU%+t9A)9}K%mtN}SDt^}
zqJhkCnc%u8;VLP@8@|F`M|;Z?_b6@PGUoU;3o2bUXZ14EIlCzk>0!&91FAzUO{bqf
z+C5ciMRW!{&n>(0*Lq))0Gx>DOr?J@nGQcb3T?yBt`@^SzDskt^Pl{}!;TA&bIV52
zLh0iGaTst};@(+TMTsVcT8GE!w2h|sU}!a?WlEN4^(ff3_^sYsH@NBpGprw$B#D!L
z&>4MEBvpY&v1CU3E2n80D=G$hs<mPHFgH=vO<NJ?O8bYB<wW)zrlk&_>3FDriJZ;0
z)lT&6mPhdTf<vlkqAev}PTh=;Yqa$o3ht&?js$o>I$2Alg@w-_Kw;9JQ5s&-KaX%Z
zJV>##TQW!NpZGI;#o$8;{h1{62jgS7MwC1f>Y&-fiM^?BKFEVkcX0;-UQ2nUcH!QY
z8{YP*0g)>WYcVPlay+QPvjfR5_*~K$Ou+bEw(lNikGnjIF^(sd`8!ab2}RPro4+j^
zW(}muqa0$v`sCbm&FGd~cHi^T@=GkbHe|eX25bm;i+d*{u4PL2SPTnKv1|{>&Jvnj
ze@i*wg+;{MRyKUAFYLzWkwfrqWq-wea^)Pq3t<_i@N@rwhayj!8VZ?}BF4`+-25<R
z5*sI19c6S&Ql;D0GH94Xtx_M&PYO`8odR^TylR=YA^~lQ-*%F3%#a!TaZ{-W$Fq!k
z=(ZMo*{A)B*~xWAyeL!2Nh#2|98t)E2%)i@6+aW9m&1{}lxw9($$Axy{kFI_*Q}^v
zSvIbeOv;SYr0Q0|>odrJM0myDuHM9AzFc>W$q@@}L##f?uLS6N`cr3(T7AY@?0`Mf
z4(yI=AWdjWa(j<~#ixK_vVnQ4+1nHkx8}?Ni+erhL14;Yz>>+9-%JaYLDn*w6a^fg
z?e}FPBB=B^<p$RjJPRlJ`2~GnD=;jLc2gIcEszv;n;)}s`%cv;mY`8{*0Tqbq*<SD
z8JZF-R;OxOgtRNrJ7GmCLsc<R3qaVs&5Fj-8_P3MQpuGGb?{ef?w+^R+AEDp`wE|Z
zvCvC^>W>YT+feggkLlze+SW*EljSN4<?dGo*QDhZ8x|Y`$E%z&w13mYkpm5Ge^$22
z#Of9ONkniyKdJU+UrX#;6&=EztEjL<Yv#ryt(nHOi(EJdlJff{V2##f!vK_*-*?j{
zkJD{s8SAC67{P7?W_Y?~a{IJm#N@w~_w!;gf3C@rV-!C0raV@JVop^3LL8jRTcL3!
zhJnnKaHaYZ$H{`+K4MwFO&HDD%SY&L@UvDy%-bSe-j02bnvWBZHjF(JZ+{V;6F=a@
zA!Z`S6XfzK=B5hKK}q06R|7z+PisSv_Mn`z`ZJoS?6us|l4Yz4l*mwatixZGEQ(1p
zschGhSNIv1vf0?#&pkWi%*G2AIym<KaUW0#Bw2-P>}?J|g~@n@=Xd0ER4k4D?SjG}
zT~b0BwJV+YXn(9<Cecmd3I*D^Q@@4LlHhq7d&ce&9N(-AkMcsVNCW$45WGWL5m<K~
z<F~pv&%3Tt7<|_0?5KbEC7l^%CCr>@$IqtR&&%{CWthPR=$T=`BXp~*$#hA$d&sXz
zsko`Qe_9C9))AhJIx5hm&J0yDFq4NXIMs&Qnf+wT7lo&>*Vm37PPKN}uBI6=i-WtI
z2y*d<v!JKPb5UtXZv~2Ye8laDn-hA1T#Cs}ZSA=mv6g1VVQWl0|2$h0+jrqqazEpz
z>d2bfau*UX*3IihJ>aJDswB)FR2PhG7>r;bPL2IEQ{?c&(3ifCsYj6NWroK<G*;(G
zPUd2SV|a$GS0;*(DZ&g`{pwfUSvD?v&ruDM%P1b5oSIMBI0YJ8QZz-d4osoYTuHBO
zU>NciK}ijY4m?OLWJ`o8H|ssQ?MfbqI;mXfq?Avc(?Y<YgtrvNmFBHF>>!f+V{+^g
ziM6ALl&f`UM$kskAb*u;lFwJTNd%<@rIBZvW<LBft9+x(?dPXwrD-PfdG=<rTtx*@
zs#~yHhhxpRkpW8f!$__xf(6H#O!4Z#am$^&N2+WC;qWk&5)Y@Fbg1e!@|_cko;Wc^
z66AjGNMA&%H5(gdzKL`P3qGDs%X&SF|Jbf`C6UyGSApr3?zh&}Q!ucxtvMqYXos&J
zRdUe_`|T5@7xXQ8S}ZkgDTsojHT+3BzXAKXiE?0J@oP|;QiVSE8zt01o33TRDNQVN
z^ARjEivRc-vg2_uHcgx7zYD8ZeY*lkT8n<y-stsV!KgV>^txyNUxog7;L<J|r=xv&
z_kWuFdM(tkpp}}b+#E0NUu!jM&7#Cpf0)_TxxU}iV0(2E9qNf2KXtv2rhnWY{#ExJ
z3M^wJjx+&uCj&8M$*MM0kyEd9WkRjFgS7?d+jTG-kBJ!xXp`Vp5g#X!5-tP}_LAYR
zibiuc&l32;H%leHk}aTsjyTU{azp#Ko@9czvj$%1PU9_HSGI9h%iz`O`aqDq6*KGf
zB_m{Q+6Qw<)OxpN#mV#!k<p~l+TBsmAwr3?gF|4c5WyWZ88hvEj*$ZI!WW$@Sw3g2
zFVQow>Cbgr__b$ZyH`r@;FFGr;J>GiKS}s{Hz_z><yc*#<8YCi<&=-mfA_9`E?z@K
zY{$>qu2_9UAV2Sodr{`(Oh^~pmjBBrGTh)*U(IY!fg+sf#Y^1c4f<fx6(;O~6|;OI
z3StE8Z{=%sn+m9=X99HFW>vT9S&h(oca3JWJzK$8MOapjJxkFO(8>dlO|V?jDQkxw
zn6H1jx5JBJFp@-^{l3{xF)~)n8`^oxSwCe1rB=bxtgxKZY|1Z-HBO$~mIzX62lIO#
z)^50;GSJ4o73j0G?hjdjGW!6jL?Q`(cJKo76t@dCIw?IUhpy>7K}fIvl$}bsbw<x@
zK_<ILpy`TOWK2+@^bSm@QFj#iUtz#p{+!Fofx(Qpeqa$P9OOCT;S~3@e)Z$3BHSte
z2F=eUNsgFl9<1smNj$>sabBkNNWaB)k!5yJ!6XlLRa?{lg!)I<gkXxJsI?LK*uW6@
z59-`KvvOK$V`4h3lC_#%zZyLZUvXBBQ`os}a4Ym<ahZdCJu>7f)*QMGSlVjJQbdiU
zB}Tr0b)I`#ZI}E<1nnl|=tfMKP~t(Edtf(I2>DA&UlaCv);QZgJL`xg*TBbbEso98
zNBZ~MB6u%HsD0#hL44#Yj?bHSDL{)owOul|!E%3|lMqk(Rg<oD_Ihg_eiiBh(uvy`
z=d<jr9dB<{{LiQkw~HDva?U>nTormtAwG=nWWoE>o)1p!r*&=j-VPf!Uvhyf@UEEJ
z%&C{q&&)ds7Ny2k)ldgs&^TGUzo$jiXcn-yMXBf^^2-*_1L15C4f|L<jRB)y-6_}0
z{OakG<CJ<UCwp%P;dp5OLA5q-CU1<TXI+h?%%|wI!=IL^HPWhU5G+~V{wjenoSdaw
zWkLqMqonQ|Pr7x*F3fK?HXa2-TYTc9J*+n(;G6&9FiF}clK&9cK<KCY)6Rt%<RNuj
zvP!l1&b=#Ijn39&#V^1^LY^J*ZzC(g@nc&EzWsuf!x2hXKq2Y=?ZM92y`3H<FP*O7
zkVzoV8f<q6&*bw@hKxqRA!+cIlsW@G+k4+CXI(}{6a6t=E*h+{TeE(BfVqOdGOE*H
zr@()Jnh9L2;j7zy*K8s8d!AOF`f8^ZMKD_Wgo_Nek%)q~+1H?4?8=hBK|~4E|ATg{
zk=RFt1Fav+!9}1u%mbk<1Oj<9L(T&7l@FS)OU6)9dgiokMJi#OkM`1C<e?Df3HmYh
z*j^E=8<+IXtIbQ@>k6rIia2M>-HpL6L~jGu?`H!*+qcPwi=Z4Q(9aWgFBz4|AWDM9
z(aG_iiU7{feO>(BH`E~``pwJ{P)3BuU9D1~1X&?U2&-ur<k`-{ayEZ&h+!tp%Vy+B
zRQ4MPCNn=eI_3*hX^a2)-L5b>tV~931N8ISpWAxC<Qm%6I<8KEPn(Jvu;IB<vC5D^
zCOK`LLFQMrdddk>w4LsZyM}iTxt{c`@i#BfKME-yYGw{4fAZ3!d`(C9xCQgy)tq{%
zRxVEiZ;`V07B1D1vT`Ku8nFXWqrR-eGAs$URs*6~_^_eeM<=0c2+2759JjmYxya+!
zUE&eVw0hp(HC<BjyF`O<))Sg{mGn0|wF&hR6`7iI`ik&xiAHK46D}5F5m*?CKy(G(
zc+hOElxpujr;^cP+Pnov5UKFE(D)C5x|I^mZpxfnyl-<>Nb;Knnmr7mdUDeheHX+Q
zYAtLu{zULTwi=f$(X+ld*aV*2;zV6lmNm>r)R~AGgC;UO3t`iL2$wcb4UtqDNJ1NF
ze2-o_(sPGTX?B8}0-?|4%4yP~0AitKVel=NR(twyX(u*owwjPjGtVasE=*bm14$~d
zV@QB7f*3;Z_)ggfUB<(c=wGp)=!$drbD6YiA<2vn-^j(f4SvC4b9UTY@)e5H>6jQJ
zETCc2JEa~mGpYPj@Lr9c`s)`7ej3*Dho?1Qwv9?`zxV}Uzr~TRj9qk?15Ofr&otno
zjKyn7m*dU`qD!mY8Y>)0uIunnB$@>dG%SjqN~7i$f-x68Gqy1M8S?-69oa9~gCOn}
zj_rhTSsd4lmSlU!;R-oPug+*wu7>wn^hwvmTj#r8AvdB%TRt-xyO$N|`7>qdYV#eY
zah`f#zMQHvKY0mGbVUzc0l;fA$NTA|-rt8<7*zwIW@<^U6yH|8>Wf3aM^UJS2x(|r
zuC0cY+%}&zeAq7F$G*6Ov!i?1mO9+qmV^?xp3hZOTbgMaTc5RtmJrB|b$PYu8B;qA
z#%sL*A)W4v0{tU00o^Z((P*95#5cbJ^N-6RxK#ICO?d>P-ff>{fSWVHv)Li$N{!h^
zoyq6#pJb%jI;<{fh9j&l^_g0u#hG2#@4K!o?ta&v_j^7p?dy&W>ykknKH}ZD+Cpxl
zpbQb4Ezt}b@YxwYv!=ItfkTwG2dh_Z=@kN*fLW#<ZxLFez+iJ0Bkz^Gi2RTs{Odor
z{UgD@>&HL9B#>pt0V+^FK8o?DrSGUdn$F`w1Ik3IC^IwZWUZP`bw{TAy<W$ri^G@y
zc{Dy#ft0aeXgg<iKArIlu__Xyqm)W;5Pbq~>xuvuRjD?m=_(K&ZpaKJC_)-gv(96!
zNDk0{>FL|+J9m*YmOr3vr~AVE&32~)H#Hj^i(q^W37oP3P{><re#U?I!tJ_gzv1B%
zIh~5I)viU=s8089SIh!*AFp^2HlFYUo4l@YTtBqO`ctkAn7#zBS?)Yg_Ie88hsf=B
z4Ah_6=0(!r1E4-;)IYagF{6fJVg4}n)DYqhK^g~42AfgzcO!j5<sqb|M}ZE2%c4Ty
zjH8aD&gx$Rfp=+XFga^%6mvK5IxO!wf90O|%Y)qQa8&My)S{pJz$1RGyF4P3MBs9J
z%00Y@S_UY@ZU1oC7^Y-(Z;|2a`1m0Wwbfdeacf0o;7e}B(jMkspE6veUv<Y_`3-aj
z^>#}M^5D-{XTbxK3W>-{T^N{MT4*83%&2A(5`Cv30DhPteL1`V{&&WLv>lqEw++Ef
zcKlv!E=?gsuG9R{X-nZ;(2L2a!i1synFQT+_KQqq&G1T`j0e?V^d_4bxai+B=}bAA
z?7Y*3$uJ<ww9JIaztdG7zdwJ~TPq@~Jr0}m_Pdnph;21C`XYio`rdF!>Ql5ndnrOg
zz6Mhe(0g{xVt@?Fi1Q4iiObf(jLynz@|69wgT?ZurkG$(2TvgSFccJl$1wbDSPh~)
z-48;XL{Mpmb!v%ukd@sehcx+=?~jE2V{jJQqg#`EmdAU9mmBqan3hjoUJi*o=vBf$
zTL~31i|4R3x)K4&($%_DYr1%85}1@`^zm&`;6`KnmFRMRQ&ZPVHzq^Eh!?Parukm`
zHHt+e6brq7s$U<pQO1ba5CkX4oVGNN4+AFkrcaGrv!{aeL<&YLu+_iSU(zt)`+qie
zoL`mmKifhddufD>XWX`uG}bkBr`Dpuqd|*KV4ygaN1KltUTIldrr{h!X{-cx@*okL
zOeVg82EuREEfM*ETt|37fg>AHDOg9Hg=JgOUnPl_o0HVN4KqiHpa?jGYcBjo3aReS
z=LJW0sSkMqjd?&8Xvyd0B;vJIGm80?bdF(SbuT4oLB05m_O4<_*J*M>xrRqNbBP)7
zyX&?Sh5&W&o1-@V2^}8Xm0$_*;exM_1$_54Jda36OH;Q$!f0!V1wS&x6m6xda$z_o
z;crw4ws=)|npMAFeQSF(p1JyJM4ZgEU~&5ja*bAfQp_j%T;&~k9wd!=ac1|rMh{n_
zA8r0xcPySXdSseX9+_=2@_~QGwFNnsuhu=&ZCaQ|McOhq+(}e5%J&0$lCK6e;g(fg
z!Ir=JJ4rY4=^i{C&<GJ%Inh~RwXp8Ml<vn>2I&73M)*(oIkp}(GVRx*%+~2+a=FMp
z!moyBiz=S^)QCn`bfXxoBGK~T9RnXqwEYklaBae&eN|URDI4x>o<-pl0o}a3h|k@l
zu^JVEq6sw=W7J%wqF%rpQL%CvKeCTqt?(gx#2b9XZxS~WNVtNcgc6}W?U5Dc`TrP7
z2%s@@^xCIxAL%J`t&c^}4QmRo5Q!q_$|MVk-7BuT8<ARJ)7KNiIFVWr!R`F5&Cks3
ztslCyQx5%AvA!4q)1)X3#B(>om7P#?RGXs0yg0g>kyTSU`+yi1menl|!g#et@h-?O
zhyKrImvP!L?PW=+UJz8;;}*UHvvW^HA<-Aa^=e}fJmUpJc2nSMqvi?>1)j87ATfsG
zyK=mw=(9^QkQ0>EY7*e7hV7(m_v(PU{sau|ON`XWa-DT;UDWi3f9|Z>T2}YeFC8~e
zywOQB7uVnjK*mET)6~fq4M?Vp(-b;^)no*j&tWp3&_5tlJ(SwcqL_--PRNmz7qw^v
z>e=B(oD)g^=;E%b`k;<TD&vV~n?et9i`OzDu!=YwsRmvf*DcRH#@D9(M}u*+8_sb6
z3pB5Kgt~$=gFdnThAyMFZWynk&p;Ww5yWpG&G<?b5V6U`O0q)5p^b#?)fnb6*m^VW
zlc7(jh3cc<8vAtoZK&b%ok)xdE4SM`W)3Ah90j%lT2x1jDh`6_2ntleaRBMZ5fu4D
zfM!JXXY<;pt9fS}cm|4TMVbCI4-I7iuXs)ZTqD}?O+gBNWx)aDuU-0a^mxbaGxn?u
z;T{N1fGKF1P-Uwb%;-iic<<W}Mn;Ii_JA?iu>tm$NuRv>)h;YeR!KE>y%6jE<D{!@
z^Wi%h9#4>pT&Q$N>5@)t)PrK5XI-Et4B0Ms^&h;5^Bjl1rI1fwyquD8GZ@!^6IuvQ
zjWBR>VAxwxj;a-@i<YpMu&SJepjFvJG_E`dTtdrZYoqelo>1FoSNHPn|9;}<$QJKo
zxv2zRnmz^?|GC7#bbEX9+Oo;NzMofvJfhH~K_po#*58F|ygZN3AO9CkZw@~nonkt8
zd@(ya-n_iMx!Up9gE=BhzG-=J(fLH=>(@W^G7)^_f9|iE5F~fXW=to+4RBdDP7(kS
zvuDpAIc!lKp>bO%Z|HFjbCQkFoW{<~ri?m>fB%N$Uv~WvKe=>fHc&TBJH+C*98q)y
zvfp~`_n~A-tH`~cGLQ!+Aj|EAE8#TY_>8cFT7QS&%^TOR&YarM#VUH_N;3SQ3+`Vg
zW-^0Yu{|c&kY}<;Fo^V3Ox`}OEo}idLbcFNriMArvk>vhctaWT3<Fs2Wq0s4PVw-W
zu|p*UGOLr4C7|{sWdB9wvZ!dCaN!vZ{Q+r)bYRC@)7JLtY!xV0>*)zDx%tPa+6msS
z5g5?jiw_J1B|DMl5e{XZUkxc7o<0o)Mq@s%#phy*GK3q#O-1C*iWG;*)xiRCXwq-6
z;=Cdm46N=<OP$w00w%UKG4Z|`uD6cLXif1mYQ<685G(hym+eg!J&ykKzq5go<w>=y
z5>Lks|D6Nsn%9DWLd5j1TpqXe3x>ToR%yl$_D^Ut{QI?{{%gb?N<mr&LpMV@Soibh
z)eiVv)Jxn1LM%0$Yt>gb`<?;J;a$j)R)D6aPeBx;A_D*I!%qZZyhSe*q4J)3p3s;;
zxZeZCAH(Xu5PMK3>do)>!mOtn*=3$VYiSDvUB~9Y?xM(vWA@u{^@qrVn?`d}*XJku
z>W<&csq5s76_Zb~v80^_doy_?FL(MX?Zl)1wFYC^@@Q>8xN)J4Ht<5_#EIjLA2+FP
z{I@fbzdtl#AoJ8UZ_T@{$*Ox1z=xd0e4ic|Pu`wqdv>^ePlFFGkf;Be?@6za-Z)p#
z6a_c(mw(I9&pK+5*Hag6;E>~#<COE7uxpUkrBr=Pi-o^YkCMl6dW+PHiA>jQzOU0^
zihzE3Cj_TwAoK1QC{PDYICRald8mAM!gp=OTqJ^<p3pT%fwnV4I5#cBoL^PGuAIF7
zNI>1N^tfTDhh>?(PW}e9J&=ozSQzf(wY;UV-b;vzcdW~TbyOvOvQ^=|aQaQi!h{U!
zaGGhC$fzI&8_Jyp@%a4EJQS}Io?&<ob^jZhvSf<X500u7DPX0j#pp+=#%p*yR0gJ(
zyU(!gZ`T|=yexlGp^6W|CfW2|N{`}2E4J71YD&H7Cd>e#@6FXob{<|t^qwG@3lDNJ
zeCv#*_WU}(=iZ!5PxG$&-s|N|U(a32)g?+lt6sNzuWV91THorMe@%AkG&@<3uJ*}w
zIZ_FaCXas8r~|BLJ38tze;v9A0bXxkg!$~J(Iu$<{Xwd7M?(^@F3%b7YXZGNk>z9!
zr`_iIgX!#zL1&p_-Uqj!1mHZ4x$1B&nr(5)mhJFEEg{&xS0xaYk|wUWf^~DV`}c6=
zo6f`6U|)B(brzrB1h}j+2+1kI1B1x4#zr4<8KkWh-vCo1Sc9ceRE1;&9#&hF?3ZCe
zHADNWmuran9sT3RBS~B2zfpP&{##|Oo4T`67@0QHcAQs!W2}87t-S8%hA}SE=1o7h
zred6r>F;@t2(c=W1$n>RX~&%3s9)-TzVZ|o0-+@GteQW_^S{6avI8wZ>_+|1n^YTS
zb&D3T7t-gr-K;dJHV*f5hOj}8{#;LUpks!tfKN$J>cnzExi4i%^dtUAj2G)bKZVpa
z#c?KOwaVW~yj4Bq;zuy^YVl7y-KSg`QgaYHobDLMX}p+8_Tlfb+rPnOIg=p@Cd#_q
zz048Z9*HS%Xus_2K`U608OI8TArfY{H5afn?IZY%KiM`kEr3=sP9*t4*}5+ekkLf=
zAT7eXRgK_;;mUfI%MVST6|Yc-==Cr9n7{Xy1&q@dDE@I@R?}2hRkK=4F;s%xzFF3(
zFXpEcryc(qN2k^yk;;EbIC&0OAJd~{W@_$@DB+npKD$gXZ$_{*NMmE#$7Lj2j0PN+
zvyKMh-^x~aV||hAa)4<v{uI+hn)W||a|Y@XDzP+Bx#VzFO7qYYyIt&!LP36V+U}=W
z`*Y(>kYHHrQs38wI8=1y;`3AWweHGN%#>%q5VJ<&vQ0*K=s`(}49$*sXum}9(tG*u
zXi#6j+o@ov>K&=?-O^Auz`F;+9s(dU^KhJWx^0?lsDjKSDrUoItiN^lnzy#@%UI#V
zV-j|h6B`JL#6h<WE62vbGY%?bg;gjHeMxYRh0qnzr@=&6wFM}{LQPGB%Tk&v%XEAF
zIE+e3zPehVo}R$^cjEp*gbt8=5?ZvPCwHhSE(}l?hPKZQy-H(}X&B(8CIZS2r)2Ds
zAvDpqI7CT8=#dO1#(ibWf_lRt(TIH-b4p(}V{J}NkCZSv=0L!98jop%E$UIOKOa0q
zNqt#^g)Cf`CtQcw7lC$poKpONgiex@b{bW}mW!C_^(E5Sox2k7WSCM^R%lZr_)#%_
z&{bg7m3z@;@flSa52ua^Bnz6LrcPjmoRCeVE&eIuKSP8RlK4~L{+w12khIkOy+c9p
z&8;u30K|pSTIUz^d#g59cu_*$^GL-q1$z#TR7J279US$)9azaDq;yI{qe5Sd=`9p*
z@|k=xb9gc#>`>$ikf<Su!8wT-a^A=`93u(ZPk|3(4PZqH5v&gpINmp=6r_$Gwv$my
zYb=Tx&Wrf%=md&qOjsPux4wtM=pz95?tWd!g{ZV@Xz)~ys9Qius#BjR%&&k54DwH0
z6nFt5nKCcfAkq<%6sm;bI=(Rp>!e3<h_rl+hy*I}5h15Ys0q==X`s2WP-#j|XBf;8
z)6LYk69s0mkgxOu5st#Rtc`u<EzS6M#F~JDEf1xpbc^(;?yY*MK(Y@7cIQPNPC9pQ
zK6VgmTs7+bCtSSJZ<ILpqHetwLC>)lEuvNILUGQHTXrRvht$)OiSv@*%lNQJG2kyp
zZ@r-zh};5*eW!BFxRe-g`Xxy9p~bQizlYSjIrLH7Ie^aqfBMI0%Az~*gd0?oy%$az
z+d{S&k+>&ISPy7@q}PWTqZymINVh!WOxzN8jZj!2(vjc&eiyZ@6;ZLd*6B^saz&)l
z6LpHE!!33lWTApX#fDyQD8O`znvv{3A!QA0%x7`*`dikNzWhi&FY?S(#RS7-9nQ<Z
z+cmdyP(+*rh<9;F+_H?)4}G^=!Kk@VCp9bTp)v|CO_AEmV2SY7>gq4(+Mc%*EIO7{
zP0!({z^x5Ld=Kxhe2MR%IZ-HS?jROrwyFR97df-Jod<E~1zK3+R5j&3Xhf^(k<7!s
z>sQ4ap_+@kRWooQ>Z*YLJHs~G7}W2==k&GN@DT&b3X}hZ5*W{-t(xFR04HRvK@0YG
z_-R!7vF~iEzl*1cMM%M;bHc@-3k<+23;AW(5#@fCH6jf${}vn?S*&g4pAr2%QtHYh
zBH}v&kBp;GdcD1k52kVvdf^^E``^h%5J2lr{gP`(!1_|Ug`*t{r>hKO!F*(hd``E)
z(E|wgy6H!k3wHz=f$BbPO_~b2f2e|lH!#@9>5sv!X`lXR9gr)O_O;2PzX=F`aQ~$U
zxn#?5&S6s0xLK|E{Qnye#)yWT#SaGlboprGd39}f^|U_iP-~h$Wo;AWhqIKFpY?+N
zSglrS2E11Y2>NGoaieZ+4A!gc@zOi9Kbi@gJ68B4y$wGnZ{(3Ne#Rh3-@>mtgn5gz
z9{LbJWElW9g74CSyg3v6aq|uv7lL)g@WFZOn(p*bitCTJf!xvO&6c+#A>a1xrtX0&
zvN~3>=VJLM!WP&1XQy+nA2ozAZJlNt@ImH~Km|-nL)hx?k0sGxx3oH9XB0<1vtJeK
zV9M6LxAx%=xHel$d>p28jV@4Ig5&}uk#w2RLq>9Z%a?Dd9vf@CFL%TDOFv(soG)Zv
z;~3W7?-(c+g!=g#JgtL7$wIP=9(KhiKaLcvZxmmf-{N9|@2A7432@F&Ba^t|)EJ!M
zfX*lUY0(~X>{bbXr*B9tZg275NuPxB*ndSXi9!#%q&2w2H!y}qgebL9u&b$JK6R&}
z>n~BB*dafvgJW2zTO$hM!|bk6en;etj~o16z|<DNDuyErmHCmOylxq~hRKCt*0jIv
z=dg6qpoBJd{nH<+r?B=41pcF~-@A${4KQc!ANXU;`VO1c$+2&q63ph~w@wN}M%uH+
zMEf^)6;3fdpEV9#mCVoF6M>yT+*RRRV#=i%nh2sy3iKh5X|X@g9Uhw9sT3Ryv4{!7
z?w^8>jHxN&yv^s8iZ5iIW-45o+5~wBxLSiU`xg~#f_0tqx=r}L{k)hCgd=9t1?WI&
zpqR^zw_V}IuQu5HI;$be?}EgI`L7Tm>sB|e<Q^7IGij?mDT~FsB%wE@(v5>IgH!V(
zpuud8g2ZIOk?n*ht#nif^09@!{@7lhz)xq86d@w!NPsG8D%y6Z==h3+28?J#@0kU(
zoMmuHVHBfcc=*oU>UH(sZKeQVo-dkZa$8&=o@wi9{oyM!!XNu_$S@#8`!B5ZE3a2o
z4lG$czW#L3`Pt2Dw7(xro?Of;ihyz3qyLDjuA8?aS?mH_##Hg?>3E3`$KxJK8=1QO
zS4pb>B^__pW$dP$=iXm{6A#HY2U2!C3)zyT^Z+5x64sCswX7^Q*9(wq%PcXRq=^nD
zj9G=C?nb+v{ai6Nw7b^V^ZIyFdF{DCP6oA6`n-GsdiG+(;FD!A-X>8-zdE~rt}c@n
zSx@TJ8U%c(sUMfL;ZU7}#SA$}{T1vu?!tY!n_Wv$HY>tDpgi17NV}D{sB^!%jyS0H
zlqz1dpouEwpeZ5p#sn%i{>ESX6Q85FS}=^k)L6wRi<3J3BS?1|9L`l1%r)z^Ppj!7
z@9gHQ_#oW9%)ac<8v<3`F)PpW{KJ4nsI%sUt*es{ctA{jZ5GKgHtXc%+xU+H*x~J|
z)PcFv_Y^#2e9&NtY>Vh3&$G+stUj4+MomU1b~QtWpkF&?yntPCzix<d0V8tRwZ@1k
zqRp7kj-Q83bmll}-uHY#H<ZK*5-AMZeK}0&i(VI{g}0GVTx0iFLFY*fMQ5cW!&IW{
z;4eAmyS_LB<0TH0zJV}0e^;iW^rsheM0_(ib&>nLf3z4EEPffa)0v-5il~aH<)mT!
zYKt+Wp}6{zGJq{IKl+S(6z?H1NxFAi;D=>yDv#!OJb|H$(aUka04+?-ajNJ*Te&K1
zjF^jWm*4vzTF<L5$E`8m>pVxl!v9g^gFk(1M8aRfY0Yi<6AoII*h!4gZ1zVq27a+_
zyHFGe4)pZRS*FKtc3sG~7<q^w&U`?(UBl&b?zhRUc>t_#!6|d%ndsfaiKQZ>obd0<
zs~4={_JUE_l;IB|%q?v1K|Uld^<8-$#e<^dP1Hk9%IQi+(nWQ8Grv>eRW4hw@S&8A
zYc^ZI5%VHzUd#1-BR7JEx6bYC-pIN1)^t~m!eBkS3ZmA5X^-<k;zduReQ`b4z>hhn
zNiS4T4uBsfyf_b|waG3ex}}#Z$Y(8^Gg{<SXtbsIb%rn3@g(G=QIw!q#SI*2qrt#@
zTHC_*_&E0dV9RPR3nN^nbUpKCWDbRV@64(FXUu?8Aa6S7+~7F^Ol?YMz3ZP{%E7C<
z9%Tf@Kt^y5&UqFj9CjYve!Jq`Ij`}xL$wm!1c(B*H(&F&<J>65z2>sCnVWPLvC{MK
z{3DwADLda%_0c>(w!8pJpu3j#M5^@IB6jjI>Ad<cj%&Gef%1EI>dXoJYs<U9+ious
zc=A_yhERZCA++)7SJv=8s9bdSP8`z}^q{%nSf%n#fSBy=W~c8v#6HaaF~kkJ?Y?^(
z;OKJ@!>^z6=n%sPuRV&8FktxO-&DX9qbAOcb`8~)llxVpov0%1d(Q=#svP*APc1B3
zTBGY!!sPj6(98#678%)o-=0ZcoHGpE`MS+LJvYQ7e`~(w%dk+FIyb{oZ6(TWyQ^<u
zX9sTm5=V-MS?{Vn!;sI7y22B!nLbEUKzL##-Izu>=xC@If%T!^>EK$;rDpls{?;G3
z0cCUpX3bD;Rz;{!9<{v|`98xw*(jLEHmtU{NS%OEU;ea*$W*MKHx-SN0o$anM%<sj
z;B@^bX|Bz**}!d2`dJC66w?f4u;u(&hY()xCl~OOF-9>azlL1GIJY=-Xjn1k0x4Aj
zrSIR?<OGq(0?w6~C}`Z7nee21jC`ND|LKO7&}!zi$=qE(at)@GXqRY%(DiQaVmiZx
z+915srM5WK1T)duyDQk%fmAYm<=&&#u&t1X%n*A=H{|8D-{1(@AnR6ot{Vl-Bc2NQ
z>GWIJUp>YP6Qb~CU+q7+LgAqjfR)}lGb@?0h^Og70wnxr6&PG?(`l+W!EZ8oLszLy
z*!1~s)Db-)sA=C-%=zU?zX!gC?FsV^a*P`Ugi;=mW@~OmKxfVk)XzEe(n!)e<sFF>
zI_1e#OLMm>`Xa2<hs%GduW7k??F^^$S;!l<L|Dr$VWt>!`?LmnE*A<g13jto7b2Nk
zwUESPb2rzJZD=0K>aRVwb(`)#LKZ{ovSB)PDcd<MCqBZnXDazAZIgTPq|YFW*#g7X
zA)isvU0NK=a9x8xsZ+I`+>F0FnPs3@T<vM0FC?-b*ykS6q5YseHC5KfNw|%L;j;T<
z<nCFGI`=gKQ8RiPVX;pJ9uW8PQ*RX2&fyQeI%HgSy@NW>^7fU)BV-K?%AOn*CgrcG
zbg0)4n7bkw@`yRN5l>TQa&4`%G8o^@cos;4(!{Hzwa~?Vq)Ya=zKjtODir)60JlfX
zT;?y~_dOtVn@9{~1)xxtob}j^;hGvXNVivI+m>ydgXG$IQR9tRfYP}S&^}xW!u;*e
zH7^2$$)To?<INY$zT`{R25*ZYtODWu;6T9_!!s<j@zJ<?S&?5F5)sDq*r|Q>n*B|;
zYz$5ZVZBmJo~kUg>)JB?^+C}0Vm{{a<}VN_qo8nl+GJ{*{S>`2zl=7=Ym8AV3)2=*
zZTEAQ&z*@Xe|&8{z_ZE@$E{h|T+#&|e$@+Y?^abZE+<1#ebTDZc;n|aQTqX|#kjIt
z%-7r6TrD@TRY*BWji^i3O%HLI<*JbSTnM^1VYif+W<p6>(~r736ih=hZfH?O!+wZG
z2QZ~L!qu{#VgB=Zc1V$nKHH6ZgJWnE#48VAZkoP&F?CV}Xh%3X9zLc<Ig~Ur{~?v%
zWl<{az<IgKI;ozA$K+>x7xH5hcVsdAjC0@2XSIz?h+XY_kBi}X@}Y#c%m$HxeZrw<
zB7?CZf+3^yd);G)Kb_!%g|!PS`TlF7pzdcnwVlfn_j%I{6>`_^VfLG}gx`#UexCaZ
zN(9VP*WLsSAcrNfy7cc6xrRjK1BE@3`CR-%u4#N=%)v1bZFO%<P<oF(mL<r3kYPB?
z+72h{pDL!AN@(t#!=y_=z#i&NH^>%htqt%gLI?jT<<_pIaOj18QP{$P7u6a4(0`HY
z*#qO()Akn_weMuAg<(zipIze2?L~0@W+=+pAJSk51n>eJyktRp1<)LXcTGz?Bo92C
z+f3;s<(?Q~nDin|Vt0VkadoWBpLpse5#wXKe#Ai;#XwTDU+%B3`VOw7Z-~WGSA}@f
z@DxGuXV7h6@xA-ieu=6ABFmU$n6FzexO>r`!wF0#bZ(g?uVVt*0C5!-pV+_XVbvYy
zUS(LI@?C#_x^5+0)|+<oq5sB#{ir^RWOa$}xjc~LBWUlSz45!1@ZM5P6%Z-)veA6{
zFd#}JP7`9Y@6|$y<l$eKb32^LSet-v5xH-_;zu3~C;f~e5_fC1VKhu6=Yxq)XkAY6
z-agNBUW`GZrb<j)<I8oJeEHWe32}C2(zXha`^yh>yvR+6>vhG>rk`$~?;oC?j&7eH
zv9~`<VRx^qEkH>%ua0=)g26f1{Fjlna1%mt(AJIKRK8@FFe09fJ4R*6Os;NE+c`Jz
zoLWa6DAYDRe?-C)^CId4?z+6%yxvbv=$M5@$X%|mdcGQ?+bwwdoMj#`lCJ+ktOID2
zm$i=`)|Zr?b!HkL(-m)2@UFI9^b<ipA7bj?usao^6HGI`MN4U%8+@t>@jkHMtDc7U
z1an?Z6sY!$sGo;6e?Ii4iBJ!l33Ad<s@Q56eMq(4mdr9fgYtbAMQ7?C6{2?T?2$?3
zFI$ya6p!kaRmvqRJI8$HZJie4_yAh>%WJTyUduusB!C`)2!-T$F}fPrgLy~AkCV8W
z9j9_Mdy~&km51lzr;sNdxgqY-e&Sk(D<cTI=Bdhy6EbR3=kABZ#zGpyKrCXhM9jAM
z-x7X9InO!$#A1J_wY{nlVWQjLU{Q9S;F0F4%7mJ7^(jkt8IMV@DU-$J>i~?GLnMX>
z#2C?1Vt&mLewd(QB)J%$*Yn`}$k$a{OdK}g^*pi_DU0{>youK*@OQOk%%?Jh?R0=<
zNp0Fxn{lo}s&RpmO4SN(`r1ip0E>WOMyzgg&uZOlEL`pw?FsikXzamIX+3jsypg_H
zxknJCy*}iCh_R9UZ&17GM!@FJmB<WS3TKq$nR+vD?1<@qF8Rv}<N@H7`K;##8`IMn
zl-<5>jUoAW?f8SQ6mAlX&|Nl^p6q-VnJ%E%dD0!h(#HiwXYqPZLJmQ1%XdD{R2Y>U
z{dxB*4NSL|Ao!y$rfh{RlVa4{(l?udW@z+yo=k_|e<5Zr-mMU9-T}Wa|MD;hvSmW`
zoaNh9b`aP=%)3sqb*e|CGfja{KrYXmMb4a}SU9g#Vmui*X}#~izn?FCU>CqGy85rE
zrV9&5cA$;7`~|D<2~LKq*)elp{(Lh1e3Vh!7h=tSKfC?2d^s#-U*|UPZc`DeslC1%
zo!q9N*o)V4@pH9faSmWKf`(l$Hh>N%rV7r%@=q#H#qu|VhQ*m43CuZ0h#E|`%PjaE
z09{n>tI4SE3zy8_T?(z@1rLQ)rawlFsWBs(uF*khma9_lA<sPW(lG1b7kVGc*z&b#
z#Asag8S#?rD2aGh6V$w+e|tOX&h7RP(PkS;<5BPk2q3`Q07jZXE%ZYhUOo8&m7XK7
z94yQ`4t#M*Az4gLe-*<28W)y8GkToMT9-puL^H@Sj_tt*?PS%VI<2o@b%*VM<izix
zOfZQLekn`nbAL%?&#t4AW0xUX^c*f6W1#x$8!&B8?ltZpjnXU-Mj<jH`PuW$w$t^l
z+~P;iz@D#(9Ux2`gGX^-JNRv_ovbdE2-M!GKvBJ9lb<X<f-+4;MhoZ)o@IFaEs
zoDBWfg3*QkcBS#=-;6#@K<h)zKLEndZAX=Z*;Nq2(oPcW^?&2p>0m5e8xO;Gub+1C
z!~I3{Ux5|R%7k!*eYOMP(xPp-S%}$agd3R+;QyMLJEJtYZKv0nS6sKfb+EH@J3l~!
zz46nMZ4#-e&klR^#s5PvxC^)M5@W#wiWW-5{279i==q<-Xi&TpxhL*pd|a+661SKP
zq>qgRs}zY_*O2Jok0ffzUXkXfg3RUqm}>PCi_C0?$ruyE+U(CzYsm^kI>c-0YF>L5
z0RDu=A9&(vWbCEYH>qTlG-pnEsKlEd!a*yQg<esYxiyTUo%^vnlTXz#d&}+=&8+_W
zXO`E#K<;pm2_-dMWM8SvoN^%q)C3wejf^e_yG8l<1@#*hk-ZC@7~j=O-&&rLhOTfE
zCWgAIiP0aJG+eG}OadK8jJiq&(*$0ApiSSc$?@L@C9Xw{>zn2O3E69#1;2{>Bc0P&
zlEsD&*sT0Y%Pvn@U|@{Qxw`|r*~j5*w`g)59TjHoArMF~X@vg|!nXdmP^fFsj074W
z!cGZMBL8dID!{XvFtG^etaU&^l=9?%ZbS<%TxH2>1aQ;m68|r*-Z4m%t_v1z+qP}n
zwrx$@n07yH+s3qQ+csy~w(ZmJckhX~asO0Q)XyEUYpq<FE7SJh?a#aw3phU*m0+h7
z>zu5DR0Qj_YSbA9%z}8H3OO&Ql`_3_A}|)z-Em9c#2->&!BYaA5_|VoSlWF={%WN~
zth^JCG7R3TY1m?f(K;Qrs%8ml7FS;QeZ<cr>$cR5<T`7I*jN>q5-LCh%sVh1k%N43
zsUqYGkQiEjE3t=WX|;^_g?56f;5|ULjhOueR(kGuBc1i6wnSQtKIL&|*Gk%2^H!UF
zWQ({SbfpBsh;&WYmLEl(Mq|{{;lvaeOk{^#G%aaNBw;DF?x4EFO3Gp>K;;ykU<T8=
z%akG9TFTK+C^jWqa5oF4S=I#4v4;D!ymNF0h|9OD%S?(rZxDmCsW<>av#HQnr;}bl
zIcM2gYwmIg3<H}9Sw@NG+7E-YzHWk*D$qY;*OQ@S4=Js<tgnn0%-Uikls*Vwv@tzm
z+;gH^&sa{Tz*S>3O#G&x@doU!MYRtI9L{;l*Iw%1lsENW)$^(?w(Hcz1*(4}Y35!5
zU}E7La`xMp4FvF-LP3QfF^ZOzg-1`)FX}Zmx3;%qFZ1|bwg+R%h5JU#a=oF4Ycq|$
z7un;suGQPOh_;j*&rrxF7~4q!e#3<zM`Mi3Ys=j>qAW7ARB@UNy0_+U3Da9&J!a);
zRp6NO&LL7N5J?UxGCkwR)g<j=ay{b!2u6x+jemNt2kZH>OxL#@RZlV=%!vt606nvV
zK-s;t&(fh36(X%5+F~LPvkjSgGvFXsW#EzoqA{Rc2MSLOIJqoGbcOL-WY2b^4Eah0
zbG1fCT460Djm4n>iAajv-*1<|*%WUY$Jaa53XDng5dC@_RN(!MYb0=w>Zr2-6Iz$f
z8#<0`8~+uV4L)Q()mu7_4c^<7Td|NQ1zQf?u<B#%{`_Ro)AGUHI}@`^rY|REc`QAY
z-U~rR+e{^^W^qI-a7EO!7AqGm@zNX)-svao?-I^pR<DU`{A;U@nJH2D_s?mm8w3QL
zlz(~7Ej&AhWBjsN!xo`MrTTDy(22RLZry2YjKrFLY0RpP^#7P`_G!EDu4%dZ2P;vF
zpl=*pL#ve{=6e1fKjCPZr)R*Y`s?R2&)>?+>tR&hvQ(*c{CN$RD{#|I|Asxc7|)vw
z^9F!8!KkmZr-bBO=g9UHdl<h|%(m+DGbe+0+Qz<Fs_g5h^)_<ul_>yGSj*yFnMsFJ
z$6a})+bGz!ZRs9sDj?zbw-jN@_I!(~UDh)p=Vh68Qhu8LK!++PChFEc>bBIY?MGmF
zz%TFM7(W0ODi85K<*jcl63+*Wln8gIWU}$s0I|v$^{I6!FW+^s^7WzY*>;pyXOYT(
zjQ+0bV8IWNM~J!kG0#x>+-2vb^FNZRitBeaTd&@MGlJ8O1rM|&sh{^Ca2EROw|~j+
z!8~^SduAPSSnSVtUblm>0`B;~FWPRi-x#YbaGbao)SWU$2_CBfU04>sX~xqM3+kJZ
zQD~tlYc^l64c%o2pvdCu?KgQ;RaqnYA{6qD_1^VN@=2B$O#qwwa;C`~5{tl8FhDS2
z-vT|}-@AK{*Y%$%Vi&2-y`_0VNLJw3{9+QX{j8{x1CI`0w5m5-Fz4Oqbdx*0B2s)M
z>A&i3eT>^~`!(iKb`TuXkgm)_wKdMAmaS+^(aQ)vIU!#rC*FZSio*rnTAJ5|PEgl=
z^RQQKuOl8}_X7kHhF~&rRe2f`-~Py9oW7LYyQr<Jj?(a{+~p_X#GoDx1}V@)GICHi
zZmY-Z%!<+hanu97yrX6VfN+4|{VCw)V6G3tlEpv4%Qe&utuyz3nQ3EYLZ+krq}9`j
zCl|{a=1{ft&gYZGNem));N+|PQXd$c=%g)s&?X%Jb_aN~Co&>mR*QbF3p@C*<Hgrj
zuvTfeU4&l6Ui`_szT8HF-jU^$!Wnlh<Lc-rFwKxX`COi!Uf!Yqj2gK%V-vaC*<ZHg
z@?b<9vB-0~(|-=xFbc{}ETN-Ig`^_%E#U@&5)cweOh%gI!1$#|Cvcwkcpn2eVFg)A
zA!CMyeg)tF^V28R)X)?*vf8%Og$f9kV!Qu7kR0Qf!<U%I2|Z;@IYrZcLC%T{-S`I+
z<zo*R)uI%HB=8nq0tyaQGRR|_YbePinWC-*$m6r`AUR?|BItw4V>zU&o^XL}6jyu$
zdHZ)i)R(3o;8uM4yP6i>ItxPJQcjyw@<4nVS^%(I2@ipS1agJ)6G$MABL&niXn_B4
zDih0iFF`yRfU`>3paYpx!wXM>mV39>0X2pYvvUn%6)Qrsod{d>lnBE@RG=`=C-Cf`
zJi`Lp2bSC1DF?^$hI95I`|z)k=oNv^0a?~U{_{fTrZA*sL|YvAh5MZkTHT%X)NZW}
z1`0SlMu~6sUo(Vq4*m+$Ff9E=UeJ2>;By{xBu$y3v91T@e0!t*FPjv*Y*G))BW`6W
z^&3@5clf?@E&MZoLUp&=gt->)&Y#;&L*s!G(-rrz=QLLWfx&11a<{>0;2&ytDz=oZ
zpufdyTzq{KrYtWc2yDuNG>kPx{8Bg~&j8>JEz9FgxP4R~m3Dg#G1TA9!Pu>!tI0n>
zWZmshJ5P#(5*aRU)(ha_%m0ysF^ajKhbTa3_3)I+0c}fR5cManY_Ly<o}(NyPRWLw
zkt%A@%$^`o1igq>-6eMBugt$dmd#bhM}@aBoRZ+QyC9BC=%@hq1$hsYMj~KlJ465;
z+<fsyntko)NujGREvc1b#Kw^*8{U{j-lGW#*+0CMy#Vv~ihM4L2h_#_?HcXzuz%@Q
z`tZIhQON2LZOEQudZOUveNaIoCupeK(i{?^+K_ri-3ItwUCf!20(1mdbNhbA6Pl&?
z*X(x6bD5)^ndAM_b&Hv;X0j-lB|Na2Q>Cl+7)gw}&h%;51i(}PBUln35+x&Kqmxn9
zOsuEN3QQFgEw>G8+c~uDO?qmA*^^Yu7Uq8+G6W3Qr-QXE#+JxF9`QLKV>93*2Hqox
zG`tx7ik^#rOg=#;RUbimxqXGfdr~n%+clgHhCc5D-#mc=LxiLWY0q=iXURda6yPcs
zy60@7gt*18@wRsWXYN{bADo*KN|AVbaJUO&5mqJe>+ZD|?W(n@1BSXR^WUj*6WWvP
zM3JV%5-p5wQ6I7atMuYyaIqxFVKoYllG$wO(F3cT<wh3b-(SJU3oE1#S&Jv3%wwQq
zpn`Wu7h;_v4eM&`3OMZT3A(Dy^lwAMJDLvS%NqFwLZShH&^<NL1jXPvrA+I;`B4Vm
zLm;*(nihx6;({M0xZhfHFZfv{#&;s(N>v2L<4W0ksp?Kd>3#))k8Y=+w5qOaglcIN
z&vEag0d7`bivHdC%PI@juqaJ7|7zL{Sg)2N{*WgI(NzhB@2d0e6tM3%#c}-Zof#Iw
z8f?EiCyJf`#2hse%2ml#zTq>H6(NgWrZ!9L2mA7~Ye%h4-o*Avk1con<R#`MOn|q#
zPklOlBKrk&-4$_2BPIMgjQ4Re821V|duTkrdpT%Joj6$}YkXgqzrtfavkHjOMs5=m
zMe}<tI^*aW)%~l7jNJ^d6sulNQFtM~H?gL>X46v#R8>rl&_skrUwcf@6NrjOkGP{3
z?Z2Oqo)6srD}BCA<~%XrJLd8*HW51>Ol$Z%-IS}BX8l@iyJ?=u2s%u^&0yfBxs$|s
zO{~dm$bx;;cfSky@648VF_A;uslDjD`JK&T3=Zl|JA=^=b=s(Aj|?N<VI5^g*i-0f
zyx^e?uz0~<?6IU;>X}{PC8I*$ew=vSY6yM%)O3n=$VnlR{kqn9n21>uOjS?`41?w8
z3vo)R5IqKIMb@1*dKYdQ+uhrZP5Sz6Uu?sMSI|nT{pTTiTN-pXdN0FL0S><$Lw=`?
z*)S|reW$g^Q6E1ms-c%W-HXXTWYvWDM(++05TAp8R26@yWVVz=pMg#+d`AvNypiz$
zMV=EvZ#gj_QZ#E&wuYYB9bmoD{ROq-sT_3!I!jrrhNsgpjh=vP$~6NM+oC<nQKUgR
zqVjyLdys1wQ2n5=jNp5Kfo6-g8HRy((Nz$(Qe>h%u=h}78LGd@DoQzig^mP`6p<YP
z0Q;(ElH;WW+0{DoYowT@n^tF{j^JvZ=O^?N3!4?lx8n&<)6=JV8nhox<bA?hz(wmV
zicFZ?!-EL}>*1zspbDr(nxNKM%K8V?9{5gbV%h$3*R_)lbD3^^hQFA<gBT!q00Q<w
zoGZO~R7$zGLne%Y`cZ)SBDy*0RXGR;SimWubS~UJg8=UTc7p(WRyOe8(M4IiUPO=G
zf~E+Ad+t*alIS|XXEHwct0U1s)|D$KX2YZXFC9B~Gp!BziPOwC8k%tXYDQ0(xl#ES
z1A(UK8@NGE2K9wsZTJ(#u}ww`xUs`Mi+fZF+UIGz&c|4m&lr0G@Z0%Zuyi&EAlqbM
zh`*ccsm3K1?5AiR#uIEeNy{$$PWivJv$0z=%EQ!(=`(uR#!e+qgXap?$q8Ow6$3+7
zFdmhA;!Daz#GPwZT=gV4BC<riJ9Ca7fyRN>QuISvYaUJ`y8n9zb|5Kr>5!2%Gn02e
z7m-jc_$-2xT0MeHXbv1efuySgrep)%AzFByWZ&)bZy)!vXGpShrAqqtP<nO9-$jOs
zZ;L3Zs}C$qw!6{LA%aS)w@1D1>AaX(dxn|#bd`xPpuy0(IrFmU_rGE<1PAr^bJ79>
zcGvStJ0E==56VOOS1o5<5AIT*$m`T+H9qLSq2|6h*hiZ8=V*d}uRE>U7C>?DgKq$T
z%@uaW0S)q393`28o_;k0+M+W_Nw@ZgK~|T6irsa&$VkI<t$Sp#LQmq$95W{xQ%LC0
zyfVg@C`Hj%%XNp0tpcX*RUY|Qr?sFDI^w;uyK?y83dpJbpOUCQ#ZlEotPw}71!@^<
zXN4xFW`boH3u^cP(vV`%6I8k_X3NN5adiU26uEv%S5B)#6P!;$v%e_H<YG`4;Kim=
zP)A8J6NMr}AX-;ff7+FJ*7x&Q*wpGZ*(m&6o<fV~V1i`H=qsj`9LD{i20HxHqD?;r
zLieFs-JkG{zJrGEbGper0TV!j3)Jgm!-Mr`^9uQ=w6in7Wuq`JCoOqAINNR?{UR<s
z$aPj#u7Y}CizxHpel8NDghz9V5;86L97xpPn1`w9W{Aj^Glu-NmRXlLo?1ZzUB(u5
z6G{f%slZ5^j>{ZQ6dyo|bF0h_pF=Zba@uQh+6@<gkqEp%m@{CveFu66$#-RFpZX%~
zX-+~qu4V~XKg7#AGgB<qx}z*G_eSAV)L7Oy_8^s@A|PWa6~uH6dIpuCrZ6gRWce!4
zX;FWoQwH}1^ij}He-!upJ;^<5H#<cSUH&eRG>J7=WUDm&TTw#1kIT=HQ|zSXYn~hz
zZ%v$n#62vf%|3r5<kN_Q31pHTh!sXR@2sH`Qey{5ge0t$80`1<WGVrU-w&5rCDrcu
z03HAC=)m4qn-8<m?ns!IBG^QjbDUdW%p~c8RhbB)SI|w5A|Jvgwe1!+*Tfx{IiYQe
zA~~>Oq+7z|KgY6Oo1KkVmbf_in=*sXyzP!t(yLaYYG!QmiVi<r|BL#0jU;b<`_yjc
z)w2MgcTZ9g6us1!r`6~f+#}7z;a?+~Xp`zyTPyC%^JPHZOY4x_40HhGd@E8T=I|8$
zF7mDE#9$Ea;v9F$&Xj1FV6l><zHCt<f|`eJl1o@uzCa>9n!rt>k)@Bq17OHi0j=TN
zKFY@?UKqj9I|0f~VxEHpsnoQQ%IPcDicbb?!qe4574A4$J8AHot8>orm#?(LZLp-;
zgrF0kqihOeC3J;4BgVSii4?A1z-sN6KA6;;kRe!r+m67&7s;c*4(}c>ESi)fY`;d8
z@~Du;nT(86pb)idLM<V;u9;Pk3u`YAw8*cuQ~FYMWOob4&Cqm(Sn*Y;qPT4E*Z%`}
zcu5$w2tS#CWs$~b?n0aKNbsrWn9^e=8=SijFF_krHx$!h{8QDw|8ywMpeY0JIy!^T
ztAK4G&H;Z-oViDSHEnjPQ&|?|rM&KOqJ5?}3RDjTv_TXG677@d>1B@(P&acP3E|Al
zDt~sI#7^sk>E^o!QA8>@h<|rCT}%T!WIZe>+2bRxFXj?OdG3E7ZdI9YNnoR7ah0)j
zd!j(fVS7#SR_qM1U9J~GgH(LNTwY70cTDyoR?b-VOa0ZE$#b&-Kb2X|Ce`f!AA&jQ
z38&|}nSG7A<$aZqSU9<)r?>1T#$$}x*59nkMbiNZG^c}=t<3lG!@c9T0x)64&I{T*
zjFizrZUjOJ?Q1KiQ|Bunq|RQ*JL%B5;VW@AQsg^kKIdulNc8Cem|yXjV5THbYD1v(
zz-L3ra-C-C#!1H)CpjWAyf7gTSYh*K`P$&nu0CC$a`=+Z=M(2#v&E_YPDW1FGW^|d
z;wSz%Fy1m)C_cgk@$mNj1i+x(U7raNdgF*vYoHh7s&=7_YH0vFb*1m_hloz=S+#6*
zc!SPF4ZmJaHIIr*&i~iQahESENY9C?tx$+>boqp11^%ST-jM|!pdtyUd1g-(GagAV
zBp8r#f2(mNXhDB#?L_DgmsnjOZ<3WqB~=Vwc`fein=3!#=TgZ6V7(s~*)Fzfp)xPc
zCC&eI{MN!nTty4`PTIpd`Dtc?A{|AAISDLr)?n)}XP*ZAnCe-o&)i9Hqq@aAjOa<q
zfx|blzi(_o<<wg%QBBL&kvcI^fisOYiM@5yo|1s(nGh3w9YKPSkv#-_ly-Rn!0?wt
zOa%QAp#+FN87u<?=1d<~Eny46{~&nmM^5;EsmphR%0p&efoH}ZpDtproK{(lG(lfH
z+Z^O+Dxu_a{)SzO*f?qJfQDA92v$hzpaB*G&}Ke~Nzx*aJ?J+R^1W`%8U2N#qSHp1
zm7!q4Qe>jX80_LR&Z**GOpQFPki;USq9Zb3SFh9B`I{{?X0)Ftd@^Ny8yE7ARkU2>
z>wLzukaWwwwl>Ob>Q-JikTx(VUZ0ClW(|BpdqlL(19u6i)ckV}STP%f{`^3iLugVG
zFcGfc7wIMgOLU7qW3C+Ec4^W}Y&@2{wgVYLA5?!EjAUWlX>H6${yS64_73Lcr!Nk-
z^ZPv00iptW2mvatKoNp@;#+<C=xkhz7gBCk4>?K${(-Tsb&}|0JBT-F=wEPmn(3Bz
z{*>ELS){=XrFYWLY>r(a->9L?&&mitE~>ou9?XNIu&O;3jd$G<m^!bJJf-gC4~c^<
z_2r8;(RcHT#f636e>lg6T=tO*sSXjH{t#c<@^ey!hrDN+{3SK9tBWWlO!nSjufL1=
z%=Z<}-&y8lxc&eYY9DaNs-g}5W01l1igU*_O)tPapW47wQk@w^9-XkTY-H6AXlXiO
zg!)#ZU#CG$aYg%O$AGnH#Mwoh(0a|Ja!<d6i|(_5oe_V&T`e-?FVn@k<V_h<w53H*
zko8}@2}hX;b3bahwm-D*pSoFE8<8~8R|xn}Q?l>#Gjq9DRzJ!@<TqcgH_Bh>d@mmC
zI0z^1Khs9+ZSLQYo(H?bvd{Mb@>)E@;glelK@PYAQOicRjie4Ed`@7c+E(vRjYJ?R
z6<%`H;bvVkJ+i~EAqJ4B`9HGJWbC-5mmXvgd;M_1`-(ct)Agu_V!~o1lkSlrvojQ#
zCfd3PA%soUU*o^%Q*IHNkv-$=C<pda6^Z@!xo-veIP=Uy-S4ZnZJ!kYNi3p0orz)*
zenA9v5mu~5$sAW?5wUIz`VTF(TQ6;!<}v1>(!95ml9l7)Z2i?bQraQa27jR!rD&XR
z;@O!oUWP+4x|A?E;tR)@RKwE?cUM!8m<e4D%}l*k7pCFgx*D4EWXrs1B_JR+1O!y%
z=*q6Tlf1vhG=0h7zl6F033M@svTPS$qE|8V*MsZlvJVPxRE^oYLSBr8#(AG!blY=`
zPY#gXx66)o4x{Krl|=gBr!-QU!o?uo?H?}aH0XE3D=#jta@wCwOfkIhvsV8`p$h>T
zM^W4H`vX>5y{&{#FOGF09MfqQqFvUjN?HxiO(MR?QaWa5a?p8zCMnI{qNFB|TxS=0
z#I4xv0vs_Pq}$7d2BcfzAxL77QDTu`s5&>V4L&Qd`qgh})7p0`#tZqSK{-CM$aBA8
zro}Fca9R($i(7s!?k}WL*=1#4n-z<%_Qc9$VL7%0#TQetqM9W5<;$DHw*LX92kr+q
z^L*ZPULE&N9u8@_&y;45w9?4xi)!mVX(~Y{%9Q?(w<3<7x%P&KitMckfxQU3bZ<(W
zPUlR3#;J^(h=H7P8XY4rE&#NNu2x4wwYvIpQ#Tp}AC^a>tBQ`Z?1`c5{G4+MZSxAe
zxiE-Ra7y%G(K6c=h|y}F-0a7M5@)s@1AUMqgsjck@emia(QELTuxLJAKLI6i38DZL
z{=Kt()YB``CogOA_0q=RT0++n*c;$Icy&gOAbwW3fr}^^{y|%C4yeQx#&WA?o+IwC
zK>0C~v7H4wgU#;|Dq3S~xmr_-*p<-a0Yw5K6@tX_&Rj?$V@ol2J;otIU(W&IUlCEY
zD>`clusnj7kA4+lz6_UDCDH=~gp2;>_Enu}nA{@LhjHVu@N5kQ5g>nCu^#|PyeJ8x
zHZg=~wp?I~E%5M{13D*owuiblctd2N(nfQ<-rIEd{YM!c0v|}KTb3*;@kp2maP(bf
z_wBp?nk(#4@P6iTUAPh)hUCnh;8@>Nw|2zrFPi2;#9us^w-J-BC235(lRl(5ZewGm
z;$t5qSuS3Db}Xb-J`(iE1c=1xy-7GrVsAX9be8#c_lNDY00t4>U6F5GdaK+)-QA@9
z+?SJT<~>fDXA-Xyd(9p<8)x5mJ;i!tB!lW-4fDUbShOBSz6<#}+PnWI-S3L8F}~;~
zp-tf<6?O5T1VYIqL!1z~TaJzaIfm`X^zM*?29h1A@T9&q|J-yIrc)XpsXFLsgFR4P
zdmt%OzDhn;z*PGm?OzURSN8BDn_xzMm;$!F9eL<=lf>qKjMGHf7F4zlfkr}gP4ByO
zEL)-a<QAq}OiyYnzBDt@ni1ML{nOV(S>R~p!~8YjX}hrn93&i{cDBIi2R4o9smiC;
zXkyG-LWVno+Y2^GeLkyHB~2gI!rkc9AYlhMzY$Q?06Z1R0_TJhgICQT1~`U~cVs@`
z7#&gg&Pg1&eah>ks(Im(@NyH}Nx9vpj2~7(^wr&v_9U?gVctS*kip)4gV(_AGK)@0
zxk0o7Mixe?%Q`cYsSR3|iofo<VK(ojY+F@*T!a03S^=VN&)?M)2hRorKTD8Ay=kC7
z@zK4<fSTR{S;hJH-S8UGlqYj6hOFt##ns{6{-#WucIK;eU#DT`dH~Vs+rPhtw3paq
zL!7O1&2-3?9CMPVZyG(q4NKmmgo_&@9xc`{PrjN2p5V6$Rr3U2B1+K0<X|aCYWA{n
zolXl^&-)2jQ1Xet*wK<Cnu$;A*v$)%m!8Ws0P2PiOD_j0v3BWugs^<|m1}IOIQuD3
zoWYD60nR*Fv3-;9M6Wz>k~>11$0jl$a&Ob}2N0?#KB_1Ls}zJ6S&4NF8cYlU05m03
zh<%j1iZY>i=W87(5@v2GyP%+CfqnM9*XaKru8V*0Ye)K}Dh-aE32<Eo1p9ExVI!1R
z?G_K3ny2Gkm8)iTqEjF1&Zyxp&%W$O`Wr9;<ie9d;5KX)tT=hQ`?w{Cw`}%uWv{e*
zNx`+^IvDwcNmP?n1YX^5Pm?C{dj6V5ykZb+AN{sq))1OWQ{9`}NN*_Nnr<s)wUt3n
z(8lga`<jyn7%_yf1JYucT%ey&F38tpB}Op?y!-i071D$Q`&}NnK2;NC&I9ZwK~4kS
zZAJ@qu|KP(FnhUMF=D0>5f$)<+EJPp`7j_DH*EP4sL}Qmhk1+TJ0E!%^`Pg^xvAD8
zBec7B-w_uqrzYy5q^*-MPT>CcSF#rrXZVF73>2keJpUW87_3=3mO}%zJLKIek>trR
zn4s!L6zbf&mM$jl3-#9TXF=PAQoc5W!S+|ATlsER+H2R=*7Qo0JQP>|m2!nH8YD;u
z66>pvU0p==FU{@-f3eAmYW}E>WkrnICI`~=U_k`4Ayo<gXmBuIIM=oWb3bMXEp%Jr
z7&AVefpjnrl@ulP=#tqhWb&zhbWrp??LH<t8!3yPAgz!uT{v5~E3%l^w#oQnQU|Wl
zjo4_d5-bwG=R$dFTxZ$`8!$GU<x*DjB7DLXr%l6D8ZkRCHguz776Pj7PnVe{$_`8g
zyq9Ln4vYsithQG2dtGQEEuI}10|3Jeb1L-<q-!Qftpw6|pfMo1`#j!jfFh%hK|H*O
zbiJ0~G~+3vh<W$+&YJ9bWN@DFB#6yVGIsB-g7T-<dQrJn!!ja7<j*UsEw|##d2uH@
ztmV4C8ftRxXR&I0jjaki41jff(RVh1sMA$Rv{ON|jqDEpd*vLF#xbT^0fd^^6|fjO
z*(@uK?<%<Ud^DKk9B=AWULSLBh~JKE-BgY!^e%HlJRPNXpH<1ZSY;l;Lqj#H;CAf?
zcGTVJa*+0Ss^9B9>CtHhG^e*aff%>LaDypwwLDjB_{4i_e$r+FPOmqvcK!j`BkT>#
zhXpd<WdvyMt}C~htuho21JVd0msk;^GKk@Ut->`K%@Jz-LhhsrL{*F!A&E~lY>!54
zUq%O*F4LwOgZ$w^IG90O%IY&{St}U>df~oUUM6=KgZ`fvlb1<jo=7KIfz#ZTq$Q<@
z1H-|q!nOj_<HJQ4B2=~=Iw$1Ab&Fld35Kj34>Q>5`D#Y6DuB;>WkJ1F<Ue`}8NW1%
zpvr#TB|`Y^unc_UK<N1f@GL9;Eyo*Wg88@C+d1sMjTG;N%b4g3NL1Rz3cuq{Ry?>-
zOGj^CdWczKw;M0@Yybs7Q!WRuYRuZ2e!D*9uRopyoeIpYZAU6hNnY0tR21{4nCkVP
zRjYxy7|=h|Sb&ZMR{1t|*L`gWqZQC8FC-WsOoA3;E|4-0tk|1@B+pYvu$md8b=C)0
zabzv;xgjv&7g!faqpQBC3tTiNVc;aCe!zwdrRXe^Iqq#oJLo!J7@cDcohjJUilXl2
zQ6sV|y->C10s@`d10d&i!#9~T7qxUWe{Wo&bh|bIpxKq=(|i(~rEt8%ML9c-8}^rv
z;=}}fC}Lum%IZr9evZp()h3|F$sgjxU_}=n63(;k;g-N!Z1I`^tJEySYf%h-Bg><p
zxZ8x$*o2Wn7WY~MWP|l#AkB;Ghtkk@WZrHkP@m6Rd3d%h2Xv1<t#Y@}9SSTX*Go5g
zRjk1*J7T<^Q?}`0Ju)D{#<jq-f8w35tix+7B$@^oc-H>bv(aD3Ya+bCdp@x^CjZ6K
z-?8a23=9Sj^(8S)A<V3*R}ACTH<8>`lGR*ER3(_WjfA}AqM7iaT<k0cj04aIxL%Xx
z6xQWdRS2+-C*3K7m%o8>sIT+Z@;~DZ>{~^BHQmJfJKuEH=b3$kM2sm1%l%%srEhE9
z*)kVFYD9rwT{OkH7Bc2a!2QH19nUpmvC=;HN(!C)nbACP4KpBdSzs0;vEUhjz$ERC
zr9lH30~wD6=I@B5&q~vTKLTI{B6`i{J#yH>Eh;py@+bzY>#g)gx#hXb-|YTK(wfKw
zmS{aeovYgJokAWGxXV}hm<oDG4WsT*CRuXQPd60CLgR5?;kEO;Z4*!;tYs|n75f`Z
z_Ak33*<G{q2SAS0A~_1!@mfifxK^-c=3#7VLGA;|&~X<tHw~!>VFSoJj(@w#lQl2(
zcBER_Y~%*w_upF~cMgH?eE3OYWG~Oi#UL(7c`?QvDm$WbGv(Uo6qXR`yKeyCcB>D*
zYLDj&6c-%AG@LET!-8!JP--Bb@p{3?xi48ElZV0+V_g^Q1rf7u(Cr}ehrZoHQGO7h
z(q3)_>-n^M%jwcg7zWOXSkzAA4^4UjKM;Hj@oFQ-=Gm8&*2@D-4Dc{fqQa52j<KLc
zBjE{7W4}(UTgO%T*|W1O0i6zEc`e!4$D|V1imb<Fd(@}|oW^GZXs8-bmB3(veCd}z
zH7IlJcY{v;oqfU2SZ2Nmsw3X7r%h+OI47Rbw51yw=x$}uFb8{I>3W#TFu_@|WMgXJ
z^Kj|Hg(TQZ%TPq`0TgEY&uT!s(EQdqE$+-4H!sF?pJ!)8{lZ;GoFwv-a1#d}D35X5
zNYT0PyEJeENuVN$s3say`U-vrs^$i-OtWMpW{$Gw%Du+&?67-%&?TK;CSfSj>B?FB
zFmqp7rd!;trGC26dZ0x(T+D;<)?)?t*M-NcP(8hIO4q~(077TjL%Uo_s%g}S<DMuf
z`=GwI@_Z;q`zu3VahMY%Wc+@P4|y31KIk^ZTe15VGh9|P`WrEIwx&5a@8w|p0%ta)
z&+2`R*45*6f)ce20uRURBuQpd`U*4|brAB_b$|*(eGJ^QVKuZ3QAv06RTA&;Gw(j+
z=iitX6Q!PffRDIx&(fll3UD6VdYUgZ8PTaf8MdHnO&L!nmMXwWj_7&48prq%;(748
z^uy3%!=3QZ|GkG)Ugf*`M_~FhxffK|tlUOdrL--#{vR%j_AM@Q#~m#lEoZA6C4>JT
zdW{BO4E4S`_Sioko>%mPxJrBW>z=<9#Psc?a&3*%t(J)IdheY%*z_IxGTS=j-^<i_
z_Ib$<B<Tx!5B10^ROr^{P38Z&$o-d(j{#J*vemz%7S;H+1r+5t5=6Rrm{U7A$?JYo
zp`1M?nRhYAmv0HiZ+{KgRx42~@K=GAEA20bqYfXnFC8~3xaf1V*gnJ=ySG{^=5ksE
z({611c3?d}yWFw<kssWM^tUwmZ0++O;F+w0;m6~K>U#Q#G4SmRdykM*Yqln=Un(54
zNlW!k{IRpi$fB&Yu5@ia4?s{<tdVV2y}HT$*Bs_8umJ@;$SWS)y17wtHp$JgEd4hr
zn4cGbTFliPMeqPougYX@X^ewvt9^ois--bEly<`p%m6@Fds~Jr?nw+YJoQ&`DOY^j
zKzFo(;~3uy3#3|~^u$oJ2V)bE=To0Pv*%9os#a)l#z1BNYlj=Htym%@Y#T|ekDB{7
z<#Z+Ik$ClSMK9FZEL}Q(wW4@eD<4Qb+aFp;Hhzv+%Xo;KSR>is`Wn_rM^dSe9HLoa
z%k}*;*8V9$|97sv@4+NGfo(YRta3BBYSy9vR%7$hlR}z$mZL#KN{BiWmy<`TE_p!}
z7jKu(8=cGxzi3FK$`kwWpGx9rts&W4?%v|bJ=*Qgr5px5*bJKXQ}9$pxIgFg&u#L^
z8Y(L7(^JbBm?6Oj)$=r>y~x>ek%Zjg&n86Wbf3$(;69=qTyJu0u^&Cb$l#$$hxOSZ
zAERS{RgWPX-?!OXq(%<by|{XBl$y)}m(|jN*gK?2;Pn7o$uC)Lx~5W=*A@E;1@Fh$
zss4t6D44Ttgp&s2Qusfvef%hDV1k5oW9a=^;m609n{Z?H35T(i9~zdB>#?&ufK5++
zg4iF{vdwopMbdIj()88bC#c4L#?nhrJkoTbRv>7MelQS%_2}_1N-2Muk@1NvrrmV7
z;7Uc(Ue=jc!<`x&k@}U#<XP|T^C@}X0cjGq161MyGWvp$L&a^9Bkcl)2OV2}7i7p!
ziAo>Uj(hlL-C2}1dbj{o?QC=u@M&%&>pTDF<L3;X*HvRxk2Im@#v^)5%CWF!arf&5
zek=wICqF*!JzJJ4BRw5}gTO5uLAmIt!s@MX%u{@NhKNS~uw;bS<qea|H*;8UitVy+
z9r>+5=E{EEGO<oI=#(B>Ke7@+tVsO<BNr@Zpa{%{cAME;%KhNy#EcLFSh}kYz;7&E
z(FdT624i<v7#>KUxA|r(IWuVRWdAa3a!ymHaac$OfFK~zNxDBM;!xVq@^e!Q99mF0
zMJ_-{@Q7xPQRcm<;^r;Ar7=pL9DBcWs%_5^>vlP+i<77oFQf~zCXCHBQx?t4Z7?x+
zSI&*2MiW>miGA)=fzho1I!iF2uo&fjfgJTc4yRtJOH?oNDU6p5Q~b4&n4YzbWsKfP
ztZGJVTWsm~vg(wIP|O8a)il8lRRJ%5grVV%2=(u%a@z&xq!7(4OFdkUAqHC4q#ug<
zdUyB-zhZRAvg3U7vle{HAwahCh1zjt2om{mFL($+DHTdFrEu~LU<lqqcyet;=LYyC
z?^j3diiVg4fT-}j9#6&<ejn{b5_|*%|HZpNMCnER0CRlz&*A=cEOBY0bZxo72Qi%;
z!nn)qN16jYb)=WSj17M<VZAxJa{Wc;R=UX}Abqk$=XtUqlh)DOl4i(Vm5;Kiskqon
z6zLu3B#+ijx^oswAwv4(CdT*c|Jwv$X;F_;mz>5x_`i}H9VO3`yE3G4ivaVH+Ac5T
zK<co$?N-A&RrnOxD^nerD=mV7WYH;#niP3KcMHR$rHcTQ01Ud9t%lTc(pokY;HL9O
z{PE>`4N1Q}MzJVCKoFx<8+@*uQ(_HFJf&A~pARP~A7NcRy1;N={pOW4&lShSV26Ks
z^#YN1h&Zw7v%_7;rRQWV$`(kpQnmaZ_N;3cal(ys<tBxQbVWoHzlM4t)R~}mE<kG~
z>H$gKJ5u?`Pva5=*1@CJZ#la}bt>lZUk-qbPg;d@3qdGP1xGke8pKSi5e5DOxbk}Z
zP~CG!_K44VOb{-o5yKiyqrnKs$!<jbQG1N2@<8~>`Mu&@j8?G{;%OFOA+Ta_la9Eh
z&wU$>{c&t)Q=iJ%q1}G0;xfelAG`1?l|(Md5d%|%pXMa`qY&*(*+m3ssql)QA^}V`
zI@vC;77-!T0|6p=DYEJ|zX+gVbR;5O?a1eACv6u97I=%f{1T*&G(l4|sUQUbZE+wh
zhI<C&7DQ1geGa0mSV&t>l0}Q5Z3}JNO%Fkghu0nHAA={%jhf?(VV5ryUorYKoe6
zkYb}cPKXvC|AMo<d1t^`k8C#MDx{_U*L#4_k5xf(1n%W!i~tf+pEm%(qqiQu=iuwx
z0pXqV$1XwYWqo&ISInM60o$-0Bs{B99Z~SRbyI3(fS*vHEFa1v2qV*&RJWNu+B`I(
z3Ou(g@`#V3E@Q2tKzBGW>bk8xM0P`ME{@(i_>r39_94hQu9HHer=HMkm<)|Y9Ryov
zkBwk+0&*jF6fvtwR2bkd@1JTs_9F4Rw^p$bEHth{8FmB%L=*@dY>8OG^;K~OAjc>>
zvr6Q%j9*d>0*a@?8G2N8&KF()3&)DJ6~Pgk&ugRp-k>mYND^?MgSJvP<>?s$iCvhV
z2Rujj?=$5$5URl@0Ib-43GS$ir4$-P-qT(-9g;|U{-<430~|;4dpgSNuILbzd?ZI)
zW2>*xd<z}Zfr>Axqq9-=k`n&RVj6At30tlQLrm7!$cT$7NyZo}0N-ldZz8VXW7EWA
z`||cJ2~G4QD|a&u@&f^PZx-lkwM4#m`AT}#1cU{KbHK%0GuY#4kCQXhd27<7v5vEp
z+bDIp^xM^X02djBp^g)a-L<*4Ss6EZ1{yJFMz%Z2y1i4E>#x4&m?}kDbIi^MZTB&k
zE=A@#N2faDXw~+Jbh2H9ZAgsK!175*oW7-Dt1}r%!4wX=Y^^7{B3>Hr#vHQa?lL8&
zRkLYEZj=#b?m3kH!)bfjA$|L=xjogkBXFW21h6tvz#bM9GX%8~Ae|i@4cGyRg3rks
ziF1>{ugw2!987Nvu329W;YOz$raMK6f{cw`6#EMnD-S@x_@B7TC2cy>sa63EKAjFD
zDd&{V<7~<T*dS`k#${O(l1EnwLovGe>Tzn>zq#GWZtA`XqoiV8edMIV94jiQ{jwwb
zQ=(TK7->hOL%W<Syshpdq<*;F!*{rA``<{*4CGo_BDf~PRr*<Trb<v}XR4;zE!T>_
z>ad31x&%NwCu0JS5UGeO+aU*UG7Uxwm>yuQ>*p`^Cj@PJXyO5S9)yQad?M#I9FdLu
zQm1k?xj*73#=7pgtk}GE%XTTT-z)n7-1HM!(RPq7O(SCye>ib(DA5n$uB(~vo6*<R
zRZUpIA(LF&H)L;GgqXJX1&y8{tq$;3Y#{tK{-t|-jZ-TYF5Ms9E8M%3U`K>H8vy`z
z=b43uVUL#E=gmaIovq?!u0N1Pgn5npL&|5krYQGKTJo*BYYP*0`$NvbAh8^LKU%r^
zQ-ro;6DmP~TfJoWP#x~~lV?x=uAJ3-|8vjvqe_~6iC`W+XO;Zap>}GY*|&91ILO>e
z)x<|^3H<t97T>x)s&G7AiOpbcQx9<Yk5+{Gy!Y+6KkqH%@h8QWjdyUgt&EO0$*LYr
zAX-ewq^v$ZFY}GR1%SK-qmJ=0(w4ELzc|)QYzc@tir&r$JmIMjqUJHbmUSt~!Fo6L
z<Z|~ix0wZ98MG>&8#>FlTgHwIq3$E|1_7V8u$oP+)h_ceSGqVWap0>v0`gP4s+P}6
zn~CU8zrB>hGM$gzrr}&Hs<S2?{h;35vDmX&%A{KnR%d_TQoZsPrCUY1rlIX5LtjNW
zM$Q$+MD`iM?@_WSHzx{x+m$tc>f^xIKse0KEeUlP7Z!`t+kuzE2VY7z2mUZi^FM^`
z18OZ&VEBU$idiqceREtE;L)!R&jWGQ)~Q#!vRM<hC7r(KqGeqr4&Ni5Uco6_Bd{f%
zFvBVIRxWv%W?$H^oy}=k>tEC_kBA`wCPx>LP6_ndR-u{Hw^eUjK#u98%9T@wK3V9j
zxyiArL!K+grq?s^@$tT4e{)OREHj;LeZAI}o~nm}VIBWc7VM%JAlXxdI!n_rRz_|}
zc8P~Z^=`FenHD|xQSuI#<i>N!--(>=VYH%Vs$R1C(M;W~wJ(<of#()WEpx<3QUyq@
zNc$q`q0(Y9hYf#QKiU1j%dR>tMR9_WBD91WR%f><Raq>Rd9+1l19|T@>02r9yVKZD
zw6l>a_L-gp?P*0A5CII~uz-@)Uv!pQeS(TtKxmD5+#P0<koA4e9ksEylK;@RktT<X
z@1!b>oIf*h!YD=l+o#;>%HR8F&-O@SuTQGa_rN}yO_x0t4k)@va`=mGr&N)bUZRJz
z+3s$Dls&;-wbv)2Nklr@Hag9=?Um@crpV?&xuYj#zAnxS>_37otx@);_7}p?5dY{|
zM*u$Y;L9;jiy*k27W@9rV24l>^=%W(3pGqqk!Zx!C<VUfA#U(rwW8!LTtZNE9M$f+
zqm%)aSRj$|o02A=0ff1=(_c*2P?TD`65!MtkDf=j_LHkUd{^Y`_ansI6sY#pQ6W!i
zX0>bROCOCDfp4v#I)FYEfG(Mo%k|4iM=J`=Wo8w!*K*bhauM>-A5JLRB*@Ynf^rO%
zwW*TNH{Lqf+TWfVJlxO=+rZ$F?aT++)7FvN4!tEz%t&Be(4noA^FEO(VQrf}2lOy@
zO21GP+f^lwagt^pCvptrjH~CZpVsxXQlLWg_E<bDkdv`^gqKj70s7{=*eIxF)6j0e
z(qk!dFA^`!SD15E!pzYDZ8ZcPy8k*KxpdM;3VDVxNcQwf48<=hx6UkDyAwHl)i!G4
zIw_wVP0}fH*G>}J7tq;8BNe`7-i7NE3TzY~P|qwfUIb;AhgAPzivXB3<)Ym5k}MF)
zX4K?Gx2+}6k^ew40%#brVd;g=K=n>6FyA^&SVCGSBQRE2VVLFi?gdTM4gx{<NW!DU
z%pB2XvmhyWL{nXZ9BuhJrb!2h{smcHyUpjGjFwQR@(=JZH+Z}Sv>E)r8fiSL;9lC_
zLNB<DI&bXGFSrK4L@#)w;l)H785)2>G{)Z?LeYOa>;cE2RB}F^kA4VRH_tTvXoXsg
z{g|=nja!kP^-7&CPlv&So2A318svexGw>0Z(RTLGDUJ|(hgWzW_ISq6bKST*)v%Y<
z4rmhidz2$_sK~thZG+IAbGJV9Z&@%j<Aw8%P0UMhL@HY0bpI0>qswe;Yp$h;OniZ9
z`a@L%sGmUhnl!J+B7VybWeXGgeZ#;}$RjvpV%{(tR+7cqjCP}zN++U@Ce>s--)_Dm
zV%7r61P|`0G4P@+FL*xpCBjO}o6n*Rks!T4?8|3zDX6ZnQk*Mv$AoP^b-Fh%U4|a=
zEy=|_W5xl_n1~k`Qdb@>LVZvm<J^m)p|c?X@aT`sm7K@eY?zAz=v;-ut+S!V!VE^@
zSghI3<3UJt@4Z){V`d&Th4t#mfA^`7P;ngj1BU6dwJ@d7F#V>X9?d#Gn3E0<>kAcc
zewCKOQLG<%x}Fr~ZUWh?bz@Rx0*A+Kwchgg%Unfal3!tOi(yG09|B+62dlwmE>#ZT
zJI&xnU7ADxxT%qhNf4Z@n8%Bs=O{~(Xa|oZ^r_sewzqVZF{a&F^{B)58m#)E)m*3l
zrLgnMr}oT!!1YOjGcHG>Qx`Ft2vqp&-|G!1n7Jp#kEC^EvWowOw}KEV2@941n1b7X
zyyo*?i?z)fXk6!h#FH0*UOj%egntS!v31|O^1F%!x%^YAGzny)#h4Zcx#Tra2Nva*
zf~+K;h1U}HwAOX@AD0CldqPSc1qyrcvh>Ij<zqIcxM6Mt@uU`5^Ng32e3-Y}h=d#Q
zI_RAQj%NQ_NB`PB(>zQUlAM0q<FcVtqKYIH0-dhR(iuW9Bu`{?X^PN4vx<8FHKMLl
zN$qke-S)<fZ|`gAkbtkH0isezz1r0c_otrB@-ecE(Ic0)(|<#%1I+KV%eCGMHr&h5
z!UQ?xrj2KjJ*cWMB|uu?a9_wTl5p@<jPQ0pI0ZyeO`NtQZK!nu^7_QFLURi_?TF4u
zf$_w1?j8<WZno6t@3gw>q?0Egp*6+W&F09!Q4fCP&XP#K@FyHe!q$3xg!~^yO;Rtc
z%mUw!hFM=xG5KXk13pK-usVO?F9B`JJ#3v_D63yjlKq@Vcn6D@U{}|f;dok>+MVcJ
z|Dfw09q`Wj^g-C)R69D2w7R^<*|*mb!3XW}^9B;@m~h9F)mR$KKbATG_B{qw?Fx4t
zXOTZw1ghDh&7zV6NksC?$h;GzX1FH^q=yQF4`~ywk#ry7^5UXbJUIV%za$M*I*2AR
zy4;As_(=$<f+_=Z_0(uYcGJ9(P_@zeA8UD{S(i6MpT}-!!XX3@v-givBl*zB-ONmh
zw9+a3wk`D2W5qY%e~U`^(hHxo&21JK*9&ZlrrrZR{i1U3HMqQe79H<Tr;cMvYdWwI
z3BiQVS0B&1>WNH<(Blnc$t1I4a;AO9ROjz^%Gw#)6*aS-ZDHZR6dT4yNu4s6e5O9c
zc&-gu)>qb~hhSK6gF@mi8j7bjI=nLCl(LmYu`wmDO8_rLSg-ler-_Um7uGXMhwZ0{
zV@5|VDr%UBp)3jXUVioq^)>QS-m<-UtoEm<?qZabTXDeHK1{ky;;O-7o>4&vvY(f5
zwTZBgaS_0(;z`TPFU0vINOibVVbpir+Aj%h6ldulVk3VZ?*yeNysGzLw%wn#4xy(}
zv~Aw*uK~S#e|}p}H=0vQE1R;V>qSHUHFFA*X_h><x?Z6$|GUM(XI+6(?Cjr8ZE!@c
zy`?Kf0V{g+NWKTQf}lG-Z{J2%?)Fc?^2iFt@h~Wg?yUtb!xC|TXaWLxH*zf{8?}D6
z&Yf!Nx;swYIx(U~`RFq{4~{1ZUEtEqj@@sd`U((W@0OFEmJt#+G<G?Ca@Ojk#0%F~
z=C=(}oKQ@Kys4Bf0G`?$1WV~5H@ispBbT__$^Vk>UQL=qIUYfY`!!ij6)cEBcCWkQ
zYU?D77T~uOVE{jNc~e3gBhfX2`+(yC=g0iO%s$;PAnBU)O&sET^^LO2lyv@fbfmF#
zeEyRVPkoV2nI>}}>`6Ckw*UG^X>Yj;9K#RbLrLj`b~MYvfPOUl1ZSFiXCb|&C>Ut3
zJ^M}UuGuSP=1L2H6^N6U&4o2ZqQ}rFBKBtlL-XxY<7`Wsp9J3Pc)+b~{cfx@(di2D
zPU{9HmQ$HFF9GV6B(N1-z2rCDnfkA+mM#F6qaXcFa3B&yws$D=r9W!+Y_#5b?BC^V
zam8^BAX0s2>9kt3cLv$Hc-1lLermp*@%0i}yqj=T;Ab^SoQEO;!FMhGM`+)Jq+4}p
zo;z)ryiP8bh#<m^uxnwAjWBP}@b??I@zh>lj~(pUm4a}7ELcRro~8$%{yQYzN;*I?
zj*6WKJWqQHq19;qUE|BJ9A1=~Zi;mUT7Y~ES<Y6kLmT(OV8P99;$KMRLwyc<QkkH^
zG^E-UTz}+Qfi`7O;PTaY8fGguFRalOri04mnGUn}Q|>*<;8S*N)-cA7<p=8eT&nsl
zx<8h3^Oz=5EZ&Mo4$;fQ!I_i&1sedIbV%{+gH7SYp@?M@kEt+S@Xc|pDp*MJm|gsL
zp92mNs9ay5*X1X_K%jk-1Je_(Y9%jW>W*6U=VF?Y;muT$rs0bKoRZ5b1N4###N>1r
zb#21_RQKOQCE7c_$%^EL+Dan=Zhi%(vYe6NUn{ZuqzrC%)Ef_~twwZw?+1XFbi9Oi
zdm26x_E#kDf_0z1z{`wsnJ-<yn21p1L=R-7Tn2$d1OK~X_0nBIfq<xRy0gaYe1*9h
z#a|D;8Ig<ZO6Bi(_ZuLBf2#&}&4PC9nUv5mN;3UA272*lZVI-t6c;kNSFr*3&}++(
zsE6XrZi9b)s+VWu^BkX3b~gd5gn)x_O(5NG!SAWQS3I1>ZbvP%N~QOw3Savum>f(e
z&MfVWIoZoi*%YXhQN@Em_iyS<al-OlS3B+m!-GgwSF_hm^eL>wNK@1Kjz4bV=b3-`
zfheq>fj)aZ09Yl92&kWmMT;w2gRVN}zeF@6YtUk-f4-`}Yb;o$s{lk5LpmBJRYRo8
zQ!b(+_<BfKXwzMyAb>m^FgcD}-ev$hEx<9%IHps51;qLR0%u6kSOKFYAYcTzu&SP3
zv2B}}c}6bAr2rB)MWf76HVG>3b5K#386e(Bf=PRu0^U59;c)txjO^A+<M%h6j-EHc
z%g)Ly;C39nP5tKAO#sbhU#D50aVF<YAVkp$?qiGDIyM&2E%lrca|c|HJ$m+t*d|G{
zcZ~Hp(1vmTwhIP{Y1|~Uenzd`c4zy8$CG$=!hQ7SlXT}iL|AuY^f=`7@{@F}GO!s~
zw(zGSL<(hB^Rda_0)U@D0S3OVe$HzViFc)~a+37k6f6U>Isl+H(lD{Lb8iK5)WkCs
z09;$EN+CI9fX3`Kxn;Q))b{cwo@22S50R-dEI;}|mBQl2*r#=dzHQ0HW-KUi)fSyV
z)1G|<JNs3`JjB1F-4lfYZgrP(Kv7cj8x>ep(>u-b<}xjojf~UR`T1VS`dQ*&U8<q>
z>yabqMtCUa8E{1or>0BA>oD*&)b?)QT(LZY&Gn(W?#o%Z+^C_FHAMy|tf^0TjO%yt
zL%p%0=gp~&7LlX>z2ex~VWLI$HI_$D(Y)o(Bp@=bG)N(mwp425ORFy30M>kguHli#
z{Kv1@w|NYkqJQ6ct^QeW+pRri41pKQ9m|YejylF*03e``2`^4+FR;XzhaVVx8xJ)6
zTmJwt@!eediRmWqd(5gXe;DwY`Ug4VX1ldonWss?gxHo>hblfofo+qdw&PYVy{nc^
zYu<2`wc^G&4(Eki=dvea-9spJ-C<Oqqr0aAb=|<q)g4MA=O#x$50VaE^1xasRh*qv
zS#5tE2>{L}MPp=d@{HrPMlu^Rwzf}^o+ab}HX7-vdsLretJQ6{i-flzIWny6kIaa|
z5V<t@`xPI{n6b6f71igd?N8A%a3lA9xo$DD5y2*9{CbRH_jcK`7xZq@l=((at{6ew
zqA(3Kv8(C8Jw$4ca^^vLi=ww1WqNszlrNu?Kj7qyQGSxD#_P!+h9A<JDf+p3PB)z-
zVC4T$^-j^5MO)KwoQ`eVwr$(CZ9JV$$F`l0ZQHhO+x&CRdB2N)?7Kb2x?5{h)vP&d
zCRmGgK@1OzX3K_7JmoAe9|V-WmCqnO@kEa2sKdf~0G|SF--X)AiM}tFsO$JSnNh6?
zFcbjOqoXhZv<x)wgBshmSJz+~j$I!4c1)*@Ab-xJO<bxe$(l0`mZ$ON@y7nf!43lO
zhD|}G`y!CxEJfByc#zMvAX*zTJ;Xg0*N_==^)_)Ph@A&aDXL?zOorXZXb3T&gqd{R
z!AH0bse0uRpJAOtyuTl{kvG9`k^IG*OVdBH&O(@!KQA;0>4$Xs5GWq9t;Meq{&P|1
zsY@)KmMmjRtiw#Tb~gi6ub|#I0Wby_cqYL`vABrKsrVO?$!jWSQm0p}ZKw?fa7LVJ
z*|IHRL~aenY*u}0ggc~SaoSFr#elb@I!yR1TvXVLN<YYJ7wdW2ag5r5;=6u&k>gC0
zM7&}?`I+^7g<}Jd<+7O)d|l`^4GiGE%CABpGK99Kj-qO*@k}x!m3>WvL3{yZEyCRj
zHm|@juVYVvZ`1@4AwwJEDXpDG;Vn=1L%Vp)gfXy+tn<w1Ag;mRV+u$!9?Uyg-ksYb
z(zaQ!E9W4I6y^1B$<+w)Bq9EogwcP;Ou8&K@NP*JUR0w`i9(7YgVEk97}oO|tl~<h
zwcG9AjX&S)9`p>k?#ZaWwpIbG@kbrJ8gAti>;jRbGy-CiE<_@NETk7xqx^ej8RWMh
zmr!+LH9-_=1F;<3wJg+@;<1>2FzaqL&2#7}3BI3iLGM6ciC%WG;n!_LQ%2wphOjxT
zX%0&8h<O~_#}NTXh`NYh1P46<&W83U2-I1(uUt<wrhNg-y@<g7oKgXCq|Col@<O4z
zVA|smv>bD35xrzmpa}#B3E&eE^7&tc6QLx2MsV&WnLA@(A@~aJB~@1&f>kpV1Lf@x
zCDxcqK4|?Te1f1D&i1~AKI$XCi)kX8E52OwDv4l&yLZd!5o7rF0*ZAI1H(wyV)~*Q
zC+#jV@*yGk8ni6yZI=M$W1wraq72jQ<rHk?NuJ`emzB-V73JI#Y86`D21bC5$EcdU
z^C17Op$V-PRL_J$&#u4N47Ah@Arv4R%lx!1>q=xo5XD$(AL}P9ESD@Oj0zn)hX05h
z{r*n-3zyD&qgcgBRX-+&De~C)?_i0PH0?F4hZ@ZTQSNEk;EDruV$sY^%e6QOn7|?-
z(L;z<ey96H7hJ+D{tkb-t}n-B4oq;GM`Bj0U-Q6DqEfD0;zQeyn^o~<8q^}dRDKy=
zGdSlEi^m0#;!jtMqm)d<6V*Bu5z8l3`UtDDl@d!Ql>1|)ERs#A;C>_eP(>^jOZ<iE
z^sXI8Hi8UDzUcyJnEqlO$RRSB1rNg&ihgl1{inxHtt2mh{^^YuwkLsV;rAOw%ys^M
z2a({KC6Nk41;o017eZLMz|vbTy)iqCZ`uOM?L$;e(^ciMqtqeU<@W5Vp+*TTc7I0<
zbJg3zhfC(kKNrrP4~gXQvIG4OAb;<FrvQ@|0xV@S7iGSho<-9Tn^7}?B}_j=ho?pT
z6VWa+I~4b`DC=A8HS5gdAi0w<$jR!8bQCVMf|0BO&7#|_Kr1KW2(}|jt&)?VUdydN
zFD010ImTla(rFTm7j%N+NMDJ=6hOq9!d+GyEdOb~RBggLw9Q$HR(TgH?)BT_sV7Mq
z0c;k4aW7wPt7De7*$D(<4C8sMTsih;IzN>@69U~mOui2&IvwL1kF;2ldbr7X2}~2R
zS?2Tkv=z=b=XRVreR+;r{Fb4Il`h@8z2@xpAm%-i{NUQ^9eW1%GnrNe1d0Zeh)~P;
zlVAVc9d?($1y)F}$*6h*Z#w(xY24So0$@+2ieN6YJD+t~PXxw4f<8u$Gy0-{q0jbu
z70K(wVA}b95^~lh75@C8-ast#-2r<WTh|3bW4`XLRP-t=C>#H%%V^)_IQ7f{0q`|8
z*{Mg5gu4#Rje?RL_2mwD$5$RckR^;<(A5&Bi=jsONkc8_qlzq>UsN(%)&Cmu1}MvZ
z{loYod(yZq+^ETnCa0)n$C^0+8@OX?t-Ed6QAG9|GwROi)=_9fxe|>ESHO1ri45X3
zsCg*ZGey2lG^)#v49VYKpCfgNoUzlGdfS0EiKl3<r(gB=&N%4c44*KKvb%1P`d~$J
z^wkt6NZc=xjUZ^Q<Wbr@R!Hv#0XTIkl@bk%3SFq&a;BjdhmWDKPt8tAIO*(C$V1)x
zTc6bX;D6pg%)JZGJQXppO<9L+<sqRQzxGOS`*Tkf(<7Nlt0lZ;n}=dj-@2ry7Or~n
zgZ>si&-e|8DzEdCPmpgGDU}X!1zEO?h*qd*ZXQJ-%@eXHV7{Y^@@UrF1pvCoBg6}-
z#26k%nu=iZoMx>1-n#`7KlkcmWz;6{ZSR5~Yu7lV_{@eg`sgE1kYe$omb2vv&DY|N
z3(EZqlNCKECtQI%>)Ka?VjNp5WELSm19m(Hc8oJ+T4rR(TT{*EZx(w}ZFbp9Rk&5a
zBrkY04&_G`gCYLcg}#s-7hr8by%u3ijZg*<9zAlnE>t^b3%Fe_RCg*Z((zH?tP|z4
z88YEm8MfbwKBmkPt~mhT?ftSZ2R-tAC@(IsM-cwEj)r5ipb_IX$`2PEDsG?=(IbqQ
z*Q@vdpS{Fi%hbHRCU8{S&K265Ff9j^u6(q(pVtJg9I=~rB33Nj0m%F1fYF$U1=q*`
zBzq~^k5IZMf9$@#!VVP<w3*jno7^}Ui9ObG63@3|NTC#lGP3%L8-|Zm3;niW=ShZ;
zI9n?^)b=C{7a@PqU?%$9En?stcMGE$iG~v!FuZ}(+!ZRris$Ayowm9ju*ftj^MbYb
zJa_9S@NJR9bGFqo2XJtu5qut^l2cn_88gD2XC<^d%+1Utm5TYD?+|X4CboQ(IC74{
z_JXt%v&hP#?PONBG+nAPNaMA+ipEv*BAC{NVkRvsqokHE;6lnatI<b86EQ5hTbyl6
zM;Vu*C!W(}pBw$yYe$$s8YX7p(#e}#8Py_;4et;P>@;Xq30N!zDapJ`{;=+`h_TgG
z)aO*F-({1!E$kJj3g!qjDD2e(ioCSnXI;%Q($ybzlHd{wZK(bU+J5GV^KZxAj2b(5
z(zcEvWljv<hBz>vNm52-J(~0(+C<T%g_D8On4deQ0d4KF!N38$_)JvLZ-%F$AKW+Y
zBV&!1X`4Hl0lNM8;z91Xl-s@6Vv*}qtd3mh={=8|G{;Pe=XLxQsQ^?8<O7AYwh3y?
z1w~W>%GuYo`c#K<%B9q3=#FF;(>1~DfYh;~QeJBzlWoR5>9<jt55<Ow>~2giWxe*F
z+l0-y2%q4#m#2ze3R}va2a&bGZ16wWbSf<c2RyaY8XA3f>sXX;O1CdT<kpk5qy7e0
zi&@Yvx^aimC}lhQL`U6CwLEIqS}F>jnEoq6o9>+-Y~1Q(D*t<fQ5R@olHmaECb%g}
zPO+tM{!n7V7jq!jB}s~(2;o}ih_qwb>EyJ_s|UE_xXn}_oVdZP0z9baw?ZB~L$4uM
z03c0vt=Q()aX8zT70U(N)}AumMaoiZ@v+s#Q#s7~o@cV2ekN@tY|Yn4ukQ1Kn654L
z-f{S2P-+!zlf(aqy7d~OQ@tPV>9WgeH=N)gsJ(8+#=Y%bHd?8kKbjM8)sJFX=rK=Y
zb;F3Lb?LA~d^oS}D=u~zCW^A-&*F&x1Gsh0SyH&aAW@8sHAsF%QFdr+s6{;3$q%ci
zVF%me?z=Wm@8o!ud^!xeU!ViDni5!_^j4>Fvte!tocyK(r7e7+Ge4oH6g^`^TX(p!
zR%9)hE4zWkYG|XcW1xtRS(AkN2Pl!|nLTxv*x&C~=HTYvVX^g4@oV$fM=^_L0erCO
zMy&mo!Oup)NK+e-Ul0VIbxG}xHsJk}PF0y|)7DUr8^h9;DyF}S$CXFARa0%(UZmK8
z<Ru$jWz@$_{MZ?65MF^sGV+JXlv8i&R6H>|4@I2_n_~RdytqD(FUPbnHUg?TU73Qb
zX|g9>{0t>`ZB#tT;`8Mm6<gmT0TCCw$C@owCu?XZjCY4B7L5t?3nTE4J?CHgA$L^i
zcd2AAM$PF@8I|f3vqM9yv?$#QvGVq;k9-JCP*$maUi_HPA@;H*N)69}BRmIs(j7ef
zUkSKKDjc*<EM+ciEm%y>Tr0o$O_e}e9G^^X7;_fN6^dyRyqz*aNOwtZ0i*U)$(5`^
z8iu6Qjv#(xDH5VN_gI*9+S>YE*8Hb1Zo=f@RMW?*?4h<gs``5s3WG8pt`K<+t{tJQ
z91sl<et6-uwut-5eMAd==?wYur2GkGQ7G3`RyH}f@C5N*JJmSTb|-c_S#ZT=sp?9)
z*JDv)qtmt!SixZb6z0rc0BvD#Yapm{v$ErwD@!>D^_pJ?fF4NvgBaQBP5-k-@QL3w
zOy@2=vgpHv6eJ2DRihr{5tP(zHQy|1nv{iID1lQC7~Di3%*IH!oI_C=jWItxM>bx~
z*{L2xCVo%3md1n(aicXB-)7W$P}#(71MbE8*U!9q#Oe})HK|1(0O1IBz@KnJ>`w2>
z__QA|1b4rJ-9Sa7^u@<W+P^%&^0My8;*~+|TOVa9Aq=YfC4S@YtcVRd;nU-SXEc7c
z4D4ildjYy<{fW^w;k3P*6%XDIY^Hs%Jdh-Ye)VSqu3PG$`p))?!)d>fua9^ezrF6T
zHb59F&u=ZcX1+Qbu%uwDWV|zJmwg`ZcVJ*9Qm?isRZ-n>n^PFxz$vMH(KZu+Vp;%N
zJVN}1u$2fcG`uppKA5Vqr~&HC({U<YfYiOx+e%wmA-nQKL4m)M32Q%MSFK=Q?QkV2
zan{rwMe5Ab@|6|(&Mj_Axw!><UASc*(2LpiYx!5M-FVbEK!?LnK7Kljw!ho)LZ7t<
zpn1f8K`_t=I-c=spt-&VW(+N(2UFf3Q>iKDej|BuQI&riI_>!*_1d4Ue0qKtGEo;I
z{BqYUGXP|8e&&Ak0VrX{knLOngGH7Izv@1ccLtmB#tL=F!Z)x}4lH6@S#2rhFsn{0
zdo($(<8pHcFqkQ|qh^>Zb|;B1p}OZ??@v7F0tH{8eGw~tQGozU#R(ccq8#=8qG*Y3
zfesDe_SQ4}<{?JON*A4{yDP()zOZQCQfkgFxh<>F_^Gy--Fl$LjA}0Hm-K=$p#Fy@
zU^UD-<Zj>G;lv#6<+b=GI6%@yV0XlK5Gu1ziBSVS-oU=_rCynpNaY|T_d5yv*-#_6
z$t{8e`5TL@ss&}w_TqAp>x}<ElV_*q{sB%jVAIU|sbb)ZZLM11qx<yLf;Uwe)>=5n
zTv*?kQ@g-`S;BG3)*arc885#it@z98b+K5c{BqdGsIMLLJw$#z7H;;OIry-gYYpvg
z%_{)3oek`@%ZmS7F3V99tc)Mh`G5vqB>dUcl)h$_&3fLLs?5`fa{H)jmzMb$Ao8~)
z?9IH<dnx>QY}UG5d6EF>;y-^~@5w^F&>3^agDc8Z`2;xV{I=9EXde{ZFuf-3enI88
zHH}GqR=~t;mS$;l{<l8%n<Q?ZCCv6`lIbb8t03T=7u=ASaab+wiYo)Fp7&i^dw_Om
z#hu!W|N9{I<8k&4r?kko*hHQk+mSKnOkAn-Hh_K6Wnnb(+;nX$H6~D*({Q~Lwtazd
z)=M{R1p|F4V39wErIvC2$%r64NFZMxX+0H_OQj+6B~ai?(HihII=^KWxcz4ofB><x
zi{z*YTJi}ztO~4{l34?1BVTi5$}pf)$(1&`UI`2({kw{Fgz_(Cz|NYA-IDgBr~*Gq
zBZ9==QA*??8{se1x0fB!GLG|`rx7T80#eB!(x;IrblZPIM%!1!QBf9eBpyuNNMzl$
zQ=Pt%KJsj;QX;t&<~%Fi7s>ONfT5d;P3}vJ=b;8W?eyf{j=0tT!bH>~%-l}kvEFiv
zHmMfep9T-@1nyX9!J)TE580$wvWIN5=pH6G+zbY`^iF!po`tG07;NVfKkzAGbM2D2
zYwLA7ugppj<~veFt^M{O?b_tl_*AP*wnFUtOIK$afjRhfQr-xI2Q2{15sh-VopNBa
zCL7%rVNI%)kP2bm?dzPK%JReVBZQk*o&PI5DJDa+E|_hiZ$-2$gq^;1-?g;!p2J|5
z&D`r~gQ<y)E?a&o9V)>VR;=tfdRd1X6(wm)eo$W66;3?iVfGUo)!ppvuZi8-EP5}s
z3JPabFq1}|>*j<rdl>+PxCciq&qr|d(eMkF#U3mHrwXdZ16-G$0YL}>|9o{5^_HE&
zc}^(p_gLI7^}b60Mv-iUaS@H5itPBrc;T;@x-!=V4t1gz#!X+KJ-a_OBX-wmEVRz!
zCAeK;_NxHb3QN!;;xxbkXxb^9!s2m)75~z9hLI*m=&V%5N+_UB$O4vES68@knCsp5
zxiYD+=Z%7xYUcoIz`*#M#;5b_uqOee{Yu2-U{Xy7y$G67Ef!famAXe2xCnX)A5R6m
zN0szG#y7VVL?h8v`kl!KG1xVguje<Jx~+OW<U#07rDn?(FK;de!tnBtmn0|a7Y4$m
z@PQ705Yv$ka}eOx+&XA^J6HW!yF76BQxyblcQRKPCxd8+x?>01vGd9q?EZn%@Pec>
zN^9Yh3dvlFDFXYwzyuvFHQe+ojwu3~B7*zEY$x7sqQE})kJrM5G09!Y>*D8xKvPWm
zy4EG{&K_MZ5KJyr1nW1#@GdCx?N~#}f;*M|or}GS)HA?ysM^s``c)m87eX3qyEK2)
zS_?L(0VRPr3ODk#SRNYj{sM<O<5%y$CBkW><dCf1fvbg#&9%|^m8sAXi2dr~Af0dT
zp_MCI<2wyx?*Dt+`Li0rjTp>}aK;j&|3yLA1Vns?EQBZG1^M*P@Pku`<u3b#1Czlz
zfFW`}tqmbeCiuvR$(uHJa9e8BF(x;MP2bkj$Xg-Zt_AEibsm<}crr5#rbZ^$U*S#n
zi+{FX=v?Dt5F5zDRspMD(`ny(XC`4C{7QicLf^G*t5+#ynWd2@zInmSe#azt|ECUa
z+4iy%F=E1uk%Rai{vxM!^KD^|0GbHgxOli^7CFdJszB9nc;fNKmx7<;gZndxTCow2
z`gN-~wbeK&=5`1)`ZIbN=60eX`I_Es3=sagUwaULy0f+6B#mA+F5pX<Qo?n^jZEgY
zCz&buj2dr&q1Nr#%&FtANTaZT%F5(m=taF&`7P74f89K%g3u*JQpH>^1EOV2Y0Irw
z@iYh&6G7}E4$fQX>0c5xY7>Y5BjQ~Y5U39s4oCl{CVH`Jk+E2(-;uDb9xu_DMPKgi
z<qGp{_yVG}Bqo^TcBam{7L2DUfDZX=yN0F0YV24s4S*oE+2szEbXG^0O%%-q&^XAi
z2;bhhg~FgL)lw)N$IILSlPgS~<zitzj6n#<+lfv|HH-D;jc_Mjo#PVLz*RvmmC{v&
z0UFu3<SLuCYCfCMDO3jEU~*pWsV&A|W#wduPdYj7+o??26n}pB{rjmO-#?Tiv<qb@
zKch(u{bk?kx&kk4hb&c&6{4ShGd)Q!mMcLR8Lv&ufbMf{if<Y~={=BdOADodCWF!>
zWQmdPGsU=n$Ao87_{L1+J*@1n?<lqdLzCtiD0W&!h=^PDfV$hB8%qC<OWft8k)y^H
z`D)Ey``eu5Iqur}ADZ7_uW@SrV9}2WWt9mKxttd&{USTLZ0@KjyNCGo^(jQE4diU)
zi`gmRRRChu1uq%E2U#o3VBKdfuEa#V$mUBg8?8~@!4Bc5Me+tf<I-r1X>AOZB*`l7
z14oHmwLY=7;%#fQHQ)Y9C)On=2OUK=*pw|L3&1Fl)rs!YV_m*2bH;v3TKHkokx?5h
zc&_;;!CBKk@8Y%jx&$%`&*TKyu=s4k>mqUyu|-yzOT!FcQ-)LTE9B9^_J}ai|Bk15
zw4DMtRK(u&HvcEu43EL%dVaFRa=1if;EA)R>$9&&T4H&4PeHbFh9?$&Fiph2{@p24
zoi%qbQ9W#SUACPPJeNxo7w!67PWp2fo7>kNJV@@&r!|(5Ti><bvLgMvUfnlMwh1-(
ztQ<WS?F;~b<;<n0t7+xu?yh%nbjA0z=Lq|Wxt8HRcd;hR9({LHhZNI(iR*id?WSBO
z{Lr5nVSK3HB2ZmtC@A>SH&k-{_hYhD%ih2@(BDE=mcq&HO>2$f-vw}!>W@(z8k!wn
zVTQM3V;8Fi19qampTYIejD<U}oXE+O-ub5avc-Uf+_luZ0HdjgFoFpWmDwV0pAame
zZ4-S35Ia&f^%mu7Rt4+INM~@mtt&l}?)uS7rwu#|PR78vx4DI=%d<wev^zDg_<rUD
zky0w)GUJWB#8!#dfc3cbV+odlCAUojFwgiTG9iUJhaR;i5<F$QK9Gn}x^)&eSZaX?
z16aV1H~zq+X!_=D!Li2FV6qCa!7j_aPK9OvFdk-i6uFG~MNWtk+>QCQ=Iq}BXCyV_
zfE&N<NTR*V!3ZI0CT@m~00rr!ljdc?oMnjmO9DR<gb8}0aKA=wG^(hp;FuG0I>3pr
z%mvt%PVCRw&*W|(<~?-CVGaY<+T=-+G8O<qoFm<hNdg5=mHWm!;e~Di3)dbpf8_6a
zfM*kEc5=xUml6K_IC3SxIArWeXOBThCQ(l2Av&4zJo%TMVkn69_xUCQo2a%y4w?>Z
zAfdy05@B>HauPp`-0SflbPZ{XO?+IXebjb_{t|ALiyDf}(Xunf9xlj_i2TpNya0fn
zQk|EI7iM7Hy88vZb^4+-o}EOnZAi}g10#@NkY0A&RWxGm=wm6;`m2bWt}c-i;Q*l+
zZQS$6eyRuJx1l}f2&BWL!#Kf&QQxF~`SQa?UIpNf5<~~(PpVE6VC{ehGtD>CzKa+}
zVW&sf{q?_E5!U4-O7e19%GOgD;J{JaPtfLITn2)8Mr7aGuY%TcV{e|np<I~6xKe5Z
zmlP2m2aQQ)J`v=n%EV<A*(molT<plV5lzSU^67(rqqSP;vqB|sqW)CkDC!nTuUqjk
zPo7F5bPc)v9<VqOC=-9scaJBYOp2iRFOj_0&$USL+czCd#bhhhrwRcVK->{T*yr7^
zkBC&a{S*jx(+K7hOEJpG^x|!uTT-cZdj#6c$fKTlgxgEH%X*9e!gb^0X!ePcX@J0N
z?uK~(+v76<@n%W>T<$N2$&K{lP}m5IeV@*MXyVXr?>T*t-RXO$cr)DTpMui-xjG?>
zH<(8`lu953i5e%tB^X?Qa<7H@4&&azI+<ee*2R)M5R~c|3NOFcchdNDpKF%X;*wN9
z3!!C}YTM$M1+Vy}=!Y04qA3O8%J1dX>k6Xw@YaIa)dAkl&)iOa#=@HC!2jZH;snLK
z9y@%xDL=csIV|g8irewQ=(m64VAHd{s=>FY;-*qEtRmjzI-pf&tR&UUF0V|CC#v@a
zugrdS{_6F;?NI>?3)lZ8jwq(R0PU;a@Rv#8zw<IL|0(@V9(G4RlUfY-1JZl&47{x_
zk=Y{{n_(x&5x>bf_BSbT)$ZkIo074TvGOuo0;|9Chi<RyKaQ2dwqsW8hZ7g)dyxgX
zvlqRN0-5rQ6`(tqhQ3vGb8XcV9;fNOG;jK<80K;-zsM~IW~5Bu&Suapf(skna&mkP
z)P&@jD~Q1?jqwiRG_BvzUz>*qt;+@GzER<#viTF;82*74Iv({kGD*bpK%AMDQhf+k
z3mzVb-T6}p5(qzppN7kX7wC5)xq!44w4XJI(op-RIsj*5bTZlnXU8z^CWO_pU`D*R
z;PVi@)g>IsFjSP!wFdZ3mYAbZ`*FhOug4LMNjxj>YfH5S0mtJ28p}lmm|gag4#grI
z{qcF<Uo@urnB**n7gl4gR(N88NLasdy(6OgbFc%l0^njRx+Am<DR5S&{GY~+B=dM?
zbrU324*(nlxMy4K=g3vxG0_Xu)zhQ~7*ZWDW@I&7-2V1TV+CGjyF(SbTb19YJy6p&
zGQj1kiFtK6)dG#5U<VWdgXMd}|7RY_$m-#&5&DIj20z-fCGx^$vcQ|Vh8$A+pOQT5
zXnUJv;WhMdfc4XEZCz9{ir*nveU^~vV}+I-1)xmmJp!XwGlW6y$m?$~<m`8Av|F<t
z*LgPZUM<r)7T=$F84U`La#GgR4!Sd^-vu(NW295t(V_g!sCo&D$X>Z7UPj8s|Nlx&
zn38x$n@CQ03avEGy#dKXR5ZrNVlkZ7Ut*98*D2n-I}Onut(YLNw2)H(O%NET{VrqF
z3^h{^8aaNXT-}zitXQOrtv<eo)aZ4n$(C9NcDUGh@<(ppwb`nhE-VUuw{5!uIk%lr
zf{X%;FobK2xdG#~DLqRAPG!E%j1UilQ6lSuMqD-{7Q=P$b|Es>`Pp5TP`l_0eah-L
z%$$Qk%*ye%?g_d@?R^12>i8n0&qF^#(oL;n$vwGC&Aaj~jV}*tjM}QBreuc(P4Pjb
zP=oJ%MWB&L(+Y7I4!8oYx5}Sv1Jx9>?ZW{&UUl31->khLXS5yxqr~=+t8G%L(TR8o
zmP@Qf>2L(#4?_A%^l0n-(>|O(*6FZ<zyw^@;2w*!Q|;T8{W&Bc_?3N`BMA6rBOw#W
zFaygG=r3KMjACpmoN#FS=gkys$&-GPoXyYtlJ(5bre;YVh@_lB2I%AJ+R1hnjbt2;
z?E|pk$A(2q@^yz~ap)+Y`Y|?LJ2@`eew;|<ArD#|F$y|u%#GNs#S)%G`=6}<*M%gT
zpVU&IKh%}R?Ya<vZ-wA}P{a}+{B0;8*}psK*ucZdS;_jNDsj=eKB{^b7cMOz;l+t2
z?65JZdm8z^^UI`WyAQBTu5&FHPklWzX{=Dkj$`*E<3~29k$WK0!i=NgxKkQ$ee|)7
zdF2&*clYGxeYQ>1de*PM78_9=DI<+=g6(F(rFm?}5B&hZnp=$+R5cwo41Gg?^+Vfn
z>KF!$L_;-AQjBOq+1z*4PYrD@YT!W)>n`VOxFJ6#-@+O?hQ4BLs(wh@EaMk+=Ayb7
ztXQ9}^^jVNv$DW>UqA2F-aOg_-!fz{YFB$JXooKdwxe-Pij?VmVLjv{Uq1KGJ{Hwm
z+=MSLw<^F=kj=ChTEl-0E2H6pET1*yyD-vEZdQ}*4kV8yvUP=NBmv?NdU@sqw%53+
zl4SdBku#z7yuqEZhLKx31q|%!&;8Ye1|1y7j3nBD`_YCI(<wgXslB4jC|iq0MdGo<
zjL$u7h*^e?T|x?j?{w3fp;iOA`YX1kifq3@fc<Dt(+&XU+G&g;%d)*5D>binioR%T
z$!hQq!jA}slmo6<P<Z2$e4ijOzT2#K6ce4L%^DF_blS;?R+9g(_2lnV;AhXrUSeza
zP@D%fgzBEWJK$`HVDm9XxC=RjARlsrALRtXY#@Ik<JG|!&V_)E@aZZ(tB<O(`v`Er
zbxhD6B&#O5Kw(^Mo&ueA8-1F8Uu#d4x2|&(1|_F?U(Bq~iPt!F7h_{SD<5t%&4G{*
zEvSCVvTJj96s`FrRwW*Idf>Up@?-gfQrw0m{M0nuIAjHb@55P;NJn;p{)^JHNOv*R
z>d)U@)uJwsD$aomc6pjGeFfwM(g%PX6P=3}lJbAm8wEL_!d^Pn&qz6EFF9s*D-CjK
z9Z%bpeR<qub?qi3Uyi4TAbUwWtBlVSE%!hJw2!yT$11v@CCJ4S?sRKUDiELx=K#?y
z8Q7j%0}+|_*t!5sDO~8sm$=rs9rm!kcw(<g-n^{rLmHRI`oT`f5m=Q7$j`0iKs;U#
z!L|Og(D3l;d@<~C+3sF@8`3=-L8sfSyeDVEZ@@zzS6fkL)_h2t3#nkOp@4#;lBO-j
z%C%?47A`)ZjE^EifslA}CH<xR(=H=rP(oC0`rn2U^7jqze~Bsi|H@QbfTdS*jD9-n
z7?b2wycV46YlE<zZkYinZ`$V529dsw8kg#8jqGK3gpD*c4mBr%pWpW2uDL;iCdW@M
zb>8z&_Pf-L;+@-nh;dOSeL2S077YKOX-Sq?Dl90{a!^h{#p0;{`lIZ3_vJm@*l67;
zP9h~Z)K+#6Q04XLo_^7HGLx0J{`R4^tMy5e?p~Hn3`x%8oo5B8@7#&8hYB52>jXl1
zd2TZo`#9ybatni-H03vy^B~bP?x;}aOETWYw1)n{fUv4XRaX7+Wb7d;dy>&=mVcfH
zL!EyFmq*YMgS>dzN!M+;ttBi(z2qtb#&KP5#{Q2bvo-YgSNyE1Zn;RQaycWr`4f-b
zBGYY@*TOeg{2Wk*DX5+~lI)9HrivRGKQgpm+!}qlN{Y{V#DZ`f&LgiCQu|@Hx>`eg
zq$G#6<r(TFil{%XbN@Py>~(ab8~U9%a`z<ijCR1YWt%-8Ht9+}|AwOo_U=Q0I#f!G
zh9W!#%oG?LrbNsuGGKh$w3;%UY=1*ljX!*Ll7F*+xe1UTVvM>nnt?6A5@+c^V2N}*
z;=eJ7o-r8LXg!QErq(o^*F3E7Ru7!fGn-1=tScHnoBDqy%v=X7Ik}OroFyp4j8hmY
zd52|_cl2*l5JXQ#U$am#j18~<_%1`0H1z-TUCysPiIV^0yFjvFVXXG|)^h620vc(^
z4gWTD(miNh7Q(O<%7`GmO{$kCSjrf7c-ItuAe7N7oE(U&Y~wxo_}=}`X%P{QnU0D1
zCY2aifQ!&nNCXexxA51EW`CwBgIPTC&LpXnOPMR4Pc4!mlDZQ?&WSp))q2NcJ!Vx3
zNJw)joA!cisz@FI#T$o7KnC3ZTvrrkC0T!oEL+6YrjibFxgX%R-)H}i$FfrmFB8}W
zFNQ9>-ty1!S7XglM9E@!%ObFyl8C*VZ8&FM_J3WKr=CgZ9+qDYSZI?CkwJE3UgNiS
z#krYgcZkD{|6DjK+)<DKrxa-`S*jVG+$(!hfCY8az{>I`p%Xw5C)g&3m`O6N^>4sj
z^`nhvXoDD##ecj$<K<?)BieU_OMO6t4pg4u4xkgRdyR@>EBiw&d-9w7DuOcRaZtQ@
z3fdi)>jeWvt?6f4eROjC7ui75-z@*f27lxT3n2ZS!&drLi7z#FRvoo0zmYMVZ>c2-
z31T|<zvyMQ+5nL<xe_6Y<~x#0OjUpM?iQCnB*jzWe$usy+D59SHCbE}9pRobye&k{
zo&2qxT~OF|D;KIhgyKf7h4%03lEfT==i6baom|8sr%&dd6p!u^zlPStFrh*r=s5iC
zH`%$KE!*hSK46BgW(2?b?~cUZA(@07xzrSWdF}$zOaRESa*A?f-5gc6GihJKo7!Ta
z)%I|1ZZwJm30>2RRC;k`S?Kb(EBRqR2vrQFejqV~sZ)zE`AYGUYlS*sJ8imhdESGE
z;g!f}irIL1{M~#03xV_W%diO2DwtI!)m(r!!<Qk}Ub6Y+#QnA3o;x56#O^wW7~YBR
z`AzKE5dbXd2PPZ;gnk#BEwGILS=@2y6@QBO8<b)F$~yS#%-&TDqb&zYioz;Q9OyuA
z3mCGofd#4fk=)uW{3Y7VhKtO|QN&xL;7YO04xz0(;eVBf(1-QPh0X-;LcWu@j44U%
zufR`3f{UwYV0oG-^J*B0Ack%mM>e<j)c`=CQmErc01`0a(3Qy>&RfdoYYM`#Gg*{y
zD~2+z)PJfLd&}!ZI==WWO+jLv9ZCG3%$F<sLKu^%W}&C66yaR#C%Q{R*U#;KX6-7~
z&pv&jraGh%22HK{L?8$l6;=V+uGAN07F<XDU1Lib;G7(bKlwJ$!|pxGh-og&i2(8K
zs%z+P<34cRr-*gJ4G1v|%11lI;E~hdVA?9Z7<dv7B{uIzZ8CiH1)L=8Izllv1F48>
zc0>3PL(=>68`LEE|1*;UR@d+losr%?W)BFRo*7o<6YeeP@n_ZqfupWz=pwLv&lH4M
z<+++)?|67$eH=NG_-UhNbO5319)a(p$JfS7XBr~4mRhFw@V)Fp{lnAjw_iVD6syqW
zX&18rsl*seqt_Ji67c|;h6|YhDSt{yuG6<%e6_EnN%TcJGSKE@GvFd075LhDCo<-x
z#Y3Jz_plqGu03KFCoJ`_50^Zh4FpbsbXWR(wXQ#^8d)Q57}GSJ>VVG*5bGpX{`Hz4
zs%y^6xR}l7Id4}!)U-CyXPr@8uidzWk#%h_#7%(#Z<>*5(A&0V-+(PROja1j?6ma5
zurh(q_fal{q>12vTE@&fshRaNcuyA6EhgoYFW>XpuOU|pkB^_waE{coZ6U{^TfYTS
zf;%!7A~nsDtYel?01?d!X{TY_D3s!$spUpSJFq!bNv9dcPzl5804@E?Vg#$OG*jw-
z0v)eG%fH(>4nr?J!%(NPti0cU!d?auBr61a@9RW|>dKN+G(wI)u5K_R`hi|fF0Gpo
zZ@mgs_;k|bj|;<>wsm`bKNI>}+n`~O_?wZr?8CcmZBL4r0OMtjN`dk{(dKl0s;pwL
z|2{ogCNuI(rj^H0vWr3%oY98ucup9BdvPRQIm}C0-g3LPW)lr8XYq;98+=A3h;kH4
zWsi36%ffq&Hg*&bz+w8cZAXRF;GR-I&ar!B<D=8*BWqKB-5*d5cjg;bR5&OUKm5gv
z)J5V6TM8QD0u)dQbRTCbt%!;igo{5CMpEDCmj?d1xZ(cie59&u|1lN@3JmYfkKGMF
zF>^$fa=57b*NTqvw{baq(-~M5a?{yPZOMHsBYx_BTg+}xF0rve<mNxe_DmY(f-J4h
zEr_k$rM?-Ex|Yg%@A@qu)5?zemYzIeluQSd4=YbQe87LWaNJ&R$5;a{|55gN0uAwi
zR#!X2#OjvNv0_-z$+K34ffEK@AM@4y2#J&wCrWfEZ{v!<^~;&bT{m$$G+?cV35{xY
z$Ys~LhpdV37^R@Z<$(344=<UyU~t5uLXxU*LDcc2fK^myq9iXef_^ntSF$Sfpe~`j
zjF&?X;E(ktoPoMH7?|g1pOus0T4957zcqJ#9O?chdLz$xCoyZ(Yanj<S0*wr`0i!E
z|HDZjfpA}p))X<DmnaE!U>{7o)7UCXnM+;xxg`AKIs-lqf^Q4L*W+<qJw<_^5y30e
zZHlK5AqcpNg5eeg2jXi|?>*fSI`;hOQWDt)P-|Sax_sBly#Dl?Vc3Iip^@lw{%q8W
zrjW*|i}owtPNOdi+#rdU%-TAMJAn&M{({cLV<$Ewrp<8^WZAAioimDAHssB&+D+%X
zZMtnzVyRtv^SD)oHc&mN^+RzDQV)<lI&?o04;SI5lE!Fs_Hlg}1cVXv&f~XHmb0J(
zm<(^9%8sE_9a*o(xBIxI{j8;T!2imKremI`7*jPeO3SywF>uXMLw0q{UWg;N<P&Ij
zHKaQ3{5L=rZWd-crtwdEe`4!xurR5SQJd;(kgn2I6t%mkE@~on(0w5{LFYBj{XeD7
z7*JIf%9~5}!udwfeL;({XH0!>{JUpBCqW7fK;Ve9V9=zlKRNPSz-PNHi!+|}g?d%>
zxw%HfXkhcIvAbW){XJ4{(Xid(sl|p%n6#<ruc%C4)moZY$|uiuwO{Zh#(HGpfZp#~
zQkugPu6+<9(V@@yIHVZWPifS<0aj=ij7cP;?K)JnxvM4<CiKRzzLgZ3jOz=)p~p4z
ztoYyW>|Ml3LxR*PlK1HM%+0fsKWD!dk@79U#qgc=XOE8Qnn6Xv=>>`CPpU7WH5yi|
zezGFFFEe|{-W;J-#!z002j2H)jCxrxi%I8?%}MJO&f^flIj86lcS$)fz(Mr{Vvi2p
zI!MvWFxIm(v?lkB|6!@+>4Uif5N<L}xO<Rs1g8qjr&ZaMTsM8mqos5e(!B;FrXJk)
z2aOA)S<3opm3}>0+Y-{6O<(ZNHddp;JIfA3I88gJ3i9&1`I+7hRmX(Ol}yDnt@X~r
zk4`~dIYG_K7O|NsG%Nvn->*JAMt(=jM#M8U3VL-?=S2@G0qw7}x3nh%+M78T$Q#Pp
z%PRN0EKGK3)+WI}LI;pD(o;g~{jJ*5P@?dK*0-qcO=e`IcuAi&YBMQ}X03A<B?A{J
z^A-^i2_vFQ#Nd*Q6El1zFEG}jSz$_5MR=Yo#S&%*v%nijpq7QNz7Qy|lyLWWl|eR@
zQxAWq*yEwRk1)_OwwRCrDpfIW=>B;=VigVTnI$3=QfzNMwpljn+cY|!pG?A`&a$l&
zQR@`u_hx2jrg2Z<vy`;8obADtC?wOR+YHxDc=wv&JbBC7jGz~+7|!hw@54!`xIQm=
zHv7j)?O7m;+owqQwoNx-e9H9%m0f|HZt3X-5SXU+9&U}0pPD29!mM~`+DwNih1S62
zKk%m*v(#HR(xd+Ao2DKraQ1tqfM7y=9&y}cd^>xrCh#1%)PEiy1$7KB-9h(uvQQry
zx3}J6PL8({;`uj9>KdjFIk(bEdwY|8epIb@b$%>#68T!Yy<e$c8R%_o>5qQuZ-3ti
zvLSY8yy!3Xh=2S7^mClTRPV+DZDVd12cR5*i7k2FZKp5iR6jXcu`ls86nev`1ogH}
z^v5vbA{CBL1kiaMTQ1lZrHv-Y!hALCDo`Yl2=vqaOVU~U0Ihx8*$Vk5O7f_M9-4ma
z>eHGoaVv5AJkWiK&l?y((p#3>>WESXuCxI1+EZ^-f6uuAplVoUb0}ENo!(l78T{jc
zE#RAnMc<EE^xXz#<-O=UoO;V3KMd+#ydHo~-}1hEMgEw#R-#4a7R|G3S>WoKKmJlW
zrre6E=0gix@f2Gbz0ua*(sH4XJF@#I-|Fcf`NSUD&-W49k7%D(D=H7VX|11DKU)pY
z`910S6iT%PNSnf+>wN&0{4mZI0jG|&eq}7?)I0!1i(Dwf_pt7B>Os~nP-VpTANWl?
z^Y<;YueidijSRm48#we!a6k=cfP5p(7$bLxiboyyu{M{wE&Ja#Ikc-R8>j`M6GMw?
z@>Ds95VXnHl!`cCs%+}`=C+JMS166<#Qez7Q<`NYz&~)rA5=9WWS2+vXV+72`!oqg
z=rB0?r(c<tYXF*O9mS+mKQgyiuI~yAM4L5<5q5s1NBw;1?GCSbadgo-1qI2FWV^Y%
zJ9VYJ>goz5q3*dS!_=IG8kiM{6n>e25s=%^auSVvyrL;|CMEnpm_M;^%M6%8FvAcR
z7QZPTpi@C+I9SzTiQV3x77$;v;HYUCKKv9cst^)QK^}36$M9Tdb2kk=iVU}54wdNB
zsleyF8Wct^vr2((l07j_zl{#8wA!6S&0V{*me5lluI20{QpMn4Pv4Fqc)f05I!EIC
zyga^DO#fgOp#&2`T3J2*o7CVP8Ur{e$-{L5Flu{(9jU{u_ZkypIF^vs^faVu*Fl#;
z5nDhfNS{FQ-4xcbk%*n-1>SNF>4QPitMkx)cL2(%l`h1o?lPloBN)a!L)D=gEMh!Q
zVNJ!2rLwM$PnkLDl1;_7@~ls5AQRUQk=NfKGN-r3SEOqc&y-Yp<dJGAdggLQv(%pm
z&;Y`hbyR>%EYxnNYuYlvCe)&B3SZ<rI&2F?!R-(v_<O6-dwF~sy!g@L9!r1@DxO?O
zsT*mF$bd?YT7(!hnzxOcNdEH`F&2iwAmgF;h_*tKZxtw3qLCvM^{i2tTGx9T-ehTh
zYBf4<Gs%4((y1%KD#3EqD|Oc-`GW$O8jSAN%K}TLsy(f`2q==dW==UJcKX7>RqO9%
zzCInn?ctFf+1j7No)-yAA#HIv%A9%1TwrVCcO??0RgPl2^Rgq*j&!|c$Wv}!_?=VE
z9}f3;a;Q)3?kb&|F~NSb-5Y5YUrCV`jk*p7$6rLQ|MAaTGh1rg+#GOIK*$8pUmy4N
zl8*UP%TP%JO9!MXD+*HyQh6>b*LXI$8W=f7IIXHt5%)TN4u}_PR!200wun|xl3ao_
zt%Y*-_JU@wD%NMiD=t4q82jcxmU5NUwp?zGp8ox+>02I28qi^r79_i+-qNDI_O|f~
z!(5u=I<{l8yA}<w6^FG557Gr}eaiOwkM@gN>7}h~v^$-@DNph0yc}$%rujv1m;muk
z;ics&Pk~|TJxl~87K)5>+t8Id(QXCn)zmCS92&F4!}0VI0YAM6+BN|tnX_Cb6z~+e
zUlhOuok`%S9D@IOQi=tExZ|Kny2B~i{&nM+$Ms18(P@tDzRmQml3@=B>LAw0_z73Z
z5QBRiV+R0Cwf)Zdepj(<*GohZHJ;qm-wEoVt99Kma9ey5`Mtf6qMb?DxFLQx=QVFK
z+$NU1Tp%G%PsE3y4A$vEW_#JLJIz(q9}dp>vsuU0;*|?1+R9zlI}T2=fcAITB##_!
z<zt;>{BGM~eAQEW^BV_z>h9&R9oOdYXKy!aNqUdG{U}IlAysNlc0!N4s-%aEhNI|o
zFSNzkY}63sW2O4ADy&4xAw=oq6rX(yA+~-?$?sI0_Oxi+R01fqtq;it9XamqUUyUr
z>-%FXU;THFyAoa{=HD)}`n#0xWR^D{>1ne10c7}DL4&}39b$m+T0;&K5DANZZb7P1
zG9;MjSQJi*NVF&_NSBv8{w!g$12IC!)W3_ccDss*bL?^rL1FWf8?MM6^)wYxb&73=
z^3N|)EdIzXz5EPe(N3>MmuCcueB{!3n9yq<ncS|&+Pu!gW>@@t@y)zKK&_acyoGdc
zUXmLvVLZ&073zRL_#ty)%bxzdF9Vl$iPDV_@jSxL<wpxQL9L;Mawt|m-9mJsH=XG9
z)j@xwlV>tWf{!KSig~#`UJ2otgrh$afJB-8AbBaKK=esE!9zgNI3U-P%u{M8K@A;@
zv{ZkKl9D=U<6&Qq?1vj9<V=PHlh{W>?On<QWXliW-rNBKIV7jR5?ts)I_YG?(DP$E
z^Ep%S|3!qaJ>f+ZvyVB~#S_R&-QJL+MQOL{!G!D?`)K0a5(D8(G&wKO`dr39r<VqZ
z(6D7JYL{~7ApDK2jZ7I}5f@w!dp(Lx58^nlB#WQEgIwh?f@)>PeFR6|pwh6WV~X+$
zh0jqG=i34xLO0vQWlBz^hz&%5>)$G4*jSRFtCy{#j<czVsU&K$R`<o<tI0b~bY}GF
zGtMk#awn7$RGp<?-W0&<10Hw-CLkl<7w`UtG!}mlT)6oIDZG^2fDEfERk3o77&C^8
zWtEy;oYZIw&}(4T#zQZ*xx1_acoh5MsP%8btYiThuF2tM5cPtjG0W00Vt12W82OZ_
zsFQE7s?n`c{J$I$aE{yMKOZ-&qs3%8num4yA&Xw!Dl`ooa0i`NvesgO`#E?CIsDR3
zx6`dVb!nE%2Tc&Bbs0h4Qar!y#>cObTPGEQukiYnw)0I~QYYhj2!MT{>)9R_kKk>}
zO6CAnBXG-wSL+ex<DFwwM2RrnL!j*#ycv-doBiAFBKxWK4ycO@#U_C_r^MlMsCVzG
zvuqMhtWUe<nasv=r+2iYBI-bP+S#m0{(rNx$}w-al#cqYZu5rA+Y7<bpz}|FP}R~K
z_qe!iTrA`w!2{Pc8z9#o<1Oirm#nF}wUz+REiDKxK1oDqrBs*e<QqQ1K*PnR%4^~u
zp}=B<^(|CI81<6Ytwt297=|)4&Y-8`xa|G2&=Ya@76WD*GUb6*Pa3cz&{7uQxoutI
zu6jE7%=Ixep1-LiBkxg+VA`Wjq<#@x#5qN^Yau+*MLP5fE+Z6rdEqgHZCg7UBsv2&
zdCjT$%j80Son+^V_bP!LxL+S;6xg>#Sh8c$h)7Q;8CplJpBq=s(uT_`i0FD}W6d))
zvM!F(ta;@l2J%<ODmRw=F#qfu@@YZqQTGqkgTOxeo<h_76YlRBz!8c9Z=D2lBiORc
zg~7E9D73*2N54ZVJ*Zd2&9-&$ktHC!t0ons6&GpT_h{R=M>zsn9<o1RiZ!D<G2}cL
z67#)f8ovjyr~?BUi{P6ra?HP}1K#e#<rSX(ewmQo%lX!R#vpH|D@3<zr>{nxox`Ap
zfO%vw&ME;sZF*f|%p<3x!anvfqnwR3wlWSSLUnz@zhHZu{HY$?Z9S>4Hwn-O{#;3*
zP(Bo8T9BA;`-|GWaN!y)BX<qTy@1=ITe%dGWgaze?=0NS)nW0>QF6$jmNqtx_{j6l
zEI8Dm`1{W$8@AG7+@_D)mq`yM76|1?EbxchHE#NuMMMw02vtA$BdZISI)1AAt-WHq
z0!A{k*TRBJtUr)8mzQRXvl*bwz;5Os*so9V;e<0Gp@2lkuh3+WJ{85fW?H<6O&#}X
z$(!?zk$ovfdeHdt7q?=44*PnKZ-bS>$`}?eeRA{nmkq#eR|#*%A(9Hgb3s%kt8@@{
z`P;Bsp1j_yR-@WK{q{ji1=!rPUvYP~>}39|-ltiH?i>FZ2b#8rYzn~be4IjC6iZOb
z+~AkVAQ058r#&dROn#hR)z8plW1KZ}`@2q!i9UFv4n)@5jAx4mqx_qaX}Zu;*V@lk
zBjPe9$yo1eg|})!U@5}6Q@Mh*uxa57!#AV$YXPE4`C&@*UW!(2$&qVGb-ZCRPDOF@
zHjWG1iu$p3)R78}Zwav1cVYi{2mbnfM~2@j*rRPHbIJ68d%GMhhWCi1AB|bD^;i&u
zz-fHc^4o)bp40Tk*A5=bEV0zY+0++D%j6}6F4lLy@coKzSZtbIi9a3SXS_NE&Nd9D
z&!}6Eb~{X?01N-D{*B&Zn_5HvMbcL5x;xU@w~RpZq+j=aKMg=XFuLueiLCmwf96p1
zJQLpCgC6z+NdsLQ=)tK&8%&S50yi?db#En_;bXqw#@eO>`n7-!ohP@DUHok?HMt0{
zLX>$-XN=_ojujL=7j;~dcr{h``#rn+q!xv7G|6Tu)FK8w1hrj{UPf35B(((z0@QBa
z-{0u837-o<K!ClgIz2+V*Xqe^Ul9E|KC^8d*O*=SFOpy=`v*mzxnLDBi|w|cJO-08
z#+pnCvm=vvhcVCm{#^7;->LT<NwMfz<ypNNyNogjPh-yPYM9LD-UTOuWB#<yGm;%3
zeECYIiq-ErbL<oEg7c`V=dyFX@}={~eeMNsy0)!DXF#V~c{7>jUHmy69tU<*yhXK5
zl6B{9`*?`w)=o5?=hg<C$g~dPbVdOUa5e`e=0k_Ug%_kGGxwHdzrk;kjqeB^POB&S
ztE$7R%gG2l?P&}vv}vAQ*FmYnrQ8q#BTZsU8$^(LlmL!IbYU~hS1W9V1#N!mVBQ$s
zZa*GhEP!;WHcc|v+Jiym%{sguw<TrW5Hm%rg|e#$P5Ph?ug)Rz>~yu4Pw$X2{mq~@
z92zDyqY&zfa4q#2gV1A2HtAy!l0}E^ZNS69db>VG)PhYEDtcBI8$;mL9-KNGw{INC
zdBfP@(vW!Jc#h-J6Fqn>J9W(q+KRyyy^nK<FTlR(d!r-vSu(2<aRY7V-nn-3FLfhg
z3s5=t-VsPYtK!^tL}AXh+j?_KP8-KT$Mb1=!?pe4*W%&jWzxp=<0k@wDroQOz;hV{
z`<?`5msCy;s2|dTM4A6&QF98RHE-4JK{s2qfWhVdoS`Q<bJk=)=L~pD&wg@+^%*i3
z89?i_F#i2o8L3G8gv?!N(mXwlceG�kAj$HWj~fP?bEkSTfZv5fIB;i^oZ0&T;9Y
zZ)w9)<KW{_*CNd$;eVw1h4t1xv7T8g!$jkRdL;_u@mLs;Pn#=(MZ0~D66yP=4gS%o
ze9hh+jQF%^iPeXV5lPx#iLbz0{4o>a5paQb(%d@V$eYgE&u_8@QxfHk`^b4D-62Nr
z;;EKA_RqzMlh?1b`~T_MT$MX*^P$hTX7i&UBd+XaaEE^h&e^al5o1ztdX)zhjek0S
z-QBx~lxkNmhxf@CzUlr0;uN^a7wYC=!}^cVod8bHPst!Q^6cfYZqSp{sz1T3lB7_^
zVr$Cnb!t<6OkQEPv&M>|#nxi2qvH7g2z$rq%(^yRyMl`CRBWqa+qP}nSg~!}skq{#
zV%xS|NyVz2=k53F?%h51_{REk|G(Fm>pYKn%xgAqx#C)_K3?=;vea7`%?=bsz~e~$
ze?MjUB(7HnV5EOGYCj*bncfM?L-OL?xy;N<c9u1srDbM*4kI^XMw{Bd;g8F}&D3OR
zv@n~UD9n%z$r${Abj(`C9$e;<xe~Uitu?Fi+NgjY29dUaPXqh>BjG4PRfDIgYKUDB
z-3N=ll6<+D;7)Dc42=+LWgetI$-Fcfjx6HCP!U~jyjvGx3Job}+j#Q@QS<+)NiBV>
z8bJ6p?uo~A!5Yxp9BJEl@47R!1ypbeHFQ!Zt6ii8yNO)=z$W<M`rXS}telF1bqL8Z
zHSQiO%x1r;?by!G;j+^fD1(oMbs6MWv1$FuPHcaQS~}Zvwc_E3TdgQMeHa}LXPQG`
zuiL8l$)$LRSme)GdLe=)o!6u%gClu6!<j`9@1xtcr3A>~IyAfWTQAToe#`TvhF}4q
zs-IU@>O=_<Ip^YfjN4>uC0fjCnj**Pe3R^J9_bTO=4oap_2P2jVbYYj{=u@eN*)1A
zD$U}MMW%3d&v!p!n#Oq8mh5Z9bTrzje!oskNd4L}=cPg0`mpKxn>!+D9LLGm)7g2o
z>x<X+2QiR@pU}~*_rT|Bs9_$h#X)N6s|!K;yr$j?Mfsz%$XKr7*V+5z(<F!PGhz*s
zFYEl0jQZED)N$02Sq#Fo(IjA9M9)qtW~|fp*T3sGZAa|H=p0;MeP)ng37AxJz?69^
zbw5%4DQXj1PIQGW4dQ}X9`|(ew+nBE)LNiq#4u>aaE<7c{<0gPn3EiPzrfSwzc+9f
z8Tey|zz(i_#hT`^;gPS-^ef??%a-xBo$t7UloM5ZD-{gL%G3w(sFOeJ(+|$LqlMi`
z|9L)VbLf_9kL&O^0knN@wMF#xDWMSe*tcSA{)CCnHSVPTH&daLIN*6);x^45)_p_>
z9-G_GsA#pLH7FXd?bC}Pqsw5CEc0Cd96o#GDl?4P7YsY^ajGJgJ_I`-<{iYMSD|6w
z@+~VEfw$*p=?!DBe#!Gcq42AZ4L7e9T!m;@UWS^5rpohJcls~2w=9sr3A6a>9v@M!
zKj`ICO660GBpJle0HR<g4OzvK9W2F`^opA7Cb-`rB`t>bpw+DSBM7)gjT=m|H=<);
z5ax)!j2}#PV<(Ts<JFl}SZp{NiF6=3y%^iYaRFliV+f@T!uwG^OTp(8LL{pbe^U2_
zJQC5#H%+K%3EGtL>k=Q4au&kq4tM=a;{&DdFpDeBAkrZ4n~;g8$d3a41%6=03^z^J
z6>0Gq-ClTaIKmXcg=-4Away4+@!~BJ;MWPr3If-etnuCRWtBJs!5NOJ+=!(Xh)aRv
z&kX8TUflgRYQSAuy|Bi>KvPu0?3B}||CQ@A-H2VaK?eb1FsUazk^<o^QznRxoL!|#
zfG!Ov;#ryA4)E^7R}*=jhHQ7*R=|L@BPu<3Yf34iDWC<Tg@kMc+mk`-HT+Wku{_9a
zR4<Z4NfITXza*T!vjZ}1V*!)+i5JjeRBETPZ*f(+ZO*2$)P$~;GjQ4xJ*r|i%1QMG
zxuZhd?dceh``bB(o2%D;=TOE^hVRPM$9Dj&c~mP56HQF6pLn4Yh}Cw?a5`G;qqrWL
zxX@O^0o!pj%MXgQK{arhWyVxWtlMEDMv$}&$=vy3s9Ehm3R`tjvy}L4c+a*1q672Z
zsBI<ML|WBB!k?E9DT_{^WL4(%;3T#_(buIw;l13rQb+5eh>-QmfxQ~_A-e+>Ne|E;
za&(u6b(d+m;yZjHZ`+Q38>wTfW?ZkWDEvz)e3l?XjbMy9myb^}<YiW?1fpB##I^+?
z7X_Js7X&+8vmGeZXvT4B#Ioj=TXSvh)Sxw6pf%=J1M*ig*UZHhS=oCSuQSP<+FE{t
zY=bZ8q8=;Boj%e%I!8bPjsGA8pat;yH&Yl@`-dsu`z%Yn$bSE>$Xr{<wU)G}u*YM?
zHp?1>5)!gbt7fQv(z(Mmt_%{&dHyzRX9mKlXrmW4wJzOyU-JjYM%=NK4aj8+9QLJ%
z8$%@nJ7L;QX>QOrG8Djqe67o(q(wkFqEm|g+JOxwfe^Z=JSPp15K};nLK}-3kM-i>
zbG83ap=})RQv{$(c1zLN|FzEB)~|zD4Gh-}^WU?Ou5+_KY0O;8rC8i?&8Lit{GGy)
z@EWXTQX)6ea6@B<63-IOp;4(kRjM>6nIE&7!Ik5>fw-o#T3Dp7B_6#HOt3Ecj5Auu
zv)A{2bEUj&z#>a-`61FYL<5H-7n@D^u-P#j3gLy2>y%7Y9V{;<T4@QxS*bUjDcdtF
zuF6Pj4>S)ki)+nX!7!DON?wUl47;-EeFjGuXkXSV&+M-O0^iVTgUh}N;~hmDNE+w_
zP%cDhgE--;WF`}z$7uhu8)eHE#z$(5u2htUy%#EDz`1n}QqaQ-FFsB*CsQMJbH%%N
zyyn(XOK<`<Icb_^c+kj5(|sF^q^zvig)A}_x+E-60Xdd25G(gfB7p?k%mi3O@xBz(
z+Lj{`2(yBXmlLQN<CKeC@>$<Fn}ohcV`>P5<N<&34QVSm?r17_LkdsBzX~F263JhY
zNN7Ni35}&-<mg;ne@0;nd_J1e6k5~AU*J$;!Z^lwd&S#6mOqwN58v9yGq+}Rh;Zbz
z<j_a&&*9`;xtY`FH7^JAoKy)9WmO&LSF*VyX)WC&ylz?Erfhp{<GC{(Hc4hJ!Ujlp
z6c6GC_!{!L4-&6wX0@xoQoKF;JP7<d-_HRG0_Li4q-m`|o+GUvR;U3nsIhGb%Id#*
z)6d09RP2)3p>^9Mb=V}9#Y9ML+63m)tI&g~Ro5D|h}s4g#fXdVFx5NBtYUFa8FXHx
znIDM{O1D5vmz29W89rxPWh@H$tt{G#*{O|BN@&-Jof4Ypcmyz~i926FbZSzujwu3o
zurSCD2t_r6o?e8-A;Sw?gYP&%^_NmOCew7C$yQ#|r)Gy?MJ!Lx_qNB=#*KAImA{Dw
zp@xG?Woc<dAn8`$={e}BN{!@yCD_|DbYQZ79XZ|_X;gWF_a^%@?cF_w=QMS~I~}z>
zfwDhDdwp3J!0%f7QRWegeB?KOYH9#nvCFt3TVG+K1}l!KkIZ7{s!6kZR@f=iF3CgK
z3Kv#r|3G}_3!F%|Zmf(^YSW4sGV21(zXuH&3`(pbB=frm!FZ&Ye6rNzNOLJ?Gr456
zZ9$b6vV@meH!RW5q<QCaHDyTYjoPT;zc+Dum2T|N06j62s=I=BCWt7dIpPEQ7I4#F
zC1QLGufRs54n%K3J=v|j%RNZ>Nq<exaWzs8xyu}r8r*BobN(R~yuqS9US_p&>rFrN
z!XP$Ww`~ntG)>=-1J636Jvmi0mZ0N$oA9|w6-}vqi#nAvxExV^F8Ky)iHI@Nv%!S4
z?P(Wm&f;hJadylhfoa7zFs}+QQLr)oj*O^DUxAukM5n9|C{XN_n=R-nuS!vp-a4Tq
z%RIur9ms&fBCpmQh${Iqs`1WE^;P4TUfos%RFc+hp&qRtKSLyPnYM5pyH?W$(bHp>
zM8;UmsVrmTyzzCzWVrw=iWJLl87>49r7PuEzzGg=Lfv7FRO=buPV6kOgWrC^Grf<?
zArIx6D8L21UR}J|KjRD8Uzjw=o61uV`63|~VVE(U)vWu#E^@4jF-a5#zD9w_zgHK-
z=}N9#)|2#1|EFiXZ^Y<gU$ee*J=u6Mo$hAryEas0)Mq7#sxRF9Iqmta$LzcFddK=&
zQUWLGuu}1J3`Zp9QW*kZ>+hJ+SqS1+%BOTsv%RW?<DSPC_@a#}7!N~T7`9gJRY6QJ
z&53on1ViHSgdB#f^Jv!Vb;$GQa9FQV4KDXNvx`4Dq;I1jJRyX5qu?_dG39?jc<OPd
zE=U$+{>kSY<2lSUL_S}CV}vgl(=di`=yk}!l#$8162K(bYYYmk#}#$9;F$Fj$rsc5
zxwZUN+ao0P%#`Fqz2?01W!TH7%Oo2N!zCgV7jV6;MAkS!FoC{9Z>?pFbZ$_heGViD
zXDg<)Q%cD}=Oy-b>r1?(l~4USAIrJkhfH<U_RFJ?$B*dF_yR8*#}uCkbOiM7)Q1A0
z$&$FKq9`EFT?&qDabbQi!948hrcg+>gkEcVUllUiC2IB0{X4&7t1sd>=+6rkskE+A
z5KKTvOye2_2c2ZtE4RpnFlae$Ts4hBvBK8K;q%{MY-Jm(Y&Yug4#zpJ$Mc~k6w#9`
zd8Eg!{o#3%#bq4{Huc_sf|kdpA6zmhUB&kmgD~Kb%yc1s>0l7@gm-YiFX3{hk*Fl{
z*s*lob@SBSBV1y|Ngve>R%#71Y;!Nstq}-dU+B|#K59$8P+bZj=k=phdPy9={QZSE
z<k8@^Cz2nD>ytFiihW6E%Z(k4Uke`AqQOyq{o`4rYA<bi;7ou#y8nmSSb;I|Z2vNO
z7;i3B=Z7<VphjHQa@&j8&*A8YxB=FrQSrB>SsG~ZKW3$E)6BI8bezX2pD$o6BnDtw
z19i$aO7J&~5@0VwcnyGlB{F)yu@_y&Di%@S_^5cxzDN0Kkhp7D;i`skgxwQuFa^Va
zcicRmdy*OSjcl!H`*>E39OT=nw9$D){9?i)qKN0IKZ0hu8d7lUIo@mQ<<#Lv@87QO
z<zJ+(j|^QXVVYqrJ?-B!6pT-9`Hul_7FNfkFR?K$S>cbZH&aRD(TC1$pR)oy1U+P7
zi0c@U*~Hsrzq2$^SaBT0{dJ-}Go$!;YLRuBZvgU?3L87#e9^O3;7GqOh!ydF*ed$R
zzSnPLb^4})TK;9X;x@6exHvcV_&)8@)3_iW^4IvR)Ua-Q4);O!`^SFwkP8tYTWF0q
z)ceY+LD#mqjZ*Vquqy$>{sB7))y4~xC?%(FBSZ<4yb_F*g%nv;Y;*-4LK3}4#(;1Q
zKeTkalOfPuv`Rgea)588Pw-)^Xl{kg9eWmaA!{2Z!{Iep6(~qLPhMDJl+jCzG`g=k
zE|lQYI4g=q6p17XEtpW{Le>VTc#7iOj>i1aE&K4YV;wFmDQqFWP7Sd1!=qVR#u2Z_
zbt_{t=~>umLvqeFdub`vVptx#uN>$eO|dli^4R-ozi#(IKiKs=T2wLHGEu=aaU`+B
z!1#GqV_T`>bFd|wII;)#P<%ZSQ|UfPQ3o5baHW;8D8-vOy!oLfU!@5!6@y>MV?sO}
zHgLl#NsZCJ`XXlfB8RKWgaOxXQaadNMK$YlSxt9|HBniV5vmY2zbu&sd9J_+agdyR
zD}v3rKz*Wujq$<WG|`3N27@U$xM75P>@eJFkbe$2*V*aE6ww7Y<5)k3R-2&~?6UpY
zveJoq^5QCJ*<r4Z+)@F`gNVKwN+$8(P3{smzux)jB=bqn17_XOg(X_9yCM*0e0aDf
z$e7+y)nD#auOQ_B6k!i&XD2a;YKcPSZSADqV}rLNDnrhE!MBjYMoV9^X_aN&#hpqp
z@^`iciZ;B~x6c(5TL#Z49YOKV{IG-f6{0Dh?&G2Ztdd$eD-e(axo4V_9R%rOVrohu
z1-n%~d#X5Z2r~Bx|D4dFD7xH@I&74Xu*aCi+%Un^sDdid+a^<}NplpYSt;&zjhbEx
z9jxrO^iC*lbM}RM&?X>0Y44~QM`{$Pc`@dr)2&tJqfa|1cgrsFuXLy`kISrI3fIJf
z0#zcI*p<O9!3Xfnjq-5uR$~kOSTpdA%r4sUGxv+OGxYYqot9HCjyLVk%@#es%`)+~
zi=q(dGC$4Y`rfIm4M$A<HU$=<<E)>(dZD|K6d2d#=w4#9UlJ^1%G#Y0*-8~4+#YZO
zdc*3N|Inq8{Jq(ElkuteaI}zv<f5C2c<+u3vFe0?5eAA|;Z+*l9o^~N?U}^5{Ac*6
z=Ti6rv;Vd)p@vmp&WrnTq}>zqFy8)2cd$V~#6pjW&Qlt~Ogr4Pt+|Yq;wmR?Fv={e
zpnG{uHm^xTg<wz<ZLR&2dq4j@xkRi>PkzF2(f8u+){n=E0R&`j5>g@ybL4P@7b4Yt
zUkWOpH4Bg_|7yrmES<4ma`G~#mwlW~S1h?ZUxLS#SP~K@l#2$dR3XE6D}k<u8bVcV
zRG^V86|;J1#59ZvYKqsfwgpQ;mnGY|bJ4GS9($Ye4OY9U*6>S3_EwO0dc6U`x?J2h
z;W3}G=`k?df0bsaw&+|R@8xs+6A{<Ux%yx^V4=ggFc{<n&SifE*NP-&RhfOTK5Ln~
z+No`EZD8q>9yb3&4_VLJJ)jZ3T{5_*@wN%`v9AMmdL!N@mg-Pft|OF~>scfNY0e;a
z#KV(cL)#iwBNKD=B>B6~o9!9+UF}Caujg4Uw>QV$tH#&YY1+2WUkUKbA^vfX!~ro@
zq=Ws>k8r5{-0^;oA<h;bMU<n?7;ElZ%hCODm9NgX4FEM`DzgB+^H%)YB|{%B<8_v1
z{;y{VUt)HmSgfupIz`9vO@~^wo;6-nQTwzJj~1pr*R_lIk}3yv1yhW3c#O!VS+2yO
z?yBSN2`PHi9v5$XiyS8J!{s|A2Ox0hlrdXFnNpc@C#h0Fn_@6ulDY00ooW9@_j)cg
zs-J7($Lz=ac|+F^PGZ<A*wo0&DAgYL_DV~|uP2p}0%R_#v#K#ZHz$=aFba{ni~i3p
zudlUxk^5H=Vtp0j(Ikwq>4xYIL5?C?A~^d{$mOICxV^%m(JB*l-Bd-9zmXJ$$Dk71
z>vm}yk%)^G{ms%;Yd|=F@7n676*^p)y37$i3zZ}-SM>CsYRAoY<_9qWvsh?C3HoA&
zZ^OwEUqqyInaDAW$gOWQ!l@_)t`8;@)@T=|xh-6`%A2OYo>NfPJvciHMpJ`$4;_HL
z65M6qQDh_%;mHsX9%ux<K=tPch=AEKZuB}p_UFVyhwREFF2!~L{iDp+t=!o@#Jg7x
zx01=<GYwQC6it)O%%xSJaqc6asp~ZjJELz5<s5~vuV<s&jl(=ERC%u!g7^K&eu`zN
z{r*h?ffm5Ed7KRD5UGBS7bg|{WuP&_i2G+N#%6@<{}U(jY<%Lx+-JZ@Zr4+lP@=%$
z^phv<n+G~QKF*~v?7?8brZ=>~VI-Dp*yqE&8$KbVh{@U|$HNXB1g^FYW7GV`Kemj;
zNu8krJ?#J7w^hX%OvVJ>_ZYOQqL%+}dxv|Y$E$4ataX9{BG~$r$uX5LD=D5W@!(m(
z?x+bc`{amwq|l<JO(m}fAbTulDOR#5VH|>48D+&DWH1qI2g!fy*CE)8y&9KRO@h++
zB{c*k{+GkEXUYiLBBUmfYO%v6oxgGJdzLUxZ#d>m^*UfQk@Rd((IP6K@?dW&7q8)P
z*&%xn_fDL~e(-X81ROPZgWUak7bWK<b_cpHe9!d8H`lS695lF-Ysdb($s@U>-sMnw
z>vG8NH}zfKY;BF(tL>&)Cs!c08EFN{B0>A0bfJ90e51{;&YcORbFfbK4J4a(k`T9w
zb1&WGzcily&)a(`jIGwkz?xQ%rvT;>h}W&CLu)Fh6+)MEH;H!X*IF>kSjiy|3w+&5
z&H<647EaPMheyK*vQKVI%BT1v>@@w`l(lBje*ndAf<Jb5#j%!F9+QK2K<jee-}(1Y
zqIs;Ria1?+z>EEouTbo^Z|G5<1564c2cP*0j>6qlCj=+r6+iKnA|7C6>U5B2vQH=M
zJVqR`GtJsxfcV=Vw^RzXPavH-hE(;X*&*x)I$KVbG$N7Y=kg6nCGX_x?kuhK-7-=C
z@PV%As?2$=tHW-z3n%gOwXHqeGR>-!U+ZS^o_wg@AX(1ia%@ULNGeUIlPx12vaw9B
zBwDhe@*4ZBi$W+x>A;!Be{jb^SS7qbFmUW$!o#@LtUEJ%1<Nk;k`R<!AL#!-#)ty@
zKNw@N-v4Bbc7^{B#t4wZu<5=jdR$AFFL8~){fE&sPlWA5p$06em0qWp+7)3HV&Rj9
znlD@n2z3WwtwrxQR3Q{{b?;p=zH!t~1jr3@f?}%C=4a;U-(H8&I5@}qSFJY5ch*8k
z@*3>7DyFP|E@1!oOwPaEE4grQ*%zGa`?C8hi+m^h0s&&Fv%uxk@xkS!Yfm=gl#W3{
zf&5{V86{0z>cM4RPSDJ3rl>+1wmZf#9;SCotqdNf<iIhEYKze2lsNU<*3D+Qv?wow
zXdF}dX{E?9s1#)Cq82qVpK$Z`g#>MS*t9pURUSv~vN~&}+}6dWuY>d+d6;Or?0K5S
z$JPu%d;r>ty&To`A_O{j`K&@wff#Cl6pe47a>1I*A3<ewU(q<x!i3J4C%Jk(%<Gj?
z6zsx@Z=a5C%%?oYvy%ZIGdBlv29aXulSi@f1f8|X+%qqH;v0)3T0^>JMUF`wp?))E
zl=7xME<I3Z=a$xt2k$`vW5R62p-g)ry)lh;0bm}jq_@||*la1!@Zu-@C%p8AAcyGL
z-_S+iLQeWp6qVcNpEzJ7RQ5ldr9H#U5f4bjc~WFT9vKqc(LU;?+NR>6dXEs5-dOQ2
z4SF^bs$vve6oWX#0W(+yOSXAF&UWJ;CIq^aD*R{Z)M*&NE2RYdm*=b$sb_ed(f{pF
z{ac^9P+N$vP1b)9FI@~PoG6QvS1R!Psg^g&fa$ef9aJVVATBRpXVi?$s7SjGKjc2u
zX1Bb0YgC`E*cg4O@g87gZIy?uy1en_-;Putqz|fs`=p~RCs4B8vN27D|2FCO;^DV#
z$$Uz5&ONtT6|z>bv*PdZTbJgPeRM=*vNQ6`LOxPB;!pXi$TFN18irk?Mu={~A7eGm
zTo}d)%|~N5X4sJkqd)WJozltPW2SyVmy}ZrN7M6Szgw>l6g>+~zEl@>g3(>1X<#7<
z#e(42{`p6c0z~Zw;;<A*r6HJTUQjsPO+*z75z{OjJ$Asb`$13_aew(3Fr;dVi=5P9
z;+8L9Sl71Dw73<`)wmNY+12;H8q7xIzLP7EJ+u$lS-aDcS5MuV6mlRm?c{48wQR$j
z2Pve1@R&iAX+=QwJh5H!#S^)fpEqZbJcun0Mao5=0v-a@!rnw9lU8k19%RzVS52Rr
zy1~ZG*A}5_JNrQ`=XO^)c<Pa`lS~5xeX9cXr=!sRokhfI93xL}5nv%*zB)E7m3`11
z4!Q0sE{_B1^{|aMWLG?ico17;9asaaW`A*$G*o8quLtT&Lt#$owqX#OASXE&pi*(}
z|2SbE0>rEVV<SHszWr=ZliO(Ys!2cis3)<x!4kxh49NT}CkA}T^5<T^2wTH`FY!j_
zd@1m#3KFG6i;~-+90z;yzKjCe>0>~dN+S(xMZ_5}0+t=TD^#op1}_bn<Kk;(Lxme8
zb>F6vZUv9^1`CFIDKv#yI7-Aa!Z<A)jEK4F06RmY7hm|Y(3;h$*coo6b(F91h#0>9
zehHna)}F+P3Im|}>6{=`MN>v<tysr!9UML__mQ($73eW}W{QhGM&0|v^fKNSes(xp
zK~nWZp;?e@H%P274zLGB;Dg^oPkU4FGB0BK>kZ&Jqj$grCgBzqIZ_Kxe}``4*-O*h
z0s6*EI)4k>WoNCu+=I7&oT=+jm9o#3yeO-GaT}5-UV&Z1YjUk@as@?mjb;%Sg-1?0
zHE^MUsh-2yv}4PRBC2f#-zU?dLM5NzFXsDO*L_%<T6yfm!Rlu5yx&Z&pdQ#vVB0Ox
z3}A0bRgk(CQ|`i#Be!<*@7}uI9>Q{g11Ct}LfFP?R%vNSh?+&e{n=DxN|PfCI2;G1
z+B|XP+=oQUE_`f2%{mutL`3>h$2l`AVRP3eoICV?@8W0YMWjt=KA^-SBRn1@O%Qlf
z3N2=TiydYe1j`^ENJ7Wj8<hF`k59MyBX9p5)e(D7USupGk7b`C7usPl?0fDT0H^(k
za|QmvNoIVgy`(J6`S;<Hym`J{rLw_d4<&9Lwf7d$%AW)`<|Ft{(k)l$cK}e+{^!Wx
zQo*4~uEc|W?#4+TfR)KfLAhW}EIty*w+24>jmV)N_NLOu7C*vw#UG~YU-4nz#7qvc
znU+*SCZ+vxb~`!K$aL(td?z=6Mi0qYbk3W^x0%Rn<BFw<D9oQklR>nBE$@x$VZ?Dy
zdk)VdKZhdcG29m}KTYiPjMCK7+x?aGQ)N5#^OR75JHzOfD#Qc>TFjp}Cs(*oy3WCp
z83)YY$W7z<(vw$#cdACdiJ4Y<M3-lr>bocINRH5`Mq|m9R<MomqiF&Fso$_lh?F6b
z9s6KWY*jOAlpwu5x2bF#cLpB(zF8ny-%h#}Yiro1PVqpoh{RKvr1l@<4wrJou7^mq
z>fVau)HR<?iNCrp2lR|MQb;hNjN?G*Fum$E=r>!N<FE4t{pso>jh_Qy){~-+2ATT9
zk@xpQinL*y74W^T!N<V2vGo5&8%RApjslS#q<UWMwfTI$ih9SA2i<5I51wk57ICm-
zk>CJk*xuDrmz`Hfo7s)~dw*ZMMHXf6&l}B@`*KY!x1oS+vu`wrxAnx%Lw_9hcHuL)
zNsnqImFltNaG(~lPB=&>95vK5^bAfAaa_7a(@_D?l{nsC|8;;wPTbv*0KF!|0w0rR
zk{nVKs=chXY8h6i?7g~MT+PIrh&Ahb#>qE4lw(b0OFz%}tAX!%KMS~Z-G6ANT6)qZ
zQAEP<@)eM;GL51(;a>i2d}Q7D6xBO`lHrU^Fm5<OpS6QEcW|h%Mz$%?+zj3fKqr=V
zt(jcNh;*0Z_3Cxo-d(y6inK=q#Y%BBBbpAp{H5Z$rc%uWewHaOF0|q!3`y3n>6d)j
z3R`!O>qM(7uFjU8^!qFl3lx@I7+JaY#vPo@)H8pJ8*{Om#H@++cpb=9{P-0D=7|+~
zI+&#k_sMldId`E0xqsnQbm5d95Zte$YFxOboX6I40e<UMrdsh+OVH*L2&r%Kri!i>
zEF!C&{cv5!k-)o?wE3&tne6skNd1d@aEmroQ&88;vNOm)y`*!1-HC_%j&-5}mAl{@
zC+6QP+iu>o&AYI2?$$Q3%k>17UuQ95>T@294|*~nQ6{rDj>};<zeaQS0b%!wWbq$9
zlb71-pC-ldx8<A3{otIE?ZI<-^$rh_09-}B>hKsh7Tg5U>f;(jc3kAr`)F?ao_tiM
z!{AZ{YC>Hm7MIvyo|hiG=bkn%pVa~Vk|28x1}E?w-zL5hk7g4diR@?ebs9^SM2P-r
z6(U>gfLwUf&YAW%BQ<UaAiIa$Y>r%LrZJEk{2|QjAa<d>tMlTXA<5M2_o07Ak-6ED
zR)-swJcc3+bV)C9mtOd@;k7$MQ_|1Vv(NKVT}IVwnq^aXkWZ<sa;#T1hy*G#X>{KH
z_t0!N2T5=K+T^L-4eX)rBTV?zO879?ZzHfb4};P1anjnM^NtWyK-TLB!X$~7o+mAJ
zL+Rn~KQc=cgAw&?KkYR42AxDtZdn>45kxP{;COGWC=dVS@x$sbB}%;^DGE@u`OGm7
z^b|>~6%nSApwt$6X{E^(an_@^5<1c|-sn50a3C@K=wW*uJu_GP@EsooyH(_898~Q(
zHsg3$dkceoq)ln72CUQ?J>{`1qi|MERJ8HJo}cm^E+y3c@O>2-lUXHTetyAigs!To
z(Q!;{4Wx1brT=>TQ?Yu{m=}Um%SuSe^!T{kWf1mL_V7x(K`avz{^1^IQ};K8Wh&@k
z=s_&FET)V(Xy9%uf!hv(<nHMTeMCuCrw&{ysCV+q{YyBdJfQy&F_1o)26Ns|FQD^a
zUt*BT(C)Sw^ZRUD(LTJ!yWg-Dki5b7O`iE8`L6FBl*MPeE2jV}PFih0?q96`>Vyd*
zn&SR5MsrHJ3pfEzj%k1s*1}qhwak;snf2SRBF;ZWGsjIvyJd+PBKj;<wH%`MAv-XQ
zHCw>lNVXLz0O>_!?L{8EHOH_7pEARjBAyIFSbjt0r5QBs)Hi>++xB;O<DAax#oNOu
zT+m>IIRs00%ST>6gv97k1AP5T&OVW%4o=b(!^hrXlz{Vo)Vz_SexFF!*XkdNbg=hs
zB3+PS4M>p%*l3E)dL>F#D#p&9B+$x2o<q7EQ=_o=fEpBlj&0or;_>(4%xXM%M?-yp
zLUgu9!rvo+F@RB&Qijd1Suj)J^C^rcCztS0DkhI4m*|q~*hiuEirkg<XY6%%cOF}l
zWoeN(aCpSaFm`hM5MYS4?Vouwz7>EIgmXvlhrLMY*)yjO#lUO-bw)_dpuQ2j6URK?
zI{zDBi8~2Fhb=$QqxXEAr@1>~1W|Xgu5Go4Px5O2W?sfx;hXvu%a7&iV<2*pl0QFJ
zI{Anv`?0x7Q$tC_(s#$zC*$cC?#YfFWQ`L$!Iv3xxAv-)i_JjhoK&@5ju?rlshW8u
zhpX*_lCFt#A1j1?-!h48H>ns1z;hd`X*3pC<#*rZpoqJb5hyWYlh9qpY4~^_+#Q+E
zX`0<p8zaAeP2nD_Aw^isg_GKo?QHC<04zKsGM0AUdP}83NfZftVf)jKsOVoi73%cE
z8H_D5c^WkoYn3Mx9FJCeD8!A;+~`&%{eP1-S{r`lp<-;`$#ABUe8u>)pxt+h4U7WI
z4u-bL)MQMl2Q<gY9Fo3KJpO^s6qiS@+J$aE9@D-rZMnmRpnGdv_oFd}=>JfofmG;b
z`q5uv6_*EfUs<4SypA<Z*O_ozse)?lSmJ?N;$u2)=3}33+@A2~?oX0}O{f0)ykS~Y
z(4^a{D)AeGd^mr%MDC=?yc6z<|7i^H!D`7qvQMUTi_Si-(aiZPiB1(8)qQg`Pbrks
zDs|)-ktzS2u((FPn}@nMPWwuq^drlPKeloBPe-$aIph+_XIIGT$k<tl-{@7OISq;)
zkIwQm9aPIR$-<xfp(Ua08hx@lPO5g3x078|iG4a$Xp*-f1L{Po{lh4RHINfXVPas{
zXttTAd+<EteQ&)!QS|FOiFFHV6J<QelEA9_Y5D7Vh=~QRKCB4#h>w&2@xlboqF+H}
zl5mD%fSE$ZG=&n<xZ&ERL9A`|)XDux<!#gOs(qjzMzP^bHPlqa&p{1~Im_X7a^68X
zb+{(gFmkHOoTd~QgfA4LJ3jP4x&B6?2r7R7TFJNaw_n=U=7)(ZW@<1Kx;^bC5e54c
zRNs_@oFHSV8(JuddoKNteB|*+Q@+-tawJ~0y2f|Vzx{5AxNNI9Zf<mWOEwKFp@O97
zpH289;>PiMf3-L(fUgq?E{!&dLias!vGG0UZgu9P#_-xxtkQPx`}$W8AYq!j|A{MQ
zvQ}7|&DN=m%Wxn*&q9CcdDO#%GDfOvt{mFbJtxB;_`;<XV6*kOXbIcPL93W9Gi`24
z7%!T!Z(u+Z2Ks^tu@l%0lvOwS>*6cdA-o2rZ^3EKRF*6ZTMXN4YkX2$*iel5-Ay+Z
z!7#SVp9~`&v<p*3a6kQgb#Unc{KPoOQY&Lw;c6>#L>2KSck7r&2=UyRp`DHO%lqLI
z`(ua4BP4|BgITA}o58QUo`ykf#|c&HnH~OdjU|;qf)qYFK0k9K<3=n?A;>v|$Y<cS
z3*ZR9yoiVM;i-X+5Rlx;`}-UvlzkO~$^Eq0ra5W&HxW2U=~@Q??>HtVg+JJ+zit+M
zBD>B_i1-#}^7ccX$dazjewW>XlkDe`KU~*Ya$@_grp>!w7k@1{>d#5qQ~ygc4~K3M
z9Bm!kdq~GCYO?}9_}i<vtHePmVNbK-^WsVq`W<=2N7d*y)n82Q77*7>w_s8uViH3R
zYxQwS&R*p>h9?rk0OA0rN1tB(7e}HI$jSPU3-ML4xQ0V~^<f@pp4GXju^gGB2C77x
z4~L<AE|GB$kvKgOU*yqDGoOA`0`vutf)<<FFsI#H`oLi~Qv9sEF}DtK7h^6cR(kfS
zRV^L}f`ZZd?rLnrg%?c4=|a289x-g5IUBHm>G}_hmu|{3V5>uW;G*<&qoKlmCrAC2
zz~);x0s|5o{;l@q!n&!syFcpJTm&nhB)axZl;?e9bOQCFtuAJ8x3A6&I8Bz8#a_=p
z6sIrdCg7w^xBe*B4Tn^&KLvb_B0wdYxgj6g5G&!;{nm%48-cD9B#ml<wt2rB1TzM5
zjYLCiNz9ce2J|E3(b@=)GQqLr{={GO&eFuPuA#issi0u*`-ucCm2q02CeDg}%jCX2
zD?deXJ!-MTO;C0{5qDrht2UkMZPOL4<)pKx^DDeYy;EPkze`IxKnOG>`Mclt_XrfU
zxn7V+Wuky}mkF&gdDAwnYA5s%+GVal3~6??+&&`;4*<tZ!23I%wS>N*xjA1vhho`H
zJ}Qk;3VR`{x@wZogv#RTYi{-h_-rJt?!D|H%L_Vb`Oi+m#&8j*gwTwsG>=E>2043I
zjuE`ZvCvMhb~epr1grAI!y$JI!WcWqcN0a3c@KvZ*3||`jXVQXdA?D^R#T&A8gF9=
zU2kklYCtg!##17rT;`o<)vMR3ZOZ9a{vyh}6~EH{kX)7>m)k;<yZ@{9J+3{~%Y|hl
zfjU8sy+8d1R|j#z^pTPD=NUiM9L^EKW^DE4e|^t>Mx;i4H`#>srsRM6_7Y8DV|9B$
zmR~;uu6-6H7l?&IOGhB3=psbtdWia910C%H5MZ{_dD<Bd^D$_|=9JuDCIqk2t4~@W
zU4%^Hq=!MAUX|rLKd$Ut!Yx{=Qdg@XT|em?cJAxJ?BYSCp&sW1+|{mc37qB28@=Z3
zm9Y4*?pz(@NmbxwTl(>03gj8h>Ez?QAW+1CgX$**OL6`_Kt=*$2fDr2?c==I--*ru
z1iKo=S2xm_TCbwn6c+sGE==VfVojN*w)1}YZ4(A?-*G{kI{u!Np(H=dT7KQjz2hDP
zkZF!am+`js+AL)!Z7m%dCE+ttCn(!&u2#bMZqr-+Y6rEGsAteZ^N0jf%6aD{inHXS
zSh_VPs)NnbG8YOWcd{6<IfExn3$<<lEk4v1Ckx}b`Yq&FoPF07L8q*wxh_(aW5l^3
z{o{nAqC;eLPMg7%0S6mdv^J%CrfWNdPARoM$1pXe@Mw2&KzC~x!R|0_U!E|tqqx<k
zp5DVtTCJH`-hx=)?haL1V217^r}3H>GW>6vD(ooZJt)nmuaEs+CPEm7yW>efUIw%L
z$8@LXogp)`3r&pG=xE991xJmI>Md+-Vv#O>^6uBZ<GW*}fVauxS#GxL+4VhfR3COA
z|KSr_Vs}c%KJ9M8?x~Dlp87)nR}`Om>eD6QD~DPPjJhRI`#8nwGhs1=Bh{NQ86PU4
zD<l2`u@Pc}o?$NpM7Q2*sF=u(?;dCTkdH*|-}Yz_p(#<JRS&j{A(5)UnVZD&T38=D
z<_9o-`RuCK!JLq$@dx}+8G471N#%#xF8C*QkN14_aSDj|n_1;jd>)9dfh?T553%iZ
z>+iFGDC4MljVfi(^}f3-C-G~h?!$op=%~3;bMlDSu$T_8GbmEF9Za2z4C+Jm>(xXP
zGKEhB&75Pqb9UJ~9Yd#xph;(oJ>s!dvA%CDMIRE7U2(hNq^`ciC`&2#|GYzV%AO1T
zn^704Nzi+3vCTGZrK!|^3*OCH-j2*z*6n`S+hqSUdjBhV1CwcEH~)^_Na_DXZ;hL?
zm;Xof=8dBiq{B`4`HK!L(eH9x7aUD>YnIK7Q~v&=;BP@*?OVZe_%AIB>%uA_`*BR9
z$yto9>ZO?f6};KhK7;qaa(AaUAQS!HVs}!mWEh>aU(b>e?dJ1y)!{fMCpZ*dG2YzO
z+{blpx^N;B_Qbw4t9ZJF3wv1c_Y1x!4PpmhkG-Tp#(DOJA<L4b%4`tp;QG)GzKyye
zKW6DRRR76~>-T;JgNq+Xm46@Yl3jBG8WikX+EDGctzP9U0jd<Ok~M8P%Nuhdp~J0z
zh4MoNVt&yfO*_9B_~O+cBu2W(c?O~##2`zk!O+1}LWV|Q1^?u6flT86OysUbxFrhJ
zu9O>-RMzaX8&08Unfs#aeJgBmlK?SFI8pd+jgdz{7mWLx%-{?$Zko0$Qc4<)-LGvR
zt35;f?w{k(P7G<Cuczo<&guR%ZB6PMEuoUUx-pIOnG3Nr0`h2~1Th`_^+wENh~>v9
zC7gMiYnIzwLm7ouD#3D`HT5|5Wk%){nNVEjIc!<f-`J!b#+bC-CCD4-xWRHy1PEGM
z<@4n0E2t?%%B&kiJ5=PdbM2i()1rW@2<U?bXVB(Yf=Kx(LuR5N|0hjEwc>r3u$<R(
zR|^tWa{CEGxG`JA!_*~6J<JEw6m<Hhx8HYB(cH2((ngmEYWxLVe}Cq0U^{&!{z4ot
ziEO}3n7Z;e;^{E7O=aoB=sEQ;Kt!Z5qzP>I%kD_Ag9ahiKhJsKO|n?812C?^RU#JD
zT_`}H;dr!>HbCj{?>c~E{ZAcW_^bnDx;bU0%PKt$lJ8tI9d~{I^Z`~4l;nXd5A(MD
z{h3=7=IUO{6Iok&@j`(>%uuGmxtguxkid7P{NWtS+X<uC*!i=h|Lg;o$5Nktz?!+Q
zy+g?WaOq_}R!V%hybIilp~Fp_-GZNjE~K-|vi<tAs2HxHd`(fn7_V#y^hU1R1IL*g
z6q>&(rP3&d8~TyXBOB#kg+KUvw3mDty|*|8$EoU_yC;|Q*;6|f<z|~(96MCY;uYPc
z;9*yx9imElRkLeInh5VEqLkyrtxnS<$pYY(&uGaMw^-arQZn!vr0}RhYwW{F782vb
zYX)cYAGzC43+02d<k`WFv2G1(w^LUr8_;T$2hM&EI?<(;iK_{GE0~qt5i@+@1mZw!
z);Czw$5Pem45_t~y>7oKuM&!Bp||(^N9d@P-vK*vq+Sc(C7Y#x0jW?^%(Dj8Qg`Z<
zYGCeB#5=RzuvZ)v61F=|Do9w0<qYmJ&kD=1zA`)bwi<?9xTIa>73{$?GYB{Rh)T7p
zND*I6M#stW4&x*|SVeKlojQPCk(=vw_9=|Fe-dxLQ+QgXgf0sye2D#SM!(feb7PgF
zProdWs`{)T{|NCws29ZB^8yZJdE83+_0XY?>#yRt^FU!zuQQKf(|>HN>uCi8wBf#d
zFyd6)9$#64f)nJr`KKm>OrYcVSbXmhboTsU+#_~L*N<b3tgi&R*d(u31sf~|H5@VN
zuh#YFxvH|8h`+^nm;0S?U>A6+*+Ip#eok@{R@p_A*lMSsS_+w8X@>#!dMrH&zP!sW
z{{0*Z*Z(uT<}ZU6<Bbf7*iv)7q`7qNTethqE&Eft4P4guGZ{R0l$OK_ZIThy`Wqy)
zCgP2Qc#i(sOYzlGP)Q!tq;gMK>_@&S(q+^6(*1RJ3(VNM8m^lW|EzFABfl``s`q2H
zYk?MhqTi*vftZ67e?UZml`zw=aL<pTyN;EFwJz%$m?rxgDE2q%Ch-HHKY;^lR=Gy|
zFZnm63-_(0s3|<(8?A4J<6E^#dB4sw9>L+=MFy^Q6L-G1yCdK7;rhd;EYEVu#t%v=
z{d$||*QEpzT{@ksppu;Hm57*hLJ)RPl0<^*-DMa@-As=5lIGwA_OEAnx0PO-KfaYn
zNk$*e@J>+`T-+dN9)O@_b6c`+Z=={XZ0V0_cyBYIkuxs6GRg|<Vpc7Vfk03U)gh!+
z@{;58^6JlgTOabU+l@(9FUi!99&j0yeI8TH+4t_pod@H|^YcTAjCr&+_iPv+>@@@1
zrAGcpqbSyMWP0!e$g@(Jy;29pvX+ZC3FFejXQQ)BY1eVLM^h`KBXnIS;7!wu%p!a~
zknDB@C3FdjT~s6pS~j{Rcw9M$SNEUO<c5p`?!#Rk&kBD1(}LrFm#5qbCVqEhuKlrc
zmSkgW!e5Cor-Db45xh>SMx_4RC<^CrMM7s&KyNMRvj_v9O}HHJb{%|R&Mu5k!=cq~
zKE`<ExzF|6aKEGEH*^taXWi(Z9%WBpBn-qAu2QbfbH&6YV#)+}7fY7%4zuuca2B$4
zmvdl_=z-R8>PyBisK!Elv$`|<Ydwgs-X7>Nrc7ib@dt|?JlI24&K8$7sb}6&QO~iC
zftzkDixPm#CxAS0`TfH=x+3TaoZA!E+RuEolDCB^2ao;-ctnRcGkOJ)Yflz5t5cyy
z)$(4ZK-x}dw~%7*+tsW7jKi#w3H&_$mzQxFW}7S#kKlLNd+eVEsZ9nm(9qr!l)(+6
zzZlDkx=JgjXmvnNoo2JW#Az6v!=+BTMLzO^>^%TN_8)q(PvD_Ws+447EE|l{CVWgT
zP;~B4_0BLgn^Wo5^(3=J(0=H(b|pk@-(O@CUqmbh{2-wZ9-BS5H%_Aiv6iX50MC1l
zKx6JGf0~fNoRlaibql`;1qz`*upS9SP7uEItD%_I7+dZd@0L}xO=laUtnS>Nj$a5E
zeTD!;yAm~9iLx7wpBj`#z4WfzfY@WEVY-~&GPJQYjo}6L>4pNfQTN{{o-=k&`mz|S
zYWm+`mQP<U5`pDVn$h)a#fX^X0jEQnev>7%C~H=aG9~u-*qdOpdDST5j3Q+44{Lhs
zwxDF(A*o9|H4`nD2lygrqK&!)NeNY16f$7Ktn?-tpXvG8&aP2YII`NejBD91Tbs6_
zE`GPR8s6nQp(9j8KMNA3=3d=0fn?fTk2K-O5{_?Y8(zU&eF=i55}9BWd|Q9<&9Rib
z;1#2iJEBvVW<sqCX<926F67yH`1XFQ|Hbg5;XOFvSq8?v$!>@(s6XaT44mFQUm1YY
zd>ih^A>OxTj+vK&bFU`SINiXpKch?Fw_;SdwQY=g-ptLl!t}>k_uV?yxk1&ZFY@xT
z@j{y)QKr9k4eBT^r7CZ;|Ll<NHKtQq`~z0vOZnssOGJ`1OlrDn?~WQo1KBRob4^}+
zA`N={_dKmXmQw@F!|v3C@aKpRB*Oqjfcdw{D4v%Lj+UKr|4g?h!z3{TzeyTe7$5W}
zn5W2_ly!I3jA<6t1X~{H-84}Gy*XZ0hZiLu#Fs*i2F-8vB`}4~ucwvE7~)F9Rh~#4
zB$obmVRt)t6vN3R*KmB#JO+Xs%U=T$0S&(J^u8w4*+$4cHQ8J$fp<$dy+t7T?}ocb
z6^q=hsFD5On1PSU6OviUST>{2d29!F^DU2p`ny-;xOXXXb*x4-pB)k@hWVM4e<wBK
zu=XNGVf1I&lK>g&g-fT@W`h*D2ysU|Zk|4z$M%@S46a!2{_F0>ct6Ce6_Sm;u0*U2
z$sN%Tu^n}#zfN}vYy=l|pj$wRO=b0kfS(FyQjs7lqnK!pUJDJp#UB}W2WMZE+w49Y
zYzbC_4*5#Ul!m(`5bTncY9d`myH2wa9MOtlBKzbYKF=9p+{!2RrQz!OB5x-%GiiKD
zNK6Bdp&7~*Y^a&>7c%Y*%)8KAhx1bKB21kAL0#}`wtq63)^hE+L(u>cX4Mg2wQiqP
z6@4ka<0l{J^p}*)3jM4C^Z}&~`dhF|!vz_qaw>=j=GiVwnN^8U6JKzL<s{Y&X!j%G
zUCGkjkX~5Juw|yUO{pR+E07~YPC<9yCqwXizT=0{Y1U8KG94zjv(SjjjeOK-FP)p*
zb70yjp_6A=ZI&(aDgg+V6M44aJ=_?@O*LG*QiyH3g7XVU&#aoB;`Fj(TclXmhZ&*H
zFAEMU3P_RfNP+a=3`{dM!Bn<O-@Cu{QK5nC>a1M}dc>Td{iNTBkb}GLP0~%3SP5dI
zg^!>6HR>`#$1o-nG+Jg4s&lOc#H7ft?5cm)y<k!Sxs8399iUZJMeKGen|Wx0<%)&J
zH7BnL%s7BKwimDYrcu|0lSn&~L1gj|-@tSg*yk>Q!sF4|*%3R}j}l*JS7xljZ9XZm
z2z(Rf*3*sVl34CB(^Z?IrOS~n`Oa?=M#tMH^8(-IFjDmhX5~%uM7O-@Vj;S=6`ifO
z!8n*9i-}!|VS)TauFt94Ox-`43NEsA{d<^<^;WZ{p1#V{&hx#zjM{Y6L7aeJxq=$x
z9%9*4M35LyWp_k<Yek>BuHV9xY;h7$BGvTC!OOjTepM>bIUP_XCxA56qmarM@KakZ
z{hZIM={98<wiONcXoe)oY&tmfu^^TY65&*%IoIXCfd#~lG@LLl+j!Ps%aSBvx)E-7
z3{(hiF2<WddIZ)ohpuLLxM~O`383DGmIKT}V56H930%q7?5Z%o<Gx!dJ-ash3dJJn
zNE_N~#c-GpiByhmt88TX^-t)UU?Tru6rz$k?3sij8_6cM5q#X}3KXPxGw3v&8Ep|)
z^)$KaeE=LOR!@$#N)nvWBo7)+LPxNCuxn@n6j`VWA?I>lOSpT__gBunSvzO``1zy@
zAMYcFyL9OwkIB4(9!&{D)FARl8%$iTZmciQsf_DYMeKy7UT*p>+EtUMas-B5wmC*g
zgui<H5+*QXi@%*8$Z<0876_HxvjWwCvf)wOPXV;^y!mJG0@5HjoqQktIIntw+t|D{
z-2HY~>v`GlvyG=8ZP`&uZZn=5vhFb;@;#h`{+dR4za_!1v<k6nv>Qe=!}MH53dfHv
zGMD~-)N1)}eAj8zm-sRD|HivS19eF?<@>IF-|?}fij-;r7Av7aU2}4*UcM9V$LZ36
zmmxBk_k-6qK61agblaRTzYt#{ZH(MG-upqbcAKoqKxn1P_wZ=!Hi@b$wXI9KpaG6a
zU2yjT|17~B!w4fi;_lcTrVqba1-rFjn%^bbucDb?S`3%6G^5s1&wDhxErpKyaLW0e
zMdDRV0GA}tDoQtML4RgDvGSEYRo9vHp(ZJWqYbZ_{fA0lfR>T=*|kC|zr_YgRBNk6
zp&XN^i*U=#rn|_(dY7Rbw8|a3X!Gx=X$=`MKs=Phcd5pZShdAQQ=O6KdlLbZSK+;e
z3y6uVF(y_H=vc-OJIK3k?iAk)+M!P+tfv=FmuWp<_kGu{y`^C7XS(a@=G~w8*lx}i
zzM}KKR0i5c2HNZAki&e2fo3J2CLiv4wJNpih4ddoly?9%z81`qt&LOV<v+DlsDE&@
z(Z{$lxX-CVuQZPN`AV-`&GLCxdEP78eC30$KU_?8{}|<&xf!Pk6|~h*j`p5k#AsZo
zO9h~=1LWfc57KFgy9O;#G=B}3Mz9hG$REhPf=`;f-n^N-NZ+SrihUpl>rZ6^R?|=5
zn^l>iqwJe`gyU>6wNuj<k}I_*@<C%@5mBS&<@Dh4veb9Zvdt^gP=8DE^2#4Sw?H<S
z4(+)zE2cF%h`T>Z<s@)N3r_}fmV6@pr~qVSjcy%c06_LhGcRQl>`orNJoG<DA4u(e
z7J4vUw|fa+3=`)*2^b;y+S!JGI~g;%J;C>9p=(&)qUW_AUy70r==1*Dr)&a!^t8%$
ztU`jLj~$Cx%A*+3%BWVjY%_)(S#TZ<O=X>vmwrI;5+J6`oLqYn;Rj*3A0cIz0#svr
zrc^k*s2y8l+O+M=ntrH+C}q^;{aL+E%~(3jws-N&6spjoT{e3N6Bn~>)S*w@KIzB4
zBOh`-Q}OFTT}@3Gy@fHecaplUQ_b3JyyLjLv9Bq{5pNZ{7%w@+%gxT0FTI&L&(3=j
z6DeI`N`6kpza&FSz?pNOOM8ox1KQL~f|c(DRAS_E$Dd8y8fTECV&*L0lc6^DdQUhb
z#Tu)OZ0Fd>SK6QYU-++mF;8lv#7(Sv6~AJ;l=T?9@tBp2Uf8O2nfb2<Xxb9*MHBhT
zp~*FblznmAGx)pevT57%O8NU^?aqj_=N0Sshy484%%KrI!W`O@_%)etE`Xzhr#Icb
zxbR;d9;EA>_=nWK>J?5JRngcfZv<)Jg~Kpr%&u&yJo&jGF_pM6=jWt}DOe}@@MZ*B
z5qkzcLHw5Jj8cAy$Ib`XWy-Li6sqB(AF00c?L73DC%4h3i>B0m`_<+4&z$RLuAZM#
zeL?v?lJ!x4x0f7V!!@rG0?<jijhFV~m~Nd2eA`5L@uvPrRJoQyzV8p%<LBYOsC;@h
zbZCdaT*Q#sIFRbCU@Eax(ah_lfSHkQ7**DC$hY23cxizAwX)-voOg1Cx@233%lh*1
zIBvf|FPx)RG6!=3_1N7vFxZ7gPM=Sew4D66{c^Ubh%e$2h)kRN4J`V5JA+7F{vXEP
zF*@=F-1dywv7K~m+qP{d9owkbwr$(!*tTsu=_H-W`<`>=-h1ZG+*!3&)rYK=k5&J@
zpS^$E!0zoy2Z*w>s9k{l5$|~w#J^TjsCX0n?nLLKv-R`2%x!^yi`i(SmDs|22`9Ry
z=+5A>JFSXwwBY7Y3YMKFfp;fdZJat*<1J%@2qsGRz`tBt>wxhi`pjA<r;4eb;Mlf(
z8EHQQ0uR!rlYym()%{ra%<b+EU}1@lkITCLE~=3^W(n@nNy>IYCpls<j*G{lB0%78
z${3)zU(;V-eV;|U4(#3Y4P+ionD|m-x)1@E+gBuGT%1_N#{rAoRuI-Al?S5zKyxwW
znAAFiikX!#iy3DQ*;<a!g$G7OH05DWT(EE0U5k<Djq>r8NiJHepNKXHz%|wnn-5QJ
zVnVa18DsehUMw&I$+7WX^VyRiMr^_GLBmq14<j+q5-l6~q4pQqb62&*NEJ$<{fB}e
z*xSpLdjQ-#j1CM=5t@4<g#@+9&+1zQ32Ff?e_~-^3vjtnzb24S#^{8LNN^Dlj#0`Y
zeCbh26ZyY{r9hm)DoD_r0Ps;t0XZU}(-@-CTjoPl!qVUI`F>d-<R$`g{*h^7W7%Q-
zlEB$vrq6c4pW3`*!doEbt3okEu=h}$A&>lAhnViFacBy0-i8n`Ou{htpcj8aSqN4v
z_@3>+6MT5c-t*|8#cZmK;0xp6rv~xLUA#QjF7(MJCga-W(9K4508T;>lPG&M6`>wK
zv;5Jq2|N{51YCgs)M06qYuT!jZVw?>2NU)Aa^INr2)~3#E&yNV0kQPCJwTp^bghHm
zOE$Y9ef*3;<#-j{#lQ=~KSlP#`%p22=6+pE<X_F(AEoAnKx?ld{8Z_m)|N_RqyP(%
zWxp0EI@u9Lzv{n61MuVS;X3XT%Haj62VRXf%~oW)oNqe#*=ZhZ_ByYGB2WpT1&>gQ
zj#HlAWPI+O1c9VuN_3@-r9?|>OL@D4UJJFP<qFa^{NE<J49)5w6mN8f5>*PdQ0p*K
zF-ywX|D0awxuw<scTSkd6<Ct~_Rs<UI}@Eu&j{SrK<j_?&{ud$oA^FVt)BK+tN2%A
z+uN)Em*`5p>C%24Lw@%w2fmNPsnIeaWnMXBAIrvnndm|<&=eg1DIeuDj4pALYux=Q
z3;-7gfR>G2<$R)LZb7!yFRAeer2?%I6KgvW_jPwJW(gU)&UOs;K%{m0dJxTuaM%<9
z3x+nVJUs%Wqq7OVc+AE(N5UsV3_`m1R!1>vednu=mt8!J7*Ap6EgX2SPw9x4)50JK
ziSw^@oXyAD;=ljRK?nRFEp!XIMA9#|@GcUVQdNsST3g9_C&m;P>7wBj4$X>~)r#YI
zn>_BQLvFYQf}=4o&F>U;LrBSL!y%+~{K}Ob=Eb3pT|E6LL(!Jkr@kNi#H|nrG_u4U
z4;M*X8zCezK!TDO23bNAL6^|qnFX4}m<#8dO)3<Cn(z)&6gMA8T@+W4^d+J!(X&b<
za7uU=*WHD@SWX)5;k^`rv|Q2tG^CV1@?jiBgVe^ya~E#SH(9FDx-*Vlnygjn2vLPJ
zjD7vbegL~u?(skeh^?{r1)<k3->%|wl<Wf{S);2TfJO10lF8)0d#-LV`UP>M-V*dr
zN`~5Fq8k$EFONvsW+&_jr$_#W=waz7$HY|KlU;31%W2p@F&T}2Vlq!)h9HLP;%O}3
z7sf&1kwMNck)MzMp$>_R%nr<P4u5>VXkZG5z@X(L7Tr0O17Z}*mWTEK!>R{t_M<p|
zr)Ca@WTNVAFf+i7pp4p5G4%}Wyg__RjD8`-OjLKoj8CAsk?)SATWngmXZ)Y-dguRc
z*L#=_@6%brwy>+U<trj_9W9Ww5y>^qfMquV80j+CP5-s)Hy8n|@iti5b`xAF=U0w8
zD9c8Ktb1qf4hnT_jGU2^p{U4Z3-#QasF1awXoF_n#zXSuu70bteiK;auGh6Qg-#@l
z9>(y-_8QOj8uNdCUqE=T!9lYhZ{9vKu3dr3ibac~U@=Ppv)V~rACi0af6P4Q0stJP
zlMB8C_X&uJ!;b7j-2Zs-$ixKr4##IHkJ`Ka4s!TM>3%J3)^Qy*ucn{x(h=11ksRWI
z+LW1#wV>V7cCwdJ`wAma{t50VH#94;1X_AVC9L1zL=_eQ2@Hw5A*M3T6_~dkg@)&3
zO#VM+JYJF9fxvLOIcP3!bov^gnVST-KXG#g?gCBHe-3;mOImWh)eelRuhm{cjJl7L
zJdAD9x&=*PGlyxkfXWs!+9eAcc98pX-)8a(isYk8ts;W;nq;$e*FmbSlh0d0kkhZw
z!V=i)ewwA3_8clgE1`sN#?^EMq31|P&7fjw*;Gp@>daRm!Z@W=>zj{n&<q+ZIbm4^
zKU;Y=6*V`?)v&*NpE7+aw9-0&WGN}Zb1TylcH4E1<haiM8NYh5&~CC1r7QWfT}^Cd
z|JSUQ^Z?jrV416G0Rp#ynx<A?h^DFHIVYxD?!U%BrvkXyZmD#s*pHiT!($cybI^Qf
z$+#pl*k_8xgJLpzF#yry11*b8NeIiYG#zrZMaz9u@4r<r`aJ?0B>X?jD`wV&A`gbg
z48*)R3B>-r;AmIo&saV=@|OJKLLYT*FN=drQsp&X&DZVFZ%eET(sm;vPowG5$=y4~
zDMwxQ8Ad>dzOhW5cv=uKy;{A$wpUC1sv<j+GQtTu6jdbzsL#q`JO>)`r!{`=^5t93
zT5)Ig_LBdyDS$~&;Ektppa$*is+)Lsy&9UKqZg4GTYvD$gvvMgN@MuP3Q*5B*rLlA
zX5aaB)NBH>n<Kf2CJee63-gj%kj;d8Rm6$~Bg*{_m7i;u|BIaU2cFFOz2HsLw5XpD
zhFskHM>;y7mk>q^OF)_cT+RjgQt!r4_!XEFv~wjBO}w2XOo+5O!Z@2E5=?0*oa+cZ
zrEVRc9cj8&DCj((dLk(bJ_<^{jP6z)gX6tkL8lI<IGu`alx_lm3P_&){?-lIZw}((
z6?&+1Tlfw>A~8F}j8-u}rE$RLr?(ZNYWapa9?$~5!YoB#q19mIm`HO)*oOE}6jnZ<
z7Wsf-A4dC$paNaTgIw1&Ifo=vL<R-Gccv#qS3)ww;Hh2<7B>|7QnB9lJ2PN+WzoOA
zyii|(3f0Av42XBoZ<gm3f_t)Smx0skIqGd+g8I)qP5~zoIkB+DKf=SFsVi&Ix7wOf
z;adP5C1hwSouXfejC~}Y29Pa}LnhzHtYFpzKURrxY%n&cDhx0`9H9K4K@mXozcb;R
z;nMw*-XA`U3NW>Xf&7zxyc)$N0HeWE;i|j};uAO;Ga`buGcruvWgPt`f{DYK;JJ<C
z0d_A8#4yV%!qXekBsy+Z<vf6YDBH>-odgB^`xBqt?gkCFE)hj2^J<$+!fzj-^6Z`r
z%SCVUV8}<>gJC2Nj08N6X+q=8aLjh~<Q&Vr{<1j#&-uXWp}Hw7)J7r2v~ra~Rmx~g
z`;M347ILRvyKa!;|IZx(AVpjgk}a~br_?+mALO|-TU~Dxb%MSv;h|a)-Yg=czL_L{
zyY=B8K~A(WTXolM7&CPT#{c<2f{Jr<7TbffdbDrQ=x?74^)-P~Wh;Mw3fg}t+n#Q%
z#bJJ6?sZ{M=o-kQ_fw47Hl^O$$(vlef{Dy^iIF|YFeZca{!O?I;15akG=MHP!!bWX
z`$CT_uRHDV?Ed#?aQ5G`0=weEcTql8Tati;5{3O?JffO!j7{vh(2_`Yi7`R4P1Xp_
z5uE;*Cm(l1W)beNn0IuFgiyR8_)n|4dR1SlMO*Rz=AzK)y*8ERivkV$bCeO##I^KC
z)6D@1xlq#HT%^3oE$kSQoqW;t1dKfq>U;m-KNEKQwS{ZoKC535(5GbtX+73!pg$j0
z2NUJ*L2c7D{*G^B8hJ&x^JaP#qO*GuF<f#wz^uCG@-{cCRJwxyWn&Qxv(SKr^eD;+
z@HxN7tCKpiEXIlVkgf9sgmLqadu&Q5?hJ$tOK6K^efZhLa3e|xFjHc`h#H5;riUy_
z%JoEQl#)OTe}E_gg)-Toax7TCDJ$ZXepIh2=f+0|H<4t(vmtHTtWY~V-H-@m+T!A5
zXaIyxM*rBNJ$WhTe8-anvfjn5E1abho=~mvYbK$0F-VN^w?*&)TZW<~?Y2`4JwK7B
zyGXjmXjy=3el?huM^q6>6P0X;e|kws1=ld?X1;bGhfTpRZkl6nhjW?oDshh7OmxO0
zZuNrfHvZ`H%+<yta=Ct#>ZdGS6Uza@C%KQDFx>tj;tN^uo0Ae~k+Dq^i2i2zxtJ2K
z>PnGN9tfeKC33<5VAHoAzy)Cidn*Q6`%jS23;ShBgS$dKY218&0$9Y(5GaD3{gr$(
zWTMyIDcH8oc>vi4&2I}WlrXu{TxRd4*|stNQ+HtdAxdRq&~$m{_Gg4!$ejySrhaIP
z_4=Wli=(g5H4(pJ?)`@e4^xs@v7W_ZP(|l-j6s*vSKT%MWWOR>)u)=&0y-rc_JP)V
zf+kh`k~@HJbz#byAl#dijW#OEi-h>1w%yuD1b<v5SAF|@j+!yWvD`%+Y!#gZCxzme
z^iK<RYD$o2yRs?j!yfWw$?d$Xh(vHeqQWiPnN!esk(1OcQqm?UGjq(~wo3z|aJ^#z
z;?26vg<&SZPfadroD{)<xqWFtOZD_#_<Cp>-j2*cbO*mahDO+{pF~8#M3?=;!l+Jp
ztx-8}V#BJOADgXDr+euHeMU+({yqJ2!&;+N*-8aD-34PSX2mR`!CyOZ=dKb1*0^$}
zbX5gy<&>_(W8QKE<UA65ap5N3@Fy2QcwkTr$v7L3S*BdI5FRW+K}aND?FN0R)^fG{
zuM)2nTXR_{@>Dkp{OXItOC&l7?MrRANz8*HiTf_iOBFA{7+HN$M^$f08(PC*(|SZd
z|Mw(T4M2^uIs8_vA3RP)Dg)M8cqd<<8u2xQL}lE6U{K(L6Zz{J@c&U%SjIG&8*wNO
z#z7y;iG4`93rDPWDBdjkk7`02i~pTg^)e#Y)34t%MRZQfq2w`yjM0A6TeD=Phe05p
z+GJl#Fy_Gv{hB>~&O@`Kyl9@Izm+_IwglX^sSdxROixQ&jX;2BkY^Y(df#f3^16Gs
zWYp<;&*ZFyCMZE?Q<Bf?1mERfDnrB)A2w(fZbVAYZJDyd^1Xm;ht)*M%_(n#1ZP#&
z%V82q)yD~hUdJ|8@rmm&0!V2cjsxd#;b|4EGH&?go_=aKaSO@czlsocR15q3X%suv
z;In5&i|Gy|lc525;{wV4RFQ}At%tM%k^bnQv=CY}M-3_v-=CK*2szlaDbY4|{Fe$t
z4yT`)|J9D@Bk9daZ$Hr%Pq$H<+O(0Zv7rJ(C$*<;Ff28CjYUq|?R*1Gre#pVO1Xe#
zvltc3iXro4U5yEY{f=m<ps}o_4Pmr>Kdd2D0Dkhw89q29^2gI^A=8^Rm>$fx+|^_O
zyAf$ycTh(&C5;sPw9<m-pE2=K%Bo^{pr4bXM4Ld%*FFeSip;rAWPCg|6n*>nGuEU7
ztp=QZcu$VFX|`$i+vKhA7O$ez<*ez4&Tn#zwAb2p%Y9Nc16n-I7##2aQHB7VylQB_
ztls=LHTiE5;x<U%9^J-JGqA|7WEc#k_Gxh8duak$oxz5O_4%I`1i0^SXvy&G+`@#f
zq;6DU)LE_ifuy<rNlQB3E7B@6wXpx>vEj+aA}MLz+KkNX(^z<#c%>6evD^<J1t?(u
zu!SgF5f}0J(;!bWY~9l|N#F{+-vFTgjPjmqzNiUgp9OjfIqhJ_{SMZl-pq+Z!^JIr
zHM6&!ExZLmTkeGwJ=tKzJzD8s|I919k7;3?@Thn}_(b^pE+5fOS`ce)0E$B=W%^o!
z?4M{&?V#iE&YK<353;w|6IVz@02|9<5>X`HYvpe@`nf4>P>2RJ(9Ilc*@NgHi=R9v
z>!tF!)1UrUGWxPjtP86Ctw;A@j!nULl1J2{dMQ%kOK1M<H3XNsQ6}3)nl2`CDQ1X#
z{}kSf#KY$oC_l;`-v>+N7nPiu@>;5S6f(Q7d6X^dUb@A57zW~;Z=S9I=3s_=jB=BO
z6I7vT%8$$bhfH)VuN3vpz9ReKM!ZoFi}u^n(S=ggdYodGV2T5o1cqhxy7E?gU<Rz8
zBY<#N%$twNiIUbU#A8m-qj|wXqAd4!Gi_EEl-KdnI*lp}s}JjD1!Q=C#~azK-2vD`
zjKo|J^ORJnK!}42c;;|`qnLL9+I197eGY{Tg$xl)qOkt|Ja}=1@mr)%DYW|=Ls^HF
z&kkalD}d9+ch-%3VcQ_QH_a%!CB`VixGD8x=Cmy>yT=c+Uy^Cf15z>j8+HHAk3id4
zx*!<yJD3x;%&atZ_+<dCjZ@<H(q%vOwsv3<zS#lbwTz6t0A|)Wc+b4?g`k4V`;hy)
zA0ES*y3J&<!y}enz*Uu$aa881*#khu;hvM31>aC<tZ+AM1|!F9glkms8B#R5$U*EY
zGbp_f8kuAP5mnKxHyns#aH*HB*Ybj|{`D(dxwPx`biq;YW>%<GY-$VF<D6bS5T7u~
zM<oR?Z^OJMKDED>?>@2Klm2~E`r(Yw6^`QVe%o;6TsBvVeg#m)|Fu;xj_`N5xRY_7
z5jqk-#HdS7xE@yp!G*ikkxRT+SJLh9+*rL^UBRa-lDjcDvNI($Y@}loR!6nE{k!k|
zHfXOjX;RfAi+cIOV<=xlEu87aWjka!KGp~bAonshW22RhG2C^i;RZvaB>#O{Fhf^F
zVxv72<q~d%nK0iW(gz84v{7jTo}xweqoGcUR?+yXbXe$4uE#x-m>dk6zXq*RNtOfp
zMb<qD`FbiAvk(XZYH`r~ZZ98+zKL@(KKEY++DGcrx+F(XUGT`dMF!G{2|vYd8sHcm
zzp~TCnxVhVlvSzm{rD~=0y5f>hBG#9*X-PNM<AICqjfeqNM~!?SyGp5KE`V%2<&Kz
z-?lhLrl!cY^r;q8H=qvrdC#k96+l{UVEmQLz>58>AFkf>afn2Rzvjvbstid<957oB
zT1FI@80k9~r5~1IgZ1KZAOE$L0yuUx35)$+T3MD$RQQWff~Xe7V5>QV+TXc5j4EWQ
z1Z8{e_pyBL+un1P5H3q9le3`|)dO;jicGXEw5Bpi8C1T{P>NXV{WccJ2~P6;2P1$x
zhlN8lcjm3KBAe<>L2iVQuj+U4?St?nnUwe}psERI%{?n^rN3)Ah5+>epp~|=-rJQ@
z_A0BS0aw1&bKQd{U;mal-$3vzuS~XbVr_2^vrs&_+wBi)7ou*dHZj*y6q&R=^ae~$
zAR(i69bXtUHqKCtmzaDvo|?<!l~^^n6<-<d_uh`eNw_&fi)%t;`^r<68WS^g5q+I$
zsUH|SU{p*YI*+4!8xO<__y&#g`bjmW508~<50kww1PeXM%PJp%7z4D9aHN(jO(A*x
zex`-OPPH4;5j6R?-rzpink*y}*N(H6qeq@acpf^DQ*CZ#F5_i2-ag)yt8jcY*&l*@
zgGQ^^5WhjA08fyNaaqq!c(e)E&NA9F=&RhsVIr?CjVQClq<TESMNR2yhUds-g_W%z
zgO_Z$jxC1mu3X3;54btym`C4<PnG??qb_NJlhP5aOV}!}qRx{UJxOrmowofB3@J}?
z$@H?Z%CeZ}B90`E*bG_csXP(GJ+?^IRWvSMPTGmiPORg@-`4+G&Idhm@895Uv-JF@
zoIiS;Te<}p80k(BKMGI4H|g$Xu2^NSXCMs174FvL<>1TJ865nDd-;Ph`PpB<m!Ya~
zwOi&Vlb{6I57h7ZlExeU4jZS(*vcWxHvx`EA7$(_sdR3BqbLx}ye)IX+^0a%P|xS!
zV6m8|jAq%-a2TEM5MrIQLYkw8krh{&-3v+bX6p^WgGmK9R4Zf^(#%I1O9?h~Gwk8?
z(QCY}j%H!}&`}$Agm`?6+i*tf&{h_Vfj*X8J$v!2)a5Zaj%Z=s_q=qC12%M|MX;HD
z_D*z>vjD}vTS<X-YVI#)Gy*licX3hNzu4ty>^PwR(Mex}r)3-$77{TGUMT5TnAFhM
zyL%Jh$u|VCQNn`7yB}OjR@Q^!$7GNB<DbhFYl~hI4US3liX)K`b5M0rjyT`zUfq!b
zE;PI+uWvMJ-PV}}SzZ`tq$(W&E)hzCkw&kD2TSRgp2Yh#C;#8*vCJ?^uA%azt!nNP
zBBXgyut+ZA2?;l31)iBd;UD*3zC?+mXub{r#6lB8mW;~3ThYivLd9VAd-UJiBpCOt
zKzg`QyuxQGxq=JgNxn(G(^0MO^EZ9Cgkqyej`f5ilrW|dWE;2@BRY3>tabx9C#t0f
zCuiR)A`{+Lb1HmZC2z)g8vM5Zu83%Tu_zz^g>c5@o!WGfeX4U*;H*lv+~(;FxceTD
z8)$uEp!RU9J?oIy7zVWKv#>McT#t+&LwzNf5?p5Wso&Ge(45;F9K`#|hcLiD<WLOy
z;{7NywxFv2!lXney`-eQGQBPwMj>>H@fdh1f%1qdh*3tpL*<=$hcZ38ZCgGhMjsGg
zRRD!)=i9F#i)sxM;z)99wh=7u&;!(^i<#znNkw?vS80eDj{1gP5!nWgrV*QNj$!P8
zIuZCuR9Cta#V}$8AVS_G+$e`u$i>~T)`E?UcK+$sqGXLN;36k(;Z&Gk?r}J$D~0W8
zRNd}9&5Ng5N<-8dWDLzCdbkZN=H+R~^(;9^1KsZHYwFs0l|G|A7E12+Fb1I17kj=@
zcnAq-An_l3Y93#eZ}5QARF9GRRFwL#Vc2RE?Eg%)eQZBC8il;iT(oFi$6zlTT8M5G
z3h6kDBe%48$*8zRmtByPLO^hfRdIlT$rAa!7$U0b#Gnwx_jE%o3?A2$x52G^HVnYV
ztni~vh@N+-r;_v@rF_1g&IB|MNYTKx3jcijbM;pK`%1*XT$O0}mp~ta>0I%j?@@;H
zI{GKkjSp|&>kK)crK*C8lBE;5>M?z<x;!63>lJ1=(HW%MeTy9*JMEvd*byH<vV)sC
zFHTM#36y3Mu78@hUEEWCj>Y6!dnfmlaWQAhs{l;tnF=f7fW5e$JwyQ@J5xQP%w`Rm
zw|SVPH<^y#r0+9@V?oWPFIA5`mq;r9SDaS3ikZshz%n_S2%!=THCF{_*n*%3W~#I^
z0fo#E$-kKC@Xeau29F+Kz`4(5962eVSvDZ+o?-_*nin3CEy?%wBzhR>t19{I-<-*d
zfl1G6=(#CKF~V+vo_oNPu6wxP!016WaQkrj*Q?MH(qILE+$OPYqHmqo8SQzSg%g)v
z?>)lMa`WSCbZbKnr@PF#unJMP5wD<OM1>{p#yt_)!4^?`kaM5EvFv>C3iqZ1PMeOq
zEu@|@IYt`UQBZ--Ptk3R!EPJ&aJeMNBU=-Fjycu&SGw@Ch!0>2|9srapMY}r<;OB}
znP#oZ`ID*1U2p%`zAg;mHz0!B9_p{d#`&o1y{hF$tTT~aT>vX&Tk2F5D$cx8@|`Ps
z-T)#y?2m=jyqhnjT`R|OB|Xo-kP*EC`vBh!TW9T+D;&q|_MV^78#Uf;Yv-=?N1fAK
zuS4$9g4oLNIZXg4;rK<oL|$HG+enmHhBcxj$HXsEU7zmF!D3BpaJ0(geaDHAvxW3G
zq;Mn}D!O=OdXDLeHKiU}k1z5eZ;ub+_u7QodjhMszz@T}6shDPmgkq<svPubfNAp`
zHtE%n17By0H`}ZhlA8IfEeRHW7=B9Z;=tqf+X?w?O)fxXC>8FPn}eQN5rt1$b!jR6
zxItO(=PT3Vg~j5<{`tC6&mavFyLhtq9J83{Yx(fkom;$&znaiOL<yR#)xHdMF!TIu
zMS-elB^AEsL@^hbGBq8j_ZGwKd_l2nRB`qiM9ee#l}*N_jo6rd<o9^wE7sowHMbAg
zUcVOhqZR;$g4hK$U-QB*pOGoe7Pjs-Ee_@di<Xi7^PQ%nC~()P-|u_jVFj6C_FoRE
z8H`B&;ZOwY<*vblf;1-`&^KCUV)TY)N^{Zz^hAH|il(K1FTpq?qtA(v6EVv5!>>#(
zDC%9Cvt6gGvX*B$em?IPEcJw?PPTzE=%uw|bgl!&7cE1J%~VQ*hAd2#cFXJ<ZHRPF
zOYDgDA2NDO`$hFBsTbjb;I<x7f4};-4qy_Li~MO~_XJu2-OAy0^O}#B>iZx$CU-z%
zwl2RXrBF6QS~!lOBQPCz96pUM5qbvco<mH@pW{c9$oqbSp?j211O@+7dYW4R$e_u$
zmIj>Wl?c5eU*s9ZcDLazLPP#At+P?IF4yC>#fKUJ+Gz@z@GE{G9;N_E9!{eKH-LFl
zb>jm>jirb(3^wNfLs>(t3WNJ+BXRc-s;p9l0JfswLh}!kN|bND9mKj7X1PjD(zF#w
z^rBhH8;|L!WGcHO-OH8s*}IzqIJG<zU`+?Y>!D=R&TpZ-|68$&L*6T5X%dpb()<~b
zXgvLa=5BSp3QIq~#~N|>+pp4<Ksn>(69J%@3vkcIMwm8sJ^a<-<LEdO)-(*qW36@}
zo9b;8#XwBQk@)=AB9KHL!Yr4mmX&!q+39;Ck%5-y4<<g)$@g3qH$^z8oWl7HAPJRT
z)c=>F_{m=#yTDkf(0M_pbhzi<gu0=lx)MtK4IJbgc)TC7=nLF>*FzHqG^c+wz0C=w
zA9qu}IcsM+ZW5KUY4?w&xf)R(OuT=7wW1&<e`a!Ti&Z(;bigz6rA0K^j;dZ9mn`)F
zb#;EvB@#Kl=T=X<yi3`-887w|0VFTpnFz;Ldiq5L43F^eoYL?rw|Am<q8Hlv?W4~s
z!&}TcsBZrQvj)^Ilk=fXdf%YJRWXjz()JH%Ex{ww6fc4EdoJ;{>Nw0e0iRS#zVx<m
z{stwJPf>b~95SzACEHvI?z!h>6vGTN;kx_plbJhXE++Z$oN2P~B!JTMLckFV1KT@E
zgNy&?ok%^t7^o%3P4Tc|v05!IF>65C-c&5Z;%bwB=r>+1<D2C=2>uUVjraz&`AV`j
zhgpFRj|ICwyw$EnRhiwt)*B{r$g^GzSN;VZ=oRJvsXMK9^z>@xkm9Gq<m>vg;{l}0
zD8bOdlp=<Xr;&m4LEkUb5Eci)f=`e%3XSu+7o{@v7V<|ifh!cI28-NBy>J)ywMP4s
z|KyU{l7^T&>rBe&3;eJdZoBOI{*kD4FaOL9O<7wwBmCZ3%=+G0T#)|G!NP3flI&W#
z?xEdO3o0QrI(Vxj;(T+-0QLR1jce}+s(5_&c<1dJBUu}4b;;wGiZO#~+KR{3@YKa-
zZxrwmA{Uy>XBgI{Qxy>*OGerDlWa1&PiW%dHi>p%g<Lok(?UCb|EWDO;u(e-r0W}+
zQl0Zj<?inH3s1uHQB|J9BFx;mM?8))I!*5vBY&wK9HqND03fcG4*QT0lD3kIS-LZO
z`*V5)Aj$S<1R%wDCecY(9#qjYk`7B-?!tpjN|K?KtoJJ{d!@wiLIg3-8IZa~)kC2x
zLj%qU$YfI(&{2ZII^OciBXXGTkhw6O*#9tTC8vlO8m4*IO4Mlb@SxTnuCkvf)|`TK
zL2VC0^YQFX0unxN&oIvm$n!C47l!I^FLyMS3F)f^>>A4197*dd*`~`#w)^mo_c-t@
zAInAG^^+mBTFdSOd*^z~tydCI>%nN<fYh&wHc*xtyp1?0%_^XKWL0kdOXjjC+q<EZ
zWbshDH?TweTA(dt{K3O^fQZXstL+=k_rCoOxZJpA7~(blN9Xby2OvPTEZ4N4wZYLD
z#sxRS!Lu&=TGF8297eWsIts<z9%$85Ed?do?*xfc?rndBF~S)53Bajg5a7ep2B1H3
z-Y-xx(mbSpHE3}|ha$G9>u}35@26ctg5xf1>N&WO55b9HwpNK*y6*67E8XQO#*hL|
zkdQa2<+5d{%al&`(YZtuDmOCe$|pqD9rS18^19wesu$zOk{7R<{-PC1BdYcDdB>uA
z8Iz$|Uq)k<hz)F~{VV(Whln@m%ni;7O+^qnT@dqN06+ar88`X)YtRoYpb(Koc!x^J
z$?fsOerw|A?7oRI`x05Q8GpMwfO!?OP3H4LBgzWl>Pa0h&^u9BCqAoE3$a+TcR|LJ
zKu&_mH-l+%|AY;*!t*BVpm4i`$i;YU-S$8%!bu=f0_#bj;PM>?P}F}lFcu*Rh?$=e
z;I;Yb=O(DcGjw4lm3zsy^nMNlHba?ya^Ed;`XbO3b<}}_Q3b5BE;qneLB_?f!s<&&
zw_u-r%#yXp^}XZWm1pSKPD;9Um0vbpC%3V$k8&iHDe%QC-$v{nGWe{E*Abf`8>7T%
z#96QF{)l{LyXb-H^6NyDL<kY7>kTOsN~rXq$*W?6rE-l1UwdIgg2MnEn;aH)!7;MA
zB|pwS>{BGqC;Rt&k{Ezd3wb*7;nqqDHyG5|kY}IgPlno9fgq9wEF4wT*bfnuvwDH+
zHOl4R+`2%;Flag=2YtcSE(>5z`@A|01gQ4jBmHWdGBNXXRXQME@7O}vS))nBe9C_>
zs~H5)9x={xnJh+Azgd^9YA>*7zVYGSPtEUkjsv!q`f-`Hj)ef1?S#xT4HJvz7r5ug
ziTeO`?tITE+r^rdiGYQ+HocGwQwGRJ)c&Uix0=+w^)bhMZLWVcrk`w1ArT0hhuooG
z-B_c2T<|LRbDka-mu*;eleg{{LV90(KX(t6qdrs1x6J>4SNY#EpW$Db4;d)JfPj(V
zH1Z*!%?$TG&3GfmrptL>p_6D?pH@>QzFpcq2O36>(^9_S_EO?vzV&prFjA_tcZ>3I
z+Uc)P=d6YUB_AHf&kIa0D+}+9OO<kkUp@G<YRPGy!z30XaxN-^sl|Bf2~}J9CQa96
z3J%e%mNFhNan%pEIi|p0OXN8oziro^E5*~6<#u-P@tE}B<L?sd{IX`m&Upv<G)$V)
ztuW6mIW8irC+gBfK9oP`v1naj1wje}+p1aBUo$%v!8>+Oockj&Y61V!3_T8Ii0o=(
zLOKWh1l#r)VnZ=2eu`hdGc?wV^q+MRUWwp`OiF+^N>yPu|A>N}8bZtrIR+ymHYB&O
zRIv|X9Dz=@!bTRf9vM<PAqfgZ#BfsJTB)5F0mjfz5Uxu|2pyLxR~R2zkK2laoU-#Y
zH8#){W)Cy%C`dCqGdYW&1d<j<Qci_>H0l|+21F$WC>XF7xlHRj%rRD^6!;X`CZC(z
zMSlV8GN<*=X5!d48|>AOpPi*IU4IYPiecontMA}rC#Et4J{3VGEDS8u_HLAowiCaU
zCPYZn%p~2tu3-DA5xdyjj(9#xAne?4ErBPN2FCq4q%J2JLm!BdPxb^!c4?v>4UH@i
zVnOUyz!m-xfPqPZiP$=Ar`DO-Jp46yD}c7%XimyLin>iJaPc04%80#nPS#E{n?Yhf
zOosM!4s9%k(9QG4+V3xZttXH(g>T|rYFPmva?bDe)=6N+^1ZssIv4{jh~}20M5TJA
zl#&f_5omiwIRVw0)pL%T`C7JE5lfdR*?L%Om6YcMQ~q$Pp+6CYsMoY)vsRu>0)T^y
z@pcX<TJ`BT`O#_8)j@E1>QY$dUpq<JO@`zINa=@!P7PcK?t0Yfmask5UW!?Y$KR!_
zm6S^-@Gfkxfn&M=&03x7&bEeR%?(ycA(o%^WuS>W^+$;XVaUIiwqdgdgMXE2_=fQr
zvV_cGiWzs}jL5LC5tJ9f*8ADemIGpj2Op3%S=6}h_8A=<Br_W(j(U7d+Vb*9AmeW^
zaY`k+2BV`v*%0a3L}SfF=Bn7hY4=7(nWM;vqsazzKu3Nt>A8f@nHY`K->vIaJ8E#p
zR`g1M5o_1Cq|(Os7ITEiKq|E-j{ULn^}M`?HcG7mrnS{4K!7U5j<Vc;WCT2$4$<R>
z72N*alG2D;NQgzP`P1!&ZrI-y8g=aN6C3TIH5)hIao&nuV)nLMM)IMgKknl*Cy)U*
z79?uQxv-oIMU}RW;&EnoC72grNc4B#J;r-++)92Dg?1ed8-a|D!2zjsp<9@N!LA-!
zXaTRw3plw=z$Urn_cb91<pjXSZC3io@&WhVYucd4*>?~VoMAW6OM(M9p4{hLVfz<)
zF#lBl4SwF46(ZPHAkrgCpf7*GgVIinN3jZTh5QVU^U~&?L$Ovl4ROGis~p&kQb%~*
z_x~{rHTxB8KE#9x0$vMUAdTz<_56GK*!Q`&D~Dsr^X%3BoT?2)nIAxprwX5fGy>oz
zk8!ie4O`%#Cy)Xu7E1}`GoGJh93R2TPdcIGq{2hrxn`+tiDT9Rp*_47glH+?ZW1@Z
zP;oNhz&{in<X*bIk(sbt!8O%2$K>`6-ECRCV5O`mJS5HwtXY){#rCnB`>THBzf4>4
zv1{i%jf;+#8T=NAL<Z<3R#&PeT~5jzJy|Mj;Lt(P%HGYe_bu8dWCd6YcQ9~Au8<{9
zK#G@{8wI&dk9xnCB$Jyf(&PwnojJmat4%8Upjfss*{jsxv8NtiB2T44rP>beMY`rL
zU6)T&<gbAXYih5lR%LZUt22h2&`Fe`GA(w^ifMGB)<N3&=mLr~!e|MO<w49}QFZ0h
z6=*4-9q_kQ9i^iW*Uu&U6p&j%KCG|A`dCg`98|U9Vy2l?TXkAS&pJ?U<{DS?+mM7Q
zH9%|cImG`pnS=J)u^SE!q#ksNjI7k(=t;|equa<TNpd9V*whp1DJSGer?TH0>4~Kw
zLAc*v&V2WelZr5Yi8IgF4eb4n_a&sgFMHgh!7AQyLnzrH3MIy5U+vGP&ZHP_RyIW$
z@?%qFvkC}wh9WytRK;3Ly2P?t@ZZ*=ii9P^=yx{6RCmYEy?39@DU)4CR*d3flo?UY
z7U?7}5!~xAE1K-q)Ug)CODCMjVFP}Dy$F0rmtGD8SUit)TQKg8hv;&3nE70nDIG8u
z?RwfCBK&Bhv#n8RWL~-7ZqN4qL{E=1BYVtCqR|VhcPG{h`-(u*s{%%a-ab}rr*Mdl
zyd`5n0rVF1JHgWviJ|U>=Hws)!Vfp3VY+FrLXVCOIUpg5232UY3GzvRltLev2&aL>
zAndjPZ6yZcEIKI@^-Q?uwGS5CJe<8YBPrBTApP(0eK@iv`i_7AvEd3aq!pnrX9usb
zPw?h)K#u{iWC5ObuX(3mw0(%o?s>qQWezKKYQG@L!{20QlXkEXZ%J(;R&Fnc1PC;B
z>t>~KFtmrc4T%blSGq^ewDw!}ee(yLK7MO}N=7tU0DaoV^>707lw3%})vOU);?$w}
zYW&P0h>r)-f{FS4(#m!3SoG$po|}a7bpyd6S-WlGHAZ?}KKJo2jv_S!)DAj9vS}GG
z4a597IJU{NKAU_sA(KH6Hz_lTgN}t1?zqE_E|QfK!Yn%ocAs#F{2`|c-vJ^zzPuLz
z68&)`-S8y-b9~Liv!BdvDZHewcj3GAvD<+gD9yuR$Q$y_79!=fDouUk4?|OWj|dOm
zKjOv0X9md0lOR_B(r|2HOj#u4k{W}`aSk>-@t@GnM+MHFhjO98B-LyjTnEAR0r@{D
zl>N!=+<bx_tWH1}e#k2|`F>`5J=!h;vU5YW=Hd44+;f_2CB7g1UYw7xGpZk)+KF2`
zy&|~YN}WBZcl_iod<~&GUvnAkY-d#7#p12@M_h;DoX}+wvIjpjG_J$u7c7=5>7;ts
zvZ<-0=xZwFDNscqzVikjb;<DKCqo@Tb<x)ADf6&+oTOM+dW?DtRwxTZ9&t$kHCbKt
z5H|2x3A3X*lX7QN8~V$^V-v=KpFzj<|ES>;Ey58`u9uOC&Ke_+YK{2fGWZVI#G^#q
z#8i2B%!~U)XEJ6<NplSvV%E%R+l!9R&b$9`Nf+3smo<Mw>>_(T$U^V|^BoVqqtk~@
zJ%$%t1WO(yz%7RBgB30U80&G3ygXr|AvlnnW{t|kp*eOAY=zl;3Q`JC(Ct?Lp_E@t
ze?!W>^iQQ=AHIwZzuXYKFTI`+U#tB_0zHgIxGt+<v|--&49~hc9^LyyQ2H`?cyiqR
zp$Lw3i4F02e?R4m@sDnxpcLde?3d3=A}|i<?RU+1LRL;)zi9xlHgNIOtRVO4UF!`>
zPWkDSD0Xjwx}0In5?cHSUnOTXygI;5-b^H-hh0k7z8_kzS(npGkqm#j(Am}=x~2m4
zBJHCJWqmKY0+^eQ0@odQD}2lA;e7Krlq2*dx8yGL4<nruYZ5xXlDSuGr*?A+CJDQF
z$n}e-$wY(qcewzqY4CFBU4C=o5e2bemmvKr2?AMb&b0{V;0}!N5*VIKD4B;v{hMdO
zyh#V5{7u>VXSY||-JmkJcgFA#P7F}>yNJf{WLKPNl@~EG9r{IV?Y0wXQF~*U30rsg
z$#IN1&4YD}?*+uonMG>}bTu3EM`So$gTwEaU|9@q1IB;~zbCi-AUN8y5hY2qfPqI)
zvxUcZ$uqb3>4Z9Ix<jIF>Z2%VN+MF#sAxyU;$!=P%@7LrhuoHEr?#x0F@GIXCwGf9
zhupJyJ(E5PuzKB3qGNpkUt=h3^9hzlyHXQ5HBGlIo+qtW5SRuzky+0APTKjxdZkl(
z{XF<H2gLvwuWff(GWPl8OICep-rwc2n^PJOcGR@*ZZa2!S_liColK->p-2m!*kb&~
zt3%0)J~9{8>rk&TI{FbM66Qgt3Wf5t>t7W-vmK(z1Rv3QU)nMjVuq}Y#o--ZvQkbK
z*{?fcZXRSL<Dbq_6=}0F?6eH|;m}iWL1gqYCJO+2%AF_#oHhqow93iLd)!>NCz2-?
z&W+T2R&W&ZG!7(CF?NT37RL&ad$+*c+3*ZiYfN+8Un*5)8jQrzSQd<Cu*bt*5{A1&
zQIA2V8o&Oip)yc+=JhgA<1{W_=dr`Cm!&bncN$k!V#iuSp3ucTM7cZRXqt#1ccEJo
zb29_5X7eYxf3LSHu=4(Wz1mA7XY8Mx#3DJmXxJo+zNoBCWhy%ET)BjKS0!Wr>gh=6
z%vKZbh-yD0sEEYFM4Zd~n{SLoZ{EVUJHg8!$}%|+bM$kBCt&yC5)_Lfrno!KOLpZX
zlcZa*KXMAReNzBhz9wiVQ%#M!cMUdSrcxYW>d({$Oo=u{SS-6dI@rf-rE%<(B55Dv
zO;XJHo+V~7!*COC30}PQZp1$74^gw~^@C0wff^Eoag`AR-EGLj9Whd7azu5=SLCe+
zmdKReG&EwUKs2<6<Iouw>zVZ^|NEKxuKE$Lfy&l<EMV{qHu=UOG`y_PxMZY=3YQh&
z5<sv$kB_gYT3GQ{4PI$URb}bDvcl`1Ta5~G83|7-t$<{bg5+gq!tXQ$m!J!Ju+uQ6
z^RUH>Fpa$-m;Awji|UW%%Hm;@Ap~{b!hW8NWc~3H{;s8+R^pgYE;6}soaA#s3#!?g
zq#0eE=yo?|qnsa;>Dn_F0--5p9J_!;;R(b0N+wKSGJYg|#?^P753xdLo!r6zFUl|&
z2b{Ohyx*ceIUps9dGr~LjuTbAe>W1nZVEU|AhDULWS<>K?FaXwoIU&}JWl>tTA(Dr
zJ=YY60#QdAD24rTAK2Uo_0cI;`?+O_i(Db74V?IOh+`reA<p+}nM1vjmWm6YepitR
zhkI_>?d;7_9j$kw@NRdK=9P-<_x4_>(ChTGfW27<GFF%74--kLf5s|!FNrEb<xd+*
zrqErPRUQcgRO&X}FQ`ko<=5f3vmp*u=h4E+2iYZCZ<28ebf3>Ox6{17@YEu77qOK1
zCDsl93OlhCU>3XG%x<}Ec$j(s6A%zaF3>@`J=&u~vj%OXN-6ryz6{KN92j{3Vsz@U
z?O3ap+ac`Jm2x993ro0>DUsVK9<7fu4<iEgzMC8oM|IrId{saLCqe-SEKVz|-!e?_
zs7{UR_Bb7O*iru~@Ws4OHWT<4H~p=z$tGMmD<<a}?w&ql(<J0GprM*q@zdoDgUg)v
z)X^3tn5R7$qhzUcd9(TeD~@xl>(0w5NJLCC)*vFLVyL63NnN6&RN@2CMGRreDvL2g
zueQ%9cyCw{#mE47gKI0I$YFk?ck<5H(u*BZLAwfyB5uRzVZW-D0~Rs?dis2R(-98Y
zgWx_);E*Zu@O$(ckor9~N&-~mfcU`+zEgtAG#mDnvp-|i)I}CsTBIU^d`Sdt@G{h<
zWq0StfB2ivUVyc_-;XPnEG;mKw<v;D0iCKX4)(ntTo>n89(+VFlcT{<QbccmXBKE9
z+NLcn?mXdA#jebXXK?<vwW8pgBfq#5(LC(mxgul$C;o8)Knn{}{_RH?piQ6q-T{@o
zOyfMMeK=I{&1njeEO$YpA8aFMQwsxy5epObP=s*6r_ATVQ@{-Lm;r<)`B><5&Uul(
zbg<zv)k&mlM9i^6{-Lg4%Tx({9?O{D{UcbdQ6XA%-;!24_Zlz-QM3SeB*F@pSwSrB
zrNyx)Z%-Bn$hWZS!pu;J_P2ZMgf8Tg$R4IY3KPV&5wJ<{@hyUQmu-YL4yssd$|Wx8
zRf5H;BY?=yA|@V~bG8LmS!gJ$@zYDo>O0l~FZp_tH=Qr75Jk(JE3@Onu@B08k)v--
zq(vV$;Qd3#=?Ly9e1>hMShZZy?CDrF1QF7QW4k^H5Q;rNX7*bzjgZSpw;+V@QTPoq
z-7p}g8*ghx)JqYVFW1MK<q$xoH!#S~L!{I_M?&ysk>+A?JYq&HUsi~|%zX4(reR0J
z>pb)Y!$cY`bEd5A#h@W#KT{CWW5#^3tI)sASm2uTC<gDrM!Q88VVn7ZKTIypEOtEl
zd>x?<NHwS-vJG)BJ>AZmlAxdNFyj6nMr&v_;YYN8PHFctEsrHGLt;M?dhkVT40Dq_
zpO2@nFP#Q~SI66}0BvVpMxcC)F(tguY}x1cfYrLN)O+Ec_(h@R^I2)rTYOEyjq-l<
z!&C#mExkg*(OLa!8Dfyq2O?%iIzZT{Q9;ND%x&Ia`+zBRo@@<Ey+PAXB1<I0kUJuO
z9v7BfbJ7CLA_~h*F^iI^-RUuqNAV(=M_HFVf)^*OREGL?RIqW3fLQi3zZbdjgaW%Q
zW|y?Tjx|uxe<)iIKEq-%{yd5X#tyqqd(F#tHea|QL=u8{7lowJsShvE8=xhEtLIE$
zqwd9*Nt`W<zzfWH*8(Y63-i3uIxqRg{l@bUFeTg)V8)?I(ZeQ@v$_tTe6P<UN5?z=
z5#z8V8q>YNi-5)#@%gg#9(N{cIOb+HS4x8;Xz)%WsCoA}C58%0ntxfm*xuHiu%R23
z)B@zmgQmxQa$uyI0ET>x0O0AStC^;)R%<EPtKanJ)LgP4+5?j9WZFa-BH}Vd;--|3
z0D?KR9He`sVEn#}SWwMfccT#;d)HIQEnTy&i%$h*NHXk0;v+|AP*P@CR5_MlM+`9~
zxnI3$$V_77E52N2UPDXaMNYVYCM)&>W5anSRT8$UQ8hNeI9mK<1}GC&Lk={pUYrW&
z<4YIg2g3a=8&8OiH{*Ep_lYc>7h^M77xdr+_oL6Gj@zM`%J)u0D5J8g4uQ+Dz){dd
z|1>UQ<V_)u5F92j7izz4G+O{EpXimudjs3HN{d;!?pxu$)m(5f>RpG&qlkgH`qU5K
z_>)Zj(VltYLqIPT42b-I9aX4bjf8Kuz7L$=dvtyNfL}1t2xtk$Mhoqlb#R;)-VaKg
z0eR*X98>W3)~)u=fMh*msK8UNrzGf+Rnz$_yd9FY&mNu<hJGlsKJrH%SLZSB40>Ec
zspocCz+G1!Dsb#RK*vS#S~T_?^g|OW5fvcti-Z`esDP%{2goG$i9VZBd9OgVcp?U`
zg1YujWb({46C@=?^b{X~w43H`GhGWFVUYns#u9}*GK2ZF*OTEMsU%aSIGtM8A}jgi
z!n||>3*>|1!?|#6cCR?y16_gReH)>?CI%fPsqhPRAcu^;Zv_LD2;Pkm8%f85tw%%M
zYJ^NXNa9Shfa+nkd~MldNY+1-Q6^B0I16lS6QDl!!L10-@h+5RcstC_{aFc`>mKhV
z!#Fh&U~JmSjyQ3Z0@#xeSROGm5S^{}gVB?l?Cv<e-AX=fW3Dpn^Nw~-{8pC=vLH>g
zE#V&T5u5B@#K-zwCcPT4&siFpf5Saq4jjEtx5bKh0Nyczd~z>Y7Q0vV;Q_U|F2IHy
z&p0ow(Oa_5Jw%M@r0H85&klQrBRS{Hu|;U~u&384_uE60@xYjLc#8-F8={cZwc9|j
zMHBe0n-J^K`N{Fda>A!bJUh-HuUS**vs3UYf!!F^5})b_!RdFCd{?h<b!dxtuaiH_
zQ?EcK03H&P4i+Uv%JsW&HBRvH{PGOHzDyNN+57v+9!*b6w6-XzdOj5}Rmvz;%NF#M
z$S3=}faWOTYI%M=Pv1A>&-=G>8Qgm4!k@y*uVxPKNNzZ1i?9P&eHT`vHwpVxDi&kl
zIWUeTj{w1m<DE_NMg2UVHN_^4*V<{7JFIf&fbYx4<g{kuy4&?ZcK{zqb@8LMGGs5@
zZ*MidE6D})seQXyilitU1qFWJ84j<R?ZUr6!EDnDRykv?OG}sHuin$uo`7IGn+3(t
zu5a6G?*wu_zOyx(%~O9AG8F;FzY)jWd3u8zDtJ2D@y*W-TGaM31&3ew)wMr*^7rKf
zKqc6GC60I$DVBJj%L<WVn!kJ3{WpZLosA0-JelS(Lq#13?1o$Fq5tOOu2H^r4Fe^u
zzizKnZe#_Js(&dAW-1{ipPssAMrpRJ-Sj0EeRiAbzO5On_%P+J7aTf#DLcgR?8$b$
zq?V%+a58zs`1b4x5=IhH)Jk&P#^@C^pw$Byg~3`&dvZeOHDk!LgykYxs=J3#`J>gT
z#61-xEdbb+DQ!Of>wnFFz;ZZR8dPONS)LZ=6f<r#`p*Qd8_hHgz1eim+32=@58<8s
ziD0qqyx%gd+~`ia;c`2`P)sJY)7chq>G>nEM6d>@HXo1rE)0OtAQ6%yBsb$UtY0ne
zg{HJevgel|`jYPW<Jm6xtKnqA&yL?t1T(|hB}q!b6z=Iyl({sg5r4*v?Nz}3EgV-I
z=~!oSc1HZ>e$3T$O1f=g7J&xL9{<GgeVj+Px^<a@A=E&D|FGXkHr6lwl4o#F?R!}t
zTV|>6A*O1>26jL#8-O%3a5<T3XX#*D#iW<nCVnX*k5r1B)HCbd=^5hXq&I$EKd~e=
z(>v6wwC2avU3CpNi{x-t)~X6?bdS$T8sT`x+RU_f*G4&tZ`K#bA-C2{YuOBKMy33v
zCDzrsyUZCe#JXxzW&FD8bn)dQmEb`{5k|n{*~U0gg&|-Ap0YJ3G%G>6eIuEheMgGC
z`DS}q(sjv(70l*ZddLWWI^k?!#>4+E`%JwK0m$j}zUwZeGdxyzl%;1o!8Kcjxorw%
z$(_MXA)4IY03}+NDn(9;aKqSCmo8hyEUBq&^~guobXDOOt100$^`VS!Nd+36jAw2e
zecKkWoFt%LK`go3_L;Gob%bqI^tq}_=<0DahlTM@qF4igm?OjQ0}`IhPp(rd_nypQ
z!v$0vQc%Rm=7KPdxAV{xoySXruW*G>>lTti?i1>TJ?cp$`e`;%U<*IQR-BbC4Lely
zZCoX`QD06WKC)u;tnXQHpQyrB<*M&dXo;@Kk{&Qq@yt}=>Y(ans%N;OU|84>PX<S2
zy@%S8L3`Z&b+n20xc<?=Wofs#y=P&!*^vz7@G6k;pFtLZz?dWr!`aj&-S`<k*gjwI
zPNEtIin+jU2AKVyj~<sn?%tWTon@Ecz6z_C(oJ{BSH<aUH_7R$s1!Rn&ZFk39oYbF
z%~RIdDbMuM;aq8ncbldq$)iziYgT%|t%@a<N?@C%0l~`j9C9A1fV5wVK{aXGa>rXl
z?HmkL41*SM6UZ;{Ss3eI;Cb7Y!w|Ot)H4uCa}ZDw^pN+vx?*`DVnERXP+~wUkH`W(
z+l33fKUWnXB!77$5z~4vP|?u#<AVbj3}hle;`%p0ScCow3{`#SSY}%;1soPfKyXR3
zi|K`jPBH$N<v|Ba#*HV>d&uTV)}`0*?BvUa8Oh8aMQ2U;ysQ2B%PY^crt>p{B#(6U
zjzUKo2vX;!o>jmSOxf7+$88Wn*B5^v7%#*lrXTSC#n(Fp2iASjgRyPfwr#sRPRF)=
zW3ywcqmFIcNyoNrbSIg7zyH)s&CAq%I`^FWcB|H2XRW>0n%{RzbuZOC;zSRsjeli>
z0B-N-oa^6%gIrkowc0U-IFhl0T~94I+UPp_@q>R#Q<IKHwbJ2>YofB*eNaQUrN2&)
z_lGpe+Ju+0l8XKmxX$XC^Zu)$U_d5Z;+WEk`OnmS(!8LN;#8!}95PvnHuLRDY)`sV
z@B5+Bal5#VU4L>4bG<p~%Gd)&HN><WFlx<zvQvhE;6DFw{=&ier=3)@Dl6;zU$FGK
zcd{g;#Pv^Gx9W9G=MMa$b=04=(g|`h-Cx=?mW<F!vk6jziQ7lVZHE)zGjz3?VN#C;
zX!>vXPR6emdT9oc?pW%w@3hd`bcEjydEvaU$unATmP~~LxJz+EW~`BPG9GDE0dKxg
z;uQb9m4`^3C?P>5Ki|rSQ>zoPaM0F-?)x7y5MO$>F$+5S^{gjV7%qpxpdlF)I%|4_
z>}UilDmy#>1UhSeBP(*SSGt$!2sSG+Imk+SIIKFYAraKT^|Ktskx>FnHcWcB>;}l6
z-0<DwAIAUZ`#uQ;HmR7%bRc83PXd9;m<2USZYVA?oh1eAYw_V<BXOloHk>7}(cLTp
z;bB2;ZzP^uxj%N$y5gZ$r1PKU#dsZUaY0cec}BBNi+L$(zirdF`8`pY#NvP51A+dC
z`<j936VV@DE{i1FOkudUvZ}4@^%9*b=_=`B7*tz*aK*62us|^_gw<Y<#9v4j)r5$q
zFT2aA;NO3;0kfrDTQFsQVp3eu1l~1+?IzgL$d0%F-+qfaEI&}zM%?}S`>!~@uDxO;
zfJnz&xipyIR%!NeTd0epo4C4nN))Wx9NXY_3y<QF1%E`Evkg_Ai^dwqpUda7;P(I^
z-7PMNeV%?i4$7sen@*u$l@p>190Xli1)W~~TXar{3za^xQNThbK_^h&>f+VE$zfeD
zrsr}snG|4Ew9~Pl3s}#(TlgWIK%NI{Mx0WN;kIFL)><5zF0>50g!Nia_xcSNK^VQJ
zGBH{y?p#c5>(|v;r&{W?%`S@MMSBI%3X2BEny%!;w{W5!{nSH;+-R?Q2o0DRW-N6Y
zE54!RG$7P8t~H59G=3*3i()YQ&VrgMs76veJ;bQto%ny7HOM}b@i)(r+JvYztK84e
zo}%=|FQc!p)tC~Pou#R&n5;IlK|s}@T4rb<W9^3FD2W~#VE(b@fezmRN1G-HV<jo3
zIsHi&ftB(Vfer~FAmoIHIr7KUuN2xDRiWxmED)F}q*~Ff>$Dq=oHDKdvfcA2g;A%N
zaLDzKxAb{iRI5P|W%C1y!wJ)4KR++@;a%2bnm_2frmJXIj9`#^_Gpi>kP{24?K%Uv
zJGhgaIScFD$SI`I+y%0L*Xx!JN)a?P!<dkEUUgk<T{NxDWBwT-<is@Z*0~q$t~ypi
zbb0lm$eG}U36R1Ao8(&8joxg~u>+BDYKoNcQ?K8Za70%KqmD>q54;ilvtFp`J#8J|
z+b@S1nz!t|$a9Z(pWTrk?Ng*W7QbgY@gMEmgGS^9&Nl4a(f^eI!e(9Cn_bQV!e$$g
zDa1l@`#RR{>FUwEvEX3p!*WSmY3|gDf}=6pMxpy8RR9z{y@BUG*TWwDxBE2ux^iPS
zX<V~=*WQ7^*&>O8ku}A6c-HBlgy_aCgG-%TXYY61u8VUp=W55n9x6|5M=NaAG}wN7
ze8aa8{Uct>Hal1#<i&kJ31ttQSQ;fn#N8^DqozzmtC{$NtX&bs&C3F|WY~PHSY?#1
z`8UW9S>GRHS`>yH*4*5LS2VX*x{0`W7fKsKubw&}KgIQTzWW>}>8w(-QCI}%`}3UP
zoPl}7qF5oigMfhql9Fg)w-&5J$qlxYIzxDn*hxP%5{fece<VvTc$m?R<5noA97n;8
zmN#7Zy{q&@#9jDiWl`59ul-Ta+g}yJMj2D+NP-10PPenso`yQL@gM$y`@w}_g6t}W
zr6`V~*wWWl2{;DWQKs1qt^7qPOADI9Y>Bl+w>h?WOls%JdQt7a^I-`WowJL>VhQit
z+Z6$crf$>#oMgD`%J!IkA1y=(%7vn!g(N&FD5Iq85SXXrZs(9;7v4oFKcebtxuyQ1
z9PRpr3gwJ^V8C@+p84HLAzS)hM9kzUB(mijD}yR*27e}s?P`f=@!vUaV}$at3IE2`
zLsqW1JNR;BF=0rIpjqgGJpF<i8<R^EF~|u-W2#hMAY%ly7YQm6hKw~VPvG=9hH>CA
z)(@d13pq0?BQy~WzUeom<f<xRVNCU}7U8{G6_!`2WvC%occK>M-c?EwQ-xksB$D;f
zl^AB{H1kr&hwQpmJq$f7w>R<XpcAPK>xa;RKWkwjV>Oo2^pQlrb0H6N&G*TW2{R#N
zl-0Ib0d*C&svGW*RRrNFx}VpXmn%Zx+=CFO)o#lXF~zjr;kIlAXYsUr2k9`v8EYuK
z@gb6itr{%SKEJ&0_)!?aP)q7%m_Q~O`I|?xbf6^ud8GiTFAf=~x2*aR8@WPtehvp}
zaC7|gwjN$1^FYzM&L?*}=9GKUl4ho@`dbezumc$-zb2_(VlI7JkflUS`)U3B))l?I
z%F1{QS_h7I`%r9OSAOnk>>{0B?V%`au*=Yiz;j-H&O7U{(WKZcd)<d8R-fxiv2-Up
zTQ*u<Tqhv2+Kh2-IMaQAf;+)S1uv4oI(J|g;JC$*lk;s53e1f}K}xurzgVryRW0)p
zFm;Bx<x;viWSy)OR6uhpN5x9Lt*<S=V?b!@{{4eU%)#$>vpi{H;VKG@QE-AMy1VU2
z_JOfZe`(iI$TA*b5&zrf%VsOW`$#n*=2D13F{tRnaO>Qdr0M>IUa!A?+~p-u^V2-p
zGn*hLv%qwu-5~qhWDV>IRr}3?hKR2%@Tm`#%%pWB=KLqnZNtlFToh}(+@lN5ecio%
z5C<(9!QmFXTo(e^Cw3L;{*uMOQx4+ru&=MnChN9=ehFZkWAQY7ma{89soR#^dSqnL
z)$}@!pjA23<x>KCEXw7Y@ZR4#DFt5OyOK5ap32%Q4@K}P_&!HFpL)x+bcq|_u!%f2
z&#Zj7x9am`vIFYh);eP)Eql>FC$Q=j#JJwXRL*O}zw{2Wp6RV@r@%dvXQwFiaXREO
zlX>qc%#3^^^n_%zIG0;1p70e7Dzl6bq_u}RgsCi=Rh*em@K=CML852QjA0OhqMs|T
zFxeu*=4y3n4|KinD;spXG<F48?#h?`Covm<GT@h9+aqChTA|BR`=?L8G$WGkJ}u>w
zUqbSOK#|FSh7*Jc1R0xXVg5y`DiB;!%FrTIa=2Qv52e=?b|7rUY*=1Ow9&^dQ8<cS
z3Z)Ztpd?VnkFU0ov`L$cEWTgGW#FF@uLxeC0?9&dF~!QNp^OtrtNRmRgs&bD=jj?#
zC)MTVA_Icvph{UNW~@!xy>BTpP42E0Stzc7bxx|3sE9Onh#CqHPbheD1ahv?Dqxjr
z#X?L30s(So{1ie`xlYwy{KEQ8UHVY%6gyfVWtOqIwi;dk?|fB6vE><7c*z^#EIE-k
zn9VAxccbbwJ*F_7dmSslF`{>)Vv!JycV>ej>@-(|qika{D!HGtg6C`(@~<N>jH)it
zh;MYW0_+a~2h~cgew$tT=$<mUi&*E{y|isdftR<LsgB7TBlDmULPRX~!c;m%8tc2k
z<deT7wk*@t#yFuc1~NQ7Ny~n%F<#|Rc+x;~l^~9MzYnaOUT!$>>(yu#gv4X7_5pj&
z4h&re6p;XteT?l{)Jv*Avr*oQLD(dBpbu-Np4R{MwC@#DdoV+%vpB&zj+(Ckxpi-m
zr1{-?3$%c?b5>9Je$A@sf<9W0nmITg90dZjUr*^yb}QUWT~vb!3G<2<CB$8Q)nD7i
zfkT|WegT0QY1Rj*i;m->DyUSNOE=qB!yd$%c;i$%QY}vjg$fBuObNtQ9&{c}a7Jkw
zu@z%R_~8%?IxjQ9LBPW_Z@6@pr))GfwF0{W{q-Dk8X9M_v)aWMk7Hjwq!47`s7zn8
zE3;)#Ahv5v<*Z1zg){%Wn;vb1;h)4+CVUQ&=AIf0tOKU1L8*DyTUB2m_a|3QP;kP0
z1ZOg<)t8KtXd4rUk<?EA&N_fEfi7uJ@e!yms4hjyeZG^Aiu~}0rOB_kW5xKvv&@!U
zFz)hUObXr(9T)ca+5~1i6mqn%+XwM|Q=V68sPYKqeCB0HYF<{v;V>P>Ie6`%G;iXB
zS|E4QKNt8BE`wD|4Q&#D9BuM%b6+wZQZRLr+!}j)D6GyOb3K?>`iQ0oac&24HJuP1
zvs^i4Vus%DUUGpPLN#t-%0fbMwAe1{pt1pZao?JAP`e;nXc}&7H~Ub39K_-hFR3p3
zP`4f@dq^uhw(S=NjrTo+{?gr1=<(erW@i@>&m&;n<o^c-^mt%(?AFQjI&Ej4fM<)6
zHlSg5H>gF}xH5>;Ff1O$lWETDwerDxs7ptBf<o39loaU_cS?}84~-pZLJl0l6Ncss
zK+_eQXhC7nSZPAo7eO&Vq9|3gFfbYvoIxN1f`SXTzFB|n2L`n)LKuRu;DB}Vij_d3
zfdc`%&|zHTRtzaYQ)xp;)gjTK&)|p!bWjvwVb}=nHzL$q<qEc`h`Rf)tG8oT+S#1L
zZhLJKhxzbKLYyr-=Pk9h6y&{w1=ngXy&k*Aib&K3v<owMi{?eJ5hx1HZDIM$Ydl#w
z+5B`bB<eg-GoL!pSWcctCBgP7SC9HVOMrui{mM+0(zA|LVf&x5FMB5Qv6+x(ODWB=
z;KKFn-;?$w4VBW+HAU~z%`$z`LpdYmN2Zk*D=xP+B)Vw?g09lBU|6EA=-^s2sd=eN
z1`m7J2I#wne*33r?y<JPYm5~U3?V`DTM--?q3Psv5&zE%Fn<{=Pa1l+3pJ%COkj>4
zF<RKkz~9&^Y+pyqCwUjPl0ho$zEYeVo|LjKEKT0<m4INOEzu$%6O3~xHdPZG)RHN|
zRlTlZHGlq(SNTH$1&U)Dl?vSb*<6rU@M=Zj98F33JR-vlZiU`!y-5st;IOc5t5P#X
z{3DhPV?Ox<IN{;_L-_a7=xV+kT>vY`B$3BG{(VdQGRwV3|D0a&jR7c}YwI|@_uRA&
z@wNPxxtJ-UH>s1PhDynBx|z*A7NI5rep$dXywoyL)IX<&p`0CRi#iT<E=BV8=G#J6
zahj77mhn};xG~d%=#y=`tP|NUOcZ}_fMQ8J3Sf=NVKzXLeewRN2><Q>3@B+A`3ojc
zG_b>FTl)HrbVS`-6~sl}B$Xr5KF=ZxqAs#lxl#aV@I`dC<9J2<l_$4>Jhomj57U*9
zAk^&dyJAdH!w`Ohln2ARyu37rQH6T1{a;bbne4MZ`8hO=6Gc7Xb!@Hm`A`<%@h5nv
zKg^~^s6r4*lX`<PA_s(8fi0>^B261O?!#Ux1+=RU|J}CjD-XQpDh`?!C1w0XJ5(gj
zbJ|v6XM!xoOv~3dTb#)zJ!@q(k>)3hg?|Q4>HEFY67UP0=_e0`vl1q2icHp_!)sJh
zvG1=&w122GZbb7{)44-zw_=sq&M;N=U-i!Eu<IoQ$sW^fD_bn(01H$BU!&Msl&p+^
zn_gcsFCykYsXiuA^|ni%HS#f}b)z>}!U=r(huj~1>4>U|1NoKmVUx%|5oYg)OMhrA
z+12*LMZ+QNrL-~Q_z!kX5lsS!7sG#J1q1JO9xq~!jiQf{tRq{4``%@@{;llNVX+u9
zuD{0_PvV43IjQk60eLiu@3SD(-S4uFnEXQ{)~nj|Y08*s{<uF7LS6aBnPI;lBG5ro
zem}Z=8<wp!s5<HYj#eE6k<7Rdh{HU2!pByX{oTy2ZUTEfEj~LpmqU(UthakyT0@;1
zH|eg+x~U9{dFsh6Z8l<qr!Q@Ur6??(+!wp~I;wV}Ard1H6L5rm24is8rvSlvw(jb;
zKf0JUOeri+TT|0qNBA?#u_mI&9ZEKLSUOy18(Ozd+noF}T&=6^`<dCf@2K+KJH&%~
z#x|+6w!^_kL>*ihh_?0JCPUv)yl6nQa%r+p@3(k)0Rnh?7r6(05^nx6O{*ZoWebBK
z%Lh9}U77ARbs&QK@_;_IpJH8Lx%D;Z6oeyk{!tq%Ml|<i8*X9Q@kt_n5RQn_j%n7z
z(mD9MvbJLzk=6pQF-2^s_DTt~bM#<<9XmM4dtuR!5%j+aJIvJ?L&twAc_px?ot8?$
z<xMP3u(7#R6-_L+`H$X1qNrWGs`ePpdpRc9bRpAVVF5KC?U)N#wY^Z-p%hEple*l(
z(3AE&=L4K;7!dNQ-gaS#_GjO2!BbmR8mx+nhKAXPZ(%Eg6QxV~CTXT~ztz|^bObqm
z?%<HevBfY&IlCLkyB@9BYu=Beecwh4_N-j2Sp3eO;dV`~l82?}f!9F2O>JMd?PzV3
z%Ug^)3J;({V7(AcubFWFyqwCdoF>G#sD6|ct&1?;cK5yn$|tI{<_%7#FVFp$X>!Ch
z=Ks(LOA`2&c5S#$1cemnj;Dw?{!3!>Dwyl0sYk=RjHZ|1zAjfL);p$w$G$d)_fyuS
z!WF+8gLY1&GcOQU>G7+|->O1mpAsctu+P3C$p#z_qR<b2w8#Z3?x@&jW^1^l(OwpS
zv7j*NzYRz*kT?vEom;(Ek{NnXJKX<d=Dc~5#vvJWlgv`%U)2nX_s7zdm1dsP+W$u(
zH>y97$ap}=R*F;s;Z6j*M}6jSa)DT{N1}}+iVf3lYFogv$wClWfP{g-Hu(Y8+C+}4
zY6v8d;;{AOuu1peYsgn0iS(&Aa`#<M-)B!)P+m2Q50422*4nL@rHRmJ#)HE6K!Y{J
zw_GN}8jG@VSdu%)IZXF~=NyEs^c^NN?dwwhACxM;36wg*O0D^B<8yoT_-nldc5)&4
z^zRq7yHwZ@9rgSA{nhZ9{|`z5{|BbpKoRg8rW@?@PiQ<eZbZYp59pK5H3KCHa@$Vx
zb&GxcmTMl@bMT*BD(1VsXO>lSSbGRUI;DhLZz_8%k1G;CVjQ-IzY%&9>n~@*oi(a-
z$X++%H_`vnqrE66LN37dRJh3Bn+)io96K%HfzdVRV1}ux+u~@?1x%0+;0+4M{+bcB
zy8%w*$RE)CB6m*aaJK2aq((8QPe?0f(icC;1b*(2V!!>i@``En(S;(=APwvrz@!;h
ztP042v!~;s+$jmlqARfl=UyCP$8Rf7$!RUebW}(FMi>0Ge`c^sDd|bkC;s25%u{tY
zSq9y@ITXz`V@!ZkJPW%;tL^&akDwn-KKAv99H^olSJbX$5a+pe18fRcWqFTy)KPEA
zn#UL4c1aJEB6nD8P<~{AuDf(vrh4cnD<mYr6rrnh3c|)3Qjohyghi{Z5bidCZ-}^h
z5^^D`3(n=#mj(|SK2pKAvbW)YQ!)^Jdds!!WoNsalm>zYUeb%cb)3<HAO=8*2XKu;
zhEmfN`_+O}k^Z6ddP2qmK`kc(L&93RQ}|o5>9AbAA{t<#M2&7dv}7IKeqxs`>jg6z
z<)>OSe@Ul#wW_Lr3Vct_(l*{#oag6%t6D00{^|q$<KaU;2hTlD{Z<ap{rhEEt$eD9
zrTWP|#0cPBJ*KP^y=`>28<#3Z7nOGy{5?W%zkFD`!2fqG@;R%Y2Z&bqiT51?U8~QN
zAI@b=uEJRQQO@{CMLR6kMT+A$M#%)`N>Mj5LE%B&tXZ4zqada%5Fx=)o`%f6hv4ch
zHCsRWc2Kx{u%#7Z&bBalf<eZ59oCi7aKw!hcL7F!61V6bhkwBtmM9>+Oxh|$6Hec#
zBwXz(c?({w+ieu*`lYbA!j825Cgm5Hqb|<<Kv{Cz+Ho{q%k+sbtDmR;#%<T{|CHzO
zF3WJgcmZLa_BS3y0W8r{WDu$lf^1GGH$$?}=F#Bzthy#2SgOAA+~b4C5cAbhsvn8@
zKmeetw8KBkog7r%ao&8tJ*`yit-2llzRv(3|Gd<^Ja;GFTxxH!d9(HS_9@>cGCaja
z#Av3wcV(#jI+Ei&uY&cJ8?Co(Fc;hQ#(nnZ{Zg;xGwsT}XaCCRy4j`SWomuy^ILn)
zin-U_C<9LxkHR|AG}VqR*oqmsJ8s5zC4OLwyc1nZ1EG@>JcU{5?lX~@-@_6QB8Nm8
zAALRAc11mUvV?Ea8l|&0Li7T`^rW9=pP{y=;()5;vRnb5z3gT!^|cDDUYp?(Z7S1q
z(q4tpIx$k|HD>EzwTorr{D!3QaZdd7mrh0SA__DF4lLy;nzA9qg132ik>%Jg|L=fY
zi5bQ_SUbrR&N*LQPOt1)LK#+H0d5OOk~k(_mM)ChtM3C_;WRg!0(43lP;I<K19F}$
z7{>gt6{2Meg@s%AnWaXBhqCDlRSjAld6-!{VX%NeZJ{EZy+yMoK~4Am9KVMHNV|M?
zT|6%jAW%*E`|6FSq$UsEYD4bnzyb*A8yxvtl3t0ik4#|AF0{DMq_OYu9})lY81JQQ
z>7y>kWk?rQzNDuAaTwU7rv37L9U9(Y>7IWNJO9SJ_mA1$BAw?ij$hqm+6@$a`b?{d
zlkN?Vuv1KSEvP7Q$9lZ?I3GZI#&X92NR<HpILs|oNrJ#?)gMQ_Z*TzOFH?IG$Z_xk
z2<d(XiT+<=PBIHBGi4dS3wnsLj7ci799Yohk|@M%nvh3|&B620pfMeQkU+Uc)H^0h
zW?|S1;dBgz3Oi8V2o7P}EZvwiD@lu=`B|mY&B^jx5m&2{EZ%5^|Ehnas}u3_GAFrn
zZl_Mm7b6#kyl_RPFct&j@Lz>9S@<#cgk;f4sCW4`((4rl^_MkJp_xdM!GMH;XzG3u
zd%`%BOL$f__uIzHxj?2|QN@Bc-0vo>T?NFtTxHEp_yt=nSO>zPD39;^edv>W7fHI&
zjozyL9SP2J44L7Q=1ZKBkNvERYJ@Q&HQjauND68!vV9brn-xH?EM0~yUe|P!yLH9p
z<>TD(_L^q5RPk2E-%_M=#kIEcqVFv(vB|*3V)fcPFh6rgji1wzCB&_#Nq!_=@uzYL
zV$V7iTuL&vRp+}ItnW|M;sB^Mai^FnhcxL>l^)&O*W|hb>mXQ*;*XnWzJwS(dv=62
z7_6>8GOtWXFxbGm3dKGX=L&%$O(BHx(okvWI;oVNH^%I3<`Q20U~T#)eFoj{MW3JA
zNOR{G^vf-QPqJv1CF1$IW%#&*i46_815xDSqgo_ycRVTUD-EMI+nWrEo4&IemB$-I
zG?mlVw*(|&<`RPo+o6VcLpE{f-9k0wM&E3Haz-UEpC$nFu^%Z4j~xLT7w_<aC%w%)
zH8W;&)~DpMX~LRoQS#+~L{+Tfqc_sXAP$|%#!Wm32?jB3ggRbOAhLVrp&I^~T_5ha
zr{+8x!91lYB9xU!-jV&L&$H1#G)BZdKUAy8(6ARrGmS~+{)y^PzWIl_EXJ9u02}HE
zeXOW&K{OMH)tG0YDI7~^1Afl&E({g2UAw48+PJDZIrNk-hB~{u`!iX>cBINI&^rTz
zT0>w`;ujF-ap*~?vXw^Qnpzy~m9C*a%<hS9VKa7FAV2t+@@P6W`5m1wbUOc6QXg(D
z&eo(%lt#(-#4v}6=~$Cez)^mOBc2Sqiis>P2W<ZAh+u-jZ8xUcHD&b=qcn!h+rvVw
zqy2h}3znOS&zzRC@{MILTvKKU)amri;4I1#M2QA%^31uN#y2={lOP2Y>Z*B}_lRhu
zYg|OKEyQ9tS;MF(ttDTsxZsa4QX`1#>Plz}2!8#HoidMoJV47Ku3O*m6e)`}l6V+J
z0|-NT=VyPQyeD%>kkfJJMmz?&CiyR;s}jzSixBouG@K858cEr@y|)z*@SgAT;siVn
z48B3-wrK<?OGkpXZ7Eb=n*H=J=j)(}Qt)otMJM$o<0d1QxiQHge(`RYU*)|aSX(Co
z$)g(?IW&2FoYl$n!A}&;43B$fr;B)Z0Q9sOE!jI3Z_!dpb$X|%pZA<(nQD*6x_AP-
zf9#o_`!YJ>59sK{U>ls=-rOWm@&qrZ!&=ulsf3G^t&Y+=(H<h5>nK))sTDKj@HFb+
zyzl*?EZe6rBlcj+UYTqd?*197QrT;m)#j(I|MEosmXANxZSXFr^`@))7G$D@55R74
zdX{I5?@5m>`0^w<2owC>O~y70bjU@U!*+ZkST;-wpAxN|U~j|_y*XoVv>W1bg-X8-
z<(UM{0exyY(`Pro_$_G<$3_3nJ&OLWOFr*Sf+Vj|8ohrJNfg<1NnkL0R%q=;_&>ar
z`@8%9z-_~gjz*|)0`NwiGM-Tw!8DgD|BgjDa@sr$fhD8u(IK*YOTiK03w!E#kqqlC
z%bOsUhr|$zxo4oaBEcD3;KyuVI&8U<@r+KVHKon%LZv&$h7{}?(hTv%Cw%Q-(R|<a
zp_h9%64o|UZZlB=9Zdu*=4FktOY`9c2Vw7b$8y3&ECNr*fTQ0;V3CFdwoe#OVM9*!
zQz0pdUFlGs%(XcGt`{YKQixG9_!TY>NWJz~UL(Li5BxkQV>BHrM))a(<^|o*ZM5|K
zA;~-WHoes`P<dTxMUno<cu{D$)Y6)HigE(g&^=U$%`e3habG}Ow%z?*<@$SMq;UlJ
zWNcDb2!E&|Ki~l3ZSqTgyT?FrE62iOV8K0~cUAxs5z_0~>Ua};w!u0|bjy$G<ZEpV
z=Frgj7XrJDT+J#Bs~l?xbR7#1@q%}o2(J7{<6;u+m*#}hE*Vn7HRFlc;ly*CY>A&8
z8jrEnuuPlPAL^Iugw_=8I=D#PAgBGed9>Y5Xqk;5V6f^v+F52}e5*rz@B5MlJ%v4N
z2aU`InkG9T*d~dSJcIY`AY)+h!H9q9j}GViI#25*QWVA9!G_CtP-|-H(%4r?9w->_
z%}rt_k%Cw(o!~rVa<069tTo+-l`vPKz=J;aW`q`_+BBS<j+?NI)jG7jN>w|q%mA3a
z#$0+L0Ef$*-LA_(p7%E*b^<NIP>Uf_q7DlDF*jseDSMR?_It=k(hqLSrG_|W2<Tuv
z3UdoP!3W*TJh<QV<A`S-d}ZIx3)d9UAFs{?4~ab%x&<}V?)Rl(Pv6`8+(Pa7{1&=8
z5_6l0#lt-fhh!~>ATb-#bPMW2%=KsLpzk0Z08{qAX1R$wAI=XQee+6K$OnCb8@LH%
z&(Z|0cDf&`e;xI#qo5(axL027{$)H)MFc0R1d+5D%^;BiTp16#2$dMA%NdEw!I&r3
zS(}2*k}6uG2`SywY;&^W7ES)DWP&2;XfzjHWE80}{~8W|rj=u|Fj`1nL?|&#w7HH1
z;NgH6X^m=&&QVY^I5IjHz5S}p)t_8RKHs6#jqWxs<f4TFpCF?Zap{ezx5uD%65rZH
zkk#ACwVyVUu+@-`=2GiZ<~M)U@~ZvUdAuf#yy6Z-mxXiwU2a>pD~s{i-!?a%6Zxp&
z0{xksA%*lU9k^xP7QMTD+5&M5wAJec{-UTGo%3d!IfKovpmGd~_Q^EZmX1OWS+8ry
z$tLw{PSxM0*O;sC$OO{-D_5`K{1^eZu5e%U5$qCWkPasK0tM0G`=Szddbix&+nZs8
z_?&%-pWxtqPW?@7)FK;sEi)jf7z8LkBEh9ey2S+7cQDWr@Sy2A8xwmEJHn5E;RyNi
z33-wPEuxYeinnz4JN8Kt$Wvs|N>6Vyir=KLc~OW?{vW~=)=pv!LYQ;n&y_%_QVjGl
zs(xV-Lbe{LaeV+z*v~@Nn;iY7*tygl9*68Dn~pKNBs7Apj|gxiDkT~gExAPKIImue
z1L#p?VhNEs2rkVVEh{t;+>BJ9mzDLB<KLBW68SLh9|VnV9@w^*QzouJ9=}QBZPG;!
zghD>k2!bTyk5{|nclhf4q%&H}R=<XBn>z)C_0r>`Wh8it{7nN&BN`!r&J|CAIoPJ6
zoNUo5GAOF76OEOXga4JEkft^Onzww885&lCBo*Jgly_oi;(qj~8Gtr^jk2Al-S5o@
zC5|RNFf3Z80%yvIzX*~~-;hra$$dBTG};KuV40FAyv(jG<Hk>$_S<ma0{=rJUDF)$
zzT_S9g=gLO-#4~k7Tdom?56s^X!8vaN%|-f^A13_-=Z`3$xbXZm{`^{?|uwiHD9wn
zqdqx)oPU6V9a&&Od;-miydW??L5=(sPNP?M1E+rbyKK)<;u%Ay^*{`cFN1mCUa73I
zHEC}g{@;xBoerzlNoI~<MH-TWYPyK`U>C3A<l|?ku_4G0tlr~3B~x&n`uXa3QSpBA
zS)*I}m#(>;*}6;DNLC6g+^s!mwHBC<Vc`=^a#UHj&QM^Gfu9PZO2E4(;@~<l?T?^>
z{+tSwEBOuHG6VURKPoew19)YMwD9aUil4DW#`x{xF;w0fg}Il3P({vkmi6BP^iJ~j
z=bpyXyffBW)csIDXY3h;XYJqu9@%y!|A+?X;`K{c)xX2%e1pM8$@;^;0^*odism&=
zrlgKbmal-vl&s+%)k6>QX8p*fvW$|$_jKQ$dA~$r%y5zYfcL?20*A_QE|Kz-nRv{=
zrS`unpeKzn_l}LLw){tf%R<*4`)m(+%R%0lZ_0AtikHx2GN+n4mW}z%r{*@>{q!N?
z%Q7oDv+R~izD~MS`OJ5}HooKoQSdfbY9+*1w}4Q{g$+|@&tB7G2_NLSH>!TSevt98
z>f4}V0w;QksDsBo*C4suJ`=Nc5D2t+Zt)NjOh0%|E%B{mgl+#@_?e}GZI<^0SJ$-w
zx^cb=;~P^5-ixQ0wKGpBD2XFU$h&()goTCZ1nRBrqL7FiafEZ()<iPR4}#(uC7wKD
zPT(<dk@=<}K_r4JJt^r5>X+A9Ulez`)jw0~&Zvt%)6X~-{hA{ICcNXnI*LdU_Ch{E
z%a>fRdQf^;W-MM8SVM-8raOh4xtP$PEubxdPWOEFy7qp1)1EhhBj4PZY+n-nWhTr(
z<Yuqg!ByU(B?$^xx3EI6!-v;sj+>Ht9)M)~c8Kuw@59MbD@SPNsPz1#?R<{$PZTPb
z-!y#U>1dJJpoYjpoIu0r=0f5(h@EC@>F{}zBL!x}p!gxPbxCt{RXAxYGVQFkc&Y)K
z>#!q~Srl9tW2xo?o}V%#Xfsj-s+j~2>Es~N0z9pda7i^HPLm|8%r|$q`MbZFK7p!a
z`L@A&nu26`9HkXeTFqCKabMrgnP^|%9&y?<3XUf+)q6ETn)#&g=KCAfS_$M6derN?
zgq>`kR|k<FktgTzchh48Ss%n}P&CbuHRR{n2t&<AeuzWOv{2)0dbU0v1$PB%(P-A;
zctKNi^dkp>`4Mw>qG;mIOGH?xX@J^@87Z#;x9Wt+glTG)$bs<p{I*4`o*TDw2w_##
zCz;v%BB{_dd^q_D;%T`m!bnD38|Kb;)G&SD*dDt=B@d2o@G4}ua;>7L(l9+BCm^m6
zVu=%M5}}|&umutv=WJ_g1s}D{Xtj<egm;^?N(`~>bqky(UvVXn1{=T9<^W{ZK>_VE
zZ1ucW_fT)S9%cPKqTlpYN}8gZ)0>J((#xKE#eWkkl#a1DPjZBM*x9OB{m_{f;Vkn7
zF-BJ2+`Tk*4A}h)w0vInOOcG5)GDS7czXwBY1xXALREx_I@>y`>Ggj-yj`B-MO(eT
zvN}=`_NBxswN&Wc#urpPxdH1YN99h}U)L*2igEY1w-Tf+?%z6Go4K33ksnje*W-XR
zvA_HLoC4DhtlZ$q^&)$r^a|V@7d4t+BlHx9yp^bjgYQSp_n30mp9V_|n3^&TccWpF
z(m}^bFMOTme!~CALxwXkTeUf;g8pghP8F!KKH@RTLU<rx{Z#lEX%GDQPh)|7<&8`f
zhzwQT<a2x3E8FrTLDD56?+Olkn3+d*VVfsrIw-VMxY5uPdt13h`Ttc}215Q*S&HfU
za#7f_?R}gz1gn>lU20(^T?u(${waxD$)hCemX;z4U5U`3Tx0@sz9cYL@dUd$kybuD
zb*x!S+Q0gLS|5LT<XT6i<>!G<!%I0;(ABj(+kOkab`sd56)o&hw2p>RiBjxPD!h4n
zXI1)SgDf1*r{nNqnLN{eXuXHqH9MCO23v$FR`f``4h)4Rlss=q4Nalt_VG|#)qEL2
zD*SWAh^W;AyGlS80{O@6|5%}sY~2?ty!T8rS3mCk;)JjcpbWS<$Gj>(76{&3g8a?f
z7A{KLqA~s_vmm(X13h(tz6KS!IhSRC{F=ef)QB{Ujl6@VJ0#7zfH@TY!E511gNwqV
zSXujZ^wd92uM)f3+W|>5Ujw3(cIe!r;uq^1z5`5obWHDT-8gWkt{O&YmF@L=N(^$9
zwhYnQc5~@kneESDVWy!~6nc>FpAt=!?iai93O11@LtZ|sR4qfJT2ms2@PJ9N9X{0c
zO|QgHrP@71{T$-2jKof9lR4JwHO#%J5^(b6nX*nGiA*sA;@HD!^ca-BhBVs>UI5!&
z9l*kh6Z@fhvXd_$%+A9=<^OmDH(cdkcuoD{OVi7Pb4oN<D3+b}W1RQLzZcHK$RE^)
zXv#WZpR{gvAHyLH^YckkvH~apeR#nLGCcezA?!kzs`Z2UC*OL|dHDTPUrlPOt@<*_
zrgKjcwfbE8ry$>u>@;K>+gAjx+xRXHJb_nF=R~T+Kz5b@dGl`}+!{p67q}4{Z|+~-
z4Gwn#Une8<MUJqiE|V=~n&wIETYl&*=L^0|{WV_pOCQnGWAli#T0bqZDNfb{8MF;l
zFsaPI$KD&2F_K(DujG@!H3mHRAWeb2y2y&%M>Bs~T$vKt<bqV20Z5A?%&L@mU0{{S
zrg`NL#^FV0bD5pvvTlWDMO8`JxqjRq2?nj%ro?OXmSzi{rj5v=O7ZS<smnw`551&P
z+_+HGLxn{G<=VsPxr5vT$i41@^KaeKXVJgW@evh+Rc;U7%@bLd5UhEipY{bqs6*-w
zJaP&r7`?~&q79eqN|!&<&Mr0^w}G~q`fJEymls1Fyu;E6@S?`Fju{=;TVr~X1W^)p
zh>O}j*NS;JpQR5+A35p9<R|VLM3-b9A1d8qC>uu{d!m0r?Nha_67V&*D2?pL9sYlH
zM7wl{Y!0vZ;hkUAUYD!>oX{)&^0eDI=F-6eyGfwQfIND&`4kC2lP?HV;RDK-&iX+~
zn*;{t-?R@I3RJYb-=y4*t(mAu3i6(>u&&ZUDRMzuGu0x=I3gK3JVvwmDoG*`pZ1L=
zxF0Mbrzm;$tL0+U6m78L2v%%q<gWPJCB!dw?$W$)=l(06s=vcUaDEor-$K|dHJGCl
z5iUpm?>3+6A)P1why@%Ft>d=ayGt<VVj0%YGBqX)>#!w93n_5mc(y4T3L*9Yv$s^+
zMozvL?6bvlFi^{#fY%Tc-n~80Wa8S(HTgl0+puC%dA<O$htEmz+cs==+v2qDNW3i>
zfB%|yYBn8-h6EhD>b7z==snH8fRKYu7WeFGlK02Nr)JXn;q5&zO#4pOE0WE7dKu5t
zo4y{p>+Dj^+bl7*>Lf`^AMleW+4C<slH7u$2*fPt3CvhR=reAkkUA$rFR5(C6Huq_
z&O(Gd;%U2B?OoGC6j&;`cj`Lkqqm1R9oV^wwd#mQr1&!WjtYrbQ4)R4Qp^OxRghl1
z*j=t)Fw@gigOLloVjD!xF2utVw9Ceo&9&YCtm=f|OKJ0;kah1@<MshB{~2(DEH`Se
z+4lQC1;}-=+ulba2k+TFz?0~hIo~9g^A&vYOmIQ7+JXNAIb4YA`F>@jS*f01Sdd?E
zTTup;VJNwv$tjvHezd>Rc8J7!2KmqbiyWQ+PVBHm(caP_Y0x;u7t52m2j{p3%<O$M
zG<QblZXOMjr5rcjh|;=jQ2Vgz2TpGd5yBvAVi@{%(JcQ)ixFapd;>SmzrPN&99?$*
zz*G^fCV=fuFCdIPu`A&y71RdfaD(|3^mL=HD6iGlCS=*5Q!OLLc4}9;;HlLL(>D#2
zCV*2M1VE7Y5TkV;os&+2NHzCH@Hl$<FSxpg5Yu4mEWC<P<t;2|lEbb45GKcvVZs|T
zl3-1sPdZrhmvDOg_aI(H+oKSLB9w@qShsv=G`Tao3)$F0sMjtIe6!88n1H{X*Zb8<
z{$M&$X&ewnRl$Xdmnm{RES`jZhtvk7#+tM+GT!NB>N_$Z+$m+8!1MGtB|IZ3`r;hq
zfSQW%AtT&vmDY9H49DnmejV6A3YW##PB3Q~#oxL6ZUh_LnU-#~4Id=BH8xa8S6aNU
zqsUQ^+R5&sCS;RK<sFgn3q?2gj4(l^V33{*@J#gB_n%y*V~qc}>L9%6!UsmAY~hVo
ztfiKly{peU77JNrhh>cT90ev(m-23-FBgRbP(#XXbvFpn7Rgqq&Y@<!!8retq-!04
zZ(?ULeFH?BgBio+>^SNtoh;LuH5d(<C(M`!Otp<<p_9yt*`7p6f<??>0=PUMa**pm
zb88Qxv=<fi*>D+0aLnBXk^u+pivpFuGET4~5d7DPu(m1Rnt#Ty&H7+^*GN4WV5Z5E
zqcCs{nIMj9j4X>cH;4!%+_gfMwa)N)c@W72q2jAGxE^+TlYx4X?=rJbc}yRJgfM`J
zrbEaE$V2Ub!TzMrh*Yec7V|!+R>lr|4}8Z(GsZK$zWSPpztQF3)(-rWsTk>pZ+KNw
zyIuG#Yo3KvUx@FbE}esQ_6M26=O4P>Rr~WFn8P;WbC0(sdTyOTaOvFyRjNVh!({C9
zpcfw38t{MwP?8Y-5>LDV66Ak-#k$ejB0VpIQ)GXAKWXUlQatha^G|w*WZU!vp-4bb
zKwvUJz+5EDkj1(Z7-#_400}6Pf>`I76*jCm8e;1wbRuPcNkAMg+WoA(H$&XkX}1MC
z+w`@??a7{FocJuG`3PzMms8tqo)yNJ2V*x^MB>?h7dlu~&iXJu*h7IV`*Lu_UV03{
z^h#_LD`)Q2U9G6Z6yG>*(u<Mw9XC6(rRb?aQq?voO1(kLwSeVlU6p6?$5x%Gk9uK)
z!ta;AODidC8*~J#a$#!I!TQCQ^iy#qXxD^0*w75@$N~bd*TwSQQQE$|S;peWJqA@v
z%2<AQo1tM$ZviNs<b)_e?it81#+{CpZyDNYl>~nJ@6LC479%v=)V^<2&t|lPQCEg;
z?MkUGD4jH}VZghA#-F%N0u{ylOdJ&B!G(nk1q+ud*C}^rTRf42G7u+R?d~!Sg{qj*
zDt}4@LM$zMwvw-GG;PbYnYGd5&jBiD7Bg9G8~5}{AX|Brr=A)@*Tk0K@Y8r??lU8g
z=X0n#6*0fV^^+Jv0LBxEJDNv*VziA+5MEJB>2)3?3W(wVbsLTyE^{}^_&bjXD7#;q
zrhFD~gl?u8Wil&v{ufvghWIqq@ztv@JL5GscH#GyX_6>6=0kp+7-;2h5?Lm=M3Nv>
zj>sHO*ZGVdefNWC^I7q+(pyK6pLXUcgI!7iT=BZ+oBM+MEl6sk_4!)D!5z&L^Q0+H
ze}v+8CctO^Ji$tYW?EmNEa6rr%i5$(FA=kXU7wdr+dFwY%RTKbwYziwG_=kS!ZR8G
z(Ofnbl=4`{$fr&htAa7#k(|iLx1fI0rUy4sgo~;#YC!vqwve|jL~FUioB%rQOkI3+
zC{voxKbLUQ(39W98~L7D+kA4tuZv>J%gAzS4}fy2MVw=phmYj1fQFGrHd>Zptj{nF
zFcqoqm1!nKTX#ew@zp|^7Seeg43T$r@o<C8hn$T9)fs^p1v4Gu#w?3Rf`Ni)NShGS
zs(UmXDfq_F9X%BfngWJE+bV}!yTrZXZ6qregG4{khJ-sK*Pv!in{QYTz6tx~P9?8#
z1YjNEomZ4Sn9Y7U|IY3h>0gCvxVJ<VV89DziGze>)ROK^*>j=1u2j`1qFRB0(;ZJc
zRD)eXTTz2S0e*!rDzPv_L->L9#x7A4>y)}l25g}byV>>5F~qZhmOzwXNM{M4+8C};
zlpx?c(!<$XoOb;#&x*12_0g$9dv|F5%3rs&w|$OSC2Ctc(kO3?Ll2e0n}{BY<8vFG
zJdjgWE^a)Ew>ct^6W@rBjozlD*#47gO1VvFCXG)+b0a2fR0s|VX<#|8QEDRggRtzs
zS60*SgKVGiWg;>K*y9NEKKd>MIU63=7LV5Qyi#ll53^P}7jiO-`Ny|H<J8E2_ZCsp
zzWh*w9`vQ%@8)-;@l@+3+4E-<kS(!uj$k%eG~F0%b63A`>&tUj*=%Ev&xMz^rvg{K
z3d+G5*jd4n=Lw;|U=)E@IgdXhO{3<QZmr1Ka0ojGGWt6(MleX8zua7THk(;bQ>8>O
z!fmG;0L6V9cJg9mY_<jZYFsElc$bgb5fEl7hA{p502?k8H*&NfdIS?Yw$f;jAo-DQ
zja6Xk`;}vG?d9{tVWXgW6C#zk_(X1tpb7tH{>-WPcSJRliUwxuiB11fFupK-`{*1)
zmRKJNldG#h#tCzqFR0bicpF}mcd-+dKqzs9s8;uzyj=KKvSa{i2?PP4;{Vi|jyCtJ
zhX2?;ZTicpL|1woGCVr9K+R`%Li!%T>qM+Utf)w7`Jc9=2a1wRjezTUmTqEgS5`sd
zY(U?VDN6}bJi@Xzy8fH@yq0k15D}6JcGZD=UhR~L?})uQtUJdIvHrwCOU^xFVTvek
zis{wW<p$~GsPgQH<rLukZab|-nvWsyF5f4b@feAN=kxJ8XCl7UycwS#_@t5_^V@iQ
z?q6iagTV^;1MNBix-|a0u+{kyyMCiMD9YpUa`5fHN;O`=NHCKdqySZ%@YWLHeeA4L
z2qaiz0Y;MH#Fn+$|B4}|oM`fWFLnthxA)4f%3)fmZ^(W*yFmXxWTFm<dX_;p$tCKh
ztI`ge!xChru7eQs_K*JPZA=u-CdNx2P+re`bW<G-3Eh{_j2Ebsh13$VH+}WryDlb4
zQNg##UJ$14>^~&zf`zR4B7XZ=gCRtplt>YJTc^yHGMrHgsA{v3EU=M&$dCnGfFK4f
zz8SQ_yhf!a?*R;ZV0AFvLdTYAo8?c$)^vv&Z&v0Vf;QsEiLj8**v(g7qr?@dR_oR9
ziSv5*+ruV7C1FxSK_212C6tL|z-6(C57yZRg89XmmuH0GS&g@rZg>TmwPCPT0qDI(
z74%+PAvBp2*Di&}81<uC1upJ=;DF`!2h5Di9v#J!BjC!pJ}z1WA82&|=P*NH_y=Ta
zUCs-8vmwv$JIEuZRHuKkH(_R9`0CUE;s-jp@{tR5f5}$l*$MLrs_@+3N?gWARV?CC
zYm%GizFr;{*fNv#Q#2D92DZ*2<X`rv|6&0ZvDK4w5?n|<6%jq1wSoQmPE!^GQl{JI
z+SY#;)4*7_NSpIkQY+Nb=JhSSia7?p1ZDHo2VPoX+^flLs73(Pc7&wsEKJ4pUe~{E
zZpYx)i`q1u6(oK|=UVfXJ_o{#Nsv5PS9b{w5uHSgwu8_9!I$_@gil;B@cwMA&lrRU
zk7eb*U(WSMHj$^Ao<coJ;Q{49@?O8Fb_Ws9NC1|2`aCnb<u&lf^g9MRg0$Wa0cq+^
zZ*@Di7yF-2i))*7bm^a5pK`H2#$Z<h9EQEbginyeh3nUCPCYBRIc=xB;Q?TcSZL(1
z9bG<uw`=MtSioQ*q{P0-ja+eHfzi|FO^#2F#|wzhqRv~Cl$MM9P;wnlAvh>dc$7=?
zqXcv#ybe#R-wS*;EW6*0Fg4I`g_(=Yg^n-CX^S=^avN@PT|8SnIgr%{$hf`2PV25z
znpU{RKB#!wE`r;zr*OrdY{T_|!nm9tF<)H{YZ>8!?{q5e30dG-7kZP8BFV&OmX-7D
zt(a?q@nzNo1pRsL%)Qt6=Yk02=GwSlb|i-7amiu^q!gAy5xvbi^HwmrXFqmJ*#dUO
zswZzeORGml`T4N({6H73P!3sWGzx=Fv5LgAiMS6RW4?vy84X_FaKzsmD)3k_hd~nC
z;P#k3<*rA)$O!0iOdQ<)0enPm|Cac=y>#pe2w{KigDZon=$6DOC*yXwo;?Svo&lkd
zBp;)(1~kx4iu!<F@@kvD-yahP;f@Tehih2%nn{GGJPA?Hz>l34rp-2#kSVV@Xio&9
zO@equT^Aak0lzYsNczVKwaXBjHW|JJfPcAQi}GWtOq|s+Ap)m86wjG<BAVUk6&1~I
zaSBvD(mRl&px>e#?gA)?3_r8-0nlsI><l6`EYbP+rKPSSZTG7bo=hFw%RIus{~a82
zIC>9M3cXEkY5#U>ve<+;njS6yLF-oX`_7qz&8pC2?`?nxePkp9nw_jfKO*hqrWtLb
z?MKwD3C>rAzvW3SQp}tn;z#0w9yOgV!JACY?*dAn;820N_pM4t80$np5YQs+2ZO0Z
z|8{4rqnr;hlFDnSv1$sj$_qZ;v~?~JF6+<_9yu%4$CMy{Nc`CBEY<!SREcHYSwVL#
zDeh8bV9v7!b~Yrz9=si6yvp^gb>!qCUKDLscc28-5-~1IywFDZYm$dRLK$vz5uScn
z3G~yY%I1(pw9%bKx*+X+Hz0~wv>It%nru*as8}Edu3k%?28pkJxl$tJ)NA4%#(YNq
zKv*jf+G?oJhMo1mNjSc)*MRy!pFK5c%aBKTW^OcHIa`d}2F+vlCr1nG7wKbqcSw}(
zqzcgDo!n&fi^mw7Ye_*$Xsm#!%RT2Q-M#6<Cf)l$dj4iuLswLp31A-Z`6^V!g_~8B
zC$SDnITFmS$~$u_s*ig1-l_kTfvZB7C^UD@IhTQ{cYjk<R;uRksbq(rbuqfMn^EZ0
zMwpKlvo6_G5{f9RHzqVr#PPDnu;*gGw&jiWNP%bD{PB9czOhY{0KBvxwpPvUgk`9M
zB67YO;Fs8XI@Y*q0N>D5Ih~jtNC{1cz4NnDL4=;P<=y$;MApXN#XC|M(&oUngwGEf
z0=~6?D{lw=PUdE2Rkci;XhPR8l%=c9^GZF;(DiIvc_S~gKzCT?h^3!=##L&@$04J1
z`qir4%%qXJPE^?sw?`q*bNjnZKfg1@om>Qkgh?dV>mxlJ8Q79YZBx1$ip#nOIaGFd
zKlcl*MV)nW^Kx~4#hQ0QoRq6`*6@)($qSD8bVtIyR)aC6aB2WsmL?%!-9GOju3jI<
zeEsh5@A}PxQmp2DrGT`^feMd<LZBeWl_A#dC+b@W6A3|Ij3HKFzLgSEhzhpseV=Yr
zKpi%Pfuv=V8c>y@U-1orkh5v|;MibCf$3J#k&<A7I%VRg2>BY5vw{At-W}0Q=#m@v
zsYB{DSnTM_VD#xPzFe=WEV{oJN7GPmn0pa@o$zzMmYi2L4!Rz{R=eNj@Q|O!_op7;
z-(&Aq{Ik0wmzI61m?uNGxsQx0wLN4yl-oj5@Uha0(SSDpR{^H9l%ztQyRlt?{OO`J
z=w{Ym%PB0R)AK!%;84BU(S>thNg(uchxN@xa9p$|16OYmA+U$fZ_5tcs$%_h<YMDA
znVt~|hPmfwUAjEE2jAyi)++Vs$U+HrFm$y-LI*>pWw|x^=rNIH%e4gP`=YJuPPQ$F
zjtMk;<$(3`&uooi6Kn;hgm~qTHb}M`+Wg3_7u!GOtXi6k^dvRrtG>qu_{w9Gu^gV^
zC-A>OU{2uknNhd}Pv9duPz#w!Fgj+jQG%7fmID=DKSZ=j^l#`W$2`efG`XaaRb0Hj
zY7^@yam9$t7ZJ-I<&Mq;TClbta@WmKtVkSuApQQ0m8KqLbnRyJYkn}1cDgQG41=L5
zC^<Z;RsX2i+arw>7KI_H6xQGlx)fH3q55}aTuHLwihURAmTL2qm$){Tc%nY|XB_NJ
z0A2vMT60{pRI?#Eox3yO^|u-y9sKoSwBDslv7V6chwtWFCu=|Jlo9CUh^F+=(f`HO
zJ1}S3bwQ)Ct%+^hwr$(C@7T6&O+2wDnb@{%`{a4QI;isxuDWWk-MxCP-CgNx2zXMV
zD&fIh1fb=393<OLWZ#De<4PWAjJ<#(I3++*@fW|PaT%xdbHd(Pu`Oqnw`f-qzljsp
zEA3}w8YV=GFGNyFkUUVsMKArEsu>=2Y9dM~5lSZJDGnmT{e2LLW<SBSldepYiah|S
zV2-TJ4==)>Fcr33td1OPbbp^Q4It7Dz?I<|p>^<Pkj}vT$cjT1^N;LjaCk8KM@8O9
zV|I}RUk%#X=JfRW4ob54_4U~#{dm3odxBnIHUu{7p0jlVFh~CZLEy8%HZpr^@}9_5
zt7jRE^(9_-`RC`sY^D;Wj-g#wI2yP_hc%C#^+Sc2&-Mavx2gFB>i13>BtSyqaDO~Z
zGIzf#m_!c`ZMclCX01~bwH70mfyjCTpiC-)Zzhg;jE4oCE1ij~E`>Zg1@n93ek{@a
zU?G3?i_E@GL>%)dqmTYk0;!W9Vs)v!#rd`bE}s7nW_sy<BZ?-|QEx8J#($s$4DsB<
zjNk`aj`P>LVL*j4mk4hb0MX^)U*K+Juj@4y;?V?-u|h)a$ah$RjWkR%*CrtW=Z01&
z&?sAf-$2Atzs|83P)~IN1H^q7edYU>8k0FL70tCsf0A{1AYL_(hhexT9`}V=E&h9k
z1&=f5kV5SBBOj3V^>&_sdvFI?L5JKL*{9wDf!ojusrALZ%u$;NsA>YV``Pxcg;I67
z+>C3stOOtIbUWvL*05WcZaWD_Gi<oAy@O2!8xc-7zYGxkXzt>taY`K&a0%eq+z2L3
z6bFeMB?dBpLR-i&_L^;1$`4!n-+#u~jmDvi+YJBu)FpJx``ht**w|)HV^5sknSX<X
zI}T0<NK1zMh+<6xI9wA!Y<{fUv*vgag7B9>rfTCcRoiCjPqtAfi(FoIf}1d98bgt{
zv0<YV^;CeN452b#LGsvUAD^=$K@gPUCbxTo@)<-N35P+$vjicVD!n1Yzg(OTdUZbx
zmH`b)3<RnB)CQ5kd1{~!q)4)85~F{exIY~3ZP}VfgE9#KZT!ZHz)%rLJQb89liE(e
zP`eGF-<P;hTcypwz=M&{!w9(C*x7lDy(uSgZ6X_gLppn`4x3SC;h+>JHl{y%RRVhb
zv;-o0!EVbRmzx(-ZNbxG<vGNdmn~K-k~}5F%@G@u`Y{e0Jrkyr?_HF2tNg|=mFEhQ
z#}3z^&q`<ktHztS-<JuNrK~R?Z4~A}fN!CFg1|6Akpd~u<_VowqEY$(SskG2dDO=E
znfsIhlV`&5n93ku&X+K_s*M=%I~Fvr|FLu4B!=|88Pv(_C*pdsZ%+O8Z?mV-09(UO
z_*oNm0{E>P|FQP_NrX6VzpA@PGJu$(7{GlA<^d=>&|K7kI{SGfgadl@w>S0v@DCDn
zd>19KD{1%sI2q{}q%M7~dh7#becwag<-oT}cW@usu_8FXUL%1D9)psR%0vGc1?OZ(
z{q@-Ibn3YQUB*<jNE#v|Jz$6Mmh_{axXXS1(@@I%_w@sp_{w+mUU6czQ1tQyO~0xe
zzX5Q{(xRH{X@$n_v+Y&5V3;ow%|~Jlq$eVq6^~>2UWScyL_XO|!GrA-!MOpB&<&r(
zwAJj5d<KQ=5z)DMDlmj^%;0rViP0(I5;ny6Ym92y<fcTm%3hK6v^UperZWWu?#(kO
zPCN(`w^Cfe4aPT3_@SsA{GuDy9CK+5XaayxAYvgbLZB?@;CMhK{Rtzz<U=N7OtcpK
zad9$>>Ts8(O@>q%1>e%A9-RkHhZID8GTI5>?0(HoxJA0|_J&F#EJWT>d_vJhV2gZ5
z%MP|?RjW<6h-A5{W_bSyma>_KQ=KN~g768fGFZm-YjZ=g$!_>&Lx+1t*KsHxEd?Mz
z+SExjNSQC7X^TPCoit&-!3iI*^v1R5`FOX5k;I;?7mFr3JLr>gC;$76W@t&rZO&qv
zJ$03riDJrg&vZt6Dy^UlI?2Yx@R@RC)TSliiea3p?_EFfvqqHBw+Cl(Jv)0HAc*f$
zmVlzvb#^(8m@wUq2{szD)PB>r_X|*&9<S)bJVqIvY-*3))hc3K0S+5R0v^jU2{rFm
zN@sA^d%K2lTFN_AH%$rAvv7?;UVahy#AgFmabUffL{q8-XltkFU39=lQ+Cip{w*J~
z(VEiq6P<yKa#_#hQ=k=hU}gdPnS5RNp5u~;&@!7Fz2R@bIX)H>)FvtBiw#KrBo)V8
zswc<Xifal(pT4`zIQu*!6ox;i?`#=ovXqE{gy`u=Rk^M-?5C!lU6h5H0&G%OLFX%T
zv^FPNclUTSCp)VtMFk?I_{IpHz(w{wA_qq);=qtgIu}f|_!*$mMW)_py0S_GNefnh
z2}0mT0YwRhi3JqrKM^us-~ce_OG$5V+qA;d@h@G;14dXd=f@?UwBs-KOo4t8;LR_P
ztlUuYU?t?jN|(RR3kbT~hUIS*eJ=QQ?uEYi<Lj79lo#^`!ovE%ySnG=XKf*B>>M#x
zUL2)u-ST6TSJkhTQ>c|XQO{U^+q!>Abs|tW#&c65%oX7ZLt&$Ne*>J!aM7ZId+%E?
znz*Xo|7owy=0V%R{^JS+Y>5wICL^~rUja|cpEVw9uno#6?QzxUI<A5naNFehI{%n_
zi(v_3AxS_4!CNRWsyis8w~U>d(M?@iNf`HRKY?}1xX!4pL%=^yOhC?aLn&+7&)d4!
zhIGKVrIx(2xr#_east}2>b+D|Q>-D(NQ)#cYU4SBiXUBE_C-XIGXfHC)7r~2ggKXJ
z(}sq7tf1K_@HuLm?X+>QR+yAqis?8di=hT#S&LLm2qSL$`j#hAH)kl}j=Xl>B0#{(
zIC9w8ycdeP9nJ#uwN?H02vSCLU4<32XEa4kkYoRPViD#yv;k@qLUzHjqAaQHANtla
zqcG)v*)>s({!*|gshM-%EVQoF3m<5ad_yP|^roU5rw3ON9SQSVF0RVMWs5sAh8eI2
zCv?=C`+UedzqW^+ca@~v4AmvNi{{1#IN$#ok9M%%O}oNpkb19*NUwa-y+X|_c-lZF
zvbPxjdyJ7xLj*8;=}$hN7UgZp#H{90t4ocJ(wSu-E*E9uPphr8GLfkSKRqAZ<ZD-z
z)DxVw;%~-&Uj9Spyly_d{i5u!jPop=y|MYHQ0v?-c>|y%S^cFJxa`K<ZGaG_Py0|R
zB4g$vL-r{fr`!X|4r{#ItqG-@{s3xf#)j!(*(xgD?hMdX<rHB;LhEc@aoPYwCD}6j
z47CQINZUlk6Z#YZw}2-Xlq?NlQQ6ytKU~|)M9b^%wFyU&AiROow|+7n(4bYll=ShK
zh@4fa)o^;w7cq`AS`M%)?+l^YtAty|re(KI_4a;u0Ry`qI)tszS#3+aC`iW>=q`m1
zn{yJ?bq5^m*9WJXw`lRfGpBr9JBo~U4)uz<ia_HHE&y*dMGj9~&p)cw71|>-*Hw+|
zVVbKG{*+5Ie+-1h&MGOiO*-5!YF2W{L2x)3$ebwzx~eA_aS1e(RydJs#ibvNlPYR#
z)uA!z0L#Pj^YZncJ1BwazeERZ<ns~LC7h<M1c;%SnrhL%a;hp<%}>`=Ti0sy%F}fc
zi^Hdv3&9LkEHkmaYwe)4QSRMIvEXtSsAFKZ_;*-0IBUhXd>_{7S-Q;4S_!ql;^V+h
z;4m9lDroYCMP;nlE*At3un=g7b#Vosds1V&GHe;+7BMmv-R{D!)|hLHP%TEE+5z`Y
z09Xycq&0i=x@g>wlK-5F75esg5>Eu4lOQy=d}WEuzfi5EN^uJufEb7Qf;x?Jo2@fi
z_@j8tq(7td3$=?qSn2H!9G#sWL>%*)><aG*#~ikv5|eiywqp8wLw@%xu0Q^USrCsJ
zZ}_cPS@rQO@(FRqC-5hgd&cFvbSueT1;Cp;8DTU=Qq%|C<Y#xTUbvYwx~DS~6vKW&
zALS67sKi}G$qNk~%pe4XY%de9_}&N<azV;^X%dc<-JQ{p&HV(iH19J)0L(jWjTqYZ
zC`E(5US<y~W-3Y`UmsFzwiHV@X;{|D#k90A4CP^tmEwGDe%VLHP-sYW%FnuM2sniH
zqh;IZ`GTL1)_m;@khbM(q~Okgpp8+pXw$YbE&BX-MW%)zAm&JEYqC(1^REVzO_{RX
z5<^ZAfYmIaI4<@V?P<s-y*1II^B$+HV`&!~?UJZDoHw9=kCeG`{39od6I*J<6Y;VT
z{(|VR%2iDz$ny#?!q@1`FgkhP4B&S(&zj&>sWw@2HPAE1mOi|*o&~C*Iy66ImX!T5
zVPv|cFBOyRD6<bD--q3{#aNwAk?Ksz?gN(r^6eyZi1vqZDEc+pisEKQyq3NhDbh~W
z`u2}n7<Si>!DM3mPX~gM@&xU{!q~`S!o`Wl2z6?8lrmiN{0w`(4gpie4M146T8!ur
z4IIvqQGeUn<bIytK^PUQptlvC=t?;N6^_Wey1WpGfw<p&!im15h{<YhZ=(~T2WLC>
zK9r!XwN)(Ik`}%!zsa+X@?bhM?marQt*oxvB%7zyR_yNGAPKLKC+ED8%5_*&Z5ehR
zu8j@Cy5`|hdo^smm_WHc8c>Lp1@hwHXR2Slz~D8Ts3<5ok#oBw0Y@$TI20aoS7uCj
z62u%KXA4o$4z$_2<#E~Q>P)>gn&F`37Z|_4r_|8X$$6TC2_+BRw-4H|3hHWgRp9h=
z8yFe%dHd5U#cP;*Fc-y?#2c}L)c5LF;^K|4P4LaSz0`2UPsey;0{-2fWW=Lp*Ohb?
zsh;bLW$&asTUsy!-I|(acYLmJ)JmK;NaO?j=+WZ@24uG3Kb(!^f-!$`u6lWi{P!+0
z>}s@I&zWi+@I5E+8S03)MTty!mP{mCTCPa$5jQU(fRe2w5ANUV37&+{%H2lPoL#ur
zoxjO%8M0V>QxNZ!06f>@gG+Fy_GM*FMWs!(CC_`4Bk=ky=a$n|B9ScrC*X?u`C}I*
zmn<j|x5C^wgK=Z+aSw{~K=fx$vwK9YfScUr=f7oNkGGh)^6}ox@4HEZRW+UBO3Aj7
zeX7nB7UjHAgKh+T+~=XiO-~b8sP=uyJG`T}w|!A3`UWtDi#2Z1Y!6asOvRb~Kh0RD
zt=;%_pWqNw&YwJI<;yTC!XY*>8Z4bP$rgP*UFJFxIOMR=zL_$;<c7_l+xel3X{3MX
z6w{HWK1j0PpkiJ4X|cxb{Mod{`P>ziP@&vRc!nRJ|2SYZ%;M%cGnO|qb9#$^FZ
zqiI`!b&?t3J%Zt;?A8k#*QuTzuTR3_m0#8!au-s}YLtU`N;2nlqsdVKvgXgaq0MSN
z$7W=`8{|gF$nN=kn2y=EemN_NzPUH9H<2bh9*)gOgwgQ??SzaM8>@y}p`!Cfg=$9H
zcX(^KmM<DnH(QAxD`NhBXDr^BVk|$Fbc!s19ILo#Jc=8Jm~`OFRka^jl+%UdV&>vX
znRtAp-E}s1K{|$7{tKjt#A#cwm2mhU%G&L^^*^=0rO)woNtQ)xx0_Sn770XOKi{st
z>I8c>DIa-4n^T`eqBmNxZKT;tIOGxL>SZYCj#g2w`ABSgTus%y{>q*p7IqLGw8EnT
zR?Olh!t7s<k8K4gd{1AE6vFK;@Nx%aT9!%J&DaC0ei)ElnUCF5&ppU7@9}Xa^IB5{
z!cz&vmxx=ocm>4<lHOvq@jWI-wxa7OB<W&nOCZzn8`eW<pZ_?Wse8ed@Ma6mW(!$8
z27Mjyl2ePa)L=vkt7zi^F<jq1IU^<jhhV+ELbFe;+tV4HGwd(Gr^B&jz|g!Sn2FdQ
z{*W$b=O7gO>wVg5ov-v4$)X`W%=8=6dx@F>DU%AW|M>EF+SyHQuWuCz(;X0teL*Pj
zXmni=y4}4wJ8D>CkQ`okgNvN0m^;_IVBV&dkjc33l6SKN_ix@2j3uZ)b}8KeydR%3
z$;})XQ|4^XBJPpAm_Yq^f+{n{t~D+ly!glEoU3<>vP=Z2`$Dm)(E#pnLx!gGr)oXM
zw=@`NvT#wBu!XsThcS3i1py$vo5fX!s@o3Tt3)cxM1zId7u3j}iFf_ll|W9&;uykx
zVqM)0^Acz^(NaO|scETG^7Amjq^_1?ft$#WtPCmsowG7k)JxXgkc=sL$l*??0J~;7
zHIc2AJwfS5W&*jWKOJ}gJdaOzgbpq2Y=1cpUI-kinAz$<!g_|b9k%%9n3mT>QOvRF
zEeY3R7#mS;ILZKejM|)G5ET59)7$o4v_Vy}_4NVN?(Y$^5BY3RcWn;PV5S2WvePOp
zRa55cg?kn6UhL?m_<nTsP@qo@P(gyW<Q<1$AzK*jzia=l<)ljUhN$?Zm8{5F+n>*(
zLsEd=mg9?Xf`35rU_y?oF<MO2Ad&N&@r{3ccF7TeWfR0R%o3WkQ}WbDX;t8P#Q@_v
zmrw^uRXB<nozeJ*f!+a-(x^d?ehh)udEQjS+waj?&A%b@kv(6<UoWM#QF#Ol^5W=n
zkAM~h!bt$i4BF<W0&l4Ok5bRl4=tiWx!BWLH@fEZ2c~ws(gy<4Czp`tjA;*mfUxQ5
z^Iq&K=A%;gk4}T&d$h~R;OUGAQ8avg!Abc3Dplv`cXSYLmUIIc2{ReMb>5ffkZieL
zlDw&jb5#pqgwT`J=WN5{lsqD8dAj<-1vDw6gr#at1b%UE{^&g|JT5aF%L)TX%YP=6
zFx`R@i5b#cCzc};J5ofaxSOlxZMka$17@(SJr=GiHu?5&QQdsTq(nCHi|r+T3zQ;K
zw=Zw!V@3~+_QwMjR08D%XeqNoueSlD?67J!I|{T&=?a#Z<Pvz;VV0OhMlbYp+>8;*
zE&9KxQi7=}%R%Ckl>ks6i?A8e)#nwIGqIu^D4YLO^XTFJ$Nnf{L?6ghgxAV?4)meK
zpHU_UT|3fCo#nW!hC<%-lGs~W2dP{$lTQiMiQo;C0Se&QarFh3oIL=7q!mX~XC=()
zDaIJ#W*ik^?6ug@wb+UWrPJ}M2e;Z_X@XRKYCkxP9WCVn)&Eh3J#?n<2jrEtTm%O#
zWJ0h1L^i5~F5A``OLi<0_l;1%er!!yN;bMjRP#==2L%Fr^FYb|Iep5H(rR0PHo2M9
z0|-=3X&r!K1=HqTn%?`6`nDauEC(O%&GJE6J`n*cUDmCZaFTb$Q^Toh6-~f&Thnc%
zNscQ>GSVhdTQ)FyGO3o=XSLu=od$<*B>>($r{zBdXYp$tUP^u^oj2yeJfWyfY%#}G
zWtV$|x`!_XAL<;&H;3hoCPCq{{SZOf1Q;=YwX4~L@!~apdk3EwLTnCB%C)9>bvh|F
zuF9{v+CQONt?b`aaQ<@3|C;ePmo|zKfncjYw&NQEqmB$nTl?sR3yCpgU3?$@r<$^F
zUW<tp`L)i4(k+uITC}|~0fUk*cY!zof7-z3S;1{;n~8AAyk~ocW-IMun{Up;4)6hE
z{_!YZ&#t_!(JBLm5(^*}Oa`W`kOrf=I<DNBHc%Zm{4i5oz?BPqVKzo=NQZk86e9jY
z0=y68kA8)Sb%?C_8QaF~bm&t59Z&W~8vf}aV!YCo1DLk6t}$4Qc^Pli&W~NK;c|=+
z!a(n66ZV8*B+>FZ2b8GsII?m8fNRL0l~A_8^Q5bC#Ms9=IbkTyPyrcBMGstJ;2=hx
z7ZL{vPY&tx`_Es>x-KykFipj{BjD1nN~LQR21P03WQVsoxwRNS(*EhM)IO5F6BHM<
z-%HsJX<XQnAeRux2#Im2&c$m`#=pPuru45-8aePT9`6c96%uyWAw9A?0BMf=QRrQL
zuWsr$M~;1;Oip&coA~PGV)0un|6lZ*<v$`uwItxM*LijmKW~}+pWu<w370t<Xic~w
zE6x^R?mRqsAQ%bFkvtnG%!0MrDewewE^N-j!pG@84y9hrXaq6T#hCHGTGvS0^meC>
zIqB%qVWk0GCVRXQgEadIF=1Vtuz7uRJRW{<_%3W7{Y-ybQ<hdWd_UUJWx`;*hf9a?
z0O6#_mJk!GG*382wqfu3(K=uR1D2krc)TP?lxWk<xQk+$kmLB)^Lf*)JV0>4+w;uP
zPtv12Kqvrw`r$ufd7Qk2;zD<CY6Jkh?fCvaWyNKhf{=P0Eh%vf(=!*MQWjw##-jG^
zOD^2G-90JgwB9C_qVlvp8?~TBgEUekwEDbtuC3SZTc~I8y0^Zxt;2rdI+aE<<((ml
zvD`4pTOu1DKCxQoFs|ADI`Z!Qar1K%>nF>KfIFSh6GU45(78d&g~sQv;9~$cL-%2!
z+yD|1_L(0O-HDxuca~UZoK<&O-$kEmgx?kWzk7JKq6+ho4p%6>bH~`Y@|#W-7%5QF
zsZA2t@*4-{fbd?=7aDXR`m*#HPY^*syv!uC!#@ED%XW8NY6$!EpE5w|lCc5~a@Y`T
z(lQm8M~^|o+6sb}UaG>}?JF|jdk$Vrj(#s{8`5{Bm{KkV9mVwC<pR6tMQDM$f&99j
zaT<n9oC3vY{@LUaoa+>z4ye7^!yZL}=4zpxr<S5WB4R?~*n)&F+<x!?81n)UD(+9<
zndO;ddF|Ej<$GiYS1+bLtms+J^zbnt_Bp3>o)=?yYg3~s%o4F)!LrE)>uj?$oO{4Q
z({|k?7x8_t89tn$OaxXLWN(jFXs%Ph>q2W0YvhYH>++#$^QIy{t$)Sz{5_dGM2s|t
zt1RW3H%rv3SRUH^yX~JeQ5plt#^_CcC>Q<)`TUDZMi5KzjKU{$44qrvZMSPS9j~dr
zIvZN}-(@Hy5;<>aD=bXJsQU%#MpFIEiwG6e1WP&ef@Ke!8JS;d3oB?OXtV+<)W6j;
zEjjk@>Q@(vz-4i->;~bLjkmdN7T*({>|XtHEr`^Tee#A*q8bYObS6M3<>GI-91M~+
zK(2!`%LXLVhV1V9XKpCyUdDn=RcO~JMvw8fK*gr`R5Txg75VEADHC@Kzi6S^c1^HT
zfK^A(9A;J-O<@f4=KV~MQIhveSC7(*>EPl;Dz9}~p8Di_y6LFaWvlgzy2ElW^|Lfj
z8#@*j18c1g1lDZ9bQwTs+V+`1zL2M6lVaNQ`e=)OQjN|8A6uB3xJ1>aKA>VXRm6GQ
zKSCJ+KcdvS)?0Ni@l1^Nbz92CiZQIKcz|*MOHz!>eklr;vYi$$9i0q{?B@hYcgau*
zM*ikYcd+~4)W)nIYYy42&*vD@`i~5JIc~4;-I9uY6|fOOdbAG+1_078#fJsxzYB7d
zN??f4*l`=P*=iY#G7EJj_MHU{<s^G7x_V@(<~OY{YGEcE>xlKU6y6Mc3bA{=Hdrhs
zBAjpN7h0baL3+C3HgDykaj+k*=^^c#hf&ml%iJWH)@iSa_Jh>O+=_oac?kn{ys-7_
zy$@I7DtdKM%SUPuKxNdNkFdBXp~Y)kX`=>Qn9hEwvK;(_O|P8$YOxKxBwVFU1GZhm
zL=sYj`x-I}$8LkSCiq0;;S$7;M2$sq`Qml|S71VPKUYI@f5PXR4Hs2KyUbT!H?gZ~
z+0WO>F1@u|Si8w)M7mrZ3EJ@>HK|fB#+`Nchome1A&~6?jQdA$@-OPpIt9$B;U#W%
zUqiecwAMXRzW!bxUnIBwn2;JuM@rXO&evJ~Pe@au)Wk@?4x2&JA-60%pr)z@5=n!O
z#6=KH%IqcrwS}t%I!;aWcBL8_`FU#XWL@K-MLQnR<I@5g2La#85!Nc>^q-PbKCrua
zEIWLP;sZD*i9vGg0wpd36}$*y!usi+j>?ki%UHZ>(5|x_cEffd2Q>OG1~=~MTkPfv
z;nElh77e!wM0`Y9aOY)wqZ6F)Df!Ds7LA=GSWwK+^2fCg9~#LMUKkO_l5-NzT*JYi
zeBwR<u@jz*NCxKOXQ-j}L|fQ>6EFpNmh8P_UjQYGNl_?PR3e77BhCfGs4&;?s#15n
zr<+z2p({#M{#<NO3&{_WjH2=MaDv$bo8^Bg5<SiSb)a0sRgW<JZJ;}o{Y$%TdIKWE
z$SpzrA{ut(|34_oh22RiPet&@;>@4=4D6uH5b}(u<MXWX4Fh?aJ^~O5{zZ;~>lai6
z(o^yT7N4LB{V|<(bW{NCn6;J$MtX5XZ?xob5)#~pR@y5xofQ>^g{CX9yD(Kn?4Ln`
z1MerF<64u<nQ^XE=d-$k*kWFj+<J=u1+ym+s}Gfz9(lx`X4f_C`o=qbSmS4Yb<Zh4
zUOMNNEw4@XccSdI-vrQHs(*&D+W$sttorlp5xniZX}!^URd;=rs=gl|`qHV$yG=}G
zksBr<q`)(8H-dXyHut&tVxPnC(bp2(j*yA_LiFblQA%U+alrM#^TY@rNT@Php4x#8
zTwg!JhbiyT=lgYQ?kzO$IyWD6lF--feuzmb9HpgL#k2CM02VOu-#96z!mb(=#^PuS
zwiMzWH%rUX+=<1;gphDrrP?!nbzFyUQaL=&YMH4PS9inx*Dze>B8(tsxDNtRs}&b6
zZjoj>GOAG;%_t}n9l|tRDr9g5@IuvZayd>;b264{wV&*x%<1MorNqM`LhWcyjLJs?
z-31x2elmEoxBvkxZi}PO?e-qCxR9&9f=3r@QnqvMa<X2>FRC7K94m-zRYYi&bqq$E
zcDr6hTo;n2KHA=U!b&LoAP&075U#S2yV9=&r5;=r=2PRK1+P%O@xuL7qE7P`UsXPR
zMvty&Vx@EQ$#u8R%Aj>SpMNPa)$0%_GybXL{_sp{yMQwyr+LjO=#M9sKl<Mk0J|PE
z+e?GONsIOojav4xXkg!;=(_0OHK4)KlQ!=)AZ($Aa+GCZ9*UeKFS#cc4JWU=L|VC)
zd`Q4>b0#Ng;oh3CQ{<rj5oq<qw_mbqT{a;&fu2d=3a3l#FE4b;`V4H!s6J!s9c-7g
z+}9vvb$}eMB`ucbPt@7cXnad@u{al_c2E`VrR>PT2qv0w<7m9)dAa%)NQ|h}>bw7`
zbVhflzv0d~q_aTDK?LSQxQOedVG{ghjBGULMF-ogS3092$dVy*QJBboEE}Ht(Duwa
z;xwDj+Jnvxj(njt+oe>Qzlwk)Yk)ZG=b(ZCm?C~*{<iPx6G2ebNkWR}PU4AL8HI2^
z65hVKQ@BgUA&Ltq9_&RDha&kUo}JC)S&t76L;SRIFJ$yaK!m@Q7lT2d@8}LS0eAA<
zJ|jciD;l?Rv8z+tM!R>xH55$Mzw+eHct|0|+!Hxom_SL)F`V8(6){z+=mvbS5FbAV
zaJwy!6DORV9yV@wUWge5mn<^1HQ<#CU>}^o=BT`dtGxMb8}GrgztU5PC1P!rKt8VE
zUPQ?Icwk@&am}4ifw!hc!A{m;#eZvgS#!<X2~PY+&pdfeTVpYDlWk-a!Id`MhQc>>
z$ldj?r{PYu&r)S}aQi6ONoIjXP@a@8AP-twb{s5P8b@2!lBw`-^C5M#9n&W?Jh*<C
zwns*`o=yb!;rHDHH9bbP$H|^rr4#L-FQ)?%ZhB0Q`6|NhAy*T|`;?;FzrGT7C7vS|
zc%Gop10;>?`d<nuanLG^TrbLbj>f`0qtRvO3HC>Bj#5I<HKco(ZTx%xPWDE70rl!8
z5ec6MhZW4RP1=<a%h`taL$JJ>zj9BJnQ7&-MkreP+{3>-n3FV6{Tw<wxT)1Wa|0Cz
zk~D>tZ!1*WWgRfS8tZ)xJMvdmwewHTJZpNB>Z}9chj>$<=q@mJcIt%M9A6!dF@F-V
z4ZK}q<_i?}hr<5QpZa@WRXw?w02Y-qnSA-_Bv#oG-HRt%oLo0JokA(#!Q}_qN_dQU
zMDZZRJ?YFru&2t8-7UKrl2W|Q!F0|b&$YLt`*8)NO2r%W>&gvm?AMAhEvxfV;?Ez!
zQ@#Gv4IA$2ZO+JfA<IjZYA|TJW)}V*9`I{!ZG8T41^V|-3kA?R$O#u(wfWu1B@D#b
zdlyset!9gFF!o1ZhxDJPOd>ol`0wd(N?6i-+uFI}2sjOA-S@Dkog6VT2tVM>K~|LC
zOs4V)4xfF_!3;VMdwG;U+bMS!#93`reu|ND8zJF51(EOiM=w&ssK4ZNS)FEw;G?GQ
zxcHg0=yAZ0kBR{410Dp@B8-4#RUnWTrYq{xNCB{JlPDPdY=U;%p!zDZJmZ;s%y`>x
z&j-rmQQBk3t5eyl=U4mn@B_rB>g<a-U(1H&dDX^^>x=G>-c^lhGUGvsyLJJD(QB`}
zNX>;zP4T$Rq;CcPSRr1I1d5xtx+Yqg^i*rQx|GUim3F{J-~}}64VRX29w(aOI5S;)
z8_L-dT;*D^A+B?<gEe|xvzvUigp6M4{vWrp^TdeBrWj~NH7ecIPcJkaPh@E8AWyc(
zl{vU_LohkZtP$bcy9!TNG>UbPp<zm_&LXXB8oi}Csak!5hk4UF-diY))YipR3TEf~
z?o^5*#-BJluYt+$mudxFTh_C`^TVeXd%BkW1bw5{mPAVt`o@P?)rdNvfkfE5zm_a9
zMkkXIhj1Na8!2G-65!4R7ZEKw2Jo{`yCuA{YU4&!TO!WV_7%xXXOhvWUCOjWz6v&>
zyR?6E4!A^u-idb0e8_;fhOBrMo%$>8n4HW=Ph|s?TUH|2s@iBQV64kXfaXg_82-Zk
zI-WiA84){QkGk-XK;_-vEZ^UQOM<<B>BqTcmAw-4L3|-i3}(BX()`PwrwzV!w~l4~
zM6^<;H};Irp^<yG)~nhB?xrs_ojOG)#`mF1E4kogB0W8f_HSRXctS-7n0HvXY-Q1#
zcMKRXG5d_hG8%%q&AtEh9hSWFsd9e&PXSCE6pogA$wgpkDEF822BRJ!8QN$jJ<u^x
z)xP3*z`VZwI{Md0KTny;g>t3Q-t-nQNcZEK?Q=24rJu=3Z*#@lkidy9Q2G(}on5MU
zBOpa;6;h0o4J}K)*&4AwSdbS=mgBD$50ot60ttx&=#)J`Yl>1k4+<I|AVNxr!imzw
zzg<`twrq**2QNB5+ygO!z>QP*-!;XJb7EW26vzX_jROVk%8mq`n0Rw!9;3@K=thp<
zuOzXr02$sT!TUlSZhDl4o78#!d)vM)dnXVqMMg|w-l|*OCL)}i%UNPiFqz9~7!@!A
zie~mK^sSP8go_F9C_h~FlP7)!G?$OHz&nyAVm%u^`JKIN^Zoj6;FW>J?eiX>ph+x{
zc$z@Ah`sXq(OYidC%N+d((+xO?!~zQ-vr`^i3UnVlY_HYp<s_q@O~1)<)<&qb?tL2
z+ljz}xoE%K>@CY{=ng9BFTZRHc?B@-4cPURtXRWKs^c>DbrD~1+%6gJ!*cdNxED23
zJ|+iea&L<aso!!W>8vj$r%bnAOeVjgamW(HEws2hAsRye=6(Gc=cK(R1v@Tszut2)
zAhaVB+fbD5d&g?9F%+$S$GOs{^g`(Yu`*~(cB1ljoVotWwx>0s)Gwb4DFpC(nzY+*
z{y!{iK&9-T8!Y?kss5$fB7rZ*3*B<<u-*C9+m*f*+gF705;+2E97ZMb@hoC1+w35R
zxvH6!MaN~v{)Q^RkD~|Y7*@bBR_adn(1{A6Jr-+sLT~}LkdFDNuw+4oBK7YQ@z!6@
zrR&c$chS&)w3qiTfQYuc#l!s>?v#rETg3M(ORGX<2wgJ2=3z*|zHrfj*+qaXEOg7M
zZ$Ftz-xSz)0@25w*?~<65Ycl){N1O*?aJp-X1Lj@PPH;`-O{koIoe4%WsS>BX3fO7
z_QQLGLk7sNQbrl)zz^S!hJRJPfagVf3^kUHjc{`TRs(MEK?~IRsdMqKRq1YAa_)l-
zUQ0rU3JU-3MT66)VXAs2svJ?f%4P>C^pDV#15qk#N8Os3j98P<BD~ga)opK3ax8zB
z`~PfA5ii%Czn{N#JByt?TOg&+!`1fD%5>Nb9am{nql5CqjfR%AB-~;Uo5jQOnpI>(
zpVNu~91$?4be}pl3owvpVbIZ_BF)U9I&wKhqN}=tW`9z-?f>LAHQ~EiE?Z?_w74Mc
zg>Dh-MEZQTbZ-Dnvl+PL;G7STX3Fo<!zdIok{2WRAmk|mpIa2Wixu8oJh#i1WIEby
zD}dB`z9KbNc=hKzqd^;z!7Flc?a_H6QTuEH3jLs77n@EkH0D83;Rs8Eq8=a;;;8?c
zHUz$z1_RC6ogX3MQ)(+Rh93lM0IX;pt9g{aYyJIHdb^Zbr*PJc(5Mz2wWD!jhzqTV
zJE*0E(4aT5z(B<L=RiYJaIK5AB|)*3uRDt7kX)Yoe18c54SDW7h?Api(=%GP2u11v
z&UepC=g9v|YT@qf%C|e4pf`EGrV*lWRhuOr8bqhP>?5!9UODpy)#u|4lhRi&7Sl|z
zvyulH?*6y-*ENFv^UFHr{SqLc`k`^*{fv9<80JOO6W>H%c^15RY6U`^vwO-!c4N!^
zSC-L+t5{6Ak4=8Gm$i<i;`pm)-ZQ{a!On+c2DLT|2JKSn1{;1r0%B9DqpXJ`ClejW
zyQ>3;8Mh4RfOA=4Z&HKRI`)EqHG~qhufM+(B7)>JY1nDPuX%4`@@6<D@1yah$$yH;
zGG(Kkb)p!a3|V4{k6T?h=UK`7jXsjn7CvZQddgo*q-NV-WQECO3+e2AbK3y*(R{{}
zeh$7W`K^Q4mr&Afq9i;Ju_<n+*{7j;q|wbk+2=m^L`qjzkvNR)!C|~jjRidV7_Mny
zL-4fUtDpPB_>`+g3R9%7`56FQo8Et%s*#tP?Ov!nG@pzV>>gVEv6KDSg6CuU@m>|R
z;n_HIg&nU0v20J7QYp~O%u;~yIw$h@<^cP{H<#~#1ZHj`T*B5^783`X(SSY^W$-_2
z^JAK7xgIT3!}QVX<S^klgJj1;@YCk7C>0EYgtx&&##`6_Fd0dKM00e)fBQUa96<ig
zX1$`R5EX71l+cz9Hy~C<q6!FD4apF>Gh(MkBu}YW>I+Xi1%wn>gRlTJF$fIAGxK<8
z55G{O*rMDhHD9_4UHIK8lW~BuQjPv)I5@JU?}o2|wLo{-{eSmyUp)5+;Dk^Gc&S|{
zBm2vjyFbGsghib;bOoRC>W|DKsNAPiT$>Etig1f@T|9Mx7!H#639A(2l;RFkP4c#B
z^9do7LX!Y7NjVMwrkj}&MUB-<!Xt!+LWsqexmAW_4a9|H+ts_NdrSLCj2nY9w3HBT
z_08l4U!1jnmJ2|#Uy0eX#hzI7Zs;_T8q#RE2>(jyQx0^XhRdhUU9er5RHSm66LOf-
zlMLrebr5L&TM1Q*n>>@CUJ##q=>wgS^m5A_f;c0OtI(8^1IfijLWY1(pgG>}D0vaL
zEq>Dhj$bH=FMLeFIH%I1GD$Cr1+fJDr=1i3u#Om(e@#AQYnHa!an_toSMXIv+#*l3
zcC3~hIyVbutCs;Vf_*E@oeQn7+SBLV9?#alAQrG}x^=PN@JJ+;flvS*7Y>)}Va^^B
zKNh=EPdEXFYoEV+;c^tZlHc&UUiYWx3>;s-mOkame~{?N|6tczTibfv2motA2mdl)
z4(O3_aT)FDSXC>uGTv>ncURS8%2Nt9o0E^rU_o&Cx6)FcBlb<-2kX{1Vsn}tSNf<L
z&Ya7palSX0WGQpkqnK6#G$Yp{(Z$8|PWtls)z|3#A$U{~OeKvceGw@Z$)cL8McyT+
z>K^$U>K=tP_|J9ucYgQpU4T4})Y2CXpQk&9J66yO>r}%wi6G~en<K`4<|fDJZK9Zo
z4GF0_ioEaj6rgvL?v^M4$V955^lr2v7AdaktTMi?u5K=0ZGOfozZnH3lCCLVHG1no
z)`1uhRYwX@8fyLTh{UE!94C$`*SjGfPNRj{JkZDw+t3~gpD7W#2A~9*iQgJ~yy$g(
zTB)i>r$lf`uM>Y+x5fI;Rz>H8dPAK8^MZmhFc>PzBMR!*b)(u`<UE@inZR;LjVzI|
zE@5F<wils)a9?8tng|8L-(eDz;RMXXMQ>35AP;mWh{!e=VF_g+AY>HAm;OF!tXp>%
z%GT;d{d5x7bk@O)I>7wU<UXXPi-g8_L4{#I707T{riKkGB2f<A6HHS5h+b{#KHfE)
zJMIOihdXY_aO%ug6j*)Djwcle>;CNGu7vGzr74V8c{(tk-6Ko)&FhFo%hRI`@F%fC
z+4ZX51zhN{xVxFZMFM;QE{Z%((TaYsTDx-fQ`T%f-P3IB7Xa*DI@{h<E=v<2e*Dk7
zLHcba#frLB5@epjw#wC6NTI7I6{Km37y2VebX_$QHX0st@4SZyC5piv`WPoTid-j6
zI+~XDZ-%iW{thquoxOBk_W`?lNBxigkjK5sYO~|ku)s<I%AYqnH#>d$jOQ=ekMCvN
z28Vs#H|R=N2Y}H0U2`_1mfy4$!yGoh`z7L?4Uq;VX`!!!J&5LkQ{Fq9zRyjWUGGrJ
zMW`gW3+4Vt(QdVlUUCSUe0BfSX6cxQ<DUfbm7Qg_687F*0+6N>NHwDb%{xk-m6DE>
zDTbcEB}-?E;xMw*eOyG)OC0u)AdM;*v@4#>Y#>#6X#jNRaC%`N`qLm8!Ed23BDp+m
z7o^M~-?+PI5qt{^40Sj^)6Mna&lJOP%tMWUKh@2w-t?hqM=mf>^B9n3GM~8<4iSsP
z5~|+`NQ~-kS~{@=?CKdgBED9!b4F1h?7IG9E~sjP*L%7yVP=LlCk&MsiZR5{ubh92
z(k?vjA^}Dt%n_W@S)^<VV*T6T&eGST{9qJ$t*{>^QRv>bMyxD{ulChAN%jm>H>p$%
zuDvLOzQAc<>#bBIpM@ou$(6QqbdOZV(FrWaA5l$&Ug0Uf#0D|2vAD$nnZ3-nqU1e_
zfCtBYyc%1Ujz%{TU1O>=uLK2-KN{bv7sVqTg#gIg5pv475>aHVGU*2^3ysK=QFyMo
zECO-abTkG8IpzoKMSG5%Pn-IoNHq%wKutA&)}Mc4c27!D)RjK1vbU}8Z`ZSv8vnfg
z{ao%NrR{7pP`&2c??fWBWim~pPK5ku<xKC_z3MV(6d$^I;!5Vyvm>UJ-z0B)Ej=HV
zgaLo@OTKPw!}**xsRSy>fb>=aod2bMy-J<ImC`4@ZL&b@dbtAn^dVG-YK1~NK{f>R
z>F#Gl##KQAd35^StPbitgE_3gN+|Z}!g%FN`g8Yibzz{Z?^QDuV`=*#H3^JR<M~J9
z+w6>Aao!CVN=<=EQeOA|XvET4V*#lIPywPU{glIX?!t7D0gB9twBe&go*L4LJSs_<
zDa0O0qu$a?@F4f*h#Yfv1`i>!Sb=CVC_t{3b@1BiwvfUQNp&;2D5iNrnfE06Yk
z>Ujf!Z_1+Igtx3Rx_HN%lFO%ps)7Q!B%7=~$5cR1XE;98tbfBCN^72T7r37Kw%}NN
zd2g5{%KzO@sobZ-rhqU3oG`>L(yINB%K$!ZJSlIv06RXJ1`uEVf`-J(o#R=TaC-Gp
z^U}GBAgcY@6s=Y!Jt<JY?DTSvCca7D1|j*hoZZWoIwi5#QuADw%;6G$O%i3!QW}lU
zFX^ejmM_Q7^b<xNfyKsC)2l|*W1B%dpbGvojGt=wu?eC14OBJ&%OE-#{~-*o?XS27
zd49-5lPr=EfqL`ZWes@2Eej(aT#;G(1?@|5(J*3|ejn}Bxv6xr@;?lEjA+QO8!_?$
zZdYsO>^hI$<&F02o6#SR#gekcQhk<@>I&+%gUuUWg+e7R)Q=hC)<)`*U3+W|d@PS%
zC_9Ud2z-ALuXybN9(h2S8rbEPF4virrHIH-!c7y}%s@ylcP~I-0KyxR`%2lRJM5*Y
zsgt1I;z?SZY_BI46Gu;<C^BewB3o(@RI%*XkT#--p&kq&(j+=q1^ni8jeT5NTnP+%
zN+0CwH}=%D8hWVOGR#D>Xb@OiRcl^%ny8WAH3Hn*dr&KYKutW;@qy%|;sG)_GNDoJ
zKoEsHo(dnGTC#}CiV$p*=A8%ECJN0ziIHrp>iG>~&nfjayiT|>6QGul08N4c4IXq2
zxAjH!%tK+q$|RWDCsDj3P~?Pc>rFSwNS^!-qoUWx-}wU7slT`0M$lgsPhD-$m1-PG
zbGtpNUx-ctGhLz)?Chzp(&+*pKtv{yHj;|nW1#mr1p{El4syy~W3FC<-XY@g@sMQ$
z>LXt28P^6C1ZZWBN_pK<wq3^vFWRL2?ZK(O5vtYC){)@sIXs?htY-$Uai^=zrd#`J
z7rx!kD9LAYNliRy41aEk=dkZ7J)KUpCl~Kt#TXv}xS8IZ>BxbD5zo5A^kg9+(2#`@
z3IhlAsJMm2M5%818-@K%V!K^0wizV09ZqA@s+yr9*dPI~lI5?GKu96f?;=28_<=~~
z%qRf~K!g?hywKbPaIo_yDfTq@8|odp=+<_W#E3&-;{^Wofse>I)<+PS<NmWhrY!$(
z3BS4le~)K?D}YonRc!~$g6R#GwR{7Ab3CGlyU}0oXkB7lA%DYIr7+$$J8L^QiE7^T
zK$j<BPd~idp`5YOv^eLI1kuXjU1pq(Wha2l%>z2E=Jg*~EfjoYn)qtlaXjwd2eS)3
zFJ1F`FqSh1;$;Okxfg0*NYncUJIn{EGe~v;1X-_+oxK4Qpg}NK9s6Akeq-)H5AvbI
zRxD6Js#_cX1RMp{X}+b~r7h)aJ~%kq5!t{3g4J+e?j}pu!o{{c=7xy@DyQQ`jf!?l
zzVXbS+rSu?BgD^hFww+C{N`17*(8#w+T$MIb`0vW4wt{+r}Hf;*2wEpHYuDPyclBu
z4z2?ewRG9+?J9V=$J7^q61AA)_V>a<*6LX6CVq-prefgK_p9YrkS{%mwQq)b@G|2V
zxNhR{R}kaxNaK^~3|_X*sVFP*D)}CoGpI@0BGjIEq46A3KhkWHhX^;@YM4*ZbDM51
z1JW`_dJWyF&{$1HYeVGN66#6k?9v(lhvq`<!kKNJGPYK;MZqa^AR_e<S3~=#HixK=
z>R@3jJ@g{Xv)tQe$~XI&-hvC6y86*IUdFR@@$sTt*(^=W4!=i2khbNTgC&l=&Zc;K
z;>P_!HW<3c{$lAU<nJvHpx@9#)z)%suV(soy*exXyTjG7D%=C`{-)gumFf|Iy;zlW
zJ#jH7)x4h*?#w5&w>Y8^y@JXaM*QsBEV(0gm1VmMS`%L%DUp95vBBQ(5$l&gT!e}i
z(`NNTIe0?J*u&z{q>vUykc)7U!J&-4I!xj%XxkYwo@?FIujCLPO|hP*ZE?M2)cv6M
zStfrgnJN?Z`V!aUy?hz{EZPTn2E9uT+}SF#*sB4Z)R(Dv@b1BbXgs^}eH}#fmY=!a
zD*!C^cJU!0cU!lUMD0Mm!UepAU9mllL<Y=R!Vt^RiK8JP(qHpSf9@8x7LY(2dnF1&
z>h$rD2EB#NmTY46ZVrBo7R}a&B@$;h8YfFQt_j;;ld4p^hB8Y~+s6PBKTVB=4}W@2
zHPNuhJf=(E$TSCrZ8s5%F_?E~5)_4w=*X@@8wN^>qbN#_l5KjI2+=IcWg-E#$(J~Z
zP~|bkAnwLcXc<~r{5W>Yt*|)trqvoTwHQ2bZs`;AqBYx8R`Gk$a$Ei>O$F=4HQVU3
zbL^$L^-kSjg()Ah(|iE>^3Gkj%jZgs<gGH+sl7ZJ^{#cld(|9KCrb%d{Lq38S=@V&
zyI!@J-r_Sw@XNiIc62G<7eR~dLKO1v=NLH^JR^*W3wG;zu(4qjOG5Z-h)<1uqs|kS
zlWFhCGRHOf-vTB{b+CulsU)6=f0PDu9mvV`1+iOCjqTWTja2}J{jbi!CNvE>e~*C2
zky{N1|7G#D1$H*ElE}1|f||x%+adGWI{(Ju@ox_4Xa;QP!I#j#k-C4;>q{q`P7}jy
zC7e1)y8WcgJ|afB>3-dWgoVnTY(DY~R%d_Fox$7l7wtkqQ?cMo+lh7c5T$Z`ySFcg
zY7t3@^6(Jegrou--d&f1-2BgY^B)v$_sh?hvza;Cmu(}G9I9#S9e9sC5wuSJj4f(M
zm*?i_`2}^vW(}bWz2n~O$xC}>d5a%YQXufO?-~1DJwfx$=(mKfSw{otFH?;zA<*ee
z^nPoQ8@RCdX5I#dO_<>xA8^r64`9xA%xIWE31`Q&ao7W(I!>f7JIBYXzYYT-8fH<?
z*tVyXk^0QY0&F5$^rOp#YrRYf*W<Y!{K;53FBXhmvJ%F4^9Bk06`dO~!@@0bGHv?@
z2G#A2klE3%&K$UiW_je5TUWr|K15qn%Te%f(?$f!m@h@FS6uJfSww1XUYl+nZIBP|
zLN^Lwk?R5MRKcmX{LG?S^d&3TuMI}>rm-J6r(pa`(Q4YRR|O1Zjgf9V3ft9aFr8s{
zcjkeh60=;s%)`wnS(ksOIW`gbJR#Rogzd7B`%-9ao6g?;CWL>&J^T*xxX|==i@Y<@
zKW(hgb80sy8o>WSYD!3nIAy{+gNcZz)3DgGvRgn++zk{TCtG34{>}`cuQ|1+FiyP%
zR|te9n8@JJcV;EIem~^82s!a6hBz-T1Y%DRa!`U;&AGROS-icQk@-e@R(G=n=vc_U
zVUwJyy^`TxZXReEQB&FO3{um2wR-jkI`&@(u>Zide>k~vmplAx^O0=_#Iaww`KR{+
z=Ua=lW{cu1*fo54k=*B%xVkBf>rOm?5WCdBRF%F`ZJwl*8hk7MOd&w6u}c_6M6XI<
zZa|wI!Zhkw<mE^|$ORDrW=*He10exMPM6IC!35+$S&Z}^j|_~~o?+TxOm&`nNOaCH
z#6S6vCLnGRI}y%a#h;w9YZM3w=x=W6KQf#GL=}<|6BfQNtsUpfW6h>}P7d^+12-g7
z>w_nu%57$Rji!-zk=n`U6hcD`$JxtXxol5YUhZ2M*rj_SFeE*W$KSWa1*OI-q@&j=
z_W|u;nq#F?(^lG<4fTQ){Je%k+(5Ith{ec_H!SRC@N)HLqwy=m*X!F%OQXkxydgJ?
z!RH|>RRzT0!mMhF&4Tm~(iQrpOO>H$$z{kC^Z?F*(GHozqgl|dig%p;Ortmix>d*w
z(1~PtmywQephxNIxoUa_kcH)ftGjTL3c!3a#!08O0MVu<k7goWjHl-Wo%ldm5V4^!
zB;*<yNXlzuALv;A;D|Uj3&(~Li3AQNt~LUE8nKxrCu<=NT!?zH{x1yl`&D2N0cL|*
z>HdS(4^?)$oIkgSIJpc@q$wYwEhh)HAI=7be}8*=K*ESMw`JS2p~192q8`4RR0G!S
zqf!r3h@rH+IGE2k8O|B5AQspBtxU5}x3w=r!1kKwgPw#*?t0qf^BFq4D7<YmU{m8O
zsoD5{VkG)uP0ST6+(CqU+I|7~>g-Yy^)#Uvb;6KtfWPK|ZPM06j97+BhV+v5o^)rQ
zpJ#7d89^=!ucNl#OZ^iq8X@Osz5!Ua#8Bul>&l6his*h-=q~AzZPvV)hS<+S-g2l@
zV~`S47npZyB^PCtqO=Eari?@0vMUniWOCY1)m}!DYs+Oai){7^=~Yr}=^@TK>^<g&
z4EcNT$3Sg9bm!W6CGGrO{l!OSo$`|5=I|($d3P}u??0<zGCH`;6c8&~hXin(K)IpJ
zOqoJ)TxsBIT?@UiMxX-@7p^A(A<7bF)Fy@cYZ#EM$Enq9o|knd%5JdbElt?OZ#K=J
z*>xHBB0<v(&kZ^zo!n(ow@y3@KLa!qAP8T8u2cY00HUP`97}rwNm?*Oj_*e-Gu;|_
z=x`AXJl6jJ5^vc!&f-I!$_*G8%f$$%m3Y}T5>NCNFITUJ#yH|@=kzK0b!|5|qfg)$
z4B~VpLVx31^y^D#ai}ObZ7G=2%<J!D!Y40VEF`=FOoYCs3um*{#>d=;zjx>sZb^W&
z|32Px@wxUWq78r@|0@<fTM5-vhtgbo(!2>{vEY;vDw-m<X8_~hIT-*Ax?1fux&fT3
z8saXVbR)6~_x#a(NYF|;{enJXgy{bw>>k4_3BpBD$F^<Twr$(V#1nhPwr$%^CdtIM
z?MyJSaWi}GbI&>V;Qr|!tGl}E>1VB4U%g*Jm;~P?)tjPqmFbLNW&#b@)QehuUNzTD
zRm0ZSzDUN3y^B79w9;vaP7BqVFXyK6WIJ&O#Op7f`@r*q0_NjXwga_ON%nycV%=Cy
zh*MaFGi`-BVFh+vC5g?TG*8GD;4rfffjHP98}DwbW~GD!FhW7vtkwIz2PxM@8RIhX
zpupXLW`|?p1>=(NMKx=ZzPD>%)R^qtFyz|D9%w0Oedfa!u|Y5C9A~OyTCES4RdmK&
zQ^g)V!Xc<!dDZ$l>5`!jc+CjcbUMw$+&(#%2yokbrh~RAAf$R|?pRL-KwYHx){~M&
zm64o!9N`OlpbX}N7A1$OVRl|R7AC7r)3`nVhD1jL4omgoTp##R^&20F59b(rXpT;E
zVqaI{Cs-<;qe4W}aKk7l6<6MQc25_n#Gscuwdtk)ijvk6qi4nX(_ry@AlUgf$CV|9
zUz<j)lOV^Wx6a`0`+1WRz`A#%+|oErHs}&!6`bHp4nDjTQ=3O|r1_HXjBV_IU9}Ij
z(Te3Sp_#1ul{DHV*%Gx83w(sg9@bhyQT(eP<g6;gneF+9%E6g&(c8W8r%8iyI(mb}
zs9RmyS_w!4<O73TM}mNK0cUSKx~^}kUGxRzl1OHyDOnOZPQ?Bg0KJ<|c4lTKV0z{)
zA#02TVGM(I+_D0rtlh&i$b8~)(gdK0zaaP&n0sg$O<>hKR6|Pfg{lovt)_z4DGG0@
zx%yi<v;*;Mf>5AbDiuLO8j0VcXRFeQlacZQwO_u#i$*4YO*r%NHjL0en+TQ`0$;>z
za@5_lTL|h7z4|#CU<IX(IJWqN2ZejeUXRa#N06K~+W8)yVMm;Q7B!(os*81Vovs@l
zEDtX3zknI}NdJ?@Weu8L14DvDr^(ECgJOW>VNi?RC?DelQhh=Fcd%tW1#CI1E!|
zK2yN#&t@ssW1Z$&hRt|)<Ehe%_KUxb6{zvr8C-W3V+$3JfW#F(L5$Rfa4&aXmso-Z
zeK9S-c82Y7ndfel`uqq5^GlYh=yU3wIC8}+Du;R}`)=~h<WM<E=Ft$XHejy}|F$-<
zDmk0@O*~g_hjezD&Z^9rkJ1i*{6|8Lqpuk8Ft%9TVYe^okK$3D8kYB`$wov$a3EUc
zN5^F23aDy+fb*FZRS#DJn4||v(H*X&M>qj3%hsETvU=}svk^(@)&|xwmo+@-^nRR9
zYfYSSANH5<8TLxzI%=(V_UOKZ(G?%74TpB;GRgc=BDZiYKwoZxhet{jfjX41SQmQA
za&e2L;W(7A)*f>uJISJ!Tc?^@kdf=iohioWZ}`V2U<&Eu^2#lI9pxs-$9cL;j5g?s
z7hl;^YYZ}-*6A4r>p0R(;Zo#Lk4ZCFlt`>;w-nsr+!gQqF$vcTM9xBDLIvotzuqv2
z@6z;BBUwnX_L6F^_Fz!wCUKu$gA0LazQKAA60~D$%6DOo-;$pk{Cc{2Yh!(fo8DJK
zm7Ej`pu$pBEiF|dqHyvov|8u1sv8NTZjT4+BMJXUZlRCO|A&zjx7VdAU5m(tSU2oP
zP+W~@01_J1P3g4~(O}JwP!M7&B%W@e@MVZCeM3=KBsIvI`gm_bge75bCg{o_p-U1c
zT_yZMVkDC)1ehW<k|z4%N4&7yYYY`X-!XJkfS4}Y_=Qqw4Ek+7Myo8wFzJ3+xMbl8
zD<S(qz%xO$U{z^>Dj2XokdUk@Dq}d;7HrIv;SGmLi2rHVf@Y|Yc*|0oOBJYXF#_&l
zUvOxgPSLzwxPKgxPy72GG$F)KhEjm@;tMAOm8Y*#ZVtKkWK+O^pAlsN!75vfbQMD~
zplT@`jS<ExJC)B<$8zGagcEn|T6ALF@0HfIjufVF(Ie(cnBBt%QSNA5hnQD)lvQ7q
z-dplZM6>YydkKY7_9eBf3RC2UowH$826p|6ch`C5OaY!V-;ueok`Q?Zm@LAZPvNJu
zgg8w(8t4uEU5=exN9V@Bi<x=T$|;7%fS;(j7mdL_0X9jQoI^T(=h>HTIe)Et%>=FT
zYgV<}PgKv`Ftm7v$R|G#h@5Dq#Me+?Pg^;7YKuh@<HPO4f~r2rJI@OC!UD(&rXhb4
zxrNT}Vj!nSo^n>OlKpz_HWVWqsukOLNksjH#0D61%_dAc9N~D(hG{WitUR-y1Beo%
zCw?X3`i&K2eLnnZUzm+*A=i(5x5OEi2_`X%@t6Q!mRubyQoJR#&(iuF7Jy=Rdpngv
zvA!VUBrKta_q1!5Jo?^o`K|SyF`U!jrt%-1Ba61VLYAu6F6%seZ^=4pwl-|bGej{0
z-E1eX(~btsdG_e*><3VJ5G-f`Ajr%mqHCnn&5QElzh1IvPQr5{Y&e2d4j8;8ZAr*l
zy$%!Ry5m6}>6riy%b5XIEO8^u<bsx|YFd_b!VuqmuJ+NqG;NB+N9l>`8sg2!E{zV_
zSIuhs^~)4E+(T=WK;Y3)GB#3B*+(qWqGwhXr5#^?^`n~JjUC7AYhGHEsXQXDVdJin
zlUgRhF7uaKo7%LMeh^xmXdljD*kq!NZVO9vDKn;&sq}eMWu(_o(#$m60T5yUNfpmI
zsZ;*Fjd|zC9;L5CV2<=JI<>5p?+N~ZnQ0aVz;pgUK(Kj|hZKeKJ?U~o4NX{o5GI)i
zSH@Cv-9veM*G}V)aYk>sD<|+%QoFyra_h7*@xhH@)w14sW46TOZ6f=i@ea|qTXVbZ
z3l$ve2^zG)a*7T)1hHZbO|qN-_xauS9VwWSpWyRSNYcM<=-{DXh^ThMa&?(}ph?b9
zrAURz;DvFZNu-1GfdWCGVD|GVr+ZP1p4}4efr}G}5?_g8z!DZi{@cki&``$j5<=f4
zpc}2ENzPCL1L45I`$(a(QRER9^~oa0%Dbdcq?Ozg$P)9f03iSFX;J{G5iLyu{f8k@
z4CTMB@&CSKpw?+XXB1fim@B3<Nd@A+8Y;O13j>3JH_U?m`|LkUn9co1F`+H+YbcPi
z;8{@K4<h*ZP{IH9kpIgw07MnHC8&Q@%ALM|K+}m5KN0>t1f`wSQF1qSzZmv0Qy7N`
z9*}5Mhy9otAo<>Tk(@!gx&`*bpY2`x73$x`^(%1yRU_}I?;1U3CW)b3uv12mt=fd<
z10}vS%!B`rCxUnb*{$jB-x2=3r+xqds}yMQo+8Q&?`Oj>m`D?ykQ`E;$6Q>5n}xhN
zc-~1VMQi;_v@>Q12}ng5Hws<_s!MFLPfhsguYq$=zz9+r?~;Wm2pk>r?gU}2f58N*
zEwYPMrTGcspm@zvIlE55ZeX4I{??_z_h`cJ04TuTq&hQo>O#t1DyS9Vg90$`kk%op
z2W3RbyV_81f+P*>0mQe_=WHxkVL>E^0480eQ=_R>pU_Zlt{DH_7DAHU02QcTMWzT5
zK`r3`1%MK@o7|)Pq3J+G>TaWigJQT_{Qce9@K_8%QRKly`lAL%$kFL8OS0S$#%YbT
zRuT<i?P-ddp~Dkon~l(2-vIj5YX5Kvs?=0a{K~&`X2aW*4*A*=-4A$WlzOPudX(XV
zMFti`;k0;ku_!WRd;6#x^-|j=v77x{Ec=F7t?fY=gV4GsK!DH2Bh5lE)_U=$$!@Y#
z)%N(T#5a()&VsjJwq%~Jt-t*a3jAv`VPxN1=*Abjo#iERYMi=hh8E!H$aq_Xklvk=
zH(nZY$MGqQTC2L&_>()2bo=Tq&w1#@rjkq?<-j_VuZ5|D@#$ng?*|M0(@9;2$lgO#
z5q@PN$xM16f6sTu#z2x#=g7K7{VA#A2|wQR*390-+v1Ps2oXUhxVL)wA}=~l*%T3w
zl4sC+8=s?3=V#?@G#Ws9N$~W>T)9OZg9l&mJ0AgzF~!({oUB&Sl4`u<*4vS$@yyZF
zV#vk^$SttWYFBZ~8|^vjXX^czaQg3nI*B^>C-`pjV^%41Nuos?Q->ih!{m7+it=h@
zS*yXICguFRbgrkJWHhIeJNN7<l9}cHjTsn<V+HUvqxnt`u)YOvPQf}E(_HQSn<Q8{
zvnuk4TKYppRPxpKRw$klHC+*2wng1Zeo51wY50&p{C8*#gG0k|t__1s#E&RiDa}=f
zg;9C4fH@Eh^K|hyj1pJc7B43opw{$7{tHe)HSlN1y6QiPN;UV-1>-yB;KkPMBKaaY
z8w7xdcw!tvEd@&8ZnFBdGCX>%t4oEZ6`P>Vq^fPi^lQ&7=e6ikPqEt#Be-MQHIWsw
z2?8S=mEAeY7vwg1I)Eu*GvnxymN=qkp9lP^06Rmrb?Zbqr&`!hH$2WGsA@+p(arcq
zq!S^scPdiBDF5)KLp*PiH=1<EN5cXNO%K2T(^f)>X6Oh}(f?H+@4oIeZTmAefftSm
zuVI|qh|ugDLaJsZe%&>;-lxch(aJV`RQI++LZeHg;;E1)5Eoo~INS<KXGdaBSM6Dl
zwVKw0R=FFRhG;2+rsK$!6qH2Obz^h0{ZQ9w+azCY{1Wcy?ml_u5l6}4aVchFGYaq?
z?ra&#f=@!9|BIx+f4hJtbCS-xi@Ao~I;Z@PM^OTC;+e_Ra9sa}Uvi^;6Unuhw!ijN
z6}g_y1h;eSdQ=2+<`0NhLR)*+=x%Apyr8BfiKO@z3n}j<CrLC+OwE<>+zsQHbiRzd
z_+MXC8+856QfS253(<c)f(a}*8uI{ZuoX!arHy#WwxpsXr^2mFUWqx?izLwMxn%`(
zoU1#a%F1Q53+Yag%_L-nn-9CUS4E42->di@wt6RnHC${gl2=7g^%6HYxJ431qtarb
zArwoCS>i>VV$-%7&tD>Qf>7ql6IQ_d^gm3hs-3*OD45-dzxcbPF*s?O*k=eBCBIK#
z=oz&>0{%xc?#WZIH4+)5?#<5RISAAwouX1rq>V3JSw^?-yFUFdy#p_`Af);-2n5{N
zB8}njv)7yWke;T0qR(u~)~Iy7J2lbV%P#EB<?9`ccG_h{Yge2DT@QltL&i3&|EPeg
z>!fD0P)qEF5S?c*FpA+#3juJp;a|IMX6=@$3LU=C3FkQ1#k!~W*=T{K>ekQ{xYw%$
z$$)+`_i@MEYT#(*PlUlVak!YO<{I7Wx8${^Vc$~p!-{6f-`k-{=Qy>E0e)iQS|-xy
z<z7Bg*Vq>=>$%w6-QVAK|MrRWj`&EB&qF!E3WoB95!~l5I1gYIvj?Obx1#ScAY)`~
zwgsbYyQe>E(qV?@&72BbxsaE!qZ1;zzmnD@4bI%1BU)xjHg4XgDXi!o*Hcs_d#E=q
zvQwz86+*{^1jwYMnLlFk*>&>Qn1-kS(pMZ*7V<{WIA+!ci?>9HEDwZ3Revha0F%t7
zmh3Lv;p5#`Pn>Ufm;iA9!8=bbtslt2A454i-yqDY+4ixklD&POkF<N}{cU>=1L$iF
zwBygUJ34u35OBIc*y{IADit}w&NV#@xR2|q3Ro1+8=T%5l0BQH`J|T&$<GnnSDxj(
z7kz5QRnb3XP}~iCWqi*J6XYu}Hb(Vw|Cql1QwoUk3d)MJdIkuKj86_b>mt`VNSVLr
zcWs_K<uM+<*dm?_d<3UBy!axFiTs=qz&eq9rIdOtd%js<|DMjC#llsZ4y4-pd0z!Z
zA!}PPv_LWSvkXK`c<!o#8Je<hgCx(}|L9zNk*aWk1Z5nTKc&6Ew9aoAw=$_UDMv*Y
z=^RfPt^Ua`779RdHaG1YDD4M7L@?K^RGXtw85KUxU%&K^!nb$4c@GoXW6HZh+=s>y
z8&@Rzj#A*uMmt8w`)h*~GhUK9{=79E(k+mQ?vn#*=-$nbX7Lns%X5}}dJqJoyoYAt
zXiSIu4eI22k2n2Tzce=!7!SB8<4lghw6Ti5Gc>(80MYncT~YAEz}QQmPL;Q-+uKTj
zXTGrQeXv(ef15B^@GpYBsy?Zaz)X$TTn9m;e(m+~JcZN=nItYooF6Fb;!3H^zBpmq
zqQ?mflcdNw4u5XA0z_#h0`NYsgFaE&2;(LZbS+^q2wI(Bqv|X2q0)m2#f0}1fbzyu
zV<tS;05m_og9XaQ8om@X?S<Rt5w&H>k_6)gqiRZn9hs1UlbL~)I`QNQOsBA8yT0{e
zIE}`aQ6%$@c$07J4dqT?zjZD}eb!>s_`%$m)0;j{v3$wrTc<)~DBPQ=2#dE{qlmXX
zYvNmITy6AKGi4;4rBW`>3X+}}Vz(SOtz6jyfWNOBScyJGIU0H2oY0GjpJ0bYexzED
zb`8YkmSJ|-Ex_f-%p7jwDHLTlSUsU0=`Rp|o>aRIrY~o~69HulEHK{xnp8ur+YL%7
zL`^V*B_$vTf;8-h5*{U4cI|o6=@=8(-8FwZzQpUcEdz!$pee6+j&i2qW`Z}up#nhw
zz%zICShtMrc$TdCf2V{9iT(|m(L0&_)zu532gINixy*D|DYT(iBY_<9B@ZOf9!WhP
ziRg;wkvT$tb-tufQXcuXtuSYiTx|R?=QqC4d6912Sd=e<u0oMYl8HqcJ&hb=l7T#K
zM{(9cOlkG>j+fF?n+AYQ05e3ke5wW1d!`c@HsF}jcs^E!s}V!nAcd2o_{RyN0D)<_
z{M*Ar8aQy~&v)?q%OU&;Ri!3TV1aD}_^O<BOre=FvL&`pEmqLsid#csS>wnHuIyMZ
z^ry@+2>~5-<0bvagmh5<6p;Rda&7zSq8Yfpc=B5zBhEwvgVWuW)1iiIG&>CJ$<{+x
zk^X6#WExp32q#k+hHlU?tRcs89#Y5cL79k1y*ci}J6@~dJ>%=kI%bcAf3pyesAHcU
zSaNf8WpjC2R4WJpbhNy-D5|aYEomBL9*7dawjWICDX4RTFg9O7QHSeynycxhk$5Z9
z{vLB17IfBkTo_RuXVxX=&mxEnbu@z=9ME>^8A2c<UwnyKm~UYX7E1^Df|r-svrhfC
zj-NLFQi9%f;7iHtMoEqo+iWS1wm8T|$;tsgiJu?XH6SuOF?dd{EkxjT44n!}9R&(N
z+-iY`5UEnGgN-u%`AzDKEM*7Mkm--wBmcbvdP*-dbclFZ*7#$uI|i_N>Z6>Pe@Izr
zj;wr?W)4g{A>;79&22b_`+`i62yxrCk_BMVbwj>EWpDbunA8+K<VAVJQycK}Ty=T8
zgk+)OS*3+i$}%Rk88b5;a2|Dj|5BLcUN{okq4d%m1GLUO9KZ=}fW_*~(}aoyH{1iY
zy8$B!eUlRszM)RE$1qxkJZPbB&L=@DM)FSnBK~(v&v55m;hd)fV?B2i9Wu?m*}#>*
zroSf8z@y4zf&BMUPPt3?1YycV<9TrwdY;-G{VUPuFwIbqm>BpXz?RcSMGbY06cZ`T
z>4j(;0w$Ohb!4jHp~t<&iQ2gp^8<G=#cO}-0qKLnU`JZ&!ac$J7VKNvi+cY@X@97f
zH~vhq)bzWL%Egf-_%r}P!~7ZIZ<-4fl_-722MpoeF{05GuRm2l*_V&-9xfD4QuQ|*
zC`bq3*f++IO-I2hC+|v<#pu9Kmad)GXqjcZ?I!tNgw6vV0&HF)dBwb9mv9PU1!N_Z
z4>21_r39=L0QWsDR*+yWIn}aHkA^(}ot>VCI{Nze@96!T=-5&d)5=M2r=O2fjN2Qb
z1w8`85bS$;^53RzOZgwBzJRtOO8#x?fT^JW5!Gp%Sug&tsNSTR+c9^+krpz5kk6Ay
zl&aecbJDS;%yptdgJwVTQ1v~CL#aWhwioN9sYs4uPwXm#@w=-LJ$+EE2c_-8t9Wrs
zQ3h)g&%RdpTfli{Pz``>k*(WUuwe2_u7SC9u)H&d@P>kYGs?znGJNMqQ;K|zrelD5
zEYre=GU1dP2jQyN1VfIf<$1dL-q``z0l3E<fzcQ)0YnT&PLoIliYNso^SbFdYyO@;
zYNqX)$jAR0cp!}HcbFYU195T)%@Y~Vh(x%O?=*p9tsl~x_!Pi=oGJXnLpeOzBBEYE
zMt>8?m_TzskSPvThIlfFDNdaJzR0^7rShNC*8~q7vzVtlq@R2@e$^KQy<7AvynH1c
zR+h!$EkXeVNCg(&ya}5T^tGjm$b4_gaY<dt!FtEdCHhN0<!qu)UIvb>yyVXRbo82L
z<x#1-Z)6O)0RGGBX{H4jyJuEnL?=QHn#jmpAZ7XOYB|V3ypOC#p~m*lj2ydfV~O~n
zO=bg>Wkf4sx7@gtWpFV4|Di`9$?ojIOiZie`hFe4PLXl$+2WK_3tg`{kE>CDz)3Qf
z4PF$Vf8brm^ja6MiCDFJ8q4ywD||>!FgYVY{Vg|LO{dr4-=N0HfxpWL?wfJ&cew;F
z9-j}|RO~x;*`G%L3)N%Z{O_sWmPe(}3<o7-fVq)WlK9~RA-_y-er@@!WOxAr@9FF}
z-&V#?s>w(#R<o`Z-j07(4|up!{J&v6dN<!-PTrG5=2IUrXOcKxBE4I1PKi!-QTqfg
zoo=fo#+&z_^v@2=don@Od!ak}s|wv)!36U@nUb{v-bJQ}Y<@iR^tbr$>`G9i*_lx?
z*xa(|Mh%TS>EZ9I#3i&!Kv^MJCVuhS_j?2$g>L-r);FXN{|7_`Ga2^(af}F!e~6d!
zoe8K9`1IZ|{%x$A5m7x6m%fGplG)oH&HlX#F=AZDkkqQ05*Sxc;txi}6EM*8lY#L8
zDKVQ=I@=8~E=k*1yc3PQjbF*DDhJy5+&O7hN{ZUR66k0YNIZa8j32{#QOipqxVH?|
zYsu_sulgLWAS|@zss-;umjAr||HSnLpZ|^Plh|JVk6iCAHv7%>E>@JU{{z>rQ`xC@
zS^eXB!2iG3H?9BpKXE-99YyJuck#<N*Ngr-_)o4s<^2D0eYC@W=lY+x>~xe;g!sk8
zVQ9Y*=W77L$IcrL8CuT25jESQY^J4ISMcbUlk+gMZ+X)tdff;8zAVB&-Q0P_>>{3p
zd$~_k3sZ!ja$?K*st-XY*ZF+F;hH#D%iM^)8PZ$Eh>;4l9Tx)I9(ZD1-^Sqifhkyv
z%&CFij8^M}a}_?kXl3##SKN{vSALq*cu3!1&^`k|idL-LbbGO`dw(<S8o&$y(J7($
z1IOZKU*V~M4RAS;IMTyHVVE;U4KN#Mf@fWKkY_DEPt)%m-+*@V;eG$7E`#N#MXbm+
zo<963t=l9XpAkvU3+{oEiX<R}dP~{%w&^I4OFlUod?dNbC=K`lw9pstbidzkm6&fc
z<W>zR3H}!THvyp#e6`ho!f!nCE&MQDXqu}pu$=7|4C;$EgM$e50t4R()kMOd8-ur;
zkHT*x*{>oxIv(!dTk3`=>9r4tT#;f?F>D4&E9pU5W6qgiN5XQ0iLgOK<n0eI(bTYP
z1TZ)@2QdB<QsGFGXsa;QjGVL91`z_l@_^gefQQ&S(v8%;$5W97AJR_HgXI#NOXR<8
z(X4)0?}H<k@%A^o&~8((AtU*^ud5)Jm$$yATBP~MEE)&lq|aU}+o*NIBVQv2Bc=)R
z?foX29=iWq!}m1|PVNnMv?Wnu@E~6Igdb|~qyjQ~zNg)OdUOzwa~a&18>I%&fv%I;
zihlR0QqiQgjtW$ZqDhVMFU)6fjeC6ah?YT9;VHXHN*j5qK)z4)6thQrVepZ941%a!
zoyLL8EA(ge^S*(Y`IY{cp)k%KG;w2Zx01E{6SEUfYF`N_e-Dzm{V)iWv$10D2j6Br
zl3BD%2wChVK~}pD&L-PEUzjgo-9rqodTs7%!NScK^X~Mto-lp<;i0Rp{vcQDrq}Cb
z8);mm1-;}l&a@$`O!@RUJxX0<A+1a*>~z-fdtpYvD{5uocV5A>h}u(SjMsLGu#`N6
z9pHk9bCU@`M`<w-wO+xDcWh(Lj(YbO{IxXUC>g7G%k0VmTR)=LHOCLQV>?^;6?h=I
zgUrdJI1u4$5*IpO!Y1@mm8eEj7{|HQb&h-gRnjk?jVf;<cXNGaNIcYp#8T{mlDR)-
zi$HWvO;85_yK!o)b>l^ngO93FV~NMz(=$5#6%HMhJ4rqfCuG=C>hA*u=R_)C8@M@|
z+mTSb@5Acz@S|Y7f+-Ckjz5B~l+5<|FS^On2jVKR=)F>`>t3I~c9-UHBUYY_pjI#s
z8T3oT571ggmfUDpZ56#$ut#b5mQCWEvjDS2WHQ7tDk%x!*8S5lDlXu*IQ64Tu<T!Z
zTzl?Vu8zL%Zn865b3<_WdiO12DLN(EX946e(SN@?U7*3Jhk#02<Y?pA_X7CmQ6_<M
zbjWBR2{+@xfk@#3lmASbLMM)$C)iGei$SQaCBHmmI#`(i=PTl$O1Hno(Czf&fw1^Y
z8cUrb70VnnL!XB8Y@xy<Q8ZRIDcE$C0M^U%V0r^nb<_)BA*WDYzXjEFskuAnMx%N)
z(y^m}svJ-k$+`^vRaX4R&CeF6x$2Xnb3v13CX0+Q(q$!}C7HS~4D(jkHIBbW6T1jK
z)t{NCx9^M&tX}*_i+Fd`WnDSQamzY7S;3K|LC;^I>JjK_ZS-K9#Mb^UYM2%cHJdf6
z-5R^1;TZ7eg73WOt72h^>hH{%y6m*0V?aFS*CwD!hgZqBD<*SMhnH!8bMiAh?sb6c
z+=IAYpy&~j*HuVKD9!N{)04$lvF^wHx&#?|ATuNz88kdE&tUtXNksk)h@BeLad>s$
zghRy>sZ~7ciKb_eb7*aZHSA>~r4{pv=FE5*c8xo=3N{5(cE!!~l2Ox%i)Njk{1{va
z-x~l9Gl*=t^Fqoj<P8SA!zpsy_jj&pJ7qB@p)Ve4qbhy7s%Dx1DUSwj=w7z%`w+(a
zODgJ&*zJVF?FnO{ZZ~&gG-tYIyXxnsbzuMZT3<N&6=%Q~FsO-0Q|9>f|DmU91sVXy
zp!vUwmAB1$URO&An?YtH+8y3(0@wC{gtx3H_eM|$!_Q)*JD-I!!tP1jS=$}d`0>Ei
z8Yi-Ev(c@6M$UR8o0+FteUHPhAt}+(FQ*@2b*}7wgZs6gJ(%+#P!VsoYw)fCty#4c
z+vh{|MIZ$jt&9hVivWLst81q%E^eGh)%5&scW&_>oZIeOMc-H#4=?(P>tO`!$>T}m
ze$qb7*P#1MaBap%C!zx<2|Hxo!<v<8ZWWb-fO~ALM%y_}T7SK1s&j<(b4c1A&{Vdp
zLl67sd0lx+L)6|GhatIb%=uIpaLl1;RgR_@t^bBJN?x!c9(|I}U-kSnhdV|rfJR&j
zwi%>hBUa6f&pAQVsv}M&APfL7@}mAm&=bZdjh2Bf=)+dq1s;=E*35CdzhYfiNS<Pi
zG62<f7KK&wo_uJH&tRJ$^;dOw7f|)s8%MfnCBp>vag){Q*@L;#FrN-L+hn{OPI%0Q
zA5^K)?xtUU-epp$anzt2#=d<lTpS`#fRKif>Y|Kw6fWfrWrey$JM;v|+@cL65$YwL
zd5f&qhbz_{Lh&LJcDFjB&zK-*!q)lcdSj&jmY8zVX5dJ}bf08~i@m*J-Wa~~StSns
zx)OzU;~0QH8y<3%aqMxfP43f!vGg6Ei@&d^wB|AAGZmxv(u_mL`7GX7F8PT8+OSH2
z?(kbKPq*pCB)XsGRRI&g4vA|kd>exQxz=SB%RS2H_UksR7eD54vOnpFz+9k!s<M(x
z^td64htYW`(S0gJi!%KXl48^?@wQcmKNaY<#iXP8?`wnxy<#i!GZIjLTOXhgG%Wb(
z_li_5yhhLEj~UYOqo}2M3#59j-wxFbwK-XDkE{Zmkk(HfE!Sg!7Cz48PKPJ%HtiWN
z&AENg*GThxpJ`8$lT=InFYb`xnOtyq*qsWvI@J)E7Xy~N3!aC?704x%5-obnQ%XZl
z_0GHS<PaYWNBj3XxiDQ$cZh8H<N)YWr65x_@pX%bi+yc#8HDL1p4khY8PT{aIO!|f
z*il`{%bAoHy%udiC;Z(eoH<{AonJxKR)-GjQ8qfRQhb|yRa7%E%j`96sz@6|px_zg
z*fz0EVY(3Kb(awkI}3}Qec|4ceHXDci_SAtU@VG=<ojh|FZNVRnWnjwJcqCmt}^Et
zzaC%zuRwt3uAy%zTYqn$FZdhFSTC#STufO;<=<0U^Y{orKV|+I2G$t_RvdKjm9Iu%
z4Y=-x09FCK97DQ7{v(2mO&}miujx?-f=)`p3WR|zz+MNB1|+W|A3Hxf_>V&MpIo6v
z0-zskI233d@#7~nDUg-+_O3lmG3BQ3HVINFlJ_^HPS4^D$qaUTzLOoEW__1@k1~l~
z7JX+=_U3(nyJ>FKX($CZNa@Bt(%X+?^Y<LCiQXSBe`k74Ugr#bZ35cFlK!sj=0S?G
ziLN6hZ4)cz+d#g*P6>q^Yn)YmVBFG2Y+=GQ?vMUPx+`bM_aR^PjL{Hx?T3uB${NQj
z*cbMD#J$osXdBw6-zb4JK2Hq6lDY_DsQ%5LSaxd;XrPkpeC5US00~oN-~@CTJtQMT
zCNw!WA#&$Ujzr0r*dj}E8&Y9h2HcKOxK;L)Ty7!RX8&UJbbw<;0{c)7tkp}caw?9J
zI3R)*$CZnxo{A@?hsxVy?_c@ckn<2(*S@*Uz}EG%t(0oTrCYPVU@wP>@Wa#nNjg_?
zoyye&P=#uczOT%85Z<9%c!Dn}t~qf(C|0gYhNX)EDy$yMm_$xI{+$?#58}6cvW~P*
z*w_NsTI{6a$@%#k5$euJfiX6y2549QSg?VFZ<-hi8N;srrVp$xPpIeP$Ox^gx>)I&
z;6MT<z&+BvdI3^Nyf2f;$b0-1N@%w@o*q9QK#|x&R$n`(+wyXcEr{QnCea^%mSGX2
zr*JLInEO76f?U%wGOyAHO;X7N3&G>pNX!DYr-f<-4E{9(HcN({`}|k)9kZZiN^b3Q
zfO4`?I4oUk%_xwHwt;^&ia#|_CsBmH5HbtBJ#luB`dRbZ0q>pGs~-Ff=FvRWR~)ew
zV03O8n~vM6;Dgvogk1n_G6(j3IdgCXfh|mZ{J02>Gi{1E;$xV$I@yM4Tlx7`*9fCi
z*$%IYlICs61Ppow;fTPKHseWsdI91mSc_p8R-8wt36>(RIKE_>3l#BCM(?eaSC<Yg
zaN`cdvmMh48muAA@x29go(+owk24V<o?Hn?6^tGAYekskh6EW;uCM7+B0bMQiESZp
zT>`r+9_)ky`msfBb2Y9xfgk2-^#b24ZHD=rDivf&oV!eax55gciEv_i5A)sW!na;?
zO1f)Q!!AVA{~ey9v*0}>4^MRg4F^kjvN0r%XCxG^!j^#gpd-LPU^uci;64$cZGtck
zH7rGCx+))uU^$B!$R13B{Eiaa;GsR8@`nyP3U@1^fA%IoG*wj34fdS1(Lnk!8H!yu
zSd>n-Jfnz46<qKa=?#WC18jz*y{;bQS+<(n<_WZ#^hz>5N0CPB=pG*N&1ylBVm|E?
zVOI_*yJ3YEH7BF!l8+uW?}-l}y&&u$8Q+<L>GEBSEY?wvs~<oS{FVc1Ap#YFSv2_e
zYs5{{KCtd#_-py^A%EivE>zoCulgkWpIn<-#4?aMCIU`$tIk$Q0wG5p%<OHi3%1wG
zFa^ZsPDDeYg9b*yt+uxh@0W_VW{*25W~J1-AraYdU_snbZbFOG&;<~{Z)lu%G<45y
zS-uzDkBOBxUpXA)!xM-z@5dEPniQF!3Xpvd$;Oeiuy%+SJj3?eth!Tni}y@p^~|JA
zdFl&R?EZw6=(Cs$?l8JhVv;RS{eY&!gGpza3Z{i8elaJGY0*w;VUL`$(#ukxHxKuf
ztIZ+V$pK7h0pu?26z&SZRFf`<-w~{~K)%yyn!h1G{R^uI)y~y8H*ul^C4_Tj*&3q2
z97D+vg$$F}%pEhc-hNBI*rI^JnmlMb$MD$IRZzlZbXYsRAqVZGvj4eOkqcDLoiQkk
z;Y~w6T7*($QwIkDG&XQJC##hX)ya1I1mC`WK6ORpkCL+4fl6BdkUD$JwuKU|jOZ7(
z0@%D)LAryuIvW9aF~zbj{dB$NufN}Iz|-Fs!Ec6XXe3IRE*8T|aP9)$J3##HMzuua
zL|;7YCJg>lH6Y+Kn-)S_yO9W7=i+zM&MuDv1{2{UC~7xN5&Z=9w~F^f$cwM9W)`2M
z7G`Q{lFCW(P6`27-fMrL65=A2u8L1L8pU%aoAs*^eox`HCZ^?#CM%t7emI4_M><@r
zEO8?5kOe$#0uXBJgs*bZ(d;HpeK8c4Z<oAE&Psx|{*qNkFjSjHN<M`KE5gcK@;si=
zQ$FsCF}0d~@c6ALrCRNFb1PlB5q9Qp0c&}57Amfebn5_IwXz?93g-d1;CHtDEe;YU
zb6+6q{MRL0=PkVMU^^c>X0zZ<CwZ4wqG58LG0HGugEfRv{$z%)EGNJr=<dz;+Frz7
z04u@f^BBM(PO7Htj~MfmzjhgPM9ex7gDgL;8N-i`ym~bf(-uuO`lpAars+eY4=GK|
zsB!YkGKm2TDtXBMGiDjJ2)m9@a`%C7END;}6nLa8jXSx#KYe|f03Cdgh^45;VZ~X^
zU;HR1B%@e}(?ziCVOUl!iK0%zW`ufVd!7Q-vk{T(PMD}L18leGw2``?{dxC6!GZbA
zxrzxEzR6ZHAwDH@M!N2BF6xx+rIZ;HTgT%aLlpqR0LV7b16ld$erAN+I{ctB*}~$Z
z_T!w9w>SKhk%Zjvs9yv8$cVu7>C#N9;ge<soU)EdyA=KM#*V1wWNgXdkt7AGcXS!n
zRqZdH3Q}+52A@yw2+?KcLw89f=YJZ4w!m?7d2zS%#edRyk=k0BM9Mngga<95Q>s0C
zTATuAXfiE;I_e96A<vddhCU8qI@aR)A<t^uv*TWW@^i;b=xE!Ev9q&iu=y<KeZ@dv
zguDkkF=(W-zNhMUPT%nD=WRTfT%2D&(I~e|Z`l&{t(ahoz<RNXb$W8l$8Z@J(rt%s
zLB+cyxmq|G1}(y3ly?d@>i%UdjvFNxZ<+;QCD^!W<Nsip2rLj78J7zzJ?v%_R$nq~
zsBwjV-IC;nWn)Xdc9=${KcpOE%sgnzlC9l3`-FQ&C-QN|Atv4p_UDBK#ym4Rry2~F
z)a5l~aZv89m?|4pg1|*l9Azb*3Feb^idp7xQ)ntt8l@$!3+_gF)C7A>8>|J>otgn4
zH_DB&5KjctnQEr4kroI3oG(`&579!QCk9485UZvlFtkE1P!HBvJBF1Eu!ReeWold9
zZ(5#Iuewmb?_$C-Uz)&i@Fc)H*i|Mq_J#l_%(sWdFYl>?kzgsdj3k?T4)(*Hdt+;q
zWFS*$A&`fPh|{jIxy&p(P761Fa6<wFd8y5a`Fjw;kw><4cK($GeuHfFb5RdrO<L!5
z%VRhhL=~d6EV6jC8;%9OURCpF^!7K!MGiLidz5)v<+vQT>vbD5xak&b(r_Nq<4@f8
z9400{yn#-C28N)@vDVY)Gifk2*c2`{V0!?zTV`i`2V)Wj^r+F~3cX7>BDe!0AMOJS
zG+m~A8t=<GB0z{4$Bw+XD@T7;$2&A{^2N9uM6m65zuzXNRbuM$bKf^E!Y@z6A{l>q
z_*}f0U)w^!FK`&SQ~2Ykzk4*W-AT%`l$!JLi>hMt89Q$w^r8fE+a!*fDKzk(lI4GP
z(k_f~md<4T)x`U%)JmfKwDJS&?6tk;KSY~7SvV75qMf7ia@i33CJt37qcY-Yjl(u0
z)l)=5tBH#OnYiP=(LBTGv@DlJSV3g=;&lW&3?p#+Q{-?7NxrY;OjY5^!Rk0C`Q@>Y
z7(O9TT!RN#vYgEPFd6<8yyf2#{nij+$_x`LfXB^rLcHc-#A_lX(H{a}pOlsJH_bpP
zRq7uQCM4)`FWsi8RaAuseKYxcR}1OJ@0Pp<3kradR(*E6<+$a_f7l+v<SZuE#_^jf
z;v+$L&)I{I#3c{ve|8$QW1AJvH}*MEN3}^CMW->w+Fq}Tn+7ok)1M?6b3m+O=WH!7
z!S&^2bu|^HW|kP$Rn`PdkJz_#Oe<;GZD`<fT+9f+H1!I<_&f-9=@8y6Ywt>HS<2;q
z9|m|x4V?&<GqNDGSWWEP)03Um1`=BgF3BXJMJZ%>sEycp5Am)~k=YJ`BPJu!yf;BD
zsctMNEporIxhzL!<vD}1!~4D<=c~|m7+bp}K_o{sLbyg6+W}N`nkV<q^?LBZzAW@T
z_I<;=B_T#R*filCCsY@Z6juDLz%}2?3%)o;ilmEla2~+cu8yI<C21J^i|B_OCp)IR
zmb)S`W)?K*K+&y2P9g>KTt80Ga0I<E*yw!9Sh^L^tq;A&CUedgRSgN%qEbv+^r*N_
z<T{4RB9~q;0cd#}Lt%c%ESJ)G#Pm79y@Ezt@w`A_?s0bcxJDd!0S&;GsOCVn%)K`i
z;MXiAjF2uT9$@gW9C5qhG7mDf=zowS)b6RPe~0j`CyT-kuHSa4(;beZM+&9lmmTf$
z=+yoth;SxK;gm_SZn`LRI<X+6!9>R_BXh16`Jmcd0a)z?pMGqRaf-3xro&}^wDyO<
zncA5iCL$?h!&3f$XuFG~yE~H%p4(46;60A&z>&S{QmyM0=4mP}FnOLvL@gZ0P(Ru*
zIdH;uK`D#`Rz`1DO$C_J#1~~Wm0D><c$;lQG1JjW<PO$bVz2X*hKNqUF;t0gK(D#T
ztQ7}m0sM0F^CRm~Vago$&<3QT%II}42B^<IX|wls`V6wUMM1$U2bht@(`a7_V#z+h
zvFk!jE>Vk(vs?KW>3Q1=y)lB{p;HByoTxSVo}RDw^=xI=xKZ4o@RJ6BDmNcgpBos8
zv^p`KrQP7Xi@OU7n7$`xx9S?#i!MQAns}?T0FdxxI!k+lb}wTC$H7xrrH_gsW#><p
zZ51#uZL>aepAlh_uO!hdH9apqH7=0Gkv9yqZCBp=JW@Sy??t?o0v8qYw?9!B{c?tS
zOt_pk5p54P&JIZ6^*Vtk1B63g%l;B<WW<Cr2pxZI=zvgVnouk0Z!-r$WRt*RA1C0D
z0`jUjWZMGsA62lYjT^+wU1&q>;cgV+^C5DP*W{SHdH7q<V9?hCsmsZ!7xx#g#;@yz
z=KC<T5QeC9340F2uj_UVho9ZuK`V_6`+M=?!SgP7S<$vU)-w~wK~aMgg%;?U2DUdm
z<cR2?S}|-d_kxhu`wj!oz^wlYwNX5C0>;?SL21{)#vEl?XxM5P=IbY7Lk;(^ws4?6
ziwO_aB~MF-3L>v1FfhQzrKlbTX&#{2E<Y;C6e!fX<5y0sI}p{jHkmoHWoNqPY1DCY
ztN-|gHC8%D_>FN?MP9ZwY^?uU>B@0CsdcZ<#>illNVuZ(C1i7~v$|`e+Zwu}1dKvC
z<p^N7jq(|pWYDKOR1FQ7N8pG1@N#wk`B?2&O-+BzrS=^v4kcGJ-0Ongg;i5~t}diY
zCn-0s0Ir5^?rR(JjkM5FQg0B>Bq(=II((~8mZ`G=O-|?|rMPu?;rq$RHVdALKxd|N
z2%Ne@P$!4h$!Nc&`FES&FM;=C8_;m}q0yo>!5J)bUjRPy+ay%z@j+#|B3N*06mq#1
z>ycjFw;C?r+J;W$gF|gH17_z7Su8xk?Qqa?wpr56(|s9xSHZf;LP}JWxIUy$Kr;ZJ
zYe!A$5)Aa7M;M~8%Q^yYWMDFwaZnQb&!SDg{kYm(en{UPZ#&V;4546s8Q_*TLl8(W
zNhu<0zkB#kpAe`l=k^#T&gl%dpM38^s@?P5WkYF_EWr=7W$gggSD+O_TT7Hb^5t!=
zG(RaoSV0e_Gumwn5o}mM&}JoN_XJhcz#yIr3ib!<;_>gCI7=Og7Usd&BKo4Y>4Ni=
zR>DgIe-kfDYt1FJVcLiVS^>B<>BfgBM5W5+<O)~9Mh9>*t&CzszE_C<VNV(AD+jA8
znvOHDxGQX!6GGV|S(M~hgc?vBvUPX}vH){-PJUh)cL|F;APdFV8lrh^WJTPQ7)(yw
z8~hAo=+p2U+(n@oNgWQvOz}B%*h4^kT_twzIt8^J_{Uprg1sEEGzWk$zQeg0-G5Pc
zPj}(DuzHrD)(B^mou-6GwqYENuB4Hugk(Y_^kZe!6o%)3t)5vo4zzhnKA@=h0IH{;
ze^Z6YakBs>3Z8`C<M*nmgSRl{K|2Zd^N|y51R5#5;Z3OEUr?f<rBZ#JDQxQsd{O$e
z0v6XqmAc5v!~jFT^dkV}_!!qM5cc&5dTRG??&u<EkYA=<grpN0yY<IZvy0`stPuBk
zMoojJ5oqGSc6TUvrqn)lQZkT2?>N=KvY<IoZCeFXF^>1t2x&;c;RCn{<!leqow8(J
z5v27Oeo6n{i+4lw(#rUON_>s|%RmiuEh+K_Qf;XD8nOJ9_Jt165!2oC-f^UvUN{>{
z-iaK-L9*tIHRib5*ytQrI0{ORm2VDP_#WztMy*M~O=KzX3etix{|Ovyv9H3~!6|vQ
zC`S-Qv4W!#VzDp9$_yI6b1X+NH<$zRDjggmK*GPfsexKQp{>8IlmHHAYFd0z4tqq(
zd6p1p4k=02Ag2kKg&|H&mNfKa=!EF)116G|;6Xy<yC<6!g?j&8yW1ie0iOVl{;<k!
zLQ+X;YZ;nu>fGd5zltJpUvbj{x4Ihcd<z#p%JG0-+%+H6jMo6Q)KU}vMWn9qaaBYZ
zm;tvnx~A^^0nrVFqf0=1;L;^S)ng(lowspD9eB&$PCg4TROI*7U=ttsyUtzSG8E=U
zF+S!f9AtZh1q{V@-|3budS3{KL?7>5>iUA;^0wAZKnH7Z14(IQamqN3lp!>Ck%*kg
ztD2IRo;sJ+lA-t(_F<a=)i)k!=@V>POdKS49I%Uy<6`m~?YmQhTJJ8&XV>AOJ+`G$
zw4RR_vuX&?ZJEN&qAScx#I+d+MdjE8=E%mC>E+6ngmW@pa=koS=2-3<ut=)lab{47
zDNBzsE|8R)MEhNhO%n58q~OtTVMSkyrvAaAS-J9}s&+BU;xg&#%Zzt<eRE-a5rxu7
z%dz&5kqlMwx(gFO*Lv<d3<_#oM{96I4n+j;198NApuM=FO*I*Tuh;C2goT3H%?A0L
zZ_(da_|q|N9=jh6Urr`Bc~QBC&d3vFQ%_oe*Al2%Poi7RrB<qGgRCKm?*Jizh-;=s
zoqU&4h$-E^)6fRCp5`MZ!j;yLB&PZKKO4p}xINKk1TnfZ{Doc!Km^=}E00b&ptYh9
zyzFk`tY$=^XvgpLUa0Er8k~*b?RLSpYv)&5UZ-LdXB-;n(S}}C2Mr~rN#8Z!Kf(Bg
zg{FVXCmfXbW-IhKiI7V3;#6=Uejq8ia4?PFw=4G7{vbJ|fWMp`^vL0X(eUpRvy&K8
zj=8>MdUkcv2AV1W)`iD7%!gTXqKP!xVczc=$?ADSX$;5jeOwu9IdwxnINKzm=?Bdl
zJ?dd-ze-G1l;rz-cYBajj}^&XS;-8D#mibNR~l{evoKAJ?3EBn+=O&{zflV7=4De>
z0OjEtYHaNjcI<d{>ZZG&bdOfmS_4h5t#T=Hdzj}q<h~m~p8Qz!PrG$$#$%S$9D^ZN
z7dM|C&7WvaM}vP>6XnWa&i%TR>_GVUYCWgRBO+r$O`|}?0ppqU*cE4&v2+}Ax!z7_
z!^Dq&q#U1<>ONxqvX7?BnAfk~nz3sqM1la+eRH?oOXP43hz6lu61X$xb%)Xkll-v|
z=+7<^q9Q;{hRTJOZ>)%a7dOugFCaUFq$i;GMtWPq4T3d8uPW6eTx>p}QLKITho3=V
z2sRQT-mUtNV;%To?U|p*o=KUXlOv^WJCQGU;mwLJwTDhkw@CamBV<UYfp;Qei@tHG
z*)ezV7fKxTqAkq5N1uO#Y$+98bbD<z(u;sr8YBSNp>ES*@KiqHgMlhm`B3ljEq)*5
z6yBsDNDdaW%}Ud4IvMhVTSBTN2eR|3R`3)k9EeM7URlC+RPe4lvmgP@22{to%Ghek
zLe?hyoBMFygQyc`y?~jo8_2Hc6z~cc3UZd4><xG*LI75)XoW_CO>>!?60o<LQNG{_
z3^qWpiwu0dZS;TVH91zw2^_88gCR@GkC&F9Q0Ea?>(}YFIPCUKK?l+~jKywFC$=rO
zl@iG(TrRC@NdH%`seJYVn|+=+o!)8+EVRSUd6}`t*PGBhPbp19aQl^14>AC#kEJHq
z!bK6FXUjQTEo;kOjY92TO?)_i))&<-47fWM>e%o-4RnJv+B=IBFebdSW~V!DhD1vF
z+(`fE$IyGTz{e=^5$b~MZT9YQJBi)x`pSaq6uNqE+VVa1`HnEKN*tv&UmKvLZc7-Q
z36!*cO3LPyU=}`wJ&Ja1KT85?jXhhEm$furwo}JIP>NEd=#n750<{ckPi#u2*hl{Z
zY*MR9IU3R6XH}=-hqz{-`Yh2C1o!?^YswiOx6+?|&(VqD0ar$!PM!z^a+CuIV;Uqh
z$DIuGp`${->(a68;lTj&h&r5K{E0YX&&&kY;S5mtCQ$RzxPUS6gfz#5;FFhnG0Iu2
znKG#+0U4YUk=}WzA7{%IE+ayku7S#41|6Grwb{Dam&j$8yw)>!yI7DjE3y6e<jV6-
zdUw+YA0M$K?NW+^L*je6;gH_Y7ATWwB&-SOS%Fz#oY3|#L0MF!1qEViXwW&CmSbwV
zLsK)(BI}xu0)XukcOyP}+T1)keN9Be7t&RwdSY^pW~x|mhhX?IjHLV004^>4pCTc7
zmZ6{3g+CzTVR3OB;|=t6HWh2*vp%qV;<vp<Udjnqk`~&lQ;vqo5v`u91;23NtNGX0
zUx<6)ewnA$I>TuqHT=s9B>Uoj8j!<qR`6C_Uu2+R0YqSP8}B$kgnBHt>+i0$M#^W&
zz@Mlsp<QTTFGzTN=oOvs+*C(fc;)kee}M%5;*5z$k`rNq?oXELKjfZ{z=v8cUN8cI
z=c<n<K1_;N6cO+oEt7&w5+~2==HqfBLb{b6r6kVtXvyQFylS|6wi!4Bho{StDizO)
z<j5lt1z_8b(!j@%zxz6i;nqvf6~-Hi_(8v*r@cp-pOUeLqkzak)hj*wvjVrqtGm2g
zVVtsx%m?p6^6KO`$H-JnCkCYll27hZqieSshJ4{H0J&OEGd#-OzMsOwAEySgBl-c$
zqP33@d?FfR-Jii}BTODmhOf4eNAN>bt3q_A0p5H&Y`$DIf0Yf6Gk`uy*`z|v1-lIl
zh)9)(>{3Rj%UcZbMe)HfnT)q_(hS{RmZUK`{SpmH_>S-+kR(K~hOiud9%t8iGxt+0
z;?y+uUkABF>L`#7HGYYfcCfPIjmr8%FGx$#Zp%~FcQgtujzKGMX)0*x#yjG0)T;w(
z02x&pYq%%7BJ#;oN;Yx_o{rH$l2*f}Y}DV7B#ZhDNw487X35i6Iot0<UU@H}39u2W
zcJde^`=ZTF0z>DE?<maG+r6%{r;V)0MB9*}5hgWTw{I$ap8@c=`s#a=#Y^Ci&m7Yc
zl>Q3S5>pqvMT`)ArmDcb7r?POio|q)yC%)fk?GT_&10Liiz&5h%Vq1@^eRaf)qtN;
zP@X;@6n_4|GG#cLm~mPc>6q?F6`1nALYD0rn0_GX`M(Tk&=raNNV&pzhrGP{XOogg
zl1c=p<k6r!r5RQxl+F1~LVQe7CKjYPb4Ptr(+mx6RUS-i)L?t)T!S7ONySzHf&_%#
ze%O3j!oQp0SGOV>Y5cmiD7Yp5Cr65{{C^z@;gu}^YMpD6`+sJnogAtERHWlc3I?iW
zN;k)4kyq5P^=AkKx+8EMC@s_>u*yPWpoXp(UvwYv8VZKe|3IW{z+lXODiXe!G^gMy
z=wA)abd2U>iVp#EvED0uxm{S08MBWd)(wO&6gSXim9r6w*dYR-SI5J?>@c%~fXi^;
zHBhAGXm{q%Jad_J5pY;YUyo-VlLjaX60xw4?_V6SIKd(chH1M=XdfWNqJA!nn^IHu
zs!}N~__WF%32;dQj~=wOJ!l$fxr{4j`{}fee15iyOhxc)4nakTZK}C&1$?@XQj};J
zUA?l0j1bFYi?zS;MY1GDm^c+tYmCPAh-$i8h>vSkTDqekVDkUa^^VbzhFjENr(@f;
zosMmH+(|mNZS))4wr$(CjgD<69Zl}N-^`jHGwZDN)>E~r{yyiNy?5;`xzdD-WCZf(
z4=grL>1+JQpBt`3uD0hE```45v^o)<HHxZKDi-$4430a2oYd}sZUHOk$!wBZm3l(B
zTu?wf7jQ_B&f!~&vZQb*m#(83j}MmFs&~VfZ>m&OfepK&GV<MENk7Eg;k)i#WyX=j
zHrQotG~oc05`5R|`9+ak+B@0nm~XpDd<x(G9Rp9TT1k?%jLo|sPODZ|V+YpxHrJM5
zxBI$0uSWdZD)$W_wBH>>%br7w3@x}hPgH~FVGKn#ysUHl7oPI!HJT8x3mmd8Uc~he
zmtkN=gHN#^Gu|XpfA{_LmfuMREirEc&qD4DW8}*RAv2~G;P=msk`??bW($+F4GLEc
z66vaxe=EYHmhsy5e!2dHbkuk8GsT&~%d5!S3zDFzb~G7?{<-<$ongRW=V~L*%v_1^
z5dNY!hu^u^2dzWGxgv-HJt-RduoJmSqC1J?;VQpiP?ca(f`#IIIU(1x;wyyKLm}H7
zZ}7`gc!Ho2q|vKUM`M-s>~!iuzHjVXA9s!0zMf=Cp$JbQNCV!{N-8dfS@qwg@IbYK
z#zbNeO4oJ(QG4dn2MvFsHEk%BoV036c&Gvt-CNG8pvbk;^XXaa@?Pjp=+bvegpiG(
zO}q8!A$K0^9g+yJeEN4Wg>Tl>?vVdE^NxuqdhU(oTYtU_as>uPw|v*ARPgFgaudjx
zSkOHm<3&E2F^{BH<7{1LG7~des8r}Nbkx*odQ}N1e(c6wq=}9q)CMTkD3+S}WYJv2
z{Sp7hUmo%X9W0`khiN2;E*op!$INo`x<qRp)8$hA$Q9Eapv+*vkcSvReZO{Fl~~@8
z#&u{%pr5t+B@FTLQ|O^Qx50_aqyF0G>o(4{6@qf$(pfS{FqE47wEgcT6-SLMxE09T
z>C6D;2kMV*YWDmd1?<Qe--h0_OhNt>5TlVW8@X#%#MyAatz<zNj+z5~O{A?XXYYh5
z2lqj+-0rmJdQvaaqFJZIJIPr|bo3X;ph%3{!YSItP6IEX%o<H5vhFCZ&9QLJj9}$q
z#r^;uZ@&&y@oIA*&#j=CbxZlMU8>Khv=X2ZGU>U=UXA&;d3dRED9|nrEgsspqO?Sr
zI*Cp$h@UyN?}uS}+#cIiHq(b=`tK!|Qsas>03B(y6GHac#&1DO`LB`+soc(Xu7v7L
zXWr!2v93&Akt-)-)Q{qQVdje5{XW+y<;&Qxgs*92B?J3UprMl`(R2KtnYLL|$#_6q
zLD7G8NQj>niBRoq6i$U>V2&@vdzrIQvt;3Rfr6J&-Wj#t5#~I{+ltMa^5AqVT8=|u
z6hccT`BXAd#YV-#@<*KP<qw|Ux>Cwy=d_UJnI2UNlgtZ_XeBO^MuPJvk6IQzwHw`?
zt<WiqY&OaC_42;G&!swQzhlvxL{S0ghI=u$6bx-c^Dk%)+z@n%!U&Xchn>?IQ7jTJ
zTG7XhxSz~02mby@t2c;864go*8+v2x3)8wy<n#5&`B^+L?_X9$ELh==Du}WN_77@<
zuPL*J{DneJx5I}<f5X*-2WQtd<F}q)r$1c^_DVE?`ty<5L_8jSDt&uIr7-Z2*pfLj
zWy07s{gLIL*fO(>>!nU}aB7_7p@nuK#++!1UKo5VHYbsP|7iD#B@{uwCi>4@bu;y)
zD)(0~^6tU6*xab(m6HNt7EGd)KCuXaQszH6Up>kSsV2lZ=%Yu1@A%OK0fDl>aU|et
zC&wdIC!SqG$S?XriBl}b;Q_qxI7STI#a*T6Py^(c0E1XRtyZhQHhP7`M`v7L8mOlW
zm7iS6#~ik2)J;JZ&zGQ&yx9dTMa3xf>umGXrQy_~+0qok{VEtl=N8y0e+qe2fuORA
zPB2-HFpz&W_E`It)TA|kaB%O!Cg>FCn8XxVFJQjOK|*w<gHgaqJ^}V}MEVwevm->S
zPH*&ABP5B33Z9VKuWrtf5hjw~7?;AF=vZ*sNvk`|IILC@jOH`(hjl3j@AA-zo8QgE
zmL)wq6N8FMBO2*<I>n?_vTP;q?&?4SHg56@y*Uh^=$+)_21PLv@yt(q&eO^VA`gn`
zF{@CvM^EQ<KiHf8fCCV9he6`aC8zVLVb}0YZi_Lk5`lXVS5Ud59W%U2Is~Fny|Fq%
zP*Owzg&7+OG3P!tst2R|YEhNf?evHHL8gJaok``tk|wVVh=l71+ztfhZ;{vS*@<Tt
zoFLpkXrjw{*)EAv@ZIWrR6a<E54}2QVZrr>BaW9t{YF#vQGtw9kMJA9KVWiVO;`$;
z0wP5W86>@wV#--2GwPH2y#~<Tgmivkk<zt;N}FW^4s*{GiMgcUQem8jB=Akb_?~L0
z+)b!F&s!{5OAK%L`+-Ua8l`{CI;MNQtzqwR#@zgLKa@hymlYlOjwY>b^ipX%h#$zH
z1C*eb*8$a_B>=4DsP-mg3N+n1#R-&(p_NjjJCY1l6m4J>Ga+~)n_cB?JlE>GPz0O)
zQ1^Fo#bj?ZSRSNI<?|48QE#G*{c%7`G$mz-2D_ElpH<Yt@&s_7#a_?3O;56u6qa2;
zLjAGPQ}55mtmuzOxfR+M8j2GNswlg_3F&=tD8j5r3fOxc|2`Y9GSZm*PHeC{qCtef
zZu!#IF<}H9WHK6%k1o`NPf6qF2D=jymV?UcZBcw%KWNN`G+6q!gb{*@FDyD5i&LO6
znry^e>?!B(mW9Y8yImEb9|k;~8{HS|c~p_;Rt?2K$xxAWNTA6H7Kv%?f1$$I1Hw(u
z_LuYG09DV5o8D>HnxhMb^9@f#sRMX^G_lk<)`Ou@C&E0ofvk&ZnN8}zh;%LPPRD8n
z+1_Z9K7(C!`$ihJ3mpA|`}6mwW7cM35X^=iKB%5U<B!L>o^L+AzmI>aNFWt{&VN>J
zg8m;&<y?6n!3&T5H1n&OZH#T#QgTeg0omUSZ@M<!--f^$I`yrZ1>dhdcn)Q=5`7&t
zRfgi`tsAe56}qp+9uB*G-YPfGo_cH@TrKrZxQ}No6Jwa$l3ny6@ffUdx^Epn4#M|j
zz(HvK+x+~$^*Q8gDQMwgtV?#GJ`hj5Mx{&fYs$%F)8D$660RX$c<d3jHc11dLa?vI
zUCqXIdf!x4zc1&gehDsK6}r*~^NoVlmGGCBjm~NK7U4>{J^nv>;>{Y}p{48Qr|%oN
zL9i4C6j=4^6yLpy7qLyTv1K03tB|3sP_35iS$)G|<cD;vx^darYQ{8F#o^V=3{z~X
z`M4W^Sb<^R+`^-Gw{kv7CER^*M@FtkLo+AYPnBs9_tMoUISg-7;F&0P^LG|W--hgQ
zgo`K|N;Alkkywt|b3w}?j3Jam%`)1|*r)QW`U0F@LrGJB)#)5Vw%(}`H1$m;nI{IO
z6n)6ofWh)oTlZcdoUu;kNt+z1fdzoHmB0afqWusNM&xuwFSu@)<-#K&Dcn@W%?A@2
zTFZ)(@oIj_Z_Ovd*q3|MCWKqp#n}}?k%tf{%z~z=T#F)_)yYB4f{uwGnjLF@%b>D<
zdI=d)5Z~x0tI)(mkee9j(rRP~8%<?3h<4=_IrqkhGM!M@z4FG07=y^aWFnkuYg_`l
z*AL7z2DU10gz*LVKvhrE-<g<ittxJUp?09$S#0fSmAVwF#$wgHg5VIWrFH|$*6`=l
z9z^>Pdjnf1a;aPHwX60S&tj$PV|wkI<+$)A5Oog_;1%gT?Fia({B^xbIS4$qYv(^8
zJU&k!4j2U*K0w``i#M!!y<$OL>ga%&_4loPhT~ChFCmDn9_{9jhh`U0R59Jlrc-!q
z-iG7fkFGI{x*M_ymj7oQ!*s2LGJnW<{v~A(SK^wML+eV&BP3~mZ@24yau|YFZfJ^L
z_TsQPKF8TY#w%Yv{PB4W<uN%T$z7*<j^q@}E((G3|I7diXh1mODE|azcsP{)m*LU%
zzg%a<JWL;f`?j&Bz8<dRFYqE=7n3pj=uQj=!0;l#^gx}v>xZtd_kT9d%!7UUeM>W!
zb^)~wL#Ij#K1zse3TfNAuM}wCwH=4w@{u9Z|Id<Cw59tVp>~2(o*Q%ZFRh{p-B^Ui
znB{-R0AyoeXcSbjB8}_6_mQGv1}DF)jaV%d+o0_EFhs)dR_XFoSoEXXW#2)H@0tAu
zU%IB3YiOc~A69;nD@+<#G*>pXR={_fEewYjyFl>`Ff_g$&R{h79Z@3r4(KCBt3k!I
zRZ|1Sd!k@6uHqS%gk~px*aqJ!oufUcYWZWeM1WBy(Lq#yh-$M`y#}=tNyj~~iBXcT
zEY?QP|6=kxsWI;sSD!yN9L|kfTQ2`yN__n<C-@He7>FP?@n~#wS%y?r)mkQKgC)%b
zEG{<FEj<Erc+U1+Ow3d^%8Sov;`u#13|s{T104J2Yvz<mjsBQua=VL&vvpHgd2-!S
zTOdl`ip`<bYdpMsvst;yg<;nAaoZwxEG*M>7$U9ogu_lwxO$2r6}zEkG<V0+H<PB(
zq*PPS;vYa)B+eEskHa8P{P<h#V6>=tF|PTRFGX%pyq65>61@G=4h*k@6`{ga!52Hn
zX5xia*Q{-o<#EOR-r3j{VWPfMIuKtA8c-%HEt14Q8b^*_TgZCJT&v~F;(~4_CB&oV
zwLkP<78i{<<PRFwx3#utU+L#H8PtbQ9~fP&F@Yw`9z|fjKi<wLo*`ug+rEM+{inA)
z^nHn>%s$#Olzd|ZJ1WD*RiH#Y(o1IL^6yv?OB4-DhgEPS4xjAJKld{u^)7mL27nY_
z9Tf!o(q&dd2%A`dn=P8PD$KP#h^b3H^vnmXTCVfOi+Om#ZM+e=Jw1jQG@zM9l1JR$
zBz7Ad@Sk6;_NDQYk_k-s(}@#?t~@Mj1x4nz4ys`TVT9oC)PP!_(!NTIf5s#CaNW3U
zAHAzDJ|7q`9jQ^O$2>t7f}DS)4In~ua+k!*w#XWfTJ`xB9;YeaAVcZhu>hXf<__c)
z69F!oyka?<H4sO>k6tQT<=%|`|Ga-yaeRXB+g|8&OHdke@A2AxWnAAq*|cvZe?{R%
zrhN2cYcSC%S0fH3vnrizt9W9anj8OISyk*ZLQ+Hb49TUUv=;xq6r6`K3UI8az^ViU
zs4@Gn%W6zceA7fm=Pjh;bBNfHFlna;qXB-4aS~o1LL_xsT`{$co`@<0b!hIWVr!yq
zW9;ze-SxnLQ@qQB$y`b;F%t3%i3+(*l!s#$6W<lXqklS3r3)hE7c2|X9EkYFd7U|v
z?55>nVoq+ueU3H0>*`Q(01+e5rsypZjU+(_`{ds)?4u)Lg_f#!R_?Uo-tT3u{wQaV
zf7E)Tonudw=pF97m+S0SK|_ZsyWEE}o0fZ16%3~-Ax=AT<&H!bfB0a3o5}|b9B0VQ
zZyH>I>ZrtkP0qT=oERZn9}E$p_z_kp{=y<{fyhM>K;QDWotk=068LG_RXpJuk5*Io
zLF6KG>dHO7CTp!?Y$J3L1BKgiXBziY@L!p*WdZK|G$P-qM|Q3g#0)$>MXGaln&sH<
zFodX#zFrZ;82i_!Fm@mxppOp!yVJoLcY?=#tH(X$;AzjKeb3bulghBl)y3&HnH6VS
zl(~>Hf?S&-uU_Pj0zjQ%8?lRFoJ-aXLg5MisNrr=q)xKMuG?FY@(GNX6kP3x#(2v4
z#UJBt#W1m@iJ@QrLs)-7tPmSD&f?ltr4+U4r07yafRGrS0H4;S<oi76Ght3j9Q&`u
zdk4^a9Z>InGC$4POYB269kf%dFv(voLHjP&_h~j|;{%{y{4@M81Xrvnf|cJS=9nti
z97b_WZtiAR2a;;L2zOF;$jZh_>m+xY-z)fm^hX^*DHAMq#)s_%ff=6*&ikW?>Fn_9
zwd&<eV%77Ucl+HlJg<`wWcRa@E$@AgVevcs{m0_jxb7dbcgVH7b+=~{&|6~-qUP7`
zQSEm9dk$<bP{4Kb-}2TPyeyh`>v|c7(e+H8?dq0LmzO=O24%Yym-D42^4zIVro76U
z*G6J{G`>#33jbq(q3O|h;_vAW4SroZ%w$}C1SS?yShWF)0rP+OVvc*ccoZ8=E1a}2
zru>J8;+NWNQ>Z!TJrDHRTc89Sw+y!7l_g=6Bp`90QGi@`M&pPoYnVc8IYz4mhM4tt
z6oWo3{?b-2kMBs>v-^sJ_VNi-^n5970j7ya|LZB?KqO-Z?T!}oku>KoO3`Aws<&_G
zCJl}X87Z?jC_bz<B5X#m-;tD64VwoO6xRfoHfG4YZ6*(w3?hF{Jmh6qwP@o7?vW?c
zw&DRYd>;tVD?CzPl(5VB&aUwS*t*$Wx?TS!!or=<-^Z{?C=eK#c(X7ze<QZsyI&Y@
zg_e?i)z^)K*&0kLMb8gNosvbjm}?C6#=%5C_o$)K-K8_uhm_Wq5{3_n54;+cjGQ!g
zT5%_!Qxv2WWG_ujX@v}UjVbtyAh)b)S~!5Rr=xhUR=VcXBUhWhc{e_e@vw!66Qt=6
z6|U&?S;f`r5H$N4lnyrTOApz~W^d;jGd0(lW7piOIoOVK4qllar)T~awMrF9uwq9U
z)Xi?`E!i1>P|~G67weoG>wlyuX{Bw6Q{VovQq=JRVf#TlR^MxSVT!l0Bb`7&ZUG_K
zc{oW@&@`wX4x_mn)&FQYaC79!o$c9K$OVp!GUnF<1%!6(@sK%o#HQ~o*k}XS7n2U}
zc&nAsngXADL~~k<my0?ybC#2yTfIK&++=1)!0h<#X?WIJ)X~nPI4?!c9`zTbUp%^M
zDM2~6w)?;FQjqZ?hQm{t9Y<4{0;7UqMd~mrn3ev2-fF|8Q$W0`k8|`;ddULt{;GJ>
z&Bm(x2_PIqvl<8T?Xh^N<6ND+)sEs!^tjYBnzBY!)*Bd@Y-Glmtq3Z=gG<qPiI)*l
z50iV4INKPYvk?;qCwyk8bwRa~?2k$8kEMfcp*=-`5UU61A^V9zB}GFg0j~dhN4^R<
znxMyANzDEW&I>nG8cG-=V1kLb8&((t<nY~r-int==vDz(B$NK0QU!XnLe6_lQ%C97
z!V#LjEr`@iWWx~Z$&Xq<FPSbgS5(aKg51dJmZwIINrKOd%%YG`z@cw&c;mfiF8m0!
zTF<K>Pitf2^oE-@2yiySKItmF>MZc9$jT_qVwqQz&>~_n)noY6Lmgg|yaSJ2fn4@v
z8c)>=GxfO7S~zz@F1oJ}5Ie+CxK9S2uIwfhc!5+_*07<m0qf^OA5XtE;+PPjnUaj8
zjU*x<VSN)HjQ_lplvU8wA=XWcKgLSr;>*2Hg$}u)X$!)Z09z$nCr?%0cjQ;H!wStB
zT(rsr)M~<@qly}yR@6#}N<U!|c-85T%_=NKTpQaKeoHWMNTBkYp$09coAsLYrfgk?
z@*}c&R?M4cb#hMlgrQy9h>~v&$}J{1v@*nMx<!)Z{w!<wOXW1=xEKP8&RY|mifL<A
z`~bN;N@Z8E4=DaCXdk0o!CS;F8Hq}W-prJE@9{aKoICcC!`KCDCm|j9zH4eC7ixhs
zJ%Qx-5hb<35IQxlRXk8@&XX5Sf-IFh2!ZhqLXS?#{)QGg^-Ql5ExwT2d=?>qR>f$R
zfk6`~F55)}AqH~-lcr!p5NtfL7H|G}=+g55uNmbh2ZTW6r=2!o86(u6AMoe%3`r9l
znxQJ)n$^!<{N1ou-A)*R5XD95&o~(q#QP;(V#Caxg5o#<$KJC|ebB3Xb^1|0*V(=q
zi)b-8W-eF26?-GJJF0AeFpo=TesB0YxRuJdQLO}hmwT2IfmYq(p;d=_m8nCW%Ujdo
zffQEgC*YPNGHmD#V=piu0`V#&>%Yn+Y<+nggyU*mAb@_A`W7r!^1z9Gv(fhfh|Uh6
zuPDg6Y2IYbq0Eua<?}+g-o!Cn!a|rW#ox&q6jV)0?6PA7jzP|>jGLBtfPI-|I0fxW
zPgSO`IA(7r>!`51gs5xrkAyPf$<y7?{{atF1`d>7$Y47G(+a5xVg<WLv*o+e7^+w*
za+Yptl)?s4E^k)Fw(ONes$6I`6mpb<DHkf&K#%pN@f3By%K%;-**8(H&H!cdpp7(#
zUQhTanEG(tZ#5nmVv0G9Idxg;MN`;h&&-(zNt)2l9d@>X>A+eEv`-%cuan)B%RC7O
zPz9ab=TbNiL%3zL+Z=0)Ug_6j3M!gzzf2J(;j<v`<3x)C<!LA7Ez=qHjvF2a%v%pa
znt3Wu)-%<{PPL93{4&+VsL@a9oiu-Q(c9MSEB^awRCVx*?YlPsxqySz?AT;J@sqG{
zk^wB%7=9jgY4hHhO(G_-o31O7a`GVqpu~Txf~+An)wsE?>u=wklw5Mwu4<W?5*gp;
ztvF)POw|rol1!v78oM8S+KmWsLe^xG>~f9AE+gvBV7X>snz-(V+hQz+-|2^8EfUfX
z0_>ldK*vKM)y2K|o(>|84-r4GrY9WO5>Id28uc_JHuBgE`^EhX3biVd$PKyxI!r0n
z1xj^O{ydzi08-p@D`FN({2cQ$IUanh7p+Li3mzpDk$WS}weA<xB5Om<HJ?5~NYG4m
z2;K+ZD8OtD8~9;jF6qHrF96kd$$A877_{y-Y!IDhj$|hGoyQS8OrS+IO(5nwsyS1D
zRr8e<C1WY{CO4RTyg)S%RtXbOE0V)6N+5e)agm23d(N~F8<oCUFNWxwOdjRn2kyY}
ziW{$({ppP<aVn^kwpfba^43|ifCGS^D0OO9{2<S>FZ;|<{>?Tx=Umhu_NQ6-@{|B^
zboN6yjB7Xa6-U9FxDd@CjpbMi-FPVl8&odycp+_hp0OZ9t^WQj9ZMR}r)spOdhmF=
z6Fh7Ka~+?UQ|Y@(5qxqvMOvA2LE=H&th#3i;zim~iD40@tly6{3QjZTsi(4%6p!BX
zQB8OcWql$(GB=1fR2oTU8W*+L;2PITsM9KISpVT(kRbrPBB?80E-LRPICXQNKt{3Z
z#ztWMu*yrzubsksia-JEwiXmtt(wpn7^YMH<yK3iyCBHxK6PD$o?hzOf0{%wNgSk>
zKUL|6bw^c%R=DR>oDgnDB~a%IdYiW4;2pk4VVn_H>LiUf%7n80Ff?%>%kPioepA>P
z2K(w~4u3m&l@8_tl_dQ@5mG*fI`fYoEV_DZtrnI$#cRvVZ(sq#d8`n_;Q<SF^eO5+
zU!VR8UURop#$UG)vBd1%fk4j{>v%VkHfP&4x$#2j#eE919tN&ge)RdA!F~JlcGOg-
zLUnT)O6K|@VPRGuMcq30vl#VyqMyTRl!Wvz?#Xy(O6p&T`ipRhwDd<p^CX~$oe)Fy
zJ8+l+fB_>dYZWjTf~A9e2lGerT=IN&-=uEtX~zXiD?y3<5h2NeRysLp3R5k1MR1RH
zt(`Bp`RpLx0F}#&7U2m(27%h;@cPlbR*P18U<Y)P;LIt_b$!_ii2)sHjF)*Q53C;H
z8_r36=_v3J{paxzPpK$ZAK&gk?8dlbbcU>-%1V2{;RN`lIU)@9){8!$!C9~_nL2sV
zCJ#zV5)qPA(4q3BC`vWKqd%0TBW0}i+)-)wjbe&ghsFQRx1Sa<;mf|DQd_HEn|jDj
z_QudfpF~TYqwemsje0`u`qXJtN9C!5t|$+8P-*2|kFp@WXi$k)02Y1)MSjf?NDY$a
zaa;7!{0hkbG>^57Jp5#@N?7Lry==X7$XzuP%?FnRm$LNL#|ejBotc}rJh^zaxvO%V
z?<BwYJNav_qg$DvipSM;c}(Q07jTroDWMs3D2-%8O49V)BaQr9V4Kf}msG{i{P3!@
z$fz*P92uK>lc0;M+pTSjI;TfM$~vVtXDGv32??awX{G`<iVqm|ROE&Gd6dkc;YRef
z%#$;Ck*E6`nO<YD$*aEp)yqX7%N>TCgD?`i1P2`L#{^WJUW<ISjxNscRLwGj;zIvO
z{~XHKrjOE3acH(BF+{ez@uB<4s81jhHN9-TKbBtF8W0$L_^p2*^yX_0&(JN|FXyVY
zo%jl)7-2-;j_Xc_kY{3M;bQ52TYArGMc1xp%=%5hRK1)hId#$B9b`k0uXd8Zl;q$J
zI4_>)=+njTs+$ZSwAQ^O$83vfp22AZ+#Dc%pY`!UwNFqSTO<Sd&x_9hf^P!WNT3m3
zYy#G3;CC1j>imSWz=X=<ALX`%ve>#ZfaAW5U}u!BY;qj~`gx3|QG!G5NEp@_)q$5j
z9&!SU%I}N`K_ugZTj%dEH}Q&NiBci~isD~dOb=vn>_V-KFP)%qf`hvdcpepV*ary~
znBv?vh29Q|0kx?<;9je-X7?r<R0IxdrD~tI>hnMuU4v#OG-arzjLF@=#Qs+Vkj{s_
zjI+SB_m65?!Nyo|%y-k^vtBHi_X*UGR0I7!EP+5ROx?Gfz-tEcQvQ{>lrh6SjwRZG
zrM#{m2>%<(2=vl#&0V`b^E`z6R9UC6Ez6kE&;srjl=Xf>F{VM026J6Clq|5PC6ncT
zfaLK@+DH3w6WkG$NSb=C2(i2i6!Y5nBv6E{<T@xR%afc6M{342V5px@hnm!KsUq>%
z%x7@W!4<=e{AVmGXO}H`Id~=f(+E<}qa2&bnP!un3ExU$4F!5uIc7{-MIw%e@+Az9
z-kUXTf1HPqLTIB8ZAH(fP4*)z)zfr$A|&{BRG?Ejv}DBwVJ<9faC8I|eEG|Wmg9Nf
zXHW7!7pU7VXbSPW-DL8u3u7%Zk?(nu_RZ_(*N!@R43~Ghp+}%sWe>X|)OBJotb9+@
zxhj(2*6tg1%>6guI8ndME<woi&Hkl(QD+flwDF?Ce=8u&$!2tU>cYC%6Wnd<Juikp
zu?@Jow`tQ&ywvui!;1wPyQ;s}eV)^5L!hZ{&N2^cWP1P{e>kR+KYD*m{bFr8)I#@I
zSBw&jWGCF{H5lDb6u&9Rn`jt}^zRI>jrdJcAH_Q|2S+0Q5W-1u!YcgA?Vs$#M8xiC
zg0VOnvdZpl^mv|dc6Z0IzASu*r01mWvho!3*v(@QB{pQl6kQE0;OXbirsF9|bH3K(
zp!n{Q#%E0+c)Ax`s7McLvgVPB@{fs>d}Z<G8Fat|jte(#M_(Z1ccv5t{70i07mkTj
zY4<(`yLgLoefj?I5CmUT8#UsGF2+Lwau_%iFoc%N+#5S#CikCGRF7vOV;}Jk_xoPz
z9V6aY7tksIfGqNpqNbZ7yi?Z_n4F~%;>|C(^0{D6Ue8Qk`@P8Mt-_|{PX1tM$0I7M
z{?*~PJ^kgLac1Qty4@ECsZw8^{Pwy7vw>%_ZQ*i&f0eJVhL~^fp`@BtZ0@sTTK~yd
z#4UV_xZf&j%rAYlI~HgkkrcA;rq`pIQf<0bJ?}dXtaVo3T50?vR0xi9G-pSu$ZJ?w
zSJQN3XmOk^g+DuCZF?yxRhbr~bhG6E6lFWxJ^80FO?RS4F#DdLOt3}7g~g?*Nnc>j
z#F3L9lyZ@O9d&b7yf#|0vY>rc0vXL!>@1tHRe(`|(LKewILjJn;?sy(*AA~Hh$aXH
z{BM8){=uSBGt4~Jgbt$Hjs&F5f{dDsnp*oI=w66hN0|RVC2cu_mal@G%jzo2REPB6
znt`m1ah*Z_?>R=R6+2_6kuIP*yYA4~PfJ-yd|K97*0Sn%G&44{jB@hIs9v=Dhc_z4
zKZU-()<$$j6$vaDW~eL}EEvFl9`bV+u;9<83M8S#q4+GJ_UH~C6Ie{)@LjVKF6PYt
zw7r-*=RNJ*sKt#WO(adQVhh<xXsHX8eitoHUfK^x>i@5DQUXmHy{YGsMI9*}NTZ4W
zTfPR;&9hdQlK(_@Yio_8Udr45@}aDF-Uxmd`Tokeeg&j6x|?z*wGo*_`8xV^FThh;
zNkRS`+Mh`G{b8BP{q~JyN_Kmga1O5v9ZW+UPul0~+JMCPo`kUaAb4Nwe{0!z(?mnS
z3oi=xSEY8~LGb*F9gg{`zr%*NIcziMwerR8kGloN?BK2z#Kib2nDC|s*e}eDU#5H;
zfciy&YRU2ePlpXiGH3Lkl=_q&-*_$F8Eos8+&?tsCFLr@ktb)R&=PjMRI(<8%Kubz
znz`!vT;e3Jsq}5t_XG$JzeyT=80|(kC8rp}Phl<L4|@@Tq3|*ZQy(HEzgUac0M|W3
z3(nk4V#s}baoh)E_2NGfh9<2%FWtWT0hEW|g(v;bJ#Yxc4<WCdT=`-<Tr_yO60g8o
z?F=h1#{$LaC`Y6ipi71+4elTnG}z~PdAg1o3<-4=iwK)#RWsDB34@WDdbUbb#s~$w
znIrlW4|#bVPxMu1{0;eKt(w$s!<4L5_PC1~6i1V%GQF_UmOwMaU*29?Q>qE80X$i1
zh9l3j_c{rZTg0D|yaKdYk)_7%Am>A=lO~x0IwVI+!<8&ehJ5#>m8FQhTCOr*At#WP
zC=wFTa!}qKCA#cX9%HfmR_#FJGhy1@eGGqQcZwV;K1>DVz~(sRLXohchy^61++4QO
zaU^5OsoC^`{($?ORh}gTX+rTj!24_G*NT-#vXO6Es6LAPk_`qW&H=JT{C{o}WP91C
zmCIIKt6*-_-(K150oCbol>a6qvzhw^_R>GevmhoW^RKw=wJI=wI~%(wj^T)8s8ZQ_
z?Z@=dlQKnxR?Ag&*|E8(A{LVE`K4nmicrX5DMqpeaj9<*HNpQ={6>!hC}f)Qhwr)?
z{IJA64O8T!@+pI_Ps`+>Xv;e2HPl(QqV#lXn8fjoH|SKf1Q+Uhp30(!j2SA*Z!oTP
zW2jad{w;VUw-YQc&LGsZ)kod=S1#JAjf`|@q>~|HJa%Fg1La?`R<f$i(%41(J0=g=
zua?(BN13&!%|q>@BHT1CKr+zwzpHkvOyI^AO{EAZ8#<E9T7joXl}Ds)w^Ld{;algt
zisW8fF`MQm8OK&a)f|cAN3b2f&EF>v>8Hzo*)~P>yl}6KjwZ_)W!g0CFmc>VuJmz9
z6cL);Og5W1wW*|2z0xHy>rFb`pfEXXVU&r>a2u+`R|a%@7{)*xa+5m5(*Es+NJJTx
z!##GccV@2)Hx^PlA8Yuj(STJtKTCS|)tk@1p_iBRNCd~kcec7kE#u`trVHW9O&U=c
zx3Vymzt1DYfxpwRlxeH6acWa69B9;e12BH<`TP(m4ASS$OW-Gfs~3?AHXbGkny}}{
zRdO3<mv|cv;n4tCW`aaam_;)0NUnQHzoR?NyYEg<kSa%ZXH#0w@BSHeq3Y&qzE)H$
z`RqmSd(*$0MtF(TlBkyuK>R^^)QRVq4554aUc}gRv#DeG6DDxpr9m?t0uqB$^7Z{*
z2Rw=v4F7<B2zzqYO?~e|a!tQDiZ0y&14}T?LMs6AocJg3ha@9_{k+heBkVyrmD^nK
zW)kD2gL_9M+31#_l8pFe^ra|Kc#b&97yH}vwA6-{O?hZI6m`8yStgM&1yqeXXYVbN
z*+FC%M>;bpj#IMv&_}myGxplk&Z;ZO1^1%MyNO$;bNhjGy$22c;T%0f80cN(rZ}(N
zLF}=M#nBJI(99(I&-jWmu2Rkf;W&#lsl3~6{7dlf{V}|E(|MOrPU47Mw-$-B=smO&
z%P{BR1u1MV%okg3=b)k=|6UH&?i;5GkmqNL2a1%_jVOX1<B$VakyUMNpj8a2e;D49
z=u9qXZtT3@R<-0SnNc5bJm(|~q=ifGNBZF@eOrkJ)W+!_v#sl$WC>bLlu>ygelMC_
z6yUW`8CS>Cz%>!Au&|jiwPao51}b$Tow2lvlAr<J<%vHgItGsg-M+bdbX4V0CVeY}
zZ4Y0VgsK=W{-bHzT*kbEI?dK*_|KAdAc_qzxt-11qdb;_p%_xmFtUpf5e?R0w~DD>
z$wr+73_#<+*)sEaTU9=0kEu;B|MD*4MH2BBw3#y*M5XGZ!oY%saFP9b!nd6w>>6cy
z*DXPqU`21P2l*Td70F5&9M~~OP{#i_8t?<b(p=GF{hcECK!{YK<eHoF>3#Whe%zYg
z=usb%<vHcZ81=JI$xOvKt$aShbaLq*H!5odaA=3c)Ss3U5=vgRkE-ZAzfNLVLzv}!
zZiex<$%kr!gdBaZ=O%WTZr~p2Xva#ATt1;nv@N@rYgfBLgg|U&Lltm;%v+7>%$?Gz
zTH^FVDQ#^Y-HzoFfLP7P8^wDXYCBnRCy1xB=8A=BEO^RDk{??Xom9ecjG*fo;w#Pt
zzU~LR>61UZo-3a6Tw%+Zm+#f3JuK=&fYarlRq+lntwIfzu7*4<VZ3JRDChEx*!J`V
z5^V8SO9VP8kdl*wO!;)Szf(MO#?;kZqr<|fUqRlRt$!{|a^BWjJKM>-sT73e#cFWr
zoS(4?PG$`lM|CL8GI%cX!R*b)E}pFcvFi8fs;xj?_N<;EREHl~J$Vj(0pD6KTV4i!
zgBK@^A!iQ6k=5?*Qhzl4-`n>d1n|G9AV65Jj9;K2?NXJmA8SE1LbFwiQEfMO98284
zJ$)3*sPR*X-oHFM)49U@7%qI~FwnTBi?6^=wWPr3LzAwLF0S$`UHNkPjK}VNeQYZ5
zD{WjjTw;O@6OAsMx>a-ohggXV?ApidAn?i$#l~HTR$U3^7r%AU5_XuedsVf7D-Ya(
z9{lzSi+dl_*bSbqU$oQHT2ETxuYpjLX$(jPui2j;n5~v@suqF}yMJktkZfu$)G4!e
zYy3IrOgW=?isQ;mfY_)&?mt&Iso>{xy`;DJGzZeewc_e2M!po=2<RKRb-!445E9>`
zaP5MnJzPo}EC$0+<lNcFvpns+?1IC{xb#(cRG7kJhc~+}d3xm_=ftW~tO@c44Vpq%
z53_D!8tnA`bLxE*>9KMY*x)irzq$%m;G$+nd!j>PgS`P_Zq<E8c<I@9FQHj~N|AO6
zT1H4@kztQ5N((6>k@*yLrRL;O1xq>gC~8_5T%8`()Xz(S&V0DFXen4+7boknzBmjA
zgN<^{YO#xRH9V#4oE8N>OtOa+T4;wlpJ@*4h7{qN&z4PD6h~l*xnp)$lQ3ssA^vyw
zh8imH2lfr#v*b_U_FP3YqhOk%2BTPXSTQ&$brMocE~IQc3dVt#HM74Paj!N<zZ_nt
z*Le=ku7)<@TVLAeEdP3`ih8Euy|`4e*@*EZH{vEIYTh&;m)A#Py+XV!S;Uv??5#6q
zQkW+46fWDa{eszNhPfvOHxB)F+9dh&WYcxI6v)Jc8m3YeooMY(vtrDrUkqOyI^XqV
zPo~t+_2>@TJ9OC4Vpu(N&=O848DHN4V>`!>v2fGw3ct;8<=tjbobcE7`Hitz!%4C<
z<#a<Ag#>BQ8Qy<gQ?aZVBR;C^slJqrYQgn6R9}ZM^J{NSK?}D{mt|*d#v2i^vp)V?
z3BYeII`8RgoQbGt;kCIgwRZnTR(4=W;vLJf!bjebOB)Z;xJ7mK{1+tFbs=8+hXz%u
zzn80t$jsRQ9I1gVF2L^Q#X_E(%K-a$O%x*UUwz$(ECI92)~Bb3W`*i^&_T`I6_qvH
zZACb=CqFGyUxSoVX^k;3rGrw<q_U{T3E&n9{4nwl4u|{*)N~g<fcn#Wf0oR<>`yQ^
zv=M0)jm@!lTc}%MtN@sSwd95xaqKH$hT)1UjDOgx+8x{+*Q^M2CD|VTo}M8$!P$Hu
z%a3JYJ(DbVjhU}B>hus&Bk~FcGOKoI%;+#<*LtqY@rV~8*Z!q)88nLvHBs6bUZ6)g
z@xWt`G#D)xf!=vokEzP_j1+m*$3KO6ny*P<mRGD!N5F6HyiI+RsQ|v-W;x(ZChHcD
z8GBDlVY8)tM`#%HN`Q$6Ej&NOC%&o8^*X6T4YQ&=rnGwo({w)W!r`C}!}3>V2%DQd
zVQfcB0eEx-@>*_hc}l}rvS$k;8{mUGU#3wzf{Z=bSO16PqckF~svZjmVnw`iu4FJ@
zF*fW0+kh}{o_{{xxgG1AVg#2btDM$Q3mvlO!%gy#o~SBrN?%KmECs<8KPi>G^?|eu
zX-wV}D$s^O)drzfj=rh-p^bX!tVSpJp^dBuzIs`%R;SsuMXwX#%o*BT4WM)Q@3Hoz
zN!AAQ%=rp*g%Gh~nf!VI3A(6X+webMJ3fqaatyJys9g-wSjXPiy~XHe&8_0S{8_P7
znLn@HY>?d$R$OSp7S!@Y-tjFQ7B3QU$df?b7CCdWKQW@o^V6+Rr9`5k(d_EPg>>w%
z**WqU4@J<dfBVyF9uTm7ssF+8>Bcxd-E_5gQkxrSg(XDSaOBd!71wV?`o21ZRv+nP
zgi8&jN;1Lz$f}ZR(B!)cVS;PU^yRy)Y}pqaZW)GOBX3<i905|lvW@F)l#G$THCU|G
zPM{c>-q@DwZ~dm!NzR?MyypATKN8OQoKNny7Y_+}&GkKL8-Ph5jj#QbRu6a`SBz5V
z-_Cg#?J%A)h(!D(0}o0<|58Y&)|$Kb{Ean<iXa$vz>P8CanIy)h~i|}m!YoBGl)WK
z$?b@ML`eq=Hu)b3hKK(mn|~$y3qw|h<&pRxvFjzywP>&587HyR=PqVN^uJ5nGH4ML
zTCLMBz(d84UH5{E3bHQboT!-!d@7CMG@lJu7tdKE<{dbM2csWfDK@*fI(i7nSUMXt
zsE}Y4H(QYqkm2~~O&dTR+mZ&XuHWYb?*yoY#L`66LA?Tt{v2U<-5m244!l~MS?cms
z>e3|Nj36IibVgg75}ufdNGy)#KTd4WCsfq|z0~>FxX0OYCGe0TQF7z6vY>zGe)x9}
z>tx`)*rW)!4UAtEMNyU9oJ+6{r0=rx{IuDumlj8>YS%Sm7M3E`efUIBsVm84`E-pH
z#g11bl$W{qIoboiaJL`TZEhJ2!8=pWeoc^eHhrFdjVZaL7p;O9$-+i&AfNHV^EDg*
zNa1`A`gU8R=?mZxl4%J-xF)|aScje$kO_%%(xcsf7jpM!5wB}vKEg>MZJ?3w3w0hp
zew5B{;(?_%o?{Y#Olp#eku{gHFpejkU=k1`nTz<#xe!a_C(Y2=<<d^VHaV`^hv@qh
zgoe~)j<<h%92Q}~(Q#Y~9G||mcWk!@yxVt`_C+CI!wOPuk=GXt%Hiv2JKqevEGjlD
zktmlg=Pb1ilxI8;8Rd+>Te_;Zi|{V(5F1`xk~EhlSF=%%Sf(PpiUKN`Ur$4pfWX1#
zZxMlm(c;>3{Ie(5YrRp3v~7fpuxd_QY(@@TZJ>SZeQ|~?SQ1*K=#-sYsN=whqr}m&
zl<Os#nwmwUM?>zkjz!<l^$>0dpQl#O!WZ;L-ShDu+);L5h%kVs(yo0dGc&h~c7am5
zRm)OB`g!CA+TN?~18*jvomoXjyW~)aHh2djzm0DXy<D|$le5DfmZF0rB3-F0qYe3;
zVUNAbxrqBN`$*oAPL0OxC$PUc$1!<DYT@1OTyk*D)7a}?q~&w<<2WSKl;BdfQKu^L
zY*A?t*HoXrcQN4S-)?8{z|B*%{?HquB|@WU$zxqR2CE|0MdF|$mgXSdb^E8iQ0xJd
zC{Ui)jc|jzjFk>koB$ky;J<Ur|N8Tx;O`-ZaeL&D@{(*z4=SZ|wSefu01Uk5kl_p6
zvd<nGRy>nY_B4T{d05+-!vdxW+az-Yk*@g9p^^(Wt4N5T0MPJ06a5c|2?>>RlAz3P
z*Ed_^f%5Uy^0QUio&?~oMV%`@=%0}${uEuI@Iiq}iEs?0ZH@0?&w+6Mj8y~d_qLl<
zv#vv=hd@IWdNfP)4}k9!4dUSb{(fI20<PPVbDyoG<>_=c+E>R23Qj2Key7;;s7Iw+
z11uzx)<@<g6WQBnU=Z*sa>=fy<Nh<Z&@{LOk$bYbw~=Uxij@%Gq@R65?BS4cvwBlo
z$zO}ycz2kB{nnXl9$C^4G!jI2BIBM67YgToxuKN2WV5H|0A3aP(J%3i?i-3apHEw_
zzmB@f#PFqClrFDT4do8piJufdle@WbrwPKlJ)WAer?=S9je>Na72MR(y(+p|UE@*U
zzWzk?3Jrs}?N`s;Qk_gUzF!rKU<FKIc8k5WU88wx<2L`$4(|R|9qK&P!g26(u(3rQ
z$$zZze$A!h7jRkgQS3BFg(fePoSnWXeCJrG@Qa^Ap7vq-!B45$Zjq5aeDbrexb^a<
zl>%*9^H5r}eQg~}3aMP9>^dH|`yw8GdOz{gzYn|HUOEikO_ZhcHtQR~kIz9;f_9~z
z`n9CBy*)=dojdhW#q8~a7}zsyfh<pvh>0`39$WTeJYeqgW)aWM?0A5AP_F~768SA#
zfXE$rMXqz$K6tqPz~ZFfeD&5mSG^1hl=sj3^GaPNtw|JA$F~1uwxk!I_n|{e^S|aE
z<Nkmn6hM>%g8{Go*T7vrswd+Mu-nPypYplcD4x6n_Bk11`|1U~$rmIg6}9erbkxUS
z?4~E#3Yb9ubuef)11l6x?WYio7p{sqFuC*sN!AGbV3lGB3&CP1m2k37Al>S9nUV^M
z8cuqN^DTdWOX2?cX7Kd!k6XKhqAQuGm!+FS;YxiW*)CP!6cR#9AXfb7ES=aTu~oB_
zrgGlQ?x!6Eqxc5sump}V6ty!&gBcC8ZwV-BI)Gq50pAXGI%rUjrYz8|9^qbiR?i#F
z|7*&{bqhSPez`vp!I3nm<B>vB$Ht5=3nf7|RX-1Di~vm5AbRjLG@lT4uYj~E@%K-n
z?F;QqK()=a*)s<P>+w&AbU<L|5H>c`@A?%z8ysz&x$MSkIm{c~5Ha{aTl(fZ&^Njo
zlYrd{>E~X9EkCQ$iZi)MuBur=9^dH?)r^D4XkN`vXyF#t+lhJ#vp7DoLXDw;-Z>ST
zq|cC|?KMfq4J@-R5!b8@b<pJbZ6#<5M9K*YQpxC-W~1DfCKF!P@f|RzY)La%C%pVb
zN;!)denPl-L>NL?T#BMrt*RL_6*HA8ML@{;GlD5HJ>W<AO?<_xFm7nmy#!ZWExydx
zQXBQU8)9d8!*HnT?UGr-oj;`brDt$tg+NVj&AiH;(a+9=4)i+ihpu&bG-KEJ4o*|8
z%)hfXUcwhB^Va11i*%>y^`DV!HcGdIfGb(h4r-6zi6cAFf=5O~%1%r4L1-N5DKMY`
z2o!^UJi*@rlCd=clxgeFTfUn_2tXf*{S$o1G&U=D9s=O>5PClu%AB)=HU+XRmCEIm
zB1DkdOZRbJ4GAF|;VyJUhz!Br>@K<jtD3o7I}*#EJkBmLF)M!UDjOBXg>xnwvEoA7
zm`W6Oa4I#q>C!m>{zXJGTr=PUD41;Wu!-U#1Pqee^lyFH#ae_P#h}Hkq4rAIC4<-z
zpZJU|{*6{^cBk=Z7wbj*_E;cHzznrQsVf33U83JzhV_(PQ2$*gb}fju2+26)DUbI)
zhnYxkKnzNsgmzonC(KkwbP#m*PIHy4l`SEeK8Q6Rv$VS4*?&mVO9J2+RBifEnDAH&
z=k!aLk(jSmhGBU?#Ha2haqR0#EsJP3D|w4{-TeCh;z{W(P87dqA@nDM+hwlFM`MEy
z<@?(5db9+{s(K-*_y$ADSnf`ts0JyOy2Noa&M;u6u>|gDArcVoSar>%*Xkl+p$kj9
zZmZR~1n>jAB)kifM0Q?AbHN&JGhJ8gSKJ(%zvnJDy~rnDc@B2k99>;2dG86{IG9~H
z+0dnQo#9=aOMc3}ckSqjokX~YV;`+(;P<~5K;tc}`a!36n%yOeR})x1Rl6k!zck)F
ztUs`~{>c%~$XZI9v|z)j-HKqkVO)9JaiXA>>(B;Zs_|O>f_hjGeZj%=myY3}rZVR)
z+KPDzw8anOB22NHlJmHPTm|Z=NAeO{igBmnnJHR@@UNwFzJD>J2)v~TJr2BT6?l=r
z3fa5DzuY&VnA~OBOz^=vn%sOcV_DYMood{<zXhT@ABj%-J&i+S@M)rJ;b1c+nRFt6
z^pXO%gY{#Kd63TuU7So{;z4^-x>jzLLO<bP7{l#&WMQBOroY+wct0xX9P4VSVw{Q9
z8whIk@>YIK1{6Nv()d0&9PB&f3T$P;cIvg<8o1s=VuIYrB!hx0%!~D87@842HxxGD
zJ}tZMH?6j)`IvT|X&(^&{UKEEXk`^nTVo7RTJ|fwI4X_xvb;Gax4Tvz8@yjKow{Dy
zgg-1Lgpuf|gCbNM>GL6;xvNjd|7NH-lA>4nVSQ7f%hlsSuyps>mUUzu!Vd;kn{^}{
z{lM#9+y9@B_qVbB&#yw4?;ewINYPkGao~;F?M%3nbPT*P&gyMh5Gtm}=m7F1(*p^l
zvwBa)uTw2Qt+F>pn-IbG5cNFg*<qB*jurh-44DQ8JGdmXTG3Y!a?K5vP7kY;{kg-A
z5;?qFJ<CG_m_|Nuy-<hMwXf^-i>=Kx*OW;JR@s*}O}$kJn`j}|`kqoFn#v+*FVk`k
zW#By*NN^T^YY_Y(Rq=BS=w<d)%O2?W9_a7&aYw}F<i9mrp`24<xT>z=-Ymyg%~Pmq
zO3-AZi5(MR-n@yGA}8wPKDgh3Jj>t~)rO=x__3R~{4P?_^;cOEa!vY_crrhj>?6l}
zJww-Rt%UtIN}wL`ZdNrxD`lrrP{_Z~9;x|y5nIR#0e7%AiBJCzfKtU$tC*~MoEih6
zuSwSsA~e?^o0dz-s4*cB3aj&?QWC3kl%7G{wq4vdQn*vgS^~ARur_BuGMY_ukAQst
zG)PF|yk@~bU%gN@yrl3O{B<G;qV&$>QFQ4=jd0TcH#O~q*^U*U)VdO7nFtu+`F}&r
zyLiwM=GMbqZXTPRwe3d#d?uA(S<RoQQieT-Ryu;l8s7K*`>jz=3ifM8EsTT&F4F~b
zPR>HL*|!<kL>$@5P%Zf6#V71SI4_)@B)k7ZWlNi_RwvO%GkB3;ygKG-SVc{{szJgX
z8Vd8DLSoP{$(F>uegt>70a}0DGO?qllwEes+Pq!Z_rn^@`F@o<tcIOMEOC;_me^qp
z=T9Td$m5c6v9Jt=l}Ah`%5cT1FjexlAtL%XDpjjkin;ed({5$O_enpv=mx^1th`K{
z5PB(9a9C6l5q5kr7L#!jKV(Or|At&Ut(aLIs-Ds<jS9PB(g`&!1AXAXbLP%A?<I?6
zINcdkU>wOD#XZKhE_wJ0(bM$FS^vXlpKFF~I<_`tvRiDqrn_dQGn{P1RcEfR%Rj+f
zcfE=c^W`~x{yo`r>ZaLj2h-pzYh7=~<SjJNJz>cy344g*n)<i!+J4h+-*t~zW#5%n
z`{Sr=W<AOQjDr0ZU;-hcIXsiqF-QAJGba1HqZ9UWP<@C#*4&W5$Q%MW-$N8H{{NSi
z(3{xK_N@1&H$ngr4>k_=Ml}Cy$G$VEEn*uJx@PDi-F{#o?sx@ZR7<QWLz~VQQR$p&
zR)a|tuO~*vNaA9KFrNn)>0&P`+0;qy7%G3xv4f$hlZ0OhfedR)8#%^YG}_fBj_eYC
zb2_}TJ5&;G=yGd4+SR*pNZ&Pf%j^97gB+f$-r?2#KK#2wtFBsZP^QABFvl&FmQ3Kw
z3!M<F#|5)pqD6X6H8fAD@8BAi2APwe<wVNptW;)Dvz|KBf({O-v^b?8q8TbzXrWVq
z{aCqVS_@y5Dw%QvrEGP}orPJh?ognwX8X<)&v7mn;w*|$LA8bq&BVw?!BzwnIjYZy
zpo#ib%4kbuaG~(G+j%)o)`SI$m25M|KLs17M+Od4zcd~d!5Y6JA&1!>v=Q6IS$UmU
zJM{ikgFk*ZfdYn6jFUk{A0EPfA-%a<rFv&ZMnP$*-}+cQKEwe6TE6N1a9Wd99edtv
z6Q}xY@;}R@S*z&lL;J4lS3fb8@rhp`Z?*DR2s<sc*%)Rwu2Kd5*Ah)?dpZ7%Fwl#4
zha-B#-#MB#wY1|UJ?J2q2^*$@IGY*ZQ|Av)w=w|*UxAW5I90%IL9D5Kc@a3P#ym4?
z@a{EozkW1aRV(d}9ke(+>=e|?x!Z*)&#R--s&s+>i>!AF52O9Qzhk4ZZL_iM#%5#N
zn5407H)_z>wr$%+WBv2|zVF$4av#mn%yrGZ_gbI5Hg>&e``HE^_s{AfzRyfF%GH0}
zVyT*zJ8twm2D&bw9q2e+++%JQMX~Hu@s%U(s@xUDELgie<Hggiw;DXSz)D(h@@PXd
z(sURkz>jCc3tzW%J_XyPK_&?&8y~KE+FT#!2$`QK(W04PAwupl=-ZKYAyKNuI6`(<
zMoL-m%P|a$GKtkC;b+i30u1HvXLG}x85Y5W1>+f^8hn1+xO*xj9OkmSt2vK(o_YCE
z)8ST0c+@-Z-hkqD*=4Nn?ep6|0Mx_~$<h^B;A|BwO68}TEkLpRB7P2K`LFWQh=k$U
z3-Fd?#F!=C$gJcQoIY+GZpvJn)H}L@`;6c{kU9B#)bOMf%S8bg&;^LH-E?q<EUfJL
zuX1C*Rx&t87ReMT#wlR^b}Lmr>0e#J8b%S(&}d7`%}H8~_=jkiiq^hdiZo_6!q2=7
zygYoLhPm48ZNn|yMvq0HK3SF0mI9;le5(W<H0h_8&(7VY8DQ6{Q&hia+U4;HN2f7_
zGDp!ZQm56}qt%e0k=~9-?*PsXoi=R2%5)iM8$_}c#VdRqGA4xZkm|A`O)t@EZFPy3
zsl=QR99ta(+oxJ>oJ{>}&0M&D);Kl-+o5n(_T613aq@P=2y-_nC~V2MSo_dqjeVjN
zlDVjXH=G|6^ofd@iURZ;u|}Kt-=x(Xqo{u9)aMUcHY_Kd3|Gb-a^Kcw@m5vmoS2Bx
z^mq3iDilI8X)17uzkIvSlJ^*~_UU4`Gs-|JKK6>fOctr7n6-3F=b0G%{_MR6yy&-V
zD9)wC2+s*9xgRf_JyBk6mO&w1tZjb2Q$c0>|A>|sAWRnN*yZm4=&UB8ttRnG9)Fr6
zWnoKWetxoMo9k&hy)6&<8H^|qel41D5j9yT!zxeax~KmcPCh9@V*VUy_7?2yH-vpS
zZ2oa~Y>_$HhfT|f_K|>T91KYZ(8t3EW2A_7vq55{m=&ybw2)0?Ug`x}Q$1mTjs1E4
zT)&lRb?84GRbW&h^r>n6l~HHh>AbywVd)TCmXTn`g*=w!h~elFrA|MeQKWemfWmg;
zZhKFR=9wcPInH9@UfrrWVYbfD$agl@Y;CG<aG*vIBba7Sj<!d;=hl)3P%nV**D}1`
zdj6*Y=Yg&=%leW2&;MR%@bi<)F4434HA5$b3m-g7GwepFNJ~=S5;?nblIhX<O}7n8
z()Tx}8o`W%JnB|j@gnBz$?Rau%@ak0Q;NB%vL{iNcd06`v`ioWMc8KrmexPg<t<_A
zG}fwXG)H$%tuN;cSc&1s+O}JDM=h^VvNoYE-bbrhNoy0F86lYb<;~Fm>X?Ilqt%1&
zM|^q{D<s7vJ=#L82ge+FE;MXsj7<GeNZ$!_*VK{!R4rP%<lZ5Zssclpu3l2Yer4zN
zNxp^a-gTXC;hGW6TXc~7o9-SPUj`F<7u+14UYm22;be0iKsorrvo<c$xeV2p{nsI*
z9eg~sZBBUz)<{u85}i~qcP@l$bDDESKPi>hJb0G~Y>1N<(_^sJ&7|}&!ZV`h3TWDL
zVtlKe4Nmf+s4*Y$pXfcc%^}>>f)ds9^_fQg!Z_w|;FOVUc1tr79*+->kZXw}hE;zp
zftGWHFqMJ<%4cbOg(zFkDzuZxZ<P3J<H<LRcpHsl>PU316H>U%v;wVN(zl|wqVI|T
z2ZqEoMKAeiR0!lKoot`KHg6nc%Yy3G-b8}xj(9T7IrveTJJq7d!k-Ysh>#nakUz#D
zgO(lyF9PNUhIM6RSGh~+G?>OpW=j=*mglwK#EV7%ie_5;TS`s=%Yzt9;zj#sW#~>K
z(g5o(a0qdsene{1IM1Q;3H6OD9#N$>`kX9|1G5T9dqt)s_~_#ZEN4~97V<p8+^`8E
zA=4GPyYr`<@&sQW@pQj+Ry8Ar2S;o~khf@=qBl@bV}w12B5vd<U9V;F$L%y^w9LQB
z?Y~t4yalm0_4Nc~P2d*haAbhP_0`Bi8+ozMpC9mJJUkn^cG`h*<nGG%jQJm+0{OEM
z+(eFHO}>Tm^22iRd5>b3WltSvt%mSresLc9QdQ3#9L{-~*3so33d?qASnG*uViK9}
zvfHQC%{P02!XHOWW+eaeyS;?geQ!w2fkGb`c>)e;KuVm_?x3iwfL|pP<w&|Nyb>p>
zdy$+A8=AxNIg>?7JqZjU@#!nhitWenP8L4b@;1zaTT3~ffqgZ>shrJ?@9MWsJDHqg
zHZtEqghw+dAiM+=WgfeIHy%JK|1CR6qvMEJ{yPDzDW5h1v!hLXtN=WBILs;w*q;Dh
zLwOfz(^rPDpgUfoJ)r$fG?Zj6c6_!<ZLdVezNYm{4lZAW^rt(MB&z}q(w|wVxq(4r
zlcgGYzMqtw29L14*PsQhyv`*!Uzoi|oV)@1>jYoa8r1%8-PlN*IVbgKL+>FS#(9OY
zTlvMoKagLDU#yiyAN@bl;QMPofM_avDMGRopQu4ySNkB0O7TcxpV!x<&D?s~ScOy&
zG?+t(k6y%Sq1jitoF;+~9}6GJl;OdAQ^S5{@ae{}B2$6JqNtVco81a81;#h1@gPz~
zBKk!Eh?HZFhZ$QbEL{8L^TwlF8*clSW5~H+!|ML^xi#8_bcV!&tp;grfG0_nR5DjN
z4{3A8IG$JzMIiojYO~3MD#FJ5lg@S4*ZjoR7(^*?Jk@6m`oSc-{v>l~++D-{hdSg~
z&>ResA>q+Y7oY}-Eu@$Llb>gnoyd_FqSy6aspp+&Yn4Oo!AthvJ{Aq<opCI`Ty-Wp
z0X`Q4-7R`{Uu7MAhV)Iq`6NP>v<70cyzDV8@I+<hnx>pj$>8nfM0?;zDGt1|QRrS}
zg{>qFUI5z=Bq!!C!4ZFaPIg{C7kZ23rOSg<wxBjR8H01X6M7$6ObfvfWykO)fpG_E
zku{HAn!=GN4q?q;t{*d%d|5_ZeL)f<*pFFyTy6cDB;a?!Jsq(?geMVCeYUfSTwE}v
z!4AVIT1lWE^U3iA9XpgcYGZm3=8PqhBl~rv?#_|rBx(1QU9b853{L270aEUG@SP(P
zVxd7Is1w`k8t2UrvKz@uGa;##9i4I{Nm!0RYd&vXSUaukD(^S%KtBm)(>nJ+OXsW-
zZ33YU_HWZRhG|GZUUIjyY#<WPzv6{BPA^#CQ<UQ+az4R$1RU|U>R$n<i^)2VyF7hP
ze`;l7=l7}HPt3P6og<B<w1lZoF!Q@@-V=jcJ6TUyHz+#%DTxV0AM$@e>Qv=rRW?sj
zZ0!i&Exzl>8*%Rqv;Q3#Dmg^>=&P@^O|1U_!^}AI5U1P*4r$&Rxr*Wk+inZ1=n!5B
zF2B}X=n`nf@qmd+kel&<vLWXh3Vt+}%0f9~ugDwcpM5{(N^z@mP$buxLN(iLe6SkB
z9AsH#mtR%MTKSfgsRT(Rq^w(APcIY<)l8yB`&4{NJO=fhV3T-*GO<|t2bvRh?Ir{j
z)CVHZ(>sS105SpUJ`FX+ce%mN@V0bR%u+KlGumQlT}U9)8A>C|njDQ|8DZY3C@eQJ
zKh!(@eT^nDgelVx@MNH@YBVPJw}b?9?Sx*zA`+s6%9|Ei5%oI8iLd;P2!E09=zA8O
z^c2>c$)^`&k8aAF;BvV*u}5dFUY)J+lIVtEV><XBz$qG&HA&4(5paH*#;sp1KoPZ3
zPzcpN>`g0NVX^?BF*i?HD+9ZhT242W&WZJ<(_!6H#VyG%JuSM!TOQayAVhYz!~V6j
zG0||n7I!o*SASC9c+T^<D%#g)WIj#glq1FZwE0p!NX_<iTm$QOpVAnVB{(`ll(BOc
zYStbM2=ZLm0YcV)(sz+_{>j0mnXi&>m}SEAMMizX&VwDgFri`{$(}Ugu%KvMgGRoX
z{3fd<BBDscS&cDRP+lp2w{|@s*_b&hrjMEGxHk%nFb`iufc~T?tGzWkLy(gGwvsp3
z1#Knf9FCWSVNKjwhRdfyS~smt<lkCN&zcnsIMMl$H49II;|%ZfrFGBDx5K^E<yF>@
zOMsZ1&WE%37DZgfwC;v^CxT*RPk@^f4?BJz;Fht)M29?=MKukoe6L);3>z<BFCZh%
zVzn0LBp9$Bvo#%_4Sz;V9>^&k56R&RM4jm5>r%E-T*^GIf}HP|j*`&}#J4Dg{m?f9
zuGKp}1dKD9f*85s8mmWy3BHny?r%Ur-8IAkFsNS6k8t|LD9<1dL>n6jPMvw*nzgh3
z;?rJt8I2eGON3GX8(mw|Q-Z}`enq|bt4OF;blyS2?cm>Pg$teh{Jv&s1`SJUvK&^~
zw-LT~Pm(mh$@e@mCf{v&m*iFHe&kaF#tzRzIL&-GZ|`7P(Z{@uP#r37o$x##^{|~;
zh(9DgWv}9t_@2%Xr+Ghp?!|EoYJZNYV|~ps;?P~(tSs9#F*G*A87e_8$8c9YJAH0E
z+2M<SkS)2T3<>537E#{V-a{#dOG8Q;&hvICF2<Otc<pvph-r?rrA@qaY!ns)7xwP~
zPYoHM$EUN7Bb+eZcm!X0oUEW6t&*5AL*~PcI}Ej{HRm%kbivck-;IOfj&Ral1`Pa&
z{f+fC2ioAO>^37r&^DJRjx7BWh#jWKkiDvN%v2uA$YMk2zelVf-nb@lxg9Up@ATj@
z(%BZTc>Xhqk+~x9^mxW<nKj7<a`lRvCCf@A#dm%za%W{g+I^1Me4ezVgDjObxPcEN
zPFsPLh8xHr5_3?zk=E|y=+7YTyGa^Vtp{D1QXZgEO4;(TyT%m@Oa|r>gfn9y)mq8g
zZi2|7tZWCE%2?*;C^0sh=t9jS(3{uWA`)bRU_zZJ>eP*giZXK&U6UmMl^nw$Dq~pz
zdeYyd1x^w@9HzSs@|es9N#5CK$c<b@T0zR(skhWQJaY!dWC9A8+W#84FRvy<rZ+o$
zD!x;xqpZY@TX8oroQB7gG+BYtBHWdmX3Pa$^>>k2@bc>f);Mw0Y=i5wzB$FDsxBcw
zP-a}{?sheZ8r-K=DWp~bFn^vhkAC~WAl^^bD^N=EDwi)MK)>g>gUhpr>ZW)HrmR~%
z4wFqG&lHHBmd_6rWT#A9O0#|t-k(?;3E5q_>$n(QXBMjz@_Ydy*&htOSiJrf-AK*c
zt0(y9O&y#+<nGHM+22x`nc{8B3bGS%()$mAOrFk9o1W+F;uw0sCo>Ka*&bX&LH@o(
zem>fWXji<|IqP|6GbefgfVwE=mW1l${Qeim<+^I|BB#lW&G$!dK=&rX8OdA81gMeV
zeBII04$Jr@Pu1tE{d;*p6|M+0*mz3SQ&~$f4?Mi}(r2In!DMReS5DRQTGVS*hKvPe
z*LKl$HR_jr*Lnp&p=vRH@?TOBdxBtPEe!Z}ni#nPq1*irOMGt(|CXjz){j?x3Qu@A
zvL}f$oZTIvkHUP?-lUz-G(#=i0<VpK4r1PHDceHx%p+Nbh;D^EKh1yHdY84Iva46H
zy#e!E1ogo^mGmyj8=C>^8y9dg8kqaOj>JeGq5{!H#zp{IW8jNJ<uf1gC|xA~oa$Kp
zMCqLB#7Yi#wiWGdDjriKwMV<;fnHw!9Q5sHQEc-Os(8*q<GgdbpIGt!Wq=%H*}{sw
zN_3DtFJ`a$lz}Lay{=YQBLK#UY%mV157X=**NPOYVlDA8Fa<?OQ}JUnv6*fU{zX{w
z0kVS(-vwxwycgtevN>YwrOasWE|1BL_v39JLT|Re-x6WOqx7VI;dCK$F`DW^4gJx!
z5YHlzy`gWvypepc^I_K&Yk&wf8nBMa>!hQ#YrNGH{_NmpT%N{FWgkiniVGNZq&|KA
z^Ajoz>}q@IUzk-5WlH9J{Ly%oJN&L--M@B1=>?E}T@J#KH@ku$6`JGTZ|-4J_Vm>e
zC7{1zqM^gdq!?1=-SQJ@W#;Y_)HqT(S*2^-;se+HjfW`xz|JS1-un;hJuUx!%yhiT
z9GCy**wheghxH<|ME)SFri1%cO#&#>yJ8Vju_*oWbxAgnU}rr-e5Ke2b6YhSU*a>?
zw*V#;jGpT)5pjPs=&*X^afBjq*aThRT-x5fzw*jhLqQC{-;`uSdGrhYEoq3Na>K@`
zf3dpqaU=XNAa5b$tB|Ff536zsOk#9<YfRjjs2fbbDyKLhZOtWxlmEx{h;L@0BGDqQ
zKB2!J=gcx_J3;cCLTgDsXcQfFs~b?}O$;DBYMsIhEYN!wsW9#l8wHeUX-*R!nX;v=
z<C%`(nPQRsOe^z8MaEjrkJ>OK33!w`%9P*hYab@YQ@*1YgS(`pFORMcMM#q5$~-ff
zEO>=7+F}qwouM)ktmpDB-795q^Vq*po%|^f7*ep(Xb=xGGTe<*%kJ7|lwzbKzy%nV
zS{LC#ElQ2j&J*>z<%$0miEn2=0(Iv&{`CD+;6q^YN8pGDX%qU%U85`>oQAB8qu7+R
zNbBEdRlYnRZzH2j0Won;u?$(}!bQ+ETD7-UhI4b?NW`Utx&do^=9zwlfkn*QH0zXN
z8@1aKe}z%ks4-qCwV?43;o7M&jtG!#(dk#!V237OVX%RqZqXfXT7WLceLa^?S7z+9
z0u{H@vD;j>%#7_L`}Ex?v5EiKVBZK$l(eTQ=$*mHZDBQJ;vCVA&{^uX<GEI|*k^k5
zXF>GUL|r)ER;D=wbtflLUUDo<9<n6nxcBh#SA_uA3+mD*z*kF`g7*|v3{YQxMGN#K
zPg#Hk*5xPsTCZl7{XHFTkt)EkNZbRG8wG8?7nv9-Ys&W(F=vz`#Kd;PmYdsf@!dp`
z|4b7nm(2JVJ^M49z0#J*HPt(DZbfpm-*#d{^V%CG?Vq)`pAy??Sn3p17wN48^Ln|S
zE25t|4j^di<N>{vP)|29?*Q=_L;lpJO*+Ek-=9mNh_t8-Q_zNa&_bNB*HnEivXvq-
z#Uj+>@^~kOZT01%!}|tEUg^TQ8S3(A;m$OKArf6GD#}$QQuPDoqz?yvS1SZZu88N?
zV(l~zEx$0=IKHw5(ydryhaf8w?HCXmzB*il$Wf}|G1Ao8cX>2m9_Y!zikg!>_;t5j
zbls%^9m5)~fzxAznHHov!TU4GSao~d@~Le+Sb!w`Pp~|ilKV9w=<(w|ryTTn^~jRc
zO(3^v^JP9ZY8qav7(H@81p5a6<^fKJmz{d7@slz`1Va|}8_w`qkj45KmzlBkYY}^k
zKrJHLzkUt;e?NqWCIREb3{bZoy2A>m55rKw-Gx?;d$Ke`1oyKy?b$uiWs?KrN+;23
zB)@>V=c?{N7Wz*L@pvtmSo7<f?<;kjG^_9GnRKg<6KRrij6+@HPg`-Lsko2DGKj29
zS=-lxHV!vnrPPNtXMIRy+aS_pR35I`I~%|9BvFfgIr=Dj&H-=J&~zDTL@(Q+0<^xz
z#s^W9ymI5nzN(7+cLfAm7NCChfs%~&`MV4x70p}9JgCl5)famI#3U>l);2g~G1LyT
zniJg+1=q7IU@A20i(kh=Fu4ioszH}|Y$FVKi-Rp9@#Q)S#A6wr1U2(+u?5jt@omdW
zL5yKsk4}<*PXe)L6<GJR_)A793wy94X;+JMl};e$?I%eXSdNUl^qOoKD40VC;(yt8
zp3x!G5Bx70L^_xn9*~<O3BR4%5OV!n)?)o-ki{YG>L-Kmid0y!ev4r75G#Z7A`q}z
z=|P>a$JwJt<Frlv=KbU+aLT~h&Ixip;?5!FRPc|I1`<e%q<Se{ai}2ikdu{A+o*Re
zluL-}$MjC>m7MUa$=`hzb#s|#f%3W^2kCtJYvOjiqo8snh3J8}xjZH+hgL1CnSG$X
z^sJub^gNEEZ`*3*=0BT^p~P50UTKGC^>FZ<tW1{a`(NE_Edj1+F)0_lKr2qSfr}&g
z2ef^V2$0}68<tf%kvbV!_-q&f&KksM%B>RU-<psO#S6cf7Wl8UmTKX`v_1<OWdt%6
zR;k-L+h+=l?|93ra&!6OU@h+KvMsE{rd|Kl;pd8p?5$TYVLMRP6x*Yj!QhlIr%$i!
zM@Vw6Xd!i8-Z+vBZ(*vFN|7xNhwxoCYd=_keK>ApRp*c_g&hz+T?PShJaf<+hkJhV
zu+Z{zZoVS;2;B|JYljAhe_w~(P8n+gGDHA@RYIRycxXAFuCKm&Eywr(;}ub*({SXr
z4HMqLcdCqwyFRZva+e~ds%G<84NqUK>l&R?+O{L*lxl**U%Dj1A+f^3`uY&2Xs5uF
zxTXwGU9In1DI6EY*$(^YC~rEgiv;)s`*LHp7v^_tnwtnmzL=bQj1Fu|x1=iieJmee
zkRVNqKNc%XLYzbvrmRY)-`Gy0Pq+@F7Ad{y8gAoxFCG~G;S6QXA+z_4p!1Qmfle0~
z3qK-~ZwDg%^`4YeGmE$DMpq9I#_b0BxwMcoqL+up5B9OjeKt{3NG=FR8UE}Oqq~mc
zj`4v%p$mK@G4WnIP#C!7t?jqNdtitQBsh95g6Qs#G2AkREV(YXMl>@ikNC13w&R}i
z&-0arl;Id6OOX>uBa6&!ox1XvL&C^As;Fwj>~(XM-N6>?onOk^?p~J1Zz2QEXypTx
z$7C3c*S~ju7;!+;vin;e|0sOV29bjYnl~U)-Qn@u7`kji@{GWpS#Q3^cf?<u^5NL<
ztr;eFaTrYk3Qz|TY==m;j6{u!%o&zR82@}*2|jk{lcqQ{q}c#jbl{914EalRcpbr@
zVMx?iyQ8|m(CW0^f7^ati}MHQQzTcjv%X(#eBpByB8s6;dyG53ejJ<+(65;n79h<T
z_#g}@LnXGQtLe)(1GD117-ZBuCd5ewi8L2CsE?z6fp1UX-z7<I_oZjDIdYnSU-LBL
zg>R+(tF2JbCIY`_u<v?_aKl&S%%qUr_PW6`Fh9e3JH8R&giaOWnS=!>p-Wu^MN)fU
zK+5LFXdc~Th)aeqX;j1DRaF@VeR}wc_khy6`a-r<S482!$JY+k6@`~eE)IH2xosaF
zo$5!{bfNRPFyO1)y?pkn^z6<3DhDIXy?v^$R=s^nLZRs0o7>gYb<5vo#MjA0rU;BA
zO4-37d~q<bZ4WgG^)>)DC4z~(f5L7i!M6X`TO#Z+BW=-S!-PT;vH0Jd7E(6N$_D$H
z*&*3N!!~4-^Wp`>A(AHF-PJieutM=7G7o&j7o<zf`LoS~@P-;XaHuNBof$asBTA$5
zD92`hmttO`+Ur?)+TGt(#zFpQl1rnSUi&$)rER*KeUZy|cnY*E=l%ch(L84x-oy4o
z>o3&beUJK<s3qinL%GLTFZHR>3;d7C10EMx!RASD0=5|nrqcg)lxgE%9p$K(^-Bou
zcKyo^czSLjj(??HHhx}*Fv#X{Askls2*mO+p_U>vo5>nBinrFIx68kuH^Ba+gAs)z
zUQG-gwrJ4;f0%0weDw30IOD-(BYjhlUwL%|ndxTWK?=-rHwZ?21>(s!Aq9NwW6<Sg
z;yiLTMatveVaJ}sUwW85l$T_RSKo5)^zD|!G8fG<l$_}*+U>-~jD6$#yie}&at-SB
zyKZXR!NUDd9Y34&ZY06Sw}Cxn+{Nfy&4Kv?PYy0;J@m1PnTV2M5w<%*r7d<Z@+Mt>
z!bJQ)caEp*m`|s>?7B;BH!Oq`hE+wt6X_E}J?pB_IZ)3ekibQ&z}UmbxRPvdvE^>-
zfo#hOwG!q|{VT<ZwXcms4)W8l;C@e>y21g9Nn}TyI+;-<UdwUK13(#VT#A#vAD5M{
zod#>4Cx`Az1ohl=RJ`9Mw%}V}$MXjKx>b%y6k2SlksUq4RPw)-AX@DaK2<~|-)$a@
ztfZwo$MEg1Jm9jHR(cNii4vOa{~cYp@0-k<igW-^3tbAu+UaByu5bP5fO8y3$tuS*
zmhiTfwLSoZau~+?0dRMYu>AmD<aIqS-e=8K`n0Eo0I!lY+en7W9M8N-aMwjJ*UMeA
z$Gj@|@$_GbSK@XS{-ys0?9PbEf>R*H(QGs1&<?s|)~&FXlyr)umdxbgap^7!aiGgw
z{xVq#*TvmQ6XcXmrrU}h+0(mRX*f9WYMITq;Hg@9FznM;00%Qkx;2#3PJhN|M<|3+
znP$#X{;b&{>&=4cxBij31zot9afR>7uFQ0&skXa=Ka!UGxYM<Y2`2CHCcR?dEPi%|
z`kRY|RFD6rVCz&r_nVP)<cR1(*C4~HX~EKUO?genoE96Fw`vnO<1jQxHO^p+vi3TD
z%wXrOwswts0^A)GLyI#4{H-z;#n|!!Z44CxqE+Y8gwDbrWUI-D?y*Gj+YcE6n^aG_
zaYseUc|E>sCYw5S+TktPGM`JmFRykxF)6{i;fL);Se3>}{S&kNOn>drR(vGoRfSS@
zMy=TtsG9CQ%qAFJB)4}<_1@Zn*j=Mo_GWueP=ht21J;cVg7&n8mdxT|-SHn2O6@k<
zztkGr69uBS<31A62;4bxIHia*&o>XHb<a%VQC3WeaClPVKU;ApP#l=kHwz~xc-$bN
zE3=?SAcI`p%SccB!d6W1&N9$`uuj#;)2tcJ@I%<dx{Uw6aSXop*_qd)8#8TRGqQ@>
zigY`A0)7~bWzh$>Qhvclu)Xcw+Si_0#XN?QX-B0i?ZKnFa%gzSat`kxUQ`D92sLBl
zwk>}VavL@>9hmmCIP87Xj6&+oSAIq91m_y{vPDZflku4$n{UW?2;=XGB_bU>kwqf*
zr+_`|A?i3?6~K-eff_G`5=?fR-Mrl`V43=O0yi=#-$F!wMDUA@L>n9WYW{@P%Ov>a
zUC2``?si9ENN>hDTnh>%tCAmB<47f7!8QSk`5_{<DaqI)IbrvU&HC4|<Os#cQXShd
z*S?OG%3DW<V=O(i{4teN8Q0`QGrO#*Q0$LlfAZse*S0j0KF|zAw`)jCXhY%=I;CaH
zCxF@-Y6qTKrwNR|WzD*-9t?hs1}8GX+1T_Kb<fxw0|eEN15`=UNTkJ9j>Q5tb{fHm
zvxQR00$_yh%)^ZxoAT|}&WO`Lz^-PZLsD$*15L{;qq&HD*Sbhq*+w)r8677ZBV~{O
z6403cWHczCu&GK2YL@#(P`eIq%`bFN4HV<XNNjjOW=3V0i#axn{89}mf``@_8%Dqz
zrzMu|RiZftUjmP}@XV($sY*#oM%^ZEl!jo#Y5z`Lf@{{i81KQf?>D>wuQ6up^7u32
z46M9wga7jKpl`>nVcUyAO+A*0y7R)wo1FE>9s(}faoVC4QZzw~uAQlB0L}XZE?~AJ
zLrhou$R60=L8h~!&tqe6w~EV68uYnMz^d*}Up-;_D^aJfewuuTVMab0or6bps(nSG
zDn-!yE0o<CZ^G7v5qIWKv8P!Uu|{2@@BRHXi$-pJrvb=a@tA#2c%IjzP4g|g%Gv4t
zdt}e$mm@JR2O05N{<D~sf9h*VC*UP8?<nN~>#gL4=2H~Px7=XpJlC_EC!C!i*)Gzx
zcj2|6iKGbX@&`-KOw+?u{TQVbe)scZtG0A1v2^WUOAXj}`3T&?J(j$;j!x`eUcW)4
zSkCT=UQk-{i=6LzR200JjY(;`Hn{EQjcde+pOx0}xJ8<x-`avJrW?Z9Ab|EV&T4^k
z=1FLWN?rZzSk6098{b!s=lrM`P~$hD5d3UoS>gFXjS7xea$#A)=8i<>@o<<zIJE8N
zz8)lQvx$jX);L~~R0VH23c8)9wvy}(S=icyBYJ5@x{=m9vO1*sLuI_hZEAzk3H}lF
zR|RkCdV^)l2oM8{nm+pyZGbwCa98_Jvo^H@;awc7Y~cuH3%Ya`=o}*4aC*id>%~67
z!p(!U{<u+Ax{SrFhzHT|ZXAWzPwaHxI3Jv*>f}u0d`eHGR7DWmHP!D@3S->Gnc{wn
zkftIWB%t+fR&v}i1%ein1ou!j7^l>;X5p@$$scmUNPD|F5{R;mX@FQjpavAhDTy|v
zyjR5v$a}-|HJ;Sdei{aI@O)}0r<6emNcqvx<4^c62I8^z*XfseR#x>ZN5Gm_Li%Jq
z)?{zomAQbmm&<l)ty{WxiZY=d&Df#54z|(;pHQeo9l8)0TxUIcBB9LSU)oVgbUT7r
zMaM4gl5N$Zaw%+F2Eqxd@H<5Zw%WwHOw`^y1>3To18r{Jrxr$9SRo9&1G^pt=xA`C
z#6aA|jptFJIyT*!^YGhz=^ekqiFVM1ytWhn0%`+-ZoZnkAw>$<deP9SROWl|%shgx
zU{D&VqxUBx2M482zJ3>5^dMW;xBep6h&b5=n<V$<4C2)Bz&wrvmxn;buyq)z+|FNQ
zS-JeTuFce{63e%Ue0rL{3RW>NUD-}S?lj?K5q~K$KOpf0vtma3DVeBj<*{*pl7B+i
zsS@rfTBp|>vGz6`*3lrwOp)_mNecbF7)5N^%)>`xI|(Czi|%Kq%ZwYKD?iR7D<;;S
zJnPY=`tCms0?e*b;=cZxKfA|HsD&a1x70rP*}sQg+0@MR5bW0<CdLRACH17BHrCLg
zW*CZ+5FHj~s^!s08WMi5DzrLs@eXy&?jL2k=}Q-n-U6dU@VGN}+y|{?F`p{VqD^cN
zu|OQWFl&aR-2}qIitfjoOmc;q$=dJx78m;?*0N-}9l$ukt@PA*vTm3nGRLp9Q(uUl
zw+Nn(rY@Y}<O2*48*ymg%cpXG4CWe)qrUaqL-^(|5YlQz*Nrojic`HRP6QbDsjHy(
z^E|o!l@$EOD>JpxVH$YG`H0H|nczrcQtY_)<!&K18oRN*!QwaRD-FLlPXFv|=V3o4
zR5_332e!ocA_s~r2YrSxqf-m0y<EpR?|7&JrjhIQv4_6Vo*+R|9-BGJ^3{F2>Axc}
zK)r=S=DEH0Z@n!q-!(vJj@W6^7*aFI5>9Y#W!xsLh*CN6>Kw;CFm8}&bmX1ja9b(&
zPfcAg#A2@H)<F4_Vu!Z3E8h5nBA*saB1^{&6j)PyQ}WzmXjCR&agQuThfR`q*os!#
zif)-rb;yTrJGBgiH9|~n5Lg{$r;AbDiPP30A7u8h^Td0ETzmo3t09a2%>bV08*sh0
z!58Bl?#K2B*~}DI&eUD$wfgs$Ey+WWQCP2O_PH$LV@9OiaLdwo-qLt&NFSV+hD{<7
z9uOkQR&wrf+#7`7LE;i#u66;tq3&JZK5VT<rn~nSnYUI*HcM}ld7FN~=!LaL+(Ike
zG}2Z@n5)}7g3<`?I`Z<Z75;GAElx4eEpgnwbRw@{^Sj!gv+w`7LI(@ES^A8t!N!EJ
zgss90I7B0b)?;V9=z8$0O)_*kJ}xq!_5exn^ZC;%frf4d;S+jL9LpyWYi66KwNYni
z(Nvex4JC=GH&mOys9tSr+5;V`6pPo3p%jbNQcdQpuMQ(b_2yc%lC^StsM^M6o+fLm
z-;lcTXLVUa&zQwT9@)ohKF0T<1cn`%_Alt3T5`Lg|I_+h%~#r(2k>nwA<Ob00cDTG
zER?ct7?rs)OH!O<e0GF;0K;D}jTO!G0NFN!OMA5Fk6^-Ma>XQ1NA_<u>p%Y$*yZvp
zRR{H^Uz<TQG+8xq4b2q3t=!)wPb>KP%69%wh)8phdxn82Ijb?5YLhY&co?v&LgN)-
z^PvhZ8g{OpY)qjsB#C&>10W=!Cl1Y2f)%*>+8=!0c^>nw(TB#${y+21`UlU@>_isF
zrX@X_l#6wKO=o(qp9XegEzeb*yuyc_YLt2E9;5Q~1iH5s=>`@Jan{=RVpzLC!90j8
zU4JGjEG(+MDKR#^k79r4yeL4BTM!RW8cdzJr{TjRDp@)40*ubJd}2v6m+PJk*Z%YH
zF8kg0+8oiJV|DvxJO2q)Pg8aN-2^eoY?E#bJoz}`Imr9h@9Sy0THwu+=sH#u1|x~~
zW!{x7j*y?@L2O6I49jw*gWzqD!Jm%Bdp>pE`NstaHo4~S!4gp#H+`b}qj;F58A-%F
z@{anwb*DrC>KOm=5jIVYCTHRhzn2N3a@xopQw$?0^1kY&_?sPP*-YWHll%4V0CPl-
z>pX6Elb^T!!ARSWOmQ7J&vspObTfyzzBP0-Qv&YQm@NO?nlNWNR6*Qe{?>evtt=NV
zJ5VAzah8TB>-jjiao&==U#pUvr9M{e6McU!)nx&yXh@mD#$ef$W@4|94KnF>o8FML
zS&vTp9a%Aim;tJKnd0Gp*_YRUo3`&SEYMoUIJQU4An{x@ViN|A;xKSdwZZU-C+_6{
zLDy?m&K4_w4)Y|T@fkSJ7FOlzMv}g0)V?n4>ndCj!49M&j=w#noZYt@PxlHPF`rGo
zn+!mG0s^F(uJqyX^lCj+bP4XknO9*N_11!iJ)wtY-I8(IBEsnl8tl{%zn9;7>Ix4t
z)}#-Lz~Qq1Sf+c0DuXZssw|~ixGjAqZIQaSio9CsGL<RCxx2B8o0r!Ebg4-xdPyOA
z)#mdNo$?GCn|nu+4cZYVF~Mtt(&?DwpANv|(Px}j>4T7pb(77?8U&n!VYc7#t~z`6
zmp*jYI!_mv^0X(Y-`RWRd4JmAY<%|il137=-<h5moZwc#gZm;(+F8#N0z$IV)#*Rf
zw@uy-Wg&Y$2z?YzE<C!>5sY)SC~mlIgbud33bWfXW05W;yqAJoD^WxfyDq$iFj~NP
zJ&bg}#M4qJocr8YfzG7sPq3W5xo1+q)xn)@SyOpns>bjgnC7G84pG|nDP~_+5e!Eo
zw%>fb_q%Bq`y{GT!`H3J=9AzykH8z%MeIFnQZ;Iei^Xk1V6V8RJvExJjDzqKk(O<v
zShN^&jon6f(@+#O6Lydhe{Y{+6beu;l^@mGl(>01{pTWzOErff&3Nl^7tybY!6>Jj
z)<`W^I2tR9mP*Me&y+w(usX^8a25YQQYMjG;$I-FlY}^eyi`cG`qdM3LK%e}or;wm
zX3z<8Pp<@;!yQD$*}-v!^%wcfE(zJ<?*tlMnkxDsg!WIWe|ivKBeE5UIlaSRx_EK4
zOXuXtSO9JLRufIj5P{s4t4m_9fa=iD=rr_dkuaaQz_n~o7Z}LeeLd6u46RHc^UQnN
zqHC$pkc_Ub<CBu`THue|9$|moKkl4EHo9R05_eo9mj~XXq~Q67UNM3@WfXzjM1$}(
z&4hhnl3zWyxZ!SnXIT>%k1HyDidc)nHAu8@Y~~$2qt>PR`tq$qt>0wwQ})60$%w={
zO0Ng}E0wP&a=y)~ZDyWuVhL%m6cp4Z9e4yM6YsE`E2_4Hqj<5cYf$FoXg)CVHEPnN
zIg>8kfIA1e)Xpk#=F?9xFR32Z#O@&-Du=7m>;d8dNlx?_EzJOf_@$#iK(jo_?h<->
z4${VaWl)UZBFP}Ew2ENKVXF-efM;y5W42CSAH%xuyC*oX!72Ppb37!~iZLrW2N@R?
zaC3!c`%@V8ve_d{tpZjCL54GzgfyBktNm~fyAU3H+3foBvdF3Lj0y7bb2Ax<?BzIW
zFJbh4t(e_ui3u?9Z#f$()HZ^O)2y2E%ph}?VYCbhv_omZltYytABc|0n6~qJ%v4;O
zfM&n(0<{^F-M14f?dbOvKE3~moECJ|OauF<OtXF9R_b6i%_jW6j<>wQb*9>G(~5SU
zOxux{hpF**1w&J4;BpJO+Fxo}PvRjRuECKz(&|5eOLE}6+J?xBIiu{}=|1ztA(ZF=
zD382YYD>Z}%lKW&_gB(wIO_0ilF!GK)F?~&<#vO)l{`{7az7(#6$sj}E(VPA*knTb
z`$vQF3P_6dAGF41`is3}D!tBZpy9iq`rg=)-iOJYtyKgZBbhW6N8dK(cCul4DEY<o
zXB0_*lr>&o$4}x1MSR|M!dY^?e2Rap+~`AG6hZs~Ab5iQ5mvl`eI0c4a1Wd2VgEYv
zOBsjx6ug;6Hi)kuO<EQvwpNA>2?VSs@A->!=H(4|5AS;=tZ7n{-Kw(HFdQqBBx7dv
zf5@L@;8%BSbzo&~bR^d`{&TwX>A*UrL3v08s7So@iX%i-;ufn$@*;_t%6HaVYNIYR
zmpG>v=dJOxS`R~*<%+a^>jTK2J9<*@uHXf!_$O9l?{)o&A+AdYd(V4C@Ol`UbcJmN
zp_j5fHC4E`icS>oj59uq<M3*dZJ|hDli6nAdTihP*>9;drk%&KGPY#V>8v>;z{!aL
z=;`9@Ad3>?CgYJiJR4{p_)NcC4hn-uh|6t$FwVk_nx!lgYi=l|b0mqH5;%{c)!#=M
zDFoIud;Tp&6M6UhwJ=cI$@nwLJR37%&SJ_j3yIh%%l#5%;YKy%kRL6xLtz9m#j=~l
ztMQn5MxudaAX4=gbT~Qov=!!0wRW@s1dixsgSWDJiDvD$U>x>shTmRWS-%{wlG}ml
z_)vncgboWmE#}WPKSQvG!P`Lv4J+xEb1eICQsn_Dq;cVp^54d-nk+Fg9Y`ODG9X_Q
zb<-1TX+br}I1PxbjuALu=UX&?j54XvZs|wlM%i=Dc&n43=J)!&crcY@mov42a!atI
z8<{*yQb`oLKbC2IbJYAfXUPcG{zYo6=iP~alxY75NF?x~Ybp~>NeYhz|FrW?PUrK^
zP;md*w3IK3USOGboy~5U*S?m`ocOVVnlolux2fa=2wiBmJs@b+gsV(WO-=TUa6z(s
zP-qlMUBs}CmO+_BqIXt{5XWx<A5xoB=BuaW$NC#A?yuYYL^t;S4fta+qWgPSXTFEy
zkoIGXKYgIQj&iFysrO`=kBj2_dPS3nox1M*G`gFi{jRe0X|T9g9~np*?rsy1i_dwS
z*LZb$?|q9W(V!H+#NjtSl#TD~n*(x-46k}yQP5*-JXe|W&#SdXd*+HjQ?Tdx7l9!D
zd)M#@!KzmacHI3e2~eXmZ06EX$qI}jA&@Fa8D_jP$j)&xC>rmK;3!&yER1iioRH$5
zywy<d>}6{o5Xv_9@C-@)?`WEC+`r{K5YeLa)IiM@UJZJ8!^(tmv!KIKwYCc9=7aP%
ze?M17ntDxOJB}VQ2U7E@Dxjo3Vnx;nI(rhmT83Ad_crekb8KH0d;&|@ZwEL%Hfw|U
z$lgc?lv)sTWXM~{M3wbIHm(}PO7W<sg9-oU<O`YKb>|Ol>L2okm+Cp_*eadC;MBNc
zq8r)&<(Mf}$rv%9s(#gL*%*;AWXq%P5$?N8uSWG{i81nw0l+H^cWB0JU`8jfcdLc!
zMqmnYVPmEZ;IY58LRC+<Q$7_8=h(-6${b>V^tnjs)WRZpt4h5j@)P6JX=R)q|L7Ka
z8WfB-O$agh#uOQiXbea9nlRCi2$gJ4c}VtHwW2!i7il&oHajtZj{cmgo4mwp5gCZW
zYeFgYGJ>t@48X^H=a9D$`#rHZXis6baS)!UOjEGsg_S3{r|z(c2dE3lgHegf9ggI=
zpAztD;eo((RFbklT;7l<30*F;k5BB7SQ6J}qvU@dqj#aKFJ8_Hm!em(L*2^S$z-#4
zm#*E&(jqeupFFFfF<9Ho66R@y|Gz?uy!2W(%SZ#TsWW+8vD1h7kY)RT|I>hGyb99x
zWbXTVKfp(!2vhGV<6i=CJCs@i;qmqIP7dT?Jt)Eh>uyYu$0%8<{If#3Ka*MzA(4Ze
zT~IxUb?=3iby%E=b;2-OdnJd|&>GQ;B9i6Z_WABS*S>Vjh2s4zMkP?*kQVU&Gqj~~
zf$8!Z8`y{hHO>$ODLF;&51Iq~jSL3Srp`y6OhbaZD1~(%7HdJ5^P=qkp#~Azp#nQv
zFFn5xJsQ*3Rvc3J7CC8ev~D%Mp0#d&IoPz+s^_>OxFS$IBVG+taN&UNKGNIgenm<w
zgTHkCGQA1dvy5TYR>(Q@IB30j_X7yQx`y*gh8?JX6VpwJR)AfHTq$~V$R30&8Nf;O
z$^CZ$?{R~KID_;xGNT(i5AX19Dp`1i`@x<O%1bB2Ts~8#0%z>tAcY96kd=LU2|WMk
z`KR7ZuJfuUt|5r&I2=z}Aw!Z>AzL8_r*$3>klFCtO2gF`O-!<Oo}FABfs+`8`1S;h
z={ATG8DxXX?5|cO>(kLGpS^lmpY~|`*5um#`u==(|90QT2So}QakA5=qg6HJzB?Y}
z@rT;Syi|n;)sVyh=P+MGzlyAVQ<&D{vzKsQ9m!)ic(xCJIC?3CkpU_gfxm7*m@a}L
zVW$f7c*lTu;BYcovt9`Z(5BZiXK7k+R+~p9eDu!1f&}BpELtt@T`P8C9E!fmKFGl^
z^rMPg+k~Sg!h8gZtzbX{g-jk$S9+!VfZ+s#+xQDrwjV|$MHrt5Ub@C%RWmwh0g|r|
z{e*=GMEt|(uOz!c4gj&kL`oz6vnN=<?QU3>`7KdB@UF#F0PM~j$hi1SC6@;;A4x2J
z7NG9A`7S+oZpuGYXjvhtyIW{U-(mVAgSD^bEU_4pvzAFZgZ9(em6flh`Fy8b|Ajs6
z`#_U-iJ{MZQB_u8of328ss`hyViL=f2oh2P9ba|^ZT3s=bk|dHMv00i&;3Os>za#I
z#WZ5uul37*SiSr^z}6m%N54~J;<iJ%hwicM9SNhAx_yTiUH?kCW#>Nimxg&K9YJ`$
zXgWMw6HWAb$|0i%)K@k>sL|O#+F8w^=KuoHzXs0%5R-wg5?P<+cbZJDMpRMLxu>`p
zmz%FXi#Kay)+H+`%|IuZm%iu0)zg6o%%HjP)$zrX@^eK6VCyri?S12M!Y4<Tb1u{_
zw}qQLwLVz3h=tLMIgYcNu9MT^+s2t-d)QbTaNs?RHMQF@XO)bxx$HS;+rp3EpCKnO
z#Z@g}qbSU^4NhF=XIq%@L^P(moqLG>$6e|@Tp0A*r$7J_#InVBKBTm2Q@U&G#lyG2
z^c2@#`;v-1KySA~#bPmC_?-98Aw-d3_<z8oxF)F+%P?v_FQaU8DqJ^eKE*@=1sFM2
zE?h<oNBtpqJwlYw+~2?lr_{qN3#2ynPwdW)bjO#!>RtR{%#*@R3}ZE|<Zo`HT$@G8
z%nGB3V)nkvdA}U|r-)Dvxxu(lv6TBCWlG2%UmXzului`_`QAZrG+o%koDAZV!3jgz
z?3Crm)r8>z5a^&Q^l`R{IPjlKRHC$v!$|TN-oHOOQ>)X-^s&pMc$~@!+_>m^P`+dh
zP@Vs?z63?5p*_6Nj>i=<;?kVC1!eeopVYwG7EXMGQ6IgR@1hPP)5%cae~N!a;pICj
zZvsK)XuZ6Y{AI^Es`F9aCMz)Ak__v=nHP^Ez)R;0w3X~(NwL5sIDJaW?9!Ux#)m@$
z?f303+vAf7SlwWx7k;x!{7K~Ip**<oOp(Udi_xQB<X)Z5c<MPEj^)WsRJlCRV2z-M
z7va3vlybzTXo6^J`ck_irmzU>@B{Rez5%2we|$>EyP>5j5YjSIB0&61Ot@B6RYrnw
zT=Z%1nO3<f@*J44u*WB$O~OUc_$T?l!(9}caUEl7efM@W=Ea06`LE%YjiG)^C+A2G
zOU5qx16rJ6iUwkh3H9J;w{SS^7$;u8Gx>Ox`|<w%%J{c&m%2`N=5SH00<tk&0+<jf
z3pdu?UMCUyj$P{ZEf{J#<4-uy`T{?me-OtzuzN_)3UW0zp`|zvF-n#-;aU%&vkd(O
z?1G~BqCRA9BhS59qS=QaBqcNp_T>mB%G26F)f8B-9j|1wV6vH5mK{yb6GmF>HY-bs
zDuFK}&75Eh-OgsQoy|J5&nc=x18qXRcJlrEucVz;Wc|hDp6Xy-q#?iXpdJuJ!%`}0
zg{GeSPCX_xO|`=!z7=oI<Ppjt>?HfQ-qpJX;M|)k6%Lg(cQcbS#}t3qOiuKZm}hbY
z;K(d3+R-CiR6&**G2IfDYI<26g7M;fKZ%pbR4H5wv#TI5stW8USw4uv1(rAdd}@_L
zI$3Y^6D=osUX}0%FELmP@hxvBkIp|E#_p$nO<VZ)3=tkzYE}`27^e97|G+<-O-T@|
zrutO--EI=lXIZf!7C_s{BlNf&JX_3u=LEwJZNWh!+#-&YeHq<lg~DY*3*z)4Rgq#`
zV_Uf&Bi@1!&!a+Y3CmR*3~VW75-*5MQ<l)hHV-f*8cTXhv4f<eqpT>dh0N?On90?K
zY6!L<+|5RhPA+ECNH^gQatGpec%R;=uNCY}A`~R~?F}c(;@eA;t)!dH7YZ)@1r=QF
z=a^nL4gNyLbZ=6oGJf6!$LjjXqUnCX%rp-b1j|UmC;#wE_|}><0!+O|I}$t72<3q7
zl|{1l0#Y$}L1;$5yV;Dgg!PL-Jt?DJ>G{wG<Hx?Eso%3jQ%Gi%^p4VWjj3X%8sxz5
z-H4>`LjETUuu=|-BvxHx-j6tVy$Dx6rf9J${jtcXAaAuhD~marOfv~W57h46h~li*
zO~Q~)K|*CY@{Mr@LQ)GQRQ83*-Y@vg2pc7To-=P<6Tr1UB1(6s-tHH+!PzLrTg(Yo
z@Ke{vu%wY>VUEm(BRAmfj<4v;H(;-jQq%NqgEPxPYaUkxhxlqc1gc6QDF_oV4epO*
z4+d5B;oS)g{K>|e=+&&ke9|1tx(FEi__rT)H1-KRR1We2j$|XBbOb-_h|XDY`%5me
ze35PV&YB3PvW6x%JIHx|gmI`#NL7&AEU$lDEo~$Sfx$vZ<=Lri+Wo?dQqxoGdO0q$
zdZ{h+^16w9(}DD4>Vvz}Nr(ZDL-3;z)!8C;@!GK4X!au@`d6J&;@_Nl5@y+W+=S%F
zXU#J*(>PQEz?dR*C5qwvIGYHhCF7Jyqos$&W8O$UwiEC>;HUti^jDW9cvssW8o_5%
zQWz<jqvAPb2DTKi9Sc;qy(n(b*_{k-r=F8D3&G$4X;yV&*5(NJ-w_O$?YhjE1c5D&
z1JISf@2=9AvQ@#HRJgSSo5;L5>Br=sTAUx}UnA^*s}5{%S3&^+-_s4j->*}#F3&er
z{Qny6N<#$%bsq^8Lw&z~#skn>b=|%jn5MXOL{^bvaRFnYgvQ2KUOmIEE|5NoNI34D
zzPQm?znFfA8LpWQ46Kudwv<XB{JIwZ?y85gmZmT)UhXawULdu0G%eB@+ts{Dh0c=_
zyPo<6u=y?A-2L>7Rsdacbz^Z&R|tOC5JUc3cAgE#^Sto1U}#+Mj!bV*_KimVuUp${
zN4R<_j3pd^&noK8M?Dq(D&IgDH!;=LOiq2a{5dQrc+c;!PecYYz<>KA4^$S5wl!#Q
zs}{R5kX9iWj|f>VonM(<RGVt`jB64(y+YX)$XoGS^$fx=z6+DQyj0r2-;L-|*nPl@
zFo~A>HzvVrho}o01=poAPpi8rMtnGIC_S-cDdIo!OUQM3)d_jQu`q(X`WPJ5dbnB%
zrt%$fq|Gd07g0%!5;n^e$p%T5X}8pvW>%^Oh6+_aCa$ev@dQKT&|F*jrSa^$cY5sw
z@Lo-k4I&uj?4Q--{*vtRlxq~&ScER-pVx#=1{p`i&)m19%HKwczl}wN{&$|AX>JIS
z##*8z&&uZrDK*DT*ok0<zQ}#rk*+B9s)Oxo+GoE}uhN{>ftd;aO%@#*zAQ4CnW$$6
zG!z=2(7~g7XC+>t3l}<#Dk7MEF4N^1aCM8=uNbr<5=4g?Nhy-{=3^lrl#ql9fP@@=
z7YPd+_aqU4nN({P6vd)`spnM&X{_i7yh9fV@Us>M$+?@nD`I26-Q(dM);sUBcr_K?
z_lavW6v$viU#gE64C5!yxrDKw;gSrKXCCDt^oZ0xCOmAD0*`7T`F-8AbY4{kY!aJ2
zP@t5-p?M&W9YFX>z5ZBbgqR|G3?&kt&XJ8^F-F`PecAJ$E<zE>$NSW!FZdJ{B<}4u
zpP1%AKm(O<a&J^&{ER2|_N4vV$~4S)a?Oj7*@5T3oC<Ts<4itD|D%f5Pa|0{)Ha(G
z-lt&f;@$Xar_(vKQo!I*@oHxWxC19xa8(wbLKAgmpCdlB3ZgqUdH1Jdxol~@BAydz
z8f1G}ZoB$_Y`p_;C12D%8rzuIHYT<`u_o5UwmY^xaWb)O+nh{n+t$nP`~I(9y;rYS
zRo~lH`&8ZT+kJYUjkWll3B4ANLe5~aWQSeh_&4<45q+ND?jNonZr>N3+LSViWib0u
z6!WzvEqBXCQI#&PXs#)=Q6sF3uW?eWakV4`Y+AY?sV;6!4=u;dx%zf3-S*NKqoD`^
z4f%JZRhD6j&By%uNye0q>w~Py#E-M$?qB;}p0{g8<}rkbZaeL`CjDWd|603D1xBie
zHhGaBn;(g=y0J55C#zaXy5H`De8Jz-oi3~0zwd>CsygeE8uTAjQ~77qSeix%Z+sB#
z#bidp7vO}!n&NO%o){P0xyb3$27=ZB*K|!71kp7%bfVUYk6D~ih&Nu^fw(5O2abcK
z9vi+rD=*I;agEcMURYQmgWGuq!uWz16^)9;=Mwc4xBao$#$s|x_c6xbC|a_qPwAA9
zE)L8r2)AXD8mh!lW$3qLl7j8`^f@mC|E#(MskS@**>CgXrW#NP0>m|P1bsY!{GT_r
zm)Dn_gzdfJ5T=7m`|Ya=&9QfkiKnY{o?IZQzI$14pY9Ti686&r5@0GFW?%^CwQtqN
zYu{@>A;IE6Mkm3kChXSPGB|aIl7BsTLNg0LG-XVls(XWS@3F;B-<szPXL9X^6HOkc
zpC>sdZivf%{6H+TKx7kjsQHls{99Hj3EqvqKAqm+;jDbHD5_E%tMDp}Mit-E_GgQ~
zcDCVHK5SU&+@>b-J~kDAfj<TY&=$#+M&}Ce8S}K)O^8~0YpG)-9#YkdfVQC<uP<s}
ze9tA}Lk6$+-xV#ssAc5bGGVE)<#f*#s~Zvm11$}-)%jPo?euU(B-^FHK?OsOjs8M)
z^|$?pjrg$kk=Pj(2TD?QCeJaF!JlmuoX9zX;wi~g)OTo-ApZ{VTMu|D3-q+5QUJJr
z8sEe`D6^{8Zt8HKn~`hF!Z&7dIpNC>=sZ9UA-^9c8Z31Dvk|2~;#u~#n=FiGd2tv#
zmEW-sf_cDN^eTmv3;jh4pjfB~QxFToHE376W!#DtOCG7ITx2npnSckNd~l&d*h#|(
z!+5#X2H5|z<-mQ8*536|<s(2pHH|5KHLAY~x)O6dg+OH>{I}#kY80QLG*Hf_p*dNP
z_j}M#u=q?aF63&9TO_#u5(9et+1+4F1D+RsoVxY!Tj)OfFNPW(K;s*33*H{1^{lP7
zvlDHzJeKwzu0X4XRi=105Gr_qr0A3i_R8^bGU0qCIQ{29vqk+rd*oGWldU@z<MdTS
zvKIeF+Q!~N*$fhs&BM{*=6qIl^<KSmy`r(h#}#uo>k)af$r~@D>DHzqgU&4u3cG&P
zTmm|#kktsFIEl0Xg!Xgu3ZwLAG;5Pkm9q)@gkOk}q8VhzFjmP=@C~%_aF3Q7!k~Zg
zM&#}M>K@$;B}8Yp!|KRd87Q||Ru*yyVc?ni_FnZeU1St^1Z9VZ6%toLxWwG8S@hho
z*q@kFW5i3KtwO2`o7M5`7Y#~je<zN!!<f}J-DPD!V;5uv?Cqy9t1Ti#9r>L~_C26K
zl<@*CoPan**xO|78W)xQgOOG<MxCu~Th=Ka?#Ddmm5|etWBshvIDxthi@>l-Xc(As
zaZ)<%`yoW>_Z!jIiXTAkw3xvCTa;Syp|83b5EXF%o9}uIaev&su7D(W58tbH<0h{V
zXT@eK86*Vk*)BZkbOP^EELtjdFcQXX1Y_KyHg@|{s@<y?ABeA+)*F$#QPYCnH49jL
z%GRU!#(^D@JzfP|dy-a9y{onHA9?DPO#(BKU86Z1)g^DcG5nMY@<$Gj!29OrV>MLQ
ziIxL~*no9tHLevcEOl7ptHRBRHy=QMV|i)!T@UCsr!D;bA=K2jtF~~sfb}?EBHTRw
z8TE<>=L3?;U7pddSx0|+4oL^Tx_NumTtg(PFuT}z9bI&K$27Az!ZZ9wfKoxcK(ol9
zOQi5_>c7^rrVl0FbJcRQ#*5c-)B6*jKPwlVmznG<m#YhU?MgP8V;VOL-N$z5%Q;J`
z6#&M&XT>vL3Laes_uPNoXJ(Shle5sb&p%aayR%Hb@0U?&%T4P}hx5mM>OX`L>Chg1
zaZ0`jJsg62{petj^vPq&fvw$m3GKnNHv5(#4AjA%<bAACISFb0k;6~48M<4>o&8LJ
z7E>8&|NY2y@lY|pj~uB_q}_Cjnn5O}55!GUDhzb#c~CgF5Zw!kq#`B?V7-#Hp=v#C
zRn;68E1Y$bf0|uFVvs)7mq#=9v6y6nZ>Dd-y}{TwK5XWfpQ=}`1~Kd~gVu9&+~J&L
z8ILchXZ`VlT$Im$h{Q%)L+T=5*Bew9y%>pf`g2cu=_49768UrX-~w4Nf;oAlADHi|
zv&{Tvod7w7Nj6%jkPA1c2)T|g!{KmQDo~?EJOeuu9&j2|#+BCeV_k5)62TSJ1x15E
z2vL1OC_=?gG(H)P`|exmnkAb#D|VBpc|+CRF;{fsIfB7oAxyYjxG~jwV;j7}QeB&d
z_HjhQHB?Kh&y^V|h7e11&BVg5{&Ov59dH<JW$+E<jVktGPc($x(2_}eiji3{<(8dH
zd7&Z^QuNC51ZTbhC>z<*fP%mRc-mtL{W){)5;3#2uN9;;v^19DoLNHF+XOxYK18M<
z=;l5-J~;t-mZ0?JjW=Ep-=av@{00U7XHU7ktb>scb)B*S9ECx-%IR7nI3NKNHKU`@
z%oTYMLXTpPa@Fcb#cy~JO39T4Y5-=iqrIa$4u3f0C+8;z6Gx*z0&zG*d^{NE%*6>M
z%@k0;>wY4&CwiuOj(`*!$*^yf$x~5uybP&vyFBOp$`IEG%^jQQgjYUc2s`FUU_!u$
z1<IfLp{V?0y&+c~tUyPH1|ZT}5t`dHZ-XqrFVHcep)!|;N7~S1gBD!?eGHgtHqK5E
ztx8W3P7HH)J-?2$xH%0B&3GQfGOHrj&~o!^hm*)WHh&Xlq^4WSCf@uK#_3i#GGr*H
zaH!517oJCeVTbu*R^wNv4|mLLQz4!vo?e57fhd@a7aQQA?MENI1za}NS%pU?*8rvo
z`SX)wvN6#yocJ4}VXPogQjw5^o?&nm({9nMRR!%^AEs~3<!4nMiVtKr{3~zWqPM5`
zKa1hod?X?>P-)FsuJZyivG}?rz?Hgl^tf!uQd}dP;?_Fbx-1*pSuec`xkgxqQ#iaG
z;+`j(`a9`WNa`-yfXIQqlrg%DUheg~Y=uD9<#zY6Z&9nMNr3UVH${Ri+2c1$ykzSr
zR9LIYW(ZDYaLpTQWd02@@#Y_F95&IUnUfdI_hU1obOA2by0f|iWY}DiASiJ@l@^T~
zB~83<+VkPj&%wZ>BlDNq9Yk|rcrS}l<!c%>59L_JGW_EZ(7BpU&8OrxsW`Gz-nVgu
zAlS`UWJ`Ec714kUuY+<-uu}Jbua11z(`Ty^n)il&M(XP4q%RwN+zyU<Jyd!}1^}b)
z|MMpDaGmtY5(xLgx;kv%Xl`G9?XI=*)&Tjts%<+!GqraOA8ec3zA+KrTj;cb_$|6K
zd>J^;h%t^)lx9>(%@o@$QDu7NSnd-wZEM^f=tFyHFXYBn`%S|sSD;)0`<<5?Df{1w
z%c46dl0@3T$mEV>-$q^!V%Xu)6h5gN1UV)F?a%o~RhUosd$yazU_zp7Pt_0gSg77Q
zQ&J@Z+E_c<WikGm<Y!ZK7RPU}>Z6+;gJpmy@mOCJb3ch@Pp;eo=IRg!?eP=lscWeY
zr76a;afRjd_q&IH^@fy4_prCi9CUFihZ@noO%0to0$IHI?j6-Imh3sU-iKo)p|4F3
zRS`F1y8<@X;6ii1`0Bl<XP6H+8GY|NC7HEF8FKudC^;NCjI@rxUSk6f1Yg|nHNeV?
zQ7F_eV|K(%?LWx*V^N8S?0M}y88RCT9}NE+D3VJdvQU~GWo?@^5u|P^qw2PUe_0rz
zZ@XLHhmcpO=5Odhfi@9)TkBa^ac~fJkVaC!6^G$_Q{nl8ytMoswrT6gqFEPfeU=*?
z&0}-6_6AGBecss)ts>tHzx0};4S=nz(^f(}?PnbJq`GKhjgfJ>GNHjGd*hO-%Pt(!
z0Q!Xm8gBntcUOwtO4V==%QpQyWzr<c)!$)t>+B#|`-GY42mumMLEJR9R0E!kc+$B{
zxU2_`MP0mmvm^1@9J-PaHigz@qiFp_DG>rKg3^T0204BRh5z&_#}JTFfbrC{J&PtL
z`T%zl+sEUJa1^Gah;u#zV?PGIr+fs5)T2SShsgXQ$rx^ZB9U-%a$#!FJcuB)K3LAL
z-4cE^XlT<K45Z?5C^}JYI%<bCi_bCFBv!bnzqZ}fq2<(BW5$oXUs`0h$%p^`9DCJb
zAE#@lnsF#^B;ho7g#T6mig8pi^vxG0>P?q3Ufx>E-yT#WxrY91W7GVLgAj!}y)esq
z%df2$bofS(8I|1rPuKr@L+}?ByXpY9=={}!_F-2vMTDL40vl8WyXYOkj+yZxqA_?h
zMn3mNg2mJQEXDkcnUt65NBhNNbdp9Q&D)WZ8of@Ka388V0<H@KP>&}BA!H&>vEgAH
zH?<r^-i;Nn29au6fj&zBr)nw0L^`taQP<Rj<&fgbL@JeZgf@tDFrCN&w30$i>U*Ga
zO9>Ndhw>o?98|y6(-s~u#qL5|et^N#cPJF!O)HfU4L8=k(c;b{00I@8(gQGvE@-WE
zGtv-9qP6%?N8~$z+C}c$W!%?|A@aPLLN4;G|He3x`=)@A0iGoW%NyQuz{-Xt3=<Da
zd_y221drMr0L?N$ViAH3W)zkbjAsO6XpPi>A`CxEViJ9!SHLv6@2|Gwi&3owu|oW`
z&)Fah(?6z5hr+kHpbsIW4}J8L=Oqqbn)KkiMdwCDZ=Mg3t_lHeGk-;ZSc?td%!>?E
zTgsI1*j2MApH%z*<uAD4Rt8PCj2j7cZqZ>p%rug_sZT%CGy4aLdI1=$Bm3$2>9mt-
zq>Xu2IylYmgpSEtwo)|D5nRz<w%Y#d+XB@h#-|?5Rw??@n~fphOqNXa{&HkJK!mX@
z^iwx`WUPR;V_a~tYV6@jhW1T~#4$Zi`uBd!AOf{AdLG?nqEJefo+&VQAM(HPAkU!B
zw|;Nqu%v|BM9On<2~!fFkHjYT*j<82sw$k7_aHP-fIHMkSH3Ti)<IAY*lsFvGH6Gl
zX5yu{4E%diP+MoW8=>dn-mr$Xk9*#EOoTtPuo&Q4%J_Ou1hNCNGZ5U8SBOad&al=F
zkCMt8W?U>4!dWV$L~yQl&Bs3`TwnzyK>@76JFOm#WpLQ*^+<z5pcC|l2_I@SPlNNE
zy)fM-8WUw7jv<dBjM3MymKXh+$xma4T+jR#nAvU?7ZaY;+e5^5oMvp-Qy|+y)Y5_K
z^BvfXC_lNFD%Moh=TbwKk$yNwKDO#^Xw6dmo4co2!HXX)ZbLPpC;vT<N;-qML8oxZ
zalxvQ#}Sj>HI6E{ehH_Jf&vqVE=3F$WrAJMvW^*@o=g~O7hT!eZEtKwd59_9a=wV^
zR}g{qa^yTRhnbEV0^7jAYrciJ9V<Mpwke=LOxE5v7Huxkchwz{yIo%NrOr2>iN!{0
zUl@H%?Zkn($?LS-NRGdSvSlAWW?R2<%3hzxsuuO;BjRPX?2LtW2(Ea5M@`eb=8bUb
zRAg;Tc4yJ;#Qj1F#etVvrD~2QN!67TD5HW_^-}NnSr9cONa8Bn!po4VOQsALGY_~3
zqG6Y4Q^Lq}6)Ti`X~v958gLw}G>n+W{xB9oE?j4#+#K;GK>3EE{&pjqOdZOd5OxdZ
zmu@G2oc`ncDg;jW<&zA^v@5*Bfi0C+qIMwacNYVAGi@(d_;~Z@eZK5`^<H+G3g>xo
zV>cmD`%{dap9*Jb+^0VCZvx5{m;6A;<jU4gfu38qmKoB6i0?81djIn*k>J=p68{}x
z4&g;J$BV_1+t!UkWpe;_)0}V_`8TReO;+^LtZvWF70UuL{6?))RgmqlH6mN`5SZau
z6$>$VaTt{XP~YUWw{dRL$Ut#)`u5Y1C73>1V03cF{W%?O^0K7JB3?+;Aq+Tv5ZK!Y
zgPVhz<4A47ls2v#Bwd3sq_YSqQ_y@G>kyvLdt{|~;W4BSWD8)k_4lFwtbux4W5p8^
zjB9@1w^942a}2kmFSP*zB20|D^Tu!0zc;0SRD)E2SYU&9$V6>%nkg7Z7gt`n-r|lO
z>9`9rEVR*-z^PlC1#OK`6AVD>Cy`ueX+BT+vg~a^7TR|PLW@9)G)G?-WbI%7`##fz
z6aX2}sHTXUP@y&hHVZbpGRXoDtG+}9p2t(8_-yvBPcY~>a=2_E$Q#L-gxbaR?7jHv
z>gM&?oGyv`$;xHSy2MGK1&xuj=OS1mo3fZCgaigx!kO9vV{yAjXAUU!lcrkcW0|0t
zu!2DnICVgQS;%s45&h4sOK+xH7chbq$L7b6G6>EMT$gT39Wp7|x-*5h+b&w3^GNAr
z4tTuw{~i#R%AnFGwjs8$kj=RKHyF5DrVU*ORo7Imjg|DB_9w9ku}LZXPikZQxH^bq
zqd4lg67x`ca#h1v0AUYq&ysV|s=L0DT?BJBY1y6#X`nJ}N20hW>ksE(KB9N|_rIjI
zC2Ce<l~UX=-Tiw#+7a>KM@#%%<3ygH@6N9??GpMt8o;Udna_Iqc2&dX-*SyCc^+jD
zg3izsuzohQHCDKE#-O52zIA>64d1TM2wOq{Ktw|!wgGn+xW2A>Jm1dvJcwbQIUBv#
z+Os~)?-SbluqWf(nLP2x#@_skRpHA-(NaeIV*tF|)qLPj1;_9bb@K5DE4)>R1hUU@
zk^u)e6<5227g9{SgJjt(4p-!?yh*bB&@w&SPf6K;rKyqkZRrlr&-v%Oea7%%d<N94
z6^qp1&R1CoK$0_C?hP`7FEi#b0mluRn{pDppj5<8qVO4a9lx!faCu4VB{8GZP?L>C
z;y}pj-&Ie!hQOxUU6mE5|6qRDMvU@-?KdLT3Ww#P6jQB<xOZtcU(e}Q_~DXgiTa#z
z=dh7Zv|H7JR`^{HYD_i?6q4mKk(u4TtWPcB`mOX=fCWBwLgQ@eT~{wt2EpK<`+{Nm
z?bl;J`0Zrel2bfIYhQX_fS|%qZ_b)tPEUZe>r`$GrpJ$J-;_0+CX+l`cDl1V)5|(L
zpwv#D!i=M+8!*aLMG2NV_mB_Z)4wL1e>dVuQtwk4e!t)rpBiB0BZ^5-kr1pM@Q@Zr
zEnk2G`hVV-nNr;$PTlu45P&Z1I11!oFVH$S*b5CD_B@a!zeRBRr(NBec^ZcJ^a)&6
z@jY}2)a8`UNI7VWafUTV+Awl%N|HliD66Gf{)jErgl2Rc7atn<Wi?Zkdlltb_aZ7Z
zjM=|*A0r*sLcFtr;_%b`EASG6rB8zE(h9l+Jmh^7Q?sIg%`fBgj73pnkc37-+v6a=
zvSZJy)lZryg2N^TK^bv@xXf|;##*TJ@U9KZ_gZh&I&1o&NIEU>=5XNYA$yhqf41x-
zHyD9_X^!OKkEuOxcGpGEXMpwl`}Z#4yr3|1G%+e!PZbPI8h2;XFg#fT7p5Mxl|TEb
zfM}To&QgN;DA#Sh9nqy{iIdNIy8vpGv>{P273Cn8@nxJbYI*7E!O5hpp&sM}7-x%8
zdQ7R)G4z9~kgSTQr5G`VCoA{0-23Ano*cRl{!CdTJz?4+eVyd=xZDtk`>KN0E`|qh
z?Z*epe>c|i4^_!D?22GX;k+Zm;MLb}zy?g{1ZVr-3O1Uv;KcSq1BYOopZ6#!DcSUH
z)vRqBJm0s7XjX!?@(0QNKU3*O@GXLDU*w}1<AKq)7u_fjk-w_HStAQ(&ko-1`I2=p
zqq@lpP^XYzVKIq^!nwnL7vDTD3|JPzWbUL1--72M8o;0)U&q_yI%N_bs8%bT0y=J7
zJ4sr~zIz(}MF9?hkdJS5ffD-p4kj$qjR8Vxjsw_==>MUY0d9QDzsj%5Pa2f}3mFN0
zyOBq}kjR0;YCQ#h>0c1%{<{l*?DJ*xzb0R1(t!m2@z=T3xzwe6ZDA~MN7C7W7()cM
z9%#DjJL`-QzJSPgl1QEW(xPN0mec~3&d8R@Hqw7x|1TeWv2X0l2cP;?Gi&0ZpZ|}v
z8V}KC_r_5bFY@zT$x5~c+of}MCvx!LoH+8V50!8spB)!32m6hCA?9r__otZge6aaj
z?N!U)0w=NKnd!NT4&{L{?tWTxC|XEbz(wj~g4VoZ4z$g&JU}&Zs-A~r1L_pa$fr7h
z1;zRsHTF^k1`ZSlyy{8M+i%!!2lIqPwjR!{hU)Nc9h#$D*W<XHBqZ2@`ND>nXitWT
z?zj*u;9sbZeZN+D$`$wz+NH=PNMD%hibv+(hC~GW!V1L===m3b^X1Wxr+xANV<{<+
z%kv+2OOZC&OLKX1d534S46-;&kiRhfPdK*a9EECskb<hRNZ**J;VKGoo+)ybkOZ68
zOc5*94vHZ;U0$4y9djy099=BW5mzaBbCJqjDFt~2<_qr3HW+ftEmfQ$vzD!%-b!YL
zx4<~+hzUp>fw&FG!2hD>OJQs!%ySZi80QPi>XzWtP+w?@p%toqD@QaLz8ONLW`QII
zCtlV1{|@Rz{Qn=6t68-K%5M*13}U=`irC$O@({o)#PWq&pz<!{Q+^|@NO#q7<h0&W
z?A10*_2!c0J-c>VdTcmDrPXDn$OIzI%SX2zmR@hy>{mW-IngJYrqSk^K#3{B2sHHr
z+RC11pbbIZkMAK^RFmqfg&d#S@|<6EJ73ejr2R(Tsovx9?(phsEH_HHs~S{1SC^C=
z#m!S%AIwWQ@Rpk)I;xKO6~%YH=z3TxE8}vPs+RF?YuBciGM-0bJdw$9?ai(sN-8z2
z_On*(Sy_;1_nMs5TVCnD@^l)(do^A$Vv3_@_lYqgXc^(gf=)k|;vqB&)N`Y)XaB?W
zRm;sVI{Wu><AfdtvR;ZhjX~7E86_m(CkmV?2;>_)fVxX91p26hAS$e8S-(KdoJ&Se
zesh@ilA|P+YbSG20rAj&hi6k!rw>=PIO+*d$1@K!d`n|)klHKku7ytpr`y<zYQ~!E
z#b8jKXFUpx>%IUFQ)I=;RiI6HLlyWC$g3MR5W`XAHjwz`0q$PxGG9V}Are$;B%Q=p
z7Tbjf<f5t1IRuR9^ls*v<)8m=8cqm&EeF9u2Wuq?85Zn^)c31DVC3GJPHT!1hRm{k
z)*2Lhwzp?(gIN$kP3+?p0nLKmil~uMT?JCA$}>66hsBT3V^L;8q{8B>y$gsMcDF+<
zvV$qkJ0vL*5r4YkM&uV|-Feib{Mnd|V}Tj!!YnP_E(=F7nrL-9h+RVVsr?(K?vd{I
zp=u+w8R=5*y7%9ZPlz}y+<4voP`MUQNYV4^mkm(LJ;fNBPhS$MeV|H;`ZX`Qdo3gl
zS$y^a=g>vo3)#qQ>-L|apG&&F?shsy1GCPrZ*RY0=AbxheAbWj%rI}vgtujMkpRJ1
zdA~ipew0~FWnr>Jf~m|W&DK3~hryyH>GmY<|C46RY^gYeEZey5%U6qH{H$Gu${9PC
zKjj&N3f&z2$uFg~>WsabfhI`6hnsQN@Nc}+9VcoDcY-2iKfT2q@r<9(+gGUhG0e}5
z`c=$m!s%}2^|E+1bmM;Ii_`dwlbG&RLkqr^yO{3SaC~z9ePQ?f(98K1)2HqI<^JJ=
zZ2xgr-|GnU{sQRodGh`&b^CM|!jPBNejNAPBl2mD^qU*w<vD2C6TPe+sva7g6LgN=
zTEMS_RUh<?N&%s3V|=D^vQq0>j90+xY;ACdT)Qx$e0BBf|L691bPhlmaBYmjN55xL
ze&H!%Iwq-))~_u*G7qxR#xct1xa=^rj@w+->K2I?4Vn{PA-QBc?|$r6HDEGHR`zlq
zp>-I<KT*ptwhkLYm7?8G!Ih%*kd0FvI}Q|4E|rp|u>2q{hTx{SBp9LD53P<<kx}_1
zDJE~4Bj7KF2(}8{^8j4;BWb=>dq56%;>Naq5e<<#0`3XugShe%OZE?#(Govwb#<?8
zNY+9W?YherY#V>xKe~plBT3TKvhLd8AMOo$(usAEf2kwAO?qDM@HK&-Nl8V{t;fUM
zgZ%>gC8P9PQ-#AE{R=8^D+3s?$mEv!=DvR!3zRDx1|O+ZfrjbX0m^t$Af&?XR`1y^
z#^QvdgrsC!-qJn%y~&cV(9#?|oQ{Dv?8InL7UqMiX`(Ae@|PG^3Vp7Y7X5QhG2LCi
z16);#`o^^{yZq^~ZDtmSmxdW>&9-FyC$Bnz#51bHMK0SVvq=Nj&nZpfon3z#{ogO1
zFP+pZ`Y9J3z+Am$Ds;r3TzI>uUy#3zEwhj1KpIsmiCY6s_Tkc}8xoe4s(D*>NWD+A
zB?L+2KICDI6+sn2eN~B9EFSNSI+sr}n{2DvLz%T{EW8XV7`o%c@fme%X(X~xEZIaR
zMnQ?k_TB6CY}^b)(4ku}e_}@a3Z*mPpzCfbs_0ZN0NNTqCvT@E>X)B>`Zl&@7Lc<3
zX!bA9$G69WaKr=Ai8*2|VTvd$L23${H=s_C$=04lxS87RS3&N>>nYVeP3vKYSVWoO
zg$Fyzg4-g5Lu%h51na>^HvnVYp;ISFd<pecC}F`dlvP=mpt47Xk5*yk5jJ>0gRm#8
zVd?os2zV=nv{)JL^S+8ZD17|GBV#Jt4y6y*k2#VNq*MZhuJKdaqE(Uc_IH^*E}JiD
zQs5RTHr9tuhy0<`>~?*7H~RKL41(uQ9e<f&;OYWpg9xSg9}cjSo`acOqjL@3KyG0U
zP$$0Pc?N`#tg&mqR7D<(z!O0K2THxbUd<c7b8ow*=E20!l6R48Qs1`BhP#3M3r`A0
z#yN<GzOwVWcmExjV~G#AH9E?wH+F`!tE!uJWdQEXIuxxU3?|A%k|^0~l{v)1n78oX
zBFlZm^hsTI61F<wLRNQ})>{-1$(Ta|p=n$p<uh=VE|KXKy^|r(>0bD~NZuKs9x9vI
zJ;0B`!VtEh7?j+`b8DSX*nmkhGgBMiT!eS+k0?4W@oW(^BO}_ihAi^C>9eONwJ78p
z<QoIh$IUpZIRm&XObQA;iDn<4mdJ-A>5Mp3ZZeWKNCOun!^hY}o*@G65Xevjn6VdI
zzpI_)I+Vf(cKdXKw^SSw?$r<gGUV!M;K$Ns81?ikp?q7sueSDAOfRx*C=rMO80eCV
z#)S?$HyATsIeiNR%6P?c>ud?<$vqgUcA7Lk@oJf>jT5^{<8n!3IsUOPKwS%e{H*4A
z<N!JyK;^&QUk~EUT_o!;BV)q!O`tr-rMe}MAr?|gD+;DZ-5s!<va|ty7klFD$ihJB
z46v_kp}T0^SiDDx{uTMm!};`Il5JO&_zVsCMtE9SyMuy~+m9^&aUEme|9M@L;1ujg
z7dzFO&d?e}2jS6@jwo#LN5A8wQ*;7L>l(Ku@6>UDSIP+gu{Q84Kyjz&%4{?HZ5W9)
zC)-ur%lBQgmOy8}<z){jc)YrL-u9c$p{ZHhAZt$b+MZOO`4T<1nCQUutgX;TP~b%-
z$`QD~Aet>Mu8u8P70m))j;7~&T-^9{uJY1o&f#%8I=NjxguZjSJc)lhJT*StzW$oF
zPd=DfVomaPJAuvTzoJBTz{UD}vv;vPbNAt1O{jIilCtm9I4xkRlQZ3zVQEpTD(qC=
zX4Tb;?jDDT^h>lhL^hvAzqAJYi#c1@A(<Cz;W-4B;kq624|e46ZGM*0A})VxFRC6Q
zgAMVq9sZ+rFHYlD8ob$GKSs$v#6Oqu?g&8a3lHf<g#S=ak@gEN{Oy$lB?_ivd00c$
z_svtbg6=BtDIx+Cqpm)L3jbXzfB-^HHTskc6Z9KMj>bnn)Vl_mKq5vb7L3?pi1;Vq
zO8)Q`*+f8pebSvhGakahxj4TH07X3s4KXo9ALO^(xh9N2W?(|Ad5?cWE5QV%;9GD4
zcU&~rvR^`LAgp-7+!YzkmuJ$gZEvQ_vV3Y6Tt0VR1?bo;IF-+h^DRo}zI`R=o36GZ
zSqF6C={A>{h9JOu=j*>Dq$!TB4}Eo<oZ15@e3uHpvT;;`=%H%#zBbwAy-f4D6>#`)
zCD8kKB7Z=3Wb9i}g>Zy^#^=$CmMQN{U7~vyqIOa#C_@?jtvNkVRKX;;@PZ-+BaVoA
zkmtSE1ImXX5|RH!W6I`}SBcfG*`G>*VD2<`!%Tj|AD!gLYF#^Sl<7crB1gSH(5TJ3
zgnWa_hG>%V4TO0@>o`zcw1(cn?J17)J6j3x^3=6yw(`DR%c?&+mv3CD<S74;zo_&E
zH4Izhni{>$9RHG&qbV)d%Ezu3cTvaDZApCy1EhU|65Bds(*>mP%b75RjF^^582`3l
z%ws#6r++VEU*Qz^uL`e4^Ch%O+!=tQg_y<b|KgZOdLPF(0Lo`Pfv@_$vyM1xwdW_V
z<i`AgacRQ|1jF|H(*)q7J#YD|4-eTC%xD3;bKHLxaynD?#|nGN7iaZ+ebPY~zM?a*
zlia9Vslj_33<f&8ory+Ecp3MDP-w6)vHTWU0mAhRnFzB2XyI5Gg+Wgc7yxDd1r*Pg
z5P|xBUkZ&)zdcV@to|MHKFQMk!@48oMxuJz!jye}CpQV%9$+mvkskTdGgz&+@Tu=Y
zy!xsZ`DqE{GKmryjL*pgHgt27plkzxm(km^hplskX47ep?Szkh<LGXy=4k84Qp|m2
z^-}svbj#_51r%nQNTgV+chok4B*}d-VgMsoab)Eqd^9khnQFH)`j#(RN^Q-6;)<Te
z2q%MuSFj0d9?gWgFsp|NO?D_=A76!oO+&s$Sqt~G_jcY(HGd`w0Ufo*^$)-$2JTp}
zbo6n2Q5A(hpTXJm<abUzHDJ0~%s*e9k%zU9`A1+V<ke8_6Acy|jDo&B1%1rsxouLP
z+c0eTr)gft`*t84XS5%zg13d^ez3i<#`V#Hdj-X)BA<IgP<%e>05kOVR3rJCLg4II
z@|8re%WU*AaKFN1%Qaoc)E|h|{>sL~*=woF4->3vQAjSdo@sb>zm0UswKW}9tf!2+
z?dKqk1>VH2odXR*(ILC~mfbcmKU)kZb0D|jhMi>bf4nocs!e&ANr<5i{CoI&n^A3f
zPgx!ZPKL)^7oU*?(0tYLsYr8%CREbCjA4g~qBo}cQZl%;Zkegh(E{BZ8sJW3p1fGE
z>eX5|YMcZRJkF_F37wHbsR%hzt0F>yXU1f8i;+gn3%0K5KO=UkkgL;+Tt-;X+69W)
z_MJnGYaX`(A>F>kCFs0M%1?@2aa;Jgves56-Z0(RB_sBca_(mYCJ_{o(c_0n2gTXf
z9ainx6K_}9FZ!;s>jBd9+9N(4J9hZ%gqJj=Ze&!k(l{@66|2}fy#sF(SK&RFcSHiA
z)38-rwGiahqHO9Ft+~DiN_@#NI>}zg9D`x0KkqqcHWjAd(g_MdZcbqf{EFXR5N=M_
z;+$chU0}+HOIBXDTVcygGcTC;7Lf@x7D&2%?OFWaJP)Ri;Q%-Ao6|UbXbom<Oh}Cx
z;0a%VswH3;4Q1-)wC~wp{94u0Y@z;Co@YpIX`-xI`q|g>VDXF^GWbTrEjhrurFQh@
zH2jiaG`l1zarQlLpzS^Ohq@N*F`2F~7_kpqingYU486CZp&wh}0xWe6^XspDI1~9o
zUb)4))D_k|XTZPYdz(;~o`zyhhj{i8*t%Pc(>vrH=i#!(@u=ft87zW6e8>jnzCK*j
zwOSn&?Jxac;jE9QgyZcbjrJ$n<I{>7Bm^c$lq{CKIz85+32TiwmOo6{7Z={QjQOVM
zV*@j1<}vDV7m6HXdxNV&iiB`MQ(Fq>ow0APDqCuobwF$_7NR`~WVxtj({F=#r!{#V
z7OL4w`!(j8k!;ro6@gaNK*V#KHBAZ&F<k~ywK<S(nQw%1hOHNd9XjvOS;t-Ge?=~p
zXj+uX=*Cy$ydvDv5hf9NMc-)uAdl=XYP#|HwhkX&Oz)xpx;wmryGj!!Y)AMm8Hkd&
zo^veu(E%tS95`-$qCKyyghU?Ht<|s{UJk93g|dPFK;^xhK^W|H4MqYb%4iy*453ru
zQPep$&O&-p$LCP!u;Yp07NNw0jtNO&^%bTNS6g;NIKEUsy*;}@x!vE8-5MZ(S@!Xd
zD#Q-^$#$y&QT5AUriIzGU6X;CZS*mjPE?fadlOKISBa?P*9~IpU}z)b=-w*fi|S4l
zGx_ss@ssjp-#bXTT{8!QCDAyB#DKPj02~jB1HpZ`88PsPCcyweMY_EyXTd|hbslv6
zUw`)hac5CupR*fD@NciDrgSS@%^_plw%iX59=HynNJ}&~u47P$X`pb6064F$AqliI
zrU1MQeSWDeOo&(n4iapv<8Su2bJA0O_0S#`S9dAB%EY-48ca|xJ=>AwOZpI@Ox-JS
zBiqKaCz_mDm<aGRirZ0w>sXsD^XI}F4}Rg$zarEq7fF^1el+e_ps!*pQYjW!{tc%j
zBTcqg&iIHtZGi0T1RE-U188t2Fu0l$&e<6Vu2S@ss7*v!lV~ui<Urjgh+@B+Zv>zt
zP-~24x@A6rk^K2u31EOTMw-4x^fV$=;NZ)M|M$z2N3)WD&PO1a?dNAEgI_|ha?OMy
z-0LmNGHXa6AYfxNZxIYbA?y@M1-y%M^^8{!VcdY&6*k}^UBS5le$ddI_V!NHi5X6K
zC}3VnD9Q-ba+tS0KcfFTPfqr3&$zFERgqgvyAB+e`>}~$&tzcK_b2ypDH=GLfJEr?
z!B5yYRXz=-qE4;?{^0MgGujC04ULq8_uoPhfNfP++(_Znrp3aep&iU>&%xnxe_vq*
zhat4FZePr&Y)4@0le<07n{otkUpg=G5N}Ka2Zl2VK1oro=Mu*$Y%L)mdhzV922wJf
zJ#FHy%I_B$*Mu9ow2+Ewd|~j|);)(BTi0HbN3)lB`*rIiY<hv+&u_eGbUmYMXJJP1
zq|sJY`Qb$Z=XLeJv0!C%&dd~ZzM{~~f!2R|5M{c8x!r)kzK<BGt{7~nYV!?Q)RvSO
zL9`EwsfsbrRPd>Bw-t>lImCM!e5a1te{AX}?jVhhmR2dJ?zrVye|mUh7^WN#srRc5
z{veAqnSF}<75{rEu7q(-9JG>PrnAgmV_`7w=f=0n7FI|K`+(JZVDIkPdd$kfwY=l^
zr+uk<Mn)KD8*Osg*}kI;$0=DXuN*R}-7@W2jV23}z`hjkd+`zG2i}(PIys5mQZNqZ
zm>`A5a(bv9WG&K7&=noxDM!E7^&wdn6lxspMb98D{cT0zx8s@1P@7EaoJ2?*#Fz4=
z%tB01`DP&NTdDuu<{}!_zKtaD?J}s=^Xm)IhycTZ$P1&>B6%0+KQ~38WTE75IAuRp
zRGG3lf}Lh^6>}BCRcKDBIR2(ma=6zvJ;Avqg(F&=r22KfteOAQd+Zif4v@^2G#-5+
zyt=1evE;Db<AI(-N+!-^vHp}n*<wi@sALt$VwJjZ2Y#YSC;PaD=LcjU-?nTc<I9+8
zG6AWpPSPc*t2Xe*4|`m2FpfOFE1($T4jycy8G&^8i)d{X&%gR9i&+8By584A`5g^1
zQ~@p&WI67X2eP)N<FN_PJVAUl#hAX4$=<HH^X(zzXo6rcc1Ina{J}FEXg4;W%>Ygt
zP&bUV*(@SmY&SNeA_%tx;4d1<#a^+!6Hr22GUBBO3JM8<(bEq8^+h}d8R5Sp6h&+P
z0;IaNt+*RSvu<B0g1;0_LsPn$t21y>p#BlOQZg_ZNhnZ9gF<(zd3bzt|2<toWQ00t
z?2iytcsj$ndYJeN&UH0gRegVXm=Sll%Zze6u<f%!Z3C-urE9ujjQwxQ5rOc8vRSfI
z5d$@$YRH2He?psL$=cf!5!Vv?GKl^Yp$ylVqBKJn+0o4s3m3?+T0@=n_+Lo?;wz_A
zFVgE%?RY&#P)>H8zL#F<&G=Q`tA=YjX$5vYB6~z}mS?-(WCAixq6yvZm$2(P-})c;
zf}^&ePj00<IAw0rL>*z5T3`voCVd=J0GUr+$&fZa6zQ=1S)-D*twksnS%IICmLJ`|
zRA(!`zSM^NlQI)ZX#RR@r^xBCQA1eemkhRHjy=s@2IYx=6qd?Rct+WB0?X|EoT{?)
z3@`m3L7G3;3g)!O2Z}}x=d~2ROkRy5PI6C|%Qg8KHS_aBYVY)z_dwNxR`IanCBLIM
zWSB^Q(qG%MO3y@a8FMKaM8{`IqtM~*k;BHNAVKt9*FVb%w+b43*lL#$+nKpay>KLi
zj;10~mT}*h94}G6Xnoj56He3=T~O}{N*4Y(fz9#3wy?=T|FGs0@~NQRgY@hlai#y^
za)-mp?6IrA#2Rlz1_{JhlBQe{Hx!Lt$ili@XBcfe-I-b-Nl+1P%F{ZP9Uq6t63$th
zBoI|{Mpu;oPb9{JEqKWVGQOJSD1I1bwTX2g3<h4EKEs-Xj+gwV-b&SBJ!!gZ(s?ar
zzU@`KkHFtOBk8huI(#+WcB01-RN#YP`I<|Wt>MIPFS(=uu-FrPYE+%Y=v<A%>OTHk
zb<!9W!~*YCvHX`c`5<PP#gH8S1+5n*rz0|DW+4&JH%$#wt`&)$V$Yy2_`!Kokla<l
zPX4r*^SGt>ULx6}K&LUngp68SP^p5v<_~T_8M(Qdtm8e=<>|r?vmrv+ThX60T=)`7
zZk}T+&mrl7z?+Tr2IHL;@>uBVq;rOWH^RTkJRQV+OTc&=LKeHTn?2iYz@VD3Pd^Al
zIH|=%pD!Wdd!1FXN%7D2){_8?iexY;42oj%cCp!7g%V;Vh;ULV4B`ARc-UYuCEp`|
zF;w!176pu8ClNe03`gNFRo98trUNc?46|<}8Nc^c01-xX$KR`atjKk*?LWj<?NwAS
zIKE^J8?JtIE|j*~kjJ4!$-;R4cL|HnHrW|6i!ks@Z9oeQ>}ZY}_Ny|hF!1TO=H|x#
zK|v7Pb;OvTrCGX$UqK{t*|cp0d7MoO{y-}%4oK=(jVyjX;zcak3t__QiPgCdTsUdj
zxF2W&ZoW}b5*vtu1>s4&4c#E{IT?xG^&?Y_VY<aN`&@sX`M^8vj4V9pAOGU;AQ?9Z
z=iP|)h>W~yG1~R0&H7>Kx`UQqn~aJ^7hapC`tiDm1&bmDV=P@FZnuC$9JE%P<8(S-
zJk7GrM{bKFsD|oE;lDzk3Lk095~!yc-eC;_q~NqP97=18-8}HoESKG^Yd0^yhqgnn
z>F&B4fQdY`*5|D?#eS!f{IwSkA~yof%(8FhIF#B1t8_2QYg&V&AMWS&fKXGY1J(p?
z^Bu2iNz#q=56O^zX4c4Dt#nk&-iCk+#PU!j471fvngJ|BXoQ&6FVx|1%Vi2g0TG`B
zAh$VD>>C3YxXAI*_)QpSf20%q^>zx~0GFnCZcnP>a~a}*FuTtoT$2xBK3lFj94R#V
zJW5%#Z4Rgr{r#q-#y#pQL1@Ys`_H~-j;XlBz8_`RB%aGZx@>-ZyT89Huv<xu_9J$u
zoAOE68y5N)EXH>JHnL3!cnwem_2{FG0bYQho<IaFExB#T8`k@*FAs;Q(ihzwDgWrN
zA-fy|qmX<BNwfAn14peH=18@k9emmlf%wTql<q`!6v6m)2;z*EiVc%Iq05z+SKitu
z-im@|$dNvy7(vbMS=>nV4IRD%@$H<Uz7SKr(L;^a5$?>0OgeAOZp<FZTKled;K#Ez
z=p|X==}s$kE^@2B<B@(}nmPY={a@Ml4j)`s*K*h<oLrcP2(=-FJyfIQ36r7_0*8?(
zjg}(n5Q4Q``G0JI7o^*Ps3Syq<JcW~q6+%>ZG{8q&5t$K-V{0V$yvXX`Sb^eg^~3K
zF9T;Z)w{^RozS>yBj@DQT~06xfZ>_1M^duc@3h#>RhC)5xd4gJTO?dvE|_Rg3w0}%
z)~9qPmDLCJI8sPYYj2bCxmoz5G7&fiKe{}g&eGonMioYR9pk+fY$>`)k%^~CV<`ZC
z2nKUXZ#Z|lop*&N5hMPxyh8^cXx8T^Oa)lq!P`LnYm;paKBuokGEZO-JxK4$BoB+v
z;RsTJ)YE}*pKsJ7?iK$=gh*s_p6N*Uj_U65&hT#ZZhEFWTxqmUYD>pHbE7r|BSs*C
zoIhV|jtXZo{%YPXEn69dJuM}vNXmL1Et|AgKImBJvrG`B{T}JDn#14ifMRG?(i%@s
zhWz$=I{p1rvcl2r)&+Rm6URmyW1vqwlgA&~?G&!WnRUeu>@<5caW_ktn@0T<uOfgD
zpzxdLxI1@W_VPLy($D2y0Wq|{-h`-)<T{XE@eB=u{vBT^y83<-L6b4YW}|sok~TM`
z)0xz?gQ2~<PaGpX%-Ot6R}JZ)itWw-6EeHoNSqd6RsT&aBL{Ge{XN!@<7$^wu3X;M
zF;ZYxft~J8LV?bK4Ra*u=D5kf?oOFC>qRAhroo#`2+rDlAYHAa`0z$BuX&V+^5};n
zNB<Ps?2&mI$-|!<7eVPELt$*#yx3J*P|5m0*ziG0ip(^i{R^rkC_O}sYp<UVw~3SM
zS->>i0(NsaN*^E)JBtgh2*9VJ`{uc0&x~Alm5d+n@Z#nFH`FCHGUx42-1#ZLvi_Mp
zRY)7eX{mxrf99yGKpk}K!tk6NM|bmQ@snUMv+*#l^D3>&<QBoQGP$T-3+$Kh--~Jw
zbTYyfj7{AZ*6gjn(S#{$4Y85$GKbu>YUX)2(?5nqx^f_rFJp+8i6urL1<p;AA01NY
zzQN9-)_7qjB4x|%?~W0omR@A!;*Lyu8o#AOEu};#WoF8FvIM6V`tME?zOTQC-xHx*
z0atG>NzB(*k|<SmJ%n<31|0Bh0M?r@qx#gi*tTWQQmsS;aZ6oR_9-Y!><r@+N1Z82
zfAv%3w*Y&~b0Qk+a#ztWj9X+spQ=*>xNmiqmynlpMDr`Sohw2jlydwK_eh7P`pH6<
z3G?<5*|zs0-(cD)le!a$ZqgBx5*kQH%b6%8k*im0kVTTt*T#Y@D%^)F6;=vXnn$wQ
zi)(Lwfe1W>B;AHki9PcC^Ul==v&HKsrE5Bt$O8O3!hCCM>3dvm_!_Gs<dfnwULsoN
zG`^8%$leu+IaQ;oB&c_|7gnDgTe<Q}=p9EwCT-{*mn&BxLEW1p_?U7+WlGCD9<jWA
z{y@gE-+ILLL4$?NCZ%Y{<oTXCMVCZQ4%Q(u8n4dcaiwxH{t=3$y}h8G01~UA4f_w?
zZWHhpWT8siKB1LbZHEiRw74DjAt|dH$ga!5Q{VR6OF@sMBogi7imcN7TRG(}PgZ(n
z1^1xn?zKYO+*@JLuH*JgyaStkV5_*vDLc-ph0MIg=PkEr>FBS|tw-l(0ZR_&3bslc
zbc930wzaK+t@B!FtM#(t$RWp1CxT7(5oO><CNLp_8*MAG<p|>_&a8VBQHkp*ooZGC
z<9O)ruxoN8FE#%i<RBZA`s`6}b92ML3gSXQcB{#W!Jp2;-Qf|(+4g(5=}Pp_Lyyzt
zO(b#?&Ki?En9exIfybidTh(V(Gv`PDqT`cyy|H;*I$}qqB1A-m1;q?hdtiMUd;t)%
zR3Rgql6S3jXheUTB1QWH&nz5_wHz|@nJYwy&?=5xSe~*cKrola$^eQ%);H+e%Z6`=
zqGY#MY~1yrS=su9C^{kO+RG%`R##$@FJsRr2$Rzq{E4k|cOG7Ahne(-sDgmT_`Xah
z9@VtOSx4EogJnLBi>uh+i3f=k%oQLgp56O_fp-*1hQtRs4L$d~UXUe*zsx+3Q?mt9
zq+SJe+<~DNeFE70@vD3VMcbiKG?Hx9Q&yv$y=<T$jhXvQ6<+g1Zove-y2iEKwXZJP
zvXW4vaD8t?qFnBp)=D!5iz7|wr|7iGITiB)!t%q6d(&f|bXmA^@OhyBWg1`~!fysI
zFhz{JWScxOkr3M{5Cq!#9H4OB(lEPocw&shILY46`1?wU{Vr-t34)0<G7KW?jd;w3
zpja%G(Sh`StZ1(zWhlf3uU>9!iY~FP9*gn_lzwG2j{d___KR|Xfr)^s-@g$`xq!Jw
zWfMvR>2b!VsTsiAmmZn;)&&>@zrMGIVEle4U&bTj=Tn)*d0rbCgM;|VJ0MzJ_?yRJ
z657jNZedt%xg8_uluX3*Yl0zer_p}Wq^!E0#+T`ZDhl!~?VV5yHEZ-rRp6s<;Ntru
z@UgtnRaO7Fqqp}HrPeAdtQ{00h3{%-!ih9S5^g%&0JKnM2qNrItOJNg$wqgeF2BwU
zGceP{Rn-pY;TvyS&Fym{b(bFyNXMVndh>NwwG#TWS|hp5gyHCO;-d;cEJ&LSuZiYX
z2s3~|xwFbfbihZp8lGdlv4$}dS&QVu3;1<$xQt>UO8#dV<!sxpLRZ4x!2ZSqNk%Jv
zo11!qk>4E&4!j}L89;(l`C&>bx7Opdi;xfvYOEt&1Y2$Y8_AliFG-_CP9=Dp99P&y
z0_$;c?&@NsK;i&)>)mbU6rp$W>Vznl4iuHxyE(x_dAf3#7GiYLY@+1`HxAXZ+Qdp|
z4(Akkz?Odn<5(3g>x)-FUTiExR%iZBC}X2D8e{8KT}MNl0Pty8|6rDE4Ne~VSClt0
z)LQz?IEpAUz105G;w_1wum18}2A8DYw3*I=Cs)B@i(J&FWgeM1Ha}AfV}Jap$csWZ
zaFg*;AaMg1L4*MN$0(U%R`Z|^XsF6Ro=2WdM!{X)wY0hO1NwsOcI5Cj#nbO`Vx`S(
z*uiyQ{UK?+2S8pLjM^592n*FdR<XjszhB2Qak5D$5tuK;uA+>`+Fcjzn&G_JLj~S1
zQxXw9_Yms&Fsd+t5!49sV(G|;s$7Xl7qi`x#ft?J-HXQQ@H+#Wzq9XlFIB`ThlC})
z!G!+li&}26@g_@9RQ-8!h%*_zZGA{5T0@>CHU6m!S_$xChm`6-ELivMb@S2Iq@BID
z;r*5oFVXl3V-mlmHk2d`%GBKKwQ7Sp6xFy~to;(w-pkpiIZHk@<&zOK6w%JzG;5)H
zI(g@0Erfrj<}3VWVsrjbkewYEDe(ZI6qipE=SyE2HNWXrMD_keb$TiJ=jU6nVx!E9
z$o!YSY#g8k`t~z>68Kej(r2qXieSgU2^0H60c}dabrNl_F^pgKOw}~#6zPg4tMErX
zhk5Nf!W-LZGYMV@nc&hU{TqIrF9ULrXK*lh)y?=@xqBCCvOZZTy}_<^6AlSWFRscG
zE%hk14hfy3X&+_YzFWqEULhtjN%&t0k#thj?*sq^OU{k3;Z~<J;c8lIL=8-k+8Z5-
z23?HIXq#bRFeH@$0wzu;=roFEQz>oJ1DtbM*ApijBC)m*lsQ@)d>>(ViV&8-I3*wS
zbdjZKoRj@HW~%CLf=Lpej>57K(WDusu12<rin=u$8xFTBt5EE0{NkP|3Me^@A_oIo
z@Do7W^cot%TvjPjFHMGg`S8f9y-=Z#t}XI{aHZ#uz(kQOTsR-q7&6rU&*bC!i#;a%
zgSQS3o8gz81N6Mi-!5&xPs1f;ANw;jcB!Q=wITmD*O>}7XjSnpjYah`Fx`S;SDW)c
zs>UX}%~2z2beg5alFFlzeS(7OLHzGh&;tO*Y=Db#0Pb#OKV43!bCUFClr2s=Mzl?P
zqLgF(v+MC@BaBOy!K6DzunC6&5^2x9Fm{Gpo9UYbA6%X3n|rAOa)XY1LYP;4HQE~&
z@rPV`m`=Lry!Nw}12W^5)!;i#xJ_kR?t4sHNcDb{e;?L44Os-!3c-D}-F~F2{v+Tp
z*)}*bWanhgzFa4%m?MNP-!&3jV<qy;`@AsZn9a1ii=92P;$7r{st>_;fN^-`J`gUM
z={`VnZAevwiG-&&GDquki}T~EYQa2=baG(NFeA4n`n&WoA%Do{d2hS5hSUGW)j4)&
z)^ORnV!LA7wkuWz72CF*+;LK|ZQHhO+qR99=Xtw(^yx4A7wob3x~?_Hns93MRX~}S
zzIZ{jMw>E1=b=bo#&KmmB2K$)<^WqT{R?jX6O$k7ljKo@aBEeh@A9Cv8)Vq?^Tgkw
zh^0~XT6a~gVsGl7KvuE~Y~LVR`h5mMIK(uYIcdxapW5W-oqdQ$23$82NYW#ZS(x}{
zCON5wQm2O1&d=?&NmiN2(cueFGQ$df2Iba#iEH0QmE(jnug0j~B6tA^EP#PY!vP9-
zBz)+8=S5`}pG4g@5d!bg-Sx|y%l>n_&xUun_rMgP^ff;#fk2AGSnvn2x0B^$5v}+A
zko*4XKj;`GQbAw4%kKG>M~`XbbimQlp5q3!a`2mh`)olUs*{<Oi4Dnn2#YrN<53Ch
z=uvL<+stTeEO(N8vUyAS<^a;oyK&Uc&@E{Ef!>!#^aC5Pb~ciw0f#@o@a?;{Lni|0
z7XU(NY46_OEM_6lcta1|ertk^fmJ^Y&XKdGol?_a(W<*l$U4S4ku62st`<at=>%Vc
z9-i6Bao{Ac=nA?@*U864MYp1Zf79>>=l;53h7U5r=zyDzfmv<gFJJ|TFT6QQU`Hd>
z8)vgzv5iKtSIx?Sv{-b{aYe{)Z+v^#O016o?~O({Spdui*g-&|Dcf?IFm%KC6-(<J
zCN>s*N32E}^tnJ%hw8rIw~p{DQFId8OOwQ>RW#siU+4A5tH-Ou(bBWdt-@`jCdpQu
z#z@sL$Z*}T!H5oGAOH~}iqN-k1xvHI<zb8n56Nbs4;>jn1@2>6on628_8Ad!rtTTl
zg8w~KiO_q%9jQy7^si3Mpm3EcdM<i7`V7os%rV%R)cr2cs9WqE!4u~`DT!Jj>n3Dq
zHnAjSNBpy->LZdvxqnD5g~Bl9XFMi_i5#1R??ia=bu=1uIRGQ=$3z9M2Ybi{pc!db
z251Nbg`m7Z+upRf*p|o7>wDF`Ed>Rk!O5g4A3G#eo)B2u{{yc|SX@$IURNDUuwmwc
z`4`OO{%}Zgdq<&D;}h6MOAqr-cyE}c4Z&kzUQ-d25_Z+uMw3DzG%vW~^O3cL<c238
zcO|Pn#q9z6831OfxjZ@}!Ce)QoP1AM&9;?b17>?j!0^4<2U@ks^d5A12iq69S~zWr
zSntbM)X`tP)Dp<yWj?{`knF;=$TbU41ik%&Et8C;j<FQAt*^gGz)Xkh4)wucZ|s@f
zzcJGdjPhcNTo7(Jj|Imkls`84MG>gjq6f!Wt6@zT4rpww)F&bQ3!Fy6wb$pN-Hf3w
z6qg?)kI0FsA<RrJ8q5}AzF5{F>f{#l)GUMj^co-Hmx{V*?J8$c>wy7My<E**%+4;K
z{?-r@*;VT)KTxYyv%iJ`&!y5NFD?T{UHfn{jW^T2uDWW~%eGgkgtj)(G5%AZ2nnp{
z*J>#S7Qn-pFkhLG+11k_Osx(|h?Sb*THgmYZzv3i2@IVcJeeiIj2?WsO8GTm@^<&+
zd=%_;u=5Uu7<GnwoLV-Hw`q_C?Uke-Mi(s3$6f6jq3HnO!nMV+)Q>;0D-aXAM(0A&
zodyG2o;t^nrtu~E52i+6No*rLg$L{xI?~H3Az;mFT({=Hwa|Mw_$6`NkpFi5Lb12k
zp=mB~$n3y+$VP@pTOGIcf<_6^^5a0HGjdfI)WmyP4^);YuPxtR{y_Y4m<e+%ZZ<x<
zQJToHRyGLDVLH<}C~>|bG~?dt=~ZnsZgg5-CLEqMp~ARg;cs!5j`7(k*FR%}!E>B6
zz$&yE^vpuyzrYb;g*N*}=zDMf<6nSe^9r}u*_lqy)-KybYMY_RnP<m@TVR{C%r{IH
zuAm95t)EtUx;_xCgM7kREv*2(RavUC6gGP@iK4LR!VlxcG3xa&yurF|7gG*xiCOZM
z<rIOvAKqmnBTZfI!cA;D>H^#<MxPNui&NS>bF!f8PV~KyZa3agi+@Z`c9s&3vANDg
z@%AmCcyn@%3~7qh!BJSt4au>$T9x;<&M@YgneB0sZ$!B8OW8fu%|nPkk0w=lDzw}l
zlx=bx@04wNA=<ZQWq`_@=iXm#I{q9oFIucp%w>Eso7913J((2Fx1?&unEUF)RM*yK
zF&aSQO%Qgw7&GXzD_>1=<V$Uk?7*9VK2uEyeaw0kO46(VAp>M)_os<4=UElHsq9n}
zS$TV~3o9iqD?B>%QQ<`YaMjN_HyB1YnL!QXD&xILA2eBB`gxwyja}DnHkw@9Gv=hC
z+#mKw2-e%28_BeCCo+5I@y0(@KTqL_H}Cx8u)pzsj#m!06wbr}q^@vJYTRfnNRY2-
zr07hyYA4Bj_Z<L&F>C6`44fnsP3?E`y-iw}mTKkk7m4OZ2xlf2auGDo0%CXy7u6$w
ze*JuecO<oaQUh+vuz=E(=nHF`%63STV41TPXoSf0NQ(1>1`og_Um>U5_&>DR+!^?P
zX_2_$n-(LN0sqoskH&v!QC_wUP7ya+jdmMnoDuX^)t>~{fe8rh9Y|UYgxOT>mbiZU
z>5(~r>vh8;AiPRg;ljU*m)~DfsDo5_Y{s#i6K(qnv6c4>qQJ4sL>9qREj_^(1%xk9
z)$b=yU{D1-Y@3;0+Jx<d4jbpV8c0rpAlo&7khRcO0R$An+QLWblN=v63@sy^z#qYO
zhwHi59oLhQWTU%grbZ4u<!1e$h7M*hF4t*N;;_QL;*FaYe*IG9vVe&eg(K(2h)-bj
zA4;vsUJEAYcrQP;&jJ;ppHlutRo+GUUr1u>iZFMUcw!{m_%JCKis`lp3E!Yv0dSxE
zCe;JEly6eaoBJkJD)Py;jh6o+)eCGAS5I5^gbH&Z%K;@V?#MoKvDpQ?Xk_G{dT>|(
zaFRy~N-bOnNy64tP9RB174o9p|Hjv=%=1T|;qL>3tH_{-tR8J}qbh2B_U+X@prcss
zndbbyCC0;z!GnOhjy;$JdnblsdivX2V(-(ilxN_Rf@s(#G0yX(3oHvuVB7gph!~<+
zI3B(P<J0{9&3l8uR8-LAw2zWkRi>Rh^JC1O>|q>mpwRn-dZ5u|JY#4gmp>)6F4Sxa
z+$jRzTp1_v<8xnks^(v=bYkZ!;PW{*gm69PkTSgj#6>6rhByRJc6=BseU;SP&qvt>
zqQs!LK#hP3%m+a!W~8S>64ZD2w?Td}F?-Zqgz^Dp45>$5^jxc+*-`G)DC1c~=r0Qg
z8iN5p7G^<c%#}*GRkJR54*FM&hoT`gGW~)<XuQ?6c!dbJBkft-n=jZcDAvcI_+0&W
zJOuo&`H+xaKB1CGX(qIm|082&;G==1otvp7<^|Gq3+SFp$d<YN6@pG#2^?~(LfqxF
z<w3O}y;&~;Q$>4_Qt%?d5|iAd$cO;;bSCjtuWl@n8Ho~`&Q2X6pz$APDw99WwC+BZ
z5)luJ0@6xlx1DzpN!8Qd2T%`SxP!N7?^~5r&6BbmwhnIKkir&Wdu<Ac%hg=LQ4+~I
zWeOIHm7$HjH*%coh^_3gNpi`O_-hNE10POtfK)uWreEnJBx$h$nPHQZb%SfIdEp$T
zW=sD^OZ$=_V?Nx$oXGmY<yU^FJk>l_Z@0=wn0|pWCQkM~%G?=JA)u0}U?~cvbGN3n
zkEC1;iHcSSYApLMozG1jJ>8JIix^CD={HVA-?aM^*8vOYIB9xttSf~X0!JjaWaunb
zp>`J@mZUuy<whR=csAD>Ry{Y*k)M|4_HX&umhKhgRW6tK&lm*mwgz@)M4Zw{k62X;
zI`o$qNPT@P`(3Crcfe8640Y41KL;kk-)aKRQH;!>O>?Dw$=?=l;w>M6I3mPYUG-Nl
zxz91-A<elC48gosp(`}l1l8`Lrr0P03{Ix8sLM4ue>tc-A@@hRRwZM_>WK?~(E!(w
z7{j0eMtTGeGmf$$OUl!6)Z<FzpDBqBL8%QP2xQ!eT+V?xkpZ4RNT^8H^8FP;>~ZF)
zV=+)H$i(Ux$>~sRpV7m&HHDf2M5>RG2}M_l@1WUv1sS}-=viKiWq4iZ{`$lX86i})
z7%z7Z#qe++s#EzN=;-ek4vR+ZAXHc-%1b`-<i1x08YK-Q-63y2TFylpZBN}=ep)P>
z9f&F+8%R5baR4~|R&uTR8VsZRkd)^mSUpwgPa7j*qZxrDYYR2jg`Na^LOG#r^-Eg4
zdqQ#E=fPXnMft1tIOtQPZz8ia<bT2$Yp|GxU~oT7s`~^fp(q+6z2p#n%p~6wLE#zn
zHAbwgLRW|<EY+C(7Cc{fEs!fk1$rfCArt+JZef&@-U*P#<X?)XBljEDqv~%Gvm+Qx
zzD>@Q%hl^Z|MkoO=8cH3i?eo&lbY4vLLjBtE#xK;a3{BUBfe6^4}x%TBZr4@00y@K
zBSI$VR>c#Ke&UHg<F=F2-sE9k^?kMoYrK!<1PT{PkCW}71RA#hP7lAzgk5`FxXraX
z4C0>$AefU0@f^8LzXj`$Q`NY=?>A{30Y5Qpwv{p~b0riC^8zN31!#7@-zT+R^3M@a
z{G{pvn2~W~uz(*hliMjP(T{%NAbx#g5W?E7(SfLs1tK-k&UVnTztLa)8G~P~E9sIf
zHq6VtT#1S3ZHvO1YZu(}ri4X&M_903K});<;3zX8LM;pL3<}lB)7r`IBdW`HU@#ZQ
zjRL)g)WoUh(+yG;#Lj6*6<RoB=>d$+yCN|eAsWo1n=d8Gfikp<)rS4F2TC8o<|PJj
zhve5{Bgry%Ytn}(B#h?SehV`Y*``5YBR9gheEuf=X@lR|{?8RugFUr;&9E+<3zr1o
z1ka10FWZUl`L94;fjX&Ij+vX0-p;x0ep>J_J2iusq$_F4rwoOfx`e!4-@-wK)8co>
z(E~qcl1rUk8CfG1P4$ia#Cu(7*>eIfuEMg|Pd5;y#a)gSOKYUrA^if&L|XXni=(u+
zGQQ{amo#B4epv3MS=wgIXoQZmGx`j`3~tNK>Fp!DPBV_SPtWmT>behm1oS@jBgxzL
zB1Gh?k^HLL8OUbK9o$`FkTv9bjgiRr*<z9si3sniA)u`v>n`709k69`-6?F<7g2%H
zbVdyL%M}c*_9%-j5+EIm#GL&bh}~Yza4sD_ko^JTu1^`P4L^+&`1K8+E~iSsWlY{y
z5_pnO2}0}FwFa)P&e1P6WN_3!90`3SIiqX;d@$pdPy15$11fx&iirh#q{TR_=Rj=+
zHsKTYWs1H-D=h-Ur2-fdkrLTGIEF}EPbw~vzOKPw+t|S#$MD9^`K~ZL@Ej2DKe|08
zf=FtoaRx@T^T34^Xt~RuCAa`kca#i+B)A@f`YPn<cCm2^ojmuN?7?&e(=n|<8m87-
zeN7eWm{#j&Phg+ER}*XOcpjs~lfd^CK+#5xlD_v2kXsOX;DNRv9SQOp)Z;N`CHGgn
zO*z#aVAftXk8Lr(_thWb7S7!zbDQ}$+a1nIEwtHe*6C_%wKJ9s)PMn<J^VLa8&Asb
zuxrV0ysI9VTjvgYkJ6nB0qjF(v}k*PwyTQEZcn_6j`K*Dy}^kzXQ_Z`L@N2$C?m64
zqo`~?U$S)17<|2@fLoVaq1ip`nD#-1w4ZfUmwKYfNawX-a-sn+Ap6BcbRsw;bmCV%
zrhQ#I^8HA=ME7Vu@2dc?I%?MBRF-pZV`rNH%+5lH_dCZh5-=KHfhiKP>x&sJ*E-MN
z0kVHc6odza<3|@3cM8ytnjXu*H*VHT8llGgn2C-OYxF#`Mh>4eG}x>hi@3~=j@3v#
z)5|kFnwjH83$)GVM=yC$dal_E@zcMYahVX-VDAA;bbZ35xcdMYf@}KatYO+T9k-+U
zxw0@?^=*=2(rZ8?Yj7x+66&?-=pFgJ<Qc6hwo4l`6;h$o;arLX9fqb01EUD$HHRlF
z_K1y^S)+N`49!x%{R(E_mh7;n<!3(8{8GK%>k5QLal!XZcl@*j^CmnC-$Bz8=+mSf
z<tyUgyp$7=dN{zE3r(g~bee2oORk1r`2KtAkUR>Vm|MQlDfeWXyH|KZXh<*&5jF+*
z1IV$aHhI`pO>DJ6x$WSf59MZkVN#Ao6~XWxX!C4J>94a3!cYp5EM_77@$Jm%Hr^wy
z?TqTSJHd{0fxOwDdt6&-s!-h-Fc<O6?2i^~Vp$^LlPiFM;jC<Yxf-}S$o&*w?@nyN
z9F;4j-O}$IrV7<dx`AO#UB;V%FPh)i;FY#8QrCM%r<@KCd*;Ll`3F+h@~LC1U1%hw
zFAcD8bLlXf{B3?Ag_;}2obDA!k}u(<B$Q*p78mge&M(xDS<s$1$nfY<nDUz6-LG-g
zl7l4p<Jo|6jSzN2?BP_=WY8EWx=VU3vKpU0oDhv^Tmx}TLwV4tW0hu2#FvE$TGi|_
z8)tK#pq`FlUR;w$nbEl$`IUsy4C@R+p*pnT9c@H^&<Sa!rn<p+l5*2tv6Xb!be|T>
zQD)P;F8Vy#=i0J^zqGDFWS|O4Vr)ALGf(HO0$+eP<Gg+YD%hSkpz!>*w;2Yt__`Si
zKhk&ud85vBdHSV(_kk#d_ANviJ325s=p~4eGyku)E*uH3>w>vAizC5SFXrvYJY^*J
z3>g;9?gx=s&1~#e7lU*ucV|*BTmJJ6txZ@Pv%hzn9~t+yPHrwPq`cvH>+}eUAPtoa
z$5jAm0S8vy*U9h7skn{|^jOGy(1fcOA%*oary6B4wcHq5A6G7kb&vusxmS;)*RcgY
z8F-zqYRj{9xMWYD_ft3itluNlbV<KT0XqmDh2TVLsHt`Z;Dd~CPz4{P#!l&)zeukx
zf9>Kn2rk$3c9O8PDLL!j6>X%w3|BtD+4KNL)01o3PH?NIUMxa{6VCMWrsP0oWT~yP
z$~UG4v6f)GrID!*nnT;~z{c^`*OOYG(9x6n!q}PX{B>J4PvQc>pDevN(<c$QVQ&a6
zz)cU1D@f(_vJgDZ+N@A4HaCL|#3Jhr;U*Du&6CIa#7!?)wR!ZNdLnMtlu)_DKlA{B
z9DMw)*)_;*4a)qfoj4NqJ|^mWEXtGb-yJI1q1TRfPVn?(d0++vH*s^m4ox}waLux{
zB1>}2E6SOY&8(()o#RGc@~IfU5}wJkGA%^#`V{~a#sW09eoUb#oCI3W4=V4s!Af01
zdIlYiY$b#zY<*x(ye^>Jlc-*oVm82;JacX!-H~5@OLqW(MiHg526&}m;<mMiIfI4(
z|KXboPUBPTQP;S~vF7nFir#ybS<l`*Loq1(;VkdVHu#MA*eHG<E2`iK=@@$(!o?CL
zW1@tN<u|u7mBc06Mc1)9qT|8J)cvU_Xy%*Bd|3b01{XlBOmMq$yPkI}qX4jUJoVzo
zKK~#yT7&%~t-#dCPQ+t|ttH}&y~l|ZPOH@wd4*+ILN=)bTrhKN3`!ULgHl%Z9;$+I
zN~NTBD8r}ud~1-8L&Nc+?QENj$$w14*2~SqgR{Ncb;~QU8&bTlODobksJZ#KpE?JT
znU@YI$;WC^3#y!=o!DCPkrrU#t?jdIbMqNNN9fhOq(Y+X%x4W+#K)k%z$l|#i-_K-
z-`FXkLH0xUUqT+PB(|^phmc7JxCRl24d8a0ZViEHG@}F-yJ6K6-yv!5J3e4@BE6rC
zUvA5BJOd(wac3`QJ3Hn%5tzvSx_RatLV`q09!4u3wjtmpUS|OC6Q70I@X?uz*RG5r
z%+<=Dh)OpcXkc=_Qxiy}cchS7IijVf-<6C|`<N!X807TR8UpW1MCAuG-N8>PDugN4
zHIpG?p}84I_&2+~{ehH%dc7_2nyJk!c4nc1sUZYC5>ROrI0Owc^2Lqv$wM&Xt4I0K
z#28(`cX}yH@45i&phK(hzJchlolgB{s8-7AYp}4KKyFt{u8`5wP!z;>H)|{dxJl?A
zI?L5QlHN`u9a${i!t4-G&9PgrgD+;$^Ls0%tQPY0i<{08J=b-NpXiE~f-gu9nz=0K
zwP~6sDMO~HNZ25Q&B!-eP7wBuAnYyN75A2F2Y)TcvZVp=;i$fhLxY$mHq?Y_DuPwo
z7yU|s<nUZ6Bdx1=ud7I%>_6kD(7XkdW>-W|lirS)5)*UK#gSonc^4iJ>$ZvO5I>j@
z2%9d}8=3p(Bdv6}2_hD$QZ4mBvgX3Ylww#7=}kvjTge7VKov7s-vsk|zWci&k~#87
zEQt2F?ZN?Z!ES>`hJtM38;hj1pRLdgKW)jRU3Uu4e|Y_pojFeEuaPXXCB32BRFmE1
zZfx)WUG`OKta7;fmm=CHlQNldakpA03HlM+-V#<i?g)*uU#O0qd6I|wdEan+ypysO
z3NdrvT3xQ!8akZrX*?x9UntJyOsi%^-=>aLRTUc`D019iyE4i_n*7}l1_SrY_eJyS
z{T#0Z_v~CSXjN@XPTao82`%4;@jb)5!5;JhyN^e+_^}T`n0J%LtvVt&5X?nhJHb3d
zT~nj=uY6}r?hOO$I7E8$&E6hD%{<52=ng@WRNjX3${(V0Sy><cF4)Woqq$u8a#<+z
ztMq#SmZ|ICeAl4~rbH{ga3d)tyy!L=gVJO@nN(8fK!f-xUg#3N2T|t8d$X|*$=xia
zw-Z7So-P}S`15Xe!B!9E6lxS4Y(#DjGg7a3{YN|Is`BMigq5IfZ+Y765-!kYBTDtM
z=*EIE#d99ziC^rAV)xS%rULMWVTQ$V{#z^n6lW)Yqp`<cqA7)<XS=FxiBF&&jI2-R
zSnU}d!7-nPh<pFGz#EWWS6kQKbb(q%T`b0Kh=4q<_mL^Ex>2Z9%N62&=;?qs`di}j
zRJl(X?+`abvNCZijJT#H$^%p)eSP2Jf7AmEfsGf9{O{ok)Q6=pJ^Xdp?t$XFgFAo`
z!hJ!P)?>bgtqX{|3VRdsaX5*lb?7C?Z2ZB46@N50!Q1A|6!n63#Q-Zzm?dhe(!S3a
zyhu}(vcy7>Xy-MnNryTI5dDz#!p$~xWXo3i{9`Ymx0lFH5W{~juBA}z6BG`vwln%7
zyyvOpUL@|KL~qi&vxRo0$|0HlhY(;%+n2OUouFxYAqj_bhAS>1O~VIWP^GlfmGCOj
zQu@-CkK}CiI@ICs8bSJv#H;m(#KKoUf?R>=u^vP*bRfvB(iU;1QGNKqZp!hffm5yj
z7)2b@t1(akhVxl7Hj{(=7`k(tdoqCZ_I(2B*eGfYk$hecjO_<^25SJ^1STL_8uB;C
z(zHuH#yGRKLNQu$F<R?$C><Zaz{ZOFFxq&QVQ`d+bYo9dzC~O@5`vNkNQvfOD*Cu8
z(pc6&o2h|=Kemb6fvI3X2<lAO^(4NTn!Qq@m4PV6CU`TP4Iwi7nJ798g#spX#`wc6
z$_d$%%i8e8DpYVaqtOH!VVnR;hvziQ(~zn1cB{_A7>s=w5#16jnf9Cm2<X5n4YZY{
zx0US~UFHuU42OEn-=aMOb!5WNK&6Maz<7c`yOQM?y|eivQJ$~ihNlG2r6r5WP9vIA
zClv9sriY?PzJpjx!d_9&IC(OE0zjNKQLa(sDq=gQS%S_`Dl85`OF{q`5&i5@sR>v~
z9R@ia8GWK><_F$O9vM5=hm#UYZOvWlqXuW|mG&F0@>LE|mu+vJZZMT}!A}5)rmyev
z@JfL%@QqOX`<6R)D9z1|Dm7e{i{CZqQ1G~7S=YmYvBpt0qIEdhu5~{0uy4G7eiE;S
z%gC9#SpxZXbX@&zI(7y~Rrqso8&<LrcFneCzbqsOF802x6<v26--P=rBKI?c>knh-
z&a;VBP%m1>xp<c<<tUsn=a2js@gl=n|1NsD*gJ=cO*XtvQuBI}n1Ma=WP`ujC16{y
zUeoe_Oge|0YP{L*t^xn;2Li$LLl#^L@h*?tU&MOy#di^q&$tB-%tzi@{d-oK9$}^<
zWu^tsAf(8R3w7*9#`_oD-tNc~iHwu1AAYofviTDq!4U&&l=Gb_`}ZRna<3DbAAe})
z;JOx?UQJ6buycBWAFV!mtH-^akgb&kH_-tYGym$nx)1M^r7Ob)*xMY$7r&WDz&RPq
zSR1VLk_A9HsnBE-!3upu6eXR3XYfp)5PiF>A6<3Y6rdbncg+vqXBEj8Hp~c%<g8N*
zHbZ!k=GLj2PJ?YlBq5hhd2Og*x~F%@u|D;iq}VUE8-?&Fod=}v?W~few_uJLd{4cR
ztM3`9n9s#pKqWTl{#}><L;;`)Mn2Y3cD@c9N&=8!dkira^s3*@KH?~{dD*>9PR36+
z#nxkq4bDn71jX|`Ovd+nv6E3uZ%1dFNc%@o{AD|G8a2Gh2yzMY%2P%U;1bf8eepRl
zm!y9@JjNUfxPs2gl;nHvR;ZgXTOjVi5gI4tXWKXEd*+LqxIz{X&8T-sXk`w$u)*=!
z1u)HEZijr3mdySP2x01#HQxj`Jp|A2KeF<+=rW(ioysszlOw_HQDM4Z`mCT$YaDw8
zrEJfdndDu^l+2T_Vlg)g7M^nn5k3c=En@$?+emKU&p|pySxNU1Q^@2m32o}gs0xWi
zx?oOA-+BJoqOcDz2LMuj7T7n4pf4kh01VymQ%Dxj8Vzd?#_#ZUo2c=3t1r4w1OYyE
zEg!GHgxPd{aQCN`2U+#;YX5>iiN#lN+Y(xZsz^_0sb}U9`I29+p_FUN8CmUjTM*R*
zS&X1B)cd`~>Us|~nz&J;xZ)D=PsBO^&YfFq+N>15#=Q0lv6=I>vdz<_!87;;0Ltq|
zr#xk(<;FBFm7m=2-Jx;yRP8sY)BV?cZ!E#vYTA-*SZ(ozbJn}rwDZ<)h-UlC{+F8w
zK30W$izw`mP7AKUP}&&}MmJg@F9_RdpH3k(@$dR_cXrV$s*0euvA;Q<j6jQV4LF=5
z^OIVW`(lvhlW?QgLtswj;AF2@09ilJ%Vzcc(S#h&rs~|K3I`S=*+yP|e(t(>a*ntP
z<Vx&-&c2T@BWy1g{Hb)zbTbnpf#R$QLnSf5>akvBFs?-;g@8eRUFX|X*Hcv`Zz=s6
z)30%-?!3cNefr;qOP~1eE_^!hFdZIGbNBc<iePi${MjY8)ySIVozras08H5=4}_1v
zaEM<$E1ByHq;o8pS=l@_>zq$~fu8r_`sF2ezNKc*^xIF}Ji&F0!g+$4w93QzM3o)2
zmPfc5i(m0+xB{jLZTKFnIZriIEcsCx$_+O)gyIpUa#o?dHBUP479=n*Fm(y#4@ae8
zHL4q_<0+-dK}NqXV)e%-0pTZ#O4-~Qrc5F|JG4X_^D0Cj_BtboQxvGK<%uQX)9>jq
zPs3fc3FPn2p&?F{SruLa#>T9e&vAi{xETC8*eIY8?bUEhnDhz`K{_)(dSH4k7ck*7
zGPP>G8ed=5>(60K$A5VHlVL}M@MwSf9%nbTHf@i!`VNVFh7|fA0h}wf1)@muF;lth
zkVqV$JW?f0M!a)=Z?~Mm^7j&A#2vSbe8pWYl9+xlqw16r{RP%CmpLu7b0O}y1m9Db
zKU*3@0^4A&7M31-3Z?i$tbk)*@}&!(Ot~>{`6FeM6J-^;nAF-PK@Ef2A9fy%X6x-o
zW=%|hs=><E$hbvc37DMNCtTYdUBttDyKN-~bioCEkJZ?2n>b$zJza(+UqK;I!jnKn
z!$74KdiO2R!#p5z<w?8RrGa@9P0?)K&L)1+{0t{TsqN6Ve&0i`_m7ORGebfd2BZX`
z-vrL=szw_tIvEy0aG6SPP{Tg1_oXtcK;!cxpqMTOl6<^n-``(LysNR&R->_2;2UpR
zc-FEx6~`z`kd16vh3UQ_#BK5H%yTocO6J%l%)GES3p3ZG4f|C_;0qbB2S_g3oYkBW
z`QO=DE$+#N5+0qL?=*h>_Cu-UEAgn<AGl*|G3RY$=HRq2HfB||A<d$@SR*>bZYAXj
zXAgruxKt`JfF=bwr6brgDlbO=!qRQ6r-9idGVMn+cTIk~kN(`DfIdG>xSA8Dm<{u4
zIKm|nO>`p#bjG7oJN77`==TktT24u}@!L}P!=O;{?VV|mdm8TW>;I^PXpK3`yz>=a
zD2TSL3dL3}{_n<G#Oe$w0bPd}8SVV9t51892S87f%vDtej9FAhBn8k^Zk6U4w7<9J
zNyoa%=8zHNJ@2BYrk#<X9o%JbN84E14lPEwb)~@7r;K}ZYkON;ES@OBWs+u@UBzIl
zavv*1JbAI_6u$b1)F}&S$?nizQS0k%2EMcEE=cs}X3M92L)6;g7}9}qg$~&$)fsg|
zAix8N{b7FFW4S+tB}gK#-Jqnd1supT=H@JoGlc#OQw2YTB__>(3volcg^=l|%%<CY
zEH0J!9riSM%MM&KhEt~}Uvm!^yDGV5b!L_#wqMKpa0Dv5vWHO@W6>2m^cZ9LEzGtS
z@YsUaGL4D$##pK-Tbj|G9ixB9kIILwO2ESk#&r*Lr;e4Qhf(ErRQ7fq*!Uz9!Fg`I
zx?KjarbsM6+aCV5E=>JJxZ7URl7U8BxUpXWsq%pd^W2Zjc)VbH&G#x*kVkKl!0RX&
zSM>SEtNhdF?ykLS3~=K1pSry|VK&(s4`<}=Dk=WnhQaNn4Xn;N+NXZLGYWT4EkKp@
zn%H(rt4YWiKTTLnZ0QMvxl4_P#Y5i7Y2wIb>wi3iwk>Bt0OtUuQCfIslyASR!%A%q
zBC_SbE<)+M=3sKlP{WOivbOMGXH0XEykKN7C5%=5?+KMkKM*a>2zUV+lZ=0O&QeQj
z!uM3CGhnl|j9N}Kd;kLh*w?e|A|#96yTK+%{iN`Oz@u5d;AX&}Vc5ZkSqQLr7nmJG
zB1ut#&O9*`pcLlS4KL#|Dr9_f&e4WDje1n@)+<YO^5zo>|GA6J_PSu5#e|Rj)7pxx
zn)&_!DWP0I9+cTYQ1EwU<@$AcvY(m7F&z&^x^bR**-!%SPY_lhKu?V`Hl#}Vul>5B
zIx*hcf1=)R-fYWn(xZM4|M3yE)en}|*@2xGpoZzyqT7Qv;QybG@P*PGiN@$4z3yL{
zrs8Tt%&d}8GlA~Uj;ptvX}DsG@c_ru-1IveTU#630Ye*Wn?l9_GJ9?so0&jF=adCZ
ztBU3;ZlQmNZ(<CsbFh@gm;c$_3x57j(wk#tim_B224Yj00vJb!euskvYW)(~t%R7Z
zY&KS7eHy+6W2~LDTKXO6t+y0@xE#$E>gJ!_J@G>S!3e18H3W;%1TB+w&eeqpq|<#H
z`&g2R?-8*h4_EwS!Vmy}(BHmVE;YgAkbGz5=irh?(Cb=VS3{ogG8q$0>Uk7Q%4avF
z-;u;xYE-j%R%1o@Q;3-%*RT^bhk82bDTYP?*ljT7yL%hvUKtUP0BON#hrI5^ybq~b
z&b%9wYiiOj*51rurxz-hEbd3~PwTs8Dp52Uk+`jVFBx>OcLgBsCzVf|Assp-3I`rN
zgGH8qV4yPmo{BDCc37C0=S>eA?8QaVV+)2oBgr`U#=X(A-@vpWPSTbmV!{F{Il=SY
zI?EtdI;xKY&a%hbX{-l-r=JZxvMp)XL&V)2LCH|vjQL`|HZ=^xyJh&-!`I3i+C_g1
z$&Aj3)Wf(hCNw~*{X+c>53lz^E2(TTDuuVHEh-A}@1LY-7Wv(w$NR3W@OW3q=xz)s
z`BAP7B_dMcL2f1{B4mpI0q<YDNfhIO_hf_M*U1YMMSSCVU;d6whBC1JCjo-9V%<ds
z>}H|RvOtRJ@=fHVIX`ySne<vS-bPbi=v?Y&WgsTNsr~>^&Lj~bT7E$kgvp`L7ceQJ
zQ~B6W&HGlT2l9-r)Ob0LkDUZkDJN>}p;l%<pjbg4!}Q)4-C`82vzS;l0hZvz3TBHd
zWC7z?5lPpx3D`zgwxN4u@2jNqGqT(yW4(81hivSqkHz&bKBC&qwB<!eF7GZ#)(Rha
zSb-uDvP^*3n(`Yu=Nyet)r-rX+0cc{o9kxP3wx1D+Q>SzF)h7xX5I9G+o!yG?cOle
z&q2<MaT<V0hjm^>yy_PjExTCF=R?TWxnu-5fE5AA?Ju(dg`&h;$Q}+4zzVm5z!l7i
zm0XZ@>u`h(teC2;%S&z+>VqoB1v2?uM*^p`s1_iHjK7;P;I}9=CN*AN`NW)C>dKt^
zYmqpJaX?UJm!28yO#^8L6SMS!X=|{|P=@nP_*%kve7y1C6zWH2G8Cka%B6ZgA<x}7
z_J$US#Y5vd&q38Sfu<_9yomXLylXcniR~I`(=v`O1VU$-Z1aWg1H>Cqd~36#Io9M6
zmkEFzL6NY9Jc)`;rZw>pSIiB$Fx8h)eZl<TN2uEs1`&D^mz!td^jYEwAM%;q?ymT7
znPi1hSlKn<pW!6v6ujWXJIDj+sjRF6`8~se*}rf})_JszRoc&DvLQwMFl8&Hd+wDN
zslRLy$uqz+w22}Rca1TVIHJ5378=A(hdlr_wz|o`>x6S`L+*m$flrg*!ZYn9O*Af#
zkiT-{W(;0kz1Kb`PY5l79G~n*45&ERNJ^u>T}$-b*RHX)67g%DBm)ughDeaLe*SDA
zhgR7%5|52)G{i#qZ;Qp@P`=+ur{Yd(_z;VDGiB&m{M*P0zcnBEydbMQgoqz7^Y{Sw
z;`dU0C-gH$-GadVl^UbD5qN1h1mjsG_$ZG=;M1ZuI<&Lq@*<q5<d-zIz3T7R{dfjS
zMOb}U=kAaS6$&=W&torpGMu1=9ptBHh1$7Nd8SKwqzTya%w(C7K&8W$zo4x)>m}ip
zArj5n3F&nL(i{{T!PruBBj!Y_u$KV75jUokr@)-5b(tr!zE{GSNynwWnK%o0-+{#3
z0IfObAbdn9XHDA!V>p|;U()gvIgPhdixu{VZZWB>c>Z%mSZGe<wAg9$P{NP8A?1Xv
zp3Q>viaF}^*JP)1ulR}}&mR%Ikr4M0D1M^LJtBtOXgydt{6n5{zi5+zPelRdRL>=)
zH@TfX^sZo>pKt9?=jQaK_Dw6hliKGR&V5>t2>hPEkV<`jO_dh9P2Pfq9xHhj-208E
zHmxkEIu*Y<Rc9oH6{}O^@&S2`Uv{j5R5CZ%sLx;}vlG$dP9?aT$y)bxV-qh2i8OPT
z+Bv-{E(}wmOb`*-G^$usyZ->_`c)5XqXuT;33{dZUakhc9&*z$!?tV47F3E`<xT!E
zY}xqx(Ej-J8)f)G9oRDw{5hp^Jd=Rzv|>?ri!LjzS*G$j>>WE7B#hXID>j?Vom`H)
zVmva61qk}*^j}$&H3B}c<=EA)-&qvdU0`CM3z2^2cu?gx1i#p>n0amiO%VYZf1Do+
zT+Y<_GwS}(t8gfFU)5}2qt<v*C7k`O-}9pXYEK9;5<A;FwK={L7^x(qUg7R-j0?c0
zFF?E2QJb0i6(hn}>H5k^4-s5;ALf@hBt9lm!AoBFgoy`be&MtaKkM;IHY4eNW;E>g
zcxcf3Yw4pt_Hm*?GpKI~@YD-ocHVrj7mR=S!+E&NJpz(oSMrN-o{uft7igWD2(Ir!
ztw|(@M!te(QzxEec&v(cxnWMf+iewwIDZBMC&{73BEZR*5Lnaxhwc*zF!jxVP+tNl
z67ODTd#q0VLpC(q!WBE$!JIX{jOOv1lFRA#lLPGtZYl4S&T6v|U;tMl_IK6l)qK;1
z92$un*L`pkq<FOBoA#ll_0cGH4qd<fxfTV;y6sWb)he-mmR%|f*`ghKk#9&vR!)b*
z>%!+Ls}I=hAaYHj1x}nbqG$Cm1h9=*B>P)rJEkjFmPZDM>Z1j-m!%cumDkuSqjCnm
zi?9=<mo|R-Q7Q5}fXi`>cZzpYL;YWvPdIs70<OKF_qX*{fA!DIEOdhIaVR>PQ^3f2
zj=R$)$54cHCIfBb*?WeJF|xi09Ue^+W-DLyq)L(6ZT|E4=I}0K!i(ZTQRnUMUatg#
zQmePo!Z8r$3G`wPY+p8Jy1nz0%Ss(y{i~BT)Su7*TMz)PjZVQO|MR0#`lwJ31-@_x
z){febZp^+8Th$?T(PgMMi0~>utNJuHv8>FZBg{9iQ!ai+DosDWAs*wHt_8xYzjC#%
z>}&-!I;wLevCO2jB!cEB^c1FrOlj-^zSzFm7lO-oN4;(JtiIGKxz;O42g(BLTJk}v
zK)o_U*#RK#!9A65&?@}JFg{%rBvg(x;E`A&3hEgn<l~2P`ZgNN$5~x}1rO{*n&dD%
zccfzQ07d+|;vwAT)_I>^fcMi8c=<HtimxWdLh5bDPn&&%IW!-M@Ns42Z&}2(s*{vl
zs`wQgtw}lZ>x>)MjU5}C%xJ74zpuN?>zXpeG6JAUGi@?sa;VSY{t9rfg08&~_<hw@
z<MNtHV`;cQ$M~H8cp-&14Xxw#dTspZsui4}4<ZygOMNz8b*rNYDR1!}^9q~lcQ^4Q
zmHm-Wq(Mh`HJ5D>2SR6{7wmYh&9SpAJM1%9v*;uy-V<X6>VZMaXtd<Z{VM5aLhvI0
z=ma2<BqMPxi<`CNu7!#<eyU>5e9u8h9!sfYHJ;o4NcNfS;PWqxWdW9!M%JuqWhH?_
zxkcoGx{&;A`ORNeoKvjK+&ChIM9)q2YlP*1!mnD&Ojo&XO)asuf@e;>rFj&hRY<zv
zpQ?a&!yWLxaU9I+kpfAG2c1rJ5J@a2K^j0%_)-pY5a0xKhB|K}jUN|*_s^^+(^u<7
z^FFeM0Vw*<6-UwYhg0UjJ`cxNNZ2&A@vx5&J#cWDb|f}Hat%P@%j5|SRR5-qJR&pF
z0L!-Ul?PwE`Bncf$@(Bj4#m#l*jwgxuRX1M2HBWnPKx;64F|jZ2kCy$g0)(y95{ei
zT~&pWgL^Z=>CavvnHwr%U5K|26^9sS5-eLfRS#}TEZ@A^eYB)?FU@*{x{6L5=}nGI
z`Y7FBt0DJAh-sYRY8-TRRd~5CE3Dd|^Rny@9g!NK3Q>Lp8-d1}<(%<mGu-fgba_Ej
z*!OA)yh2wsR|3@3rb?_vnEb7@s$_rx8kEI77x5Q^)-j+QLa$;ADh?ZKq!x@Ut82Nk
z2GNd0ahNM!3Hbh~^}09|{6pj!2pd~-C}Ura1;SQiT^IK^=Inmjh~Z^KJcUu-!nLvB
zs?mv$3_1TR^825tnewI?k_%{0v=Q*LU!Jn(mZBCHj)<lUY<9xgBh9jaek?#P`B3UI
z-NR6_#URmzAc)a#5D^^uB)$z$JJy`c$TU$^lQMC%H$y4Vq~gP0PT<iTCtx_C6k-tG
zu1!18uL$&;g?*et04$^{Bxk=GMgV~tDf#<}A?7hktI{RDwauoqkt~R##%)P^5n2P$
zbTrFJPmJ3jR!y=$b8pq@4-3GnDr+O!hjp90S<})CziaT!7Cv-d7vjgh$X6GBrF8<`
z>*ndZ%9;F%zC?$tU(rlXMMHtMiXMT(IkH_xe&aZ8<Q`HeHEL+tlkHqf*Q>KbuzS;1
zXYOQQ$E~ZE697=bv&ko816@H@k9DPfu?K;)m|yC-syVc$rd3cefB@j9xq+hmntOfy
zaK~=4q7|zjJ7AheKKCb%)Hsta1le?c6I<>V!y3wVYd4S%;U%oT)Zmd-vX!T4Jy=dQ
zhWUg{G(41cj2vxC#mh0REj8`MVQdUMG^q<HP%T2bO+sReIqjwFrCIKTb%0je&h-U<
zwGF{lhHF_%^b3YBSO7%F^WtXH_)Yx9if7>MmJyDd!V)pGAT}-1%3_<YMD<l;Yy9)z
z=7%jDurPTI(yc6w)l4iUNXk$H%2V>q2%?bzt(k-T4;K|_wy}Y6I4qHLYz!I+vMML6
zOt;iv@4k+0)940HK9t2Xt+WF)K9Ad6F!U+-YlQ8FJESo5Y5>;}%<EV{$%Q9qscn2<
z9!>7ZkCrQZ|1^$?v&co;6G#deSywk2=1c~i^Ntm%_4DPbm0+uJkIm}QNL=`Q)WmyS
zd587iWHcw*w3KB5y@d?<Y4O3!5vYr-2a6A;tXrbfDQHrdoGwoek0qtE$*<#D+h-`J
zR2RWxV`XV$E`YiK2E}v*{y<WT`DO85xUp1hQ8%`F7pwEs-#=-|q|eMEz_{V{#mth{
z8s@AsxZxqQS!Fa1_ks~91g%!$6(qYUhtC!=!kxl^`e7F3MAcWg4%2MWOEs7rllF(G
zW<*J)S;Kj}8<ZU6uvBxX>3!;zDu+9;#G+g0e?7>Hb^sP1GUO@VZ^Pc#Y~G=jkq&MH
zHQi-d(7i$Kuo*<Y_&gzy!2Uksa)kSFJnNv~wgyZ+`#(TtlSov^LuSW86jWb9foMGt
z4v95lEPE2=y$2izc{sD>hu8P8TRN6qSZj8S?fEd|MEG+Q@|`iqvc>+KS;o>0Y547Q
z5PQ^*ngUQR<Re8uELtgEJoWWgh{8^d-VMHUAkn%L%vHkem{^gPZkZ`{Sz6iN&LD=_
zyo8Y+S>EKGiQBHRnhE5)0b%l1CB7f7A&tdlwLK%1Yw7rK4y&{lyaSE}vinLTK-8hG
zyu2$44lub>b}J#$SLJ5u++K|hsiO-%K85${Xa%&jwYGrPw124yG>Nyo3@uzVVRImp
z3K+DI7rfRSFWg{DT%wh}g<_(NP0G^lR~=RzzRcod<fXbAE0&<adzH%xBkIr#;G}|C
z=Lu_@jZBm>?IoI-2Ez^dH9ad}Pm^B<I-xL@qisXB@QIplb=#J}gAWf(E&qXI8LTYh
z(*q=JlG9_P-FsXmR&@j0L`?rZp026l`CN}VTrk!$yDY)^Wr;$<u1C<=Bwan^THRGs
z6hAQe$A@%rH<+C7Pe##5GHd)ZhK1soeaGC+nxN~1qNk^Xyhf>n!uqAp#WoXZyoqPO
z9lUmNli*8Pzv@_2+&Yh@jty5ji%d9ZzA2y!LycHlFgbwX;!iT9H>qH101K4aYA_3<
zK(9Y?G6m#EYsycmMOW9xUl`lo0N`>jHg4lhw|xINatu$BM5A+cLaKg-GksUzqLZ;_
z)c)xLoFpA*#D<p;e8l$==kiY9Auv!7DKxpUso=t?K1d!MvTb$mgUGK1h7t(NbT)v6
z<Pi&;;q%DpS0rmM0e^O;)XziQQjwNfIlCa%_{wz(4>3*Lt8?FG3!~@bB3$IdC&vX(
z3x{lECuxfzjGnbSx(dt+3)h|6hkYtCiDV9-8Q6tDtf+Kr<oAFKO(ZwU)M+L?)KLOB
z?g+^;oF$~~8h^fOviT{B66=xRj(h-C(1TwNSseGN_WBkdg^1y95we%DpV|iec~Wfv
zdR1b$Q+x7J=g(>MfRjGAaF<I$Cv{G%yQ6*bMxkkLTK0YGbl}>f4WUeju|7Xj7z^lS
zNfUYB&(g>X`ZO$Z_R>&IUiDL}zdts>7NUc3px>*YP+ge+>@}EF5iL{tdc*=q7Y5kp
z6t<ExJ9<SYltt(q>g%vmxxQn~4Vsnnd$&D%%iY>VV{E3No8oSNR717;$y7&faxRHf
z@Xvxye@(6evAW{i`|-3Xk?m&~en1I;lw;j{acRNA!%4xhv2pz6&QFZA+#qmpXu^Xm
z@z`0qJyO+@^1F*tl3a>E!l4FGI+qUY#M7Caz^&aD_6UL&IeEPN5Mjn~QGR{3b%|Pq
zHqS}3VL$$&Y}zX3YF433SVMkQYf+kB;B9T?7E&BfUIf=stAiD4&{BqqP?pMmA$x!i
zriX?IBo5jl!~CI{pyj{9*9Eh=9nDTOCrMyTa0|34@RY3?;_0JWl#%eAKR<THzOW66
z_sJ(6tkrt}^B7S7zH><TS_C}<SfK=#^KMH<KpfMqEvDyHXVf-I(K?e#(<+(SuIgm*
z>E)=a6Pl=s;+X6K$yAHejE0v}iO4YToZy`XK1cLHNtHQfx30nymB2aW0m&iZ>*o%e
z5Q3+rXY6a;j0!(Zjzt21)#5Q~|8&5+eK9K31;B{1byPA^nq)%qjfAJHG=5jB_>da|
z_Y#A*p^ZaN@!W~JEQx4!(6Y;NPJO{|Hxer4E7OQ%rV9a|-$1oE0{`nYzLmyB^M5Ig
z!PL=aay@8_;jL0-Y6!V5#_b(>@~Ekw?vNS0{|F7h{}_#UmT#jW*@yp+(E!`~Z=>Oy
z4ws}nYgIT5MS=7=gR(m{rJvV$+MEf}Mi!0VFHs~=lq9}rzD~1;d{k4_+WIKI;bcP(
zf<ZP%^M?Xww|&&#r?q5|>>z%!aeayuFE#{!Zs=|;GdTvY)>zzyo;?O24mK3j56@JQ
zk6`caxISKHE(}nH5;e?k8=jU9S8QD`v_R$ZJEG0LE${5BH?9yk>2nl_R7`Rj)Q2nT
z7%zPRq5Ct^Z*(;meBH{<BA>;0Dzs$QwF3V^+6t>GuDilYE6YM=90wBc61Q*AtKI(t
z3+1Zpq+GLdDIswi^pgY7Bd29Ar3S?9dadQvYXwopzl`oHHdrdN5X8FtqG{P8UuBEe
z9my<n@y{>M2Sil+3mN^3ntmy?c($0<Fel~sN*qnXIAt&Np>s`~`L-Lin<-v`V>DVT
z4IfORHk!6SmXY0wa$BbIjMe+^9XVU^S-Ut<B8A3tp1elK!4Y8cbEX{3W$~F@56LVV
zibf((XP=qb)!TjD9HX8t+sB4>9BJ(P{6^Dero0RGPd9s=II7VT???k#UqzCj{{@-N
z<GpkaFXtMIQH%@c{}QJBKZI#%2jm2Q)H38DWg%CvQT(UlV#n>`X9{Fsf7&}gcCIx5
zhW=J!C2P-x>nx>OF#F(yl4=m|;bM#2y1HWsUiSbdMBw4x5F3rM%BjCyXEwOCu=Y4^
zpjS-XD-a`LbqL&OhX_kIjSh{_Qy4!7){y3jHWOJw&we{@m^=@$+Z9wU`<Q{^RV!^$
zx{s^}-X=t2h9qH5vQ0%}-lFhfqoy37H_UxPUf!C2aC)jPX0%;Q|I_Gxwh+T8=F;~v
z)<e$Wyp*@CHB$@d<v7i%#l0|fb=bbL7&*p6zG}|87bw;i1i=xwFCmhoA;BJ@&@yXB
zrB^?t$_E1Kb3gh~l%GY$;u<JBTNW-@l{oZ?X093XnJD-N80Y+x6N8t%uGtpg`#$hF
zOE3?%1FLct1J?uov~?yu?zeqXpUP<hGJQBV)7xMD-aTi>Nk4_>nhJ+_Iot1=6Z0r<
z#ELjkcCgo{*GL|V9i8+KEiZFWO38nK?%LcJRlqagTeq-fa0U+b+5Zn%0;`P1bGMH!
zvNG8fL%kxnI~h+E0z|6#-r4S6+A4v@uYGfSixR#M>$+T-DFQi4XWew>^n&v2mJ_+~
z?Q<-SL6(8Igx+myEDv?lyari6IrjPJ_*m?x@7Uqbsr{i#6sT3)p0!ANR>4#MqF~pE
zbyT+oT+ZiD9+l3^@iobmMNGk&wvs8O*^6DiLv0nd^V|7W02K2q&J?i57Elc%ohE8D
zEyN{LS@~*$<}3FoxhR>8Mvf}Liu|QbU+L3!|3ZL`FF4g}gl3^QOJ^|WMr_-Q`#ucG
zqFFPHkH-LIX_nf{eI1msX6OGQ@7Eu8#pN=1PLe_FU|Q{XnY)=rAsr*Xt6GbcqN`yQ
zwH1hwHtE-{0`Mttw2Ku*byu79v@b_t80&8LW9!`oW4<Nb1DtIH2+;UFW{=G{3crj)
zWof1(t#Sy{v&t(zQXp>`b}rDu#PCvA+qI>K9tVRE1-^l@g-e}41=D76A%HY1zha38
zMkzoA4|t&U!n_FWEC$n~Fxd)B)<?syXo0$E4O#3O@H3`K%HJN*aaE3bB*XJU&W(Q7
zR9~L^hD5QUvXe||SLor-FvZ_KdBuNu6T>VK1V;8}cwS#g<{$IcIcj!w9;cpQ^S_4e
z{{5yHjXglv1mADyO>0MF!>-nq(pknM*A4up-tGlLO%f!WHz?M7R>fM{_vGD1JMWse
zX$7db631%pJNBeXItx`XqMCnZoaQXfc}p%BM!4TW<X$Ej#2)55kEbj3PeLj@?H1tj
z6=Ei{C1WmjUSOfk?VOP<4-Jp3o>AeIJYx5VkQDoF)O*wD&L)yN9>vl1Num-HA`SR_
z7S~eyYYVc*t@6YZYr`AD9Ez7aXe{elq6R3G!ck(ktM)xn&$uYtIhppGgx^HsO2b~_
zmX!Q|jJ;J*AW_#O*tk25TjTCdH_*7dySuv=?%ue&ySqCy?hcK+yDZ;7vpYL88xtE9
zQ8((X9;)s=Co@l`?Dy6c$%1m>3`QSZVhcU?z$%YSiv?FwE=XnJ_ffRjzvgMPC+heS
zjH>NX8a!2Y+IVdz*d2&_8irDLOfEA$eKoIA(-HfD7*q^y32*!wLLM;5;S+*(xn%a!
zTY2Ec9+({fWw~t1MbTuU#1%^>me)SE%&0Me3cAY426v)ubd7?*quizVr{TE7!r&%|
zj+JrP<V>lOP}87{7kN_V8Z*vV{*$88n`41uj^CcG5jZcj@Hp~KK%d-fGj#3HWnVb4
zq;yus8L9<}5x;=4Jqio`kRjn0pak6$lQX^4ADaYB`PqWL_j+p;Hls?m2<h~!@KKwm
zy9Gh>FravZ!`kwcI#Vla;<=-V&eQA5o^{~pNmIqebNkaX`8xgCrg)jl+B#T55Nfkg
zkXA9;JRaHgj<k&M5sgaiK9zz$Su4NjGbkFUHS=lK$~>CtGQD8t5UMASy{W=ZdiW<?
z+*JV>q8w~y{|NiVScmrxP^p#47x>`e=J6urMttKY;^>SEJMWuZ>G5DfA=d~4y(y@A
zn^)<+cHbA}4SAP%ac3E^>LWoW=+!WZVcVxdltz!cNwC4xkF!`ehVO{13Fv%h%G`>|
zi};K{&$I>+3C2&*DRf8^lic~`CrrzuybcMVAcr`+t2tV+EhXISa+?#Yi0PUviyW%T
z3Nhom?t;_nNSODgzoPA)zS7uO=~Y9K*1jPmbFexk-3^Ym75ZjV@#1T!!Gd4b6!r*c
zw5q{|ZtF1{bUX)Ypy2+5#9`7FdjARmKC@Po_sj~o#xsg>*S2vUCbPnvU!w??u22U!
z6rhl5`_We_Aj19{$R()@RWw&VqGDQNybERD&R&hMA~0M+>=lK9o6fC0gGw2*zXeua
zlX(5;o|p5eOhZpM9=C`mJa`MVNqh}iOE)T|ts96hstbq5zk>x2a{fd7PAeatJbqu4
z<Ccg1DIP?Z!ZJl5^<tRm`&}=9@^b=cDzJj{=sYUZ?~tEkPBGIXr?v2heqCh9rg4Vx
z{(X13g=}TQHC5y7D=Aw@TsEtj{rMu)#cw)DEGb({4t{KC(Kgy3VpNh|w)1(w?`jQ(
z+Sgn|xsf(+#w=Jf;oi#=dHeRG*Io3iTi&H0TEgi`B?-52nOfBl2ANrQyaxw>UYth~
zv<q!ItenZEbL*^}PtR<GLWR6!)Oid6uO^I2g1FpNty5G{b9%yD1_}OiUj)HNI)y|6
z&EyWxv=xLz{sK!N#)VW9=@J7~q>g#H@B1Gtr@As3Ve|SYku7B)*ua;x2C+r?<8|jm
z<v8!MauR5yPozVYe-yuB9RkuUtxK2oyP9GT)#}YtHPK(-1`}Zi7d%^z*LA(yTLnAU
zY<1-icXZr4y4t$CRS0Yei#dW+&@`f4E1N>*KF0Q=JCcU$+umz2_}*#;C<@+WPfs6D
zt=F3x@n%ZLt<9~)L<^7^wnHeQuhwGvsOfVtV#CaB7Z|837xN;HTY>RujX&srxk`x<
z4G-f5h!GN1LG8;%T<@;~g)uQQ`69Ba=L@?NXSPbs8KBqbtvvJ4Z`y32Uw@p5zMP2;
z>|B*b#8yI$q#=6>#JX|{j0I{{+z0X5xhaG-cB!2bLis7|lo#^<@LJI`^DIuZxk{KN
zDUMw@Y&rEi<KY8KECC9C;IX8Id=x5KkDD5ohGyYeqOS7$%KBSdmptX>1)O&zO*vO-
zy1K?NJBeJ;kGBe-xWAiG{Vx2j%}hobd)=UNn&56xqAo;lf3!Md-TF(JBngav^Q)Dk
znoEHwV?JbyIwRKQ2XxCE&&p*80#kN<sYzQyhsvx46Zf74G67PT+iJ#d0@^~hcc>}(
z-%^xd8t@mO<Rk<)9j?ICkzcWGMxYfppax6GM%);feoEWE3_j{G9Xs9kR<~Uxb?TTa
zV}xo6(PsZbU>yf-#i|V-IW3_#bFq?RYCDsYXD}Fn3kClwezhV9ZI*uvg$oxvxs9^l
zFYp%ADRmHNnF00j%&_6_d;g#$1}KFinTJLrunI%v!?E_|9g9NcFG6$$`s3cl-Ze#7
z>!C3s^seJ(XyNuydkrr92Y<$q32oRq{4B63T-ZL!$Zh?XtDeN-hKJMEDM-~Cdcbh)
z7kWTw&kEqEhGH0_O!5)5Nq3raa&bml5T^kHL}6A}BtWo;Q2^nupH%DS*H-rG%LX#m
zwaXfZ-$`p5ect%PD(~}S*1{^Czn;{n!~e6=H!U_t*%g~wzO#{}acyXx)-7uu#+j+-
zsE)BO&55uAuD^^CW!+fA6sf~#d3P-ELO^NZR0BlwU{nQ>&pEZC=oI?O6e`u}`}dL*
z`}_e{AqlLhU2514uJ{e5GXW@$41Jylyf8XGL{dqsXZ(ddlBT!tqv7S(=4u|WSO!g?
zF9|#KASpU3U+46N9*#$J?E(cv(oIu{ZzyQtoP2Fo+7_4EIPVwOA6*braQ_<NF|j{Q
z;BBpU!f0c<?Dhk`aATy#tmC*Agt!StdPD$;Ouo3H5jkD2EXwO%tcg02>Z(M_(wRh7
zMFcv#=R(`MKNWqfleygwFRYF~nO8p!*E?~jv?T4KTQaIL`<ABldzj_c*a{~Vm7h{+
z3nSg;kVA;jVr-%alO)it7X!-nZ8d~h-}#K0TcnlZc(Cs%6X%rT)6phh`M}&CE&GAu
zjC!g-S^dL_y4qA=2FgeXq;j_L`(n1pc8Tf=TGF2-vVwH6@k62$$bHP%>qHv#4?T|k
zW9ji=Ped<V`VI%n7K3)Rx#986e+qH<wosXfd$66c>33yXMXLLdon<3;WeUkh&7eHV
z5t`m}LG3zO7s<!SyHJ>^IueYYd1$~(=%ccxGfdJAm)kb{jQjDOtHaUOmG))*nRiwf
zX7HGN$A;qAhDtI}55`|Fj3d~Of$TW>OdwjAUB~xKW1GXE*#P_EAh89<a1<*5SAf*p
zk)dck9x1tlwB}dYtoC@H$C<xqbC>q>nI_Fx4KZk`IkL$@Pry!x9zD)tnFQ!bQz|Qr
zcEq519hmUSdhe8QOxnTgW0qXQR~+xglFNol*JllPPAjg{Iw;Nm24PRdj!Uqk<=;r1
zyp~9P7)(r>P2n}ihwiXhF5LVr#j(tT#_woZVoVH|@>&cnh(YTe@Jni37%Gp@voDSj
z%-xxn_zXk9gbR`3bmBQVQv*6WXvyvi<EXTvU<dx4=8F_prK4Hy9$1ergxRDbKj8XL
z`sI*nDtP=5|0LuiqT(d_j(64c9cm(PrXG$DEAs`C(cIxFP%_&knOA7G*vChNSstvo
zzF0C*mp?~U1ep2TO>2cQDRo1ro@3jV&Jg-$h-jV!;SffW5RFrI^cM(brPKPDkUOR5
zI!;dYk)tBfoF3KExrL)p<V@`wMMsRVW3CseSK>~^1TbKMS-vAUobAJSHoObFf<maa
zSqm`<WA>+9M|xw1&9t&26JDPu{V3Uz@&-bBhH4x77+<p4<-=%>q!F>~_R@UKO5wYG
z(3dd7_Gt71Df~4EBA`*TQ6l6uUWVGn=QvT7;K8PzBm>chG<h{@XJB4!ET}X(0Pg18
z8@^SW=X|mCpYJz~e|Jn?Z&x6~B_=5j5-^-(1z>a>L^@B#1^Td=vVGbmHlx$pBRQn2
zB?YCyal_S37KjpHG<ES+U%+#%FzBvtLZV}Fq=#Il&VOKM;D6xG>li5Th3bVi+woNL
zt5GWTO_K@Wgr-FUga)>}#HM}YZx8lD?*gUd($~-W@FA;Mg!V93;u+ga_y<1|$>$Fe
zw4|U5{5*?-!4@z$Oq!<MkM+THv3-(Mo%H>YDae<0PHSbMmkHDbg+#FAcT##>h;D-V
zg#jg4Rzv1*zp>p`QD?GKz7l>hem5EZItBa5|IYE?ja>04nmm|BPY#ARm4>KzcWJ0i
z(S*jnto6}^hWAz-cCiidF)gC@`dd6+qY#;&0!c!D3k|aEq(AT_9`L$JKHne3{hx7h
z{f*5dm9>qQym5Q}7F0Fh!1P3JYs^V(RZKz1(6R?7RN8rewk(N;SvYx_qpWdx6>I35
zv>L%4SA3QGkD0P)9SJKf%Ybx;eC2G{2aPix{1!)}wdYrXugN`E**j7<D@`)#O!i+H
zF7%P{0}%f+ES{G>H;B0VLHAqlGCx#7s*lB~9C&-|U#Ogw#zZl-8B=h;*R`|8Wj*=z
z6<%81wUJZKb9aAoadC2hI&kx03b8@Fe?X|(yk*lx&G86>GyaQe!P*6*53_98mb6u~
z<RYjFk}c|g)%vvBg&7|fH{Gg!U0%mXk@EPzSeljiLU>guiI=$n%HO1!!YTO)ezWoE
z%Em*Yb2&)3kr4SX1&>~!4C)*x{T3i7a9?SXuaLo#&W!x@K10zKIZQ6v5-|S*Iq&G1
zwT<EoZ<(pB7XQDI^QvDunpQ+twSHG|Vzt?@*hVnwHclS51oUvUC|()00|Ly2dSQlo
zK-kJq{`ryNg8M5aM<9jARX@V9<eiv9_nIN~UEUuQ=E}mqr+>AS%@@Zq_mkwTTjf;_
z2-ZMVAqY*i_L1;GR3Tys`8`93JI63+EAZ8zE;0|5r1SICqO(`Lyo%5)wrgrF5S0}2
zB9C6=x#?P(8xStNL;oT+_MlrO5vAiB0Kg@GPFXt5D@3azn1$M007=c(ic$p?2KCx%
z5uWr@P@f)<vC_0`>_(BFxoFAeu0l$IMdt|{X)X6u4lhMGzE<{c_b021z+wH_H#)vW
z?PS!S*)%dxm?~7lLZ-GQP2<deV1-#d({v7DAzrpV+#gF4joc;#C`a9}H3$!sfm^z%
zCl5o-Cbe37GXm)Qz!?f7jY&1ONDgR+jfQQ~#ncImw?A#nM|7r=axr0K{j;;|<hF$h
z<=&JVG|Ff7#!2VZCYNjX_sxKSgnSXFvE;SIh7K?Pmxl>T=kX#p=En@(_;fLnoGbGr
zWf+pYl+%c^CQ7QsDE1uL())P;5HFvHIjx&Lf=Q|v;1=Rz^DgL@HBp*;gWS!V|MSZ6
z)TR-Z+I$(Q0)=uZKi#&Z{L9*Z>G6oL@)Qc9(WY%~!u{@CgzI%-f;^-GGnvFL7;%EN
zJg(y`Wm-1{VPkrj1k6HEGj#pK!o*OybtsB6_^Dx}O`ItNzsP{pr+~m1aO0VDK&?$?
zXe#BcLToD0N1-}--78@&ED@xd5S^V@99KWq%jvux1!s5C|7T{M_^h$0w&h5dz^0%>
zi@ol{4R3jd8!p!k<B`G4d;O?qOHftKAtd(n2j&Ne0H^?&Zzemn5PvAYi?FXf9|1pO
z3R#*I&9nj)4DF@0iUe;D@cse*bPdmqT7NMxNrIp8vv-@Mmevq%OJhU^(&!LW774ev
znuv#fKl`ojv*54jTg|3yli@wWPkJIOwK_H&$u>k@B$6w;6eWx2M4k`TexdI*-m4W<
znTpXh?|&9P@ht;E-1za)q0H8-eufeZhb$7<2bW{gFdj(W`$^140j4F6!{;Xl+*(~>
z;>wM=ZFb}$G8`~IeKJ*{9V#<UQuKL#Y`GXY1#M)XKH_Zfrx5)+F0n((LgYTwzdS?8
zq(Ue)-JrC{oS$w9uF-$ymHT$N8GmKH;6z%?pJYm4H0~sICBzW&<}6FlkdNGYF1k8;
z6I9+sXP9VGcO$Sz03aGwcN)W<Z!6Rg)|O-x;kL8#XvQl!F})qM_lfF1lIG2u-47)$
zW7mntP{xA+ymxc(?0lk_d(v&rP@I)F^<!n~;^Fp9&7dM%6oa57rh*9m42!tFJk9Y4
zKb>NDOs`WUw=(O(%`+#x3y{1d&A*MJ$WVljF7ur~9!z#h0a#gOx3W>svbWe%C}l*d
zdde9eDn7ftdDRW$4{{%}zO&nWboNBB=YtGFgY)4=7?{4GR(1nPe@Pv6M+_2G<jOTm
zw{+Vcj2u!u$5d8VqqPIKrez4u-ZnBq&ui0HhhdnAd*Kk18?}CSlt#s4)QkHo7F33U
z@Hgf;`|e6~z!3v6Cf=KCAy-K@T>#KE)~&cY*IWs{VO?dTP;SYksyAp)8oWPQP-$<C
zMiIK61@(FIC03I4lAtue1mnk)t8cFMsPNFV$(nS<ZTUGuD@V2xORaE9q8y*ggZRJk
z66ouHB){WQXiKK|9IIZ4r^Eo?_H32Q6GgaSMq_Z1{+rlRF!Cu*Ed~u!R6s}BqEbU^
zZ8J=uODXV*vmry2Qrhj~jW!vFf<_tEs>KQSocl#05V#AhKr@w}<QKg)%W%SR@~|KL
zhi3f$$Sm7<?<No!rHa(^<TwLA?U3cVt@Q2^6uH+N_>uuPGvw<O-mGj>8l#{J?AeWR
zGWpW&<xHHWA}w)3BQ`>g+cKP)Y-N8G>|=UnCp418$~S%RsfaIe@#e2NmdO83M<i<0
zR~hR#AH9Jjml*BVJsu0XZQ$aRPGQ~WI;ZMFw@F6+&WP+(7K)02@k9aZ7DPA1v*&gR
zeTo1U4KGb;zu*qC0z>Jetwp;bMJbxv28pum(!)ysq_qjNL=$bB$yf3h8%;?I7uh8y
zjVt@^RAOAzNOn=Y&&a0A{+bbq1(ix=nV^IyWkv5kDQ88Pc5`9RFe-|d_0Q2Hc9?BG
z?^RN!D__%&tg&gPIhr}WC<QeKqWkg8_y_<dcW*t`BF=^Db*~wz5OV!XRHFk%s#gCk
zI*<&ru36ON7JRm*2o&3VDP5sFP3+H~mW5cJTJ`!8pe$Nkefe*DnKiDhJIU3Ql=^r1
z%N@wfd<v;>r{9lKUx`W0Mu_-jT{l6*vwu-soI-6pOAWpf5bl0BKZvt!sgPKt?y`Sr
zuAHra-{~`Wqru9pi&@2K#-R3!O%Z*P&D4qI<L6iZh9%GH+K9|oj3Xkxll?&v26b=A
z(zqJiAa>9B4{5JcL6dlL`-V5Vr0yDxDR6K3%c!7~JcH%tb%1Q#@?oB5B8<uW1^=K2
z<YAk5N@A(0^gcYZ6w!m!(<9~Bw;ZU-L{dLg9#G*h-||l5{`@#o7r|@3&*&C*Ci*0O
zh~adl_}6+3orThAr{*BCytDgZx-av)^1thR@C^gk-=|bv>zS%$W|xBMOTN;nD0A$z
z)ydx5=h|AIU1C0NcP<DTzn@$eq8D1;)H5Fi7pyl&{YrD_>1X&iu=&Nm7hGpKP^lCf
zIJZ%cYN8*7XY$*q9b!3n5GgZB1@!+E(vhWx$zJFn$dxxRsW_tTy)kc94YRC1h+W;=
z;r^$|dO1x&S4jPMkb0JWCw_HJB58}X4d8+k#?`-M9fRt&>?L`?E%ydoCfu(fssJ^%
z!b9zc=779M?m}r(+bB@XbQTQbP&-fL%AiHi^X;F*CxX6~PYID!S?kY=Og~srQ##RF
zhz&(q*EN5WS^tx_*1p$ZP);JKsI34V1SWg1VFDSG7X64!tU`@yjwBkce(WDloQB}Q
z(K6QW0eijy?s4p=UL~hEyv`Noh&{77yv+MOvl7-+tf8u3<bA;Tw7e_282l1;jUdXQ
zjFp3~!ngMqp{<FMI6Blc!%HDz)HHWHE}<_)h0E7XT}XiCO+z+b^A}a6iD3xdAsF#K
zk@?i+kYL5HG{j-__CjDdCbB>)vhePb#3-VG5heaP>$^yb@snt6&uWRghaDq>O{hTM
zDw*nPEWfo-Fo3<&d?wb3q0<JY%!8t?h`k+<&<4VjNLxrhZ8!am?D696w`9`pfR2Y`
z((HT#Cmd=<LgL!JRPR+knQ-^1*rND_+-yKe?O!rUby{XM5~kGc;KzA<Xf~4h32v%e
z*0C(j)9PPbO7u&o5|;S&*NzTHKJMXPyMPw6O6lzv{t4DWa?cZ9aR+`c$QB+$Xro3v
z!IxWJN+HF`ORFAaki4>c=~ze~)QWFTjHh@vUSmVDvAh17;^z6EI=?H@3GDC}qUmdG
z!tR=p!1<qp<C6WIkOSUE_4s(#cXeow)vzw%b@5mZGLH-ic1OiP;;A&^&vv;zfw+i$
zm!#x10OTPwsvZrD(!!NQIuK1h-bz?JP(g>^)*^v<q1$dRoOhnBiK%XoJ>zQd>3)!X
zKElOgUt~NV1M{jv)xK$8<N%X(Ordz=0BS6v`{+P7zP3JWhEf|`<Ah?kRVj#q9gJ0?
zT7hf7^snp|@c@o#2!E7BVcx*L{{V<I%6_OZAlg6hC2|v&KQKGd1lRW_NJn(<vMpaR
zb3ODf`pP|pxWH34Vy|uVi-F+Hfo@zcg<Iwqkt^UqEgy_9lwKMRzK>_7{;?q}I3D(Z
zOVWVKH=l3Gs@9v49kP^qJDQ05UgI>)(A-a71XVU~bH{!ATc?TLkTvI8ij(pDU7W4~
z$cHtwjp_NTQYas1u)EKhbMva<aek5mj`V4nha+6D|3gPdD0`p$J^f?&BMASE*ueh&
zxwd_DIhWH{H~ybdX=;Z4dmyE}^9x<KRL~642o$Nvr_Pg9SjQ7KuaM!0s@b~U$QGJE
z+FC#kDESMTKcpJlk)|`VfMCBubE!rcpgH%2PMavC`D`gP<s73KWpFwThT>~o8aIg*
zu9`H(wtX+0+eoAHkAZ929;We4x}{VDI=^A^E0-pv|21w-=8+Tpy@!^CY?w;&lxVfT
z18DWU|9CiH1Kq5s>GS%r4TW(<Ib)+tovBz9;tq+5!Je)Hy4LoYkxvXcigYsqV8rW|
zN*Mom4mk)sgou&FIN)c5u;XW3atFz}!vZtIj9^dcwCP;8rp@L|XRjX?(7CHVx7P7U
z(ZglIf;e_!T!nuTJp0=R@6aHn(GYdKgr>Qt(5;B^AHXDQL%(htDQLvM=3}@AT2qNT
z>?@BM3nZgJ$i;6kZ~Lfq$B-Ogme9W<K{k*S66=2uVU`|c@<$<#A{9bquc#kKXq)8K
zuAkry{z;Ro810vc;kwlvItW`dqT%a9^C_5uukUhTa+sijZ{ZQRf>O~CwX)Kf>#ql^
z?-IB&8oQ!?rai`7@%bMuOjg`A&yL^9?@y5k*(&!)`mjJM)U*BCWHkT;rs(OvQmIbF
z;xBYduaB2~)o_t**DbfUf2>E}*G?I<o07jcal*>`4hiIu?F?`5e7>tBk@6G4V3SCN
zaWl;g3FsLgBv+SJOo*vEeak_PKC;6{apnSvhAz`o!ILN1=53K~AqHWD((C&UBPV1k
z{GTPTotV?3^&%X=vCN{p;uXsbq(S}W`0b$;g8RoO$t|~o`bmPCv^{P9c6X*<UneQQ
zeNRqG_GqV_tmBJstq|edjY+GzxP^oP=zI{Krt@3Dk_u(n8bPr@)W2Ia)k`NO%^Q+C
z`!;^Oigpy|yf-oGA#Pv0F44FfYe@f*_ZW^xBn@pH$u}@ysVw)|SL_X<4bLfT3=agh
z7N45G_Kz=V4V^V7@>??;ksWAYVbgdGFKT)JZ*}U0w%=G+-TjPmXJUWLSdxCbg$4}x
z2Gg~R-hnWUZbK2oC5d0Lyz1#?GuY(d*O5PQG~{YBxl(tmSqs_hck+4#I<^kg4pzA%
z6wJ4`80EYJ%{&_>c<Fz;xO(^_^6#L0!F8=sxCPH#cS`u8!F+@WNccZHE+V$iuKczo
zw@NSSw>Sa>jckddTz#jh#}|SoI6!+{dBFTVso<fHu}sE!kUP8~#18cBs1=wgh;sfY
zIvJGZ*Luef@c!-zY)3FdG5Hsu)H2l_1LJB%ikF@TV3?k%p)0lM3s1~Un<qRLxzCiw
z+A<zvV#XyD3mhd!Kg)JQMzlH-Lp#JG*Oy|P+M_m4jrz3Kalm0QeZZecB@Jh@d3C=4
z0iJ`tHJJfiY}DBBUHkhJl%GnLN^OaqFU6!43Kce#4$W#vBy9Y+Z_PZ$Ig@0$2qEV=
z>{G&kCA(yzL73}&M5~#nie0U|>$j4aC0*OKsw#0qu}BOqaCq;#yCOzf(Qn~M&QY6f
zKrthuXmdU-eB?I>7DlLV;Bol`pvVS@*C-}CNB;7)v1EEXJpvh~82om=$Se$DcuDk#
zjBuP>cBs;<5R|-x0nke%TCy2_`+b+tv|MTcY7`4NW#AS{O(+-~)2v7j&2P6bbTQxE
zdNz<g6#QxuR1$0O9z;qSYRm}hQZjxNNZ4e)H(vwHDzlIru|k?%M07aSJ;A_7LV7bj
z8@v1K=OoExwZKFfeE}1^5v$gibYv_F338`M9{ksNN;Z`9BH9`hs@QMBMz9U%sFh-X
zp8_Oo9O!j2=-*`ipb^fnNuiJa#9%shlE4wpnHpK?1k_xcvAi^GB@$@40(kW7aX_f7
zKkoL`E#O%8J|1ItquMRGh_6*T&(Q4yA66?e@ks(80OG~+wdomNyci<ZKM~Q~3pI~0
zPN*<jhhb$TJ=q&NFIX*&)ptX(Vy1u>Q0jey##K9V)7u}7O0x8?PCEu=LijC{-}&PR
zLDmaVk7uVw3iIZnt;frVf2iA#;VOHQ4vbZ*<<let>a~8iTBcR462=inu!R|yM=W>7
zgzS)%EQkJZU8eSLuj%O;cqXAbLmZCr>R)<OWvyMLPuPI)%Ca^)`<-rJ{*_lH7qJm#
zh<Qo>NIu5|tsMGzI(T|R^^9U`tsdV#`R%H|n0Q_tSW|Rf*@PxWchzUY$o6j!)RVw3
zgKbh4{Cslb)K7V4n&?eoH+gV6Fh)+ZSh+}=gfQTfC0WXbRW7A>Mf3LrrS;uulfP>y
z)F?eCZT?9ij%fy#*)d9&GHC`_jVY8;*=JvIJ02sysZk{3uCBi%!4+5S&yVgLSDzIF
zqjb?r_TJXY7T;I<zF()1qzq)GdJNg~5<>Nu8X@F;&y%s_9!Yen*7!a9opv=^xYDLj
z?W~G3Od^+*14B_W`RSlL^ZIXR!$9+K=etaR{-ZmA(i-WY|Etgiop(RrmZf+5yg5`P
zeW?V6Nn!ke>HIUd>+QH#&Z5wwRD~*0ix-NQcrGbcX!RWv`%-CqP-L{(9&1$MDQ`y#
zz=_H-@G1A^B7XQIBEpJ*_$sCq&a8fi;=i0N@{mGwIMr`-H)B?Y&CdMtNjx^DcN-*U
z6zr)=TDANDnvvk1!RZB{(Z1kdyh~nZN&je*mfFN>HsO_sLNy+&Bb$pNEYgP;?E3Q|
zQe(_P_Z@182*<l#v=Z|Fwb51ih3QhU9FJ4&Hx`|MBPfmg<jC6h=`<cx+Z<H$*WB64
z6y9(xCf4ff1h|_!Q4d<c)j_)jz$@X>&kf|Q&`L%EDaTyy^`(KaH;{PLJ>6&p9do)C
zQLt~IQ$m+Gh^4}RGCkQIhQBdUK&LqRzg+1aPWt7(F{0GlP>fu&FQG{gQMYqYHZ!Px
z4cjRfTFy56;XipbIPDL~WzjLtv^mS9ax|SEyYt#o88Y{(Os1e_c5$#yz%(6KA5dO8
zSmo^l5%U6v6lx2{+3c&|Bxf6n?nbG8A%5n6ewcoq%&I!qM4<$oV(79de@_dOi@DRi
zTX=T*P`g_w?v@6gbE)}R-UQTl#>%!wPeL3y46;Vsho8ZNq3i<dC)WqRyexEv!I5ZV
z4nJ)5yI->Ms!Qjp%FM4Yx7nDdMhj$*6y?zYmJ|d2<0#|yYP_Zo+uBW9bezFO^>zib
z>_j-W=ZY1|rAZ5jRu%e$Qw2?nPh4K<TE6;kDr%=v%CbWW?op{|%1j93SgTxvDSJ^n
za>}8@GGkJu+Xi3EGMYmz)QwM_=md{aPrXLbZ2sG5{SvEGeIio_o`S)~>2o0mmMI1>
z;k!j^5h?e1-_Dei*UPe|H{phxhV~>*h}_Av(%ByKuZBV*`8Fa(R(iaPcnG`~Zp`XM
z=<M_wuCTv#p&f$$6!ItLD@oc#d#r0!&0z?Ii|*`a{bLfk!LzF<E<W5a?Z5RaVI$+d
zPC$IjzQL<}Ot}=zjo>}@IcYoVDi;>ub|D^xQs9eUnRTl=k*#Z8CkiL8Q=?ocJn2Pk
zh3wWN&cy%ADB(9L(kXXQ=ln^wyAkompL~)9n*JC^!CxR#N^`5!d{%Ms^WL=VfZ;)^
z5MCA~(Ezay)XC3=(AI`n+(u#{iQyNo+X-u9cDjKlUPzk~G^@<YJL1371oU3u`nrc`
z$+pD(-MgE3(=f3P+5Rnl$k?93qA1>%>8xyERUSsj^><ansh=qA+a<KqmkH9ZmK~#s
zyJ=Q(=p+I<HP43H8><dv#}e#ntBvbw>{_*bOC4$zi718EEZ2tZ31USngY*op7yIDj
z&mDxFCB=&+Pt};M{k0I!oee&K-Yn*ObgLOV?7j)CEEBCE(aZohN-`lyt<0aNNyYEA
z5OblO({$nU28D7qq>WZJ910-KJ0W&0@Cq4jnk~3HHtg-`3bpeVE(ch4r^Xo(Z$z7#
zXXqak_#eXxD{)N5QY_?yMN@h!<eh;fqxuwZ$C4!*MyG2-v7HYtbkpzvub18$D~vut
zlNg6BtAn^Uc^#ZwD~U4Px+OCXP6D;n$u~8rPAVp84`TsZxEQfLcb<#ARKX>Z4aU)L
zrTzDG={;8kTGe9rJ@f(}u4-<(ItIGLeP~tvXg>Vs=K=(#%e^rf81@5j(vc-$L1U`U
zPZxtIEw<RJ6Xei=&*57D9YYCLiJY4Gv}i-DfcOA){lZm}_(xi)-V|kiR<Blo{+#((
z#mS^7zGLItcg<z~+PiiOu;<?!9P@t&2|Sw#lj?9|UB)vNNQ#cy?-V+{l`f|CkvQC-
z9Rbd^A0oMObjks_y)o4MxX3l7;lXqk;w%&-xQ}cO4&gE)NaXf_PKrqO`x&w%SD1~t
z2IY60b(Vc}Y3YB$_`ag{j}yupTyUW7XI`J{>Kdl)M-Nnan{ZlCEy#8(md!%fQ9Kw_
z$D)2wLs*l0Ey&AfwA_^nrQg*SE`X3{h)2{{R1{P^A+amIn5AE>0Y0p2pBo+VhG(#s
z3ft!J;MA1@OV&idgfmf%#p#n@<G;Z0J{SD3^Aq09Cc>&qn@jm-*GK$DLc+b`^wW?K
z+~?BAy-q6lbVjQVT=r!{`5_^9ot0>0JDE&-F!_E39XxTUs--|#KHkN`HvW&RK$vI3
zzn6SAy8}lG#Y#5JiQ3=0KJm%D9u-oqf-WvKCm&CY%20%WO~$u%h0p$W(Qu10vJbm}
zh;tcMC)~p5?uBW;{H>$v?>JYtSMh5m9=_HYxB{X%k(G5U)wAxkcHtsLOzB0Edb#*%
z&xv{BMc8DHvt8z!TYTa*P}?1}Sw^xbBXgjBBC=eTO4R4XqOG(0rJ8dlY?buE2nIwq
zxsQ~NG{bs;k%|1>KyP8DWOQjpTX3i@nbXqM<WweRD(#D`t0PPmiXrOC!XFMQ9T@~=
zU7o7Gseh(D|KtOHcQ!kPle=*TX%9pM4KUU++uDKr+@|s<X6TI<8I5iKBU%idR_iWw
zLo*>ZJd|nT*Q;AD{n4hwJi0RWx6^2LXARU~+?aSD&?u#~U&!ydJWj#N2N}zDjiAro
zkv3pTG8zQM^W@x)dX0SxeZ-AVM&ixkmA{%Gj>yBRE?SgdM*8|PU)F1hNn`oedqgUx
z@IU$^?zw7tDHabA$s_YpQ=8=)-CIK{W3AdJV}ISy#^O>pNj!58O0?>1%P5soZFwXO
zAPKJlDy1WTHGXeOGdwG)l(xL>KtXTGiQ4?q(bnG5!T&}&n}T%>8TQW_3`|}kBQoHF
z++ejNh241?dMka?lpk;KVBW^R0E`-!*x(`>e<&(ck#f}Rmz1ht1LrNum_lcQOKF?X
zOW_#q{;dzoZ-u{5gRg`ZB^O^mY5b~H455BKKtd%W+1|JqY#(^8r)8mxlR5aqT?cvF
z&@z5BbDWU31iI0OKmjkkdrDydT&yF*V4OT?deDFu!ZoZCui^KoR83{1BkW=WynDKV
zcPwtjnK{D@Yyx`(8C5NQ4Jl<%MO`}cF`MnhzSTnvUjg6eo@%Ykj5vAwtcK_SLTLmp
zkYiwiSu0qpkl4=V=U=njyN@J0o#IND@h<@1wwRcUKJv+Jf=<Lm8iL4|n3Mx51Wla!
zL+&hqAYYAMsnC5x8xQZy9_{!R?+z(Odfo8s#qW0&#VtD~g}!_`*uFq#f)e2x=@It^
zyk>F{(elftevP()ukX2E4~B-zlRS+Vur%{h6=TW{3SA1`H_gaW20jd#)St@9l7Kc0
zrK&?*`6L&DtAeSy;h-&3*6{BZy^=>T>WIWM1G@}Fx#>+VSAq&{oGJ|qY&j@MF9<vL
zkJTQ5XQxV)Qf7dtmK>G}oxLAyZu?}12qx3IsCF4!d4%@?nd}Zg5nN1(QEoK={N4Ns
zfdp67{-^kV>mqZ!&Hs$fk5DbUVDhgcl#ZNnQrmGl=>(e9uAgq)Pv2YL`T7&Jwbsv3
z6yJWAZwm1UkCt9OjCkl`qCAwSD;r?%LX^vSEkWSd5j1h~j_K0f+S1+9@l)V?BhM#@
zw*|T{Ha37h-y7u*yq~cQ&C6r!3Dfvp=`!TGJYM>mO-z5?vYv`R6uZ@)p_WRTrZPE{
zU#8dUilM0)9pp-TnjZ2a1J)Bx_LZSqG0}F`2*${|e}`6`gxa%3Sb^Q^Bc_RLlM%qj
z3hRW1H9oUSymY-wgv#(WI~JmqQ)-b5ee!-I3f%`(qW$EmRQEmNc!)xo^Bt=v%nt8K
z<CQEK{vKe&Zo{H3<NYNUYMz-_qFDCYB0vzh>VP5WEPKH6aB=@*d;DIXNOj;BZ_8@Y
zC}BK^*ifJ$j~E#6=>iG$I1AF#J8U>Ot=H4u-{O+1Kv_R9OYjHAyx$X(LQTG;4gJ^V
zB`PoA_a;TL;9CQ;2(>&ZPozOesVRx)shd~=eRhuGX=yp=N#raVLvV=xC@&%3QWmK0
z({VsNRCX`wr&6g@@JuBO(!ij$|6w3it8B}0_2aH-0zDmLbb0Tu2<9z<6&Z^>T?Fh;
zM~4$_E!Yeh6mJg%Svr;4_+P@e?Z>Botdc$e29Ps?y0|-I$v&|Ql<5=3fcr=IerK1$
zJCpA9Ckyxv5v$#!V6NV@2SoQrYqT#}Y*Tr?y3xehF|A1Z**w|r;lh#-@X?yF(x^hA
z(gus{G8T#NUIvR%AM@?592~Ad2x?-!fP~b&#X}WAu2M+f5v3j)ItFsoP-FRgag5wo
z|IMX<0t`L}RZpzS#6>$E7zxr$CcyoQ_%Jm349`)xC3s(a(+*CmC|$aVZTS#my3)M$
zDa^ZtGy61%;FpiFxs(&VU?;RCn7;g*E{e@&%cJAW-qxNt1)L0r#p*e8`RW}-{I2hc
zcl}9!hjbvgl;(hs4f;mY%y<89gc=O!3l&t$#_;^05?Nlt6BNrTg=+1{+{1?WZ<Met
zBCpO1J+)@$C*B)afL12;<CIb}q)Mx6`Mya6J}!DJG&3$e(9$cu9w+(s4S@6<vGOOO
z+0o_tg&%s?vtzKIr(_K!qs@Ed?#Mgzq6OzIaiYb$BPhUg3Bergcq8Bp1f<R52`076
z!{5@f)BYnJi@PF@k#!{PDTg*j9cvIY_NV5{l@>@3WIq>3Pa%#2smQO#!N2@DV^8(H
zoGc^z1V7dXfSl$PWr={7`@~O`z1y_{$IQ0spElSN5I_jRfm9wuu{A+)2V-lX++{8Y
z$*XJGrLX#6W{c_)Ay)?nFjao$^hs>JKlXob;gaf2oLJ?#5bbXuXqhQd-E&dMCw!z1
z){Xq=4mDsI>z<xGPdwnn6-hKGoTUU;r~KrZ5J)W0Xw>24gi9>=QKnpWKD+Zk6NWFu
zATwQ&myJ8ChnmrDfL7THMPHvnV|BrJKzW;Sq_qzY{oWI{?lfWtbb5pks#D=K>8)-Q
zwNmX5oroCb>rlFmulG8DWrqADG|4Q`!~b{q%z^$4eS5V4dgIaEzb<>Y7sHXT<_b+2
z;p04chKfpp;|wO%UD00XPs?C6p%8k~{jMgcl+5QP2Lde@f_=wPpuy!`WR556bj3($
zUDkp&brQD6FO8rDTsTxZ_!dFZx>`@qJ4xZe?ICf<A&>5%ko{SA?v63_nZu!SAZ-g>
zzX_sj(kC#lnW~2lFu#ijHI9_7SRPn+^9kcvY3m=MjI;dL@s#@`*GbyQ1S-PM`4T(7
zMbqftaB`HT1s})Y1E&gV3Mc<8Cu{(ZFYTeHeSHxag#(8=fL45Sr}R@BHPEn+Mt$(B
z7jyXe0+Bl<ujF$6EHuAK8HP<;i9X9IahH37Lm)SMKC8rrk^ugnk7u#J+$?>ZX5p0A
zL}w~R#(!zB<Go0{B?@|(Ro!5@*Toc>-2e{Po;CP4po`99nhVo+Fl0wnhYF=f`oJ~6
zU~Jh_i>MEf@`h*bAJFPnGQiB?%)-&-k^GIriq#upl%vgrc{a4s`_wwkOnmpyA*-a^
z;pD>MxUF533`6bWz(g)1Fk2IBAZF(CD5k;i0e6ZIh(5}0`1_H1lzrC-t}U=uqkS(|
z)>PmT(2aJpBQISkN6m`H^<#>Q>|2cONFGv7?1VV*`twsvxpz!gKX0EWm^ol&RfU&U
z0pgltd|#om1?jJB8j@<+`<+0kD#C_GT~u+ZFfkRwc%lhespFGpp;uw`(A+d*cFx++
z_)0`V`Ti1)ZOwSA4wj>rNFeYKNb5DxT;v^ql<z)UH8U((R^w0|TWfV&8bhb--ghu!
z&+!dN$qyD{fBk;lpB(#g3Yk-^H9K7S2p1JX&?tM{|F(-aTN`>GYV~LVm8C@FQr;><
z#WAZJ|Ca(m=gVgqt_h}YxGacOnF6|gg0L!9__HH&i8lksk|IJdHkQPyw#nbB#c!_5
z%09r@(9@*$QZeMp`O+*>7y3&e2G*>&!fpw``_Wi@9XGBkv%YHbYq;F54U4R{3RAuT
zMMd1b2kl@e)V_2m@-!-z8-of7`~%^wE^NURDQZGYC2CTJBA@Mx^Ra{y!P$dN%Y30j
zE*g(WNq+J+DrtiSh_<ADFEESOp^Rv@@H5tql4l%c`!b1qvX}p8kg|N1kl_0^)xZJG
z%I}hc1xcTOa-c1=bFT<PP=~y%J<>THC~B=R-ucjFtr8VV)z9&0t*_S#uShc~{YHWm
zItRU>qQsgn;zwHJ2NxJO!MXX=+N+hjg`8-IV1n9|z~&CMHNxxOE(gNu9<(t>rJWi{
z+$TjF+n}C0kE^~00=yXV@KGopJzRlbPTsfySeNmJED?!i>(Z<TFU*7}Vz%X)RhOnp
z$I<@Z4>LY??LN$%a4mF@acQS+o!DI4I|cu4F8p(Se)thy1%{O4S`zUVTD1{rD2Y};
zAus=*14E=)`p)zAK)y?t>~6cX?dS6oB$jF|wkwSuPyFf4=wyjt^sU+jdNgnow2tw5
zhACN_rqJ1eL)-FIN@(7|{X4?NqgHhuXc=SAeg)s@`X8kP8|#mxAkl%m%eCs7OE3Ft
z>~~a}iI9tpfv^ni(VwwjI}VBkCns+lngW-EJ*M$njPoD4634Q6^?X0Y7=z3&m*9Zi
zd}GwN3Eq~4MfX}j{w#^qr*ZfCKtzbiJlbIX#~M%A2|1L4<U$iOpVGhy7zMlB0OZ(j
z<ge9WHvItsy;O^i&5i#|_T4OFlUbi?g*L37VPLID9J=)U;QV|8%wqKPN9$vDfC_8v
zO6;VYh!y*p!uX(xpv9e}6}wk(kb(1!VDP^qcHC$K^+rIUa^A91IS4T`WWRgknD|7?
zU~8fs1ruy;@f)gI4<RD*=ehm2x?xHW0E5rom(&~!gib#=4h}+%BMJ~?J(z-5ViDa`
z8gEK+(mO2QnsM7YqzwSK^SO%Mw7BOP8^QlM*eh{X`6u7kc;V>eoWWp-H>rjDs~>&0
zTY9B<nE||>q8y4HH*9!E@VDpsZ5>XQZ?46|si1nixG<eB_B_GIGjI5+E(tHDR$${<
zl6KwXYIrS4;vc{<yHJ0y40+VFN9<|uj@3up>(vsxxksd&d0d5e?vW|3H?nEWGVr;z
zIO5w6MSxdQ1>vsduoVt&)-Gdh3m!xdyfGtPsso4J?>$Cc&Bnx=9*bd7ATkiH?va&{
zV2X_jxvP{Ah_*1HqDiZ*coE_b&;3ZmT{8vR?nU;;4<p+U;JoAnqO_=bASN$pVn0{;
zT~8NIKrp+9oP@Cdyd~ald;29!geiKd!>5If#XxgZ{w@*^1nFl8%;`$0z#8`_cH#72
z{tXncbU=C&B0Ymra7hSTw5I)r{reR>L$KRpwEF?0;#`0Q26!&>AFJXFMr$FUHhi_v
zM`c=x+*DL%lD~UC%TBmDFEJe9M<iuZQ|%ARS`6;!xp1~hFvofwmEW2Pk&r>ZUYolp
zK;NXqB9@?b^cN^5D5mc>PiAN7;B@BfY@i(qQ<UAxhQJc=ya0DM!s%=_qLU{gMZ0R;
z>*nxeHmG0o8jGBaV<O}3E0OIMaSvwvtqhH`_At*qbx4<f05=536q0tGrWh&y@r6C*
z5YLVYD*vt}UbwCP_gL@uyMiSJ-_<0m)#V>N@kD)>I0V3+p0&gBgf?Xgwvc9#0{Hm9
zlkGSME`flTyhK=(ZtD=qu?6=}NEp8JTU8iQAf9Iw(XQ_oh4tvAT-PPVz0<_2L}OL#
ztS3?uqQdB&d_aCTBmuiGZ463gbokTto`=K8fhOXJy2PB_2mLn!jx-mhw+-SCXRq%c
zzt@itXL5~SNl)Y6+Ul_LU{?9UR)DG~uMN@Q4T5N2v^)>FLB3&V6f3g>%gs`4StJWa
zmQ3R#y!1SPohq`Wfu}guVLY?OV5dR_C$nqBs`vw(V{@0)4MmSToZmBLCX}~s_->f8
z&Zt|5Mzu^lb63{4p@!KXPD#yjbf6XaUq;7!4tKKUponvqFrjGX9Ln2u6cDI4q4+nr
z(sI~TsMM`l$;!o@JBpy1E#%5*czg6inl<8V6WIq3p%0nh8;%#dEdkEwHB{%vVYa4!
z0rf&zX3rVdH&W4=c)HjxR)|zY;yfmR9rOp0l>-9|5gsG5l);n~z5<a2tHIt7jj&yI
zsVkWKEd$^qgfDo=N4jC74U{crqCzEO&w0LTlWYq9ENK!Chj?UKrVF_h-MGA_Z%wD%
zP-tzx@|Kj{?e<=y$R%crJcMD-zO=Sly_0davsW-NL0&AQm&qc4FH|W@viNR+hi?q+
zB_x`+l>zX!R(T89D35Co!`doVtv?>=CosxYX{m#3zO3kS+OphW0?SpujB^!Fn!JM@
z|1I%KBPcg#OIcY{s;TetuU}3Y&2a1TT|xYaL?rC1r{CDFwv&VfCBwKu(L-jVzy7=_
zEYwT)0sokV`aqn$;$c$d-3HMy3Em27FGs!dNl@@n`cyd30*^pU{xZt$FjWkdQ4Zo5
zsyqg2waydZ#@Q7(0H2g#CirKW7u;NzLi18i#&xWer^LF*uk4*@i!zyM{>C$?E1WZB
z%a0S257+_GZg=X$soJVYgxj>$2!YbgqzL6p1ZFww7Of<nM1@gthN8p0zrk6JW4Kqj
z-#%qxZxAI>|3fDzO}7UN<}KGbGC1l%+u8g3LQs@kwl*9C&pmQJDbN#%LR)rTxA)+a
zUxC+jW9Q?EVa0~Z$7y%oM7Y^8Nmc7uWAzY43eA&Le{$mCVMh~~QuSz{24U|ybM<9v
zqyaWj_heHvI^BX)Ap%5->2B)=5T62b>ErsjL;EMZ_eZ{0)#R?S0P9Ff+UBRS(z5A2
ze~;kLyy{7y_;xt2n)luMWgJNvBy9(C3|3XcQUBfULS$ma?U6IP=jvOu-D2t&S)4X>
znzHS;f4%-mcc!|fc*o|vs?3J)0ooyQz?FIj<4qlI2xEW?&)HdP^_bYd_(`{z9wd;z
z%x$R(dtUF5OP<Y!yT7VlNwhQ;<e+UKc@Fd!Q@sGD7(K9-nWZPAYSh(Cj(OBY`q+Z$
z->Gr>bIXupG80l2dIuMY!^i9YKe&Xl#{Zd1_%ABK>Hkb6r2j8e0`UKzOThi|K3~7%
zVv4o<7jinpWh&pUNY@1FX;z#roEtUs<hFh8xC=&MK}F*FNO*A1$t%H!Wb$}`%vkJT
z@Xgf>i+}GZqZF?HIh?XJhsN#NurM{8|LMyx%FDl4am*XU0p#Y+f4#02p|W!UWX5{I
zCnm$6$d2A_l5CO<r~V<7mxvtut%qcm8OFziuz{gCxKIF@rh(t=_iCj#eKOhz#)!<1
zI_C@)ip@|+KYPDxbc1yq+-k0`nHcgyX%I({Tl7Ayh2a6Eb3Ag4pl88L`DI6tJ62F&
zgKA^Evn=2qaV+DXyWlTk*Ss5~I@B93*5}ag9y?^S=VU`d_)%bRO+nbGHUc9QNnU~J
zoE{t~^#)n0984)?eULDOwX)9%un^JK785!weML5gimfOwg=O_0`OCq-K{BejV8t$1
zpX|RKJB$89FomPmgGYa;%^DT+s}SS?@EmYR2ajAE1slRY4n4i}Voy2dJ3hK}Q7&se
zQ<0g_n4T>~z^D;q3<cL7pA+y6q%!h_QfW`7tO`y>tPPmL*vh4I{7sc6X#JIN%}-ch
zdckqGK=VPC-?|3dAo~J9%I61h`VP$4b4O5Zh&RlhJ?mXPy*-^7Dg@g0EA;Szdze{x
z4+>7WJTx>1d-fkjjq-7^0Cgv*$f7Qb6*s6TIkvoFwxBCI=s~$foFZhfABa;)TJX?Q
z5K*4_$vpEsNZpVdwSP!L{{V9X<}l?l`C=uNa70x{eyRB#8#9RB^mH#XIk@&O9XQ4U
zEO`A+M1A(H4hE3CYN!Dv-H~qqg0x{@?7x7)(-R5tR{2x(KWS&FfeN$&hdnxB8E^?j
zU|Ec^aR#V3XkYF9FKdCfU&aQC+17@NvJw}8cLB@whqc%7%<8#{fu$sdm`+VNU@8u*
zpq2*29*-=nLvZ*Z6QzuuqtypJtd3=Igfk=hi&LK$FkF|+M7F6CUCB@YjptCs8MfBs
zy#l1a;@$fT$VLh5%#Yb<43~O^(qwOZS0t@vYr_{CfoIY&N_8b#F}AyQ!@QQI<8dp*
z_h7oTpJ<unGD;O-LTKcm>UTm53t=z7s8$%15p2k|voKjw%x;Go7o0KkRdByj=}NoD
z8~qk<Xkk4o95*e30;?6EURKwjVDrN;G-+u?7A1pX0pAkE`s#F0p#Ho8NjB9rTQBqZ
zp%9*$qlscQBynh%GFSX~rXax_a>)`rdDC3nDOyL797_a)Z6?-B{z<;xPiZm!ryvWn
zvg*)4UT&79{G>UhDVk!C-;uY^;PYgOvPlt4YB9@7CS>{aRfYoac#W)#wL=DZYZh4A
zd5`bPBWeki&!y@pk~Dogb=)dvPN3|;s5}L76U=Uy&H^Gq6dP+pN-LwDLGl;tw||=8
zQbazcZ^(|+RBu9z`@snaL|U2RZONAhYy|y0b*6(tF$=1C-@nDWkTXg*O_xt7ZKM@_
zoMM*%acAk22z*{(*^{WqJjG0aHX;oSj%o;8gRwomSyHo|>&^`2VM7exMr;<xp;mjx
zymJ+sF!n@Sn|dr&I3Gs^3O4?iY$|d96~)~Le3Tb$e62q8r0~4m5SXaB*&nT3X8~58
zhgc>POvyrt&N0xog^}q`#0dBe$umg>T<n*iO)K=VzvPzyB&$Q(xnKS=Pb_w2#&PUi
z8_%Jr%otWWD_JgZ7(^*Y_U#sIVTVU354nApcU)f|rk|jD*R`o}DEyBymaOTtXyAx)
zb8VStHq#Dkf;bp~E5oLLjWWl<Jv-R2799ZbH&UbxBs;hH;vKwtxo(T)q$@Y;?P|Gw
zqdH*_XQT@#jO?6?zQDUkIn?|w_Nt|P2MB@L=tL)X%3Xa`J^ys?xERI1%U}7pJV?aL
zr#2Pdma&_r`@YJ`xeNn4i?!GGwgZE6@k0#+8Uz{>xrqBJ*N;N;v6sMxD9@mCJ)Rpc
zj57ZuoE?{6?*jbt*GperZIOOc<S(S_+;)(GCx8*nQ(#N+i0)F1h4I}g@;gHA)o|{W
zbwyfm=(iYB(6j#&8`fHAZ`aK03L3|X1GU){Jl72AL7M&~U>r17ByRkpKsvS0#K6o-
z&)dFL%}VnBV(guQJBhn~-`L5-wr$%^Cbn%&tba7IZQHhO+qOCJ$@9E>*WRb<T%78g
zzWQ}lclFZeyVl|<XONe`MIR;bjlG2*`?m1?c0@5(1;|%+J>UijbYfd@V|seLcM`8{
zIeYEQ6DLZGmEoi&5YSNmMDXxR{obu&mziRtge0b1&nG?kz|D0*^s6(Qh~u{`_K)hy
z9FtPwFjvPjC5#L!BPh2Ymzl66Ug6s(3J2=+!cqulYYW9BhL1pOKk852M6Uc}RLy27
z5v&SJfD4nTW=8SbJvG}SO$#-uZ}h=fDl(sQ6#|&5Wzsr)HjoqwkZUic?R>i+1)rKG
zYL&*4eAD_76`wwh*@=$Aq5`c8zd6H?=pzSbx+ig6sYPp7(uRs#KF(isY$~NtnYC|$
zYVk>rb}%$-NU4;AqSI$4AolO;;iUK1b)udz0Pvo}rizO0T%)L(TbHT7VZi1QL5wj~
z8a=IpFe)>RiNMmm!cTu_Tr)cU!>Q<}7{d9|(YhRdvYO)Vyy#8O&i3~1u61POC^ND6
z@D6`bY>YGAF9*;t4V){-U7|=(iSDC~eC<IkMC={(UeQ5nluSXvc&J;x=V-)S1Qrdz
z0`ml@7lT5J#~%jshOeJ7rsUITJRBCk##EBuPji`?b65W4aDK9CaHyB^n6o&&&02t4
zO>2(fu_BSA@7-jkw*K8LsM%!O6!s}vENA5o;rY!*9(&y7tYMt}J$tD?^W0dK6{MTm
zziw_QqDf^R&@rec|Fe6QrLUQJx(_%W@V~)TCbn)q8l^3|j1otgXXg*Ovi;Fr6`r^k
z34FzVeI)1)af?x;U5<^gIO^q>>pnzIl%DU12KH47)uCN3RR^mPTOlfnQ0E7$yr?22
zPc_&uRjU67_hb=bz3)h6J+)gok{f@64s0ZE1@?*mpHKqu6HC}Kr(v_&HS5Y8nL+2l
z->U^pfc}`=|6x3I8;<x<MzhDdPAg*EMRxBFcnjQpNUGpEu*?a35?3Aa7M41tU=d1r
zTC!F-Kju+Ht-0EM!>#3I<>lAC?R$=8ybV+rGK2pDw@Qm76ke%o`;uA%{Rb_;ce%UK
z#yA~tr*r!Iv&6UV{#SKYW`5fWpV(fD>2IV;^CN?;5V}NBs?HzAW9k7eeLZCI@8$LB
zqg8De$7HQM8q=-e+KT_$+9s6be^VYY?MMDPeZ$>@p`m2y)8c$!X-eeuz~2`9w14V)
z(qi2c*7l0aCgok*#E=%LVSfHA6js2oZP3X-1D*~Q(lRy}epmd8DfJL9Zraz^5d9cz
zw)~h&M>Dhkt~e+ZT(6bUzDU%@%f5f+L(`zRJADY2wL}^F4}AkM{mib?k<x5KRyJ``
zsY#-_W`(6Mh35H*#9+RrQbEz@+M}Wy#Ks@UK{>kJB<^`9s{%<|Tu<HsQ2H|a*L?o=
zGjFIx8bK!!;ss^M`}3WG(P!b>dlyaqVn5w(?@ynfPgiKAM+5fE=R~j1FBjWeS3lJ+
z!<cC&3Gt^tE+$kvM7yh+_;b^}dek2;-6Xr@?!eL%Y9WKseuy_1;p((sVM9G8v1m0-
z4I1w#)&QOY@ygG(l)-htjz2%;sAA=(4^o)XmAw~IvGLW-fg()GDC$M35#EB2M5Qr)
zw9XLsG5wQ%rE#(LS+_UdLYzmNUeX1R)fM@8*2_yTp-suAM~F++H85n`=z*tOKrGbZ
zuiu<~KVg@@qg+mqcpIhahHW=rkdXz1Wa3?gf(05icub471J5JCM2ql%lY)U&EAQVp
zlSP{~|17qVi1M>2*ZxBOmJ_L9R}eF_>3SaX+Tj%XI%%Vn#R7ffnC(s)C<ge%>arp=
z8ut864h0nG{LKu|`icC>!WR#7&gsw^6g~f1aGvIW5Qtq-aLS;yt@ySeJY9I}q?KPO
zZFMVbK##0kK@A{I3cc?y+OJX*^pwu`J(x}0;$3O8E4-HiL#<z%7K=1a9dQs$A4ZSg
zVyHY0yRuq&DRLeaQ^p%g{})!UVIODTV%<9Cda9>=;ILdT5s;#{9>XhBleJ=9pcg{2
z82I&7C(`VJtKY<#e=z;MM&iM<oe2q&zXcCq5QI%$J>xe`ASL>dhgjhH)?=Fb)~oM7
zD~jhGWD4hCSG4nRrHI@KA;l~({p;6{17GNd`qT(UM0&9?B2{<%J^s50*A4Rm;fbeR
z($yZPkvRS{z=7B&uCk(}0_--d$r#$qoLf{>!So+L*)Sq~A0lB%s*f>hlc7kiHgzK)
z1G*`E+hqu{j#=~EDOQd)15T9JadU6)8Wggcy2I$Y??)352qQ5=hqD2t+$Re!<I<ju
z-CXfl`WuzEA=hYCFq|pPsyyJzRy%*0ddW3R8MJO-ln|Ad9`G-hcrmU%_Gh21s3KiV
zKHi)E1g+WafKOr7K_(X$gp}jYL=z}Loqnhhj@hZ90#OYp`RK(}E*1$ktYCH`JyScM
zGG*B2WQb-7kzVzA!LNMef3;8tqL>u@ts*P8zy3P<Dz#;Bm`6peWQvQ6zAy+Esf7MW
zf9~Iu0z`vGfZ<Cha~Z|Je5Bq<3{s)qL}%RukGZpia>`@E%GSALKTblGs6~+gkI)`I
zR6@k32Y$gp#Jq}Sj=8_aar0{KSxoh|gY@jpu2u0?5BSQDqD4v0?cGX3eZNMxoVqrp
zkVmE7<{8HGU4zbB*HCttrOlU7O@Hzr{3s8#hln9>U>|m(;__LVh|_D-Gn2%E9S)LP
zqUAT`LAenOjxbP#GUOAcpgv<je)KRq04Xpu(5)oQ_bZfBW3Z1HVY^P~NZFeh5j`#e
zD#*=-Yroyl5!d@Wo9f!pp{LZ{MK!Rjz3+FJ@qUQl=q6-EyPP;!=575%RI*`*niYBi
zH(omD8L08{iX?(ISIRT)GV4JS+_RUHjYRQ$S?Y#%C~5g=@;`!e3o$gn)*dIhqS#1M
zs|ud7@>9Jhw?NT8@{^OWp1yQ!Y0A%A5jAeJD<hmhLz?(I<n*7f8sMR@gI-Yl{72a)
zre$YyolJ0kl|J)_YK=TO799Lgt&|SY&9MdLJ7+0IrvgbV<Ah^qo2_t^omTq=w?T4-
z+RNjLCgT)slI>8!R6yhv9Jm0`i5{)m#k=soZEAJ*KC#-KKm|rqy*+!V0U+F^$vBRs
z`l=!zH!<jEnTu*9ENbOPcuf=OErJV2OB|>k%yXbM;{e_Y1GpZA^RYjKt6;>ZpK0_!
zlO#NZ_st$a<lR4V6Vm&ZK&CrcEGL-etspy?_uIddVgK=PfY~QHC*ngf4exOIY2{<o
z0Ir3$>#?hlj%|UtnBxKaM6|36*9W0%AZ2{t1Tpl{8Vkh+x2J!)-9Ca@Jg@j$+kAwt
z1V6s+<wU?G+^rbYo53PzBBgjJxOlH@2-#4&MxgCTZ$AYxw(@3c2Krlkr}+|p4swt*
zkicD3i@a|Ja2Xi;XonvN%YTcRNAfHP5@~vvXiAyJ4OV#R*k!Wf%FQRa<{B_McO70d
zU{V=^7XlQ-iqFl7O}NPOSF3821JRWynV(!(XV_65-@xpE9i^raLR91Jow#_sAA~GO
z^{kMeP-4rq$2LWLzd}A7mgJ7=QsYNk(K^xiALa*G!0z;PyI1}$)35vpw8p9UtqdAq
zFwA+89FY_ya18N0^Q78SbD}yQjvSp(;K-tKsj=X+omUMn)Ac_Pu>1V`3MjKWbh2%+
z&^DpnN~WA+eN%`+N7MSRQeIn!iUBltrZULwzhK%oAegbt#+iqh#Y>G>8reqg5d||8
zfUwkf2$lgNph1V;$NyE%@BbV36~#e}Llz3)_<JhV!%ul)V;?8HvJx_wGbmLjhop4l
ztRCqwtTS}{7y1o8JV;+qvC~obZE2Y9{D=FU6RznI8Lu5H6KI#;iw?M)FoFXemlDNI
zU7rDR4_6ZYZ}10Dy}DC=W9+fWaiBHgBi@jkdaTnpJ7AyRQ96xc=;9|D#jil0SIHve
z1QGm$=8D-*<0+2CI3>w}$hItx9zsYYu`1jW!@;4+DYCZ+LuXfOV>3hh>~oITeT^4&
z2P@<w?0jMd0thPxmZ7j%*Z?@pQ><*!bN<Lcr}u&KE?*e;bN!gV287*cqg5)^+?JTN
z2n}~6b0MCwVn&|{5lw}m$Gld4T@LjU-koCRt58c7A){aunAooRRHIT~St&8iot!~i
z2uXj9w}rRmI8Xo+3+Jo*a<>m{PAD@|6wIlxTpG>FvkRnZXU>Y=Api^<&B>Iu#dZI&
z?`<GN<L_;_zt|RYBSy<{vhS@18KeB%G%s>Cv#ZI8xHmU?u!qd5Rp_{MCTb+eBQK3z
z$$tQzyTy2HNv&U0%cab)<xwVwqES4#8fxiyD(f5}4xC@4@4d#9fU#kv=%DpIzdTAF
z;7-Qmujavlk%!qa07;E3*6RwAx#gc%@720B(K23wrPnujH{mTMX#TzagcaowX}>jp
zCFD-a`0wJ0kYOVhtGR(&k$?oPJ|P*S`}rAbcMB99pGK5n*{Qn*G#dgd{5vz56&lNZ
zV{^vB<*I!eZCqkzI{Zl+G8r0AV_gl=55<;<`xQTd&JwKN0D^eoBMMj~D(>!Az%@<K
ze}k<oVX4LZg9)^s1`4Enf^vi8_|X(Ixp?c&6gbm6oiDzqZS!oLKafI>wQ^7cQ55rC
zC%l1a1Kyi<(JN{+)!;uV#|M^sUGGl1%|3MA)ZYDt5gkFG+|PzBz=VhW5DSR;7JNj!
zM3<RLFj_2y0WOr4RVgu{eW)SR7cboBZm1Wyl%+dJP`FUC=-YkU%hqyFN8nz{io?&a
z&s9XC_0UKJn7*dI4OnL3+vUyP0NXK$r34nyP?ABUk-5agJv5Oi@^58l0y@vz2J_Aj
z9r9EjSpg;{H%j^3n{U~BIn(`D2uZ#}`ofurGnONB!0)=0Q7+v>y29U9?h$_$R%bWQ
zh{Xyq?osAXxE^2EJPr7lrkv+E;-S}|epGI6s5;v0tF*ROt>Cmt?pN{s0k6NPI2jF;
z-5M;Tvi8)YwzZ3|(7W<FcIk#2=6{Q!_M8>}h9^-Um5|I?d6<*6Wx=NB;?UXcL4&Pt
zlfMH_17Q56nLinNtlQ>K93w%7PWyR^(VK=n!Nj0aE>gr(@+TdN#5?SCOXGHb1amrr
zMK=%#oX24vh+g7Y@4h{RTH2-u!_BG;rGl}&r^O?&6{ppQMV95?KJFk0;Xo0QxB=;v
zF!Mo(6~EY*rS&-Hc(<U-Yc*}JkutGSZxpgAK<fhTm3T?m6?#s80Ouyq2GFE$&#CHF
zc<08iHjFlon9U1?7R<HnK-x(1bvRg9t1d6JfRbhpWs_sgG=o)a5<AJ9l!cOTA!`$&
z!WkOP(Tg%?L8}EMt3uUvizBxYAZer}7HAsp0$?0!%=(HtDH8YvxdA{7UT&@ut)GV;
zfD*$_c7;viuZl{Gfy@Oe;b9S;ysJQG3jg9?(10Fy_jUjFbnzESx4cvSc(y>wVs0NT
z#PILU@bA_IxPA!s%fGhmW_64%XxKiMF<(<`gu>xPyzofRAx6V*n?xiQ6$Q1&jL=xx
zDFkJFS=k#1w2T`GAR2-)T;v??=54aD0M9{pJ0|{kx`-<Nz_xFgxM;~4(s?}f;8?m3
zp-7TpmaseEKlYfzC7rlnJOO+O1}(B7BWs2YQQydQ(Hb+P%UO9I`g3PM_cD{gX`Ld2
zB>OC3?o*gb+J9)6|F)-&@`&;VXuB``B~a*dsIrdE2ykb?k4}S70i#%Z=%GAD1RSM_
zdOVth${Oo<D;X~Ja8?Grfjo(s=(3bB)V`MkH46VFLS)Dr7K}pDT$!|l<As4;0-rZ$
ziwtT9SqeC{s8*rgXhCHBj)$P+1)M<I-H#GXeTMzVfLudEf&uOvM4I2ir`l?W*k@fu
z8pVr$8@-*l8IE#|8K<1KYE^LnjIMIdemjH}uN9Fmxd}Lh^$fPZip076z$S?-h?#{2
zos8v*FO=7m&D7&7$!2>7-8I*>q;yY(+e9?yA^}RB7g^JuyY9N4mDPt8k2j>EftL;n
zi4S!Yog$8}9<CF1;iil+rF%%70G4;-15d`*ZYBbQo1;rN3jRXCcY;1ZpBBme6nv(W
z7-Q$13jj%VS0ds|#8L0WQ#-{p&~3#mYVBjRATW|Hx0F@`WkKo4en)kxjHT(A#}pvb
z7*BTJ3b%4K<|3lK*Cul)WPMXu=jHTo-^Htp_%Jq0Iq1e1E$=(@HVo&qv#3ca{@k0G
zkhf+<uUZmM%vIN_Mj{SC|Ft|i-zD8Lyox!-8vY$F`0Kr*&&9EFIJ)N(Bi|arC7A+8
zJEu-eC!v$#7_Z6Yb<XfcM{|Y1{5@-=)FWCHm6ZmGwN;cxpSaJQ;YwAT0Nb%4=Swtq
zt!H^~BVJ_OpM7p1>c#8EO+f9FYIwVE!Sqi$&+hn{R9F9BxsQB+dc^@z(pH>Jl?0ys
zcFj}Cz)ySmlqI67i(Ssi(%WIgYmkVc2d2P;a#{$Jr+cfNhc?I1qFgM$x0Jkm`K5lX
zH;wiz*DJavjDN`po^T2`J+P*RVSYqhG>2>e#^hIxgQXT$nS9!+f+iN8!ju=u8@>Ue
zkNzy2sbrQgbSD%bq<s)iDc_33==oJ910i<=_{HRliOd+OCv>Oi4mRDBw4?c1kTv9S
zq#Xh1Jz|}ls$So9mr>jGJvE*kE3)h$aXfF{oP+Xsbe`>4WdY_4n8t!W4vV(bwxL7f
z($k|>)Dx{~K7lY&x8>ZhN-rWR$J8qfd12n@XSU6!zh@JmZ*x{|G^1%*&2y_daQj8Z
z#vXv)`5A$-n>D6IGw0C9Q!*{An^{sr=6DA80d_ohj|@ILnzV0)SHq<OPHe%(saEb6
zJIz*@;_O~H{I&i5m|cb&_~)U!--p7mNsc$w`gc${$4OfCD6r|z@B@z&;)%RAsXSb*
z6kKG4Z4z4m`!^Fyfrxm?K@5nL+V4v8^nRmxsy;*0{xjdwHp(moZc`<0_@+LcimEfJ
zJa_39_xiH%e(1l-=zkNcqgXKlP_rr;SOz(bcmpxbul>PK90@0pjwecdf>_20uaq$e
z_AwpALn(wZJMjGO`lIM+T2j@80?8sEA}X~-=*KGnNqnqoi2GPUt!A5w8{m;SjnnjG
zqFn1VSdHpBZ@X%^R}AA0dKCpt0faNeqipJse#U7iAiNzwHMz7Br4*PxVG0=2v?t~n
z+~Jepo;6;7gSqy3BRE^+)LnrOPJM_+Szw*cqit7dq6j5o&6hY&v>*PqsLqm01W^#_
zsHsi@&PjS3tV=d0nicD8rhtT^n_)osV72|4GQ!Dba_ab|%Ja|i2a#X7u25l5Em!4h
z`kW|ZMw>Ac%?mohR}o)T-ezF^8Cm{e6WS<kzE-qE7x)^jNJMG^fo#XBz4L+_tnTL7
zLj{{KPP(#mt!xt6*PEwB7i?LyAM*?2@b@GEIO04Klm~z9j|{|!P{p~7*w_&uXYeYa
z62-wmFD2hdCBW&fSY2Ok?(C@Jx|NQj4dGrX80T6~JDlisEV+8GJzUS!6>oMc%Wf36
z!+p+ed)Q*_yb7OXeNHpI(?3X_@vJHA^ks*;xuvrtyIe8&f(HEROF4z^y3_JlzchmZ
zlo*^YN|X%jp0(CH!H!qmJWXL;eQ*^c7>uv2ZTa-HwCLMkuHJx@LWEvg|5EW8>AA$)
z`M$jlZhuMSTkK7)+M9I5Dkb_ui|}5Ir==M!hF0}U&KTIhbT%7>@(n{(N6yE@&)H(A
z5t66=$|_twOF^sOjWa2aXI|Nl$@xqP;9sZq@_4Ru{z$KQNl5BV6Nx3&M9wr!Vr4Bo
zdzwB*{Lk;vwi>x?V05x{a65zK`k9LS64)`fr+vrP=NQKhf3G>uX3n#LE5BV^XCKuP
zOqX8}<En0rbX=j?bjY;lg7LIA!$;K7A{YsBhWO!Hl$0pmn-pA^7GrmmhbXrcplVTc
z`Jw~{tnRLC`YkUHQ?x7mY&x6!vtt{t@lhg|XqGoOij@rv`vt|60)*d*#kCFm7o$;1
zx#zC16Q-(U$IODw(J44SJ=+M1E57)L(2h*x7f2=+@Yd+Z3z`$<=kmF<9PtKdu_ru6
z_p*IbfA<2TZazHRLa|laEYGtpz#&qM6S{TKD;FsV-T>W=@|~je-|v3<=N$1k_+h2=
z>V^}Lshjm;{y$b|_u|nWBp?$MX`t0&L=l4FE61IAnhA>{_5q-Sz-fiAO4KR~-GV_W
z*{R`z<6+V|ugRV#LU*J13s2dGXS$)gBDZ}b(Ho}G;HKyI5xlt}>`C^ofT-^47gU-A
zyGY@;!L=jIbnz07!3NZ5Q*3Gi4$v+Ft3Udd>XH$H(0DpWIpE;sYZYH5<u#2yP1Wg(
z%9jcDzk1Rw9sQbjkY@*6)6T5L@Okh*3S{?jpd3sZY*^$yahXI6gUxw6NS`BU4mSyo
zD^<wT_gYod3{BpGnK|%N0c*Y40d+LmJA;7?aj#DJ!Q~udQD&ph1#W`n1d`OzCP^|^
z)G7biuB#kLM3yii?+{U0IC#^hf#?NzoX{Q$nC2>AUdOpQ(0jY(ulq=|kHvb%R4o!P
z_#73Us!hFTEj?Be$Gy34--}JtqK}=Xp|mQ3^m{qcnk5)tk^sV8fCXcw<lxMMKhYtr
z#SGg%l8^!1So>)6I3V<&7a;DQJM)%A-H7`5-JQN>32^=tv*9|~YC?w|$Y~oeKny8q
zYtdvIiwgKiVxh#p#JcsYnNMtUnNSkf(!EUrGK``+-`{&bmsTEBzjdn;0}8y5#Lfg2
zSBFj2a9=?}-Q4f=0Rv4416e0`gu#d+{K2_I3QqN*E2bV8L}rLANRzVZl4OzuH7VFB
zLX+AgycP$!CBc;UimYcds7!(MYX+>WA<GE+Ln9F@CzJt6)hqu(OXfbxQ)8CCUjSkG
z8RJ)7$As>PyhQHq5JWKeHF>jx!-o9ry!n$LPKfMKI0Wnp03LkcDGZBQ6k&1YFJLO-
z?IVA`glmqnba=ItzzI{n0A}BiOU~l_F~pJQ1wEV1;=%eU9=X4YQ+VmdDFRSn*GTD5
zasr5D<xF-Nl^4fdImUm@cw$H4UKAhGl5sZ{iNuzivzTYxxH}=N!2o3yD$)tU<wjMa
z;>M+Nj;0$=faX?oU#eKZK<Z`)!BYSnNyOT#0<mqh{ce(&$Jg^r-B{zFIY#Im?ZNJy
zT`a|as-uDH%@Ak0F_xYB3|=$8mz2^_FTQ3(nAfo0di3uB-@(s@W_7=M9|aDJ^k0B7
z`68l_^Xi#NKkM>(k5O*I?EHkvg#Cr=@um9ocgz|D0pSX(plAAP3aM4huaZo1+FjQd
z>8d}uja8cYj@cRHkqL804uHEYBcXIxe~`+oR^RXiOSTy7@CXB~i{hlF1}zI3Q_j%1
z4=j_hDG%#1q=|>#3sb|+fXH1feJYg5{AVfosY3P%0$Z$Hx5dGt!XcABYV$hG-;C9*
z1cCrrK#P$w<J@ZH>Ll83i9<o7+*#o9t;yl0Y(ImLW61tAid(|4w4%=>4xJopNlB&;
z`w0U<X+P83uX(1hl@zH96mn#<a#=*nP*ws<p^gIbID1@W0+v^Bdr(R6A#NLmHt@ql
zKG8Nroe<c7fW7Xripgmi;97?cLu$XJBv17%00Ip5V4!_ph#?KRp!>~=<_uC@fChAJ
zK8vvjH%$qa&jJlW6g-cCJ(g%pOm>Sx?ph>hT(JWtkqRO;zse!x`J^!TFd{{00;T-N
zBwb9eM+T)J9$MFyq-nP<^J)RKSNyfRp)Jh(H&kvS5EL;ZzOR`v6PnYJLjEW{7`+@0
zpq>cbi&*s!t*&q7tsK{aVJ4Y@%c=#gJhp-WbVb<&pN5ABbgeK24_@uL^mn_JTaod7
zk;~)Ea#^FSeU6I=s86Yi;J^$$At557kYZ>{349YIv5~k}rjARr=T*cs8P}O!W<o(l
zwzJRj_0x3W#kI>Vku>sPR7l3xYO7i>K&4H$yb33Pm?PJZl*kB#327+EHX5D{?UP{c
zToRnJa7X(@e>&MobtNj)HVoQ+BBV!_;>QXwYRuCLnr1CzKqdF>uFp_nbbS#(`nm-g
zRcU(ZlS6pHu{Uc|zh^WcNbWUmyxpfQ!gy?_{@vZ6{ntN`NhBB5E3#_~<31-2m>~S<
zZl10#6pm!2{r$Y$u-8?;vq$-1mZ2q9q(P&;2e^#bn;_{yX{5W40TTPkuhjp#t0q8_
zFOGXjic%6KE|c(z+zo$Wyi-%(D-;qTs7DQ?MF%njknJNoz8dK$ZJ71!ViXa^B1$p<
z&+n^^&phg9-wHpI*W3RRBOrbTn1=d*(;uil3!7s(hI8t2$sag!f^NX=qCdE4Xa5c9
zH$J;Pe{9_1nyRA|k(3oOsv9Qlnf{v(%gD|}a!SaVcC~WKgk|&i6D)djI*lVnvOL5C
z{<_f*`Ky!6*gyZCfC|Y*88=IL$nm>a105Tu`v_eD&45>>@w&W{ZOn)Qhz6cCal?TD
zySDli`<OUP6%r8M?H}4qUc92e9y7+`!`=Ec>^;rilV_e5nhqWY82sr0pp842tj@;Y
za(@6x9K)h3Pf*j+s`0@J3If5<7mw|@ysWql8X@pip_`U~@R0VYsifj2yL3qs(58hs
zFFBXoqG&>B1gUi?@~oi%ERb+xo26Kg6KivMrHkx$tOR@$rSdeS#a>~p0_bM?f4yTT
z(3^2BLO?~CSOj(D2z%e=e;={u<`2a(M^<{G`vFzyXR0Py`utlMSPNMxV{!-?s;fAY
zZ&))wP0yHQ^>~hP`xWz|ER+3;g+53o-#SETO-hvzS9ZTENT_NKsAS3;WqmmX_VD8*
zcy@c)UKu1p@W4Ei+va`AHCS8Bd&Z;lCWeG8<i_Jp3(Ru}=y=R2QT-RP?tAg+nAx2D
zMQZA}u1cpYDpT?<<R49sr%C=Wv1QrfBM&Tk)-ft7^BH;Ip?dDh&|bJjBs1w1rL5C6
zoeDF>NWgFIY%AXkC^rwUm%=Gq5Sm*$&jh=|cE%vQl5yJZ?zC}}_xJb8A)6;tI<fBX
zn&V30?t)*Jq=_H14y`ZJzpEaRlNYGNTWJjUU;ec7=0Gs}>ciZzIH0j2A4FlG>GObZ
z*Mmj|!2S(fEcaPWG9x_vRc*SZhGi;^2xH%S8oa<`s4DXX@P}v4bloF(+?(aG8fZrB
z3Bi8f>?&C_c==JIFtU>sMm*?)fq(!b`Uy$HDOI1$1txQ77C;SCD)Zn(DjG#9DPTYw
zCt?;1Viv_Rji+||o)J<8Q|gn&Gl`~=d`1wBtHOZ8DHi*k{jaE;Q8Y|&SFjh6ESPfT
z;nWZka9kM&W}H}-1wTy5Bu-Tt6T~F_-*ZPXO*gqv&T=UssPxGam?YCkz$1nJdz#Y!
zIn4>Z@Dbw|4#*^M-vye#955oq@4Z3Byb;n<LUEi0wJazr*^E|^Dd@exQlUln(zvTW
zcTxD4CJ@sWJuiSWDK2X<AlbD%#iq$qW6Mj9Va+mx3yI~cADX{Q)H-#qqCf9w<k@{^
zkSfwPE<&w?nCbHGWYA=zu-X2nY+Od{&!yHYml;>81&jus22GJ7JCqeTTcvU~HHDeG
zMQR-4fxr<n*liYg01T(sQC!9|8ax`zU^MWX)xtrBzz-fTxOE>;>0c@<{^9qaJ;848
z%uS8-Jvw<dK_Sxj#0Y{x!PKgh`99dg7ItCC#_zkD5Y2Dq4`)A>0Z<+~V$r8PsR!-q
z%qn7PYQ@xwAr!j6$V+HGR{2<f{>KGVG|#LmW-SrAS|pw!J{*qNC)QJgLq3%Ugnrn|
zV;Xpdtm~R9<Ix8Il`BM^$BZehqOuu`rGd%FtpqP&e3-|mseA8pMuW@r6k0<;qY~>2
z3J4px?NM=~ObewD)GM-QY0M23MtXunVgr^3o-PQ90<mc&iewcIvIRC2z6nmdN6GsL
zR>z;3mQWZk6gpqIxDl`kr5KW5sP5^$gm={Os%$mRKyCo!qzfK}_wxUdH4CHR=ugtD
zZZ$AP7ImLWOoZ_oCwx;QZz4GQOZKAV0(fypXZSKWUpYeIct9Vz6-}3$e`ZtSZ0GFY
zzKO+d<GhRMl-)d|=9e=c(hrvM+B|O>m1=RJPl6e<p-#_vwooii#Ij)<C<3+e+Q`}O
zsx}|cu!#d?QB$C(zN)PnV^!&X$FU3xl917eOXEc)+Qgwpp|UFdo1BvrA|bOXWSy1t
zaO$@xES#6L%#Dkun>U-6^w9Ny*P`=2)~?fH1oXiB_B9vhgBakMn#d%=WwM@ECGZ`(
z3vubC;I+uWC#|b|68}2QlRMJ(J;;x=luhXxZ2kkV{QFViIcX}i!O>!OP^Doa)WbSM
zWa;v|Cf5*~tEc@kP9&^RG6ddvL*lS=>7p4x0asBKIwV71?WtewxjWz6@=pz*!H3Lh
zp>6W(Er9XKz+LKhbEryA@=Ka^PjYJ-{ipNaPpb%^bSK~Jl#$<8v!HZ|5)!6-&*hx0
z8m%h;6kg9w*am8a!Ri>yhS3U+2>Gdqa6WT3b1fl!x8M;!E{9FyGO{y3U%o<|uH=8S
zx}&ut2NWgv7E$6Cl9xZ%95*zb7r^(v(H?*$FN=C(5Hkl6;@1w|dfzSfj+$h7Xy~yV
zsGZ^HeN9BqPdSz^@qgY~MqwWeSgVWk5rE`NW+H^w=l<``urfM9Cp%r)<l}W=8&`Sj
zRj_pT>HrKLr;vJ?R0n^bT3J4xT3A<MJ$C91de&;~e0|bmK^7xgT3I{0jDwYNwA5BK
zE=L-KGAFZNL!|ivGCeKPMFr0u7m>KSnYzwn>Sv`RUpmiPTll!2F8?5vm8zzH_zTWs
z^=adDk&M&&7q!doLT|6MR)46xG~XZ-UKcN@7^>~tS@_v$QElYeX%r~#!dz<f_CH!6
z;M8bzc%TYc;e?#fW0gopSyp8Sl^9l+yU70*fesd@DdiUe{rPyP9J2*z$^<V}GStJ`
z0#Lt-DY`w&!aY7WInakYzd#ZI(mLQ#9zF~;0!U6WX3-QJ2K~s$$#LMDO8v+g6E?~)
zM*T{XuQAXvhLTD6Jh&*yWq$vPQIc`gYg20@AuYzabXI>hy#kE{k(t%~InXoUpac`P
z^Y`<>oE&J)30^k{4a3|t2n)lM9W-IwKA<onR<awDl|4j|9TKD_Zx3n(c+iFfApT(z
zE(f}Nqhj9>JEAvZm{0ENy;8%;r!iwl?q`N5=RmmSHDh43=sNnTt1x=`|5)(UZQmkm
zlBue_ZR9%xf3B#d+rSavjqJtq1uQoeN4Zhc$(%PdGmi7tDKOQM1{DeRe<uEmGNS+!
zx%08%f|~bFREav<$B9D#-usOX2ubpR*fyL0<mYB6pJ<b@(SY2(F=ZqheiL^V?DI+f
zf_0r2&~JxM#E;i??v8l+gnyI)RF2DV6*(BREA0CrG$+uJBzEJQe~xfQzHwxM6@IyC
zl56K~oqtseWbG=^W^NNF);e&TMa|`h5`tZrh=Bqj^P8R<EC6W$tSdbV%aY5mFzu#m
zEP5UOjDl@suecIVYy6Jc2d0B;#LYJ6nj9Z*oHpK;fgn?xNrjmxc@4I~o+++14!%2+
zbhW+m$FY4nVwSuRhhjfi4~EdUnE2x?N-O@0wGO=iQ*9rDDf#bK@<oM+#(_x9DzS!!
z=`D&);e^)j$GW^v09SFh71TI<`H>ctBTATo`=)c(U<kxFv5F0jy3I^=45$55`_;~#
zF~YY4!?MSP1~CRjAxuV<WR#Bc^cUgK!}5{9r?x}jk_-$f`xI7dA2oh1)r#VNa|nv|
zV?-b%QaR4->uT`nJHry5w5Y34?VAU`*-&QOEck+oRcksi;6h9l=QZbNYF&G_YKv(B
z%9<nqJEmLIjCsK;4CaGOhPHE3kMLISd={@|rfAyqJ)h-fy(y+hzQ$*fo1|*ub%KS6
z$GKTW*Q#`}_#DGS@Ru$1&OOf7?=R0q>cfUb(~3EgE1CHu`W{J%u4nbL%a~)gz$#k(
zt4gtUa&BT00Chgxr=Ihgle=>^YAQahNsf2ZWc-clUg{%Fhxe6w*aRJzA#9{c<q9G)
zBnswOs+Mu2NziDo<S@ELOSUA023=D3RAyqDMX+JU%NOG7nP07#;y@{cII|j4T$CvD
z>cvlain~uke7op*q91+arv#kotbH;s+q?hgqC&JCfYB2-z4V&5NH*V7tz};5Ce-cB
z5M2}P8C!@@fRS<=ErTLZ8z(^_K&-?<%>Qt7WUd04UlY^K|Mo^>Ohv@x-~F}ThuA@X
z=IeQVVV`r`YHYsIkP7oQD1qQ`BfqUXVfu0{>xZ(nv(=l$6IyE$Wa@A-ed!HWBuicR
zb^HMas9O-ch~et-U)UTD=V_+OJZTn-edp~QrvkwApG_Typ4-(8CgJRj)2i##GDF8h
zZk%YRE&7r&1#qmaZ`-lqW0vXrxicxqAASr%%Pf6v86#0n!)q|grljWQ9)j|Gp6*|s
zcY<oebuG$ekwQEUm7G-?%rII~V09@GrOAl_68dA%O^+c0tIS%RduR6LA1VHjM=Tui
z7sSP@krYLXk0+prlaM9-|96Sv5^}7H|0Y|E^9`{2(NjslmmFG;9C3n1rt2=q7O;6n
zoJxQ<M)K8Z$>a2!Z2>heZCiZg-3P!tck#`4baqrYbq*L<UaM;CWVnI0rU$_3PT;||
zrY!}?gMFZOv*}vKkFvrh1}w>Rvvvbh8nsQA<mB~n!L$hhjv#cJwZ6PJ_wP%#n#t#s
z8yy<%u<W{10k?z26>G!<>_(c);gO}lO!%BoTDdtr&<ldNvVhzifu`ZplaNtC0RU{{
z#_k=Hp<C~p_rTI*F)ar-l1*!P_9uB=_f&Ajsy*AhJsO@j(OHTjTBsd=`|Ig(O=MBf
zAXxjWHKfR8d*S<8TSkvhzj4o!?F@0`i%n}U2AA&x?sq?nc!?j<hXm9X#=+CHJh@Dw
zjTv3&+B@y5At3C3m0$Zk1JMV`3+>)1IlGe$o6w>xbabA&b?zj=y(0){U^A$)|0koC
zpAQfl)IIO3y;FLSo3Nv0i}~>aubP`mfAWB$`KjBfUmaDIfP%D)G5q}mS9uX<Lv&9E
z$e;eCv*pdh+<SDDrnXzIB*+Th#Vhu;(T)TDQ>H!AJ?l5tw93bwsF&(~xl8=J--t%z
zH`zS4`aceQNk#z4w-qDY4A|nRiN*hCiVgCsK67nZ5eD6wDaa;}<ClQFk!L&+RSPw~
z%%6}$TUYrCHJ!0rZ;{Sff1X5L3r4UK5T>Tf-ZVjc%e7FOj>F@)PLFj@4a0bV^#Ktm
zD`J0$adGiRQLwaaxo;fg<81LgUC&*Z)tK#sf73mT*@NixahOOsuBHT2+StteO_lLk
zsn0fdV7CT_?!7Aad7VH?ms-opb!Pm4K<c`cOSIc+uA5ki3%zdq!n1lAXb!}&+B~v2
ztLph)*V#v*Mx5v5?_voX==dDKq$IU8t`1aO&S2wWvN<qRZ)2hbkffZJw8P;<J6j@4
ze1$YOKAEh+mf{c_krx{vNQgBji3wS|*|ILozkdQEOTcg($&`EiQS^B>lE<n8i&k)U
z3fQ@I@E$Y~oDw$q03RRDv9SmFmja49Hm&7Up1a_k-#(EPuSIpX(3~G2DV!2EyM_v*
zmhU71W*x!;Czn@ngOH%vpM~{?2H~NFg_3|V-ftghsIQ@Qr1*N3T>&+hQcyAWkD|eu
znGrY^On?p)2rWstOy^UC19FBk({Q_9qN<B4%ZzKSKOLizU)0~9nK-73x5Bp$<kuzZ
z_}BYUXym2n+2wZ&%XG;AG{>!v;+2~nfT0d5Nd^b^n4EpBa|r73rG0(oW#xzI`#!K&
zxLqKVp6er+poQlN>JE;Wuk{Om!p&d(?gs1r0f^z?=aW<?77K8^D<n6pzAVnJYa~K#
zASiaUNnB=j!@1y)oukESJUBv1b$j+dMw|Jb?qdP@TiB~r5*S$ky$IbepMI_#Sch+^
z6d#b!k)q`}b0^QYqsx5tUpO%Z9uc>q59q-6Mt->-8C)j09vx2fX)N|K_DmKsZ|Qkl
zBIUxn=xC!6YWVQGy+HhyZtE#Id6r8r!q5lJP&}+QNQ5;&P}t2())7gJOp+eH4MiU+
z!=}5e*1Q+yeBRxFol?9Nfx_gO6Te&Mv`18hp#Fb_e)g^ib*bo}9*jsm_;-5EQQ<?K
z@@^SVa}-hKq*bLQKO-~+FW*83*cYDW+Fh0FOx!}n<}Z}6`hiKiCAqbk+4UIcn-nEE
zJt1hbndsVVs`!^rMwCFi_MnoD5JWc~Bh|3dvSs9ogxc%@%*3Y!B(M_SRmwa5RH;l=
zw<x&Asw!+9!UU`%F2oRacGHher9+DZl-&F2qTMWPNhM66Wd_eJl21=Hg2JwPX7!Xm
z@0yM$)ubc|1nE1kj|3cZ{dZI-K(m-zTbjt%G<rkVw|<e}P^RJ~jc&;Cy>uxd+ECFm
zhwKq#i0lebIKKZ+xScdB0zb~zUjLasV<)P%pSa*Ls~X+#L5>$vP63tyeh5Be`A_Li
zt98frc?y~t_aJjnefk+<9=6$mi&E;nhGE+vbwa@!rw2>2;-KGywO_4pn4j(-BGd*r
zh=3rMMwZo6ZM!<W8Z#`E=E)-v%Wv`oeq8mJmy3OX9NTa|Gd=~oDs+FWCFic4a*&>g
z+STWmo^8t$E>WFOgCTWL7a9=}^~t31h@)J^DWq`=9o;hT%0Eoqy^gyr&eT^(<G&@-
zUz2c9`xwI!^76zuJHo*4#{A7KpG{gE1*8Y06qK)EI6$VlHL<y=B$oTX&oZ7WhlU%4
zomvF|;87N{8W;&}dgmuy0+l}h+(~hr<I`Mv)=Qh!PVhu?N&d+K;oFA};0xvr^(+24
z2@k_FKT#=uo31O1lt`4bG>?<DySipyE)HjJM<_}6@)<F!NeljSmA+yeLZA6s-MF&;
zp87Yf;?ay3=-Fg}-Q}d+etqS+N&04cYl-?Dz;o+#4#KZ<M&GRHO+@^4Pn!DAFszAe
zBiej{UsDgzID>Dm|G_qwhPCtmwvI@d0`G6)uB=DtYtP$XFn&taeM;1}7Bu_MC2tof
zw6TuTC#uBm-U567yr%H<%zr&3__TQtSx1|^l%yStqzT#xs}QXSzX9YRn(inx9h-R8
ziU#ql5a!qdkyp`rJ2ATjh)V;>#So_k8KF^@vjGylsucAi-0J4$WUFd>Xy|hE4mxXi
zJRk3k{N7t6SmSPqn+jI$7|Q?1PKh|yR5BVwCW=GpY`m?O8g4=0CRL&44)^{lg6{d>
z-tP7tKF~c7{|S#QO9D)Z8+eJ00_lRRx)X<Ws&Dal_Rw1&Jw2qWFR_0%>>K;C1dn2q
z#~G-bU6HQ;v!osZ@#0IQ(_%;*C}7%WUf%a}s^5t>X)J@lF0^oTAasYY?5;u(m0;!;
zuOHsjQqEeQu8Up$UR||iH6vtV!So+|q`*b~Qh*_3YAU=45(Suy8(APs%~H?S#VY3R
zw~#-&>ZYL&s$?3ST|JKc29+r1zz*?tQ!z1&mOp>5`sCb|)t~l1_;m!$`SlhGnh8X}
zvdcwCVD-L4b{TgTe?}2r)<7!7uH(>f4hhi?GUdYr?7w9GGyUA#wnkB4PxV%Yu$Mp>
ztPm3Tha}?&{p{bd>b99&sj*Ze(bnpLs_JAdVUE|;_M#Man_kz!sktIF(eAKABti;u
z5d3FE63WJwK~mnquU70=zxG<)cOiZGa=c!@_bBHjYO;R^GY5D2A#Q*bkKV@;@*ItW
zBV|<M0$ne5^4kYzed!+8IqP+`xXLQ{QVG)aJ5nMz#RL!>)WoxBgJvPO2HFwxby6y*
z$O{5en*t2t59+U})O;qq711Mv<Mm0x%4^}kbM~L$Gstx@%bfSFFS95Rk)l)v5enA7
z*Lm#lG&Syc`F0m`a2O0{7Li<mAd2A7KK+g(A9X{#M|QjJyO_jNd8bO1BOJX>v7Xp@
zQp}SC6wnj9K5keCSmfmAp8iFv6yqlbKTZ;sn=9f`rw}bzsFS)w&`Qe*ri*1EeEpYm
zz;*e@Y3NmKwH@{+3WB;lpIBd9O1FEL+j@kr7?1UjZJ)Tlq|)`REqPG1-9+6Q7(CcK
zjoM!|uxz(}D1emfH^&rR7jk^2usln(EO8~70oJ_=_@|Qiv(VGDa?TOpul}^=qCO4<
zbNHUTQ-2r2!56x>RE>cbrEU3W3!$vvZ)A&U;rfCJ!}#rbyb3wT>dKKI=%ryg4dEj(
zRO*oXOwAHb;x6d=G|<$IJzgvphRI1B<GJi&GG`YYBx93a?lSl&lpgg%C@#zrys|IC
z0N(w&Vs-OJ5lqP$yrgbcP4YQzQrUZQzUea|ZKHLqFYjjne}8cruBw+gACe*P(0!qB
z9U8GBh&nWbhlrgc`At7RsF_Kvcq}OEMKI4`SyJpJJd-i7kl%dqI{=<O%$*#WaZFw}
z+fDRZ26?ZR1&ZWR<VPPo@B8(%Gwwpt0j+E3am3*PLB;KHdgFz$GAkIiWN?L*(fFqV
zBNSp;b>QoxuTvX)hqtt4H|irMk||PgzNJ?!+ab@csN@Sdv?yI$kI*Ju4TC5#yHnLp
zy3VbPT<x6SPT(h(*-bMWaf!>MNvdYYD~cEAN-#>=6k8DwSTtAhUGaqDdlY^20B&=@
zB)!EOh86r8&LMy(Gye3Ifq@GK{Ip%&R<a5O_)v%v*hibnu=>tSo#?h_RhTJmo}1t9
zWgR2yX4t;Uu4IcVYZDXR(?3w%m_uLq0qc}o+2XBdf5Uu+OGU(LiT_fXh4%EP^!>zv
z+``XEm@dSL=piwHBYlXI)i0w7Pz?iBw9up=F>4zIJ+GbF^0F{<ckpYbTW|IQs)P~$
z57&ZH-nPr_mPofsOZP|Hhl)YOCnEP&cNaAshzXKSK{jAiV=Xk;*Mib0#0O4o_1zvn
zByq%}j?Pr%pmOt}Z{m(^e{O(E?U=^UVD#xKADRpvN*sG+<9^i629E(;0wqHIxkxZ6
zGuXUu#7EUVOBx%AoGlmi<=zb;Gfgl^Se2j>VV%HCa>D8m?SLUswEpjBk1IW?f)mfL
zm)(+zg!ap>a6rbiO0YQw(Ux<c7tslWu0DW0i|5GUY(I4^nqa`lGM?|)&7uhCCF#tD
z=RZcq{H=Y9B_sx}1*p++9y?OJklo+2fQW5ym9TPIm<<ApI<&<a!}OL-u~;WAudRgh
zuSFoV_#}P}17ikv4OpxzhnwpokT*Xs{D0Rqjr*~fpAwykwsyNSR9gndqD}VvB>6>(
zqs>Qmv6w;Txj@ZFbFH@SY3GgnNf3Y16cs==QNvx;1XL6$3iUM{mOUa*9-%y^yi>8v
z?1`te0V|0ATngGsqe$9ZY8+NZ)}444eZtztFkVxdo+*RQ6lR8T2xoYu^31mqn?xPe
zdumLPUmzoXItW64FXxX5>yycVhv@8xXOM%vq$@~tRj+eisFHn28|PJ1xZ;<A{tEz7
zB2J24suUY&EjCpPF1FqbQ9YY-2T~B_3hxAlip7(_!4wjyrYgf;B%iz;o!n&rv+{!g
zMQnRATkujGRDuA*gw*J~)#W-b<!W!<+u$`1$9305nJ^*MMknFNfnrE5`*YRGjzh;^
z>-bM`glE0raf)F-P4vS(rjfZUWexD_i!x!v)ZRcqHw5P`y?SPQV6&6iG5FX=tORr%
z(QCKsAX}cQ&09+Y??>rnDWoi$pFYT_`#p=89dpOe00|kS(0@Ak)tW9m>3ayHq$1ZM
z>pW$kkyouGq!^fQ_9s7><&3TN>8UqaH459Vt2E;!GR*vyy^4mz?Q)mm`2%3n+-^yM
z{$dV|Y|6X{A(96Tfij9VB#vDw<p@2TEnxq+zafq}M~u_>nhng3>dfEaAI}y7Px<8<
zVBbWL;1!5twR3L3baoHl&?M<>{<zSRl-soq<@pziHmhxGKvivc_5u1+ly#ez)S!qZ
zp9BjJ2L^L72ZOOxfWx46HyA)@KzOl?h;M#Qj?Z)}0=*etIs<{&sdug66!1}x?t9m{
zO-KRYi;t?S#2A2r*DwW--V{!&Ex}gC^v@pU9nt^gsP`75RQX0&lj0sj#=hT0mtb|)
zbICE14)csntfe7ZLlK@zfX8UC9mQ4yioWqxH88CB#Mp$I<z~&Sb^&;#TAWRAst)7+
zC8|@oOm07#_o!V#!}dD1D$17j5p?^n%W|hbZ!K&25bP4CeAGB3p3+;!gpx<p%#TCG
ziZVveXK8dk&o7+G|6Q^*=DHSge}iBc!Fx`jyFhg5&Bv;t=Gf3}?Y4V$$ytk5b(E~g
z<g`6`VRH@3;&K>gSQb!#r>;zpX5w}lJwwEke!y7PvrxMHGFv1<0)`mdU_?jTyOzi{
z-r}PB(gayPt%eO%BaSR%)_3VHIv0-H#Q%MLw;tRulWH06s42Sc8Ox5@nm%D8^Q9kz
z)jt`MR<4#vq-y7b*e5T;N`Qp(P(~D-xHdRzR?;d}5cB%wKnf^3zGYDuSAZ~Q<f%;u
zS?+EAvtb(MFGrVkX1Y?uZjg^yt^pB`M3)H6RityHg@MCN6Ci!Q;c>uqc_9YW{2+fA
zb+h0Fw!qXPhk7b^5q<;iw06<VQ*x+rE+xTtmfFbEx!`Cfm7_pAZBtcS(creL)3K1y
zp0$j{#rNstItQp2sLzx+JS_jSXsz!|-$#lwXB5`RJ+gKKl0ztC09}oSX9O<IGy!t(
zQ?5JAVp+Xz#02VY#%xH;llYe>T1}62fcw>l71AXy=4R2@)@=I6w2p^0-!emf$0jmT
z{g({^Dkk)5aA==AB_$dn6jTs2V(y7sDUlND^T690YcAkt^W>&w&W*V1X;Z-`_}>kB
z9|xLS?1&a5@9mK>NVH1u<OK-_TX474JPvSS2CDl{Ag@@FQH3m_VTT&SPELw}@rAj{
z$w24{#B6Z#ZOhmdW<F@D*3x*W%BkuZ?q`8cqG;=kVZTqr9SI<1F`>v#P$)k;IY&6B
zf$L7X<Z(bK${;icBmpahW+t%~BHe?96ypgZEmSh*T>h=ENGQ7Kgm9xqz0lKmROM#B
znN7DQ2Q<q@2pmM)JP_9AWPs#ttREE@7zkv*GDE<0uMO_|#Jd`Jg^QDt>32ZV6o}C>
z7oR{odo$tn=?D1+L_uak04-#Ilq{oWvstUyI6Z(;i}~{9Im;qD_#xgAnyO9f5BIkk
zwTK9}ZZmG^D(WahVQ+Bm!^)y7-)h&f%e&Fdx3;tx6G?!&id1MY`?@+VOt$1$9wibE
zI8DCqxf?qSLc}5okQ+PsJ;bFd3EuPf#0*A{i{wSTa}hvSDuwTi6MiFJ!N5+&z|2t_
z85>}{kqs`+hXh50nEEoaL>t2lR`>hf&BNpq`|RLAD&_>|2FV*2Z;2oO`b6TY5Dny^
zD@(56CpmWG7R3^J?&P`(+#*1QyaY0aC?eEpuFV@vHtUo(elPTLlus3#CBT8fy*v$p
zhLzh<8ifQx_L?>O>1NI!ZP)Vvb`fBA#Q{V;{(e`U&G2yzJ_9&4j?BRw2x|1>r<_L(
z6V<Rs$Uc_Vy?48!LI#~pa3Ox=$pRU!M<V?+gQYC{Gl>)ID#Vedbx{F#86|q1NbziD
z86Z}iE@HMiv{sl|@Cq|wZ5A@>-_d%oNbGwk<xlRHq3A2H9dQZ&^lQNR$kZ{z<N=t-
zW=5*n5Rg6Ns_$<1-^?n{&iC7sFjOlKJMh8pQi23-rnEU{VekC^0fD)j-+S~@`c!=0
z#wlSct>KL=JAeFgj5kN2NIKCxc&>kxiu0rt6R-ahK@29;FBBgf#0^3OE_myGlwZwd
z=b%(a!96bOyDMS)9$Zc@P_J%Jw+8rW(|L$JjC+V$qZAkqBDfu<S6OX|WNjH1p9qR%
zU1DY%s7eYG&<RO6_Q;wA{ti6igA)D~L=`J8=r$<y=IC&r_xuG)u|zvoBCtpL@`LJ}
zbn=sE<3{M`N8U!>_99lM+MNl~@PAYj{|{Mj85BpfwT%V|?(Xg+I6((@cL?t8?v1-U
zgy8P(1b4UK?hxGla?bmHRrkBM_K)sgGu<^cd#(M*dWu_rRNkyheq?vr8ORYx{(Ylw
zyd4LiGF`^d<X3%G@%?HS9op7Gwq8>kJagMLB48_6XymZ?Q&~?J&2<z8kl;q&Mv{iZ
zbvTZK>#oPdO-xMqM$8tQfT)t2-CyE_v3eG_d{|{em)iC#{8;C<)}4x_>UfP(0(p!$
zS5|fx0Us$T>jyXu>t(ut;W{FeTVDIidN+`}Yfvkq_MUW@3c9tPWnE)_K$qBb4fuPx
zW9qhDWsk!RZqY*m2@z~A=AY^u_Ffr!QN%iC6k`qXP}88lSypZ$8zp?L0(oJ#tVg(d
z-W7rM9<8`&)dm&q^34;;XRBg|cqlV5PnTp-WdpIemV~d-G@EaN56t>8aY*wnNE3LT
zc4cVIDpMcZ-!=Nq+v6Dm9oi^YxnWUy$fUB1UWLH*5zZIv7V^F9)$t9Ls$=NRabmM(
zwEE;B{oLi%%D=-XIh&k``ikEH%e+kOFuUzldA|j9W4ZpW8VRg%giw3aUOVQ<FW%GM
z-`52I^IV*sdSZ+2A(=+c4G$#GwgzZa2UNTEvo1;TJvScrE~ghltwnSOh*%@Dccr{V
zK<^8<1W@FhEXZSJn*0QceLMU%&4Mr*bA(P(k37%$f8HPTUdo+fxdV^sLKm=N@$k8`
z?;WKdfXCv*tACA;#p?VMZ6B6L0<7NK!l;*0h#q$fT^TC+ExlH+JTx7CQ$Xf^j}P}A
z=Y3z?Ocs8O$1UxS)!z5EB4$1g8FXJTKJ>$&=(eaXOz}pjj2LiB{#M<~5#qleNwFDa
zqds{LXBjKLdaC%pA0~D!cYTD<*vc&LB&@8?__u8t;L)-5MJMWOQLi=184wSKrwLaZ
zU$XL6^M%_tIh?VOsEK7{5rG=Yg4Ktte)!;z2zB1uQa?8r!ncgrsockL5Q98%?+_pL
z4^CfBH=(HUIsI;q)N;OApGWHlPOGE-`T>2Jmr0oSeMX6*0Jqql)5pGgxhCXnTVkFn
zx1N8f{>a`nPx#?Ob6#?xfQ36OL6fyR;}v<4C{c^I`Xl*QizbU(Apj|&*4Sq5N_^zk
zGf9}X3rs8w3=~>m?CZzQS|;QK@G!8b9xvoS$woanJS`~g5;mWr_O~h9TarhF9$mOE
z{!*t_t#Pfy1#+NW{;SM=N^xAI@(_)`Nn>w19kM_DO4*B*&2~&O;wVig%tOxry?re1
zO#2sq5*vgvG@IE?FyJX$_?=!*52`W>Wl0<cQuII(`XB2TAxirolwo#AIobU3sL~}I
zNVE^$`Ue;8mThhEy3+18EazXwCRAY|-$eOwdf75OMy`6h7u^axp18EI%+lCHzhQkt
zQg1B9WUzeE3JojidegU`kQWtaG1TbCXOZwZb%35ekF-P1^a1W6XnXTo3|fgAouMN4
zQI<HKW2dLb$y3hwWe@phO(1uw;FF0J-1g+%wDs0}&$ZXRg%PhE#X^O*Ul1mY8dt30
z5(>B8BcifgN_lyZ%oO~nBdfE;sU*kAeIy9Wyk6b3r?%^eGfGgMbR<aJrM88lgYal|
z$wf;ml27kKh5@g53EA-BalVo=^l+DCc3kG6Y0@DBmYc%h)bBf%qUTnTZ<Mx!wk#)Q
zI->3K8|WfYij*VAeBrlHU9dbSWb>k4IgtfUFVp*pev(%V*%~rH>r}C-rMOEj5%!(+
zZPavq59t-ke0_--tG`!^ZIq=XR`W&Han@N7J;xSU01%R=v-pFWh`f6o`uQlNtlUgs
zuxmnZhLW2)9#4T;CDU=~0^PT1=wb$#2J7lRLLbZePI~9<!u|mhtc~;y|9m^uG|Sv?
zOc*!+oUrwLZB19tWXwccXEI}~Ue!e7!zXLmtWv5F^A`;kXg9Uv61mdwM}+e|+1uFJ
zzRO|Mr+=~Oqupa>S%05HGURY3wz6^lroO>-Fo(DutRm1EXI{rEuYJ<4dW9oZuj&u#
zOm4A{*Iet1vxn17sgs)zc9zIMNQA8G0F)Qb9qAl^!BMIbYM-Lf3RZ4F_XoBZ`#?+Z
zs>D?IOl4=Ot6KSN=_(!;{WrkQ54<&m#}Y+`0oM*^6$JQxldBEhXhil;W9k3cnP8!0
z4QC-;UH=}~NKT;%a_y}|Sj#WsM9YdTCG~bO{{4y*ow$Z#or?X_HWfcZr2Gf|{ApZh
zN{x&B_6dAzi>>dQxPewRk`3p|<?lZJmCMwX%g|iX6f<b^t{LC_w+)Y%=mWe7_Ked`
zc3^Ry?X92Aj0u-n4ny-SRm+A;*WC)jN6n^TWq-Zr!x#6+`^waYRNSNW7UO*h+LbVD
zB3DYmSys1-@C>#-052RZGBxhIjb4z|pMu;DnYsBl{nQSE6>D`FzDEp|sQ%1Ebwox4
zjmw1EU2^36&G>q`)BByB>k~t0691XgW*}6A?=(pCujxo=*kM)jw!ju)2%nleO~+9f
z@zS^gK_kA&87Fnf&VUq{Y=B4tX`O=TCdv%0wKRNAqveoUEhyd%4=mn5b`BhJmmC_C
z9#bd`<7*t6^Y4*Csse$X<Fqy+<?L{}(ZLqw6S|^p&jHlC+uS_mxo>sHX_}zIW}s%y
zDMl@);b)ckA$vHDOREp}6z)St_8%9iLOo#N50I)}gCvowu|oEO_!g3smKn3z{dh^e
z&7&`1(0M^(AYh<oZXAMKn;%_I(JV#ElF337T%d>ZXnC8uR;uIklorE1j_2OD)6XjN
zuvakUe;Z}D#97`n>3umOxIJ4I5Aas=tz!P-NFj)eUq@FgR#0bfms6^>I9?FG4=d5g
z*MvV=mlT{RD_IiH6ALut8P78z?38pXA+MaN$R1eUI!Y&mKcZ%rSzg-c7h27)ouGc!
zMTvv$#iM71#L{S*>mQKTKYV^3V{h%yOs_~q3?I=7knoP4%DWlsO)TzZ09p-q%614H
zXBv;66*OE!wSCF68M}PiJKDa~c07rPY9~%E@H19-X$zK@B1D@AN!_{3l&^Cn%*v`H
zM&Nqge=6^WH!C(PV&Ok}YQc?Npv01`wv$tnkzo24F~fu-yU)1eFW4IMEzLqwm>S5D
zidbxME0DV&^p<MEa2C+m0LK#{15M>y=e$lLhh@-bicpiM<CDjQKXc^fgW*$z%%x!v
z(r3luyLr4}=f+%)?CI*)B?<(Tn5iogQ#fMu|Jj+9_AKey!x#ZP&w1VAQE#27;W#g=
zS-~DLZv!e$$&ud>-VBZyvzyR_KO>m<L*B2yJKGUG-+1VP%G_dr(b+5Bfi(p#U7WNt
zGZ^YLIJ~`Ew^5dmFy$bTe~-gKi8XtFt#TE=Tiv;8*}~_p79kQ~?iQAPZ!8KiY9BBc
zH>0IzMn}lp+pD})<NSjGR!JfQ-maNldYgO_CQ?ANj2e#0h(U!>1Q{%8{Y%_J=yOR4
zWCW4aW`R?|k#zy*x^S4sH-}SHFrdF@a$i=cjd!log&*QM-Cnnhzd&E`uKv)|&BPo)
z8BqA7GLzw#F#4zW!{`~f!=C5jNk|^&mTKFn@RKDJu9-cb=5jARIl-~3(d;{0oT*w|
zlIQ!(35XZC&l=#yj7LMV|DAWGpTmxBDomM3Inj1`?T`bzE!B#0`Fa!pX-K&S`Wyy#
zZiD2YXu`LtAxkArj6dKdYzy^H?C@XX4idtr&t|bFJXW^qbS(}8%|ItJ<8)I+Jq*)=
zPloffZly9OLRZ<nAU6_gGlpm-Vx`a%qCg{GF~<3O^bvO7iL%%k-d|`>0yrO~WJX*Z
z6E96GU1vamQSofafEoWeek}IjzKm7U+s1bL#N!g}{AAaC#5ed&E%&<C)*5lRBxN1D
zBxAwVRV*R-1m<#^vj5hqnY<!u++~fw&eZJ+G`ME%uo{|~XXi+R4o&o!?Llax*v0m3
zk%PF!UHFKka)84#eV~8u6XPVXc)XR*+Bg#j*i8wG=+2JgXqFZP`X^@ca-MlEj-N1x
z-u;;cv*L<+N8uR6`Z+R0^36<I;q;0Yu^94Y!nVqe=c&N=C#empl%%dthbrq272@jN
z$yc?-=9JVVi|7Y|zh@Wg=wH$JnZPlemLs|*G5$S3^llCp7^%C1-iZz~1I-ZS?gz)M
zfwp)ML@z5VSh-o|s_lh9w$DZMc@t@~r!)_Dik)<?rGZTeW@Gwj3giean0p5~0|53L
z6aFl*Ppisxtq*>5ypL52`%~)GrwWI&v{|7>Z<ph%j1vmq*WP|OkKeSv;c>s-Z(>DT
z?P@upNk!-~SLAI#KBu$kdamy^_x9oi$XXCEoH?U~es%+{Tnc2@$Ww5DFP0j<^3WaO
zQksChv&;(OyJtFvkOV$U)!~{;snwC&n$CmK_LsEI5RuYbth3?Gpn2q}FQCKB4ZBt#
z;SstIe`6o{=2w&J7_!!<bDYuEWBjY3dOy;=rexPxE<0ago7tLjEUKM(OFtX%O^hbn
z_NEn%zc&jG3qz3ov%q84?O@-)a^+T)?2?R<KferQGuVSpWVNq&^O0k{D1<$yB(N<h
zptSm)7_Th4wZZ$Fq(?Qx<U0udQ)V1HYKWQrSUe?@CFVx&Jd496;?Q9gbTg1}T<12p
ze5v1hbB$iTR}()|h*f7Tk%4TepHtr?7w0wGvm{$%gE}BX->M)3(xZ`+z){Oreh>e$
zYRy|yt7XE8l3k@A`ffKC$TfH5I38}?e`z$?(BD+Wz|C3L-d64EhyrsBVIYFYhuGw$
znApbkpmjF*q^QGvE*`@6^L!{KmO$kT_-hjZM3;(g4QtS<W3ba`i#pI<KtEIPy&uj{
zw3%9M314;{jSx|`(sgeYEj5oYoRFBS<T5(a1U<5rh}p(#d>xTmlgtX;JSQejM_gf%
z9ZD^YlDL+^0J-Gr*E;)4JCR-GVf`N-jyphH5rpG-5PWlx9qHHWOOEuNf)OtnkBaTO
zhZO}~c252KBcuZIH5&lwgT^N`R>a}t;>{-rz@p7eO&3$G<cP&KrsAZTk|01?bZU`k
zlYPkSEizvsCT#wyOq6gyDwR%;!+nSQ>@Fa)VnX#n<4j(>@d(i*WTR0h#~q(qH0O<0
zYjCSVVR+HRW5E;?s5`>y#HQ#cx+A#)g*my$DzeIOm)nuwaR@kL?I13Pt*@_&U8Gy%
zI`g<KpWS+>*#4brD1N_a38?Yg*7x266B6>Y>_>(?G^3M-@l`p{(??NlOq^F*&`N%n
zL^n@%txR4aiemGlN@+Cm;j=Jb;;gq9q*f_U_S7tou&A*toI@=<?U*eg<e<!pOh>pU
zd?cO8%w-C@tOO3zoTZFh;uS}h8#{J+MoYV23x!Q)#o%B(pix~{@h*w4TXHdT53qmv
zX5NnmT4@_kaRKWi8f=*F#3vdx&iCX0I<@6$%VhCxqY3-2BgQasU)ut%(8<myxL^De
zoSc%Oo{UQr=u8^rc`<W+Yh_;6s#D7kla1*Be{E^<LjZU2LFzf!f00oyZw3E+(8_@z
zPb7pUO{6S2|K_Z@DXPmOmI{9JaPX3#D7+U|K5wb<jF|l7_3=P)XjdH%#8*Jai|zb8
z)Vm&C$wBABmtIK#y-V;CA*89M5~cCT=*p4h=D9t6H9Za6%sPEKSW_8%+N*vB!v@KS
zhY}bj4CIbW(KO1ndVcFTD+sFV2v>-pIm=T}3*@HxgHV~;<<hR7T_1QQAL4VGWfT=b
zStnNG6~go5n0y;Kn<@2{&`X5o$w%n>)Mbg;8msT;1hvK6@~Uoh*aer>V?jlXDJsxk
zj&Fa-pZmS+9$wCGA_Rp8;+RfW_oeDLBJi7!(g8xEW(myP8>i$CY$QEHi02xWiE(1W
zk&&A~jS1b@FddI38sl4vrVz?M$lXEWL9AGM@CriSq<_4<e>RDWInz~ihZ&Vk0o7LW
z2A#pq)I>8$EBlNy;H|Y^l(jT}Mh?vKowzg0v2-xiushRY>4ei9P2oq2iI0KbmR3bi
zt^gb1$CFSaUl_q%?XTeUt@cBPON5Qhf~=Qx>);mP|0Uqe!8f|140s6F@~0bt22K)U
z^G+xO@B|5=PwZ*D6MYI6P7Yc2tA#|>5@{b$8Wq5o7=?5kwL?2|6ao|j_J2q-nJ9>j
z<F9n8>LZyV4V`cSWqbA^`?txzPZcZ`Edj~|O`n~C{@WY6CQn7xdAtaQ2;0n~QN(9u
zRW93kwI5ib_W3#q5JMsR4r0p64lb6@iCt#n%6vOBSE_mnzKXu7q??>`nuO^)T%ea8
zB}fd;N!3DsIh-BdSWakNl1bD<kD(`H-q35f00O}*;W(Z_ea`BMv62bYBM->72>|E|
zKn$@utt9J4-Kvx0M_g2v`QbMx-fLt*GL3q~1?jn%MYpl5Vg2;OAs$aKOFV9FW{K@t
zN<%_hc!(fA%c@|m6if#U<p}0B6MX&7b5zCQ(PXP_ONa&>G$a6g>PMz;5S<g-HrVSu
z#|=8Gox3%W^7;FtH%cdqMsf0@fifkrOY%pjl6{v|m9`(Zop@~zqN=62n=t}Y<kPDA
zF!9mDlhk$0=|s7|#m(h0%ge*qyarfriPxk{$!1ZHxkR3K$!4(~7`RHpBLX4=@Tps7
zncgrUm6_AoFGJG&L;O;Rry1swRcTV@W5i5Ubjak1r0m6!=x~-XqilJw0QZJSpNfJ-
zRt-Do!uw#Q&qJwMX_SE_cU#xgR!6QHe&x>_od!7`o8L(-_h~oYF_x7bKPyf6xk6k$
za3ZT%@wVOfDVkrOlP?JlWC`e-Xso`$8LFx)M3}RhkJI|$nD6C32>!bq4vd-wbP<9U
zFL}qr>Usny|5|Xl=usYH15khdkWY+FE!VZ=!oB%@n~WOpm1dm`eAG4nYuryDO#LA9
zU>$n-5lz!&5>ew`*;eh+Aoavc>5>huwxiSASoz{g@S5kLykLiTkcE=E>sf)ETb;X(
z+`py2T_Rjbd}9_mNcMC?ma{ObnNNotDX*Tw4P{y8a>mI%#g1sn3;e2gE!9b6yXZ1J
zKxkX3`;E#z!r0}#k0Pd;X$QZ~`s{Nc0GmOaiOx`xT?1DkUD78sHsRTS|BKM`9;XT|
zBKnu(6qQdIK2-yo$qu-s;>`?ue?8ec8kn}qBP98x&JN4e?gv?~<2eMlsZdtv64@u-
zmoCaFmG|Wrg!;U$H4qRLb`fYFkH(<0ivR4s9d<bp&+&5ZB__o9BuMT%Uelt)jQFr$
zMCZ}6M;?a4!^dJ!;^E`)((2@LV<Rm=fg?S>|Bq!s-!P<J+FaHHjSlmCIE+gx9(T={
z3r!xh3r;UTWGM7)mi3XXO%H~WkkZG8fL3Z>v2cY`zI<B-53v8_J~9|3W6ZX~J61Te
zNUEaRUsKI`{uKX=6JmJD>--0nTZAi<W2SqtW}??E1;SzAAqnB1(e<Pm@3}CN`580A
z$i;x<7mvjvlNDx>X^*z@@OaAU{cIT<HYP)bx8AiRF6bsg<Ze$*UgdNRi_3H0IddbN
zzwL=)Zk^`s2ZRtQ`X-!O<E_a@7_AzmnAzWbND%xmorf1DF6;_)L9t+2<JAa?9^A<P
z7&oV}Sv{mL7-5F2#-rF5&^LLf)lUzzjUAl75__dxg*u+T+K?a8BJ#+xs`RegAUX_G
z-vbExj0zOK?1KrahraPK6Ew|hXgYEyn{||X+u!t<0S94<(hgn$#h2-?XO|!cFSN-0
zxCjLujvd?9o70GGyg6MMo0+B-%&GJ+?kF!D@^etVNx?yUq|YjBJwuL8h><fjlit>b
zMouD-t#DqGUO`e1YBc-l_d-{3%~CC`1Nifakj+qUcpd3g$I<^<G8{0rZm*gNR*4f2
z(^*CaP>P-hU-v(1HbIMPuQJ9INDdanW{4gc?2rnZ(7&l1jNcmwpf0?9&w^pW)rAd@
zBPJ++$wDH*<Udaz(lxVgKp0^u!WyGiuC4#3K4*}(**7>#K8uIt#H~rY!|)^}&V>wi
zWYum>=ThG(T}`Z|1+UDtqzfS4Cx{zyAJ0qx`Vm~ya-5irGu7n+%Frs%PsHsR?dEQ)
zK((GpA8q(Wo>R_Fe>A+?EK}%sBqBq;o+=xyqF#k$I-L3OA|Re4kF&*gAC22=rTva=
ziX*oE7rzs*s-Sc|cr%GXw^@^TRPjpQ%vP577b&cafWvakcq}?tRhmecPK^WXYnvG#
z0Eh4yT)cG+aS6fU!H6^F@x0#r$N%v56fB)p-MZ{oB2iGWt%0I8!ql8*_=QIRMah1K
z0)!CN!R@fanc3_ggn9Uim{4i@3`cG157D)wyk$~`$gZebCjB1i1Jo=Q=oX=y(iU#^
zH@4KwRz&K07r4bue~iJ7_14PxNUw!qV6!>tu}uwBm5QG1K7@ntwD_=7+KTp}UsKRC
zI~q-W55d81|4NZYvB7#C;1oW(O5h+uh}30v5UAOhw?Kzh3X{fxo5B}Fhe&RTBDV)w
zw=F;%RjT~Wrk(g}V-JxfgvW0%ffdi0N$6N1o#@OJ7rBwlWmBF(;#b6M=@3CNAU8iS
z28o;dnu|Np!Cd0)j_HXS<iB(g+>4|xCXx&pil{#U*-MV{1y$R7OUZ&i2X5S0kJz$Z
zXVLX@WVC;!&$Pw$)aiYQDrc>*SSKuqv)1XyjitR|XjP1JTY9-^%(Kt*SZjGM1saN}
zXe=agsl*<AMh&UEWG=0U&JlA1(0}ZhU$n4_R_b8MH!Y40Q&2P_vxNIyib+1Y@SD|0
z>zlv10;+V}%btP{$v(j~n_0yT>1ZZ%y=Mp?b17@=9CAT|dM(24uOq2k`o>cFue8U4
z%1Y_GX!URnhKyOz9)HPSvOx6@)YPo{3ReAxq@R01ic^&dwfzqdih)1dK+A^q@yLvR
z$3L(5^ZsA$BlvRKjuI+`QS~w^(OOe1`6_Co3pY;b+%PU@r8}UQ?B%71#HrsDCGhNW
zSPPj#+*gge`QIuV={{TCH_tf%%N_=TxByrH?@~OJAN}#BlcYM*F^-X;l4-&ByL_sa
z_|j3o3zD}{K7+NHpHtfza0y}bOMc?fgP&2jm|^!l@{n1abfP<SmxNK{>&~_GBciQ6
z4?vP(&$d01X5To&S<X(W6jhV>_lTbAtgcgIRqMyDBHdKEWpT1nHAyx%d5#M`9iwI>
z!X7$WsDD#E;jB@Q*|b&FRj7hTUK!Xg(|TG?L^SoOtpHfMGCP$`;HC1<uA21qa94-L
zg*zRG((!EB&2SmDwS=zj>8f_9I*&=CLkPDwIUx~3%I67+U;}mWlva6Q8=K`wKrZ)H
z6nP~9L1Ssc?V0EjeZ3<qh1Zz#w6pckSzMk8biX5XOglHF0&JcK6s;N%lcgl#)#;cN
z^Y*$=u<AULBzKbnz@yZ6D_{_Ccsq`WD>1=AgH<0QaR=LA6DX)2Xs8}ARwA-}uqJ0+
z_Y_RGjf|IZwr8>){LcG@+J>8}Gl?JiFgD>H<;2<>xY~}Z@nO?xMykVH>n38RZ1V5-
zcrUWisvq_Nt6K>-*Wb|jpA6yO+V{ro7*Ut`PLqE-Ti>N;031~yoFdrYjU=J|Paqk+
z<ig;UZ|W=K1yybnoj6D_8|vKp-;olCcf@iJt(A{g=FlJYBUJ?65iP&|XR-#LNKwRs
zstBU&e_A4HCyRWBIh!;n5E>_hPTs2_I{E1La3E|d>ebw2`j)|5*<uBPEMRA?^&J?m
z6=l{<&8`Emh$93<NS+0}y}zRnZ-R%l?8Qi*!>dT8-Xc7z%a|f|Y<rF}T|8C(+;_4_
zfuvN?Q-p88Yw+VpKdi3k%c8VAmIpO&Z5gJQ*mvo*xFX)&S@q0?%-;e79}^J0ygH;O
zGJz3#;+5f=PRMGuX;nFs;=jA#xq%t%Nu`$?coE<*L*S_$Cs0a!*GN2>Blc7H`xSjA
z?VD5U&riPr%GUuR^J?*uO#pGF2nHIiiO8W>*ko57S(H|scB(KbR4cf@C~|G&$@VON
z#C{|y5vy-j76bdF1T>>~gJWK@ASPW@k;Qc=0`_w<^}zF(4;S|*j{NSDE8<iOHjiYQ
z(Gf6Kxm%yNT+F_$!;*@xQ2#||xv9$5%AxR}T<D!P0@b|0j5Q45<|ibMpcujwGHsC2
zcEWjn*f)Z>>v0e&^cdrh1<s)~qrS|yl-=NFrUk*A#uZMw;!jU8Y=e-vU8o28kY<4l
z5t&?=ynl5!rzrVBvq5n1?d1!|9PHs@?*zD?E8g<-RdH_9^@6Tgh*Z4x;!TVtmL%2e
zrBWl#{QO_5(aYWDn*|*~hxQ8_is+TADBiO6&+#~}KkeUNw>gRZ`SG7z?IUk5<~|4r
z1;y0~Qx+Nlh&Xn*7M8F{-5+M&mHDbIq!K|_H1Bp)?}yFmXTQc6)i;Nq?oy`;e1QL(
z%^3fSlY=*mhxiumQjDDbp5Jm7v14g-dF>&<*C6-_heAf2PEMTNLFL7gYNv~4qs`eH
zEYDn*7+eYRk0kBxJHEkiWzh}+8NW_@29s*$3y=7jI1meTUVI@AQF=)Dac;xS9$sv}
z;RQIsgs$}&s0sj=)S-&X;ozeExkT2pjnuv@Yn{YW0iY^-TSHvKP)_fxMv#6R^slYV
z<#^P6`QK=tE#?(6RPq;*e{hOT&{Agoo)O4+#Ij(1|H6l!?i(~9)P+z@Ynj9KgiEA7
zE9?Z!(Q88JEH}Gv0;4-Q_i$3a$cdF)6}v>URxXrMgHr+s0VYU=lU*6vUA`+6w}fM_
zb(*EBi8JJOjp1T=kOe50;FKuauYHEoRB`9OscAMG=<iF!N5u268_{J|kzrnyx&10(
zjS~4BZ>(INEG}D+!p8hfS@bIDdN5xLN_L!3bO<TFF6HL5bPCy%sT5o)fA#_-hg&U*
zyZo4CdZK_e7}nIjx?n5GJKlZZely--s7hgv9mG!!PBz>r!V-c<$yuk+(O0~#3CfQ$
z*qvTEgZJ;z5HzhMo}=C6!d)>>;CJ?^?!(&4yUVq$O1brmSfYDaF;BtruMb=>R(zM$
z!Vg)<60!)x@sE8&#s*#dT@{`+Ef07^E+kFvkXk@cf4$Y%cQ7}-zxt<t$w-sE!`2<^
zF(lbVYmW5;*RdR}L=?U_T@zcv+PIX<3Ob!U>BIJ1wKk}Dv>vPmw8~@&-2L%xgah4v
z4NtK#paXvCPjXkTMn;a6M_Tt}4y~j8hoej?SsrwyP22Zub@Kg&oLtO=8n4Euj#qfp
z5dnA)W6!C@b!H8#70;9m&nNPlGGub(_-gPp9~R#X%swiHK{5SR{x)&T`s+u2_@C<w
zbW4wrr|}rBz3x@o$V}SDMlh+pjTk38msgyh&8Jxl-tKzBu4kO>WYUboRXS$s#y=}U
z{;fE;ey`Rf8*|Mxk&hf5E?Y<55&8S7*y0NOHkye-FkXzQw_-Mc!li=l>oqR&*-dq}
z0Y5~WxDM`5jx<XbU{NI6XFW;k%W4sP(ZZm>ptvPzGMUwuVr}!{nW}KXJsj>NJk%Wz
z`LeehbMwu#3+mc>$fK#@H`ND$ba37+ZTLEC5~c_3&m%;Rr9D&KIt4)t%$;+8M+;(L
z{@q^2826z4?28!ZMy5fHO?05W$0nW!^(e~&b;`LQYYWk-98?V=wa_W)xtSJM2h)!p
z>Cr%FMYrz57E$rlzXUUyjk~N*mN6nB>^~v<|ARADsCa)0yaeE$%8z~Av*W2`{QKva
zT|h-xi(227aonVIzDAf~iO<i4M#urO^}|I*mUl^6olOH0_1ZI&R%nWFW5b7ByF}Ii
zU9#_+>dU*1ZVxv}uwkj+%hspey&Lj7>}IZ~$^`-C`~#ul=T1X#|I%-tQ!EYhRj~my
zbrUjdKid8DzbqJU@h1zme{SrH(9C&-FbCo~b|L}oLj9O5t7Ikib_5jC6bo3R)w9`s
zhw-J6hKJ9?Uw1x9Fiy$;CBeQp#o#Syf-2whXcg+|rXS)<a544VL5gGCZMWHnHka{#
z&HrY5T2KC$3ZvgloM-xsIz`_p9uhP9HFexca*rm$l;ZsMIh^zVX2Zy|J;d_ly@42y
z>v^UV8)%a6FU`}B5)5rWbfjTdJ>tjg_HOo@?NZgTyDxqL&cp1-+HW`4&tnURlMAvT
znw+E-K01)b4~^B%tT_ccF<Bsnkg^AkQ;yT!UcesV1SNDO<FxF5h%ocyn)?~e|3iae
zZ6YK64-HoM2Z9<H`!5gX?0(E5)J_!`+ptK^&~AprJ`&c-Kqz2legI-Jy?W98rWc<s
z95rpQWU#2S!r#2d1%+F`?ctew00|M@v!hTPRgR1*@%h`!aOO_xvjm{QRtF6#7XB?G
zAZJq(B*!7P$U(jE;#Hqh;|9|}{<w_oKp;mB;b@@V0LncfE@IoU5x+}poqrz`lHnU^
zJ7~&wsQ*ZQdH)_dL%cJH`Gxse^cq_!r~v!-=D!G6R?0L3jDUOJ;7=1Inj6uNNrW`=
zNk))<a@Lpr%tot)`^3siRq?lqSzc)irIc;~#`2hK#8b08Y<S-XVLq?#3PZWEXp9Q|
z<OL${Qh@mGo#fJrgQ`t&a$zQTVX@9hT_U=Mc7nU?4>1+#7C%&0XRfq?T_G`N(p{Tl
zxwKY?-<XClaL7+kH?J*g)3Ym&3E;`}1oH=X12$`x!Q%)zf+tyZkDu5n{_IJd>7Stc
zh`#&hI@$MSi%omcfe&n=r?rmOKd<T`zF!%-OaMebYi}5BU$5hqeWxc`P+6l@i~2YG
z+|~Gf4ulfMYhG9Hy&S{BAMq`_2E2kb=PjdTTX=%Li!cfnXZ5o}u$w-x@&X;@CY2ku
z($pLAGk<O7pgJN*RTfe-mHW0g0>J385saYa9Iw@OBFAFM;xHkLbHNUs20G^m;=}YQ
zc>pqIE;bq&8W~u)L>h$P*`Uvr<i`dpXbAc$gYk{1&5Kq=+GVs2-R~YYy`@6DEw;b)
z7bau-jrFRtB7!19Gg_ZA*56h`Wfb95@~W02k`sCH0F%({!I*XS<r0yt?triJk@BIv
zr`s^v9a7zQJG1Fqe^r%`^BufZP8;SjfE=^`FE40u8gA35RH?G3ruEX+y<=8`S7j(G
zQx7vyDEmA-uyJCWgwX6M__AM-YALAotbIsA%P{>QKT|1cLyeq7iX7!laf5RJg;ZCT
z3pol>KWt)EN2eTr<|V0zJIxqQzA2aWjVhW*9)|9Hw8O>)R@G#Q7R$|Ca@X%=)UI=>
z4jm{9LXuej#(g#uG|E-Qt!c`(*UkKB1?T6hQ}cVu=k_7b%nPE?O!>zq;k~p<zbg6%
zl?mewjyXN>OkWa;3lvK9u1ou#+~!)3v=f0J{Ohj611`lLeO>FX`5vNdvmM;z>Uy?v
zeggq#Ecj1AiA%F=VvLT#lw=C!5n7(9$+MmsH;X|k%j$Z3!fM}o=pj`uMNJf!S-bP3
zozOo$e77JhqSeWu(Zf4So?JiF$)Gy+ZpNwM0mMt5u?-XWKQ_HC`D%9RafeR4v4q4H
zel>LMJG90)U=b=fdW8RyBFuhGdc4&fc`7x!Hg2o{>!#nvc%55KnbLb1)0imJFb*Pr
z6+87<%@N^nvy<0a{@i1XslmwqXC5|<GLqLnjUtIWr989b*x!-6KKA>qb|@#k2mE*d
zeBK<7*^<^4neUx1H^nRsp@}VDQ7Au13^p1KuTY0Y>bbxE<;|SIQLfdU#6fH6kTuEi
zP<{Lr2)+jU#H2WnU}7dAxRH-|b8h@5`+i-c(4q6ZHtd|t-v*Yq)(hbL6y<3s49g*0
z@tM7!L%RD1EsaX;z(<kHp0Vj?SnJ<a25UrODU!z}<P>xOGZAln0Amb4fto?iCT<=x
zO_0mwY2?I)o6js{6f_7EjfzLkCT3?HMPS4P^pBvfkT-~%$4(RFf;_FfjBhrZdD%qG
zBU}Sm{&SheIJ3YC-1dwZff_%_1qpdIpeOz35_C_~d{4s$((x*Dk2PwlxmgH#tBNDj
zHt!~;!PQRd&c3)5XS*BNGf;6VIZ&7wId%Ona2+ApJ}%OwMYbC3%YjG1CIOhoIa<aY
z&D+m>e*FsU=zpH#^TXR;Hv{H_+z&lLCS{zjb!K?@Qs(aV>ijFezl%bgO`a!)xE_9&
zkexo~&08024|Fy2&73*^$7k_k;pY59bubMrDwl&zv<LTp-k^|TJ9nX0w$Muc(A)CA
zvMhvK4$p;>QH5k6mo8B#S@u7;wqaX@nZVe5x>R6b;MZWR!)n5~EYNX=Om3AsN|$(U
zkp+JZQBT5g2~>6>i%MX-yyp4B7I~SVN?-{T`ZT7tZHJZVU0R=~-mzaJffx{muPAu%
z{TZ%I>1u&2M4ytyQ&B5qA{yK_FI-5T5HUU~L_)W7p9Roah8D{$_p1VLgFB+j`JbEB
zWunzjspd$L?s*gfLqUG6pi$r;mDwOwg?d0RII}e7cQj3k1_|@{jlX1-WU44-pD;OV
zc+%}!l!QYxzhjdbV^-{us05_KE!lLW03&!O46ciM+U_n6`GCPjf2}?8(~r@AM=~~S
zqmBSQ>-rCy?s^;{Rhv?Fc7k(tR{g>hc&N`E^9!rk2kt%zed*XqoMx=(2?9ngn)II0
ztD%D~uFmrm5>0qWnMol<fC0OLWCsyXm04`0(TS&wkCRP2OXbzO{`UW=#uYWZhZD3g
zJJ4c&c{F(hLY42_$~BwF`iymwLqJuU=m*sb?;ZiKm3<wDX-NI=j3TrU)VwN-rXVv@
zaJ2|%k7fulY>*Qz<?)nFu5xK3-O#KcHCaU|SwHb7JF8LpbHfB9bu@}((4l!l&HT4x
zm@4fN@BW`@<Nj~4Fri`a3X>XQL<n{;f~6#*EjBo18?!MntBzI`VeIB5m_TE?_NAd*
zvZNer_b#Es6Z}zyhIux@h9gaDD0WhMs96(3!2n3R5X$du!m{xi3dsm=*WSS0?IF|i
zDADb3*$q9lvan;Nl-(+Qsj`9gqW0YQ%g)G1lkCH%M!{`TCw9H7YcXk%)H_O>1c`p}
zv{E=nWRExt6E*B-3m`+d2xL1}kECA?6?|sfriKs5{J%*4lIH`cFZGLYG|3yi(B9}u
z{JapinNFTv`|pxFJ)0e4#0S|}ClA4uUIp!~85`K&J2$1CWyx490#i@29OOY_9l6_R
zb2T`*-fa_ybgwYKC=259V7_YnfHfj|HRL0uo~KlA&Ib6B!^~b3<67ahV?EK;d8dy0
zc8NyfHdtp(#;+MoFQvWlF6(dt)yj!fxtaVTw}XywQRQHe!b2ZL#(w2BvwWc&f``zH
z_KnXlow3J^#3G$27kN*84Zgha>0{wNAf~rT7`HEA&$z4S#1WaY1`kBI!Me|QBfC(V
zDaGE?g#x&s>DA@2HK?KEf1jQ*F791RnX?ZYQ2d{Ba@S9?`r7{`9E4Z&s%$e-KhMso
z<?<h?+U&4vW_R_qtda@Z?#U|~J|4+(Hq`x#PyIPATr8}ue10-<n5j0I++V&NeJY?%
z&^5i0Svk{2TTE&)uaO+l5F^7BF#!hU)OIbM#2)dLQ`FRT*3b;&e(hAGw6Q5K<gM>M
zuVJ=|S!X*<tzUW)ED)xcvGr~L(~mTpW~ISjlELa`!~3CiT50kkB6N_Gk507PKIE<X
zb8NrATeFReV;ChrgWHz=Zu_gEtfXr(4MCc2EhB?x$LI?6?4R$V&1JP%2AmaljF6c6
zs99u<U}!%s*Z5L-)-vT~8Oc1&mkSVv@-fJCn?M_jNz;{fJdZsXjMVMZ9d9QnJw(D}
z5qB0;?WiY)De{@$JlI9N=fNP<xfL&2DGr=KIU>?$44;8qw_+#p_8oQXrO}`_Jd>9E
zv$YT>G1VQ<2dG-N{B-}V0_i+%EOaL~w7BrNHp8#4a-ZpG)xO|z!*3Ty&wJ@ljCz*+
z7G=EuqV??1e-=dW!Om5Mock!M;x$6hAeMf&V}<1+PKqs8Ke<b?tT(N7KRKLU@gd9j
zcfDHwQ%=^|V#I`M4)oWn_EJoPCS9aBn^-}eeyFiQ>2Km5|4<E!Aut~SW50~8ugc|F
z1Q)r6CbJ{5G=(2Z`n0<3<<_-F#%x@f6v+mb*rNUV-@5uOE;``svVF<vF7t*C?RvQ>
z4mght#@S~VaH>?aa#5;`NTSLx`=a2@0sVu(1bkOSHGcXN?2B2mCgXnKQrWgI+48&u
z`t=<8`AOcpJA?q$m;vGsBl|thF+cM%^4hG4&%UDbr)2m<n#(g<A~+AOSDcf4aR|J{
zy|8JFWVV)^kNIk^elOB*RPU`DqxQ|IR?Obyu0y&t%f{zcnR1yghKWz>prowr7}@8v
z+^h*Q->)yG{VAl^>fCq$4BQ7cb;{KX#TrXnM0r4Qv?WkkHPSb?ke?%pj$VM)SxKM7
zF?Q$rJ@?t+Ny93Fka%PpR3uk>3W`XPnMzrg%igP$&|#=yD$L7@B{_CDzj0pEnwikG
z-C&>q<-}c-$7RQS%4Duvw>7Z8d!FR!@^*h1n7bp0bQkD6A|;^uj&p`%`Cs1){9MSj
zI>9XVy#csuU^B)ipgAln<B|VhR#|)CV#?2U=kDg_cD0HpIAg`}t7>Z~r$|-#)@#ZX
zD^y4outl54_C&x?zGtT0M)_uQ)zCi>nGO{U%gS!Lip^VLmzLlDlUy_wMV4X;O1FbH
zzEh)<#O-WAqnau|jK*ApHq@UVUG*GEmg>}&E&-l|iGF8)N=yWUKv#qH9pC2cJM)e_
zxePa^LvYFdiE>@?7(iVeFAM$$r2>%)tyDbBUoG;#exgZ7wKfJCAluA^*aUx4vF>`&
zwd`@{<yC7Xm#CgNz=si!c;9dv<o0ZOzM8O-{_$LNQoOA63o(}+C4)W}7h%@63sgf4
z;F0ho3i^%I>{+M0EGKA&8?_~DYjsJnTCdQoPXjdXj=yFPmEb8R3B4F(^h#5vG7fn{
zd(xQR(01#nD1O>zY3hhlK*8DWBP4-`!|+1t{_P%PgZhoapW-M;fr*iXK7x6BE{q^5
zDx?@bOEpnoc4}O*o4?|~nm(=7O3;4^IC4B7LRmZ*=-)<8dH5aACD?~aN{uzigw)N4
z3flgukf3`z?S2uMZ>G9CYd9&#I8rRUcZIv9ymVJHYT~iX?_f2LDH%>@P^3yh4XMi2
zF|1`OoO8&%wR+<5kUu~+-;~~CaxAD5{E03yD0h71fnu-bm~S|~0>w2jTm5N%K5v1h
zv<Iz!vsP-R`kPy}KqqeLGu85^(g^cEct2z}?Q`p2W{_FkEch78V>8hom&Nywe?IVT
zD-uoTdZiA}x4r8kg1l$9j;8+pGJj<=qjsG)Thl9a!Qm=u<H9JckDMlfE>%j{KvWoX
z$$5tR3a=N)EBmwO`*PQ_P~aCptCe<4ICwpSTAGRXz3-;F6qLh2=*=S5u&=Z$=gqto
zeugylvv6g!NuQ=LzjhOl3VXF`nqPX671jI#p$BVndKev2HFo%3JxBT)kllR-@KQ&M
zqn|I)fqzDY=73YRf=vHDpRD^Rc|$}Y%1EG9Wpil$R{P=nDliwI^Yk4k{()ej)Z2j;
z88z;85>7e`)_&FR{T|_zGju2>uWh!|lD7=+B^~uSSe#$f(ToQ<kGZdM8@=dj`0*7I
z(yQzhvpjh=;gy!*n$lw+10~dj0NUGc|Kq2Sb?mQTuZ`wk4CIGEC!Pfy)PddAi^GUs
zbFAV&^rkzxV6Z11&8}2|cWjBCBJw^nk??cW$$7nZ$Z8jI20~`T`^kSuwkCO@;+_cW
z?i*q2-*8l{lWgp`w;4P55f)9S-M_kV)4AWiNLfoC)_(64@rA|n0YUXf&|iH&pnYxI
zHUjDLKd?pIM_@SLN1v6U*PA1mlrIb;NpuJ+3deXXq}2!==&@}AH+rM?I_#L}`G)#K
zjA4H^NT4|9C`-;~O<Ff5%NDdHbVvJGQ$>!-XvGY*Qf^3I>2-HTv@5r~r#i>Hbj50l
zi(QhyK15Qbs)pZL%<Sj;Ub20DTLlqd?VY;={oAG@`wsq#Hf(r=0+rMEY3lwSj-jQ>
zvat3}8WIxyUlU`10>TWW@qc~eeC#5ytS>+Q>EsQBrY}>My8hJl4M<LK+svkt-O)Ac
zT$Fw)X<7bh_oL{U%6cK8G!8{X=1$hl?jagOzwCnV)vxr=)K7b<Mc%Y&uW5E8p=#?e
zeWB{0%pllKrBSySXMLdyIcEc*YU8BOJEBlonIq7ADVYtx-p(%juq7L)RP+Nuqqi{v
z_Gva^8`I8MX##f2o>3`)0N-tdHX>~RNy1DYi-)EKtMts?pN_uhyk?L@luY=#X7CI8
zB&#k1G>#QQ4AM1mDYFEz15+@xAL=KwOCp-NPs~9M_9u*fr)`FMe=Z=Z#}E0pCpU9g
z?B#rjB&!9mw;(TF-z)<C7G2*$w5cPrd%}~ld!4$|IY7@>G?v}g<W4u6ht$b6wVu?8
zUudTP`v&F6EZYl-e#wiGgEO{~r;^cb^G~}cJ=id`OIxxyx5A_he7NF7*LQl3t4CuQ
zM6DYr>TRuGDugd5P&z{!Kb2fL7WLFz)){1iRr4GG>xHv<p1VN&V8R_9D{@*~YDhlt
z>>sVg5wto#PEu2c$+Uj%RcP99(0|eBCph5JVA)4_!2GmU$Bz-kIa{ov_`Pe*=5#4=
zZ|l3)sez?%2-2d~H5+YM9z9d4Z4=}O3y)S1v^o0+-w`Y#%Wu65Dz{fUHin?yQp1t1
zVUDB#__`UqmGdtaKXCtr`{pZI&Jz=pJ&(1MQ9tvm*<x!3-bhJee<AFF_0@aeUZZGL
zuTJrjWoUY_u;g5lJa;MAo|~H7R(LW+6_d(=j}REyGEpxp6c8*gRu7a-$VU7j-t=cR
zhKHfzk1B;W9M*$DQ*P@n6NRTH#RkdyZ*TYm0MnF?q|HS5SRgjl81o!#9#kDENrKP~
ziVMaGT>X$uAN8^zcPK`r>GB|)DD<Lp&uh5wsM2Iuuu^41)shuyk-T>D7Z*3Oe7Rg&
z^lCHC`Lbs*J`EFjSB<w_{cPFHjDk|&@WqdwRj}~lSYCPR4Gy`z43DP&C1CT@+FHO4
zSe{B|Ttnoi>C*SjS<z-QVko+m{6mrtrK@XDm~y}8BjMQAp$jp4ptTa}&XwTYNF0+|
zuU$**GwHfe?z@+@YGj8XbA;jlnuuQKtl~#O-ETEMfeXL7-?Txi`it<XRJVi_Ro9QY
zUGgi1g8dQ(dPa-f6ogOEdKN+)hbApLpic4b-Mh-f{sVPzWVXEV2CBfKuv}1VfN7QE
z8!8$eNW=>H=OzPnST3vxdKi_x-p({+K%cen;aWkt$WjEwI~`2R<@fW@Ejn@4@x}K=
zopsInjL9Ix;CzW5s*$O$24$t{3Tx#-;KNRxqaCH=+^6X4tnMUK#x(&m!DF};z`{_b
z<b2d@H#&`AW51cO)RN`~jiuOpR2>(k!-->RAU!8Z?A8|CAd-HTX&JQo6BvI{(l|!2
zFYK;{pr7DZ@vPPKv(fTZb2jd{wXB**d~K%$yG=A}E)UL%|9?4AG*;nv{XnwnxR;gx
zrkAqP1_!{xdmLTb*XLX+pd91R`MXXXwu6c&p;QC(QZq7i;~FW%S~OWkjzge~UL`Nf
z-x2NG`aTFCL}n<$vc+i(9MNwZUu_zz*sV6DM2%pf&w047u*I%u*c2<a%`9OX6UH^e
z(;u^=ulAaQNo!G<8_3<oQoo*O1{T0gs3dshf&*HX9=A^hm>37DIr%=Y`CESN3Ww&V
zPWE2`eHrZtX5l+7-RPUX5COv&viQ1vv3v20GmZQ&;#7@Cj%G-#M8!x^HVwqNsXFDq
zIaYpJmM48ly`|DxkMUbxtxe+WOMgx*CC2+UcKfP-E<YXdJtL8b-ITs6Mpo~4-sGk(
z&jb*kcG_~}M)bCJnVLelN%r#^Z|Q}eub;W_kR&yN225S|3p6(5&Yko6XKn=t2iLzv
z9FT|}INSr@ro0THeW&8=g36y|H)aqM_I!Jx68rC3Syb%3FC89H2}Hz$IC3@<4L<h)
z9`?GMH6=foI>SxIg@JgVoQq$RyE#E;AyvS7)uLwly}WGr?n_LeXmBkaXR#4PTDXf9
z$+jXSp@|4}@4y_drD852hJRS7917)8z`V_|YA2M)$Q7x`5IG1&c9Qy$9deTKPOTXB
zi~qtByLPFBr9XCN4htzmqZ1PK*4OY!SRx1O9Uk-@gppx(s^;>a_0^7TRta=MC}#i$
zHJ^4c1^v!&&_LUEyqt6A;|gPDlz3ty8CSY26m{{Gd?Vt;P+PQSv9n6|psk;_IW|ti
zIL-q+r7R13TgUXZ>jKT@<lhKL4z@GIuiG`>7~3S$290N37J{1AUpCHzLN|^`X(;Uu
z;?m$KX;3b5@mS~RnGD_Sn$8G*gMI_F)UB#Wvhq^S7~$LSUhhk_QO$Th-m34eniMHB
z`D$^o%(R(Ii&frujjUR*Zph2x#84S__ZN@F(xDp#7^1i04cf<bid`K+rTcaBw%dP(
z87Z|1z~|O%{(P))iSZ5htq5+15;wXo&!ty421aX{gJVxbA<DNG`)oS&{W=EbhKP2V
z)rrYSna%mjagK&oWwE1lG+iftTv9zuwU|aYEbI`p2-`(b5aB}+7b5ZT?JI?+l-i||
zJl(4%y0hZ5>$kUB>4LF*`&|FB5XuEtjd4oR$Kd(?{0c(}JR%)5E!dkO)P5Jm5o5tj
zQOrQb(f%8PPD=EGnrdf_i~s;vOhk1!5#Cttx0qnYsam>8q)DXxyB__b7N}b_p2}(z
z<$yAQQEIxE22aA%Xd*)j_nBujVMb3KF0V*dd~tFoQrzBOWJd%TBmQdT;kn)#VwEH2
zf1!ui2rp+b@F8z;0}6Yw;IYh4$?kjD*Xyop%-wupJZuC%MWH?$2w=8WLkY`k)l({r
zu9K7rCI*XAnm0rS0{`8l$EE0=#K4gU$9cLcM`vWLYWVX!`qL_fgugGd%&0p*0Gn;w
zy!U1Z;sR}^<qZYcPHr4OqIf8U%8$dw?lz~HQ84uv7y9JqjWLs~qQ~lRK)I%Bf(_!{
z;=QU8b!14`@~|-NfXX?%ex+_zvu5cRnh%i#95y32VY0A?-^>9mqtT~&F|wjY%eC9>
z2QnvRABziTf93wHe@FYpb%-sG=I545)Tvp$&`txQZzo0llfvv)<m(Tw_{YPZWM<r6
zbV_o~<D`|-$f;d-hH-4cluP>0s?!rb($_o}=c=dK7h8U<LZIsEN3Q_b)<o`L2(7F4
z7VlEVmwbtqII)i`xjX>Y08uv+3t85IyTTNgx0#U66X}JS$KbKpS%Rt9P&tx{=7o<5
zq*s3w#u9lSk<<`0dB`H9fM$8hcCt;Xv>jP+IJGDwEg-N=vIR0Ly80U_RZ>oNJeO{Y
z3+m<`!;$~y1Pq7gqzHX(1?c|yXl~RDTr?es@yxqkc1WfBiZx;bH_$6GPRqDqt(w$S
z+0naM0uX_?Nv$mA<30JcK0Sw?tt+GE&+2&azJ=x}JJ3}5l|>0xh;Z0$`Y3M09=o{_
zXRq#RrA;U4_mOhp7(nG><g1mOeN)cAx4}46UJ+d?!2D25cyZs#4TW*3nBRZ(!(9$9
zmP&7IGvQh!N8l!stYjDueq%r_?Tu$5?|!&Twf~vYJ^AiNtUPh%yQyW?oT;2wJaToR
zB<zj&Hx`zQQt?+UDR9%>8;7Uz93XQ0>WKn<X#eZC*Y9Hu%U;e0*dxEsc$s2TeO4|-
z&Hvtp0Tz)Q14F#g-md+2(TMN7kAkUaf#Rt7U42G*A?w96R;Vbo#k)i<pxIxEf?Fk4
zdm)|2>Eyo{k2VN`BJUq%G7Bja17yWRrowja6#CPj$FkEjV&t{C4g3Fju}}0@&R~Ws
ziQA{A$F&Z+eb?6hM>W#viPXgfsi9@+nE`670C?5290OsfZqdim=KM_`wmuzUq!4|t
z#pQD+ZZrNC6lrC}=PPGY$N#p#Y6(#u+&i))35ofz{vW2^F}jj2>K2a8PCB-2yJM$g
z+qQ9H+qP}nPRF)w>&tWB_Z#<)Ge(`l{;{fd)!KW`x%OJq%DP$N&EJ*Ub)chUlC{v{
z6)Ub<8SPlE{4M9E6bx1{>e{gh3A{?KZ~h)CWo18)ajU-8RWFGw_;_AH|ME!MCo}ze
zJ7F7;aoyv@mk^`va36zU&TgBY!_8j!d39g%JOyxGv)Fk)0?Pt-Li=F<p13}r{Sh+4
zm*5j(Tp7h~D&rg&x+vol0P81cqf(UZ;3sHyEju?_xmTYsTp^`Mi8nQq`h=}6bUa}~
zIq^ASU$)S%EUWX``(5bLYR_%L{RP``;;w*6XT_V7vPN>&ZV=iNkv)Z{e7OP-PGu5&
z@b8dk<nE9V*Py#IUH2;(>P3ZrMQ{NPqg$Mtzv=E>k2|Yd2%S{F3*5aQug<E*YrDws
zYqq{41|u7%G9-C6FTwTVGKK3dj+~DfePz<|`E<H|_gO>t${0h^6l+%hFU|%0FVM9g
z*s=;Wp|MuS_zF2@2|`Dc)J!~5zW?m`(k-DKB89-SL72L$53HH>o<?z=v=&H35R|vF
z5fX;QGrYk-kt4-3Wa1gl6^cRN<$4(Bc=%-pT%3=S3Cm!Nt4^o~j`X{3lq|#qrSmX(
zWn@6}3J&b|swUdgu%-=QtjryE;hts>QdO6)k!zWU4oMV>kF_QPpNsusRU!`VWYCUR
z<EEH*hW)cFXW;?^R6JhN1PT02?4H@&nM*Yg-Eq@v(3jxEcL)F5&Gi-W?uyyGneS^0
z*}scy8sjEmMMD&r|4solmS$0#4wYpX<W{-`fC_~O^>+bxYo8b(4VedZeS4dmJjR!6
zn6vi>KC?KIobITz3<6RzX2cOOa>VY#A99Zla`-aVnL0_f1-|#|kedt^yZ$ij@}60f
zAq>2Zx4KzEkL&&#=;c+tO51z{xchB74w#O=`KVUlOEe!7uepzBdtN&pJsT(+H7<JB
z|7I3~uD4SdQ_b)vxt+ibvkB%QZRv?Bv1iHjR);BO)?&E~ZofS{oK6o!iK1&{l>iuN
zcH>AWCeOR;BFS<VB0gYyqZgBpkEgpV_{p;%*n;|(z1?511l+xNw>SB(hXzYNgtz4r
zT`B2VJ_)Xg3*0q=)WHgUoqfW{XcVdEj+pLkHbiK7krGOPw?+0fOCA6W%dQi34c(5k
zHmXY8EStSON`N5;D0;bDj$Ois?#635^T8H=8PW|A)mdT1MyF4&&yyc@EcLBrdS~x|
zzxZ15xlr66pq02ZnwXPcPBe--#_R0eFvCYMY>qm%!8C&YDJW6#JFI$9Hp1wDYd*{b
z+wct8$5s_U!aQ+u3+NxXbR9@n5UJ(f)x+$HK=I{3g`S^6Z)Z|iC}FlE;JG(}!yrQs
z72_H#(kkjD5y_|wBJZmNfQr|MOHfl2alV=~gYr;yS<~8LZs*$BJJ2TvaAmHgbL0L#
zT6b?!pl!*}K6P`>%5EwU+H(4cbx|)FxiT-lyyphsGyot)B!+vLj79?DFVNh)H{aPB
zN>e0L9?{MQ$ln-T5)O|2vzas0rad;yl5iNFyfhJIYYD8+xSY;gk94j2E<AtmESJP*
z!P%uoM<QjLNqIVJwv{2nO4ypL$hC{4UFl)`i-74yawVq6PONf5?OaEb38msPsc|LG
zb}B%v@mejMaHur~xrk9EaWiN2wzeGSn0F(63zrFxNoOmPy%^Mp+o&V6bbsq__eHY#
z=mXKrTVf*kcPkj!8sTZw1YfPo1gy!c=O(D%0L9FEo7N5G?=U-ITEMiEX8$4^M=Jx9
zhY`ETe(kt{H2k|bQQjR2LH*Sx<2*gjMhF1CX=Y6zY{*tjYzo#`!DToK8Uf*N2t~dS
zNar+`N0}B#mD;4C5{)en7nWjkZNFQGfTRa9#>fm3%IHkEFfx!!6y<1)#9SJDw6Gc?
zsM~9e?XO=Hz?XpSI-slXFeymr%d|ed*w~|A?euB?;?i~N<lExW_td(}4RrIagS!D4
zFx3H610*_UYwlSeAh(SOGcLD^y`1-DV>J|5ck_ACMx}$=Kr>difl$i15DK<>c!vv|
zlKb*k{~CAXZSLi(i_fQEZn9bR^99kw+83wJvX!|^(~|iMwWxU=qQ<J&8YWMed<<HH
z=5va(-<1K{U&%2VlGxeS`3z244we89pWQ@^NY)-e3$E*=&O_-T&jlsXqfMETjOjlo
z5+aP9QSrr$0e=dBan8TaZhiSEL{{l~!=(GQq7N4<MJyVgV*~3kN#K6lp<Vs@35F!p
z1iyT=<kas3<m^s)p-8I6$hJ{#WbmhRXbLE9FVNP7LC<GS33G#=Y$#%p%>4kjW4g;G
z(W*sC*h~T}1$!f4Q;FyDZVLAD<E-=#@vc>t1Q-e${O-;k_vh2ko-OSz(#@m8sS!(M
zUyaEL6%k(;e6A_#yYq^a1IPD7uTG*I`iyJt9NmP3!oR#UyON`hHVflPRDlNbdHd1u
z-}O6Oik^R=Kp3}6f-f`B9$5e}8jaT((cT&)hx79_^ssN|{vwxg0PJ_lyZDomgWvBL
z8VfKKD|;rrOk{XLtI#c_5D3_fQ_!P>A3r1Cw~{bVH%pDp7UQREBC`?fd=ZpFKL}(o
zwWwo3dV=PZ{s+1F^GkMML_&_XYu0+mEneiaUMIPb0V?Aawh0Lp2@Aj%A+lU}^UK<B
z1AgBoo{}@NLw^Gvu`pzb4a1e2ff2)XpcT8kt<oKN-)5|1<e5Y(^W&oZdUEGps%z1_
z*uSowIbc9WAqAFhhI$I(f-1A1l09_!YI-3@+Iy;GKdd2dcZ=%7bkU+i9k_R!=O>j5
zT;mt&-rtL)&SN6`GJb%w^(_INF+PPb^iXtO@mOHA7FDNL2=KD&)hfPAnZ9|qWi%&0
z*TAlY^{2e~_JW#O0sM5ly#gvwL{`ZEKZuGp4y7K*ddVLQ<I8e@U@2B#qfX+VFlL|S
zR1W**wyJU}ob}`2rYqSM`PTp0k@Rz9eQHO60F~UCqXT%hnYjVn40k8a?h$*SIFP)b
zf}f*OWgsp}ofbD*_FH_m%#x^~zEsQ+e;jc%Lm%cL`+)yV%E&^GGsJ;yCyPsm2m~ku
z+BLZi3Walo^2V8Aq<R-n*lIMYioI#~-AvYV)<lMEhL%s8`)l3$r!-7XHNVfBoA>?I
zzOvv2Ho(9{EE>ufIU^Ewa$pGD(ZXnHN>34T*$Jzh(S2dXA@#llOpi{=x(gI2_3~#i
zJto3q!^Rt?BVK|_%(X--$Ls2E%c#Wx)Pi3?Ve*4shE;Vu*-*mIlt&~%vxJcdhvdv|
zZ?Vgw22N0m7~k}wAq-rBZ|b6-k+XS)oBN)>=m4@lE>rBd4}aU<`jXR5)sws!WbONK
zj2xllvi5Q*UK#uU$lvM5A9GR}vo_hvr0JBMRC#c@P<8flL}p3eJ~SAWsIXV8(rPbF
zH=vFuw)N)|TOj_D6q-2p=*l{+U3V%ASLI}Oxm%Sj*_z8Bg(_(Sp_kxH^SBy2RsF<v
zmjJ*_$c<k>b1b$mUm{@is@&|+f`fLFCFAdr2G0lAtdh%)(O1YJ8J1#j@AttB_njL@
zqRU19fkhwVkw?s7eY~d%G~2BgR_Te;RW)w(U}1uRs3moxvh+yZZ>n^r_KQS&SR7G)
z5)sCtnDeXl7Upv63U+#6bbB_z{_8C=9u5G?&?C*u>Sa?O!@;CS^!Il($v=d2^|?BB
z2V}gh$qg54lYB2k9Dw4cLVUXxi4PCMnf3RkVl|&E)mR=o+ahh&rNEH0S96*SMvJ4q
z_;mWwgD2hYdS9Q3P*UD?a6aWA&eD5=GG%LYbJiVw90!YTyJ|ftl7GXkBatUARsiyH
zWD|z*8z($YdFHQxDPadb_!a4AbfnyK2RT}mNRs(grB9FC(-Q9HoF8;Z*u$fpfk#Sv
ztcNOf&pDZgSTGxE=nkKF#M<Ykk|RnY@(x=!9>wcRc4=`i9Ki7%T%{R%69*~NbM4LM
z6jTl;B`#m@BTzC<qxiB8(?R1SJb&D7jZMMq$%(Jp?0<Ny)Nc_(KURp9QX}v9cQMmd
zv$I!0zes=Io3F?e6)@f<>=D?QjUXerxJfz#Ih3P+;vj~jKLfeAGv-hxfiM=tTq+6}
zuG%EDzc|uQ%T||3=rv066ATsvlUD%&=mxJsO)ityQJW)%2{FlN|LjA+ed<3@RW?*d
z>#a%?zfzNr*%P5*NJpnTG6Fs^NVt^Sj$bmFs$FA2)k&7&V?AK^%7_{MiT!+c+OX)5
zkZw4+G=#(E`kuaxOq0YsxeSI<!fE2`sBD2D$j!ouhzSK2VWDZ$yP`o6T1otqPRV{$
zlTO1lYHdIR(#Hiu7|dA!#(DK0+oetakL{;$KejssizAtEO|-fgGiQk;GdI%=8kx9J
z=rc7qNsSnptkPnVG2^58GT3s8fz&Q;`Z)O0Li_DBmXYcQE9qf<s(3I&;UTG|iqf})
zbBC*>h{bwEQ<nc{L0|xDbGtP7jhQY3u(S(GIJ6$qd>{Z?FBPbmWH=Ql7+SA<f`ZJK
zaE>8%7>wbh*8l`V*#D>La8pGlkPDmzh+;cIZXl)h94)S5yHSfw>YTG0P=6ww^q#Mg
z*}q<aVaNoIJ2D_oLP?n|#-f74T4S$({-Hm8D@;7!>0t!M-NWRD!kTofSA*!!=Ajrm
z7{$xbD1Z)<qo#lOz@3WanTvC?z(0N5@k40?_W;IbiA+zgr^tBrHvi+2Bw!Yd5#{ib
zFqjRGNQYnO&zsKNF@ea_g|dm#VzE<rSt;K;v-$s-q)+gPu3~N<82_^oti^#dDA!sR
z!~=>)I#dyw#SO>~7;w-kBCW;cVD+ZK@<>7eFzB3h8)8XrdPv#Q;ra7a1V-8vH|2Oe
z_luZ7d5zz>TYC8TTdwffI$@gvrLNvIdX9VUp+W@PTA|L8ekJ9QWe*#yd1;wvTSM17
z!<Z&>M%!~U6_Ok0VYUq)sO%e21cVV$>HY8xbTm75?)M$~GRy5GKq7u^9*z)$-~t$c
zwZzZOK~!=M0gC+GmmM3jW#W`+Skfx331-3g`df}#-FPuI{Q}l7POi#K^7=e(YcT-v
zCVi;X$TTl*XR;;aT;YdI%(nI2Pm!J;^dE2Gw&#K#W^G32dh>#lsTdcM^@16(M*OYf
zSj^{%*6H?!tMieca@QpV=@!$RJ}?4cr_jpWfc<^kl5M=!Xg8teyPci&#qy}7GandB
z4f|?mVR_}Gw8D;{e4Q-0Xi?sK?8{mkt7viSjx$BhFUgTE$0seDv%PHwUUj06Wge8u
zmW({QfjWF5HjxVZt%}XmV#ZtD#Z8~7dsc;TL!AMKLHys0I@vn`E)G^{4+K!n9N8yu
zF87JA{A(2YFPw;$$tl&h3U!hw`ri_>$^OYGtjYe3&2t|~$VSrZtKWg>8nZ%;;P#QQ
zM@uvJ+KzKQtQ(J$c(3R)PGj1A`=n`3xz@`hdO*8)m`=tZuFnxh(w~~9K#aJ3_fcP)
zZ?t^UBy58RS7qlJ-cLof{bm50wEYO5yhgImy8&mPY9@uYWXHn5&}wB=dBYMBCWow3
z;f5&2L&$;zlx#_8*6qvmRrennmv_eV9<|Rb?gauVmPns(UqnHSl0%e#yPG3TLtKQx
z&lQ`B63=|bynxv{-n6A(%K=F@E<X7m4My~(Zr=Xlw0+jF@nhwhIuBs5RTc+5FoT32
zU3g5PzGC)~r6$npO|ogj#-&I|FAL%F-w$3n()z(G;FOS9+`ij?7=ajTZ!Bl#T&#M<
zU%T6NY!@XLM?wS?2wk_^g?W8cYINAj)rmebv-QIe#-;C1C&~FWzijCfij|FISD7g_
z@TBCw$kKQvH{9>ep8%KL7iX6W#mmR1v2{G{B06weBpkbQmrvNQ#XBVTAdS4ieeKXc
z6_C}RPZ=l{V9U;e4uLgBTgvV(-vXHqHd)v)G9^5@&d(RG^lLO8>y<Tl0x_P(^M$1y
z@S;ouUav+99bFW-QqEGrY-&yFl>`HJ+(eD>KwZ`;%@U<HEdZKa;C62b%y@Ru`e^Xk
zn8D45Jv#`dQwz{kN;*LC4jPB+=Yil8B@O$KNCU>u(5@VQAprw&WUyPGO7WGCI<YFp
zF-Oi~QgZ->e+>geoBX9xVdgdDhJ1PCLbwT3yq4V&L>yVDW4)d1yz_Muwz40PArEHT
zsUC?hEjU5}pMY;BWEno*z1Lw}T%;%UC$cmYLlX}XgB4H%Za&Y>wT9cpX~u9D8Giaj
zG(<*L^5~L&nRw5~SqWn{OrVQrhs6!LB|U-_ez9t=t*c_H(6%8+MI3`>d5D3}u~=XQ
zyq1yupdidPf<M6`{)^+8?TMvSOzVz+j<9&(3lN~r5&<GIa5E)Eg4_1i*`Fgwl3dC_
z^B@_yOc38~C+JS2%tOw!yB@0Wm9iqxGszGH{tF6Cvs`YyCLyScq!RNy_cejC*E?2M
zz)fOj+5beNN$5jcBUrrb(H^qGIPBwq){}7Eni}5Kx#!o9FNcJ9Act?|S|uE6d|q)V
zBv+qpECB0NE!4C|VO2E9HoxGKSE_Lv$VJj!wDtb3muaK3>^zw-gF`Ulq<gr1_2-ol
z7ICn!$|VI0DJo;6!z|2ou5H#FtKnAZBDzVAJzqEPS5J9Ff~)!%EMz>T7u-rtn3XMW
zD|kG6YX3?$IVW>c3nbd$CQBVhgg>Uhdq2L{2nUe#^?(+5&=x=<gm*b4_Yn|z_sjm=
zpm{FTO!9a2+>VK_sFwKz@9B2_kU&HnI36P>Lq%-k<D7a4h_aKv-!7+rl8s!0TfVFo
z7l%YYd}LD}LRUSZKd#4NClbdRLYLkK2u|yBCG9>oDxIv46~IZ{y+oZ-yO=o$T_tFY
z5Wpb!Gu^auOtsT>(4lIXCRmO4<HYfABC&<}ABaTDv-KT%9zvAXDod{)3&m2)hA|D>
zKSG>jDPkgk-Na+7qafNKZYAV6qMxhF_=h_Pb`qldT~j*c63$#wHOaP=>Ge9WLC~y|
znUL8Tr6X<<{Z&}7p5P>EpyCc+ItG9S?W{dQN^GIHd~>Q7M0TZfLvM8Gm@?taNXW|0
z&W*innbeUnCE1o}t}ik^dGpy}ye8qk3goCJQ22Kc^X4$)4NNKbc5Q1u5mc}sB$td4
z=iU7p@I2VLMVnNvY?f*|Ys@B8si>fFAx&SrRV!!qze?Dc%b_y8@-)B_v(3-^SIg-j
z)J$U~tvaQ^-el27YUbB-_Zg!AwFlHVB0t;X=)fl9+bq*jztr!6=B3TWP50do1zYFz
zvUJJE@r;2VrCT%mnt^b~-Zw|N;<DIlmF@Pd??0rq0IB9)m*3ayiJ^xy_IjmW1NM{p
zDpW7r{wgYeG}M+22#`d1Tor&g4kYHLwpm@g79|NisHeEiI+8B+S!Nl~fSnP&_-Dg0
z#t@n7&tEWxofUuPNAh`c?O(3ITu5^cBDTs3>Ec0W>|zhk(U6dQ!dSkJ;`f0V+LZyV
zY6&x68~L+DHqzTghYGE5x5(z-W<X_Il&Uk&J-`xL&;bkfSAgMQ&*xQktsL(;90q-)
zvl+JGNrrFt{RiwMk0SYsv%Em48@58>z;7uU={f2OL-=3Uu`AH8GqiZJbI<!J=d-h^
zYu&pU(b^oOb<~JoUM~^`v3~mr-G|acFCYPK-UH8N+;uC4e7$-CA;q-8csFt$Mkic`
zPr;)tq226{s{j?S9Ce+(gdQBid1MRfLDE4w>ttnVR=$D46OS^bKJWEr+e3);;uTsA
zck-3|oj4fHM#E;gD$O#N!%Ni{D;cWSvlfa;^4>(L-()p*zN>aP2*Hg^|K9o~tNvv+
zb&8G<6@j5)>4x5Lsrf}6pOrYzjgE3NGTAMdzW3r4e*>B-JPNA$=t(SU?1)J;uS7?W
zp<e6$;63#(b%q(g9r$Z`yMN;?exhBvhl<;b(~V1ep5<27+chcFC?8&;by~KLmY`Uh
z-K>GM%KVLY5LonuVc_$~KJ;}@<jB9<#stKU^XDZ{umt^b2}k@uueC5TOyX`ggw`zc
z<!<JiMFljAQ(Z<Qv%Hk*>;&L7BDWHd3Ij}Ebp`p?U^)w8hdL8y)w%X6!kPMg0&8cI
zLU1Ez;)`nftCZC<9$2y`UH`}yKm8~gD{>tI1(VdI_sP4j%L{maQxNztiPs#;?8&=Z
z$Dk&gy3gIRT}I>$z3v(qvp<5h93qqdki6(1F#-Z`Fn6FZ^$xIp=!(XsJahSjK;k2?
zlvU($A>2L!)&rojk6|7IZNy+pOO)ltSrVKOA~(Y{UTcGw7JX?7)mr7sbRhJdjWQhE
zV&4#K@Df}7HQ1xpZh|RS{JS8I5w=38+at2AqrVTdb3Y+lcd}1%OiodZq~0UO_!%5e
zzJNX<FLH{rgKN(P^649iN7LOtqo+Os<|h$LuI90S_G1@CjD|g)Gso)nIi@$*M%^F4
zUm5p1LVfe|L;w?VrGL%mNiJRKabJ^y-ENDk&1<wAiKwz;z8Xg>51c(<l?JWTiTWL-
zZ}2eQC3vPd!=%ZSYA1eSJ@*8yCG6xC?tr;WR+BXcNl?Sqosrj?w`pOo!sC~2W-=6=
z$|;W69?~xr`A+PwJFm->;|xF=HSsOx?7P8diaa@9iR^(!^^DVYy{M6-#U9u5kuOuQ
z&|_-8g1j{E9UjzH0T=39;GdUK$srN+z`x62aA;i&+B3RuEbUts9a{jFnwlRr3-B4)
zENY$ER?%e^PDH04cF5s6$Sl^($?_h2d0$PX1s-b0y^Gr)n5<SBsyLUy1Q;qBJQ5B1
zE-}UY>^SlKh>tmqnO!LAS7N9_QLD<HZ8$0vdmmoVeB=GP`p1L@c&0z&nq1(ZYp@8G
zd(|a)-ngrkTdR;FywKc3l*XAk07$+o5wo`_){G{L;ze?{giZQCRh7LYBbzlk)PlDx
z9NG|T7iR&oR(EzlKX_9RUX4k=2?snoi_vn3cORQ}ql@!dr+l%{K!Y^ZZP$8!_xOW6
zdijMn6IrQgQf4jJ!zdk|%tB}XlBP(V##G$KVaCQ~Vs-6WGP3rj_z4hZ2IQh(xfSQ>
z)!_9vcf|Y?04SoCzR?|`bVB8bjy(ilYThqrKDn1l5C2T}%?=z3aY0xmr)t@gcHO0s
z>#iNOQA&$`A_eMJA%bh5?nhJNjRVO{a#*d4k9ZPr8Lht$R;kG0pmdkkO^<6Ie`<?x
zWF~;gZrX%>()T}v$U|7%0)$N5jaov*Ft=-k{Ph_#I*owaP9Ie9YjLA|Nb=0DjT`1F
zO<D73P;i?PgCSJrZ75hGHsTfSBTFC@Id}DS2MhcG9!>@~c>zmWtt`zP>&&fb^h5UO
zpv*i}SopUEyY(3q7~$_a3&sGwNsjW`n?O{<gd`T0R<u{8^=K(G0I&lAg%od>IuYM|
zZ&N<AAy1v#xme%IOmhmxI*>7N#<FdX*NN7qo}D3LZb20k_HJfM`Bdf6vr&MLkF&gb
zrS5ihUEVHoO{y*^-yvUHO|$nvV*vNa4MLE+jJ+#O%yn?K+6`TKsiC5LM!$%ih(U1=
zA4C;6M&1`K_rs$*0H*1<gGdm4;|*M$BhO-Q^Ejm@6t*vGr(h3;Z!_b5U9>vDC5^~`
z95yLsA)is~uqg2krmI3QRWhOYN#pbYg;zKUA~Z=SHlTTG(!X$nyTtg9us=*NM7RSL
zhj>ftF(!fZqAYF1r>V4d1AeB#*l4He8|<G%EMXgFOe@|eKu9WQq-cRy%3}l4xIhll
zJw`8%DCX{mMue^44bdIFFrINMcYg&GihfkK>JIu>(9QY}D|;~C(T{eUXj#la%SHvx
z@NIDMIkFFXg!rUw;dWt&h_iVp>@$zB?GIZcy0!Y@qk0emX*D_ODw7jlFPTj+iUOqd
zQ<<ZkN44k_fD&<+$Ty-B3gRJS2dDJc%irx!pCylR{SvHJ#RhEFEm%1DKYUX2vT6A0
zv_TY#Yym%C3WW;WsXV#X0}(onoh=JrLLYWv%)x{`KN|Elg7#7(hpRG6OcbWn@Q&aj
zE{NJ*YrR2#jw?F+D9kJtn%@yHm7!Bod3d^??d?*$fHVGoH>w{YBZwXqE!5%TOx&7?
z^4X-2=E|cJA_~->=<MmpumeN3w!k`XFrJMcTL!BAG8%7%FNVc(cyP$Sy+6=&3crxz
zy#{_*BK<CO41<2di#{IoAs$9KrDjiP5hc?{(S00+pbN<s(jfxRIZ8-dncV%b7BV->
zU!4J00J8QB&e+1?D(EmT^v*^9vyR*&B{jz@jhZ^D$|{BZN=;CbN`qfydaUJWIA_{?
z_@I2%%=TKyvq7**f*nM59x}prQarvrOXxrxEm>41{(EIk!A53kc<{DTZd49^3#AZk
zQ`-mf@;Bw}Nk{n}Rx$cRfp<#8q*Y49h88P$fU?gNgZMUuuNxdA-Q89FeVh9(X!N%6
z&lNrL<HwX1OJb{9M>Agqn_5NG-RKa_20+JO{TSRkbmapBkI|}Kd)<l=OMkDkXqgQ3
zK`XlCKaZ|fXL5Z<yv7#*LUzHGPRJy5sv5P~)Pg$&Nq6I)<EG^}vl?ZpzNw_Y`9j`+
zxhS-9W6V?a44glvC=prRVVEkI7}XqMV)ww2*^Y7|@RMT(E@5V|zamlH?hI!ypcz|@
z2!d&b4mt%_1^FJ%(%1Z><%UC>wJbKm=-+IUL&ydr=V4r=8Jmo`yV3r^tD0Z-b;l%l
z2)5?P(!kF8WbPP>p^Wno3SOi@R`-1YVnZ9>BlL~d=6AvIPUX+XP)Q#=|6Q-<3XK1B
z925=QuA<Qa3>TlnK1y^p{e84;gvzfE5WlWj#}7@KHK4&@wMJk#dPX5Oq;>;0@|Gh#
z+BU_l?SYAwgq$?QhxM>fWL{kPW{)z{gfu<arx<r9XV*LaKK21h)Ab&I8}xR7c7{5J
zL#>wqCto|cgP{>8aq%lnl$S$VHms;ah*02rOC#Xft@eYvyzsb!gd_5Xs8@&C+S*Lx
zfmPjZc-@zrhtDP9HdH8br@$l!?xbtAWUyAt0`6!q?QaIc*EA^7=F$617Loh=VODKZ
zA9swSU_j3hznkM{5wSn&1U@d{h9xy)-A>p`uNsofQ9hFMw@ZSfG2pD-OGZ@p!CX@t
z<^3}5@+KGhzM+k1@2f>^p<$JSXZtKl1?A;tk7mlcl|L3^i<ReSmDb)v6E5=*H|-VM
z=mPG9iJ70Q6oN-O>ox}^q(bI4wnt}j@Hfqb&sp}JKi_=qU`?Fh3{DhaMNSSvXct5S
zL6a1vEVhlaAzEf)556kj=$IfEJ#>tZ8UwrA@(EvcCn{~ZqY=!)?cN%|p9+je>oj<f
z8gu{q*XYo!whlhvWDfb+BWRMhc6)Etdh59i18&M`q^n`+6c=IUrm<(sX|Tnaqv;zb
z%761afmxc+{)LtSTK*>yuX=6SvDz~*<om_Vgt(DQAwN0F@~OJhWb0CoaSo=UO#+L<
z;Z0+)2ofKyrvhb80q?4Tod2iafHwW0a}nFF&qKTn^vBr^S%N#0kKM}g$TtCF(6kv~
z+z=LEnkhQC9MHHu^mv@=k<HjYo5eq^gn`2G;y!<2j`Fi2*oF#FN3r*hgSU7V9FOLf
ztkD+XSUy6dU?lP+@zoD!64H=vLnO}BwB2t)s_W1oyq6iS9Am}x%~dN;<PcuJT`bk2
z<pYI!ghK~8Zm;bsq_IG!j4Bs*_hui~VYBVk!MmxXWzpvB*aHQg_$T1oV+J8h1r=T+
z-s@_nK(mnT4WbSpEwBCyGhvZCs1lhz<=iyXtvd9Zr2e|;lFy(^zIy>1gcZZ7&FNxe
z-tUMDF?}=O9IG3Fehw2CuZ{JJ^7J<PgV5cG;kRFR#gXm7cq>(W%Jl`IJ_5-%f9E)B
ztC%U>w)An#vAle+7)%Ul&BsWA?qthP;ZxipUTb?uo~c{FqhNJVYe6VpU1;(w5w%U+
zL_bl=EKxyg1?`ntoV>7K#EHx9#loWot^)72wbd~n$2kC(%=zA*cLX=Jd3S!ONPOnj
zJ3D<~>p4pzR_-*+sx|i<718wVBfXG7=R6{MRjbp#EaRhNZ|ZJOW*KJ_;yS;vg6{G;
zJgj4UjxG?8+=C5uay7@d1Kyiam_?0{=-hZ=n)2w;0V=)_aH8|q7`%OKw3r5wE^C7I
zRla2;H{|aMfYS|QJ|j;;G@6+J5BIiGTPSm=OViW>Uf3p(V;cL3r|dFPLCa4Az6lHJ
zE9ks$CAzEmM|-ExGZ3`AYn+%ciKt)tI$ZT&0qOuO(Xc<jSRJmp(rmeNu58-ucj{E#
z+tzns&P#o54Ta5KGZSmZ0KMstCMVio_>@aHm(GIzHe1W*@W++0%9+@*JQ@ds!(8u%
zz2sSQH*@K(>^~*l%~e3SLb}Mk{&*6C9k1PEqQMYk{3EKXO%@+Lph{nIQBi;=!W-`4
zbX^1pk&#yh6<$pBP>ZpB>8y~dRQXNcJfL2YLbtwi%vH^?x64{^vfdI3^V|I_+rTQ9
zs2fCb$ZoH#0OFR$v3GzMaq`>_PZ%A3xK4kP>C9*!+4kzy?{CCa#9pg}sW%t0H6q%k
zWplxnRWf!I8p#5%9D`RT;vc#Ne`8SYn#XZKNWh{j9RkOxAMDQzxIcg3uihRSuX!5L
zWAUe^OYTDn(Qg4_{LDUu?VuzUU@Kf&Vq}p=yIptFwSPM@oCN#+RUOFai7EJvT6qCi
zaozg#T&HO|P8zy-&x_Nqke3!+XiiJL9|5L2gX|n6R4)9Bh3Sqf`Qq^Ri1qToQkCEU
z!#+Y)4CXWN+W|YKy3u*WFY`4aQ(f>R9Z1T9@Q*hCbAMmzY?sA_QS=xl!y|NAkt*;8
zLzL$HTH<;%uu>H)hof@A+Ig+3$)AKs$0@GMS!}Nhoge&DEm<?KAk_x@iy9RRK~81G
zI{S;Wp($BmG&KXPGeX)f4BAVd4$kg?A02ejbgt{`#qUQVJ^gt7+MJqtpUznZh}!1o
z3yDB#&))XI892!kZhuf+z@=0FTR+n8akDtAH7wmTiH7C9rpj)ZB%Xu(lO6-L4tjcX
z*P(2a$PM*PhDf%7Z^qMS%7}^f=pCNvmw6k=+}YiSF|e7ze0HeOKnWx=9l8&Qv;zP+
z48sTse*^yk!Ik@X%`qV9S0wTc2$0=i<goXixP59_%rYrhkSNjhBa$Cm`Scq^&H7&d
zxH<jAYoXl*r@fb*);P5r%oF6SXhYrn^{(4kv|1<GYTrAC!s7&FnawD}ji9fLI_q|J
zI{729@o_pD&_<=eDsDs1JgWg@VPHxyB91D&tXj|Aw$30-F7->1uzZ5e=@uoe_2upN
zl^CWvbdakQ)NPe!EJ!(&x^GKs+!GOC-Y8#bD>XWhzEx2c--0kbPuG6dIV?yP%@d{7
zbn^M0N`&1I-Fl@+q%1$Q6B(}>238$H^g_A;pD(YSi^e7P#O2U_x1<0L$}zj;xBOk>
zYQI%|!npu2XfN>9;N%XVLc-i?U*+dDkCJgGvv_y%L_KjqP`8<FtYK<Dp%{{*?V7fl
z9hXbCdo&$dNy=&}xE_puWF))c&lOnq3s*Y}oCQvPBwo!pIeX8$#*e7wKeWF~YY`Kr
zK2{a%!90so+&)JVW|RRh4UaSO%fG+rM2f!q$WoFL{xrFfWshQpT*n!Y6B1YP9!7#K
zw?eSObSTT6SG|3kUTdb#i#9udOU5)Dqw)!1;IXY_*ULd8K`XNz0A*ijrzuGW4=+W<
zVn_@=b$Nb2uUA15AR(eim^CdeY%wDsifwLt_*gz#l*NyxMX3RT&~Hx7H}w!E(+)|{
z6#8*<(PpZ;(Qpj%Z?cYimSdAnw$*QR$YZm9*+xx(i0iC@U{aNIYyN%gS6f`Vnm|Tn
zZZ&HvT>nB?6NXD9s7NfUos2JuL(p)~FGCC}Gq^s<M-uH=YD?R$`>3-gf8BCJY$HPS
zuYc)9M>V=_sc-_+DOhCvIWu)CW$gp<l!`LOX6ynz^SBg*a8kYf(@DVMzQy){;(+?{
zgf&;44~wP}NMC71Fb2q915x#BO^gVtW=2JV)#Yx|VOVHcZr!0WtK~`=!b(}Tf=Vqm
zMKf<Y@}1;mdZv_TFuICs6FcIANE9}@LNsZr)Au$Hc?|}*BevD^i#ro9X4RGxFGeY~
z2;4;~u`6tfJKJsslOUk=Dzgw$+84HqESvM}*ZEW5ZI|h&iE3R^-G1wi8@)mP+DBqR
z46`R0j>C`}d>8<R)?7m*|F)0C0dREdZwUooA>wGeiKRPyehq}U91moOfJpK8&W_B@
zU<InZNrV6r)&!Dj><-jkR=dUR6iaZ-buu_e)7`FF1bdAfb)t|ao9WP1naBhF^}<wm
z6Pgh~tID6n?oM!<GE}GNGW%CCupUjVgqx@?3;(XNSldVIp$3ZcZ<W@^-V}V_$hUEn
za=&Y|&A>hpsg@eV!UjE>WuCRiUh@pxBEHO17q<kYy5)jn4L1`*+00!u)pqG~H_6y!
zX`>M)NrHOo<>w8}G~_T$y0`|1AsYrD)VQln)_Mq8V)(l_{mV2d&{mfClODN~k9q-b
zDEU;G>Y+c*G>>aiCtuE{84rYo=8}T^p!`{98k56d;&2JHa4lwji&8KK8_7|rdx{A|
z>vRi1o8;2>#Nv4mPk^o;OE9U_Ls>WSZ@ik^5`uE!&|nm$_0=(g@<V50{=>4|nBW`)
zfz|v7?HH|bu~c(+g~sQG#(dwUdvkHP7VC6NplK^%L_eLe1*RlqH4Ri6DmYc8dpXcM
zk*0`%r!x&Vau?>)<Qtl5Y!w*q2DGO~Dt-lk>oEOMl{c9UTxW3B)cs{hDw1Ky!Heww
ztb_y4>={rG=h=f>xeu$rSq=66w-K)(?N@;VqlHgzc6o&rO!gS9Jr@KclDkn?<w+kj
zfwLO+1=4^ZH{My4EcGv|&)8k22n)e>fdGO;;syoE3(W#1_}f1XbO9II$t)0w4Jhn|
z>YL*Bd3=i1ZM!^Bu!UHk;hCn-P`3Mpw6bJIO5g<0t28LEz9B8@B^~M*BgMQ&ANXZk
zyy5-Zab5d2%Z?3dMQ&MgRtCk>O&Nm{3$2)LaX8-|7>HJ_+hqqSh7ykj7w_j?(E}^t
z7FD#!!I;F=!@;GioI&hQ0NpJZG$7F$fa(XL!`9CG7idYioDqM`u>Ow>*IOaaT)RZl
z^)!0hzto58wDB!<@#&i|gdXPFKvs0Ws9-(1d>_5&3ND?9^0~Jg%PtLetVR6Gl8s3f
z*7GxqNicJ<P&Lc!#SDAr_~E+#z?WAJye!Ok+h}-X_;{W)4B~MqitZ|{FhH3=U*jzo
zI>NNb<;4c`5`juX*k=dX5CMC#LFkh8!|Auw5+u3X9(lR7<nZ&b#L9{j6Dy3+(9<YZ
z%pFIZXO~S(Be&j4#@`6c+=4p^c$ZMzM(VkQ@8|B#gbd%^H10DT4EAPQze4A2bGWrl
z1oP)H?T&3*S3WysEU!d~7{F*2|9Y3rvFsIGZ`i)f9E<E6AF!4V+HZFY?D<UF_wTvy
zxf<24-5SOKE?4x)RGW_#^K3hnww+~fSFKHQsiKJ1=v>mj@ZKZ)Jj&$LXxYEk2@n%0
z5U~#VsS1$mP7p1qu+sicDBzV1ZMEc3F0MXrRZ9CEkp6-fIwF>ju?O7efHLxTl#Zht
z+ock7UfpD7;>5(BKa8t8w0Qu}z>y3mdcK27$fDlcv*opaHss59RAwsg{FVpN#4a&&
zdj~cAtp%c_O?z|187$Cun9~)wnd+cua47qY?Dan7AHei0K`cxNhhr%l2x`+eM9)(X
zJ$P99S)zSs>>vAJ>>=Ps&-0}P6B1%>Wpr(>P^(0riTGN=t6lx0VNuX=!-nv3$97ap
zb4I}dJQ6RlYL7&@gbO{WGJ7R1{x4`ea2z#{Y4u@9)LQW87cQ8ABa&YL;wnEaaV1Ul
z8?#}Xza&cZVOM`x<UM?nhKmTXX7FQJCzJOMy=OHzPrWp_6c7wqwhR9_WBm`kIv6dX
zb;H7!L#uOdG4P9#H3~9HRu&qdKLm>47?JWZbFcam!cO_Z(DD@hyeOkF<GO#hr)8rR
z4m7|Y8S*U@(%f)Iyw)bX>_v4&=-10PJZ7a*XwaZ6Er_*OS<V%X@b$5hKV;ki5#fiK
zBb2Zr?Z>eL1Hes!!!E-<(qkRkGu`ifa0&t4M;PH9GyRjWbN2Ve!>NeDNUo~1ogJC5
zhu9gk#AI>~-KiC37|~J}vcvcl%6z;#t^l>5-upTJI_f#ru2*%WO#o)(eRac~cDj;o
z57*Tp<eQ5TlQP1Gjo_TA?o_SUrD2n&CL4X;v(&;n4KPg!*_TO+w<aFtrp?*FDJ6X7
z-Fp1_b#T?qPvD#;UPuW~GIu87%5>Ts%0kdmKe$DmAo~3DRXxw2JG-^mcOM~HK-AaR
zzS`7GFM*&Fc2edl-?`zPi~3!O8geD|7G!8S)Z3PSG!^eB0Zx$c^kj$tbsb(!oREWO
zFG_}w3echlnSbG@=y5PK#~HnmU`)!Bd9~b{^|e%RxTA1|BWCg(y$E9R?1|lw3i~KO
z>w%zq;TjslHy9u2uGDF7@sZuIFQyKpZ(p-{uW-1{had>z@-b&1G=;;7WpzNr6O13>
zijN)<<MIKb9360B!CXcnd<+%V`Bjh9PJfDy1R%L%l0A^0$~RIaLrxdfi|LiifNoPS
zPdW#h6gD&UIm7o95<k{x2R}Z()sI#Kj4|u$W(*^uG~FCtTTD}D<t!4kR!?md%SHGl
zEN7sYn02Xn;}u8vElp{d(*Ffs?_W4;fM#9X9h-hymQ<(95?ZVkXrbr%?f9M$$1?)M
z2IzlJn59;!ohP1F>w^Ga{@I;{E<o8T@c#_|eQ_~5pnt58tAxRx*o5+6ApaY_7AZ$i
z$z=#tEtH>4P%SjZK`i0cF3z2eW{{j1Ht#*&2e?m=2W`(4eNU)?kY@0^F|AkdgHf;@
zH|Rx`X3&uEi^Ia@E<EU^Q4{gMLmj|SudBIcN<TIR;ssbYRL$s3eWYj;%$xMQi&F8-
zcoK8l?-Iu7H&ll{Bm6vY%knabY1c3}%n}`o(}D*o3~B$|K+>74cd$7=gE3~E*1gEy
zF0K=UZq1hcMCTWo8}3T8G6l;&I~mc32Ya-nF6l-o837=%p)XJd?;qi;9BKfYHv@v&
zKb=>-Y)DS+xW^VWR+t=Ch_9kU_=KWpm<AI9O#j<cR+s`usJp+>beuDZdnV1>E|D?g
zOGeZo*tUgt>~3-NJ;1PLkgI8xY1Uz|hpUd^b}h1&Yl$#a%J{FsmSu~2A+9alx2jQ_
z9Ex=04qS}jyj%rOx9|X}k{o1oDHZ5p<@PEl3N*?iu&PRpdJuBQ5a{SVisF0<=!Xw<
zJl+=KS+dKN7+Ws<6TD0>6wcyQvX}YC5ttqVq#Qf{V_|G#(hfs`%#|N!wVJ<>8?sL6
z4Fq+i+YK;vWq<3x+zSwDT(fZF=Tg3WGF)#3JoC-Mao1x&gqmkiNO3K*7V`nU9xQ=M
z6tI4f^1d(#7zaq5BiWJ*IFKsgFo^viRs@T?V0;Alp4)DCc;oQx6s7Rti4Dyy3y9R(
zNI#U+n*5a+Q#ThAeiZ4YkrAPj?i@XV+T3uussD|^6w0m)Z|9fFj{8t!81vHPQnCkS
z4mv#{;}<;ue4=>1Uu`!Dvrcbt9EG4VzLcj;Q;}X(bI9?bHiM_Kg<nsYJUStYD8rud
zhF6vq7G*E+@U6BSjRl$C$C-NX@k06V*C0|JIWWRbXL6%TnqP%)SsCSS;oKn4A#5k+
z0y~ExD>Nm-X37IB<h0t56}~0KtYI{3;zKC&(q92UUsHiRl2&ubBEB9-=wc4O+Gn4e
zT}Y0*<s)QtvC%TH{wB=1irygBxe7s*W%9X--Csd`D6NOuz2GL8%2xd$cxoFpd!b<P
z|NcOcTmH|9umKSn#l|@$4h8pK0vQ>@@xn+ks8<@jq&rG?w`y+nTg2)5BeE8+(Gw$p
zs?blG1<|HoQ3RnJa;xk3f9G~1hGDf(boofXw`|t(fE)T6ZE++Jp{VP8CA|>5_Xc8W
zwLHYzP&08%>$RvtHGL!Iq|(H*$lPgX&TT%6`ZbTgo{}3i2g5YbliFPPQA%CfaZMq2
z@*@I>5!bnMeSS$rQw;4Y;QZ%)5O4tQVTlr8q++glRT^a)miYu$zrbt(J5c`n_`MLO
zni{>^_hA1jggjWe$U$_qsjWWPl-rNU8(z|34HHP?$$igtcPAE-`eS*U^3QzCtwt&_
z?A9>%IcXaY{M8-4#nIGAdsdxNC#%3ajQRUeKNfhal|;Wjk)nbE?)5*oIqU(+p1iF`
z<b@79x7Vv-IOrb_o8H@bHKcQi5WmbH3%S#rG=;HZHPz3^JsAHkENaf3Y;A2jzlmSs
zHHQhvFer5V-+L)^VML7dT@#rEZmZLyzR>6~{OeYva9Mw407=5a-C4D)X^ARGEA9cO
zg=Uqm*NS%AKtF8{s8S}>aUqG{v0Gz<w0S8^9sgJS(G&;~n}>@uiBM`c)f-J|{-5(t
z`_K9RqWf6lLuKc9kwvkk>nY{9hh!v?nUu#Dv<W~lk{`341&P0$WYdC5;dD>wCBuzL
zbo~&eTDd5+rmtTO3w0P?v0C|`2OphSx4J?cS1SQ{M_i<lbdgC$kVs|d3+<I2ol>aM
zg4836w{eJIrPy35`yibO{5(IC)X#6e7S7%|lRP$Mg{ro^H=@Wn2t+T>>U^<giU1RW
zYqVJ+qX7P0OWWDu8X+M*13+q!ETTy<$V$REh5$nV>Q!WX9u+(**3{q0aN5bB=v=Au
zyCDK}&h}wjjk^D?@d|<Gl!s=ulSgVK6yr!53dU&!t@Ge%HPLewnx<wuAkanNAsGtJ
zeBl2`4#t@U`DHa~pWFl(46K4%iADKlyEpy0NK(l0-<OZG5*q-|*@5jK_=`-MfrtU+
zr^UG;{!;9HI6UV{tRoI9xBE})pFsd%HQE=9^BZ)#$XXl5R*#1PuwL(}gfE{E%|{v2
zXrpl0aW8Ulh=k?V8HTwl;<nq5eM*~$>_Q90i4Nb960I$;J~!9cK1bkG<xscQ4hEK0
zFdlR&q;FKYsHP`X<i?(at(eWAuI@{@io<S@8Qg{<<-M?YE0LXf60EKEVO|UXEJvAo
z7kR<s4Q7CG;W6rf6@sK9K*85!AWYT-ZzH2f6tyn^8BOZIkiA3Eswv9I)Zdcq8E0nG
zY1p#sz#-yL$;b9<aWPS2we#B<0Xv?yF}CyAW@i1l!>$R+GgnjD<yNe?v>Y(PN;n<V
z^Us#%mWP|k{*H2IKAM6I1hWGi0N)>h?mV{iTi(ZoPpKfcPq}dMceO%S-|7f^tzR}{
zq*{ks%(s^#J}S2suT>KvvE(-6t%Ovlrr(w#{FY&;j8Dz*y>bwdv`A8D@QTx*RA*M$
zYn^a@@2KgJ<}^7^zw{8TeR2`f`1}u^T9mhUEg2(X*J#te_;btJodLixkepuY+!Led
zdVVT0(k}sxwQz0iBJJzS8!cu{jd-Id?Wt{op&FuQe>?om1brFHZa}Ml30Kncu%b|~
z&y|iFI}T4X_!DG=Lr|$8H=H-X)18<*ARLnRE9BLTYQDt39DTUFc}WlwoADvTw^p#n
z9jqrT5)Dcemdz}GRvWN;4*epZ19y)Q-}}h=o~6`or9uL&F!CDuMPXKMe7y8#uv@7w
zCPmLNpZ(w~o2i#X)16@h6zkbV8*uzR*Y_6IQmEqp@*t-m*Pa0BfLJ?mJ^{*f(tg$9
z!%vV!7l#$_l*9D;FM-e7id|gst6?*C6#(I2NWiYhsK{?_nGDdT4rYFo4<9J{pBI-h
zqbvr6Y<_2a^Uu2hIPEANp(j{4?=N9Bc`C)8?2Mt<xk~}UMb|)9D6z-BAVItMU(#?m
z$eGjfg?Wtg-YO2KLEGY|Q+gdR>fJQDEJleyP``HW@Nu4g>u|Zo>&6FvjFnA%KaQ2?
zbTG$eWY*&Wen8lM_i_R^XwL6R_^(HaBduoZ$znx|tDBNVmCH&H!qEFG1;Z^T4ra#^
ziS(P&PTB0t%&XR%(RnSP2+n^j?zb7=MCrw-hMhKv3ydYkM~alMkh0z|Dc^+GfPbF@
zW4dp;T<jwEh?&(CHXFcj@$S$gCARp12WvqpMHCZ&>8K>k0k0pHW9zF|4LrqnXNoxe
z?Qbg8)BG-|A~7n`9}Rk6*wH)l0B_GH#qLpF^Y+i;Qo(_FRr@-mjkN)OF9m61GsQ0r
z8u<Mi8Qox+E#SwV7gbctPx{}}ynPWfyr+M`V;?v^aS_46?_qk7g0@Ra(!t;IHJY1H
zcd97>f9frxG>F&@B)5<wWYgaFqoY?K<b5#CLRaJxUsU=q@R1*1T=j%t1N*Jw)Ib<w
zE?$VZ=Wb=Z4lC7U#|Oo%dB9k2Ls*w14pIAEi&wA4qW-iz)89#8c^PTWTCo=v-NB#m
zj!L4n_CZQ^AwIjxURF^*OO$XE-0`n}{`npU=y&1b`pTGD4W|?h1s78)PDliPuemc5
zwL;N@k9q9*Abe&dti6Sx>~Yq_t_p}RNQ%K7qYE?;$!i+glaPWB$*VQdGfG3Kj1h13
z)4+o01nz!Gneq0Rp;@w$HwHv-P9VZpXHQ?DMm+Ah&u))V1Q9yz`_jiuOipVME))d=
z!2Z3R6B+S6%H@Q$y&>iA{>wK{6ea4@Blw#qKzyrMZ@4<la%_J&vXqj#j3wL0g*&`V
z{q3ebg2_m*i@hH|1G{%NvC-?JU1H;Usr-2NQSX6ytb;2h+K;^^=k?T`%5Cl%x?t4|
zd1-S@|Jkr`-pGxQrmMx*^Jw<nM;?+5cz8`3@luql?(7*l_cFz<G1Jf=F<GXgHcgUz
zV}BpEedEl}COwj3gF$ys0lAJM<P*R4Hgi#{N};2}-Vru!s9y?C%170fts~*;RqY?z
z$E{ldg)_e@w10f(W#(SG#z!sx++qoCfHszT^G*Ohx{)$9O(IU+#(3HdumHRR`c!JL
z+8?Lbx~FK`oexd#+MMuI|HYq-i#_N+jV>n8uK%mFH5~Xb@xh+OzXl}2G;P1e_#tx8
zqkC{0a4m=SZfC>)+3vjdDNjwthQ8);oGq;zi8yC)$|o{hJ!*j*QV7{r+qlN7+6754
zkK^E#G2@%QO=;)5<5m<m)x1UrV0~Av$!!lNybk^5SNCHTYJMjxo5dSpS6pSAy9Rq0
zW*`5j0-#_ceM+!EmRY3Sg%nh)-ouHy)!dPY&M?Wx4#c!%k09nDBo7zidTwbc`4Em;
zS8RUqHKu-=bnwaHI_kMf;-TcHaL-k8<(Gcl9-Q+nx?c!+%FfbOI)rrtJSYKKL1{zE
z^Vd#b{B8d1-f7M6WBU@2h%M5Pug)d0+9U4z1SZC#tRJ?g8`XoD+t}65`{Qaj*JD~f
zHu<p;_fhE3a$r%W4q39{hZ~*Dh;S7HMLeu8(3|64j@i8Zm2l%M%|QVN?hN?jYzs3r
zWu@FuT|F!v>(aQrkZ$@2F!8t?eS_)Ski(uvs58@p<OD;E8`>b;ey_oA4{XScvL?pK
zj96Q3Gp+bjX#(0aprw9`1|uBjRJ_NtoXAJ<zr%I%k01D7L=h1(?c{6eC0Hy^(wfRK
zzxDsHQMbGtJKW4{Zh=eQC4HI_5*>0^fVEGbj{81*usLFf;~AHl^<5OwUWxrF;FiC8
zaH>G9boh)bOA@@#Xj?1SEUa4B%Vjh_oD=gZa_cCegb@KYEW2_Rhc731A*5s?WE9u@
zswlC+EdGKL3sYY5&_~aIq?7yQ$=lmhIreB?{y)<L-Cqb-D17dbH)OE%bC%+OnxXp>
z9uD?mC3JY?!exaHtit8alvo|9(*PS(1(I1He=@kR;ZK%$GrdA&2^L0e(0H>r0V1rh
z;6NB0)j+Sbm7D*Mu6F>EELyro+qQe!w%yalv~Alqr>kw-wlUN8w9RSTw)JZ6{r>O0
zh(BKJh>EI+Q@eJZv(L%QwN_@58l+z821ztIf_nP`-D7s^HC1-qcn+Bh>v)bkeUZQP
zCCKeck_jB(FsoR9CzizW_2rYd!7GHAX+u@>0qLJe6}^(g!n7!Dg(hvJ#bAS>*cH95
zREld$DuQ%NQzo~(4b@>(*z{r+mz#bck3gGB;88{@5sq(US|uDi;{(gnpDb|vgG4+Q
zh1k-^B^a#TiQ7gv7bQBAM1i_=S~Qm#O){QJjPyja7RF1YVMjcdc|!{0xw31pvt3RO
zFzYDH%lx}am=`$HN$-DW;swsM%$olAzpMQJU8G{KRvP^grQl@!^)_>({|MM;G$jdq
zt4mhmP1gt*DLl;2^-Vl{4t!YATI6U_k|nBA4bfUA;DXLXS&xAyvLB#JRBn^T23u!y
zHN{wYD~^kQxOpgUu4FRIcM5%I*h7|V*W**-yYEq-HnvzvUtrjaf#9b=;<O2f@&$
zZT*++7&Xx9I1vTnsqCNsQku4nRZY&mh>$hh613Ts&Fm1lW8ELvzU6k3B3+_d0W#M^
zz%g;R>s7QWdExz64UiZ@06D$n|C=rgPR7w{>lzc$g(@y?Jv5a^4r~e0@)SIY=PiO{
z|BE_3MKV=<P?z7x6M1rxg+d1E?%DYve6G>CZJEALW?T^kUpSq*Qa*Hu8i(Z~?o+F6
z-BuOlXZrCOoDvS0-A~yc3#E)2!v-usH_(8*EulcA5N}%l0Wg%Ak&I(vEXdMflur2u
z=afbEQYAWA8aEt)E&Fg2@p04jhtNb+2?2Rbn1H5PnU=d>SORk1lkpf?A5ZD%lXLkc
zbOb)8Go;l5@`=@&k2LOE9_%{DV51NH_hMk2Z@gmkpEy5CzU#-VL5@6)8MMcF2_$oI
z=RJi&s8lMTkktDzO*ywS6#{hko{kShUSYH+q71E4K;3PzG?4lSx6PFeyqF%Un1Z65
zqTFtfG>!+DvRv$fM4*!!+I9Ods_Wo*p?Q0$w!6GasmjIT*EI7{CX#S6O3<lER&rgL
zmtTLBE;^M>*4N346oEhvrR-u+o7DPxy(9bu;%XLPZyC?i@OdSI+A+UvI<jp`Ipa6N
z{M4lD!xgR>)E2Vn$6bT$zJd}fL@I(?_T7+1$mmFgk7b)@@hlTHm;?)wAI)^kXa7E_
zLLh>)GAF1RF(obvFb!M#Az!Vho09i!<tLYwG3sCH<d|VPBfe|W6@$DJOra02i{4}L
z-+v<kZpZrB;;Ap4jqw!*8sE(ihI?+w<wK;vo#|vgKu_HFi<r1cJN_oz78kkR?(;iH
zAA&S}u^$Ai<BZ6~Hq&zmB<er0oiwjsm(=4JEynZaT>3AVId56ynb(bY<UxIr%2^R7
zr~`dd4G-982z=H_T9B?U(p5L1{Jruj$@~|<%OCW&-1)%-4VKq1Q&k+-?hs!%Wov3M
zDJ5%AtPB1*OWa-pEB1nI(^TIGN_fM!?0ZB=r$mZ~Em;&bNW6&g$u$aQr?#AR8`mrw
z*;wQruH-`1>eLeM-y8`I3K#N4YbPyl;BE8I8iA`>BfHYjD%y+M7o8ZBGq4NK!5oAD
zni|P`VQCw=6c)``9-Kex64cyja(U>!3|-fFu+8LGS6Sm}2T`HXXduF1x<unf2CzLY
zxzjEsBmie=_rGR%8*6Xhsq{VkBB<{`uO<5uu#@&fbWr(O;n5N0R7z7_JgA&VN4+}0
z5_siG>AP~*zZNdoU)*%ME{_CGMm;tFf*G_2uZ^;wsH;<xJ#C+{ZtX>r!%ZVI`9sel
z{?DYx8vRMXL$EWc>@JD2I~{W2ym>SmSp$>8&@j@!98XZ*X(44<nMX%bAv!5WkBxI#
z3?yIkaPOk`1aANG74`m(c#`sofTV;yCSaE8B`5I_hgs@f9Ov-MQcL##Q;tgn06z=A
zDpwljxs77$3)g|LU}l+;M%Nk@&tdOuPres)(~92+<K1nWP^MR6?nRxpOqe0{I)k}t
z4&>3%!2l0=guOstrP?`9Vg|)P;@HY5-(y)O-`N0^d*Oy5n0&(>HCE<Wz@L=6Jv6T?
zIJ9(#eVzbY1Mc{3MHlub=E)C0>bQjp*8)BTnNk*3dhah<vf<2LS0x?^RPIbCO+ssZ
z_t4C`G4Z+ibTlHycn)02M={oZFVm`Sz6HJ;B;u|1z2bNXLeGTrJw+FMYb5;B7Gw?s
zJ&)J4)}V7xa7_86t^Mysd6aQT99lC@_v+*`+s8F-YC{v$`Yo>6V%QOY>gR^V^qO;}
zpJG7LYrwVjzah6NkJ61^Dw~ikughV=idaybW1ul-TKYIh^E9``=kqghTBliu<CJfB
z?QXUgbrfBh@f4r6Z{xQ(Qw{PN;+5aEV>tA~c+p0eWEj~MM4%C9C`}OC*)KUd!DW20
zoj84mf~ik=Y(0OC0D^gJM7DK<Ja@->bKeYvh)5#l4}jN}Y$8K~EZ!frmz-=I&EMtl
zX@z2f4fxI3+i0+<3}pK2M*Yf#^-CUC+;pCEGVXL9A=LC>hu*;`NadIsn`;~A8ZTVA
zB%wrX_;+6;E~DxqKSQL)?;<OYz{l}2z!QqOkkc*GKU)KU|EMx6X>WR*d_2^ok<x5V
zNimcT3_U`Al#X_btpBk4JXLHs*(=xya`jWUZ`Y2S+vhVq{uDxENg9nQJPWPi4jc>~
zrF^x$6`B+cDOO_AYc7mn3BgEg)a4WGfS@41+~|!G_?pAMg~y#N%^aH?9RW6c=Sjn9
zZK>rE9qk67J&Fh6Qm%2REi_2VcGEuw87EPNbeYniOY7tNJW4N<O_Zlt{ID&bpg+2+
z+t*rTe`+Fg<13|HwEL6+>xUh!cCci2c8l%wc&&TRy_s}&x3Jc%EXq0uybnegX8|ks
zlUv{pa9SisBPMCqCBYszOl%2G@vC@VD1sqRIOzfYj_!a5^)L4GS-C@R@wz$7No(BI
zexl7wL`8&rT1u(Q;I4*RH7Fo>Khb^oP*Vm*NQt$mK<a9c2gYx2gdDH7bUbdO0;>Y+
zyX$p`x>9Ygs_ek)3DGGZ6r#>{`AQl5jX-iHS%HVvLh^_=bB)KvOcKrD*cWMj6S35l
z^wbADG8W*Oq3NKMVQ8d5;I(E^(^OYiv0S18hm!Z>NS=*`%?Rqj2I<hZe*x-Xs^(rS
zFe^WNf<6>MCdZ=P8eOerWi1lxe)VlU7Um$mpa2)&*+SH2P+E!hHH#=u(F+gS*NbtS
zO-bL&%F|MLuyh~#;Dw<MO2zDX;O)p9fS#ifr<McgFC55J=H*q~j-1o_tuAH)$%5-g
z!@U?7h}G*S$1z2#mP$TQ{EemBMs#xWAXGgK4!UOQ<6@Paj3i*STXK4PplJl6>kGbV
zAe86U$3uVAme*{$AJ%o!(_GN%RLpGr+iXMSe8O}96<LoH`DJX)C|I*;Ic`%vKrjSi
zAyoWE8UAU0pz$Obwv4cI0Ff+|UL1k)2Li}Ue>f-^R5T8=F}Lx+b~g6QNqirh%4d|O
z6SV!1gFd`tS0x^3#DRvxkD(@g$9HI5!}#I;a&|c(GZJU77JTp51e#?bC^GrwD3Ms&
z)tp1TLw-{}WPH!_5n~xHk*1gm0N$*juD=uL5ajr{3@M}IC`~lH;ZCR%D4J4-hc^u)
z%mxuom17WOD-wD{`7bmjO~Mcmbf|1eF{9oHPzza53-*#}<u*@}zVDvjemm^}=O>Sb
zb`Kg?ql5zohiNr3^q0keg2KTWz`%JYM-L$ZA=W62b&34sCCYeKewFqEs67~rGB5od
zzKlUG3zIzt<H&+Rct#B%oQ7nMi5knmSEBSsL#cuY#{1%0hilsC^GO781y&+FAa@Dk
zPTQ3~NcJYWyxO1gXK&95Sh&0g$9{RI+1MQa=@)|>9CGC)`Y1En0pz2Rr78XA{C(}W
zHWTGNgicw9Yqspi_@9~q6iD-BI*;-Sr@rwX>m=afn8-b%rTrp9F891A*tRiVNra0M
z3&lkWV`Md1sxOzmO3`a(%x+TuSeCF@=>ftBwF%KD=vd-T4B1E2(-q{SBXWn!s3Mzw
zxxtnWcSr7nh+2$^2(A$<MiDPopigzk^<JZ}uNPFYUcJB%j|TX-O(cWbg2a|vms-P8
zVxKgs&Pnd?`4--daI!#onvT={IJT)#W4aw1G|N~|(H1li7AdYpIpJ3<xNOkTdA@}>
zuhrIMLN7r%G?6=Hd-Ngrj2EkA++!_9wXgL%Pg25}6WRnL!34vtbB|TGLW?eEI5SVY
zlhlUufYtt|ISGI>PXTFpp4MCOZ9Ayx?9ePffj|>=ZCgf^mR{3%r7dri+drIq>3}3O
z(!l7Z;{F0on-R0hj7A<ZMy2*ZKh9-rO8j~+<8mEQYbFKuz{8NL{Et&@!)(`^>vUb0
zgiMxkGFD2gcnF!osk3PN)wtha(c;;rmj`TQK?;{&0W<(VlIv%}1!*r@@<NhV;Asj>
zN6s1`2VXniN>BAv8NLV$re3F+W%8u}uC83S?j60nvHp07uG98e(-ZM%-&DVY$EYvx
z{{}cNRd&+qQkK)+sNT7*4O|XB81x&c6hb3QF-JJ8q*^}os<Y`+7^n6Dp7heBw(AcZ
zN^ZL#?x%XcYxpdazsjS34D2GL6ehsRET%|UH#}WfNV`3Dah<WS$~#qF&Q=-2FcEud
zymA;c7_d!lOF-~G+L-Ht)f>2sa~k%TM`Cwd^c>PX+8Fmxdgova5iF-fz>evd9?sNG
z9Ez9tF_Ij@|Ie-ggu0qBmeedG;Jn<{<1#;f^qLE~#z2oFae|k6(kvsSKRy_;<s%c{
zKEd(;F9dcLDBE8IoIUTtsW}K2wB;d~&0Iq~6X_CELc--CoJITF7%+4PBRms~ctbog
zXETYYJ;zoF)$}l+3`^L5>UABXG_%271IAXdId*aAh3*8vpk_SEv3p=N%JKLisWXbX
zrj~lWf@2|Mr-8~l<qtJR=mh1CQQ#x`gw^KL(Q4|fCi>Jo+U50M;l+Ye?7N6B958jS
z@!Ae**@07#PWuqn?&a)eIx=;9F-k0i*yeZKLLR0#Y`6y4b@g$o9m;vNB~OxvPx4JQ
zHMN>7;E$HlJumD4!OV1|abV&)xQi|_!>z=qe0YG@@n<h;Fg?agLBK84?MFz`l?*#L
z>rDa^a>n!WlW2KU7tR>D4Xt|XUyXh3c?1FH8fs7_Nf756oo3LszZguuTW!80ze$%O
z^`YC;WflFYs7HSP2>G8`*?Hk7e)Gmm0f2Kl*8t~)TB2QmvZCib`L8GRzw)08Jo1!=
zuJ42$WU(s4Isl|ze56ip9`-?*zR?0JgE*WE3T~SF{12HS;4Wa7hZ~eCu;M5T<bfoy
zs>dM`R43h!36ghFE;~sy{A#KSja5-F0XS!SXL3sQSCq}3Qv$ZAnH$LD`D4J)vb`hu
zwq}iai=9G3v8{BAo#LeS`n_B`Q_0Py;~-o)6Jf~n6a0VV`3r!;T9W7Yq2x*mk%7iE
z?9CcUfbqgf#{2&Wy|rZ7C@34a=8gWJ#}dR*h!0}&KjQq6Bs<O{yj{QDmPaB;nCGrM
zF2pt&4d4LOR|U%Sey4(_jg~VLi7<*SAFbb(ADCM}b&GA~V_S%}VC9T~ZJ?a$qdc@<
zkm&b2i|*%NZtDB}c}Smg71(kOJkGNs)GvQd%&UlmuwDLdB(G`TqoQ(o$5-O4xtMhi
z$4@7ER{g>B{uG|F=9dDEuU3~_h_u!?fRF|VKr&Uu5maH<#cSq+EqV5XOj$xJ3FUGt
zi#q0#XhYCa!}0pL=B`PdF4L%9MA_W<c{;7XsS=z1;Y)csi5A$>wm5cJ;LcmxFU;5V
zS9}PuWd`ZiEc&jM8y|dZ*s+=$>%3YPj>+w<zLys6=&v4OQC>E9sL95Y1vWF8W<)zp
zKqS@x32k&rfn7RuC2Q(ka~10^XZ=>%9vMk1bR85k=N7%RzD1)3I6M!`$40l!#45`(
z&J|uDMOr@bohdp;29Bud^|ZUV0zHMF7v8PyvXx*<bTU0=#HTFjIz{ues5W-7pm5C|
z?>WF2(`+j_+!(UPfp(qHDwKr*7{Uty_=+osK*nmwpT`qe6`s6&o9W`B-F(oZnNyvE
zx5DK=*R=R>zX_IG@$f$H%J&}piTfis{&1TK;m$tY&C~R(;d%w^^1|1;^G4^a#9Q?@
zvU~)kZDlEtr3shw*LM1A|44mCv;7iZ!Kr(t{k(+rp?e0N!iZ7$E|$w(&~5b+AZP5O
zy3x>cM7^NgSDuRkYjJtam0tp+=^Ss~vpHNKi;wv(15(VZ7z(?FUwaxNU{r~Fu-KF{
zr%}R?wj6hQK}n2&*?--ddz5Pk33U*u?*mi<sGm(lJ><^STl69)SHE@v$3aGrV%x@L
zAs>uqm$NiuB8QAx40SsR&4C>UL}Wlz7n9U`tWe-@l-r9)RPJu<cj#$BMqDJ$PH+>u
z6rjPAkARD?ximh?5Z;b-&n@{<BCOpKbZSXyC83UU_tR7L`Q@9MH)MNH1P&Xv#2@H3
zSWm3~jYV_#sU8%B<h9TT!2DqvC}xV9GZm9Pt#CVHAMGoEFAVbtmYm=Tz}$$skmyX!
zJa_wJDlgJdA?|D)fOQ)T8|DU=ce3Vm%>zGZ0mZ3_z*W5zqvu!3$goJa=*(`R0C{5q
z{I8m6?o}Njrl$O{6!{uI>Y}G`czfY3CeBppki6cYJmq1L%Kdy%BSZz5GzxJ;{n*3!
z*x7?bPi-};+_m^!<kcDgh?z4`fU8$AT%-znjSv1WD~wcpPd<1A`a2kSSKl|r-%er@
z%MW2|8U3<fwRqFs#2J7hnHaaD*qU!FZH5*d*gyQNzaf9ip>F2wM7v;b=vcTQb=7C!
z%siUO;gm@8JWmI3&he)lkm!%OP{lr%`jsFh2ks11v5%cLTp1DoEbJB?B0A&YZa61=
zMrJ+&ia@WzrE=~{a_iu5THMA{85&@1ziT!`cms8I3r_O01{+`n-!tYSC?h;dpJh(Y
zcFg`zha40-Gy2!}CPY1;3sx;h=J;)RMyRt2i{ry|7i_{$THBeWhQ&L;O4Nwf1eLhO
zckmEYf6ksGb-SGc44Q+5zL&ZM!M`GPb=+4EeKJiOyl42=I;*>ymr`QQKq6{ajeJH-
zINuN;$U&bjI=3BPg7%v6Se5v+SOxMDK1Du%l<)T>kKQBqLnsr_HfIsJysn!Xd{^XR
zX4|ZV+m7BZ0KZg1fj|#tjtoYYyWb09RVGJ;gU4^{HAPGUa2Sq$QzkEWD|eH!$w|kM
zVulcjgrGqymW(G|DaA|2b<lt`{E=wzv|RKWp6Plo<yGvKM7x#4WFE`c*IXPklyJRf
z<^P)!)(ct39A-Jn_@EK=N~C+hHoQfUS$$^}<)|A~O#{A}H)D%>%tA-DyaLJPLu1^F
z%EQ}oS-&<9fXQ@&HXOsccYz$O<_L>}XD;tXxq(sV{OQKglQ6fV`ZJi6)bt)sPOQHz
zGoli^G?b&~blw4UhpYZ(&Gj>jxdKj7u@Mh7?dQRf>PmYK-obklGo1x;I^&0|FO(_(
z&1F3BIPSKwSg@P(quF;f#c_3y@ae~e@P${<a;GO5z~#4>oc*Wt=2F%Ra_$rCkCtsU
z3(=U{jj_s?$y)S-4jIgAY$X1>P+szKqzoHgQ`SW%{@M4i>Jvr7#d6VdQHOXUgBZ`4
z&xlNcgWb-NvmT0;Papo41^Wi-YXOIXrx3-~cm+3mCyo!Gt|~9Q$)|l1yzAkbX*x`<
z9#iEBK<RP>rs>Ca_tmL|nBd{~1=%kw&KttjkGGa|jwv^BADE^8tgQJv$}xB2VrSKU
zX%I)Vbr(E9^w8bG476ni267fWX!G-?=m=*|%L(kJXdX4-#-uZ`8H3YbDJDpVc<QVK
zn5w5c$t^1W&(!t*#_@w+G`>F7##6h+zJLX2jX<SIctT?ol%dPG&j0qfS?FtP3^O8U
zWgKdg-e%49jQcq-eYF*Dg*P8qDz9F<1r|5`cP|t$jj^H55!u34YAt85{XLqVI^hZf
zD9xiuzx18_&eqES45at@flUL`U$#p!aM}#_8^d+~#?}1M3J7mb%~-V9h#6p(?xO}k
zazGbh6ci<szA%hYQ;cykqZVN})+F1!4=9^BkN9})hH=n2*!hA}AW|?8TSbA;i!G#X
z>0(w4`ebbhTZvl1ORn|1Gh_~mZi;q<CSv92>0tjp;7a;sQ)6ACT^E$pTrP8hSAw`b
zxdbY8{RO$Z(v*U!^Uo}CxyC_Z>j4<mE3to*V-egzG+b|NcU?=wueO$z)S!k2272$|
z8*U%@=y%=mFX07|)@^V;BbE_)6j|tePp9gQz2LQ_ZtnIGsrOE&$IUK6BH*lZeqdjA
zpXicAcrAQ7)>J2qCQ3HQe-q-^^Qv%~o%OKkDtUS6O*N;_W)vRcLJ-wOa02XUV&pA%
zeVaA*Vho{fJ6goQdV?3{mnIRZMItyp9Xq%iGZQ0tH?HS#>;*7*N^qtY?&>WO!z>~q
zyxk^L2SR@!MI{2lLJYeha=s%RA;IDF$*0%NI~o%H45HeIjl(7!Y8t_zE=`%QDZP=s
zU)c}%xY?!yjsE~OwPz(ln+E&?n)iq9=!reAQw*0+nyfQLo}d$dRmtyw`5t%0Wk#~h
zI9H&;H-LxcFfaTmJTYpEZ6sR(vmlt2E6j3*0F`67_Jsfy``)oJwac$ggA)9uI9X`b
zx`@za4_CZ13C;6?P32ll`+Azkk}28-H?oKqYgcGmU`k-x;Eq2l251v$<DD?R;Em7l
z8|@GO4w)EY(zb^RVQJF#M=IF^_GU4HA)K{MM-R<*+uk^<G&VAqLfin0tX}#mzSyh*
z!}qRYkUANktTny(JCHDG{ox8tbQnZO30gNQ=?6>dK5^CRcJ=zr@eP`&fd)&P>(de$
zRh7mBTa(d#5w>zc0B||I5f3vQt~M2kL=kE-5L*E`6BlPD5#$t&6N`#&8I993_W1if
zmBskZVH>Bx_Qc09TGML}r(ru6hShvLgMo5E`XAe%O2h&ok(Vr?%~f%`@lyEnZip^n
z2MI@p^%)T)qVFnEYjB&(E(*=5H$9#jmcvk1GT6_$7pJ`)pr!#0{l2Z`%cVaAY67D_
z))(YA))EaqG}A^j+ZF`&;~dn4&vv`cBs;_j%eHT7Zh@8sRcsz3!wQTMCQ#O8#8f0|
z*+?Mj#g^RWk1wZP6dcCS?Z6-&2B=sWoe?esa0vu~STtW4o?%4v`(6<yrl>TIBh3|s
zHLcFM?}&lhfC(ouPiclOqe?(zDaFR486OPfB;6$+%<pFV>7k*f%W{UDAvbVBNoThI
z=&@_D^!ie_kS4yan!%F7GUQxMyb{Cg;w36oR2nK=c-UTiN{^c>mChzFu~eawHmuiy
z(yr5^18H4wdN6pn+=OyH_36s@ELkY3UwvC00!n~T(aJS7KwE-s?|<w&udd*IC~3B8
zo$9#7%tk_1aRB6be${Ai$j|yL*jY;RAxh&#gevo!8btK<UkuiNXf(O$dcAS~j;A)q
zJET9w_d<gHslBd;=dRx7H*_D*E06m(Dq;iwSx%T_mpB*NsiFFjHRk;BUDn&F9KcR;
z{<q<-LW{wLu1aQfNw@N4`xNO#yTf#hi^=wc#X0m#MBtFS_1SAry;E=SgLN3QIj8h1
zVtV96IBvu@I7d{)n82-Llwe+NHt$gxY_wwWm^*d5l0564$NAMk;G9Ucmm%;=O_b?)
zEckGA6cg7sc@+CG)@tHj$-eLu5K>`uAa#)3CG)wxb+f6YpSt-Xy_yrS{-)ASF<HA#
z1te7d3+;I3aDS1XPx?HnfB1c>Z=LOD^gO+oblfDc7bAW7=}`h(seT)o>^I;AFh6#M
zuAhy_T}5?WTFkZeRE>8DBzWI08u0oESTwX9QMo<;GKsZ*&kwVB1I)s=en*+_+TfkZ
z0HRf;TaT2j%In#;eQ}wkvJB0p&i(r)X4L1Ox%HcsIchWL-6V8wTnP2|on-b{Ew-YI
zEDn=XFCHV;$jl@z{FGlo7J$g9?_P|~u}a#$+xLdB(e=FKBMj`!gy}&)k2w;Qj2U--
z7F^4_<5R<U#%#L_1A5;?RN9DdF5wXks5U=g;A$=FJiDmX8S$>OU5%T8f>3Vn>nTBR
zSdEm$Id}fh&`~J<?rut~kM9~|=#$K)kRK!pQ&EHSqV^ABdRVGy1fw`r-hJv+3*(2F
zxjhR1A9n*NusLEIN|%*0kz|Of2Z5r1rLdA200^j_XDifmwR_}ULMMvWA+P?E+K<?^
zyhYp{9?K{<5$lgX=a{g1e52BMkDH)H$iq;<U0s;4YJyfVLuKh|Z`5?Hi>r2MzOfs#
z4P_Z?6OYd~^Sb)vRny23WTV=@285|hSqcZ=iPLO@%5b!_A$j$h)a+V#_WlP5DFFN*
z!711{dzW@QdS+r;AP{)(N$w$x|KneY>|X&X9{fuO+;nGGT4U&Oo7OKNmW?m<zj-g;
zCAX7pF2Y`EQ%SnmNdFu^LOwXGO}Za}zpP)fcjS#qi~<vAMwR!9)?Edhsk|on`0kv(
z13YO5HeZf>PVz3g(aixtsUdD%j}t8;&v(Hm3T;b0SMcFm{d={H^beK<zC)D#%*yfI
zf%hhMD?_eR>^I$oXDI-%t{?PM*Q;@yzVw|yS>IK@DFE7KO}T01(Dtvvt3}0<(GcK9
z{4=C~CjF$@1^GKo1Jjejv%_}J;n<r_XKEQAE^Vedu>It~1tmZ%RT}#%ugk&CeoeFa
zgASc$lw`WpN|{MoUaCbhVoW`Stj+Xo|J+%bCLCXyI>{f*Kh@RcW9w=HBp(J1ga)oI
zL+eenc`|Lb>y@8>!{@_SR+1#EJudFW6xeD$XF9K=V4xLO<x|3ho1F3E2Z4z!<$_y`
z&S~RQ7qm3PV;2DXjE%LIPpySi>&xDvXMQyW>8_tP02Hu~3Ly~Y)9V~|)NFSas*#wX
z6t1hOVG&>@Sz_F`H|I+A`$ox>dO)G{0bR)yt2?u|CKHw-Yf!^sXroe#c1|kW0dXG^
z!XC>TjDJ*BVFAu#KZG7e6;p*AA{U15uFRs~Lza?d%@R=STMJ^QH!Dr9d??OP`0F6E
zR)Rrc{g;=-gzE%v8T5s_zQ-W^A6qem2ql+uXk$4lu7z1py3IWUM^I|ynZHkeB74o-
z#olAIu;vpr*LI`oip@1eZzu5;Z)13C8+Nl*5oPfP?ni>ZVQo#BF~{}g27e&+q@2V~
zlDxSD;i?0Mh@CB3(M0`3D@sCQ+aH4?@ld<H?D%?6Hd}v^@j~JAHuTIMf6!*2$Pokf
zm(fYONPe4O!HI|ac0Beu5TouyaX(K=@sh^r?Y?B%hOBsEhG&~%7j1$t%YgL66RECk
zmTHlONT-~+N@JdWD9Ik2LUO2LW3SpJm8|}SJ4psG6blL~5`u;2>yGB@vXDj9Vv<WJ
zl2a%8MFx{&K7u?F!~6IH<&}%lYsSLBesWVo$ywLYf#Lg*s9V!rrG%s>7Yi`MGRwo`
zxcDfT;|d%VSOtsaL&i^zxBlt8-7f>12qksDXd+f64OUcLVE5*`#AV=Mn~mP=#~rCz
zhH3%)XWA^)J35cP!hNlMF$YP{3m>ZNd6T%)#}Zllr#3<BGXX4agRU>vO@wZ`w4pyH
zDFyhxdaGIO+6P?9cV>-dsG9bTALcNYPn_A!Y+E-ueupj;{6QXKxL_%Gr|rg>YJ|TJ
zBT|5Vm?&GZS+B%KXG@u;xiTI6i*Ug29iIX41O3$ABidxozS0Olv805U6+r@f$>yoE
z@oZeihW=iZ)uff44CyqZjo;sw<{~wEyLbeqr4pbl;fzhzQ~Bz<Gv9-t<-4yRVy&35
zssi~cYVxg1Q!A7qgkAN=w}BK{M%2Q&lNBR7)_wv)&s&Fr>+lf)6P9RiXY)8kJ+WXw
z*lFS6GgGu(2ts~q_%5kPd}m}7GuyZRTuq%|_G3E4s1osc5i`^j@r6XrDtat?DH4Pf
z)X<WC;|T^jLYG^nng+~{jmumM7gU4!d;uabGSgJ79vi}reAQBtQ&=+S!z;v8E-#SO
ze<S}`6#W*Q#-th~nP3dbjOie_8jxI4P;&C7-4Vc!0&@Jl|2=;Q%%d)5b5B!RbOv0!
zt}Zj9=z=DB_-cI^sc1GDyH6;nl?B{8Y6_NUe>?8+Mq&rTOC^}j;kPx@*rREIkW=JL
zqL$$epGOVEaAjuE#NQinxRoYC0xzS;&E^#)=g_Kkx=w@GIPIz;c)jFwXh85OSlMd-
zZZSmwYZme#xqQUzh-1?X`mvO{ccFSK8I!3^PN-CwBn%&_=b^&%Z3&W9E`MKk3XJ04
zl0h+2irn%ZKE5vkmU$Kx6GX5ca&ji?>7^{!-m#v3HOv3KL$SKtDVQ8@1~ruTEHbtx
zKq<D$_rB3xnp6e=Dq8tD3PKOiuBHqLgZN?J2(Dt$JZzCx^6)0=S)mfX7bO>6MHBZF
zEX;{WykLP%X9xQL6K$A+%kDp+K<Ez;cNF*nrX6YONU5%0_#F2eWcUpJuxf&6f`T*z
ztGe?H+>XQjiJ+`Loh@3zn)A)=YQ^#K@%o9w=_(zdo4S$!fJ;)opNz#|YqTUvq;FNF
zlnEgfP8d@qSA{o3Hbic$hb~`L)vQtcT~TO%qWU`mt$4($VIar;2AKFq;blgPLw6n7
zuNsPdw3d_l%&i_&!S}ZR76mDXl`+<^jGK~BboX)h8E}57$m1ZHv0m~fxZ?la%h!Z2
z(|$Qy(sTR}kn}aToSlB&cad;S#;0bP)VG|A6cpP+wVT<J%EJl`6^H1l-&zp1!ejaH
z+Rxs#`R$j-lk^R<mv)f9ptE8_1~iOJc+UK1f@`2+x(>`U88oINd?n};-oxcNNk;p+
z3g2ymfU!7?&peI<ZoJP-nZJ9G#5Ss@Dghx0M1;{gAgGc=av)6#FD{$!1jVD1bMD$}
z_Vkj#ReR>#Vsh#zfGjR&C7`5SkUj16UMU=t$2qkt>g@LJ?MWLXMc&lV@s<Q_<@Kzq
z=o?5vfZg{$_40H~nsM6x99=yYU#v_`aTm~!W%*+Q#bon^2;wr%7$JFoMK7MU!hZ=-
zJRmM30dQB5v7CUFS%T+RU2m8Q?QR^ijplx=&Bi8B@c;M~V~IEG-YiDCTHCoM0sPcK
zOHI0(nn_J)=kxec6?D<t7sgZ=Kua|`@){6Af)r>s^;uf4Q0`22S&OE~`;48jXK&B*
zD;zY~Em`pX#|-7qnNYNfLy9?OYnO?&@4<F3fHjon6ys<!q_yCa8BzqHbxK^r@HJW0
zHFc*<HrdryyV})I6?<HN(dE$k8Z;%4N;~JBkaD@#a)u}}WsJ3b<kiC{yYN+0AtWWd
z!e(hPQ0itWD4c%IV7v1{`Koz=WQSPs;Ek0~yN36KAUH<Q!}1?~#HDM$O7|Kjl1}FB
z07<T`>;#XntFBkWr5bf#O$F__kE;?##!oY6i)w>yxj;*t55(u2an6_Vd7!pD&k-tm
zrdF}Q$tuEG2yv_*g9M$Hb8z-p0B=K`6z3Xc3u=>1A2ytHY{a^{l+66|&77*4^wUXW
z-|DMvtc+urF-u^GQ_$rMg2hw0Bhm5!zE<&NkC7iJbHek)sk8WawGy$TWoOeL-a%qg
zy2FCKn3Ck<vlBli{%n%Rr<&bRF(aQZMRJHaehI-{3PNIBetK+w?tC7&LuOs_aj$QE
z7sL*+N}e$N<13D;$y>GP%Un}x^J(y{)n{+0261v$a)TEb>}MEs)+M!|L0eV=YF=X`
ztUr^@#OEcqac)Uk>%Ct%Sr`C*E&R$J%_+8;IpLTFct7a5>olV^jbDJ<J0g;fm&<he
z_2K%nsT}dMM+CwN(mFOez~VdJH*HsyVm$4YcA8|Hsu(KE2VTCDRl^yGBoB1idTuUX
z)Mk5OvwGC_U7GxFAPf#cso3=ZOtM%tFx$midw)$kK;|OlL(`v)O+RdX_R%P)_tCIO
z7zVG%@#W^LtiN)KIk<{06Cjl@R2&LBTD^F;>~sj%Nuas0wX{4(W4T~K(H3J`#nuU5
zk98!7=MD=_AtAn`fieU56uzlClO||8mO(#VPH*cp`_^n6`NpmvFnPuSXQ$kljLH^X
z^TQl60TrTQMA15{F6B3Bz6l1VEoFqFF`;Mlc5x?85xiRIqj0vETRYw|Ihl;1$9M+^
zg67-^ab{zCEKS0x7U?%ORK^Vc`}B?{=?(s4**~}+gJh~t_CtA)35b*4^40Tv_WH|l
zkSIPA@yyNN3AoPPWQ;)oZDtrL>SFSl;_`G!KVM!?B9;E|5pMkQ%!cEuygfdnLXg(u
z*5Wc!H0xKk?)NwR*5_Jad4eq~6Qu^GmwI7k#FGF%nsO=TS!UOco*Y2sYRT$_iK@`v
zNXyHhLO9}Ez5sS2_6SPKdpwaAI-q^G{Ml3xV$>8}3Y2{Mnov6e2LI3}7;Y)Dx@pNz
zQl&Do{}xE!%;A6d?!5X=n1xF~G7*IIDFZWt0XxF_Z71nPSS-Qiry~jp>D|kZ7=i^o
zH4o#nMZGnPbGJ1kf&S}q8Ef+-PCOuduanj@(_+k<h*WlnU~9!9o4*vNr|*)#d2vB_
zZpxMq<yUPfSeiNx;I(=ko|`oO#f}VzqClK0ghTSZciIR-&`<}4Vh$IjRmDpPk*r8o
zB;q4y!=+6$B7nb{r#R5)D5ofh-^nI}#ZpA!#b;P&$WVykig^E}N+}(l-}og1c@cG0
zCZAX7g)mFytjbJKGhzQdB$iU+Zwdo54-aMErNz9*)iRd}fR}~O<y!tXE|x&W3F$Au
z3@`s$+D3|#tCZpe923f0+8)@z8sg=uzrz6yrXp=9x}Y1&+_YJ0`42u3)o2OYZPkn~
z*YegM_|N^}H@eN)-VtUHl)<BrktIBpxPOxXh9`0J*ue7PwWR>7ueKK!^VELN)(&JN
zRXrS)`6gioz~@D%gCc;h<Mn>Y>`Eq*{7WlLErb}GWF*Vp)jwZ&M*pJwWubQF_@eix
zC{BsnmXcKHTPU%{#?PWH3Oz^i@4f4kzNBs}Az2e*5~GtiZISYSRXc}5@tfV_<PUTb
zl+zJm{~4ujL{w2Oxp7<XL(^uWm@SNS#bNvf0f9EaZjScPc@T-Ym6{M-<P|8A+-Tb}
z^GF}j8V}SAm@hI&Oqn!M3JD6FTsj6~*rVbv(u?w)1fq%<8$w)dWpU2RaR%5MQbCrQ
z!>$Xc+UFO4e<rG(q_nRr<~^N4>3z)5Yb=!?Et<_}oaglHlFVs$ne66eEla#(l-_Er
z;7HE^vY5|o+2~;&HCW6e1DCroHJ7jx?>svHm*cK26Qr)N%^o4!+<u<BrrVG6lpmH*
z<^Y)Y!$@jn&PAzu<TZHJJ2yM@y($CveJ0ygWf%fw)v!(qg4^`d=UBH#=nYqD5jS&Q
za;r}$0`G+P3EhMXj;nI?zp8^#@8_)-c?3j&Y5Irhq5aRa<F`%@dJGW)SNp7V<A4P8
z%^{NsvR%9Jz1b~i+2w6)UmU#hWv$F9vXeO@`AhU2yz9^DC^^JRkxxLeV}Ayc`IzIP
zZ32T5Pdm03gwY_Hx-B{N<?P2N)9-7^PEsDC0!gHdt+X<5`-U5`nIx#Lhaw-f<{}9|
zq0el|W8!gck3qEO5$c3Zg8{+lSk&`~7nxpkyVd-*U)@hX$``wPzH$BkF-3QFH%B*8
zGBLw-8D$vU-Yevk5_G8JL_Fm%r}*!bnk@2s$YAk#8i%tVr@)$k6UsMxhNJyMYQ6M!
z7FyV3M&FOb{uCwo=otS*Y!T)G9A4xs;b$d<*rCsnCG!$LUR4-*9i|&&E0tKj1}Z`n
zz=$MZ6Jd^Y)yM!vAWbCZDee2|zee#y2!`Fi&Fbl)MsAUYi=ohXpp2Tqu@w70lY~|;
z6p@8ir<xOoiXmtr3A?bEM!mzcm>$X=BES<Wld>4pu5g6$E|;&E;Y7#*l3$l|*kLMn
zd&?oBw-FxPLZ^R2KZ^)Kvt<!U`j_9L^r4=f!5p!{=bT<wnuXLiq2)tFSH_`c7j#cK
zGXC~wf^cPS8i9KVu~d=p^sD!oIg>hq0e3&;Z{Js9=}Oilc{rtTn$#J`r$AteK<k=c
z<>x-f5O#H?S@4SbU@GPZ2-;thC{sw<P5r9Xp3O%nr!v09Ci;BBLHPQPWE|A0%a;Cx
z_7RH%gZ?p<t)ZW}Rsk}tK1r&kvj<M1naz<gm89OW!DOy(%EKpBCJ@d7pD&<5%9T-K
zvc9}UZREc`)9}H}^-EY2TBF41J6)g>hpFKoqwnQBQ!Ht0vW==Xfd7PS>A7<|(0i$m
z_}0Fd24ZOH1rkeiuubX|#<w70GAuI%%B~}MeY9Y&%D0{vU<HD#C>$;vsQsI@1h1Fp
z=U0Sg?7nDa2#|Pb=FGZ|wYv?S7pnHK8qd9Y8rP=XXSDUNH|MAG25DVPj+c&?6x<#a
z{z|sKAGfGc>~UP(E03yzd=p_Ke32WoJQJ}WLjP>m$0DI!D--6TVZ=ntBgxGI>sRg=
zAEp#E1a{{LuR090S{eZXBpKm4ED000wMPABZH~|Vluixf7-_aqpXku71fj4eLYh7S
zBo@G)8UC3p4&-N<Nt%6o5{-<N;%v4g6=DM$rv;70c?Tj|H(!D1d_C5eKBNrzN@cw5
z<aD(TTFna=iB^0ywv@UllxpSTEwN5~4fg2Yci9y7DvA(OlG|E;Q@eI|U1edI^t7+l
z8(l~e8`@R~Tl6vJQLhK?)WtHH7~ApFBzu6;bd-{qYX6sH0h%tTI~I|FcyFV!b1*oM
zl*eD70+U0+w?5{Cv*yH;{xIj<wh^0PbTCjV{t_n)HqeM)YJDuAYB0si6>~aILzh87
zAhk6vs4ibc)~6p;zuv0eDub^bCgxX@l`)UmM_##|PltPH-p<6zMpvq(>ag8)AOwK$
z!?&Kl4edAcB6Uuu?h);ek*XnG@fTQ3bQg{2G&Ov+p4|SU;mh|5(thLU+D-2)Zc$C)
zR*h)z(1=up0IygCRW+ARZL-St=>wDF9TyXe>#AcH4v+Rbwr7vO*?n~*5YEvB3X2HQ
z1<1#8)!pXY8$~0hHo=4C-G#4_r#65?F{5QZR~X*pG$i1gd+67@K=fW1BL~#&Q<Uu=
zhpev^{6nRnosmIQZ>_Zi(Px1!OtWn=kO)8ANTxDPNGx!6=~s5?2PxC5B8X*^VoRl1
zaMewoS(TlrQIlb$gA>E=E6F>MXhNZZN2K_ea`)*btdpe*-ol8O@_8Niv#3y}hY_u)
z`p3p<OAxTU9WH_uk2IcB1B#o-jRSf5s0koMw4g+!Yn@@KWV1cC0Ox51M36LqysJ)x
zp+{Go9mqKLXNN<MvHELf{k5VX%=V3oi>vcPjsy-7#d~etqhrG>e?XGdPH^t+Q*29g
zpTG8|u%taDLb|pA!&_hCl1R@sbJN-y|ME)GhT^b7TS5-?pa@}G7ON5Z?}i9&-pYkp
zROKG)oyL9)X&sgXY%Kh51x9v&*pPoac`T^y77<Lz+Z7Q<O;jycLL#cYxPl6S457%k
zeC+PH^`!wr^2+t;DK^%eNg7k~h({*Qh4~%eh?bq4l+<-MwOQ+%B?`B<|Hrwh#4x?s
z)Csws0KxqyGd>+m^cCV#QO)Y#mxo1%{r9_rld`BPnZ`xpq6&^{*)@Q*_E0eA$lvd`
zCtE*nL%Q~sRcth~($()a-SJFYrZy>huCE}jaN=C9-lZ5B1t03Z%HDFb^&SKh$Ey_f
z?@w=4o_1L+NuQC89Q@239t5U0Uvbz;CN|$bwy?{-D9bivK0oyH?qAL}6h4{`iLm~G
z_Vb?)W?%HryAO+3`+(a!hpx{DiMg)Jp0_{7(^TO#{OcJ(9hm6Zcf-5_pA&>GfO;Dr
z&Skg1sHx1)9Ouq7S%PhG9|uX=!zO#1X5>i>ltl1B{zg(pQY!Q(jffl)om_2L_TC3X
zh9Zx)PT+6zXjkm5Bd|sH?*XqvXz50yIth9W&V0<!Z$6OB8UPb2S9-t#cCaB=yKNu}
zGt<|1M(gv~tyGVAUq0dFQ9bYY3Mj|BoW(hj<1`63f19?epZoO)o}z2M%d7mnDbrHe
z=2_sMBSp|_gPZ&lK@d$jwZn}@iTT6B!viINrBy`M=8Lydw|U~X8m#e#&j3czw4Zud
zi7c!&r55s*@qkG@K0x08I7vVssY|{&;yIe1^eL7<zg9*Q1Y9bRyhQ9A{7+Q~<l}tR
zmVoc)0_MSk0(Q^C=pv67!0jPmX>vz~Run~W+;f~IgD0lO<Gl`u-(MmO<9IGq#FTNx
zj*=R`UgnG!`q&%;t8oH0*s_d}Z9YjP7aBWn0D!4qvNs-w!D1f{cZqED;m+OVU~awc
zWfTIjyoOq;oux>g?yYNUS0X5h5G31ie1W~JliT!q!KM2ZG_~7+l^2&F{2!HfpjK0X
ztis+CA_F$5a*AO{f;cU|juiK*o(ny8^N<>aoC^%d_|L}-7VZ$T-<Fm%miBj@#l2d0
z0k7{i4oZ0@?bfQHvsXB;&!4;JySEy*Z-WOb!0oZ6*8_RnzQzCH@g?iK-*Wl&YA3))
zb?X!8nFs{;r?W1<^nEcvTNfJZZP&p6)Utl^rM%{SJ`7J2VGdlaU^Ahqd;Q|Pgq~*Z
zRW07jf7R^B{V7-_d=p!=7xE?E9<Q05Mu-531NigW<@0ki`s)t+1v?Ym$GXAuFg4_t
z-Qz)=&O7$=bSM00N<MC-Fw~IeI{6Mvhb#KrRWtI_>q~mPi2E!N(Ou3RSUvBJt+Qx^
z6I<zf+>oQB?rRJ$kc25eYb2gEyYfL!);{++IAcQ_qUA1;ql50q`H;8K;1wQe$fQO-
z1kf>tLd}ecD$M_aq8IeG*KV=*B*?+-9+A^`bN2~VE#0HeVR$XX7KmI={Tlt{*Dvru
ziz`YJc1v1V3ssZZLY{S`=HRzp=@j#S@c0{~dkF8uid^E$>R>3EQXw?xY~fQVAz-98
z0#aOHQ9z*f50v-x`-jQr$OFigce4tp3YX9Ri^_NU{J*Gt6B*q9h06OKN$6Lr2bVSq
zt2#;n1FFBs;TW?-^Ax)d69ZYP6O|Y~jAGWI9OVCt0IsSo^!^f-jgu`mY8&WaOFuJL
z==TdVNb?vzjQ?os1MuOSptP~+Mv(j$|I_mk8C2ug>g3EdfRP~SL`Bp=%30pByx*;o
z4X2D3bUu%W-;7^niIO4{>(=YYqKpr}7I&TBCnrc4o?09z(*aqDrQKB{o-<X6_6F~z
z2YnauQ?%2_H|N*q9<P9MKB1P<Sd_CaSDnyYfNzgb(FG{B%cYJL(b0FhF5(xXllV(g
zCCgF)vBF~^QEh1PK+=;kNy$2+|KfzyE+hHEz!&Aia3DzNRqT9*q&Sw)nY`z7yrk+(
zPSAw1;|Y6~$P)x!X#7$^x!7M4V@-bs6Lt}fiJui$H?8<8jfYzubDtu|k;dG8D)I|7
z0br=rcGZj!hGufU+9Xf%=UcbgG<I@3-nynhgu;e`_@Axpn`pmZQnlTjBZvezDjL<=
zu!3hw!zIS@yS5{<cgPdgr=>D{Vc!S|pt_`F7}c5N0&*CoRCdXr0fpN6yq$RYousx>
zP{q77=28eSeMh-EM9Ui0uquQP`LBvYLI4nIUJQar{mM0Dhmd47!cbW1Fwnb;TxQVB
zmsd-<PX=q&7w23yZenS@*+B6S@kNveW<OfMD$_loH*yz~=G7lL3>UZgg9<Uo#!@h7
zmeCf`XN-&^EqaPcn%P-d^#IVprR+)}w+!>f#%AK^3~8cYU^@9SBM>@@0v}EBqk#M1
z(<pcNX(T*y(si$>%P?Qi9R#FsT1ABMQr=k#@gmQ|n3?zdG|g>URhfm^cM$Qg!lD<@
zNo3v(e?v+pdJu1-nYVbRqkrdo0h0ko6-U($kye{1|54HBV!^NBuJ6~(v;Pn@zI)+~
zuwu**G`m1A7CmeJwLffWzVB-i;FZwW{J2Z61ONAuy~lJDZd0@~#;r2`4GBw1_UA%;
z)p^@oJ;c8ODWU5N#W^;s=>FR%tZ|y-=Gg5WV~MA(2t;TL^$?l^tJNo4K=f^~L{Rbm
zwXzFrm5|fNm(l6G*Kg7j72#(r-e7gharU**(&sBUKu;*1OT6_~y$Ud-c_NRM97Vz!
z1>0Z4OA!oF+4gL!=Uxn9VBTW+_oM}-g@D2XJ~!EkrBTec{!9+4j{Kx~fTD7rL`V;j
zmxe8leu6Yv89RtWvI*)IEYK7Bzib?ehq`FcUtYx&`2-YC((8%5+=DZ~9G_1h4;^ZL
zb=7=?m~>~Wx0eLq;k$9EN$F6tlV6Ui{z5KO|Ea&4RN7DYF`DAFSA9zgMr5%6+~aw3
zS7V3X_`=qmOgmXT`TKZox+a4#kQ~gITqINr^}4$oxD#$oR7O2}8%n$XrQW=_|1odL
zb3=)!`&`M|+Q-ye?Uxxzjt<^?FCq1NLtCDgT!0*e5^?qb-&wIgC0xbs6UK8e2@Xq_
zrLg3rzNj1QJW>+-4K5rvCs;zwa9$#BZlg<dpg?iYzQ)1aNpfv>sg@0DEJ|g7F0Zs@
zIeh?B2_7wG)9UyW@2%J4Zex&LqiUJ{j&(^@B}pATs4P&lnZ8u)pkJ1rAP${rYSh5`
z0^q=ajZ`fqEtmrJ0r80s;TP(Xf<hGOs(|>W=#UuJtmYzy!BdddI?0wxy0AcAR{?3F
zrLUws?szTPF^@(xeCUy4A@LK?kUvVQ!jo!M#OmzY4cHWn-yj(CJqch%^^tyZv-ZCq
z@?_trCrQ5M)bbMcS%9u@;9*%wS|i&z0C<ruQ3f^e6F~9jgEXvyX4-6OACBPT@YNnJ
z<|7t0<pinMF|EIz@mu+L)KM#ZGxEsIW#5j&SGC3B4}pgMA6h_D67yODiB=V%Vi$JT
zNxi*K1H5&s+`|HSbm2fLwR6f#T(TNU-Pg8IlW_RcI=aZR#w7lsp{wU{7huce2|<WC
zNVaw9>ht!!<H%4~`5OJ-WUaq9D3J)3w~o(;PDfZMlM!s!a5GbPlr*%e@QtjSMM#E$
zLjEh?&bSDdKaSy#oBK$>mfsldfs<!Ma-g0KLf9)iAKu**hiyQI3$XT3tJN(*!252v
zK9?&RQ@0xh>-0=E4{!mfvTc-d5X9NU6&jRm;4hth>H7J7eSRnCxX)Xu+QKfa6Y)+?
zBuw?GB|}0*^^EU!IIL)sjOUxa{KKZJ>qRO^E$8%13aoSQL1ZYT#+7ft^R!f7prdQ?
z;a}bL&;?rH*`^p=ea)czRsZvEQU&JXgXivr`-Y3j_Dbw{@!>q+;mEHb(Vx9~nVo)<
zVFBJ9PKaQXs}t}S`F##!Ekdq=KrB2`A{>^+T@{~{^x7YzW57Zpl0hvJ5xlY;R8GF3
zh|`sEa8o$4D*^aBLXk11GNI5vB4aS{Xqv}1d#Y8=xAjjh<)0L{5&Le4SW^WPWzeT*
z$iPUG0<$}BpY;KpFbcU~r2KFCx6!F)`_@M2b^4Msi|(^$c>yjdgQ(oMF8fDkw<s?>
zifO2B`$r?wF8p8;euPZiK(;|QGkJX4!1*!my^~KllkSm^OfW1eRVaE%Kq&9vh+A;F
z|BX~oLqH{IikIHPTE?fNBw@+ZBXq2P^;+g+lGdo0xdTurO9ttQ8T5qrBk35NQ~Hnh
zevVB_=pXA@=5Dtr&?D@GOwnW-SB$ifHI<uPRv%2(!sh6K<i=eJV;N1P`^@~*5beM1
zd}4L5|JFy2AxWLO2kM|EvO}V|prnh27Fpgpp&t@m&Ok;3Y!pIzhz5rr>p+P?LKm+{
z{404U0Fa<aTp%P;U=aLVhz>X0+2yl{3i)kiFNFHvsY+#yIi~}5EwoZal`##qehYnw
z%k;(eZ=x&?l|)=RD1c*#7_{cS7Dt1T$a?^{aH(X^gRpls^(SZ64+k;{ZKgD75R~{i
z40GkAsghuekBDbP$Ws&n+(Z-o%Ve;s0#fm4Qv%z!^oiPSAGnOuw8-9AX%=Zm43M;N
zcn76fp8pS9-vC@$*K{4*wrx*rJDE)Ei8V1Nxv_2AwllG9CllM|{CA%BeLwv5)vnqn
zmAdy<*Xnb6_gdX&Z#Uo(hD9m(Opw|DV5!c^5%=BW{NpzG`W9Ing-kOOM!Ol3H4rD6
zF?bLH<y-D@BUE#fbhTM(|3c3~qKTF3?+X%t(+CoOrcfiu0edU&re}Za5kpE7We10O
zfkVAQ13lHa-A=*5*&}XUzidG#wTd`m>6Mujrq`oQ*BUSFz+-dq&F}L@#CZyUCm)8p
z(=5-dcN_Uj0O_5w0Bvv*o$_J<U*%F=>MF4?_KGYP&swD_9me%ng>vF}<W>oUz4Yqg
zPm#W#ZTLhULbjgRQmUhoBt=ms$qFtg-y+Lg&Ndf|lV6?^m!O)YOi;DMgt>`&9@Cc;
zoJ_QbT1B!+All(YvY;DUwwMb6O%A!vo@=KB59|(75?fev?8@c-IJ>#}QuFK;#<0)b
zp?^NzT^~+r#))2lX)0N$!hK@Y^8hWE;cU9f_x^|KGR{IILLU}8%jvu!jT?zyD|mDT
z-W1CufZ_(q(1lwgIa4_nlZ;ey<Lfne)+!5u;E#{ET54~yAL1DNhO&o%bn*nGupLPu
zZKsPFiN(={qJD*%YEu0znC$^~EtfK^6cZ{gubdTf2N|=v!jZw$Zo7s*-0ZkSIN76u
zkH)k~jvE17U|L*@x~cbYGq|;l;@#SiR0DzdB5cqqzBCr(pf@)M*th-R0~ko4>LBXE
zn7iop@+$rE*zjtPD&eYt8H^QEj?g6Wix|q-_Sphv<kIVD?R!5uZ3OR#F)6c+gHD7{
zRgItO9QWnKyC?fYsB<;maQGNL_i;!#U^rm=kLl7N%!eX7dDqJB0u&&Fo#J)EtV%)?
zDih-wY8ZFQ6BQV4Jj~$vqDwmq+C179Wg*ho-tw`KCIZM7CjN4qzPP@)!Z+kv49f2_
z$W9nSw^W#e$QVL&gL+frzBDd_&_`)&ns5X6VMdJFX@;<du+HEFT`69ykhlP%Q;(!w
zrXy2fwGrM-U3w|&T05rAuDmqZdf!SZa8zM-bX~9s-``*pAmPL`KDp0+yHcQq`F}4(
zt>YfJC`HxJzlFcz%lV}WvQ|p4qceaq_}Tduq`~4LM}-U(WT((!l^*3~m*mn43osj~
zs!Y?DW@9^S?L=q>v=WW;0REE>hu!qJia^GUvH7wINV`Rq!w-dVth4@8RYxMnxbfa5
z6-3FI92_QC96=s%&RBBhufD0YE7AXB-tSH~;#-fx;T(AutHO4I|MhICqXN=;5}LE?
zkvqSX<TT<{_*a3zhEvc!l!t$_*!m(R_L^YH#RKqp`3X4t5Y@(qNUaH{M0R_Ed_s|<
zW3aAe(AL5lKy2rU&+S0*tVX{!0BKRQ|8q>N_r}Oo3TsIRIfABw{`aTyE@b-%%V0UB
zG22_<Dp4Req@L7w;#(97E$2l>*?#y{`qM7Z*Eq1^YLa(K(8pccTMObK^CD+0YaccF
z5rLzi0(N^h`hmmR=JMlf2WKJJCM_MVcc$TQ<DXlFcK64vI0o0basm-cr{T!KdY1b`
zaUG;?^Kh4ib;URqaPS}z$el(vivjon)_Etj=RNqzgvo?QEZiX~qsMy$P)95t3E@sS
znf+?v9hoVhDx_yKTEzraj|*b50g00At(@He0g$4Qxq|3C2SV^=Xq5fltqM81bAfY#
z)1^|K0=uYom#M=1znA~=cJ7fh(A)b8>^*UvJ7A#lg8v$)F6Ck|_z)}?Ai8?0Mr#fj
z0R=b&gTHB@q*!?p!vFsL-!|w0ZTP1uR%s9TA#E8uN0TvU;T{Aci(Di{D7ME7RgFS2
z640fQqPF6-;<Zqv(jvKu=W>=QBk}hl>yUEC2BaJ}BJ*)v$zY=yX1?Se{g_~>KmRor
z&eTGPG++$ZKwH}j`x^KDn!p|2tAY@1>_~+v#R^F{D8AwPoX>E?4%z^itk<0SPwSLo
zBy=-8osoJnXyS=TyS17giz2Z9Yqev52=!436_v5I_^b{P%2G3`k5aRKE`)>|iMY%u
zb=Y$)|5%5OU!$>FmKD&Sf360R?Rb%x4&m7LqF?&l$j5l&P*pQcOMKj|TTB)ir!Yf@
zuXI3Xff-%mE)1wU7S-FG@%xD4RhrcVpMyOSZ#39d3CT8&jgsL#KPS3ZDq@#GwLw`q
zQga~s?|VONFWB=VN3`IhFG0f6gq$xDvUc(fUtgDcc-Ul|^9Q3-V-x7YFNM_$g2r{e
z)!zor^B&@h+Ew~zok=0Hp{APACWG4Xk)wNn`r8$RLwqjpvhb7UOzp$%uiE$vbAsz6
zF_S${!<{-c)--7n!lglh>FVvi#XtSOXcv9>a;GRIY#j5<qGd!>s2<^^F`?1&2#mv~
zgqMbaLzKU;%JN&!aim8Qc;%=17UGT~3e*%rZ=zz!dD&-i0_*p;_x{(+7Y&G;)Z_es
zBF)V5S}Jl?)QOPuQRdW?Otd`I6XP0r|9E>_hi`NBBZMT?47&~i%OEZ7a7^KFAiDaZ
zD#&!=?kdP*iDQAD%AjgVNlxD&7wM9I>Fd74OP*E)xT)Jwb)*v>27VJ9WSU}PQsRlG
z8~dlbaZ*O5yvNrAT{E(+0&J1K8Vf)l!^@O8EphD=ck;KZh53;(?FpC8@w1@%kiXZc
z$LKhOmq0QZ2^<kT0*QPqM@;hCN6N_Mc77#|cG2Qm>X9x8>pfhMX7~n}l_m~w?&?hB
z*=$8JCygdDHBUjj*X?*NFW{;E58=yL6j|Rhv}ya~9_cCK8UU*mByTz>3yE{u=7j{J
zDoc{mfdWqgBcZ>4KC<dEy2?%lF3?EoWlwtfkZb^&?_U8~QJKL;XhxfY>sRf6{EDpt
zS`{81Ava)wmy{%{OQvWN6`T@U^lj&1Blag@etUE9sYiPcpM$ce`Yr;fuq<4mqB+d*
zTR*OUaOoZbD8L+MF^u4@TJyxoy2A!O`t+23RACJy2A6I|-J+s#E&Sz+^0(4*f@sW!
z2onu;0x@jT&Y;<VRf!kXZ|WyabfxMi1B49`g?S~psVp&4QWd_cRbZ4jEPvM>PS}0N
z0C-7mf&~;81jw=;2=HgLrbM_Xs%uKTd?W*Z(61`ufO;1kl~3TVPHI-RNqxR$f1zac
z=usniA5%)1qavA>HcqYlS2i*yR-NT~R=TdbA~k{NAbS*eyT!(WnGW3Uk!kXGGy^5D
z7r+<Q7gae%7sro>yw+Z>80R-TS)roOuLmE}D*u8HU{R!%P!0HexiZ83UGUW8)Z74w
zH<Up|XmWiD5i~W4Ei~7}ym=px_Gb`JG3FXMpY7)_Rad3NBrc&t@wJep#62*7hlM4?
z=>t^R#9$d+QTRW32}(9ada|U3+fsd4`WWz4GpRnJqp(iYqKe)Xdp#$UIZ{^L5g3!e
zQR&I1evS&oVkSa$!)TQ_%4q<MAgW0%{7#W)aNOx_LWCYndLD4Xp{pNBx_s%KRk=w@
zBnUJ-fnh9VKFhyOWg{aM&~#M+EBiJ!R(mc=ggFEOj{&7WXgf9f`{X+=HIE0aB$F9!
zzvxifGpvxc%qJ*B8;m3BzIqxFL>pynI7!PuEZv|1Y2z4QV|*qW6MzWEfss|ARPwtU
zL?h-JT~o8M8-yBWIbFzi2}lG?7PX#odE{i7eYtNugG>wwJ#4UTN<IfcU{W_H54+%!
z9vboy;$7U)>zN6?^Kyz|ik7zi!&-}a%8_Z{yO_cRF7#a(1Wm{bg352DYyzd;`*N68
z#J~&l!|9@6{n<=Bt5sk3Mw^qv<Aqn9vF{?$i$JMwLe1lhiN~-c!&{R^cDVU<4}nIE
z;^oC*4@}6>7JkYq4<rGgi@Zd`{KI*>dESoj+}Y{YmWPFmw5GL{zpMS-{pRNDpWAz(
zK?t%z)RvB>6ON1kMf2`1_^1?>J-z8)H9=ONArB!go!3I2c=NAmriOSqx2#SBUE-7f
zN-P9$qQ9AM8(szefnq1yY-os3!q!LuXm-2RFf%g&yg$kQ07RV(l^CaqNw&=lK3fL?
zet#x9ejA_)jSsIQM&7i_M2~treGGrjwZ4}oK6YI6cbyI`pOv3*Z5Nwri-TxTXDisz
zu-XkJ{+Y!=3X|(r2{DyJKW!gkbXk4d0x!aPBLZV90;@-yd@wS9$_1pu4qYh9{UpHA
zK-e>!E3?5#0q7;CCq!xfJr66tL5qo>e%u#*O#Ih-+H4Xgq#(mGC6Jsmw7Z9o4ZfOL
z)&PTzX##DcS@JCh`@f6$T)?7q1P%huc=LPrb=_EAt_=;fRF^AXLC|XWh>RdIfya5I
zcppL4%D$&hBcQz{B_Z}mWOnbX)9;n)_4hFVL<LSvDmX5zYu1(7Cm=#6ErkUhgN@iY
z!Wv<P%5A|Y?Us=8DS9&g5SCRhn;V&SJe64fN->qi$pi^u{LXtTcB^5<Hx0@74s~Z^
zL4b3#q@zrZNfXPCJNu8sRW!B~SpN~=8$obXj#)A5yI%4y=*^bhqstN^Kp`w(Q|K#z
zyeWh)x$e(jzEk#PAjE&q!2t^IXJ`WHXV%d+{Wddp_G60s*tck6<_xRP<}*F^BExo$
z{}veT-a3!4R3YbRqr=zbGOXgatCkJwq~}8NTb8$0r_Ii#;s#Il#dAx;tU&j|lXnQx
zx8B!WEF3-(s^kONH_b1B<v}8%$MEa4qQ*ov^Rd(sf%QIBh%SsK7MUeUm4IV28KtAO
zsN_YCn@LE<AbwFI1M-+V=9>((xR_A&*U*-*N1Yn-pRo-t2Mij2v<ou_-!!%pUo6R6
zEc`&iQwFoQT}?i$JsI*XSTu}5)M;iTBHovtj6sq6u#NN#^$sfm5vDq|EH_wYaoUhD
zcLf+L6W_Sd7iXeUvn^yXYl4|vPZA9o;A|3Bj?sLXp*udnAFa4Ni@pIFm+q`;()`ES
zLM?fK=k!?Ba$9`^T9B5LID?n#(8x3l{=Ts*mu2Me?m0`Tb||^H&%lTh%b#(|V(4zz
zZmrI3U8$KuEA=TrE#SMN<i5KhC^Iq^%&TAF88n(Vb%cP(!GH(ktDng<qEKuNu_0Ao
zmZECR?gyIMsl(5oCDN>PT1b0K1OW7kEUQx#>Yq9g^__Lojr^yrYPuR*Gs!y`nonPU
zyHiG-(bBD7?#I?ZpSZV6sFd929Qyt0oDyJMq)=Ooe1-c1C~()&eQ`T<<l*rQB2U^j
z$0GWJtzby2-*94Ji9_W+t^8B%jup6<XayaMLw_hK4oW0CM;9%`O`qnji$9#n5i<hq
zf_l{46@u{n5~e_-ZGkOcl0_p)zO5Ig+e0&x#kyi63mnw2#rok@?gzu4`tJ;Xk4}@r
zS!-UNqi?hY{3sQ+SeIOXgr`flSP$M#ocU2rZeg*`MAke4w}pn)m1z2f%NQ4{i{jOS
z!Lv-z!7EAg9xRGkeXs7A65WE#L$un8ZZaPm!a@St!dJu8eNrXrU08H{OpcQ8A?j+^
zlv}2|vBMn?UbMZWe{xlIq=&*=evUamwftx^1o+qkWLymQ6OIm249Vn-17Muy7)x&R
zV^@xmS!dJ{%9}vHyVX$oz~Hm&^L86JflIEsQk6^vVM0j|rqZ7YgV+=^k!IPPXU+Bs
zU{L|qT8A;+MhwL;ouYfbd2GWl$6}T|+=C@;Jr=fS(vqlJ;}{PzGjOFaAsnClSZvFE
zf&`%hl$PMg=0<#~8`FRqW3LRH4l~tZ6YYGWbc6JipU_EWmr_jQc=T<HwCvCVkT-)@
z!|)`tD~mL6qxn_3Ov^1m$2LBr5tloPmtuLHpL<)21VZo(2C<jJD-v;U{RyP;OO=cd
zA3#*w!X&%`pjwS(lhMA`$66W5sz>IYuqzb-CTz&)-Rm}5`%FfFGupiJ4m6y;ovy!X
zk~puO6Vt#|gtaIs+*&x;Qqp(5w%G9UYWiVLsGzZDO^s@j1*fwy%mLMHIUA>pL%#~q
zR&~Z#=vG|+4rc~(A-|hEac_y9>gVUM3;*3>GFkzTZ-=CScolcpMhh@G8TRzg+)03u
zXgg|njFmp9Q`3U$eQHI~q%PwNZRD5vpNWtYtc_TQI6Zv?*wRLQusFd?G=+5;TL=BK
zS(j37sbXcO_h7oLA!twrH=eYT&W0GdRxIBWqvJipRealznI%CF2lWP+!Uvgf0`hV{
z#NPd%Aqzk%6mTR#Z{1jOb`!fpyhZ_k0R^ot(!hZXm8VuD*{8-zLxTDQRm~&^Uq0Op
z_XFTOhXj_tsxecmox$KiePZrp)%&|OFsRCQ)S-_1T*w4F+yl2T^1*a5478l?+VO1s
zwti}0V4#-}@vkNT%F9tM#_yu|CTD~^UA$){Z^yvo%dkWz@M1g|6&6tayn73*^6w1}
zD}9X5|4BqKKa}10Tq^D;K<3T*Xx@+Wug1IbeDr2iyA!Gussxr*Z<bS+rWwHJpA%!!
z$S*0a{%L}CxahTK!0P=`=hg=Zz>T|HT^$6rA&d)EZg(uFV5f8o0^$I9JhdgsM=>GR
z4a|Uz-R>U1q24v1CN$@utk)N^TftF&s^d!NG39bxyr^;8miY3)sa_<GpHzc$^H5Az
zAp++2=|ICMBIVFz>a7r7KdA%lI9w!an({KIUD^276A?|*$`eX1EnzMu+>gJiJ#=SM
z&~l4cNJmw{hYgZ__`v%ERui0u$C4VwYmn;D?Y<0fR}etT<%}u#Wnc~+s(yNd&0+V8
z!87Pr<zX+nCC?VOC&R9Y>g;P5Sk7B;*#14|7s0D$ug&?)eLiG7;tZrW7~A%zy`b_2
zHRZ)K@pI|aUW{)FFAl1Z#R@vwu4`2Vc5;`YrLxrcpum{JOS#Kd<u?iAAn=I%dSsW!
z93#mzCI>KsptYrFuqkf~tn)RNuhO}yeq3PaO-zyC)aazT3b&AIhn)LpyzPsvRV~1{
z`>Oqbd|2|kipzErA?8rsteRfE=a0nu7snHrU(#L%jiVAofgQ|`Urym-sY<j^ayHNk
z+C(cXa;wFAVH{ZJ98Jq4oqv*%iFlK2i%jUuzQTi5>!uiqK{7L;&0Sr!cv|?1XBn}J
z&?kS^s{o|!Kn~p0W+2L&T-7ysBf{wDdwdhh9v16~$5MWY0lT)Yeu#g2eNMG`VOMS8
z5<H)nDIVeEESATQi{t07$xQhp21y`W`3*9@o@|xIo5sa~b?CNCT8+$&?=35NtbsEC
zL8@gQY*Ix@fPvQcJaUA_9harvkAOIFk2}M(J`r70UjNH9e4#ApJ>o|I&s&PAIHWwb
zk=^Pp+tHTj(uoo7UcyhIzI0PSScFB&h&Uu2nU-b^+wF>dm$en*|Lnj<<qOteL+1;g
z-ylr^m4L(q<Y|lKt<fh>$^JCL!Kn$k(Q>QlASWI|W!EkWV45f!`TCu197Ir@s`SB(
zy;20SYvSv4;W~cT1#`rdos%;ObyZhE0sF~5Mc)zs0ma`sjXiAYHZLong0DUGbn)vc
zdt$~M6`H8s;r>tx>YuGG?e)$0AnOEjd`)u7BC9<>Bba|~wN846BTNlMoz_K?X3rbi
zbRvy`y*vs7Hx*2u|MxLFDT?=)J|nq}$VRv!Myp7XL|wa|xM5LRuHqB*3GoV2K^15s
z@Rcf<h2>x(@DO=ccC==`$em~dRIBuqX7e}Mv?{-Tr;=zcI>?U{n_>m_Z!XRLvgnSL
zMVYAv#FD^ZTHTE%kZd(KNXKvVWSMN3-qI=9U0u=|&K<ZL8g=#0>KpNJ=rvDwT<AZk
zkzL)dIIl=q%y$uI=(85nAhT&2oByLw2c4CgjUgwIN!$7)`a=5RI|#7)&o$NIRnYF;
z>)?+vWSm+w)kDbXqfN$~d}@BgF}?~jG5~DkTjCDUu21(u5I!P{<*qmrxD6mV+4((n
z(w<A*qL@^~fxVy$95A8cSgWe{Fg!M2G~|852?d~URBArs^dflb^OxbY8HAFO1=Y4W
z`gL1UHQZWRiFI_&>A#sRI-zHnp@n!8z)+&d(%{RA%O1@*>G=+)Q)XuD1R^l$+X6~=
zc2yW>aNqh`EKfWJa;rgXJ$uoYarNoH*1a!C%!C(p+bp1i-`n>6$lR3;Sz6OFI$@wH
zuLJ?7t4(dG|EuW4dw8Ju&tx_{+u)VgI#=c+;~AdEWVUysnxp*{ZgRAnfUi}$1#?Nn
z`I-IGi^S?GZr1*1^UB#X<PE^{vdteFS)(Fz_YQmG3&`Ydn&>WR<SGS~(TcM<cBWa7
zPZ`;q@+nDLHY}lIAsMlE=3|^8c=$8DwaijdWn%dUGwP~}wqq#b1?ous7Rzv?hzk3N
z)7|VT>MJdzNd_mDzOvo9PSc-{Z)g3I(lDTmWlB%%9qeI27svNbf*pWq58lUG3h+9f
zFo*g#%)L@52yd0(J$iPnQYsc(@6B!|l5b$88Lv|ID8p}-`w_-wpCD*qLTG?GY>=?-
z@KW*{sa++yo2QM$d#!G^kg#{8=!Fco?f3=`+n{PrifH|pmhYjpaL4DxX$|K>ecsX(
z6>1TtkmPCI&Q|V2$DV*Kg<^g9uH`tT!4GOW{iK<pfeKms@4GG%*vibg`odkg9FJ*A
z{=1F3@)9G-8(8><w7*Khsn=OFOyqvnP`<59O%L#Nvh**q5#<~P)m^4sD80_Hg%3+@
zR#c4BR|(BTD>khYAFiZNP5PG!;gzh-@>Mx`%x>b{pB|`ViNpY=MQx6GPmp<20*nCM
zr%1eilEgd{%I%{HHkG*boX?^ptJM4kzcxM>aRPWBC*OxUS+{FGLIfJBJ{Q?nu!!$$
z=XJ0s^=mvn%9tn6->A=(-U0otWK|InpQQkvR@Oh8?RfXl#M?fxAHqibe~h#*mt00a
zcXm+U5YLx8?>+&TN6XnON<*(U>N8u`T}vfKUrK+G5-->$r7zma3XXsu_`)Y7q@OUK
zM|xUWA=y)>b1whFPyh5btL;8b2y5o}<iL8B_u@#F`<T*P%h3@+>~91h!KSIfu4Z4R
z--2+kC<fn1RIJg>_jCW2l;E9A$h0>ygFd3~&%}ZMe!B`l!n+8{6XF(vk&HfuG<4?K
z(r2dD4wt>@JRA{QHlA<nv8r3LC|RT)*MTbr?Y6M-l%BUWoM=1o=|vjwmRMbGz^*tt
z!rs59sGYJr4QmnPC^W>oUQ)-&G(g^Jbi1}iAb6nqaw%~<8NSuZyQ@e;)tbVg*!IX%
z@AoBMwc=5fi8~kOHn=EhM^>D&bn>y(gu1gsyg2GLs0U(D)X<--5|YX%WsB@nI<r~4
zt`|17WklpzTK9)8Vc3M$nBtUbbXf|CBqRnvO(IlQIb%}A@;fIXVH(#dHW)gFTxL(8
z6<#>Wj}FoD@L0tTFl$THU#59#Wqp=CmL}yeAsOpKB7O*%ek~Y36#Jc!xe(G_esB|g
z=g(oLa-Fiol{<!W9$dsMCU4=Z2;kij^>@fWxu~=I#a+Ty1TshZQRNoC7|~uU!px8k
z(D1L3laK~kXN7KtD-;i7612&XV&kA-ORKyBEo>6VA368B@|-bBs0b1{i(xX`KGC{|
z+DRvy_|?DL5UrT%joqkODEJoL&i?)DRI)YPwHZRq#5$ab6D>xiD0U^<)s3MJs00*d
zWnI;8roixZxT6(y<(b{Osvbx@8~Q*60IrL9b8)0CWZ8~H3mBc2&F8r#-=X{E#&wV=
zRuDt$&H<B&oP=6ETsl9U7Sl{_JYoi$Y?h4zZc^pK^{#KGXjCA%mII_PWykbfoO5!i
z$!3speLQ1c{I{8Xl~G4Os3e<}M3qw$>#<UkqG0T#%Fr63!rcD0%AUsnCoQ!GK&csK
znb|mS=I~H!(9IR$#OE?i<XcT=E7t;=#2MkF<vZ|W9CWdA0g3&P$aA6XhPHUa#SYS?
z+J6=OR2OK{6woBNJfKM~)eF~-?TMJPX;_D!TarA2?Ri-|PN4Z*Ukb&5z~)Y&ubR{&
z=`~)8Cz3cDF^}rOP(A_T%gN3ud-ofVlvd>W^wHpy&M86l&*}Fmwf$_^g2KNYhkx5u
zAUdbd9m)0Qe8k)cQg$@R3-*SDR3LhT)6MPuB|4I*;%;U$wxSV>r+-hMn>ctm?NjsF
z|C8lN=x!fV*f>I)D+LHja#ca{Ur@$@pfuS4zjU6P4A}{I|2>Gy6GaDmRKE-7tL>Uo
zM1I1gGTsiKXTrEH%vg~-T9AAIhG?}KOIxra!U*sF(hGzHO#*GoRH88}b9F0jDL;PF
zlr*u@B%{>iFNpO(5GnE5OcQxaO+cB2+#TrT@+AAxDqF#F<2TJuq<Q7k>MuoUv_`WF
zfKCGZ$4fwiPIKcSJKujxP}1*K{AdfpmxjGNO-=bWWtyMuf-O4e*{mxrP?0?<MK5M1
z!Sszhykj3Vh>tiVu#Zp>KB&Y0I^VuT1u5UEq`up~PtP$WtZx&DF#Ho5oEs1<=vkm$
z0zkVk_)E?I+NCnROM)j2AT$)hL6PK)uu_k+YL2v07fF!f9EZeHYasn9*%n}hb$uIv
zy~q%R2OGpNWwy#JVg>wUED+B0s@(Lya83f@T>J~?Wp4UE)^h@_FIY|uzWHs4Rm32x
zB_X>?C#watpojt6LU}A$9=8ceR;y<mSwFF!I~4#ao%x8bJSKa|s515!by^)U+P|nL
z0#TR!i~17kufzw0f^}|YmuEx$ay365&<X1-$?2eQjKj{&bGXB}GcmF7sNUU_$W{t=
z^pV;JCc&|<;Fke}mCJUnvUY<9`b`>#7*(c?yXqb4)%S*jFP2a9Hg*?f8`z7EfHIfr
zTt@?BblR4G$H0hYmLaFZtMea?mlx#+tJR7t)y$Y5NTZE;l5ZY0I+bwqsjt>|L1vdm
zkf-}Q_?X#4@a(!JEj3<5Q>>CpItPv6wy+@PzpOWHUxeE~+ZKRlz6~b6JHe6JUYlOZ
z^jsj{t#59&9B8&MeBU{C2grEUOcu+Hf9&q7c|8*Ir_?(fTmH5hzu6tXL5mOI>XYZj
z7v7t~wzvN{K{qO;5F3Uv7i>s?L?)Qt&%RGHTHkPZ*_V5%K>VG>3b^k+2=%_XuoUEE
zY!<4X&c1Z{(_ox8WE0QS2|4;ZO99tzIw<HPO483XxVZw9UsOs68_<vfVg~OTJ6O5!
z0iX4zjW%|^z_1mkt;xMB*~(H$3_p+c!`X#Ukfo4(xF+xY4uSGu2=lwOSasjCfh$Qg
zovKx&iUd@}$k!HF4o&$rgIa2O)3gsrFSK$~MNT!2zB5sOw&4>tT`%o$dLP9m#-f?B
z!zHmBcwHt9FHVQ`2tYsFoo+3=8YRJq=K%$7EHfNvn4>IKsET4<W<J=>uPjFkG0kCq
zM{f)1x>>pTlR6oinMXw)#}D#jA`zRl_DYKH^Y@P{3g}(pR!@3D%Y|N9+D*f>>ZYxj
zsaln{OCiA+=X*sirhJayjpPN-+$6n>j*+L3>&<w%&I-J|3;``p(>fE8d~eW?YB|PD
zE2Xnvb1?Lz_g`mX{Vjd@8o#u;n7MQbtlZqhH7t%jPGojtY^}6xXLl(czr~Ez0K%R>
z9gR#i-!FF7nOEKFG4N=MB%^Jon<KxC%@}OmGw(8YRA6(Aeh{O{ewS+5vrI$mO16QS
z;5+-|25}!@zyll=-Pr|{7wD%%MdOgb#BHEmh)TR=9J;k{<6<GGNVEQuA2-;GqfM?z
z(k~yeM>|kspAc$#TDu)wYF`gR9KRVYBZR+<8oYln45}XwoK9RfUn{HDW9e0aJEPew
zF7Q9oL7D$MkRmNh*1+L;uDq8@mmPDA+#S<xJFV3LR%inqxy5c7MG>w+vB=um=Nopb
z+K=~joCXYC$*6t2*;Q!dmg89ZGos-71htn0DWQb5`M9BU_QdYlKK(A}j+^uAdy4|c
zj8>a;O*&bL3%rOY@1!9I^EP-kuVj#OqhXJ3A2%moY%$AcYWW;0P5i8`TpO%v)~u7-
zT4y%^a0G5xlRCNjd!71K4LOaMQBno{f?j~anBic{#SczF_r)(}N?V>|v`t4HS6Hv}
z-bPkfb|vfqoc!$zn`EVi`uGwr<++_}LuS%PO2N3qu7zLPmBT4&N|h2w5~>{ye`K~r
zwvvE42%7RQMBS}kW*a;4ug8<FjkZd8D|#LQI3MgB29vB>mZth(o{P|V#8nlUyQ5qW
zws?Y)5$e8PC`nbi@TL`*FY$g(xQl`nvPm(YzrcGD<tCHDdo>mj)%Cv=*y=*;&pZw~
zeimAn$-ixQ+qen$nRm2VL}~Y59@DG{h!8a+;?f(v`*e4n{t>EjTYN3>DuhrEq}34w
z5PVd4ajL>FV6E{#GPdUqUI#Nqmh8P9<P1MG`|j--=M98*%SOILJgnDIN3E5q8@oZX
zLA5>naC7fFrQi1ce4W&3U4@~NS|3?BrW;AH+Z<2c1i@r`ae<i5EhVGu^rwqgWW_BZ
z<B@A9#sY2@Df$9Jh^kptdvFQoi4@BKAjq&#-UE#fh)Cp?mYs{98@7S(xA1LJW&0A=
z7Fbv@xHRN7ieZq7UfJ9G=AL!IVOykd2B(`X+9oB5+fJ-_bwS|x@%q&uH4XWP`DQ_3
zdKRlP{N4-l{4Elz2GP41{NaL3&*QEConym%5EROcZ)?A8F?_Z*f1O&s`%mYe0O$66
zvJxrk^+SfO9?W$VPhfGUa(mg~_HwB3sH)}s?5l`LHyv7=uvwchKZ-sKib1Y-2E;To
zT~)Feek2Yx@6e7gSC?r^RM(c##eQ=MkvI6;?!q6*r-%alyCHt<kKlwR@cRAm`RWv}
zZ;GR+W3%V_seVzCN@Y8nrptaC0L4|_7l+_B=qmxCNTrvvfGM9K>Swye{JRL2Uh1%B
zBmZeFDEaF$@aAXqpP_fDzxVW3Mtf(a_kzCg;>ur#z2C|zUB?C4c7Mgc^`gz7Vme9B
zVT`pQY|0oipbfYuBb82-7-Dc#G<HWbq{3w%ZLEzqrCR0cG5WKYyFJgu1dtMor1{#*
zl7ecgqxEb0{-%G;J9Ym7qG2rc8x$RMfhZOtCzlErhBW<@Vo{qTLP`XkKf7-AhnUkJ
z=Z1No5>6Z*TDnXs%aX7%Cs5(O1s<6R7MIIWr(jpVsQ=--R9@`kuDcB0>CPH;PCj&>
zQv$r$YAD0g4FD&Ed{%<8ynGgX0{j5DUzyVt`T#d;JppaWluESOQy+qjeAYN3vlCpZ
z#bYu|f?}%0kaz>bT5y<|IWEg=CDKZ&g=8jn4-)I_&)Vq5iUUc;#Z8J|0_KX1T^(;`
z{DNA{+nEf^A+7yV$rj9>Hn`JR*3|gARyjg?(RLHAEr9uUBG(`CZ(-|sgYETp6XR#I
zwSd`0$!@pJS^vC-xLX^{<oRw*DQ)yF*WEB*eLlQss1~MkRNZ5X!^^e+YyTq8CWqPL
zC=U|71}9(Rs^c08CDTw{?%iG1EZAn{Hhx&1=eC7Gj$tQuL88%svVQxzr$-W{5)$zP
zxRTnCuYlj3M-wV#11nC7N@Wx?IVK$=w<~g&>w~l-jV=b7c+YdIAd)!W_#>z8fvZR=
z`u|!*!c}TT%A{WpSl|z}g>KDmC332+)x@%>1Ly`=aoGfW6d*h!gB3xxnj2Gwi=e*p
zhTXi(Zhb;bvF#G<b|;`5k`(0tcFh<IaH*+56T9vc(YG<zgf(lNj~}74#(Qgp2wb}6
z0pEo6DFY|M)EzDiYyzJ#DK!t6A8N`=+}EY4aA|)BGjrWEjMfF>*0sr08`XSR^j0&$
zdXHLFs`AY{gEetO=v#slj}8@M?dChS6d`(3lx0}W)Ts6x{9fh}@xbu_WQhk!3D2-+
zk~wPYG{<kPn^*Cc<-|a)3202A0VZb%3mNxaw{TyD{lEX9=EHiE@GgQJ&hEnli56H!
z>8X<vW$mp?x$_G6K8=n-;Prb&jBp~EV6(B~l$^>ukN2&7ij>3GUSRADIhesojSlu6
zS}ghAv@e!2m;qIC)cjf*KnS)=nYXQTR54DvZy(RtX~0AuPE<gWi$+YBbEiU8Ino=?
z!U;d&M3bL8OTY6lQXL-`ui;;7_9bZur|OSY<0t*uSzl1X0J|qYjQu92g<8d6eg7XH
z_avGZo)`!!zle^3fvoMYalB<)4?wz_?zKbdu6T7xrmELkNJJ+D;H!v76C$<;&$w?y
z=K?4`wg>4HykSV|i1nQnB>bHXxI_7yIjQQByN2%BxAHD|F7;=kB+ol;9*A#>j)Av+
zBuod!MOtHlm%a)7m7zruL%a%gU*mIB<ESmcM6roexU}y)3YOb~bH;!C>_5}(XFg|-
z`Qe^r?Vgpu^=uwG4&ab}_c$d|6?gEBSHTz>#~*f;krd;))u@ga$d>*w`&gq1O9nhT
zXc^;6C<VscWVRE}>7j`Ut4ZWeNs$^Yo>*XHBj+KwaU%<>Cq3;xR=Bi|P|F>etm`-1
zG^Pz_)RmeWDn2k?MNcdlR%rqXBN=cWBdk$GY2t=&&E0xDcL4Du!KaQYnTwV~hwIT%
zs(?wn8zg95cG~VI{AC39BZ>z0>nYl5(BD~3h@hQL6d358PJR1eH^RPdQ*RKgijfRb
zI9qM5T^kj*GSSM>X3xNY;p`@Z_+ubc`e$Or+Nz|&hGKPXV9>$+ROVKFM*;hU05?48
zz)EyJJ9^w1Eudr}TMEV9=BU%BIu~xN!WL)UAsO^L->Glb)QaLXQ$1bu5AjT?Ozi5~
zafK2L#`EXMi&*-{t!Q>fn9|P2(N&R1c9M7t?LaPLeE&T1Q`rNba!Q=ZT=<vdUNlOQ
z82sG4h)TZQ0x;ji3uGEl$~xsWHZp1Kf#3Gz@{0BeUVtpoZ)DJ<2n)4;9A>U8VWpXj
zzO@Blet~B!GK!P5Hf_{_tkt|J+G4n$eM%Gb;U06KN#8JAFKgaoxNIKPr8~vS9$n-8
zdX9u8R{~N$DbxLVQv4P8-WZFcBi(3hzgiP;A1O!0W7Q;!V&<xUi#J-`>}&910DB+%
zfJy)kaY;#LNMsB&DkU~@eHN`0drvd=Aq2hI7^LSFf%#a&W%{?tsev5Y5U9d(&{SGC
zL{Rnvpc~i^{&9n0@y5U1FqGz)Sb+PB1stFKz_EWk6YMali?WXZ{V<8Oh9RQaNPkre
zYJ$E3oa-<N^$4<7Y=WM+lKcW-)utln@+HhkC#xmF=Kl)UXF2~PT>n?J4j)>Z{$J5r
zfB(OuHN3vldNEU5F$4$p+(6XjgI%kRQ4Wci6ua9rsV+LhVntL0m+b_&Wkwb}0K-}W
zK<Pmzw%><}9djc8lRgOojOzDiBJ`wxAc*4R&4wCO&*;Wl4n|Ue5cc;w(9fU9fT@=i
z-x5_=2b<VzRhe9&Q6#YaI&9pp*;P7kb3G|jM1p;#w?-Hpx5mNU9Gg&@;;<tiSSmRZ
ziA=$;be=r1V1pHCk1*jHYasso01V(wp-<zti%CW8KkBy&+)V33<j9fb99c@Q<_AnO
zC9esIXHg(lsQ#!Q7USyFdCiUem-0np9$?F9kE?fqCT;U`j&`;5Hup<B_^l<pmImZx
zjk7nSd$OAr{WU{36Qp~xDcr*B9jeE0VdX0`7$l*dK{P7X&IQ8z3WRVy;DgVi;KT@O
z%|a=godfCLL0F1OH+<KWPpWB7_R%^xo79%H1)XmW8jWI!Z?$fP`IMefx)O5<Bv!K{
z7zE84EgTl@EOQ&5Vkz5~qz)>K^VG0;A18PH8?lHC_~NRD3)0SN@R^D}K-e#?;4fgg
zJ-N*2G%Uh?ek=f4fq;-xWfW%E?gw5{XC%^)1e=`?N8mvaG##Kx*UAz{AkAfQ)8~D;
zyj6{CCloyLXvMj8dOfnZ`Rwy}or}Vv{Wf$_H_cylJWl=j&AH~18QEH;;r4^00szw7
z#bCNlq@luVacMq3%qL?@m@o~M(51$$E-XD89&SsKj>-TZQRU>wv?BIB<X_;H8VBEv
z1XJ&F%ziS&K`-`S>g*h24RD|O>a6rL+S_sMl&OhwWE7UDp&cvLo1uH0Q$Cp`V3@3}
z*^nXr1a=mYMA8vUI&6oh?l3E~_`RvajQGB(A8)tBmlh94(TmRvMA5f81o8T|j)VW>
z^_?M1Kx!roRS3RDD2R)|^U?3yZE)RYQ7}f}1p9YF;?QlbPY9=XSd~@4<ZxJpW76q0
zrEj$fvdhJ~t53qvZPI9_MN2EB?{LV8ysn0L8btw-xqLG*d!1`%n+U68p~`vSE9hdm
zhdzRKm`AWn7a2FHZ!)Dn`4fhD(3Ve{0gw~gnQ(2C3+(VxMRDZI5%4B)aGO?iWq6Eu
zm5u1|!dL>?*xbVdySyaLOKt6-K6)D36v*~%?Fdm%CZE_8C=Cdz3TWUp2a`imGdEb|
z*}o)eB|3P3cBbQ@OM-X>=f!+w^HYR$rh_S1V0SM?g1oVJi^-&s1-QlVE3c{t0r+eO
z;PWh+qb-_~7ogA$*1v`uBB{!f>zhln*yQ<tP?NyKQlr2t{}}f<7ui;RWU9(>zG{
zmXaHLFk{jdRm64de@Ahbi^{*d@clyXzM5CW)g_>B3my~W*`;2#d1`|Cz&(1bB+r;*
z_LV-wIVu5%ntE;fVYp;l-^eo#0YI%~xc1cAyjJn3JvGF-$$~2_;u&@4xj2ec<+V3^
zTsyr*e6m+(r0oI-L)O2??8a})c9fmuhy)CSC}Cod<xJ~B9xOJSd_d0O>s{JHuZFcZ
z;fuJ|&RePqXKuqIyX^gJ=HzbHtgpgZ=pv;Gksdoa&@e|WunTqgu@j3>2YBcZCu7=)
zJy8s|=(vYHqU`5-odQv9$2IvxzCBt`FT+b}L^vOC5Fg;X$tjnzl*M)$s-w$rONwE`
zBB%p{+nMPdGvhPa0hy-Kirqh4g~*^<;i!{Jtsc7L+f~>|1rIJt+<uEPqP1=woIsxc
z*1*2Aij@a?Vg<5-r6&fm4j_XagjmmK`O9t$b?ya4BGJjmQGr{O=`lZ*XU3B!7OZvw
zLpq_1YJbOgHF#SVTZu$TZRS?m{Cb~Bv<mwD%azdta2@?;8FgC_NBG4UnQlMN6`9UG
z?yG|C(t`z9fN&rxpz4J0=yaBy?L#V{`XxR<D#ryN+f4K)2V|KjSD%ZEBdn|_$)1UE
z#rt{Qi?;491xq#u>sUgy>YF$D11`DfOaMIR0D^lj85mGj9twoVPO1=__ii9x0sop2
z#r}oP0`f2%!r4ClK=k&2i(X&!7cL8ee;8dLJItYqS_;P(zY{m@fVhOg9{Bi2|5Oxo
zQo_DK9Ek}Q0O=u+tW<scFS3Gy{+FzvjCVpy{v|7-2>*~3(Fq_|35^4Cl~CA!xC$)m
zf4B<u05J9VhpU8+{=-!SZ<As4MpFg*?&E-5<%Ar_RlEy<Tt!gyAFkr<{2#6Y*rb^1
zGTl<Fzj-$4`sq~PNwAAVtd-w&TxZI!Q<-IBU*`Sk%+t=gzNvWTxzfdYL2Va;bvD+{
z%3LGW|0$t?|FPxLF`Io9w^Q}K^6p{2qJ9>uJ6h<-;T0U}QP6AY6xhttzWFK4f6
z*IKK^{1AU^^(v2Wo%+2-d<mO4fQi82CUpuzr&QOcl;B<W)Hr>w4VO4(zjq<6WYfzl
zw&W{HzGKkWm3~_LEvLg$HRE%hk`lH32lM|UpI|xptmjWoe3Nv9W3%04O@{jA+#HN*
zn1o!ra*oN)QhdTC%qfAyrmP?XZrZyF3Nnu&0TA!yplmd|F-kxEPc6-}MCx}286&O`
zTDiMGYzd8~PNgJKA&=$wbJ1r~bqP2kp2bXH#IN`u#F^YM5-5wA(iSrb9gSK8xnJ*O
z+NZ*nB40f7ilTqsw;!+l=^}wZIX*i!f_mkhS>7q+8C|PSy5ACX+)>3~UjH?f1ZWsG
zV82JOgm23-3>J(fnO`4`9`sQ$uoNjSsfF^DpXP*!JpuYW_T=Dwi)8)|0wRrQ<X#!N
zfl_Mtq53rxRjYLBSGg0+HnDEZ?Q4ML7(tRO*hlXKg?AbYdsv!Gpvo;Szgmc|yNvte
zQrL+j3{R#B{8h?$c)k=NOzA)fV2s)=5tiPumO_5gxZ^wJw;9CES({Uq5nfq+*uL~l
zHN5YpL$F_*ai0SaQz4YisyTXDn)uT~VDJ&Rk5o(oKq*GG#Nm8~S*+z-4nAa59lD0|
z!qNZ6s0&Nt>Kin9V07Zr-8*0auf9=*w$a@S;%4?G^j~!W(u;}*8*qR_d4?4qTgQ@T
zl9kfN%U;w`$yTf)l1NelLEcZ2P`Nfu%jDBQdm{-3J&vVS=d5^s@j|L{ei?L>_UG2~
zp)=>UY10rkw=AZoAORnv81Tm8xA=fp0A&Rt@e+>WNqKMBJ(cC3@Fbd<OAb#3hXP?y
z(#`u6mI@mJKbvPuOAUbbDr2V00`L&z-C`CbCvvIE;hgrzB)iQ9UmUR-8gODGU(Z^3
zY+aTm*lKgbO7QPlB&mdbn$+d@ZQG1d9hJ?UEYukY&2j^L1j;>y?^{1GSaJgrCZ>d;
zOIo3Pk)}8yW_p3{o9X@Mkp<=D{`Vtmbo_hm!LeoxK+Pk{lfDDbUOQ_*$=Qi2hIgY9
zy`hKnZ@!Qb`ftAQJ2L+({D0z{yZzyR(S=MPT?k<ROBa62|6g<gVe0><3;A&t`T_r;
z3jiQruqBGljkrIYMWPhmV(AmbR&raBggmFwAyfvFHO^}^l8#|=@4XRCP5)i|$O~OA
zz9xcG`Y-(k4Mo1M+!31I4o^$6R+Fe^3+4P@G+~zyQO<7rKuwM%P)37(H$;ZApRF(>
z|Cgsb2c2E{GT}wi)TK#=n$HJde$wSNYsqk*f<K-#q|I#hi#h9ZWSbc)3!h0fPbkfe
z$v#5K0=`B>Rn${F?JpK;P1bXU8LuPm^Gp}M5-V4#&sug45(UK6RocEJBgM!d)|bj{
z+<%pZLs>1R#aW;jQweYau=wRdp~1e-Y*B~$81D<QMT-<DPhJP%0#<MWlJWHqF<6Yx
z7*QpEuoy?6>S#40k~dGW84jou*0LH${Jmq12ObRxEj7lYw97kWGIeG;vrP7GMm=8+
zO+0u82nFlKy=K3UyEkk0ul`njdp~h>{@&b?we(Qy^?glgV~cUA#^t-deoXL3L^aoI
z*IBa&a#mELc|~m|;B(M(DSe&GWwePt7e9u|F0lrFoMXh$obLPNhCqnDLaka27t>lJ
zD{`T&E0c01=10<g8A2SqlrDcU`;CG+8ceEhwJH$@JQ^4`=djaPTo7s}YXFf_^rvp*
zB=qy0z&p>;ckpak%7Ho&PgzR8PYOPm*kOGWd5>nh3psED0I%0GOu<-)%K2CX?&q8{
zKFKc%67W!F?_Ife9IDde17Wzz!gy^GO_@<+=wEWV5Q*-25bCO_+S$cW>Z)YI5C5gq
z9FnXgAIJJ-BobN8#$U*5;vu^gXfr0CY4s4U9NJ>g4`Th9uf{yV_n6TxX533@r?c=W
zI<o<YAR<^9%G5$|vJh69<U&O&dgN1M+#+3Rxd=m)Gq(Gni&6Bg<I-fo`a;N>$EW(u
zttK2HRC#4CC07+bJ;uSO<h!pbvQ=?7eWd4^OY4kWEQH_tLG<4LJB5Vzw+!M0;h3u?
z^_5HyH1?O|J$izQYNRHyC=?-&;!OoG@<^_@E=tAr3A;Rg%Mwei-tkOaCXQ9!F9l}V
z$kH^&;}B?Yz;}QnRi}UF@nQ_f^|5p*j%JFD>`>^qck8NS2+Y|CQ}V0L^7z_X?X!2V
zlq%9!$@9t}B?Qu1*kISH1|K72LS13T&ah<2e($_kP|t2BnNEvpG>t%DiU$L<8SVWh
zMvE>3S=L>_k99RE-d?9=+%J=ZT$vx3t=INSCzA8yFYizLV_0xHB6fvBa6J+Q!D-71
zNJTz;FTo|2+ZTDL#aH5Phh&ZB{f+f#51*N#;7*r>GD&qp3(SBMDp$pcyhgiLDIn9D
zxiX}|4R|N1{_*&RJDJ%tKyKdl_YgW58*qC4l1)*5;&2qULb5xBzTqK?&bLtvn&^wh
zoM?XpE6=F(wLgnm+_#<R-O2vt3bdM!)DNqgFE0eInr~HRIH~oruoy1A^)j$jXvl>c
z%_!_l@z^lvOi>k^?IQPRQ7wGH{lv<6<Cv_IzBh^qLbxxAiGmf-AH|f-Z*HNm-`H$J
z+0_!G+lnzb?)qCQ5MS>vAsFWY5`u#9{}2Lvy}yLucoz@M&iOAPn56^~0^l$Bw$4I+
zY0pJ6lZ!^}+Prr#vtIiDqXCRMCe32DtjljbW_wwnAb0%55h%z3?xFAT?@hrzppL(w
zNFvMi=kh9NQS9ziWAmsaepX|9<knZgx#fmpi_YQ^%;GVWA;z~IQbIm89He89vZ|YD
zJBdXE3HA8PAY*QLfSrPP+|q$IE13ul=KrAax=OsC!oO&5=yhmKPCT2`Z43;^Gp#_#
z<QHq+Z!D6WeQ}Ki%$r^JA$6W3XlHEw;#wMz>eR1Tuyd&ob<f<tlWIy<YA8z*D<6`H
zM8CLt-JAchcB(cMVWKSKzwr5`T+M4yKf*l~p^&6&$fN7jBxI@U#=?2U{3_=DW`(7y
zy?Y3{<~4jfN>_=e$P90&8#J&44lZumyxRZPN~hix^(2Z3@Wd2<F21>@OUAcya-T^u
zbrlrS?jX31m$`%m|AD8w32*EeT9~Qt<8sWg>yOixyM_Y!2SJYmxJtN6$Ugg{QW_>+
z8>342ey>LhrgM%GyJmy1KZlRr?rzs0k8Pd@V+_vm&ca&T$*ZcV70v066CJLlso3!(
z3_B?Y{T`4EfT!WhRQ4lh#pmG8g~aH~^Gcp{9q9ul(yJ%&lS-j5`#t-x&(qKKKD20T
zz1L+D&O8CTM6zt#Iptr)bnxW!y<$AQjqA<_bMai>{ZQLNxE&AVP#Y-C>dFD7SzUU0
zCKP0dRO>D!l)Zrw$3UZM+vV{tx{`gE0}EH?i5^M7e9pr4W&N43UCG3WKRrW<+u?mP
zOIiZaQ6EiIi{SCCx_i1DLa?a3lfyihyc19+wPgBBD$?N{h~{<wi&V_({t|cA2ma@i
z16vE>iLWoOtxj}#M!h*u{Rpb-gzrtp(iSX6p=4Fb#JV-f?)Ma-q}K;}f!-V-(seZ!
z>5UlrLmy34P2LkWfHX&j8UN#o-TS?V0gv1O=LKzCAx=Up&c<a?ZsbdL2(ExikiFSG
zzyD|aVKMW+q&6IPkvV;oelM~@F`TV}P~6}revc{p(|U)dl{I(fv|V^7f6RU_ur4>B
zey_9^>M}F)KIAJA@VP}00K!#Uu<wolT6oxwfRUr0I|)nYk+(k%wPZ5+!8Osp%SAf^
zZ9ArlpaDi(Xn95i0Y*3o2@>lpIEM`2Q2jvBIKU|M4s3-Ksvl`7&b4Wyw5UB}psi(M
zU51M&r!wV-VseVu#tyk$Bep<uq<KVB0wPn>bPki$7qgCafTbRC$Cu)dfr}YqJ*_W?
z2M-o2-N#l1+Z}dSFEgvs9?fv$CkB<|RkV5m6rL9|)+uoXpdJrIy+P4tm)U%+RGQ@)
z_}I2<snBzre9$yY;h(a%YQZBx<V-~KX@>Aga+^s3IdXdPPW)f?_Tu<cdC{A_ejwjE
z!-=X@+rxQr10biorhqwfq`=XAA~187X^vos>DvFLm+3FB^AuF%$DD;|7^rg4Pt)u?
zhybu`N3{0_7QKGIf-PesDilpU9N4t_p2Djc-14dQ<wBMHdJUL5FkB&yZr=hWg6Wo)
z8;XEr(1+hITV6^okAt>WS!44l0~DRr^oT(d@$MJ{N=-z-`JKr^<7&H^5U?=){wgQ+
z@X-IO&Z12$+$yyl1;GdIgu;{}Rv-*1@kPGCBjIN%luAAbi62cIg&B$L{XUOa3wX(2
zb*0ODzpsc!znO4={%E_~>=<_r%aJb$)tf|Qy)}U_;v@gpoPV>55h92!puid?*OnqL
zHP?3HrAuV}7#3zta)vAaZzQe$qxXQ?zoMMw{p|?FOy(b5#i`>TT_vDR@0HES4%?{%
zreOcCt^#@!&%#^%mE;h`k$V(ti#Y{TBBl^cISe^TiJT5yecw<fHGR-^ShhSDzrA49
z-ZXYQxf3mrvjb4ZF*G)zV0X-R4UNwNn>ztw>!0S}!G?^=g=0TeFaiaB+}aR|@g8lo
z#GN$$Hh;5AmL`Tt?I>`NxHd1Si$k6IS2S;BnbLyN#VJ8rX=Ooj&<Ejz=(s*ygi}9T
zh1ym7vgm4pUHw3}nB96ej6bc^MVdw6PY2AqUxZOo0FHHY42!Tkec_tz7L~L7)`~Jd
zr>vwk=nH}=gz;-G|BtbE{tl~Qw|1Mxw%yonY@@LoCyi~dsIhH3jT+mwZ8f%a(&u^i
zx6ilt`QeQ9AFMUTy63#+4Zmf~yN92=2#n2V17SyJp*0<f8nd=9&`f7BQ$gQ$-Dxa~
zDU9-^Q#1Ry^PP*PEJBZ?QD6>@4FcS>b|}C1$Vs;#H(3+Q<rRK*#G7m<r!M8$`qUDW
zFlR`s{VdgRrU3+pZ#~BOVlj03sujZ07teq#u*etra(zP@MaYpwRe4&ioayh=`L2_^
zF?2ZG<ki{(f+;zC{@O44S%-P(1g_O6)Z&xtke2W#2J8=#!jSS0N!cX3{x6b}^y@E4
z@%&3t{(<8We2P7|Oms+udxP54^6G%>`RE~C3p_{9s(%fVkc=(mG9TeT_RS5<mx5#b
z_yFc_cR?UD!Usa=%0XB>l(D@z`BJw1AlB>F>$U~cvig^*L{5M~HW{3sa*zc13|nNS
z<M(pZe4g?L9VetwDaT2gupvmw4s#r>3VPKg({l>a{3DdChbN9qln01M{rQ56m#PRN
zJnY=awf6NN%A)z1k8v}R+nUDEhFY(WoMSExmHUZrW&lx~2eN-KfxIY{zeoi0feuyG
zoRllF*OAGwDt5G92R(+7w%LgRmzo%p{-`A=$Z~sMc#T-$ushC@rLq!ez7^=0!3j^#
z75K_R5{5}`G~^fXG4mXojL%GrD-xfcHtKANaged1ldL8gTCHKX&OK1Q^a~gC16o}U
z80UbLn1AXw))wwuq=oM5UHtXWo5o7J{r64B*jZxinFfB;l>WtEs~5~gJ~1*jA&jNI
zF*4ys!?{*iLyn0cBq>Ke5iv5oZXhQ}Bq@-`5-otH^{nFBx9w%quQN&iOS3$1jhOzM
zNxU!w1^0>$!@u1wGH24w+kS2QxR=(9x0ZZrUF5`jvYvd0+Tap)Yq6G`w2-I&o?8~-
zecoL2S=c0OZ;6_=2&m4i+pj;>u|I|R)1McDi`^<fD_4LpN(Z$3zbpo+qjTc_x5apn
z|DP5^OyaM_u!8>AVt|7GwHUra|FIZf{>x%KeOL^xmgj#R#=^f2BU_%7NSGN$w8$9D
z{|kVHCfNf6A<9S@8bXVOMwL@HoV`fsNDtZbW;=U}VCI|o=Z~*=44&y<Pf>w|*uXP)
zmwFSyv)DJCcq#o#qfH3rE;s3+7)6tPkH*vl(pD5)?6V$+pw2w$A?9)NhkgeyU~~_S
zMZRRlDasTKbA4Q3MCudi<nP#CxY}QtJb;O5$JYNPJ#q94h<6s?c!igRtSMyk5J@Er
zTz(r?>&~iaM)y+x0wD@<f(Ie3{)nJRt2sN^B@B2$1W{mP)n~gbhnNC?m+|>6f|S-&
z9^rKx!Y<^szUf8th!?~P0E`l{y5w-NhS7NX@$L|ab2&shjd4uyqbM!1WvTzM6G_?5
zAM?cjZ5(JIo4UwYM+>L1%2=*a4@Sk7e~Ut3O8Tx=*<Bz)4=DQ_L&6bC!ZoGoQ=$N(
zf5r#?aSRdWb?VL)E3qO@IDJyp3BE_J|Juba5X7WkOy|j|e#!i!kPdG%;ZbB)+3t(w
z;CO3(SmOA$upW)z<6{6hPl8miaooT4aNhFEM?GA(jQyyGY2Wfc>Y*{}-+Ea5=_j^w
zP~lMk1HXEj-Rf`?gT91fRXBG<lMGR2>_%Q0VsyK7F)Dph%XEqqz2$NzdgJE>UOV)I
z&!M$j>wCs9$t;2n5Vg}n9{=VJ{-g`BpLG2m$lHNIk<3{0J?6L!n|COl1bM<mQ9@nw
zQq%5*wX;dr7Dt1TIru5v!+EddUhT+&J-J7EFGe@3uNZL(yx<nsiqHjiF4vTH!MPu(
zG4}R5U9adKlpOAp#Fgz_wr&B=L|1`>-xXu5V+eJSPqqoi$lkwSe?rq~S{3_paYhOF
zupN1nUPN7)?f+FPd9^=krAM;`l%5<#PO_fd#A91m<{329onuu|-bbmVm%~q=!+_b3
z|2OhKIs7N`htT{t@&_&di2Ma;AT*K1N<Jcge3)%tUTs;l0u#9;Na~f+PHK7`C{YSa
zFsrisuM&YA0Plyr_{pJdAKrk!C}q9r3jzuEm6B;OTS}+ye#ouK(R<z1D_DuPggyLz
zjp{Iiec8#xf5(=NY9X!36dF=^L>0TyH87>UwgS$<fHr){<aOMX0q(WyvCFG=3u*J8
zsDjmJVg`2C_yn@+dnpIJx{sQ=YnyTBN26$AP!rJLI`f)j`>{EwLat1kM$J8Pa9g^X
z>G_o`?>9NPFW=&p6l?@VHz2v+QP?2MZ?a{^#6~y&*uDR?uH!gLuA_XksOdBVdRgCm
zuV2(;ZNbTM^n1+Qw~H10*9!uxX++NHq`}*qJl1Op3oiDgNX<n`G~+-w6Q;uW!ib48
zHb&q|Lmyv^gLZVx?~(_8rVm1;d8$OqkvWHEGk-dhn($dk+bt|<KQo<W(P=kYH6n$W
z8!hhdfJ9}1u8?U%HB3x@95UyG@lUJ947JjP&<AGkU`x^Kckw^$6C5|hYUA@Uq{X)*
zw_<tsiAT+eYR#ssVKEDg5FKk>s~)7_`@rtK6+4F(KO}Qz))$07{+mS~4SP~&Mf<ZU
z*B&0p1!4QNdnJW(<hZ7xgQgUpBUeENGRl95q^uWP&tp2}r_^!XPorV<@4UzvDyokH
z7^CO)f?~YRmF3Vp+G1YE(;XY+R}B^Xkq^f24G@I;-4~;hKhH974J@lyJ08^f7Xif8
zj$X)0^A=5pBxrC3U8TesZU^@f`XRTuxCZez_9@Z1S{9_C=r94e2dEH44fQ*V0otU{
zgK`553Gk=Lf6^oDu;2f)lW^0|vbGu=b1PS~5?!f4V%QD2aDS2B{p4o&Z7lRySoUjX
zR4wj+{uUZ=71HUp%nzsQ)U%NY7cITAYW_7|tCH_a+JxJjj%jENhmo7BcBGS>`hriC
z#Dhd2#0%C*z;l^LQb<n-G+d;m&*YsRXKS-OE(Ek^z{{;+>&v3BvIKfnLNu+N`~=k?
z4fP$8{=+yy?wciTgC!|OS)Wu{Lem$tuv3EwhO{ogLwvQB>3|_wprOq{Xt?QmtIa_J
z83!_s9`8HuIoTo6>3a8}HqBrvMAOb1?OYme%Pe-Q4>xyXlY`7)r1cF-lcG4C!X_$h
z%=_TCAMSlpH4DogYng%bBcm4^CaiTBVnfioHiZvfe1PjLCZe#kuKJyeEX7@{>f-s2
zxUh2|sy$+VN31}{Y_{}=O$LeXRCOY}47|mA2;B92Q2VJo<}q}|ySuhHXlv3nm``FC
zQVnyjVuKcMR6AfLCrnGhByyk&hn;H((!SQaqNOF(Rxbh#S&Nm^_<FMSAhKdPI>pb`
z&fbr>eB&(rs>+Z^ZPgInz^7$)LO64NQYr`NVx?Y>bbt3@dM_?q;o^4k4ExGf-einf
zu{`m(-KDBI#`vC2zrSzzpeDwVNsWhSQdXKI)t8HES!InoU+$?>G9PwChy40DuO@d0
zpZpk)bU9w>sB@grfDdIRft{YhtrgT}v!2}(3ugw&REchp;U0vv54!`v=%_~yG4ui$
zU#191XEo@v@Zg^;D{9=Ff^fiLh+O0|zjdrsjC|bl>z?JgDm1hoYOVR&_;zE~;#H~{
zNhZTSSdeVgDBoR~xtsdkds!;1f@5cfc%5O?_Eps4_I(lCU;*y?L7%zBd)sSkGM8_Q
z|MRg9`HHL-ANJ12<+x<isg;D~*^jN1Havjmlm3!T!*7bAC(~FPnQs5V1HlSDo%bpH
z@W7?dbU`B*NnbV<hqANcm-s06L>qJtl@BGb8?;GDyw3+$Wsm~RJu4Y*m|ztE97But
z$^8azk(ou#UBw#E1l*?0ip`4cHT1Y@onPq-X#wfzD{uixDEN6Thw54AwQ`ogLE_$6
zdzAaP{5G&<J?vWN7sH|0qotc?A>CHItn`;Gjsz>G<1y<bu6E6h+O@{&+T>lk?ZlNm
zO|{iyACypTuLvdUy&vNlr54-$c&E;MX;@(e#^S6g+3&U%u)FcOa6by~->9yqVHa#J
zJxZEXt5|RP9^N=4SD3dOVlUHx$R`T!F~QDpT#BRztt_*gr|vZ_kIS{BH!set*D7^^
zhBM%ZArjP(P^J*}0J5;25O?MGk&;+q1>HeZ;5(fqR!?$ka|;Lm#;NiIx0wQ#5$1NA
zRUcR^1iWABgQPPOsh^7O>Z_jHG1ZFh9SZgUM5~c;9|nx(`*=EoWnBZ%!FD~PaDriC
z3E5$)wl;M;r)Lu1meHen*cw1WLF(8|-FSL>X63Y8v4MFiRe<kyDR+S?p3o+qri`sa
z>%njbR6eV%!JaylT~EvpSo_sfvURrcl{;6vOzq`?Gjp-pcnuZ@h9iu@75=&~zH+Zw
zzs+Vj(nSbNbwD&daGwQU?>;8pypFtEp7sGKHAppzf-=Kv*>N4md5@6`R}GKWGob$M
z;e|?aIgkZxnB$ow{8)A<Xoh69l)bdA_H0W=xQdSbSJpz<sk7z`A+}9aEV9AGioxLF
zx<V^Yfpf{C%yILLBTTwTxz~+tHwdR6?%xNII3C*C#YL_LXi^Eh4)}P`1PD0fxYXOY
z<bF$das1$&mle;1kFrb{zfU=P<Jb1A7Z4gZs+wvQPetx+ou^Y7-d%a8%1YC6Tv;IG
zzk~c}F*9bhol@A(jvo~I=?kMliJkAgLd<TZhCBBqC;K7i`t0P|z`J3Rux{Pan0BIK
zyy7iSu^1hYGFhJ`S$n|hg|S2a=V8)i5P{A?{7NntX!dP^OTjl}lT(U<pu)Y$^qr8n
zaXsU;3S2vL?Xdqu25>iUam}RLf7+SNx6pDikz<&DUA=iLo}EZk8hMDaQYsc99_7Fk
zHZiHc;5u+e00M*wt63J`#=TEpx|>Mj;~~I+M7A-@)$fEjik5Jy?S*K`g|04IMmzmV
zNS8fv?X}_Ci95RHonDhk<LC(kQO#`a4B@v+CyTD$(_@me+TCT&Yt(8|Z8*e|N#UKk
z%r}c(U=$R(?eXp21s3WbW%^Qb{#9TQ_xCt1oX4ZvF2bB~+luE7jW0&@wBOH5e*Qur
zJYrV`-P4}g?3BMM9WDCcwHt2X(2CLl<#vyxB7o`d@W4Y5TL#@mTT30fnKaH~V{tu$
z)?Oa5zSy&oS?klWe2Harx_Q)fNLT%%r*2asX^bD$%@H3DT)J_dW?9l7pPEC7oi<gt
zD+_2ig+=i!(#Eht4fEO~Oa&&6hY&#Tptv=WDs6EDeB!T)I>}$Dswtm%%9zr@dh);a
zpNAZA^-NuvYu`H^cJ;u;^&pt<MAJBduU~3wZf>1^0v&TM^Yx-?dGfsPpNKjCoR5yZ
z@CU<mDB;Fv)I@ITN%%DC&32K{)5tZ-``lbj)8(S_ZEnwJEjKm?*muge2PW5pAG}kv
zta>fPr&IaypDXVF&-CW&Tv<cC*Id61H|N~qnN?RuA?y5ZOSxjrlNkE0@!)!o@;se?
z+@-Gi6B`?wiHh7J9q9)4_lr?IV2%RS%Iw9A_9o@dlr{-r<<X_*$qcLT)q#6OExRz7
zk;_sdIP>J~JTn<Ed}hKOommOkZv;QY$8{RvNP;RB*?p3qM$OWDmY*w5l)jY>at^?n
zFmVVhuF>f;Xi=wmvS=AFr)c$C{Ql8;Y&EKC`AXf08PE*QL1MfyjhJ>63id1bJX9nZ
zGaS0VcPg27XUUJPD^>3s&#xgSQbR>O;^8)88PUAy#Wpjb61=<j;q4Vt>C|O8sz965
z6PUu-I0@oETkNr!_vC}Uv+IG=50o6c@ZVw|z?m3@ORDe3ucQdxX`omS+^m!fd%RD%
zbj_`Xv_Q7?e7PHZ_r448x+VuDBn;xa^mdG!tC{1lnQzW;z}9<kdVB4NsNR#v%*z^4
z-bZy?p$7$yi~H$KuyhAJd7r(x5wsSpaXwj&sxQVb&f86ER?nB17}TU^k9gj&gSYZ6
zP8k<6XdJ)Bcx3EIsm&aJi#m&G<tDr0bL9eZUkkDCD*3;{xl+4QOY)-Q#~V9%eJNtU
zvJP*^YFul-t5KBl{^&(*ld4`Rr~2~?W!`1|X|fuK_!!o_W4CIiP3AYQF$84ZNhQ^O
z<olsX_cdpnIVd1;YPtv0Nc={^ECFN9=Asp*2r~KIvR6W!Y%?yqaa(YBHF&7lyQnsg
zVi!wVbug?6zeX<&;^)z%Wn)~h)!JfZKYPQvgD(wE76t>%V6pcaHB(|u%oNmVN!l)i
z6MY9X+$?s*Wa@Nl3oea<_zy&)Z_(a4vt(J1*C#zI89T<1Z*SCymch-NPs4T<eHKO|
zp*c;VH@DN!O^O6cpCanRHa_A3ep7Na`Cy%tus4+kmgd$chEoTNo0c{-%nxEc+jdb@
zy}j^BCA)Ldyu2)G!7^C3rSSxzu4_-?_zf@tIB?n>AQJF!7h^Bt`WEQJ4r)@~Y8_$p
zzuJ*$1X2i8XS$^edpbz1oURGePD9&H3(_FR-sY3kb6Pn%wSDzWjiMQ{RSRffPZU?W
z%lN(P#vyfda{ON7LN{tgGxStB+GWy^J>1pWCCjOz7Z7#K<&3fNTudqojTJsGv<;X-
zaJReMcWfNSfI*S!B!ZEU*QM=15tkKpc+1M&2T}`j92lq(1vH^X_sVvOg9d1_sFjDh
zoGsS`C}e6vi3Hc!`o%#eYqX@0QkU=%Bwq4emoP^Mf;CX5%sz|cww#szEFryP+moaY
zLDV;CFBxy&BQwx^b#N@>g2oT|eEFgOxVbq;EIqTR7qdp1+FmPFX8lmY<MNnT(&|S;
zeJ>Ve1Z=H=1^?Y?7Jm1<77KYbvt5gWZG3M8jCu2l<T7#l2Mt@PPXrXfRhWOa9FLB+
zPeKsgTLJ}bzSRt*(Z$pdfy13ten>{%;ti<zEzP_PHL*5hoP6Q2@qo>Qcpxd>z71~<
z`B<Gw*ED4kJ6RPHhpN@veaxiHM4A`<Vb9W|PTM&?_+bQfb3*R8WS~^k64Kd<?o}nB
zO^23lAvX{%N42O7rF=pEtl7)x^CySd=s_a`E}=$1k-rkey_?f()t*1@fjt6&d#LB7
zutz$!XP#twEo@8$#dpzX73fl+IgA6-++i?}GOwiW3r1k`a~$i0f_c^W)g8qn^F6OX
zkd34#y?*oudyNlXNq|fEPQ>D=X3i3PK&hk9!*wkP&lx`aoLYGEWW<`8P@gv{Ea?9D
z(3z%EzH;e-%0>v=_AXZUm{7=Nw{!m|9Ot09aE-e|3~Xo;6~p1K6X2YSeb(#~8O}Yg
zG_{Zv(+MVov4wx_jGAFx9<f#&Lrjo?wd0v6N94(RWx?vO8g8rc?hcHxZXd-!sa`Df
zzK`A*{qh8JY6nxSq&sKYA6QGK&4w>hX$qE`4U5ZFq5_+Rw%@=unE!xPOSr_iwcX(P
zB$YXlGTh6|BV_#i7>VF(2y`%>b@A%W{)}{?`wO+TO^NKaNW`5m%spME`B2y3UgVoG
z5!k>ZRbS#qN4H7G`vK(hwZ{>fhU+hu%&P#rXVSXXk!k_gy!0{x>+7U%8T&u(nR<#s
zuf8(Pig<n3z64;8!@+`S%_-lD8xovOrA2mb(&>k@+e4t4WrhK?X;({4v<%{et8xSh
z!8Fi(zfv4C8B)talc^nBq185c(0wWUE&)c@46n4uU;N&k0?T9~51JifO^S-tVeQdv
zxo3TxaUZt|3eoa47K;>gWvMJr&m<2v2rmqJe#8x5NaeZf-18ovzlvCo;t_ZpeAn!8
zexEmBAk%V$CQJqrJ6qH>-zZO59E7hi@ASfMR`qfszP4c8TyqJZ=GA6MzYIO!$`oKh
z!hFusVG4>kS1d*h%+C75ka~$}lw#vao7ef|EX0<fr9~RfpDI{fdfa#p6SYS0o!#S)
zqPh-va%}Is*NGl1zrkIvxqMQ*`VTg9Sf~B~&G_QAki>6*+{HBqt(X@o71LyZBV7;J
zstU4g#Q`(+M%~K<yuIg~J=Mm?6{(B$ai?O35Pr_;9Dfc^MDC0Ra^`qw0w&kDORaXW
z?k-Sb*V!o>U)&zpU+AD0XHXNQPlRa+4J;HK9hda!UM#>Mi25HS&uM`WbTS@wJlUGy
zEq@9kBGL=M;<%B_Cxwg(^ZWNt>Vbz$7R7&u)t|(%%wK$Y_(`Z$t!TSEp-yzMV9AlA
zO`R|r7pgmHeL*7H=PKWdyxR)--I%vCqDn0%C=~S&BaY=;SNg>Wzc;>r^Xi9(En|2I
zs}e<^(3qCR2EVTcwb<~jt}`#u3OO28_3ZX?3%LT;yXk!&%(4ut#2wxkRZF)uw_4}W
zg1+k}4j~_`DUGW)DgCLZZ?d0+U|@;M4r*e`LU>QaJ-B|SB^X+pmsz%S;fpc8`b6eh
zHyesx2s}RBTm<95gbfiA95*z0s$Wun79BAY-yV5NIFKghB~@pMazYPHtFy1e9{J=K
z{U-%*=vZT}&AaUsHO}YH7<IZxQ6S||pmGF?vY#>BUxeM{<&WR$OKJ%dA>btEJObWg
zJg6ietHbsL4cixWDB<HcPpJgJ%4w_eg!JoN>{nb}Np4ikhnDa#gBq1-4@Q5$9|<yf
zd1M0lZ^E>9{i}zZzc9hQ#}Tcy*?Jr`S9$@K2#^LV&Siy?0}C=ToIXrr+>_3A5n7>H
zxZb2_m(_EFd}k*(exbN8e!A5KN2USLzB6Vg4o?tccMNR<GxcbN<?Jj!fIwl}DL3Kv
z1hSUyJ)1f~P4>BVZc4AXFALZ16^RxPUCYms+m<4Vph#V~3TUcnluN+|KZ&W8uOooO
z1td<PQNrJ})y()9EUh_Z1i~PXN3KNDvCV!{^`CYf66TvEx=N(4_?1WWuvUDHwv?-+
zuh_qr5*C|J4Y>|GXdi2q-tsX#NNbDjWB!m06jj>}*{bf)d3PIr7P<(;9J-=iC;*Qm
zX)XFQ5fWJP7HLV5`g2lvPSFGHyT%NVtb=fNPeTeArcPnKmOVGiFb^Ux{8-*yb#isu
zXJ&KoIf1WKBezwuSF^v@yQkbm`T1K!_SY{osRONy^IGd2$fpX!doKbu^R3TW^gO72
z6=v<>%Ev#0Tu)@nClce696EFQN0bkyPE^7Y4C=I~IgdxdN%X!ZU7M&}E?D~l(4&Y1
z9WyyYZ+YEl`e+Xxe=u8;I=%|rv18a=RoIawF?>BSgX}t(J2m!^sAJq-fN#E#cfm(T
zP%%NzNJ!j9isewgdeo|*T(IhEMjsC=2tCh-J~YRBY7bKD<6K)@6W%XNP=9M(<-(-u
zsTaJCD$hPy#ryPQ$>j*iH8UU*2oXd-M;V7!9ea~$aB{+U@8-?a5KHoO4D1#9J$9oo
z<}anS;6Rl#lr;Hulp#%R=y-{arItqf+g3NMuM4PQW))tZz5P5@t<ce6g3=-boF}i~
zZCd5`nnjGBC$|da%Ag*Xoc>^Pzw?=2tZ;c`hO09WIHTEn`T;hP*rl{vz<wqtjvS>&
zU#<QZD=<SVen(d^^N=3#5&omomb8}RAlFCQI-ylhg5<=q&!J1^7+y6OWyZ2l?tS5A
zg@VyP4||!T&50~EPR=?}4<jp`6Us1hj}nmka46HMIqq&`y(wyKWdEFldI<#{!Ziyp
zws^A^zJSfT{^}s<VdVoS1N!=imq$cOIzD##)m3J-Q@cbd`kr^1x=`_0S2C_57Dqt+
zDylEBq)ymYktAwsGvKBOd1(nQ$x}f4m5SlRQkg5BQpW~q-Rb*4s>oRBtVcYTpse<G
zB!$<=a5gG=OG+fTkHzDul9i!jmyndf?PdbwZz4WR&vXmj2c`a62(+V4T3>)%077o>
zU=-arAB`G6p}Q!{Q4H=_6Sn!Z{1{{$81Lv(Xq&M)xWz6m&9O2Tq~_d3@vxYlztyda
zn84cF=^ww1cQT9PtM?}(=nX&=j@CxPV66n1IyB|R>%9oM3XB_A-KxwvM&eT-9ei~t
zRhM1j^%4LFh$L~o0_i^wQ~CuwWVzq%G5n(VGE6bfT*Qol+?u+JbpeBAuMocrT5}mn
z9J2}GP`EfDPz-KR3k{h^qM1%SB3rBGAr9jwvj9QCZ#B&tHyr8FO8E-;f)GU}kz#O+
z^`(btFH>Bqy486)iyheW>r-j)l=4>2(@F37a!?*zF%>C}z?g{3*w?bm82Y95#~Zuz
zvP$26)}<Dd!V$Pq^GRM1WQTyY?~V~Z_iRor>8BM<y`L4!4~9%18~e~0d=j<Vlr2S6
zGLZVVGrQ@Dl0OA)wwD(jB*`2-ojSE~eBuzS!6}$atx_mQOI66|iBdpypADdS-!2`{
zIB|qKNi+XL1oW{n-g=)Pl&;Q%YMW6-$(^>8H=TUkkiI@}(`aQKa6f+<Tr}GFWD;sM
zxLHfEF(!tLCre1_!M`Q%Ob;%oVP5@Ex44mMy(~48;oRJX@cOwYgM=VLtE?oz-;b1<
z)x;~nSL_N+(<h1ANz>j*^Z4-W=yLz<45V)PUUzD42FROkYIE~!RpB~~eAaI@jc&Gj
zN0Q96_CMv;QOH`)M$=h7?=|CT_ZqpkTBy-A#uwSIy_$GhuxUtZJZyIh65<0?G9!s(
z?br3Ub$nh8@qylU&M<ODv#-t<3{I@`kAZPsuH7l=D3#;U^ZE?)`l_onfA(FwGRu^u
z%eS|D053W(U;IBA=6O;TQ^BS8BCK}iqY=K_E1b}0x%8BG^XYs_=450p0)dA9`R8IL
zSiL(l6(r~d9ge@=6y;+(opKRW*@Glb)B4i`p>N&8pvW@>`e-(8=HY}?wAI|$s+EQ|
zXOn7z!m9<@3^k=Av-8~};ET1<kNQBl9_!L-0N#8@8T#?@%q1Av=A}$dE)S-mciBZL
zrwL=&nmX0k%=;~<)SFhd84})eUL?EP%m$81g6uakxx0s>vyzN&QId%cd)-YB1>wiD
zLN%I$Y(M=r=lYh|M?C9i6_g5&G}h!j3a8Y<OjzDRVP~?vM;!187A};i=3v-j9+p}-
z0Nj<A6Hxv+Dt_!S6Op`!xbKl3_2k$j>0Ak%VNTMN4j8b}oPyM|R=;)W@7?(*eH|o_
zcYglpr6L*kLLw#^_XnJ??FbBdIS=Y6tO%=N$AmV4?JIFH)Zfguz6U?ZbZKjQx7@Cd
z($l1N4xgU(&$x6LM^F<z4%#YEmOqpV@F7(qnnwPWD%{Rk3Dw<Qrli<3c#~x#>Vjw`
zkF@u{UIybmR1JL8tTi94HA5yA=T<HELecvO5Gf(pS2XM*9*wQ)L<}Xf^@F`ybHd>e
zF%X#F;x0Gdcs|*E6H_~tHpZ)OboDf;a+^Nn%6CbUO<Mg9qbZ?|dqCC)0<cz6|2`1y
z^vo|$u>$4Wpv+>Y=_c;1`K8djHvSYWD4H!~8xsqjpWShEb6u2thf6|qIigO6b}+A;
z5z^ntKs_W>^LQgs(57EsbmWfnj#DPazrwC|msp7z`lzZwBn4d~R~}$j>HVB-#JtgQ
z0KB28Rmmz)XDn?gx8LA?0SxR}y+Msf2m{aMOI6}E1#<;t&>B&H8_RFD!LT3E(-psX
zBr1W82fcQ8s#zh;?6yXAyf_(N?m1cV%;>xSBYfeoqURfZ=Zs5`mmI?=DZ{B4F#5{m
zo-@cJ$cjvj?wFJVTAJ;<BvC@YNmrizhD=r}OB_yU<rsyQ^3euz)7jshI%zfMr&8X(
z+xS~NnzwCxWY0bi`C26h*2!Ehu17yeUYc6|PR`_N!ldv=c1_jy{Q>zcy#tK;^v(-+
zZI!r?hj)nr!%%0`z3bcsB=<J}3!l}OYcazM(V)cN>?WXVLqK4CR^hu51yB}8#J6;8
z)SjRJtu##LTR>GkIe$C|r{^h+78si9dQ9q0ZJ9fO#UDQMlixPtu&kusxS9H*ew4}&
z&kDmnh3BzBYB#~XT{ZRg(q$ZR8Iwe#0qYJ+u5)M_u2+o>R5-7N(R?mbr}FMOcvH=?
zq<Lk>r)uI5w;$qpeg0}M1kW=-wh$j-t>E=s+_BWC5*Vfad3VKJMvCUJN<}TX6D&iG
zPs_$!va1y<Spe##kfO|0n2(aopQ04YNp^smqQr&QKAf(Ye8u1(HH6XJb`Wp0rJ2qX
zRpH6hyUkX#sZJG~>=BNH(tb>WBCCAt#@r3@O35nq({t}g$`W>p#R-*Tc!4I#Vt1I+
z#mYtfW1nw4N-(y`78nJ?<ahmtVS7`~OLqQi?A_|VsxJgoFfO8=z0xpoXoW~WL$-C5
zrQ+hrslMq;pkk7)PWXlRy-;#W<p_b^y_oq-4y>JlWqbLO2Z$o9xfnSu_6@OUoygD3
zhM);k1#*XMVV)3_^>+xsK?`fSPYIrkRd7IeZ2$v_OLLc${+r!?;3owpNvqFWFspc^
zn{7AK6ys|C`2?yvt*WXO<8hsV8)Ii3c2V}j_r(VW1)OoKpe>nBJdv~q%s|rv*A*`f
zldebu797-#dN$d?a-`v&)?+D3*l1vH^I}VnXlEtE=<*+G-$MA3b@{qkg1*Ey5Ss?n
zCgA3*&t%dhU}<I=OS-StKZq2exlR!0HIOxyjl4Ni$I>HXDSw?(=5B<}#w=<3xAeBH
znKrw-{sXE}(8;!yGaF&dH~X)h*>Zra%s2GI{_HmT?asl2e4$mepCTa?KR+d~I8RO9
z8M8YEV8SG5YC%QWz-l3BJ>W9i&fi5mM+=fYvV(oQ>SZ$p{iYk2j}l2XIxz`F_L=2J
z%OP(!6J-rnVxP3);xL!|S`MnnMY3Cn=Q(7lxbC;|s6R*~3wo*kNixYF*KAyz5DPZ*
zI@!0Kj?DX*bxEsHX_<sb39#=uKi5BH$2ayjIXRS(ju1fxzcf7U5QYO|X=P7&(~c7Y
zA>ECqHS!@w25nxiU`2EC*E((jF85Iqj|XEtJ3Zk1$h|*QHVdIws%uq7sfM?{Qt6VP
zqut7As|^GYP6hR740WF%<N*hL_mue*T{D5luo0a_i|9m=#jw!PruB|q3*vCppFn7i
z7YSD6U=Nhp#bM%7q%>N?9Tt~4I3<>D1&o8Kg?};oz^eq3P#Sc)u@eo(iab{a{ATy$
zloP8<I{)|!f)jOq|GYo0zFKf|JP0qS$m=&2+8gkNc{J2`>!E22O9klS;L!bk7wRz^
z(9PVyYq%>qv0c_qqUy$tmMki#V49LE4?Ef|14mZfDTAhUqNzuoPPzha>@{XTl4^`_
z)=Zr$VTN8BN<N&xQA~|f4Z?VP!l!I)U)%JaCx5vWpVr&<#SYA4EUn)D)^+1;Aus_~
zsJyFM#FseHjx?ZS1p^R-I@rSBA!w+z;i*j!MPj0i<NYie{{>30=8L|?&MuKCh9F+S
zaCFNqF~b%UsGzFYA^@Fp(I$#?dLlkdWYp&`5H6<<iPvD1lY#y-V2nj+n~`|j*`l5o
zz8a2OApx>mWv!x|Cpa^Aa+)Pal>o|plBj+dRKd2|V$at$wiQrzBdXv{d+dSUnR@yX
zEg^|g3Pn1c5M8d^n=tYq@Dfv+$ha0ns#*uN%~w?wT3<aifN=&WW!d#~AHlvGgx}KV
zrZm|eX(J@8?tuv@t|jE1DEy{k`MF}aTF)rXbMOT9t&Vky=b8l8!9f0tuRzRC&NC^r
z)V|z|F$`)mta{+9EP|quzAP(s4>B6lW-Bg|5q#vUn$VKPPUHNjN*Q>y8^NCZoB;SH
zZTjXei1GYccBoWo)P_;(1Wy6nu6UP*6!|UR<zB`2A${r)tPBvjT<yQaGDzMb-PykD
zIWZ~N>~;dD(EC3>W;&;4^5+aJqz8uZ42Fz78lYVc{bXV<eC=(4+FTI2e`qQhyWd%5
zp=PjLZ<%DPI>kreQh)+d&#}OSKA-dyXI2I0arrU9b`ttuiW!6nGb$+lKRIRzzP)p~
z!Rs%xJ93gy2`*U0&dxXY66IC@rwBd#3btjpAF~#Pf?Ts7Nf?xlq?;tw!pMPSc575i
zoC%VgSba&{cxNq_YEhIff7^b|9LvL$187IORb~B1BnmR%_W9-Ukl3%<9C^i8d!_lR
zbXaF%KEY2oP`+$NC1n1X(PfRV7u1K-T*8R?nbK+%wj^ue6Yo|!pLJ7~<V%`ZP{ywQ
z+sKXHc6HTlFyP3pG&7J}5cOR$i*ynPQ1DRpQ2E%gU5uP>0U>6a@7G8DVjJ>ZBVsx`
zHAeqS;jdFl1VMpFWxumbYa13@^1g~sscQlstCs`#<#0Q9)~DpX`ijG<^*6$|*sB@!
z&UWuA5`?4*+;yyE8DLTjk#f|DPB80PAd=}iVDWHqEP*W-5utYZ5=y5KFs7K!+Ox2y
za2H5lTGY_s4gHR)V=T}CHExVP114Qs`@#^x`Bb_?`wCB$kYR-<4&D(Rzn|};Tz=S5
zh-e_lf+i?^_<i`%619wT?x-1@PJ9KQw)L#%Vezfp10*<jtIKLuco>w;w@>?|Finn5
zycfspJOS+jqTF*5-%ejfMK%nmG_Oj9we;sb<J(eAw8-a7GIafCp8ufLm2%q`*`UIh
zz7AU(P?7^iZ9FNLED>z-_Bgvg;&>GlxU(Q#0bWmb`#!o_;Tzl-VRq1USVz<AsUvMr
ze(D26>^!gJ>9Vt8hBLXrt*AE|20{jY7Jy7488_|Z4-kf)zTeZUcsO;M6<Qb**50c_
zvec~ebr^P7^bp_u#txu=w?P=CNguKiv@QJk`YPodaMM!u)F9Rh_Xz&^zG8sFC{<i5
zbE3fQ>q<}bA--8L_@+B^7wrs}2gaxBEs_~76jbK&n#N-0a#S$*l=+zTZ3VZmND3ff
zQbMB8&Ire34mv@##Cr}U@}N59T&pCxk`*lW{?W%)$abLK{t$A1uMNI#U4FChb%kv8
z=Z@BXyy!I`%L;oRv*utIL9st*&5G8481ilCZZz6&$9w;iJ}iXHt(FUXBOzkI&UfFl
z7K*4i+6wN+pud(k<AJ<NOmGO{01zlZt^1_Ah4;tT4BC>V?#0|HHrlAsIGZO9XD5Z8
zn*$b%;!kqgW^COEdB$iJRKDj)3eT*&iz01hqj<{Kk1^hy5+$P%nrkQgE+-a7*f!~}
zzCTC%Ln6@`Yc#k1L?m#)#U+sBP^OBU1?_(zDK}*%b2P0i&C>IYzJx)iBS49x;^pO#
z1Hp(H;#Kl3M-u`SCDu4y^WY$?(Dm!YC+LBJJ~;X1NVDDqR)Pqxg)4r2l#0`CuK9}=
z)rTf>D;m6KS>!3lKfdVE2^c}K8PAPOd6sErIVWD0Ek@Jp(jE+_ZMFCGj0J5;cW)Ob
zuP7#XH3aay$d&>S*4YC+Ilv<2Q5uBK%seOduRI$~M2|9psvjvBI7~Z`8I94)Y!qnl
zH_+J3!pW>+*ICKATiO|HVO3eR$YR=Jzx9`spwCi9y#kLl?xOOywnw6;nVLAbNt939
zk}DT+{_KCRH;Ej7OV|5sRWYQ|q)DrmLd===2$yx0b=4<_N1MgEg9^A&1l(aOb13qa
zpy6pfg`{T9UaAA{M#=pY4OaCgAQN9BeeqcKb>h=#a?>8QrNzjQFno#NCSL2RzDkre
zXOe5F<QL^!QknCvF%e!YbsJ`Ix1)XUUuI&gQ*DwTfhL74Ifr?J4s|qpDCS@_+O1dl
zO_tOol`3e`-6Tj}lL%DK)3o<cv}Q?I2VHb?xBanyR!QsNO=#vdKGEOyHH^Vjr+|o0
zaDHx&us)MNa^xBzvnVF<3{~_ksS<D@B^D*#I8u98>QL4DR<HAUOKSq&90#@MoI#o!
zP6*b<R?yC#<1UWs%gycK7BU-JVG+S!yeoyuI=@2*^HjRZ3#_khtX0$ep90;7NCHmJ
zJTve6j3)HkLpwpuKG{3ey|ALp&`%>c?n$5YV~1tXliT#K%*q5b-;aA!gdNkgm@fy_
zWUFKt*3>i16VPk}W($OG`Y(+1m~WkIg#ktMH&haKomSZ7r==ApDa8#(4TxP7bO*gX
zumuvp5AyN0yb)<iiT$z>NOET;#g^tanyGV=f2i$QB$$5I4102dt6NhZ()@VY7>e98
zbOJwI8if!{4!a2t1qH!;C<B^0h3f)>AMrMKOrOEfr=hFDXC6kFInO-wk=d=PU&j!$
zu1l17A+q0=)CQu<ODIs`3v?6B_g7&zP5_Y+P!&2ZY`Gyd=&4NjBGwUlTTLxwyI@_H
zQBG{G0~1?oc?o8x{jr9BE5G3<OLcXxE!~?$@#Wh9l+qx{$q5<bt===dZ$g@(PWVW1
z+j~M0bG7%fnD@%2O`rwy*!f2?7|-WWHAS+?NbaxxEZ8UbB`l*a>!-(oMyNn~@W}^#
z{21sLAJf?^d{}i-6PheX9J->@Xm5I@nYWi2BHU<xwBfL9x2CJNFMk?8{_355&gHyu
zppb_#evr;?Z+)1$KO#}5<Ek}!Hv3eSf1yPw+9cydFs4(dPNh0h@VjQbzBZlqMQ|gV
z6*hzTN8QSwXe0hzOc7T3sR`hPNeoNSEEG_<QFCH_3#TOA9LmK!YOm^M+G?#3j*VYH
z+>xc9D!kp0%5fa``duYbQx#%*bwD3Np7cqL6_mNIBAq_;0MFkUHU|PV&rm-zZ7{o2
z7uod=ytdywAt$~pTuS^0($Vt`!(=-_%KW|@Ofcj#wMxv=EBl#3ug(WX+YgA`x`-)c
zP&Ioj{(|62WzsUF@!G0v`S_Lo+WSk=VA#>MP>uo(RGT%g61$-4V{EPF==a@W##wOf
zX^8P=Xjz@AeiQnXA+2I$_nALPdufs4Kk1RMko-2&@%gm}E3bL5)CL-}MA;!`c}?Zx
zJ%<tT5XRWOZS<e8=DLBJXWmjf2O7<x0}V%_x^3!n5;gOEuYzZ+q~IOt089wGWH+&K
zL5PO&jh17a&nu^Cpo{dXSBxN@kc1z=U0W{+3G8tt>8F)?$aX-^BD5#H!tl_z(IG?g
z6I^0%UGHtz&hnvld~-90hXE~x$pZB}rrv3q#}>3D#l5}N*I8ga>^gD8?bCram}Wu^
zQCB_wZJB*Z1>RYw&ky(1<W_lyR*3`wxR_Gsh*Dnu3AXVf6?DZpf^68!&~kR^ytH&1
zbi83ce8Ju=RQ^|zxrm#SURCzQ$9|TyB}|^cPSI}7ZZ43jdPs;%Z%ULIl$nsuTKzXp
zCMs5JpZ#W(2^3($U?*jhspoeCdK-V-&tIT+7{{P&H_)4QtNu7YQ%MO~fF5uC&12}&
z4bQdIKH0jKSO$z<3rJkzjSqu@#c1iis*UG(ywAd5kd8sDwK&HuA=;K>Ucx)9<11ej
z8LO>b8la4<Kjow+xa%E2hApLNsSh%?)b8z<Xp6`#^aaqRrh1G%F=oCgl30aQvD-R=
zDWtOO%f)Urn4HH93!O=N?ViM|g4|RhrAH#H`D5(!S9r<gV8qKUPeS=Ikrp{am+0rR
zw^RnkUrC^Km>P77Y$RH@<YT4ybB4SJ2NC(ka%2lEje&n`C4(6+3J47t9{5t}rpJ*$
zv>hz02j~ZIkp9A5NQR4|AL}+QFFL}uBYDrZW4DmihA}Z{q>{u+vK-+!7QT@tiT`yB
zB*IM{@Aw+)`FZ+Kb14OlcX3j5D-IZ%zR!yvwx72Cs&%@(R;;ma^y*FAn{F@=w8bzc
z5>4en=eiy2bCSl3STbaQW?@Tl)t4q~ty$@G1mu4CZI-9Y6Q@T%swf0s-?hjxM^_U9
zoa&4RWFK!Bk0_l;4!)={n6LiN;O+c+U@|r`?BX9sY^>5s&u}4364)s9%@xQM=rIVr
z&2EV~r-mGX)K^iYA1J|&e?e0s@6XOWjAj8zaJya@g9;pH82)w#5OUk*0;>&h4KZ9m
z>QCjw4a;yIG|0+Cd-^!z3-a(s>tIPy<Y8~O1~-0+F=8tuR)-nbM611rSZGE)UeveW
z$?Z#ciN7^THL6y>ns8bXt6CA9#oztdatL*f>Z;X#pu)6UL+%lAZmV0X{kSVyOBxo`
z$efn=ezLr-Sz6Ppq2V`xIhY*YoQ$gg+{pzZ6R;nzPAEaj)9=>;zoq-8^PqD!_q*y$
zEKiSWhK{P{Oi`-1ewFj^)t%=%$G9$bYoI`G%GS-gd$Qy98RS*_oGVr+|9U_O`Urb3
zPqkpWT~KZ_YWTK0L2lOmL%2=a1yxQyv+@Oe*7!Kf4r7(MR_9q^PcY{Q1eiLeR}0qn
zsC9wkUKpy_ZdZB9Ce>x0Rp{+LgPPZ!cjT58i5EC{TtB#$IWFVh_kj8Y-(22BrR_et
zPL9;E_qHX~hFg&JbCSwN{}<nOfenrp{H1IQS-hFfL=Yz8s-aT4cjOkMwQ!1Fk0wWz
zsmNoPJ6id@c~bxoPND=9WU#ViL1nK(C(0sl(!+-8vaOZdzg1177Y>zdqC!+b3~^c|
zh^%wb&B?YU#9JS9&OAPq9M;Vp$BuF(^xCfM>rdC(cZV0v5Yq3<ixh?+pl^H$`tq&d
z($UMy!KIBY`G;T-e>W<mF+<M}iDrOciJ0oi6HRtX0jv}d?;e?pF`stz#N&bgqG*f>
zCNCsJ$i`=0y<*B2YE%HFT#*I+<&p9?nT+bbiX%Pemq0!WR=N~c?~IF@2<tqD*@3ko
zmfav8$IPJ<i_PioWex@^gk(8mJV<sE3uA7s1_eGF)c|216OFvl4(?+T2y|cSZc(Yk
zd%CwWS$+UOarHj<2g{;C{EKCeaB>b~q&$T8kNz9Wf<6y{7TgLkZBWCCP=O$l=&$aK
zDgIwnR&q#tCEyu8nGV2e7cNm%Fvpd<h@<47MN@y62r=H`q`Uy-myE>a8KL@o@b`yS
z_Er5V9pQElsKgCpuxVhImMbod3kw4QoyK9WTBc=7Rba~t@#{{$d&qUYJ<RFKuiizA
z80C{W|JL-skZep{R<KWMHrhVf@LFhh?hpL7p_4hycFjuGSIesVIM6a0Sc{bH>Neub
zxb3OM-4!+Gv%Vocu*`OKy(K^p^?<=ws>N$VjNK)eN>1)Bq%Rq>h934=+o$+sqdUi~
z`p>fCLf5IUuQkEsLRwHzT8qWz{fBxpC)qnYXFq-BPL9FO!$6SK<WKznNn_{N=(MS>
zq<D2{sFZ-(MeiyHrdX%29}6*@8k4hls%$$qJVEpO7kSJ7QDGn1xk47xUm+F_f7e#N
zH-gIf0F@!bjaQF6l`>nWe-iccc>@az2SZi&0rN2F{VW(F#XB4leu^wdSVmBmCV00!
z0P?@Vs^8lhVSrdkbfuPdh!w!?UTFPU{Ly{SK0=H0pyCvv>=a(_Gt<sr+yVZ?1gEQk
zKu5tE6ta1Iw4SSl9EJ9?<qO|7*-uFrrQmLnO9tu!$q%+#$e14ji*jbtC9*Wr7pYL3
z7@F3C>T?I4dcm1=K6oTUlgT~HdaHfVEWyO%%=F^7wypjG+Z)Z;3lw1D_pYCW#!<g`
zS*B>?M=?Xu)2essxH}KE5mt1L34UIe{#di&$tH26&o*xHWABIu+VWxT<PG<pvdA;a
zonAGSWr(ij=GTthE`QSFNZ&%t`I6Yym*zgbPpvG3eolusnR=x%YS6PzMTI+aibZFP
z!_x5wX`Oe6{?g~(wg7~eSn_KsYro=W6jG2OY92TkmxEvYmO)Dg<nN`C{Uat6efsfq
zGXiqdb&_O?W$MKZ!`?zvG%*NPh+3*&+-PF%q+Lk;(zH}_-JhSJ%|qp1L0%<ZC35!L
zu6)kU2&d+}TFQAmTRJRq2RxL;fy#|_R)ynzRY8~AO6k)tJb;BkS*5>j9O68gTV3sJ
zUr=@!KqjQ#@k`p(r)|k8Qocru>Okn~;YW@(kE?nLvwxr_^E5YYd(cUcrS25jl1J6a
zH&5!3jRnW6?~_puiEA*w=j%(%(H%m+D3S!muc=v`d9HZg(Z_~om=pxWzu6Y(*n&Oo
z-fS0Aj`j{J(gQYmH)^v`a@)ad3btD|cvvUk<L7UI*-i4gHNPK{#3+|qpn_xw7f)o2
z1s!+DXe>ElWJ5BD{02r%uaB@e+^)^dy+nernwm%PPNr-^yum+z)mJVQj05*!Xr3Qi
ze%)f2()I%FL0fN^$&oMV`{CcM`I~daVUdB$iGR0Uh8ej3Az&IeGPqPB(`g-kHIWu1
z&Oxx=YDS|Lb$<$68rI`&`?d!e-SEv#P5ba7qWf7UgLl@M95r*1^K5`Xe$2M0#3i!Y
zOH;edsp_%cXxA-*(7DVZ7^;0cE9^hk8q{VR(izfWTJ%^=D=b^Blz?0pB~J1MJ?|yf
zKXVPV<^wEvRg!{O^4+C=yW6wHm_THsfl@&XP(jo@hA2Yh!{HUdarEF6e3Ag+kOd)f
z<$-)NJEZU;G<5is#Ol9&P~gAf<uE;Gr%I-AaQ&<FcpvwS#Dl{*|7UKUafK#&<v>wT
zn1b_QN*u$+Pt<gt&e3HQB?U=sRi4sIzL7(d>``FA=E+kLyUn58Dp|@vx2|4M(i4jN
zx_s@Pjsbb*D9cTIZsJV>E%?2AQAMm~?D|{@{Q26k1kLE(0=+I9Y-iVWbhOYP*MNxU
zD=HEtf$2hF!N2{E$D4tg)vJ`5u#ossRbrt?9v_WzU*v1jdmV+SzxwGJ^qGGrUUmHm
zR0Yx~e;~ys*K!B{F&j3u(K!;c3>pryPf@5Mi?BDgTux4QcfI-k>Dp!L{1&Oak>vIm
zq~!m>iNfezTv`^j*TcmNGX9RmODRaLu~)H*6-kCu#|ldAleR4^jO7TmTi&lXsS~Ob
ziCp&ABTIMYr-zitmq`>vLRp1C&qrhANwlMtbVwPsN1HmW)NC~I?2)5N?k(5opT+Og
zpOW(eMaXLq3C$St>F<SksbvWpRwqc`zUB4EUNQ3HzIxDF@_t!sTTJCg4u&6uAgGI7
zm+t>!eo@L>+E$82!Ziyu%?3(h8qsBdzD{byrA`4~ijBSRxcQ|e1$<WmAoE;DBD}E3
z;!FEq7!g*<I|G^g<S$38h%{bMjF!qtrh&~ggkn$FY~eGIuNHy;N$`m7iS8}074s+e
zlN|g+h~Ws0oTMVSBfvBbr)GAi;39;aj<3g6UgS5eoFXNn?EjUF-87#Uv84QZ)~ti9
z0+4VU(@sr(qVX6!87?T7srh3!=uJWmPMf=!y%qLw4xOXNq62pg55M%jG~p#<hd6JO
zk8F*ur4%d5o>y(AFsD}*LRUp}XU<a7y%3Q?-;-a}Td7HSFSjIz(}L+Jim0iUy85zs
z-%M^L&aK2<`Bl0VN@5wCpOx#F1Oev}4mhM0j!T$h+y1Jq2>q_+o<F_L2#XG4_F^%r
zm14P}wZxyRL%C0K?2Q8cDu1@3S*{<jL&dN<==mQ19wHn4)5BH*wHe2rj?qD0+Y7|-
z3-3C8HV9MT{f#6c+SJPeo|DxGuiyz8`C^&(>VTaLlL{Ykkwjj0KJBPZ_gl1K1YibQ
zZh{-Tq~=T?n3zgsmO3PeopsGolnznIrsx!F?E16&)g{&`CM^&MPw{0l<%%0(yt%Gq
zdTt2!9-x#JoUj|KFim{+(6%iCS+lLYwmC@ClJYiOABGnN+d<Y>OL_0np#`<u5~vA2
zg`j_>7FT=__4r-jFmQ4@?aEqs4mdMms5{DdEGhLG=-<&R3griIpnQ%6>13l|J(hwM
zFCn28!H>GVzsV(H`YH2!*K3yTHk>7Ttcm3Uc9bN!o`0h1{2p6o%X5(`To6>SCMs-<
z`}Z}5f7@oPy29C>Kle%cZ^@;~c%f*P>WS@q0lpJhtuK=|Wcf%E#6D5bR)AsNM0f5R
z`z<zma@1va2h=u5M)9flxYihb7gx`5h8OrBil;E+mp;{rm&Ku3(ZKy7xC=C4C`g=*
zvZghe943A80bee?)d^m%PjCY+`8%Lm8?~&ALMMTDci%}D4}3RS$@pBC>(AzBFplS0
z_g_6;?<57S_H5+XtqMXkPk{LW{&>hA%iwn)uRHsfh>k8%sZwQcjwAFJ=R8`Spla-1
z=(EyGI=9{?QY&wXlhRub^RD{HDE)`u#>|A2eFtbHaA*K>nVZ&uC3(fsspe~q?yYg*
z35{;;?a3GfSaJjP1~|*}p%zYe4$!=lY61|p@AIf2A&;KbT9$O#t3VSa0_K5SW+hhu
zFFcOyR`Ca@k~UG&K=XDBVFkU^?FqAOrbRnHN}hN+yhrY;TvT-GZAHlCtC468JH(yc
z(9$9B$&Go;<5ki(332BAe*noqHovLaDi@k1@ecV$XmiAPVVmo4U@@!{e$LTGi3A~$
zFjAQbfkK(Bxi1^nMa1Z^Ul{m*YHtQ;Bnz4OG!#QEPC<;E<#H%Q>@4>%_|5BMH?PX2
ztEi5gMnMnkWbu7b8@}NErr@XSz=Cbkm9*AMr9x34S~KWBN#b2V2<XRA3K2|-(Z4c4
zbQZw9<P^g;i}j`=Frs;uxR*|iRBCCzSY((|ElYi0oI@s*77CuV3oI*tHA~H_MTe4!
zYAaWZh>!6lI8Io+Juoed+NEqRo8xtWZz@46f$hH__C;Y9WCLK&Delv$(C?c6ylNGz
z)m$iEN}MChgo{8T&<}w_)IH-FX0w~^daX6CfLqpOu!lHz{C0_Q%U5BO{t}fk|F`bU
zwGA+F1n>mvM4dAaup}gZ9s?4ea<vPM7$=F##I1G5htx!^$R~TsU?wIvVyX_z2^~*p
zmRk&)_n%P8>-<aRCye=a&-v_B+CR<daLM^3bDmkTv<918`HMLaEXNelP!wmZVBxXO
zuJY<wSJ_R^PVmzdCAdzYf>twn7FPK&Nn<kVvUg@T@`n<Xo}Q0?bc@vCYtBgvILqOV
zXkd#v=O6HMry4;cOowJ4^;84y?P7#~QK&HP!=~a0q8a`YDxCL#vR)Yc(4Tt{oOB{X
zZ2(tI!={FxgmqsFD&4!<vs2LSQW~!UFP3tA=Sqt4Az1I;%-8pP>{Nv20zUq({X9)U
z-KbmIbea0-z;<bW4h|eg@eab3?2TVd@M>~5the<;fcSen0&l_QCcv-*`{GsZJfwma
znK_CVE&iU=f3ePcL#)mpExM2MI1z5IX!S0iO^p}z!jZUoqiFYS6ruT^*i9ivDGNi+
z4h60YcE@9kqM<vQEOWpPmw%RiSuS_p_mOVh6izYhP(-1B0$lr??u)C&7^;JWTu`n|
zeQrEXr<jD=c~;4x1LBalW`9FF84W$r*gwA{>s*4BTRO<fLJJQ&ZjIN}#Tb)@4VXDo
zCk(>lPEbkq#>-x9kalLpyAyb0E~1tx$wM$%6RPRt;-aUH>BPcLUeZN|9qgQ<fu`e2
zX9r`f(jg0fcg~ncW#7y`SI1*Gg%dpP0nkuCFEuMs^>>YJy?$Fs<N0~6aV}7}{{=cq
zzF8Xe9+?qJk;NQwOFI>_8v<EyX_=&DvMm!>q$4#*%=SoiE!DM*8s$OE{aRz#lh#P~
z75AuVx%Cac+a$kTq7*dMnjT|F+l3B3^{#IVZ4#@0rE*35jWZTn=Wm5o)&WHQ2idIW
z%^}_7fy^ExTk`jwFOrq6hamh+<bAV4Zh@=x#)Xn<S+0<LcTKA4<5bggsS>H3rFNFu
z`QWtk*6ekf!_@M{#;g!p_Q(~9=WYT8QYEW2ZvMQKyAny`c2$BeE=Fd5dGLWSMED!i
zS)u-abZAiuX2--sX4qmANao3GCKEwve$L=@-ON^76A4bZJs4}CLQV@tv)S1#yDx{Z
zJpMAv(H#R1r@f@(b(!S$QoKe+R;?Wy$9>BHR%^6oxm0S^BhqM5&?v6M^u^5uCyXQD
z@`=loHN#{?<A%S*msLPZnJvMEPWo<%w&b#Zk~8@{J_2ve-x|rnRB4i=NfOQ^ZOvXA
z$GLi?m7fiDlRGdX@qF10S<+Juf5qTJpiM$2$fQH|yUp&uagU<(Ly7m&)(k>bqlZt?
z+feHbQQzd<G<EF-BR8Ixick7!UYU@7+WeYGKkbP9G<odg;QchoUV+$4@Hjir3{`i3
zS~j`OlA&`GA1lmmfT6*2gWprT4%lf`%SEQNMZ<}7-Py;4wxNN$h5O&sXWX!``oh0R
zzi^wDp5kcHx<yE0v|Ai85)!gZZ(euIckx&IBaYBULw%g_7|b$PL7l~G(Bz2PB>OAC
zP03x!-Gk+B6ialt=6oBfr(@?!=UhX7@3a)-2l}pK_kA8UYT-i8c_ZhRu4%rERO{t|
zsupE;YJCPIK6DluJzl0dk7lo0tQ4!|l8Y{m^?*uY=g*qz;6}cnoZD8L`2!Q9^NeX{
z<F7aj?#@=<rD-aNBTNW$1h;+U0{5;Pi_cAG(8=OMxA=bGqWQXW0k>P2PAp@8W&WvL
z&lOsB<!-=3uw|Ln*_&z^Om&>u$zb1iaF>$6P$Shq_!m5cXr_VEbf{98qGXRy8k-iy
z{Y8Jx#h1brmxC?`f$L;ID#uOF{p9nxTsY@`61#4>QZvubpUNfTFX=_lvMxOIDAKVS
z0h(SILIMIpJKe>I>`|s_y+yQtA>iA-IRM`>C_DrR*xON(@kht_7huUH`hWJWggJ2}
zOaGN@+&v{?Y0PDOR1_Uxu*b1ob`5rIO++6KAtj(KAu*D$%iCT5ee<Ob5J-SA5~O$&
z6I~z#R_6QOePoWCXgP$y0YYF%WtIJ9Hv)<XI6wsOLuZSI>31(82o4Z`1P=Y?FaX?s
z{`FU+f2Qf*pvIF&={4vqWRwK+)E5h90wzt~G}%}<`X-RwZ5)y2&Bt^}oQ+Zwe=FQ*
z)JdmYtq5Ob9Uavd1IUzpG?5)jWP?|QTN56LB2Z4_P>A?D&y@X}%Mm#C%gaePLrVIZ
zWjl~8fH&NSEx_m$FVz5lFQLX<l`WM>2Kz;R3xQYOEqM~89JD^6Pi(F9n`VcLkddvj
zB{}0bXkr6sWNR=PQNvl0RxH<RXH|g))+~5=;>wGHF9N`X1-@mA^4QpKCedu7n8p(|
zG@ztr5ZfG4qR93-=LOcUYW2%j6l;e+;jHal;y;2$(H<B@Uz**2+nd|#dlcbswT#Zr
ztF<eE>Kz&J?+pEA#xn4v0*C`wWYF-@yOFRstDn|Nyb)|37U54&vkRlyoBFf-VlxF#
zT}#n;nyUHdA&oC^@fg9SyEu%nlZ15=3z71`yNqsd`Y$604bs$W=<(WV^}NL=Z!8M6
z2<{Hwj@&H_cRzrCy!T*1RFnB~&`8Ygcmn?Bb}tzA%uwX77cm7gVb<|<sO?a-NFL1K
zSD^f?zy-*1QQ?6|91e5nXl!~VAakeOx!!V69WpCXcmpdWj^Vm8q{>(S$~*a8f$-Cp
z*qafRjRG;~pvWFzLgA8^Vd9kV+b(ixoxz?OD2p7Fo||HS_7}E~Ly+x}HZV909n*_V
z@N>=CSNZ<1kAm=oY=Qs6>7$a~OV(gG2R(5$R-ThEOj^!kMRHF)^W&R58gFuPeJEh*
zU&NUG9QH*E=z@LPz5e{MbNldmr_~Ea)oPvc^dfq)bCB@%M;J(CFUB7=$~22Kx)7K~
z!Qg$uOkuEpn87Lm!%d!>)PaHt1yfc86I8&WU=m9(6-h_qU1VO}G6*W>U3o5)?=taJ
zaL!Bg2KV^6;qYDFyWp;AvKXxSnS^ohJrsy25Hl$d-Fk2oiTh0@-5h9&tQSd>&+_n!
z_%*jI#?N*6!s-_kRw%48DXjdW78F;4h^tRL?k@*_nR#1&;hrXR_fN}sm;`e>{84hD
z<cjAAktG-O+o04EEFdPUDzmB}qr<E$H&qpgEC~|D;c7&;me?lXLk8uSn(Zb%i{!>&
zKdwE=kmqCH^KCdgqCOKPA4<N=l=~Y`DERhVIWh4DBe{Y(OiYY{{9~lkEYpCxx#CRq
zwZrUxwf{^sYn|#!OlcoRO1s;6_}snS13X~Q+V1g?LGztSNZ-mct%yz(GH>*)Y1Fk_
z)oLrU`6T=ahrV|y4*J-9lD^dY*XNsVr%gUJFaNuF-C3icXO7@5gI`rkm(AvOP=Hm+
zX`{d}jWr4;A_*U_`H=kuV*jhpx0eq$cenR{<f?o3kz9U$=sf;<^YHL%=dt_ff4Bz_
zLE~umxXGSzG;Ipl^n4U_2(yK82ItPdn;$>@x{gIOhY0lk=7`@eSL;_8!rZZdThd<{
zCRhEmTs>(E#H&Gb*R!T8=C>lW!!2~=%&oCrDRtZlZ(^)bk;;W~p$g}Du~(#0AT{=X
zdMrZ_(`!r17FDB}FrO=s`5CY?gF+%kf(pf_tCOtRAS00P=9V-!IFAmst2lUGSCNJx
zesx|xsfjekBd*sN)lqhcyjmUkh5HAlkg`F@Hz=f#kK<oNErnVNwG?Wpv{yW<ki^wG
zYlSJYp?J_-Kyv}jg$0;k$QVP$sgo;z(j>x&<Do|!QDV<*FiPybW#t1&@`+0zdoq&B
z3zH+orc10>gE|v+ChAPonW!`Oj9k7VUB1>@`LtPWinG>+KN3rNClBS}yZnNkkVRUe
znng8>YSw!=RI_`_%BW^h&5E(UgGrxeCeVNu4_Z7+EuPL<x!rP4zC?=$EuKApw0J}#
zS3}_QZsO7iVTuT6w+sej-E=A!7w!w+`h<UqeM=ru>EG;a`J!BkTlk!)6j*L7ng9`N
z4AWb*g*V&*e|FQpk`>3a;H0vH=>aQ{kvty5e6J5Az`aau=^MN7?|h#8*9xyouzk>F
z+v!q6r505`F=f)SsQi_GahC6YsX2NaYPMqz7`x9kvituL0@^D?b2e<7co`+A0z$WW
zPr0ceEJb3R$Mt5YXG2Ux1Q5l6-fDv+tFl}i8m$gFAtUWxhTq%<6S_z_WA;%4)1GhI
zLfA0xA+yN1tNJiIXO+{-x<GKZsZ+EmRi$#CyeDSQE_+9o&%|mTv=L>05=)UR!?Z?H
zxx{?l;3ac!EA0KXWlqLrc9%uRk|O1<a|M$7XO4YBWn~D#o>FYBDQw*NX|>KCT5xzQ
zkM($KoJY(d#Cn9|3vnYBF&4*rf=Ph=P6E8+1R6Z{%x@4yTTY*Bm!*h9JJ6ViKq-;E
z$`^<Pf)T@P5=C|>MCx;YuECPazsr^!#Q3ftE5P`!5S)YYT~VNi@!kEW5yp1~AQHxR
zlaB9Zai+d-KsX}pPV4ja_08?|9zX&^{=0`n_UeAuD=!s}b#X{!W)-qX>=n}lv?A7$
z4XOBkRYd8$RIYRcg5mt&?2Mt3y!TH?cZCRhe_)Xfeu5}j#H~1g5IrF88BT=F9tTq6
zDQ0b%B;yh3n--*<6-P4-qSD@E0Nzb#@yBbpTj*J35+DM5R?!Rc-7R;t9vXsZ2%;gl
zf3^i0f@lctm;(>Y5JXGxt$Pc5<2|I{V89A`te{^453Hci3SVTfg5EFGUQVg%X}x?U
zQqCz{cg}0FK@Nd`C@YYq#j_bB@jLX`(PQ67!Gj(<dhDVQ3q5x9*s)YDdc$GEyoVGV
zA*<Z@Y;?UH=+tb%sYc!(0ygIk+dvwMxXO%CZQf7G77ZJ7!`(B-as=EI5Gi~d`r%o<
zp+z2}ahrYPmf5lE1$E1vYUy00psqKHS6CU<STQ_2C&zYwwO}?Z^Vu-r*m0dP;n|y>
zjcJw)rdh;QDeKK#5sfaG@5+?iz<if@`7SJQeCrOw-gplw4$9GaMCWmTWfMA&Sz$+X
z9?^LWuM%tY_KRNVv!Ks%I6h0ea&lI_oU<M3m6K|x9bBAoWWu*|N5;Fve>t;`OyvHE
z!w-RXv+8tz+5{KoE=cp1x21*A(TK{1eR1Kvkj&oyQUwy<>_X8R6o@^h3fPhgTZbjv
zPVq^$Vw<l7B`@UneXZ;TM&VMN8k-UEUX=hBBC|CD`&=}NBRBT|az|5a7|a#l5EkXu
z$$7gj94tcHnj|c|{Sjjou~v$Ylp-L->!?k~woG7uiH$K+BDAz(Sfyx$Yshv{*K9``
z$&NBCDzdIi^xIfA)S?=HehsX&S*ez<P6S#h7z0-sqaJAUeEzrJ%76SJE@Kr)PRh?U
zqOPM#(3nVw$jR6Zlx_buY=_!ey>nJU6o`oei~eWA2O$SgP=-|bntGl!8&;zj4-9nG
z`YD}%?)UbGzi!PxyiSdQGn6KdrrQ@6eFmfzX5eYN)@mo^)+MpRl!Ez5JUWcN;4&m@
zBq@-Qslv>moG&ZWn%0rTa+VH0Iy<eM)XU-!CoU@gfs|5eoPYbGB68G!t=7S3<<n+8
zGDQ&nWY21ywzW;^=-#bm!$xD1`Sgw#d+{QF7)$GBW=OWzi;L#KPPcp4B{#Q(3C~dU
ziTRipq}%z|=bLV)y|kz%`3*=)|1H*u2(+$s6Q9AUW;gfb<IVm3&FwYN`{AD*(j}eS
z|GDYj-G1!cKBOpxx0TF!Qv_uHrLS-v(6-`^jyaZm(fhmcyYXS}e{hY-t3X;aSrvPK
z8yqSm>w3{o<t=6TPP2Sg>jY@=>I2=2iR@bng%3wyWN8NByP7ACNqUsXJ)Oc!SyhQm
zpQt74#LI`;>~rUtQiFZ()%+Fm93q$w!w&my(th8N<``s5hvvzJnztBe^pZoBO5LR-
zLL#hdLi-~OPSzcYCHR62lNgk-0aK5E6Ppt6I>VxAsWw-dPo}1l-eh1Gn66Z0xJz@0
zW)t0Hwnl--47qV;hG$b;o^(!1qLZG`Z<_7cQqHx+{4(bPGdG6R(x6@{+U62w-Hign
z0yB>|5}Y9=wqlMquHoraFdMr}9=iJ@4mgg8Yz-zOYB+CIm1otHR`ndA5eE%_Z-0b=
z#%A*)g=2u_@b~{Cu!4KmoFUl5k@3Ry*}GOL*Dt5N1425)im97`SH6X%Qme%5m(2+f
zj>^XJy|;0KFE18fjDLz8Szf+@7nCoT5sB_dh_SZ2BN5sYnF*O`i_8Qa8put^O~_5#
z;3m#UR4syvf_$)jkW97E+tr(Y0JJY49-xxm3$Y%9hObA*G<6%<5ZQ1eHf&J8<A;6&
zHnbVyq<I5TnO^bopFfT*v!~Nhp1_4U4X8-dOaA?E1#XUGa79}9#QSO-*TZg{pPh6r
zlXB2gCBu(yV=Q0!4&_{`YtCo4mn5{0hmq7>q%UV*kH4qUjCpj6K>&Dv=9vLkOpxia
z4XtW9Whh&+;!rDBAZ`IVyISUTZEi#5hj57-99>G;ACV|U{3HIw!#_+iC=hcDh~|?U
zB=205!gS_PVLIp_k8FW#aR6-LPw$vf4Z1nd6j?{bfE;iL9N^BgLHr~BSKvkDCexUd
z?GHq=i^fFPdF(@&{=Y?k+N{%DwoTr{_6F2FThKkgb?B5M{#WAV^_>PtUV+G)Y14ul
zPlsh$co~msy>iN}1fk50P|g_5jZB#vj1y<94fam6Oap3C-G=s6Hpcw=Bk?W<DmiD9
z*Vh4|Ruu;I4CYRk8Y;D@`bi$pmPO^S{EM46Zl1MEr>EjIZiY60H_1An8JI+FoZ!og
zG1aoM)I}{Q8!yumnmn||uRb~_JA~fY@Y|-4*p=ajLgG!>nVrh9Yko<1t<mn^vQZ=l
zB0!=yhJ#|J((t!H!2ZEZb5qylvCXIQx#L(e4xS}AIT$sw?#!JO7M_|4hz}cW+M0ou
z`N1pQjh#B|Ctr?#kc{?|3%(#2L$q91V2k{ok!)Kn^7U5v>|DGCsCe_@v{WrM6KQ}F
zd?f&nBVw}$P^gr<yg2^MpkyCk%TzyZn+oR^`#4Sv0QsA>T6G5oV*>kdtvQcRZ1qsU
z{pgsF6Pqqwd98EOJV`}wCB5>ww#mo7X+1(S<<rkk<>Kjo@j_oEzg+At*AAW#5r2DP
zA`n61YAjo}>vA(6Yx$mBp0dYFMPnun8QZZQd$O%5k4&k$>IMLF(2&UIcVPqWWT`8U
z2hTR851a3&91r&9WA-q=?x`%<A?pFj?)e7mOdvMKA3SIwy_1Mc(%_X9M|+YT7=Q{h
zZZ(4vm~iWVk3<c|qewgyV3@&FOm;JK$cb$8sWE)^ygQv@dV}rA&cx=wr}g3IUdT{>
zqNE1{=)FM{>@P7lU<)yZtCLE*hBL}x!@P$S<AL3X0R#uPHE*?4tzVV}k2W1H5=OGd
zgoCMWI*6H3|Ehh4)Q`)Kc>9ygDne`75+ufIU+QarIusdjO|#qvNpyMX1>TQ{4I*)@
zo9xjw0|wK()7fArc3_5ViIxShAQ(aUGKgYPW_hTuW^^oSPKe|Qux~WJpBYKa=c3W`
zdKPxb(@BUd=KL0wiM4?(I|J>`W$mgGoL7a=-aj&H#*=1iJ)KHGaGN?(fa?v0uDMu0
zFV|~-)yNY?(R_g=S8A2Zwn$w&M+6!>h&Qr{D6(Og4jj|wP8s!%n6W2gJ{R(fdz&rg
ziE2`tz`Y|=)%q{|XZ&MW2p6TyRbsX>xamEa{RR!%ql7gOHVQjkb}G$FfzI1$NS@(C
zWQQ3-)(korIb~wdXVPW%AD<Gs!rT>(9SQ`0!dU^zvxd|oh=ajfhx4UR*P&)Rru9ND
zRzTU4{(|qq#L!LUYm4b3wst8uE4Xq6E~}{VB<&`<ndG`xB{4;#ZbN1wXkaW*5FF70
zH0Np-bIPame6`7QIF4M(^<LY4V4$lxamEw>@N|H{tI6$obCr6f)DjK@Z|ARA3`FjK
zk2t_^4+!MLNj-Ep){Ep@v>&~bhl2}&iMv?z0B!&Isbtam((Fu*{9o*5X5AM1mOP?T
z{yWue`WO4;dp^&;WotTE#BS5)swiQSd%SI=EVDKS%+PRm-Wvo95w@v4IlaE=*#&jB
zQ60hiRFdRYZ-2z$R>oA&C(Wm7NHeT|n<qbM#!oP2c)$k>4+ao?&N0^7w6A!Z&=G2W
zT~?v9si1eERJpitUp)3Eny!Yhw<y1VEB}EK3t<Nc*tAa$fsidAL>oa&pM_7V_NU~}
zpHuD61Ihojs*EkkeM!JLY#hz6)jP!;B2Z(~awsJ_+apER3Djaa!->s5Qz^550xa+J
z^9`Bow07ERZK>EssHt%owh!moLzst;9qlMH7t*zOcgXrlP-{lyWsGWEUN5JXCws)9
zR*u=rKfC%#SZXy^gjR%Op*jCHU6iF6D@q4)sjWz*%%0&@>AWxwESvipaS!;ji{wsM
zc^T=dlv}M?J2`I(Pu?GhG*xncp|B1Uqcd_RPyQi~#_tKlfCd&F>(Vzq*FE?`fsAFx
zp_Tzdo&M~=rN95=pZxu2r`x^jLP#iMP$IpFL%g##T}VYCX=U~{8<K(<fRa1RKa8V)
z44n6*<iY2MHv)+KO(OpY18XTyB3v|KHnkDL(vTXndqXsv$V6xEDu<kZtg00JF?LmD
z{u+<mo7UHhgHT9+hwSnNku^PAV&kbJ+l&8HDs|pcfbpL_adzIGhQ{nf#?XObVg!N<
z1YQ?_Vgb>x*GdCr#U7yJz?252J>r1Cyk0nQo4X!vq$Kckr$<iC6fCR=8j6-fTrX&d
zYINn!uGz-4=ll`T>(-Zl+=GXwY4qJgLoByEtI(K{i}O`mqAhM+xwuZ(vS@USUh2S}
z^mv3fTXbEPI2W89GslCCh>QZ!yqp5O;8_K7lPeI&M=${MDMIYtn?s8ozuhv-s*(zI
zFL@^&<y_!=?A+fsuREmM`E=KP=(M4Bun#e$p!;N=pTt=<G}at{f||+~NDjIdNYo)j
zNUj|&i4o<3AEOmE6L(ofv~4N5#DCts`a$=3AGly&auR5a?42>PH|{fvMHGu^i^V+v
zLN=zRi4FwNkI?WRBw=<a2;aWIHRFh(L{n6n$@@s?y;vmn*wPHAXqfwyN<e+E56Fm&
z(%jw=4Afwt2FaCw?2`5%RF1aTJ^&$Xi~VqjwwM%G_Jal)8f1qg3*rp2sp%CsY}5IW
zxi8J`?al2qyu5#C-nN_FHj!@cZaq7WT_Nz!yDbl<KrXQETAj=0=lc%1yF#-M%|0X-
zntdqhQ#JeIq^eNsJEV%QO-3|gW7FQMoki&OCx$<MyQs*2V`pMf&m{BDXO!r-v9+S6
z;;K_?)oTKAY`UlxD>a@ns<-{Xv4-Li>09P#g@fH{l{&Sngpi!TQX48AnG#n$jDH<v
z$ncf987j`pn5zDPG+m;YhHdJ!z+Q}C6g4O7FOKzc2^)Z%eSn;8Q%YprHdk<WyIiVQ
z1e$Ns7&^{>_~Q6@B>$g0f+wVgLW>Oc$6rhf&awTXIm5Q>$j85oKb)5<uf~2zeIrA}
zI9!REH`|8|oT<{KhDt4}ev${YWl{Mn|6(lvSd>|Qa6*)@s-qE1^&z%nfm$%F&HC|i
znomMh-}VD9!87<QYqipySL&y=q)vECV8M-NO;^l+Z$*dNPEmD{Dk)OAK-mAQd1e#A
z>q>#t^3dC)!3^f8b4xO>91`S?$o&xqL0{fJtOPNUF-$xc+5W3K^@U~-{CQqAjzTn(
zI(8gIaG;{Q5gT$w?Xr4Rt8a%h&QJO6XiD#1TDn~EJkJWkEMue7Z#^uFfw-l?0XUk%
z<k2I4xo196I0mjjv?47Q_(hF9wn~ZF=kL-XauP03L@@slC>&_34Q00H1NawFW{WPf
zHRH+rO1sQOk}XHdjVArRPQ5e3=H@mmEH+VxFim@ANG(bNyG|14NaRI&x&#cLGnrMw
z$+|v!zPDW@_mmPZ4s4ktSu^^UY$&E?7kj#YIVkomO4)yonHZwqisRw<m$7L(#}&3C
z$E6eY?{QnR<(^J|w#QV_`kL~y|J%YMA}&v45*I4OUuv+81P*cF#i+z&+iPulXeoNW
z8PgT7?-NCui%#=U=`kFc5H3htUmBFlrB0=}9b}wUPtrmLT8^*ME9~|OT{)+&*{WTC
zUY)10R>jm!_~U#TSEuK7H{it7+K67lXCHF1qEN%B`#Z4}u30#ZclmyZ7X#It;UzP_
z82A0tHJWY7FDl_haOs#!slmG79q;~OrqnP07yX_u1hLQ$QwIXhQC?1PPZ%b%D@nK&
zN2JyH((HCFrmJ*M(ed|#JX@OMP=g<T+{Bayk=Q2a+E<7exn^W=tL+}{boKd`$kt#2
zV**j9pibd(C19p6*<zsB(8Jg=6{aMpN_sD|A&3n*0)4KGaRdSZfxv;_mYs=vDm2Wl
zU$3W0P#;-b%KYf5-F*HpD0FeBB+7O%C=*12?I*!9MwXPCF_L)n@2}U>*!w4cI$k%O
zn|N8OU!ArQ2ysC0_D38bxJ}I<l&3X;^VKCQbZF*7qLh6kjb+QG63lO4F1{)|a=ris
zMjAxcX6!?DiP<9TX7&`tz9o;Sly|KX_Q{;66(2pDx_E+$k<_AJi2@;+AS=j>qB-VK
z(xPT)o;dNYNO^AfrDG+f7X({>HONaeIz}(kry<u8A4oaR*1e!NCXS}t7Z;w&D7t=%
z^cTO+yJaD@v_-+1|6bq?#BgfQ$o>5sJm=iA3r()Y!5#BVmn}3l^5lp(L(L`@Rm>;$
z=Yl#P!*qy8X1C}EF7QN`O?nfDs3vu9dN!@EkjH_7D~>7?k~W*x1s%$Nc)FDeYQIL@
zEE=>PHkQk#4ME`9yW0T4&f?QMu%u(LlzIPQsSflnc!kT`Rf-38dflr9$EJ`KZgv^N
zjBLC}iZ0u>JCHU{e$tGeh$7_%eB|a}FxE{cm)~R-*!Dq_Vo3vnW*xcZ!lQC=;l6n6
zO*CEgkc2`8@xp`3Z#koXY@1fquA~QARXet-5;3AsyrFpG;%)ap$YiivJRn4b8bjDf
zcZunRyJ3Qu8;A>HZ#>2}+kH0;#BFK_)T*uOb}FMY!Y<-)#_aif{&5c8i6`^tp}@zR
z`<ka$wX<zC^W$*O)Ip+CW8e&DkwE@2aNd*B;v6C|2(e6NIwp~SpG-|%m3%s{oYqB}
zS~{(<ETvmQmXYX?7!MqaY{te+sPEA3NkTS3kkppPWBp}DI2Cxq7gK2ii;ndiOk8vr
z>>;0=lk)K4<-^V0?ZuC|IcHHMA~*Iy7guW~?kJ%yPKoGKkS}LskH*3*EJhhe&x#d?
zT>JpF73G+sk*YC&F|_|o7FQ+C0%pwEjFAQoU(j`VZ1Z4;xD@)`Sd7pl^Ym#t7FwTN
zHGjSD6v&9mhE1HI?C_~lFgBX1Jml@Ry7NIvvL5uY7s*YZkBeUT>dbc>*-&Lmo&6G=
z&B1O|WSbVqii7MD1$z?y3|qZ7GREwaxmuw-?1}SidQ<0rZ8Bz$<F|bK!7yPF*0=yH
z+4PB83?%c*B0qEATvMsxXbPZ6vuwxPi2`{ZYRV9{$D$+i$vwfvfSgP4{oV`lWIXzb
z>rw^wMmTHoO$lIhBhZc5w}oZejmRp|vTL=X!pxe&MDk{tyjouxQdwnx=Yw}T&!)!F
zN$KRWekq)Pag-nm!uxY}by7JK-mvhhw0W9A`WDfyCJywc#>H>HmH+r-xrHGb^%|iR
zkhnzp)7D1gU}ip9(LN{~l{RZ;c#~{Buk#XSF=T7FpHxH*8!blXWRGC(+CrCa^dblL
zq?aun5fZ9@20|%UAfOM~@}YAVKD$ZTEfULO5{|2X7u(u5t+Ylf8LecIC1A$vudG<f
zXdk0}jP^0w$2-;zXdk0}>;fa*w1HUJbm7UX%eK>{hDt4}ev${YWl{Mn|Ki+{Ry>Pa
zl5$^~-P@bnYx1#kf8V_BkZ$MGUH75WW`aTMLq$f==?tyV`2xwoybqFNuiyfEeet`I
zgEwn`mNA~4S8G@M+o`gqf*?n)<|23bo$|?LYfHUnLS^J)Su4Chr}av$+DZr>+vm4=
zak{V`kKW>}=PgxOTLd@ijg;vg(hloZ_Z+w|_k`6p@SaIwE}ZaN!#3_zPs`1#?a*AP
z$t+oLeL%UzNwgW_inF3^N8n~QAiF|J9P(;^eTg4PUNp=%RDr#P<QSgWr5)yb>9=>}
zjgI;nLA`j*)~c1xYQcCt6cl@hnrOTB`-%Yz9$AxFWdhcF<A^XhXi%yW?xlVwLuygd
zql!GSgZ?|{Kh!NMt1kpjbtAIl_ME+i0m7fG!vJX7r7JemjcSW^{M5SMTBmeU6FT&N
zCeEXS{DLu}!N4<7ma=A+cy4}M#Pl}r6LCqHxM6+}0V8qfy*tA^GWLj*E>WB9$)iWB
z0IuSMRcoPXr2e2@PiN6@n(f$9E_7NDu)D$h5e=$n_+*uQjAmGXs>jO7?9H|rE|%Uf
zQzpCt-sDP{3^O(@hf<=mt%D&TFzeTU8BSny3?(X6blGA~-Lq+_8%jeQ=_-n1jpVe@
zNLP=cvOLul2IiO}>I``i9!$S+CYHfip!YH}JZS7q58CC-wgh&iVH)}i4Eo`x^HJkT
zSo${d<XLluL?$kP$;h-QOt~^h`PG69oAbqzSzBt(KiJ7dS#=5TADXxAX17g$q}#h&
z(&={Zx`bUJ@XqrU^IIyn^fl#j{eTshSaBI*u<-t1#pNEvg0bTAz_mw6prbs;aCtPk
z!$`<teP!vSQY}Ru<rw}1>nrz=yql@A<$UdtJ&wkh21}R5X_O=`CpnfR?$S`1$$Td1
z>8_C$jn@<r^@=D(_6bs0uZZ=3iU)2gZCPbvy`pF|7YeRfuUC9Sg@;?}fYprH3W+xT
z+im*#X}J~&sR*0?AatHEkV2b&8#aCQyn1@tiDH28C!7JiOK}WPuGWwN_8~EuIM7pL
z=U8rD6w%1w>ljV$CY7r7t2Q@)-tOw9$n}`Ug@Cy6q}f_er;<vaie|rmFFiAQn9oI{
zw|Js=B-w@Gjo5nD`DF^eBFgwFTa;_qu%@wYvPWc&o=Id^ZuK=iTy3zv%(`q=O6B$e
z%B*$gr6*aY@peJBSkvkjp-ZlYOX<Yp1lI9N!zBb}F(Bt!DJrv4OoxC`qR7xD;F~DW
zQ=_SrpTx{K-J<@*sOqABb@=@)tPPx;5@)|1OPE(!gSrXr!xjk^x$1JI)@eod%Z5M6
z)~e6S)hk!G&aXQB39efFRo6Qwr5Z0|m{Dx)PN{VnX{Cj4U%nNudMKMJ&Zj`XuK8s5
z>`QR9a$Y)byS=dE!9DEQGuVgcC0^t_WUuScF2lEKJoqTX-*L=fF}q$XyOK*A48f%I
z-u6t#F-Oz)p26w8b9;C`<W5W6`R<#hGxK&Zfr<I-soX!Om^x6)GcDjAepxk@YfIxA
zPc!Hz%~9A>R_pvr$Np~cmpxN`;V;-%CNOWm@&5t<m!XUY6t_3s1p(j+SJnGF^|Sy0
z=Xtk6y$0w30r9shzXlEhf2Bqg_#N*Cw9O}&JKonx&-7@|5bsyFXO;IG=9>>5){hVI
zvtsn;B-Uil0}&d9*n>SseY|1r__#nO`ji`dPs$gJX=(|+Cv^|4PSZo%(gf-6(@yq`
z7ilm<a=2j>a$L}Y95;+YzCH9o(NqKyCd}&+Y5kPspw*>=y!Pe3f1uyW$S|*~Oa&w`
zt7=X3w!1<f)K%$i3&tln2YXg<vDkAoxRmYh39f;L5?o()zXiAPclNf&G)N%WD=T=~
zPlK;8+TnB@kQVHj7<?c0><t=hc$+%T9=Q-=qh~VgB_U|S0coIdg)nbub@UZ_C#n~K
zqy}>*#SPz+x`S4ye_3HxrS;he`E!yT>@p$Kz90f}M_nDEZMkM!8$82fuQ`#MTD>#$
zP6-Zrv=o~0Wq?!;WkzU0b+FThT0lPtwMCe{m3H({uVw28p^jkhW}!~l8>9ZV-xgg0
z-7?B9v3n=Fb=xh7QeKeU_9#T3Zd+1_fc6x%o!K)!o=pqLf4!+~J3$MAkI-)+f=m>F
zl@&ut^o-P7BCr->=z+-9pbx@z6S+5xc66CYhCTQY_|lj!BCWzG#1-LplB+O2vZkPS
zQU?7%LV3cMrVFf(L*RST9`HTc)A!^^Gb9i2M#MZpKZwN_c%~%!osuu;w{k_0E@gO-
zA!S_frDaWef7^ESJ$u3`sh#Fdy<taB(nN^F%br!5Zm<@TyNlfi$;(6~0(P(ntdist
zSntW_W%p?E1$<n|H}DdOG<y`1zww!IL+|*RY`G(H*()oBFreQ;ggrQgROy)l9<vlG
zte+GXcrY{u7yM2n?@H3vWXqit1IXSa(Fb!hUY(xVe*wNHl@5KKdV?=b1LUX(*cd5I
z>VrIBou)juEh$pop!cRyz~WD(7HC1aK>DY0!(Nig06Qy{72s_((D>AF*-|;x1l|Y{
zng^9MA+3Z7{Z==$Ejy`cKvuU~oqYzsWtK8%e@~Y71TB!!WXt#*QlMvYhZckZo*xm|
zPjgtHfAi(&8ni9u3SSxveNQaYp1nD)0z?}FkpP-^PUUm7>}TX!^vq%cdr8(rfZxf!
zL+?o6vFEkOf_{^}LoHYXW=L_cf^(s0MJ@$wz+A}dp38#0K34~JeXisS5P^M8Mm4P9
za;6WE;OsFKfmlv)5v)+zx`Cd_;9U{fwp1>zf58q|+y#E};s#jN#glKN_z2p%RAtb2
zL@xMJUqN>ff!SMpThxLHx^f8~BzXz3+H+I{q;d%t_|i%k6TVb>2=tC5pFOWjEMU`?
z*dV@8BKZbO907jksKh{0g+53Wttg2smXaoT7D_G|p2-b-PkI70$dbwaDf$bZDHS|2
ze<C2SOKG4zmD0hkFXav}cdAd&wo37ZQs;!VP#46EL^3_IVF#?oHwPp_oJpDs=>F9;
zSU*)dl08F2K#Q(ELta4jnmt6o8&Un0Y-w5j@bFCVu(oT6V71kdK#Q)SX6T(Tz~3u!
zQwzw>s<DC=UE=_YzsBj{d*TA$ZIv+tfBhyIV$Vcv+Ysfc=?>6>Oc9Qn|B$BkGg2~)
zc5a|wi-6~+<_mNe5%B!fJQMUz-Vg_@g?!N>4Mw5FP0uv2ORA<Xd}$4k?^0`lcw?;<
zV$~u8Moy2wysj7cZbd-fIp!RZ<zZIUJ?vID=OC8Tq>I`6g60(X$(vitmIuuze~3vm
zKOk<_{JQNg6#*Yt3l8*y7COY8MIffvVgYF>vh2QYQpM~U-{J)}V2cy7jzoqnEnA`t
zm6m#eeoHq*Qd)N4cWF6+x2PQ>5D4~;3}pJ9Mzbn6uqa!moS&8#_@bL65)#m&TcN$N
zl?Z<FCPfkEK}j@wO9k7k6<l-We};^XRtCi0TbZB<H4P8U(RxD$M5`0p)=#JwNQ6DO
zyK&hvq`M8PvHJwMINehOs`~<VqX<ay?ki}w-FL`o?0(Vr1d_cyM2H9Xkie?$p@6m9
zLxsp{547fbm=}ys*s#}&fDYTE1J92LSjatIAg_C*JfWTfcouqU_)G`=f1sxe@}7E{
z;Q8s<ftNtKAnbD@AYFPcuxoZnOCY7vn3e?^WUn49iC$xe880;&Mh<nPUN7)<UMCVr
zq}i>Oo(h)4<<_)E`|=L*Lqs5h=<)&l%9k&o-HJ5$9#64%5%98IeuK_;g)r<{b;$yQ
zehU>cqC{ZLU18zd7Fc>Bf3WHgfgTMTX5tki*uht<;CB&$)p*4VY_lsxu;)Z>Xm!ev
zH+ZE2-iRwT$g7d<4)(&8BdqN!XRuGNTp(_CWs2=yxkI+_m2Ws#a1{q<9<DNOm?33D
z<ol`tq{~$!SfN*KG(&2mnOHlmxk~<nt1gIZia>VL?Py28T?Q$3f4d4Z<aT83-tGo(
z>+J)?^lqQPUbuaO-T4j;q}Uy9$b!G6^1(Baa%k@uHa!y|UZNQan$Rts0>&raU}M~}
zz@QHjMMdtEyFKG?S#9<TzEcBBPh{J&>Q1C9i$DzQ&XfUu=Ky{_5r~N&0<F$@(Ax?~
z*gLlZBbPha&Uc=mf7RT1Lrm{30dgnrQmNlc2cMA0ZA+xPI*`{|rI0aw*Bc@aH%phg
zAxCDo0Bsv;n%xH?u;zv<)1Es+I*KhVhg%>IhIb&zhbO&v_yK#P2xLf(5W$uiA%XQS
zGVK{ZLI(S1$O3~_2a3y#&>%K7!h{`gL_eTyF@Z%nVg>zrf5Z;+ZAkLLXvYOR?MNQ%
zZ4t0sM+%S;HBy7j>>=9>W_%jpS00%%SVzv(Z@I#*DMIsFgjkj10ot{~si;cWt
zHxq$K$|xS<!J`Do7?2(W5yDXlXwjowU>S{aLu_i40Xots8)6cpW|AS5yf{Nf0j$P)
zLF8`K8%{P1e{C4_40#FWIS~k?LbmXfZA5RoAqF<J8tm0L-G&ol(+AMfL?EAB1f=Ek
z4aQU?+p>514t~q&2jo48fK5C70<WP6cnxPDO?8F@GGqp_C}*IE^o*1PI3vlO880~H
zBLZ6Ilq9F$nV=)hxIiDB$we52lpwEWrtIl$wAas6e<7!2;=V9b2ha9Q6QW5oC(t@Y
zAX{r@ioVP|As18xA}KR3@R-fwL5|MSz;2v{PTGnL=<9Od)PezdcC$?I)y%RWKV()1
zk?&a}*tE0WkYhF_dqIZOMe?mK*fnR}kcGHhhw)j*w*Z7rLx_MkVo5Fo@d5n3%U8(K
zS^fq|f3f@oYj=eJ9+?$Vhqr}By%QEhV^`R4-b^wX=IDxa(rLvTtiu%-^=P7eq?O_Y
zEfBG1)k+lcT&aWiYe|QNkxMt&b1O~I<5ybvI~#VUl_NxkRvzHLUU@<7j8z=X_~O6<
zTgAf;xJm?%<|-Am8WE5ot8DOOt?EFEt(s_7f7Jpx=&M$UiLZJ=?$@d}=m*<vh?i{d
zU|w&Z!J-s_Y9!nD%l@A2Z`d`r-ynN;AP@Qul&!TxgITr11<~>y7MytyfpZ@s;6vGw
zELRcwofsf;w`1|p>R2naU_kVLYbn6U$#y0}k?)<f=zt7ouTv3-{)>RcveTrQm=^3z
zf09(7+3ie~ICjRFSZ?~#>LB-P*8pd1c5RS7v+Ff&Nxti{?GmkDXY0NGoYyzys6VH=
z%pMhKwia-Roipe4x4p}pf;qN_Nu=1T(Vx0f=1lab5|lZwMWU^XiClJO&mnd$sK~HS
zY8+x~T;4@*Tkp>V`dU_-Ipakj+9%R&fA6HoZJ)Xj0S(f-l<aH<k!I`H&xGCvkq436
zjva~&TaNnUJchPuZNr!5VDH-^-S(WG0!_9RdWd~~!Jou+dRwvgIT45<i!?hM=nz|m
zhzvVVNMzBBFWZh*9AfJ(hjde3dsmY!uSITKw>-qoW%H-%7R;SG?c8v$LAAY;f2V+!
z9WxT?wpA@cZS%4+>@gLwquBoZW%srB4Q7=rbX!Xkx$FqL_o3MSUJ>wqi45Cf5t;VR
z<b8Co=6uX{-jhhNeX57pS)w8p<c<hPJ`p?CCUV(cYLQ`|$`sjl)gbSagT2wGU`O&r
zqV2mC$+j+bO1as3ulLzgU*|xxe@eC)d}+>h6waSRB+%D+*%fn6r54*WClc&)ha#jO
z_?qqboCur~7lAr}J~%k>CX(%_h)A<zdLow{-xis6g^WY&JO>{<WG;$$+m;cDb}oSk
zR2dcNcE*6nuw!6{fQ}S`vn^MVWY-H2$*|@^fU~h86=qckkf9<n><GILe=*rL@I<m5
zy%Z^SKA%XnPnn9`Ao)U^c2rShVXL=oeH>pN^;?SeNp+EATZbYU^tV9uYD5OD5?O8Z
zJIg+katO$(koJPMWjKp1679?@kz)G~M5^tj7HPDy^1!!k+h3(^o9u;9ifz}6;Cw67
z_7;gOyAp>FHGy3sLTjORf6%z16g7MHMwjgT1Ca{)PSh?;B-=iGk;~4*7g=^(#-D<E
z_Ls&0QF)OL;}gS%zK#)cf)Ak{jUMD^jKQuOA(CuAl1PU26C<2O5V>Gw#R%1>MYf$6
z?In+F&$dXiz560Jtdf|fUEfmbsJA^?BEk0jh!l{CF*iHYLZsXIf8-(;jY1x<n;lgd
z=1z&!*9B^*oZ`85j?5vTc}u+pc_8%~R%3GH!AP#ynHGoGUiU-nya164c0du^pC|&A
zzD1^;U*(ghQ4q0Dc8frkj>v7F;5!6%z$A6ho;xYPX)losvp1dMn?Ndf^HOMb<uQ@V
z&gT<>I>9o3AR2p!f1T?g0+H`SKv$MhW1j~Ufm1Oe3HGd%GU&Z2RXd|Y<OaV>lF~_U
zg9>yq4n24<QigcFNVmP=J{72hE8;<`NhNj|(?WY=DsWExWQK#EB2`C{qI%G7MeHmC
zk#1*ziwy7uqzcuIy;MuE=dvT8oroPp@aIUE-RgYuO>E+!f7+}_v#aY%LvrBx(FPdd
zdxE#mtcnECxH4tEije;xhh}Guiww}kvJ@+5fe8Fg+@SM`fR8IDh%Spz<RPPU$ti<x
zJxhkzYayowJ8jO{u5>3-?HmjdiU^2+XCdbac4N-V&R9DH_L7|8wtz#xLY689-qu{8
zdb>!pv&Kadf5_{6iuM4>u$P<z2k^G)>>LacPv29?E0t@Y(XN+W)50GSZ_l?P9myWk
z;wE@o%gL-E0-oUFJxDPTJBL;TPBDrMyTYa~A%bRCq}AB_X$cqDB}H<Fo*`RC$~x%&
zg-#QRB#`7KW;<U@q}Zo2MJ_voRRnUF{3&{3&(T7;f0O<!5@>a*_SsmG25YX+%^8Q-
z8J{8@beEDn_<Kdbk5qC1f4qntdl$Lv>bZx&{v$zQ=g|712?uS{aKnt(2SJY(xW7cC
zf(~2C20v1%j@D_tVC1B}f#0P%ihNfWK!ZH#6!yHXuG)EnA{Y1%s=L9`(}zM-QR3Rp
z=n#Q=e<+85Oq7TRNnV48$Z8D%{75xK@R-$5Kp)jsLf;cMtP<HccAX!Q3>H|8-OhFt
zf&9c1{sFtbrULsyO(f%Mf=WXo12SPw5F6%oj<7G(T<sj%Ltq8hd_(?D%}~co1oDx*
zeua9cVDFR*<aI4KjcJ*%|5PnU_)-!0o(eVae?+>S(Jpe+I;{gVl;#4wt+@;{q@BbK
zAOoV8&23m&%|mroksEZ1=BHf;-8TtKdwfJ7^4&tU?<Nx2c5M^iq=z5@uVIVbK7T55
zft}Uj4cfY_AMiMffFy4T&XW3;qkY;~B-@$HB5-4m$PK>EhVh@4w|z&5Zw2ms5%G4_
ze|wQiEhv;R(8^_>>=wDff6&UXtAvOw$kGu3e^sj<^pI9#p=WN;Hacn!-9`Hxm`H(q
zm+tDe=V*5g_TYZ1CIA@_CGBp4j?~?TmDTAMK!2)KK?^)(DxDfp_B`kzg15Dw@FkD}
ze)1m3Ug)9U_B`leLRL$U9_;O|!G^ZQe*zw+9viGv5zvIXWR<;6MZmw)<AQ8Q5s3fv
zM4ER$?r<O#G9-JV*wi5~3hAPLOVh4T@l5EMMe{9fu)uouAP+>qa_u=n)?UxWuEr=*
z!JpW32mf2oH{>JrykOt%g>sl=M}W7r7pTDCdv&xLtB2@+ugR_(Ap$oHh`<SQe-YA0
zFPFhLcu9!aGx0iSyMZi-iC;d$yuLh~I~0Kmj{XWz<xM1mCriXW&m=NIkG~>~+!ZTm
zgI9u^QAIk~!6J|af2C>Hj5`GU2Ukw8POqH7`nb*|2SDJKIe(P`w&qpf?6OFsek(V<
ztvb*tt{T9ndev;76g-6bphBH2e-XGbTjYXOe+bOpO1V<EOQ46`uG^<2MJ`yuw}X?9
z{`L+cHuuTW1QNj4DFSO-1T;w55q4EU5xCJpq=M)4K5`7m2JgxpJopdpkS}`$-*k7>
zA)s}NKqia`NS8Z6-F=Y@_R~ASEsg$;0b*cx%+$6BHKzTY0z@7}z@EEPe+G?M1bk0-
zYIY?;e`g0LPi}47NWhx9Go1DkY2YKibBBoZouOi=$PGLCo#A}F$Oga5T^!`pi-7fh
zmqNW$pysd01v=7Q7VHamh5LU*669pv$6W+u+oziSIAzU&c*u1h=N=3o6{6+i)I=r%
z+1$fjV9$wwjx-#cW<3P@e|mU0zb(=sS~2_$qdoitKg9?h_TUjBcw|m(EqiT`P{5)b
zp@9xN!UZ~o2t-LoxWOwW0`g#lnQRF=!Uk<)#7M1<5whGyq&yT6`<(b8;FTKjhMdz8
z2k2EJZm_CH3bcMw0oglJ16ef^+@2^xCj~`7QjFXon{H${>w5^~e}{|$_o#_v$buiG
z!ucN&uuDd1pufqMg1Ffz3vzKrEs)_g>VgcSQ8)NUr*n4wUXcp9y3<k4wg~j~MBQ@w
z05Wk(kVC6IMN+0GpWO@&q6jlk-o}gxPJhmrAs2MU3VyyB$xkuk4O-fa8$9(hQEt&p
zI6gT--r&p$>;>t9f7EYUS<nZ0g8yLV1scjM8FVBOkmR#ad~TKjn(C|$^515y;BlUn
zEZ12V$dKidU0X|}K)i9e2DZ$iTc!Lu=T(3-utKF#*p(ihiQNYx5Yt=1gLGLT!pNO^
zLi7xs+*;>qdmuO1w33LRldnjzoD~DeM9E~Z(M4dVT@k9@e~VBoXQc?ah9aOJEUXH!
zQi1(vC5khxM5i)EAY*OiX5V8Y(joV2<q3Mo$_w;^RVY7xl@5{bRp4fOkp*$uRW`(?
zR>c`#q0;vua28?F-BBW!U87KBf~45aL1bvV1bfzY9sKcIJGwo4w;Ryc+k5chY@fiE
z*|J#dFWtwde+h*4k}Zn`dM5;kf$fk%uiAmEsU0TRGCMl37k2dEHQbTTd`nY?h}w=+
z^+W`6*|0W{$OZK&v=Z#9>LLqHE$zJR`+xi{86wgmkO{L(ft4i!QS4nR>?ND(CyKyb
zJ|Y+R95!a@>@vX86QQ@&L^DJxIgDv7pr!44fh6Dce+ECf2-vhDwtqB8qp&rH@O)!6
z5V%cM1nM-3Y<u<wNBPPkahqcrPPT_R4-T=<UWOB$f>!(6w&z5)En&m?+gw@aE<4+|
zAu{a>XF-Et&)!4qeOtuaR@<qAYe(TkvTfIkfc_>@ZAl>lcO**ZboS{15l_97qP?<2
zphj+Rf4SS)s3Nz$wu39MmxzF$Lgcn>bdh0anu~1v3~o3V$HFre?j{PSW)Jjr9d_(p
zWZPcQ;2mT#pA(U`7gVI#+PX-$wLuZcvpa=2Y#(Ge^>=KEBtZeI{uJ+kR*TRx+m2#~
zQ>esVr=BXUhEr04-j?myt_ZC;FPn+TZP(fpf7!O5;t)HQ6VAbR7&#Gp{hR~o_WJP=
z;$|XnN1{l#Yjm8lD|XI;$h0*S>5ow3=9xh6q(n0$K|QYEGek*6BCV1<?MgP`Tow(#
zL)G;~Xs7kn+1U(-*x3x{JfiJ26uIqFjEC6uM}xLakRjRvZEN=syMk~yBG&Hf;GnAd
ze<8M2ATnWI2e)Cqor9e4rKi#yop(@E&)3Gkga82q5~>JD2)!!^NDD0#ktTv5NK>Rp
zQ+mIIW@wT~6Okq;-5^zvsvtqC^xmXO?-1IX-}_G{lRI-~CwK4J=X}oDXF<@yAAT6F
zL*<J0BjYY2b$Mv7O-(!l&%{?NSYb53@n#_X>wY0{p|ucbcz3Akg{3V5E5oCH77ker
z=Aoym6ydl0sRCZK#NGsX35uU_HWNN0vEJy@R9T%XY+k0NH>7m)<(mnRGyI}j7jVPR
z^BjB%n6TEs(VeZ0|7W+A?Cxk0$%Ki#Def|`?W!m}4^>Q4WttgxJfr_AJwi{Gi7+v|
zc$909s#9aw&Z%DVm35L^5(J}%eW0rT6OT^|`K=8;_n04oSDc9}SUkq<WubB`#^t}h
z(6&K2$;E}XDq(oW{Wg@CoP3;ofzT9)$tL}Z1uW>M4ohX@&DUZ1nBAIZT8Oy_rO2i`
zYT|7g8^k=spC+?Q=wylLK<h}n$?HdaEi>1pr;LFYnUjY`;U?x-)30J<9M8t@Fm=`Z
zy*~d0pqYIrn(&0{X#6Xd^Mfg~@Eom<d+VE5(L<?=oi;h&`52DhL@|T9{{YOzA;o&&
zf5vgwR%@9h6>o@EY<!EaAv&3H-~C5+=o5YerTr>wT5O5Zy5`|IB~jc%u~{U4VN)kf
zhy3ggOQFmw$m;JRg|T$wLtlp)2r3DsKw8YpD;ZA}G>{aCFh0_OhLEA}ETnQsMTT$W
zEqjJz{TNjFdL@32tebHc1-L7^3rN|CbAlTG)@CwrcnaB4Wyhg5|FELv-)q$7<s-<?
z%L_#4VBdCiUwDA~icFrLX?@BQnS6EdP@1JQk<(-6&hodGUWH+g96v^rXG)Gd8ynI%
zyAB!@tN*bV{Ty$v{P^P(#5i&uVZ2GBkFdX5G*%^<TTB!*q7+tB0Yude+9}IWObGI>
zy*?DnEHf@RgOb2D)*1BU<Mw^4;~bwJ``~u%YU{t*uWVzcd_(wPg+JJS7fI`a1GOg^
zPO?bw*E*PtL!+&i0ajG*5^I!5bRSi@3vRf{f4ARH*SuVjL?A6clK1(xVtR_~H9TD4
zR=S?>SyaocWj-*{lJ=WGnw)c?Jw{bkITUphw?lF6EC0n#`I3^Es&yOgr;dklu8KcJ
zlZ{u?2h@T@C@S7+QjS$IdiWlt>y%eoGki7lj5V^?0XNKldW79Yz?ts5&3%%no|LE6
z_V}-)%ohBQ6I!{X=ED*!voatV)MV=S)1nYLVakDen+bRbduk+C1y*p#<lF9MPe}VF
z#!ekMjlLY0Grb9VWI%_l<a%-7i+ro|A6@e=NKgn%SS|lIXZ8GvmCWoeo1Z87WN-1G
z>1w*`2>)eu&LEFp_4YsHUZnXxUc^gY*4u@Q@`308&ZbcQ;g7ZJKG9z3T*`xRgi%+O
zAqeBzFAEgAsv^iYvXij1`}GDpi6iwJQx_T=t{tRapE1UwI!Ips6H?TTy;J`D{r8p$
zH}{qXclYq-;sC1e4*Tp?VFDBbTU_I3Ip3>zMZ>(YI^wE6_<qFH2OeX)=Q3*&N1-Av
z90ww>a*j!12Rc6W<bqGGj?A&lI}}wzwsJ0Olz>V$i3^LXu`KmhR2o*s`oFOJl9P*3
zEzMb?C;D`eS6HA_&D`4~8r-!t$p%`VIxj_Ue162VQBd}1r=ogd<M$XsZo{EWogdww
zS!>7^i#i+1)}h3k_|ZMpn)c}zCf0fV5kRV<>Vv+KsV{;u4K46ihqn)s<kUzJior-2
z(9Bn(*J4N>AN(2J_*I**%zE$EgtBk&WZ09RH)}K7OL__>>*h&o>h@83%Sk>(>Vv*F
z%WAzxMb}}4mjRfN%L8b6eBjm{zoGvu5k8OI`bYUArq-n1=P>*R>lcoqJWA3sYeOL$
zRaO>s@gEMLXMPfFEq#9S-*9?-=-Iuuz)et_prW4**uV(Wa-rAPmTLM>W~auQr?t5G
zRw^$D_Cs_kN7>3Q{JFR0U7QeOi=$sVedwO3aj+Wr4ng#XxE{jj)jue~z5XqEqZ#-p
zN|b-+^^>A8A8Jz2`wTHDQUF)1eesTP^Zi6Iv6+(VP@@xezApN+YzsjxRxZj0V9V3I
z`L5SJ*(|n1&|0eA{JfrAO8M4SvW{lThetj1Yl&XJ-<R+_qg+q~=Jo28={!9Rj{b}k
zX+*ySP70pTlSv*X+&q}%-}yQ7C0yL=yIZyueL*Vy`I}>3tA83BUSbF%ae=8;HPYrI
zP^lYiM|2jAko6nuAN6a1TYCxuckP73F9d!Lr&9Sv>$cBJ7}4FG8h%K7hsiEjWQZ4e
z%rsHZ%0O1;@C{sGin0}Vl8ch$qAzZX>!)Ewvncg$W}{Sz3{^DIuQR-}sFkH#iPu-`
z3-v)CtQElxqP)ywnc9VOYHVNoqGG3R1%29mu<eAeU8#n&$!P25XeJH;@F0hL`td^E
zAZ!fptnD*GYWOutl9J6LgvU=k0HKDogt36KZZ&CY-np?kusNZRU~@Xj)9e4MO*dC?
zT_*gdl*p21qJ^Cel%(8DVjsl^4myoGkMb#k7Z1H*2;qma(HF|3k48s*zi}BjYR*6q
z_k0vdJ4j`&GU!MEyxujbGY1iArSSA$9SH~I4j732Q(XZKz2Jdqm)B#ahA7z+cgJ|!
zj7%DV{_uD340o-)FWqCU0~x!N+0*glQYH?bb8eX{6)NzTY^get@7naJ8;lX<d4|On
zf&*~+yMq+p$Mw9UpDpX`i)+X*;wZ;|Am;uslU=oTlOsxihrZN|V}_3R<|;Gz2j8)N
ztF#2W<4A;0-^6YNQjp--W4qfIi2kDV{9ago;b4=Wycm^_lcqa?+S)x#<pKDJyHoeO
zdkN%ma+hK`2gBF1z<}TKR+rJr=HQR9>n8f;_aM+iD|CZR!JGKTibvlzUa%p?6kkoS
z$o|wNTkO>ZSk662_;}!vFE+igf{IinvVS8c%yG-Jt|i{4E~4hEZgitPOYF5W^?+K$
z%fP@C4*Tt5>V<?7hcOwm#`!xZ#kBdKAm$5igXWZ3$?#jgTzjWAyl0Nz>4>XG^ua=t
zKQFux_FS>5U$%}oJn6nh$3mW~xIB-b`^MWY<aa=Nww2HRtE+tOsMg$0()4fg&*H}X
z{4m(Qpuv!2I~V+GNqr{YgG^b}-J?1_m=0Bf%L<v`L%&zhfU)Zd)BLJWywboE(GdTN
z8ehlN+V?GQ$EsCX313+D8ftyL5nUsG_U9-PpMgwy2!P}#jR~2Jtf+=YNeoJniZ?6h
zQ#bJ7ZrhtNj9$>uuZNSrr*1FEe}DcZ+uz`m>ZJO!r)ML7V>;eH6OnSeP`DZ6wne+!
zefN6Q^$nUIh`~4c7_sj(n%<q`r&PNpyy)3`=wE`GBmv>cQ@5$Rffzj6((cii1VS)Y
z8pipPj$0_!1mCq9J~nnz@WMII&hdl%UZ4Q&01~~ec45eUP<WSV{rSqP7*);G)yzt-
zcc>g<0U<14lM5yJw<u6ue5~e5Jdk8QgL+G%L7l#L9_5eCG-$Y973$QAE5p7OX@aHx
zeq@+=^CPCE=Ldp)xq6f)oVCp<8zDblUQ@&4IbM*Und+9-_K=eyCi5qA-=|{~sR=k;
zJAb^sLg#g6rz2)9M)k-`$xKhfX2o@OiK*}KH%iC&q+w%U<xiUH8C(;d<%203&wRGv
zn}yZM-Nl828XtO0T6*s>IyIRV+VkqOmmcS7Ov;w&R7=~TB+i~p@|5>c)VlWhUPDgG
z)<dRR*)#}4;pQ7pK)8&iI}(Zxd_ct}lgDj7NA2hZ`H^d72E#`LOpAE$Yoo-^ZnIu&
z!}2vzoPD>og5E^OE5DZ}V3~~aX4ib(BMEe(cNg8Xho1KDp;=AN2Q1tY;~^$unB7Pm
znr||R^xq!Ds+Q{8PdMfnq==Q@uKF`R@?EE#mlM%*MDd|Mq^4jI+HY4K32g39B5pdQ
zVBFALPgle~c-0JfkQKw*D>-u+)GPY(IiEVZ#Hc=6;;BizZ$B|yz93Ij_ptO;y89w_
zEXx*F?yFfJ@>w)1E_=-{n(HykV-j$uhB}T~!R`%(CvO(hAzl{LU(?r?M*dW8N4#Vp
zv5EPPYw6T0_HVhE0D+xPfE02gL@^ceRw>n88=)z2=4ul!GdUcl`QEDFL4HI#LjCPW
z^`m8;JKCRKw+Ib6o8=7V>*nP0joS6_%Warf6Dsz#pY5^v+vi(Hq^5?yuFmAc<qY$n
z8{yBn?wm8v8TjDH&YWomkLI~SBmZMab`y?gs8p*53(y-6tI~~u?l~`=udQ!z3zR)P
z{2sGiyo{TqHDmlJJoe5ct;j$T1LG8Z5Qb|rNhi`vOWS;6IGB_L7G@?ve4lILY_q+T
z+mPh^jIATgU!^Hg3jXhmd3q#<!PYg;vea$NlBo_Wcq0*u4EKCAN(@=pf|@cnJ3=je
zp(U(n-C4wBQxKrwz#zvT*OEzxm`29ID_SHv3#7=f&sP@OIF&cT1fP=;;p+A}`4P-r
z#v2>Adl;9F8hT0Rn)MWo2@0W#LqfNv?$Pa8w0#NXm1<deSzMw@%g2zT?N$DF67ltq
zbP7@2Z;Z?<`U3^dZS;9u9UtZ%N&dcB9)QPUh5tvlfovlQj-x=~h#cyrKoUz`*>IYz
zLv!BAUEtaXYp23ewL9*Jve~p+3tduP`N-GqkfrB*VrjO!nLM{?UI=ln<wt#gkxBT^
zr8v@U1=jI!5}Xx7dz8&sn~4{T>qLjbK?(u)Zu+*og-dUlJIp0YQg3A_fJ+k*0sl28
z<z@N+d~bQ0jP~dh-8yL-?rIXr8p>2zJaIHG-;JK%#XVmc>M$;_S`vSh)+%UXg?^OH
zD}3~#^ozzM$~MKsZ(N>s<<jFUzK6$G*#k%B@;$JS0)Vuw6er%A<+E&2v4|j#v-vLT
zf5Sz`cUT)nSWrHmZJjFGt*K-uO@U$Ikz_ldXLzVZL5Zz!Oe){OP@G^?y9!m@ev}`!
zY*a1Nda}r&bR7XFacITI`{nyPhL4TD*F0(?@_Z#3S|ngXiXVfw^cjVHLrEneFEi9!
zmIN#|<b%fZTMRG0_L+7hMKid3ku9G<70&*1%Pv&lH&;87U1<?<>ea9>tkd_{r~e4-
z^=QyRaR1#NUCvdA)QOOvIH#yLr#0%b0B?~qb<R<^F~%IdtLbkU9S7O}juq?2Se8Po
zUwqYYP`_^^t_Ak&XCS-s`3Z5q&(_vcuB+q9-kaL-kQN*$DAyl(ht_Fpk*wali`gsX
z&VQ#Px|AFm;NU-HCOd&u15SGPU!Qz{LQq|n<2gTNi^z6rDk(t1X~Y#uZUDJcy3*&@
zj%j%5c(F`DyL68^C6KYZk8ur)W8J+WcRPWiffF_>XPSLYIum7GZk5&0`UB7npFT+j
z?Vc3q6}H3B-<BR)@0e#*@OT}sI_#_yi@IE*Odpre-BT)4t1^vj@;RGB1x_5MG#{?r
zI@Zh&XCo4{J~qqp{JT7$0=C|gO#O}rZEOleK=ngv!CbFGw9w`)(gOBnSzn#n4P%p3
zn1(J<s>k;i!_)S)Z~|CL>()3}kDMCNNI;7nW!RBoI@#n>&#g2Q@PVDCYSFi*+h}?M
zt>COBYuk&NidM;hd-p6B)qOY0ZsFu1Cwd+sJjG|S^)j<sE(dPFtz2HnkZ$8<luqYl
z>b42q1L8*RY?QfsL~q*Q!Ho?GPz*TtxPm`DqOXsbm3Ou>Tphi(aCZM+Y;L$J{bH?I
zeffgfm-5O-)-M(D4fEcQPgNtG+gpU;{f44@E%nwGlsiF)xNG)mr4GY%tfR$L#VE52
zcwdzTT4(1)4!a<*chcKY9~fRr_}9LEE9TBl<w%&Y9%AucDMSXeX7s$Aa#IY3F^E+a
zvrV85VRUJ*EyJrlNR&}L@p7TxxsA&C6`;TOWHgv(zkC_}-M)tbS^2CA0xNtyy|$}d
zg1~PRnND+FcnDTp_)%su;%Kr?QaiRs4}00%KU=XjL|Fl{sL+~gI?Ze~Voy&f#Tl9d
z{mxFuOA)jUey3$kwC0O}{7*8qDWo(w$|k3U4Zfby7F=!778ll>yZO^zv@|wqhc_(W
zWLby_YOrPzhm4s9g*C+yR%9qqs%zi3#^IN&?^!R`J|yReoT56Vu=HHaf)_c<Ce2JD
zJE&kst0}-7Yp1dn!Q$x|!C3(*ZA(%tok_A;?Si3&COfl76Y&N1t8$8;!peemqfC8%
zAO3?Ln)ENQu=n%(If+z54p}g&TNfx86P7Y75`t1Q402A8?Z;jN9Zfn@DbS;W(`pDS
zL3``CP0(yT+uanyYt3X*Ewe}k@il3NHn^b*V3Z;T8X08I(GlLutG%VlEet_f_KnST
zeMoO${G#(|`mUUPPOp;ftISiijkWv#n?VOIvHpj?IBtL2Yy;%gwE8BfNo~uvBd$f^
zM3>xUCr;iW3$>;MEpA}ky_iI(>~J&_)uMJs6-tXGb$2lSrh>3{cwzPGXvExu&;&p#
z(&l^lLNmSBO_apOpofveDhF5K8M2PQoc>qjyG+Pf@a&1&;sHn6%vAS5sbj?V-3+dZ
zEAEjyq5Bamz(Y{b%I9mRV$+|E=6xQ%{=WE`ENflgK5@ng0hXFW)w7!<YpSjk%)Ess
z0rx3YPEu)xtIZrC7u9S^tgvzuk59nN?EnWXd;MXQ>ukyunPI_?4?3_4EB8jM%{K+B
zxt?n3KNpj%r8ASFLyS|Gs#ZOlv<9BA!sNz~!TPPCLqBcwzCyC-`Kd3}aE*1hNNiYY
zEB|3e;j0MUWkqTK*M7a=W=Df_010^1<^xAP6ua^<@KCF`FpEmaVhhT40EmQiQ+fys
z&PwtcH~vZ>?7-M@fBizB+%m3L4<Dkvxq01lGwu5Rz-Ja1$1mMFIOVuYP6ZF>c8*O7
z%~*>r;~5r2FHq6jznIXct!n4@X6!Wv&zx_~>e&b4Efx>QZ^OQ=d|ViET*^4YC*IMz
z95KvfPDcvz?Cleusx`1p0Ek`H*|}Ulg^KN_Q|@z8myY+WsN~*0i^%N<{bvP;oQdgy
zZe)#waY(6F{^|Q{c+)~<6ltPH$OosO2}h5!QXlR;FoN|fFrugIx=s%tSwOih*~ZJ&
zJ&`l9?sfLrHqd}`W^de*_&FT8g#bxMg!}uj;QB1r5u+s~RZoDpY-0pSh8B^O<rLC*
zqz5%Ot!s^R#x3kWR+~fprgJsNsr4oN8_`4(Ki#>_6G~Tx%9Pl-Pqq3XHcWj}1$`|2
zdDIV<J1N!`8C4+tI*|gS8f>z-&oYSUFJo-eeq|9O%B6EyCyOI1@a$}NREqGo62{F-
ze?xd1CkKbL>jR?vL57C;KQ1292zA7U-FOD&_JWd8l)-KHQ;dqKJR7pCL|IIR^Inbh
z8AI1zZY8gB{|CD97BQ)9ph>?mXPnFXHRWi3CYb;#sq`lNsJ@;_Us8M(Y@%y_MkL2E
zY!qKN-}HON&gA=&9p4vRD^-c-U}m3_PC4BVEhnt=ssM<t4kk_6vdn<sm1neSion*;
z!HHNn0qrIDBX8%*zVN<Fb^k=sQkW#u!nM%$QR0H5HQspm#r^VylA~LU*?z3-O)m@=
zSA6W%|1g}3!MG2d{;bk87bLkoR_vBau`y}sD#|<i4aWUl$*ys-+zZMzSba<f5_HYX
zG2j~rkp=F71EVd#(DQ9v($;q!i+2&1#{_@0Gz#o{@10O+Ld|)E&-5Mz$D@vlZ7wGC
zdwJaE_n7c7(>ynZ8ExnZW%cQMMjXX)4W<uKH4<tar4C+JjdA5pXOcg-uUIxSlAC=m
zpE&Uh*QfoTB}cni2IkJJ4^Kr1g^P=HE4_PVD<D3&OkNw0>3f86oxLsM8CuS^yt3Y@
z`32AU=f0CI_rV>zzQcov5RrwmKbqFu%B*N-84Jps(Q);R*6q?P9&Rzv;DCi>PAG3h
z>%^+e=|5H6f{R)BN&OvM0+8@Ax?VE)B17j;*$>gL&$pL%oM*+bU8Q_bGYN9hXcJY7
z0a}uyMyjE6-`7N2<W`x<Hh9&BG$J1@TyF}33*cI;W;^y+P>9LnY)bU^{6yLU&uqGe
z4+gQ{5h~>un!%syAge!P8*E|6b-~p!gd_P|xWA;P6(nAMR_o*RUE`yFbqG55m0aE#
zl?YrC-koMav=c>B8ex>1(<~d@YYKJ%++)sL3^-4Bt8LVRhrbbMw;dOke7e2=2xRv$
zc0y0tuhU*%I>}m)Ro7>Rbay}&M72q8yDV$^fq}DoDdAJ!op<d*o-G~M8&66<5~xOq
z*Xi8jZ&QXYQ@Fg0r^69%BQX^5=^@Q}aYthQUX>b3q1|iiCYm`vjkRuc53qIuC+QuA
zDKsEaRIRZ8f;3+iTF>8II^sqQA?l=8jCyPE@lBg00T5BgZG@8bcj>fWORUnIN1W<V
z$Ozk0vLusp%~*T_bWKN~4PVf}+5E-PvdOrp$GQQOsT1X&0-4gQlq3A;p|^UD$kF1S
z`1Mk;?@#cBQ+I6i_fY!&X|{2|mag|Pd`sm}2N$M~6g<eN&h{+_3J^O6fHP6V1;>s0
z3DkG5;u~!#FfF?0OElK2hJsPR$7ZFF_5uQ}{HV=m%s!s4JTjDDBY4G3;lpJ!)}L}u
zHQyR6mTdDFJIYKpxyLrBg670)vb33P#souCh{K@tFE)jjpmFRja6bIEL_`y5Q^ZUX
zYGd@OHKPs?l|H}Dm?s42WnlD^L%i9nngoc%HfY@^Y+nC%#nHkJW}UE_Ckfn(m7zl{
zyr6%jh-^HWYxC(z=rx??_gB_|3j69L!w<TRt876!o2ctxDV*W^Cb@gy;!RB<mqWha
zmnwDW%`5wUcSax}L@*4J$&7fJjSSFG+U?{%2}5A@1NSI`R)me7bA=SWtXE<g@`_L;
zDap%hi#e0f7ZDpv57jEQ7pY(?B**y(0@bTM!50l-E%~ExBeLBpvefB8>$`T@S*;UL
zvhT+{&$EC9znwwRA+=b4{tS2(TVeF0MfE%9?9@`6eL&c~6a>Xn=RT*G)+*1!%EV?P
z9wGD-AfV6--U8~Tj9`7@HE&@s8TVftJDs8TCA^Z@O(A<K&5#+hAE9`n-+p{8rX}7Q
z_RgDm0*Xxt8QkUiJrUb!HW|mPa%BP*AdhJ<BDEmgT)O|B8bP&DT+2+<E1KJlQ~k9p
zPUkyf3EcdaKC?&M%7}i!w^>Ft(rbw~E29Ms>w9#SSE8IuJ9Q#_`d5EpkN+NKp=?uY
z2$VL<Zrmon8WVbJYc<T`A;bSQ4Sm`%oLf5}Pd9~O`hWYqb{$YuFwfjR3DCa|TkU{8
zJ{eX0RNTlk+xKLyt6`r!!$Z@KA6D-7Xkv~E4}hoNBDhWRzUZ0aMFDH3&t-2L19Ksn
z<t68HZumV_r2jBTG<2Aa+KERCN!L3QHI9T4%JDj}<moR5zvv*y16J3Z#D7>i(H4{T
z_vKu=9WE-KYYNGJ3;g6UO?B|Kv?-_9^N>!^#_7k&bfFc^*Fzau``E*)2$sfW27@jc
zfX&-#<N8l+u&0}%58)n0_0g+>O~v(zUY63;Ogn2QXyM32K&t{2Ez*`Se3aoZBiG1P
z_<j1hdd<@`He#i(ysg>hyC(+VqoX$gnW{qj<u0!q8%5i3z85)OF4~441#w@g_R#uG
zlTq(}Zn5mCSkl6pENAEXHXk%Brx4IVfLbKUwRt<laFx_|Wc>i2nLPhRg_`I)`A2H!
zs;`scXsy%x<_Dz)kpx0yV~Fly@qdV)q9Ji&vlr~(1&@N@ldIq+xtdIiISF}|We7&E
zu#<6D+7J~w*B7H|+EwJzk`d=h1~cEK+MeURBF}PCY}&V8j3}IUW~+d_u8Nuj*h%M3
zB3?eqygkRKawkO}oP4EetNxlmX1NhhYdx9HtgVo=;PG1ESBxHp@ZnVHV6?{U0qd!4
zc4(0*RVii^1amejI?%LagJJ5szC=LkePuU(NiytsHeIW~Ns|8qtmD=ZGLRv&&cqxK
zE7fLQzMlaXDU{RMHtbj!G<O?#M1SIifHJ_)@hfI_PZM>}g=cIcn^{%hivpVxY}tuL
z#7Xa{!Vn`y^=L0?o^D;WM(#p|0}+NWGTWD_U}&PGk=|UVc2CF46?&c#!>%um+T0ay
z#=)hI*G!@>q!rpgO}k;wj}w-v^-;+3kInb0MQ6pJC8q+<^P5ix27z?$DXG~8qZK$v
zJ~gAT_v|zG!Ec+5Js4VPQ@_j54@MTX;wwEYk5QFaS*A-LE_v9kQfo=mFlbr$W<8a9
zWK&$F5MTdX<02zi_8ahJilRBtquT_u7s0U;U9SwaP^^7k81}79qxvGjRS3)W)VZP!
zF69^W`pN1hlR?@fH?SkLz4|hCdRLWG_>RdONBq*tn;6dW4b<ES^X?6l8r;RFgNl8s
zgo!%qP5d46JTLX0qfF+N@B2iF?3)(iNB9_33TuWkR$}*hh&OICuT1&*yHN4BA&KuH
zLthT~ezqdXVLLsj_bMe%1*&KHm4`Px4sIqjZ7Q-jH)*Rbt^j>Amw26Pi~A#i(CtBA
z_CJ@AbPLk$f6D(>X9t<w{&Tgz?`xVEEYBF|wYjhBr+m1OXQS3Idp&Z*2ag0@Z?iO1
z4o}(%BIcV8+fl`d;hDrh(;~{?$<;LL*$2aeCjlkWpWhNEFmP|<_=Ie9`VHlt(c?YL
z6Puqyl?@tvmw*Xt`phHNfZ!qJiI(6y2x55fk<DTtK4;}g;|nZgkt|4Jc+vyW<M%g4
zsY&HN7*U{X_71^25Zqn>S>2CO)zV+wmw9X@*s$&RkH`Jmc2y%%7hci8KpL40VsuhR
z`FLRy3VSEVuPzsjIgZe@9&4?W<c?rQ|8CEsT3iVP?g;(dW@p+eRSHE`h*vZo|Jm1T
z>eA*r^T+7y*1RL+ugKbP53`V^pm2ZljO*`J;0>ACX9TwSlUR&AkH3alFQ&CP{{HJ`
z{=dtOLyH%0;Ybcq5X&Z_XEYFV&3yOMTNWb`lg1;6?rCLh75%ZV^Prz^hyJ$Lx!U>0
z8wsHH&oKg{gnTdPIo~V78xdlqwLe^~7EDR<L_TAe{GjvPb(U(ROYC4&U|J9HG9zVt
z&YgMp0SNauWefFX+htv5Q9n$N+xgnUK7ZrCox|C(<&F+3l(kA8$s}6yiS)QcH6OQM
z%#mC9PDv^Tm^9_y{hpuNal9MHX`bh49t>D(*|$(zKI;|6a+^C>BWF}t0z%w%BTFsi
zA0lN>4A0qZ@;b6H_%JA0h$pUZ^dTWp2WbFn6II4|<3;(ULioOECkcl{kRKUvafy9V
zP_;6)Kf|OGa(!+gZGh>k4r6k(Yn{%1$}Q!gr*mlrnEjYJuMF%U6IoBz%+r?Z0Cv7|
z31Lssh7D}~VyUn*g$j0iz|gAJ?5n=T5Nde6NtM!Hj3<00+p?>x<aw+s=~?g?s<)X*
zqmCKOu{?<@+fiEkJ>evw_<M)MfIap^n^KlY>4j1HbVg0N%_z_M;;2ta;;0>CK&wDn
z$bz($a(?ujcojV?eIzH*#vT}Ca_jK(I}n)Um@$NWgzg3$z?+_0*tayoK{cNwu35x{
zoZhSkK-*Q^-)b$doVel_MKOUoye&2ygD$^fRK+t0E7jKmt0I0Dzo%1bluhibKFnf7
znW0)F*MvQYdPnbHAsA5OfyCe`8}PsO%l4<|y79N(*ib1qA6z-L{sgXgX&JNBoNb#s
zpQEL%=_5uRpy{^Fhw7qDX-kz#YAwZoD5JehJ{Ox2S=UGsW|2zt>nN3P9zD&@ZEqp<
z?%X7cbcpxPp+972bTR73<Q93aHEcFl%#+99k2XlLDUrL3{QZECXRtuUar%XrJgj1a
z$a!F}vg@^U=tVhI1@J1{Zev>X6$ePFmeGDoy`I|rGfvu{k(X`kke6Xhh4(6W+Z<{$
z&jNNocpxMvqf6dwpS+cFPt*Dvo7UrlX%Z4ag$$|p)f+RLfE!-GR2~{FlDj;sqI}fi
zM78yB(-^}`cN(@+qc~(7XJXNKK@~c($}qL+c>uqGRM2|?tUXih2`7BKa&#rS?~%=O
zbYDAOd&lKEb{l{eaqpw`M&HqH)ZUguvHps5E`HE8ml92)fIE&Bb9KGodRIh8NeL$S
z<yQC9Tt4>ured<@=WSB#wy3$g(7Nz`iR!z3D%)Of<&)XOKaM?6AmerbJK<27K`rO0
zt?WdCG&3`RHp=`Ne?!U8-n~V7mCp4F`U&mWuO_VqZnm7o;<7G&(B0LJJ~`y%`e^JW
z7D&nT?;<E(Aly`@GlP*dxDOm+%rHXU)veh5`wSbUHWtwR%TIKomuadT)mNU;rRRyv
zr<#N<7WBn_7j5v0%?$MZ*%Eo`KSPNeU{a+{B@~D{*FEMkMMN}NrXR5W?pwE~K)bf%
zjgI!rkOb+A7g683@CLKAf_wWi1&A(^8;{XG+GfJ`O9SCR`4pi-XuP^H6do&vcjp9)
zP=>mly?2AXQx*qRI-KFzc95YLhYAT7azYa#pxYiWd<2hF<7Z6<8E*fN&(3ZNWsXoN
z0eQ3FPdc!7x8PFqvS&TCBg7<If%gzSKbW&^lFHuabD99g=|e+{m*hx!HLcA~rUfd)
z&^gYKy>QM7%=Q!w$oD)c(JXY$F!wzLj8KYvt|mqlrmDC?K);8)K%W>u{k{BSm09le
zXi9w0K^(9?^<~0>zP1*a&e}*y%}E3R6NybS`kqYatqj7&*(D>6`q0$y>Qv$(E^Wqe
zwrRy+v4#5VfWe^2ks0HI%=_cVk>kU+H)lg0as{ud6wqwWlD{}uzZ4+pOy`&Nn3l=V
zhp)SjGla+w1&pZIBrTjhJ}V1OjvU8gRIdyH+ST!xu;O|G!_zj8@j26GnINDY1he4r
z&^tQJ=f_Mpm8rY6alCg^wVn-O0&fkP-7IeInW3+rN6?Ch&oqU+zVGxoyyy50z2JZ%
z<<#{GXL$YI8;0O`GYP5EOnIlY?)el^6d^Boh5UF;kqu=(bBdNZ>0LjQIn$*sp3|J8
zd~$n?3$l7Qj<fKi#aiHIJRC6UZ1Co6+3uvvYW+QUoE@uF9%1&`v2kygMis#(0)KSx
zU|Kgk@T)0n+g~#@{cWxW3fC@EJ7R`T#P!``hJ%-pzvx0DHiMpNBJT$I+)!<@uRi9f
zXp}N!(7$eON<Q>d{p*k)HqNMe=RH?3`^b~fiWta&NbHkufBdfblYuM%41(=4J5#OE
z%08@-*O2LjpkHW<^u(th#S>94ZulH&xV~ijO}<r;BjURL?a0G@lzN=zF_jUQ5a{u&
zm7S;7?ATSMCB=QGTFOi{MP`qf7K!m(U-06O&4!>)ro%-d0Ua@Oonhlp1ed{AWdZwG
zKb<9Vvco61t)TpfmYAv)AaQ0YxMBGtq}}fLJjcPVkM?vfiDoK6&=miM4g3#dSal1&
zyKMqqpb~sgwsp@$l2mJMy-MOPSJ2E+U&8t(>AarW63z*f1igsA7`lcW#QmiqLBvIU
z*uS&8o2nnqQD!tS48;dzBcg)9;AvYRNm%2`e0jiLoT9*Wut5aasVlk={VRc5ul<MF
zX|K57g$b(5$4Cn)`a10C4yd(E+^tm4-|gP=u}29^^*Txvx+{1E`E*|L|4TbXY|z=O
zzg|jXHUDEnY_YiV3_s~j6aAowfZk?8BU2@tp=9S$2QkNZzHf%j#x1lgXvn4+<&aY-
zg~KTreX}spWClRlUWP-;TwwU%@n+c;hRcb*K6kbNxb|t3@h#Zcq)kMp<3*gcAjwrl
zbZ@^tN3DhaR6%I@z{3RpRCvI-G2yn|X8h7&(3u?5=G@~J(G!OVLZ2Mf!L#6f3d_?B
zS_33(Rpe|Lt$3m(Dh2zt?qSitAHt+Z;y8~d)ZnAOfOO#i#$<=aed}e~lwKo3PO^}*
z?lhc*B!_=n-dvrkU-Gw-h5Z9Hu`IudF}k)l!rf-&ZZ#J&^(be3p@uSLyp|4=Gh34@
zHDAMUR0Kisx_|E<_1&GSYaHBeCH^t8M-NX5qT<{O+%H5GZ)91j%*EJbUX}Y&>N!By
zuqkW+`ojqq!(#fDe=hgVLZ={TSQSy_>kw_7C7Ks~dBX=gdm8rKZzbAtxtBb3I>k==
ziead}W9)vmHX9j<WB$v|ARJgwss;CaAoN*A@J6~~PwH38&qJ;mFYY()J-D~2oN9#J
zce#VXn=lq39Z2_ogTUkcVB^SWglBN$eE_;`8W()Y$FOiQ2kQ!h6#j`5v+3cA98r6W
zj8>_-A3_nSbIekQNUUwplzqGIrNz4}_+!#t#iK)8Ffs98tm?S}25-ck%S!1YI8>-a
z&Bcd6)?6CeMtNgM7l+q%@`&*k(Ep9L-S2d(-K;e(Op;fs;fkrP!!E-A%ob?l(Yoqe
z=M6(l&aUzfa;n2tMWIDaH-#x_&qN;J$Q(@#awJ)Qzu7VHl0X_WwoR7*Q6*cklo&n>
zLrgn{JmZv7a;ZC<(TaU<hSyW@)Y~H8J2`!<7OdY05fV&r>W%92epk(`&vy{2_0d%3
zZ7p6?VJaa&bx|Aq`S^{J-3*XS*vWE%X8&biapaaKX+H-i)bHV|>bE(TQW(58W6{ns
zszIs<LV~l`JOpF)@bPdElZjSmrzj*Yj9aXtRhNlOv~02OtV%6dD=^tOq((jetjP+u
zDGF$>F0>35bsqnwBBQMs+HkvqBYOow)k%n3bfjC0GX?~s1R%f%(@*M3uaZ++>4n{h
zfdDSM?})@NpEIGU-{yJ0`y%#sJ|Bt>$~{<+W>9PSC)P-8;>1z<KIKs{UyR~tJfKU}
zsOoE-VDDL})r41y?42lVKM<IveXj0^B^+=)MbMr{A(~;KYv;jLLJ)Ij1d?8{SdZpv
z`GrXNmy&qTD}b(QZT<kiMx|IHP`FPI8gc9K&K{Wxd1D+oV%gK7y{`qyKT{UWqcpg>
z@xxdZ$-T}@W}5K8W&cnDF+KVPqi3^5TAPQ_h(!Eru4)#%l*mAcRRUy@233=oNV#hY
zM~rGWN8d!_zrl)L5=Uo3tQ?`=B3!s_-G-Xx=)dp+r}5s<a-)C#!D+tGot}STPX5px
z^MAoSH;q?5bCDb+P`yuf|5+ol&HkfVsEo~9T3nAry&q+yTUeJxea63l1qAyIL995h
zSS>Dtv)i0Lsc{yiP`Nfe5|SAC{mIdYsqKYDPg99rzFJ%9@Rp1MDz(ogdLXM$HHH&&
ztD1NUbPD8Eg9|)eC>nl_<P50iw~|or-%$kDMyA&LZr%ewTdpd$jffG}&1}lp*LjE3
zxTq1JX+#5@TxL(jV&U;jJz2k&Vaf|({Yp0<%{{do=8y4}-JCn!=i)&yTnyCBX?g4u
z-SKd1{?LQu;A<*vRKcv4;Jqr|@PNzXgDN0Iz<4n%WyI^X5LyqwZ?S+~W1%UBj+iNm
zZ?!>4q^L@uv$bMK<Sn6dne>?**53>W6)8}Q`dUh9m~{sJ9|Nf7K4$Ze&DrTz^}D(S
zk~$PJ>wkt6!;<@jc(9eepW#qmU%jN-Bc-5!w2#NX77p*7%H#ejE@*xkRp<Oe2~c|2
z`A+bAwPwXtVQfr!t-;u}{Q?qQZqN=4?os#7SCw$g-xPdIwM&{z%x($L9Dw!#c71k@
zz*gG(fN@F>b{Rg)n@!zh>3ndF=Ae)QeOL2Z&GEWw+xoj0vGR7U({sKDxb^iHi_4}q
zHv9v@n`w~Mq&Oo9@5p$5=sPJua4)R-_6?yf1}o_%Bq|^6cb7<+VHn4}#SQx=bGYQx
zxR#aMTodB=9bz*2yELz1{2hzy^K>0so5XuJms$$X-brp!WIo+lUXHHx)7s-AQD1@{
zoqKwEJ%0QqkNmdTc6Uyppz-`nR)Bi9ftFxfmP2RLk$FDR{UiKkmmWLNw&VSqTyQ`v
zrzPjp!&c9C)e&V%9Nku|=~#xWO)9y-NVpwW6-DTXYcgs3@!j_LHK;1h#xM38XjP}Q
z$lvt1^_53XU1(2NSFOAFmH%Y8adbzmAB_HTf_qQorN?-f`mpjU<>D(1P@gOQ*!e3)
zy;)emd_3Fk?|WBZWKt3^iexHkGJLW~{C#TFeU!3{7(RB4!9EuRt*qkpW*Y@OTkuyb
z#h_EuRaa;y^ny4{$zE-f`M)pRQ}3H;CeVofSK^)_vmGbZV8K)TgZ36Y-A-lbCW*J<
z-Np`^-;HeC>$7dXP|$nM(M)E6N961?fj5e?U1^#Pfyxu0ml8*l6532{nu3rli>RBo
zR1yC9_%MIDS^yPV9+7>ZZWV^vj^Q+PH#$DB@4pU3hdSccO_MtwQf|x@C_y8VxygJn
zkPoSbeDA}U3pWD(^^Jn#upL)U&rYrOhRFTOA?7lM_nUq_2G<{U494Nr*wA12VZ{Rh
z>c?P6_g4V+7)r*Bt`95lu%P0ab-ucJ;^OILyBW_trqg%VIvYV?7?}){-}sBO0rgpE
zvos8z_Dg0mjP3U2-)+yZ9e4o6Xf_wAy6z+#Fk`vLMkOCzHr0{HUj_gb58S5}Jk3yp
z$wTa;XM*P)0`ru}35OTGf6J)F&TAd+Ud&pok-AbSeS|zh;E9(b6!ML=rbOj3_2+$)
z_1Ox-d+6DdU%{SmHl@7Uhu!m^I;RjhBG=o@4&%-nh*g9Lz)cOT0X7(utA7*|R&abo
zE5GZ<OjOR(NL@*QaFPh`j9Lgq{Zgum`0T5kn!`-?Qc`n$&&*@EE>uLZoJ~Lv;1C5G
zUpAF}fY(=eQsQlwXkIO)03<4WuFB`K6qSvR(N&x4DKpL$e4ax6_0ZJ1(D#W#@;rwv
ze|IsL*mqOz9T~G}+lQsPDRCv%L`u5l5X>Dy_7~Y)3zuqIT=sCMfXxL-iT;ZGws&J_
zyxalpe?Gvyf9BXjvTEp|4+_$M#|a4SsXhtlp_RYnqSIHsaYC#3V5Rd-xg3@{JvqUb
zV8HNUQBfQpFYcMM;81XX7A{w)G8W)JObIjE<^FEOYL~X*ql+q(3QD8d#ab!&Z!1P4
zL8of%6C-N$+f~8VJXVY9^Z}#b+VL+G3LGdnd9`S_BO^O?sB@<6{5CPX1CR_SE&szh
zDAu+t-)U<ypN)~m7sUh$%Yfs5!`C%$m~(uO`0|=SsrX7;p9f2in^R`;hP!13AF#jc
zhVO+%H47Hb#vASL0-LMPLb*O1L`-~4qBqgW@}Boa*0scVdq$LE8n_FeD+c^<S2_*n
z^3)B_SW6~j87v6f8@pSr3Z(QFF^TXQhVTT6okfNkPKP2YAwHg2L^aL%BDP|bVcy+S
zXNrF#hIu#A2L<f!((CWDj$>^|(%ND5PXFWU1rJTfV$!=&P6@wEED<MHMu+>jBn@Em
zkcmx`z7219H7Bb`&b*3Jz2_FG08?5|hcg8@<Xt)X-n~eCVsWAjc&SsBLZ(PKw?IVw
z%+dW@4}H%ibDa^#?k7so-TOaXP!wcwC-<2|n3@hEdQD)flve}feZ>2csQ;S5Ne?t-
zhunhKT8fDSRFB&w1y?)z1zQx)b;GT*@XR}g%>z6Nq^*qF;HR}V0i?<a%R4B)`l&o@
z=K%RWTc$~OCYhu^e%q8`(_m<;J=1<6=JbP5d%7Q!U)Wq@@uqkGHd9KwX_8*DZp7#5
zF;?aEnw-ir7b*ll!d5d*Rexuc2^HFuwdQtA{j~~z<<w|*^Y1$Mfqux|$-L6_lW8rq
zt1<NMDdpCtnveRd6ZFqH<!W3EHUvsEEuRt4Q((-EJpwXlyy4SM@AyE6JaTU3Odj<+
z2AXrfux1SF`rz}Z3=VUwoZpU+W{&UDGP&F@&CItMlBHP78zv&t13cC~U17g*aV*(#
z?^)0dIWn)hqQ?fs**qB$Mwt?zfS!{58#7Pb4nkkNejf<KoMZYkO1Kn?-9L#Pxv>Y<
zyK_M0Spf9vZ#PMk$=Mp7S^uCY!jDbK#R>Pq@<1uoeY3~|jV*t8VrWN{<wIFPA=W42
z*8U?*qmvCyBhNl32XDqgoV=)Z1xq33WYSCXE^(+Dvv=u5%vYt3$?saP9;&8UEbg9`
z!kgZR9<&6PB|{FfW3QdNr~t7fmYbzcs?hrf01mdWI6W0=J`5iDzntkhW^FSQj^q^w
z4Qe?2>lqd6?tk@F6p1+|TWKBuVdo7!Nv#Y7_5!>z&Y{%xgz`3lN$1Fv3DNF27I8ud
z*E!6bmKMZ3?0#8^X+b&Q*qSVR#~ouMPIU5gc12=Hi_2HcH5(ZokZy_E+|(bl2*4+E
ze=n91am&*4Mbtp01d~D6lZ;x~k9VhoUgIpInPkn%IUU@w45WN%1u;GO6!FuN12MR6
z+vaZpI8yAg3*q9j>zcAWmg!?3BFK9qrUaF_jL#tpo7nVc1Q}?3`d(NuHEvxxOM2ax
z&$v;N7Zj+j7W@dcxO@8xJE@BR&{{dFzN=;%#a#TCvB;Xvd~uhn5s~voaPO2<>tAPK
zaO`aNGs1X#B=uoecU-C#`$&68pAg;o&z@B4>^7wnR(uQ`i?9VO>BxO$jatHXsWF?m
zAugWZGbCX@S&>S#>5h>!%V}`@{aWGfBxriM)_s+x0`o;8<d<09+1<AQ0qjzKjUW-H
zgxr$cYi%@g_~dg-RJoS=u`zDG<OpR;EF}CbA!@$CE7PkRI))qUl=75-?SHIEf`W3R
zIeD?!i2-q(v(k55+^v@^z)~BC#Kd8YSh>*)fsBUbzsilmCBOYNFsgF?(CjFtfRIqj
z#gpJvh!aXP?#5JWH4-=o66vPQuBIDHFlD7V&!SY&dyoJwFi$Zh&*pJ6`TsHG)Jj4^
zA}$9c)^gks9%NP0s2gQiviEU_#O<mCgMf1Q`Zpv)VbTj?Rx%uE_P?lTv$F1@>##~q
zLhk&9GR*AdY!-pRZaBhTT|XR};3O5f`kf1ZP0mul1uz0P6hR@89+_+>kHvu|_2;te
zk9MvI!F!_lDU`cV*tyHJUnB|So0c^|J6|lF(A>i@DE)q<=3<!dcjE37);#@h#gyBT
zmP`~rI2zbJVuP;|sR&7MfbO*YWR5Up@zTZg$1>c%2^snn^{8@m70NSNq5Vu6o1LA;
zyZ2)pm}C;7#&BZ(J3^)U^$m-<YOPmnzj>egNn>&4&@Wt-uVjTvEEs0fDzZlmzKCc3
zWMirfH0_B%#uOWsNNv2NU&{$m!;{ZFTWjfs)e*=v>1Wq?E>ybu3RuXZTHsK!jry?j
zx<O0I@4BR#jU{w;%}rzKQ*EEwXz!Pv|7L+4xw}D{a1l~I>Hl<FWoM<{m5)d%IVFXd
zTRI4Bv01&?A0Hdn-V32X^qUF{gy=UC7{nHC%G9YMCul3fXGw=&XR>GrojZC9oV|eK
z=4JRaYqBkGwwWD)(No%l+OrE@CRX<ep6t$TF_^hW?j7)K8Q(W&?IPsy&9>v07~qbL
z{;}bzbdH;()Q&Xs{>`P`5JrULhnPH9%<+rXv-(x|Tt?JKSjch&$II1$NK?~JU-rZB
zD5)Dm6yV*IP=tyN$yOutCt|WMA104&<lG#X{8cGue=^5$0DABbYZJjTAc!*40z!I<
z2z$!HYIA(tA!l#?$ZJvB2(<k;2R1nuEOh4FIrG8|IXunF4Yec2+;l8qr+*Z=`d4l|
zz1>vc5-oHsxE34Avk=sz$?8w>O#S#hWIR}!ZYRR=<4LQYq<yR-iI4Vvn7ioyU^m+L
zBzU@vuveXf$f@A<4DorW<ZPoTJOKZP>|*GRc+An(?r?!N6y^P|BF*mR3E+1FoMmvI
z^h$!6+Z|Vx|2VRunjI9I7{nm{xt>)mX}Bb^bcyuZmwi#M?p^gUJaw)dyu>~JTc8C-
zCq1#`*!m@JONs(iN7bn;CP&?;{gqi%3s0L5)CI26Qb#q?o1cV?8tc{V$CBLIlooC7
zl}7-d(9>t4Dx`y)_gcz?fd!VsOX>@i<P4f90soJCPv+N#))Wm<tc#yR4_g{9TuhXC
z)tjqKF&{21BCM`lm=<`NDZOrfF7+o^TrB?A>bGvPvcus`yrMMmh*g4WoF=eIoi?il
zsKRE;qyVNWsAk2s9Mib9iak=eRT{cmU521I`jKR7QAOBGi9^@}%`yHx8>jZ~V3Ni>
z%R7sdej`b_Y-)?>+6#hWq$8HCZ3AU%r2=?o-t#2qfr6XcnmEbmu7J0+*`mC`U09_X
zIctr757Wm(1XdS?ja7NG)~NR84uirSyC|L}BSIhT^X91aKAD%zVE=prxlmH2DLd7R
zX!>okNOYYa_3PGHF>dZ<hxvLGy(Dd)mXPftyN9IJ3~@Y#OGvwroq0dnOYSB{xF(%`
zJ_sDJ?Tvsu_wKG~CB92=^@g2aYd>&EJfKg*GE=cqHc#ANm}pKbWu}q`oLL6Q=zt|~
zy0Yj``>iG#6T2AHY@K@kNnpBm%d;BYD9&_K%C=0iI=CG-TF!#WU(igvO2gtmx0j7u
zg(&zVanRT4;ZW)swTN^&1ot8ZCX4})+504({-wOc-XLJDLL<zonmmBjH)pIPR3i{=
zf_<uZZAX6NR;$9)_0cJSY*~U7l%JJtr*YErp|gWKuK9KTdQ!wCigNXJoy(nuIs7Fh
zr?-70o6Zna-c4LUQ*8`^>itiHR=-O&cBuZuTK{C{Eq$HHJ&-RFZsq&gX+vU>dpfFj
zN@CQ^L>ZJ${rz`^AZT15vE_x_Z@=-faMSY10~Qa&;{J>>+7WPX?pA|SYz=D9S*?HG
zf9r8f>k<eu`4eaVt=3Qox~XUvM&bJ-{KEG~-{_R`LV?;56Yq(*VJY;Kl`L+ljWVEE
z(03MCf%Ui3_pmem$sh;$e7WF@{rce!&wx+<_DSVS#si^Xu<Jw;`C=&7KR)=|+&4>$
zG&6sbgkK*ma{B{nhMl(eK7a-lG=)GuMziuh1N23-DjJs&4*}@aA2%A-D`!~6pkWu(
zQK2lNI_Sn_))DW$-TDmwy&saQh_+0_*E6b;lW%7V#k8+nRn{afkI#RQ&5{noX<6S*
zl34;0As-&Xc1Ri!*)C~N4us<NB13VH)WGL4%|o`Y&jH4Rlz+{7H!dXwXP3&Fw;t@I
zGv@TJnu(P4aTHS7hnu8o(hip<_Fu^xN$%%Kcp#E_PJen4U8svwLkBqLuI{Wpl&s(h
zB?fV2E3HvKWo*t;JH9$3vj{ulS^O_F_ebVo)L+G-LrHo<`q={dq>l|SV%cwJ*bsve
zu;(P_GJtrWt|~!NACnQ{tRc~wMn6;@r%LJ?HNk|)t4Hsq<*H=sP?c;ZkuCczq{;dJ
zYgQ-ik?;PylmTh19SAmjr)_&2`g}l9h&kv=q??vSe$MN}BHxDixtKh?iDQSi$?Ph_
zd+Y2~4EEPy<3<Q#&hVVb5KNEA=>RUmIU9LS0QxrwnCoW_fvM|q=W|1|s)8rKAdlg4
ztl$24MA8c&?ANBhwXlWob_cBDZRnIIqYAh6srO{vZ`kTjSG73QzvZzOm;s0Uch+-@
z_P%}iVW_m}&a|;=(10)ncN1jj<rC+V(2iND+|aF53Zx<oGdDwH<xF&=6KgyR{7R0F
z0mcz*2)|lw!f73ThE|8lV?E`w)6yh_=Tu1ow;^hXdcWPx3);tOey%m--aX|(OuDG^
zwk<xRK(w!$B6`-JBBssUW8!tP(d-fk+n7Y`$K$g&KH{gMGwL~n(N7+!5Z`E$mH=zS
zZTjIMtBZ;L+)ad?Kn&sFYq)l>LpVU#HKjphAB)Aq=NUW-_$2*M#rQ4gW8@y(Gq^(*
zMlidX>%i^ExHdB?T;U`*l8q0zo~y~tz}o2@?rgm0|L~z*VpClbDp+9oB->HDZGG`C
zi}{Zrmki&hXQ$3&e-EkA$0Kpg5gl;>KNMFx%;-r8`GxdT2)zAcl^{{?0Fa&Iy|jkj
zgA0Fr(>UvaKz6C-|By`Zz_4MLq~kS_-27Y*xcAn1Q~F$-5w~Ct4Oz-NTD*Eju+s)E
zrKiPB-}4Cv3Q>f<pj$V0Cz1$Jiso&j`SI@r37PNk$F@of7yn1ml?O8U|8Zt)2+f%s
zW3CX&kz6y{qFjX%Vk(6seUWoKMl>X*A{C<)a!k&Ip;S|HM{bFbYt9|6-}C!-|LxhH
z&*%Mqoz{QUdn$rd;zN>nb${Tzj5FXN1+wq>kcAR{=-a9gV5}OKr|Rg<W~I?L-*sCN
zE)0=`N^7%P{`n)YF?S4;S)Dpcp>r8W9HV(D)=KkhR~IoY+u+EmHcGbS`=Nctk+uJ-
zVaPoIB{*F~qFzG~T_2z*UHV9M$(2?R9j_Zu%0m`}d$}aV7FZ|Ue)Bk)`83HHDSbHb
zkU;be&Pm_{ysd7++I{6NQXNLYseq}SYQ31m0@KDqtoKKI+w1Jb4YI~`ZL)WuVhTpJ
ziq0^zl3&%M<K8fp%%-ZsUn}4!HOxe4e8E9(chw$3mEomd%=Zy(+YGCh_3!@NJsfu-
z{K*dHEwkBdD)>LFLdF_=RB-+}=~Bk^7*@{HQvmGyk^A#L4*O#J{YJMgIUy*=cO7<y
z2FUC6P!zJ%;Z1q>DYqwGsRxO_9@AHQvJ8(7Z?jLwq;|x{240&U?1jSt)u`qaI%6-#
ziZH#6K(2}pjq9kcu{$NwVo$%YVY)I<GS&8o^vrrp+C14G8f7wQ;qPWxR(#9u*DE32
ztV=Sm1AuQCnLtZfqw8M2a)0Bmk_W19*4qpcdTn`Rn~Fm-l%5=H5!UAB;8ScT{yPiR
z-Nn|fTvcCNKb~|RHPo8MdLp(uB#kofCZ+Uqh5vJomV0dtlVRLlU-ilN-R5kn2P$!~
zF`$S$>0A!$NDZf`(R$tNHa3%LEzZVVd@`^DID3#OY0r1Xe!=A47?GUc8`$8{SCuOL
z?r$s=9yNtz`p=tp4}4F-xjCNNtlvhR3p)bI`HfZeT`<6OArKkj7s8r=L4%Le53DdB
z`Y*Lw#@wTo#Q|0>{J)+V;E*~I&;lE~p|1wGm^+d;<8HUS`U^onepZq;nmWj$&;quB
zXMyBDddID_y5G4K#BQGd&L!vyZH%Zn>Q*0wSudZN-Hu4P%z&Chq6#fnd=S0PIE;Cb
z7RI^Y+c)BRpY|_7)FioBj&myoV|9!?WAf<X*5MVNVp{LTf_UecvA~^%+rMW`x`nb+
z(ZWY8-_=SQmToD<xYUHtk2uCwbvv#B5q@bc>VEtazK0dq*vX@$mf}FaZO}(ZEV^nV
zwPT&iNlCp3nR^9#q)r;T@@{1R*38G{rOfym#8r!P^N%QJUYWVc2O@G$Pn&vDC0Vxq
zF>WpC^%(zQLF=r6{7~0|_`zM?vPu0dJIz6n#`wLd6|BNm&ndorcM+k`KoMZtW%)Nu
zK3iTJn;dnIU*nGulCf>qHv2T`0Arg8m)1auXU(g(ayFo1e4sK;D)Iwb^HzxTL&oWK
z2D=G$Oe|o-@%I}EFo7&fc-Tn?w$bK!C2m?Nox5j&%jV^vP|v>Se=|k#jPM^_lS|cS
zOU7y`ca(w%e|N%DT_*LK*uc1STCB=y@6E~LZSJUYrr<a!>)E6^7!$eymzbJUI$*vK
z*htf4KoK{3N#5_Juq|oIkeFWx*yu<+=TRE@O;<y7TbQ_%D~)y@%IQcY^I+vqFQZ+U
z2vSI@X|(oScsVc2Hd}#eMWijGU^{^*FUrF;9-SIEdIT16`4w=Pgfb158n;MIPf@B8
zS_?b9OSFAFCAj9MA%)c}5NIt^zsnQytH+Bo*Zp(4Dg!5wLqQXwU%@+q<@B)KRz5}P
z7W$uvjnAaZySkBjd=)`A?v0P4_`P=?NH>?odZ4Bnw30#?cataU$or9BW(t3rPDLGf
zknNRwj~?q}-UmP_pKQ(FbNg@<zUT%+pZooVkRNBmMxRwV(+8d<tcl!UV=}si2>}^!
zTD*Y7Gz#f$(wVrOr`4^6dz+=WuAYn$C=1;u{;lv{<OI5N|2?04fd0pgSg<buGgZ;N
zOe!F;ZZh6ojWxYLSlDJWBcT@Vi)+caw2JxCk~}y_156>SoF|aB?2B*FSsX<76_+?q
z21?L$PBVZ|+vVb63fYcq?}(GjIJ2c)EY1k1Z%w5IF+GIkxto7|OGoN}H`Ps0@P#5>
z2E&Yb|5lhojDky?jZ<|VTS5u(B^a%(O9reDB4o%=QAtMXyMM8V_A{#QF)tdjUb-|X
zP+fopQ1nW_*YIBra?S*`VKc%jcx!XgM2U2Z9_E@8R2&wo4NJqICxpdC#*cr@QGpy!
zl9TZbrB;z922BjQpWrZSOZ|#r*9*It$};Yu{HAw3CvC)trujL*M!8_fbBX+QrKyrb
zU(31I7-2tX=B4!2=wtu?BV&DQMu@TxfU%f^RufA+yIzKKB_^RF)u*tNLy>ViIb)g`
z{E;t7)V^)CJ24zc*T-epbmuaX_1-?Vd<@n7XC2X<YyWgI$MksY?FcZ*JS_Su_nH5u
z*36W@o;jv>K{<X;b)IiX6_p+k<hUu84X%WDaV<*0qPBoo{p2YGchqmXHc^cZ@LzQT
z!&;4`RRvRSom5f#B%ZC9BUbPfK37pF+bv?x_LUG|M~U3X%yk2?+8T?PDL-jKPoV)_
z)~GVu3t7Fi-H~*;=!}tB5C1YRIIJQMW*P0q!FHL4b9u7!iuXQljU^=>b)ye(=1((M
z?4XsZF*-OT{oaEb<XzE&(lis9B=~jz-Cm->?hMxnj*iolE*&Y7+};{rJM+8w$=2qb
z9+@9tiLs{K)v0FPRv)Hj%D6kb)o!~aVLa=RBJ2oI$M-I6V&Tu#b}4~N`FIu)c5a*J
zHm1qoeKzf6>%{hX9XT@3@RC%1N+q3_!pst#P6Ygjrr~YP3p(Udj_Buw=8*iq%ij@=
zczkEf!&GZf6Z^gU=f&yRx;X?ONqZ1gh^>e>H{HauEgp8bl&{pEq`9>Sa(bqH$bU_K
zCj%kfVKn2(aY<ssYTwd<l5R=tzPX1v!)8PTe2U&0(PpuKca2lJefa1ss?+Z1pdt1o
z+K)TNF3>u!dF-6xn=qb@!h`wYuc|#KE?gUZ>J@#v;8z2@%`b3=xEb>YrWqy{@<md8
zGN;iGbBn3^t%LSUV)Bh(-Ovv(R>JU+4}AMCkMtK5xZky+U>Bt?93*}#le)Bd;YS+l
z_Hoo?FgVXes#nf0WgSz3X#z~uYf{$y<0*Q0y#SWk^+;+t{bd*?TQZHs&!g8rx0x<g
z&g49&Z{_kLo{~VXpHr{b$3pAXY)MqA{lx{eu_`3m9>*BsFzT}DTQM~DM4trQd&|^v
z1v8xZBv6l!t|Xa#$Nt($^!UkZi=H2oZGR~q*IrSNXd7`&V`(@6<Ujr&urZ#0o*l<g
zvQ#so%SdqB*SQW6YbNM7LOvF4Bj(_36wyXmdipS}KKU|o@WpXMY5?iL$`3JDno~0T
zO!B2Uz6XzgfLxkc-vfhK=NLKN$>N24vtfZ3ORUqcAoJ*yn>0u0FU6(VraaW3%XrtX
z__XIq*!S_Tp{z#00x@UKAr0QcsxJ13=Y1#AH*zUflUcM}$qgU|?S_6N{1)Gbt?f`k
z>u@OH?O9hp1wJ2lTI2JK*q^gi4_}#9qD^6xIYuHVP|W>U;014Vn~R_MR1*UAsq2}9
zD-C!F+IxagS)mthkKae>5(+MPM=f|jjiXa8Q`YsI0RB6$!i+ioZ;UZ*kmLSOa!7&d
zlnQ=kxFHX3Yr_fW^l>5t*$7I?I^35VmFJ!uJN{gw!!5GSYTGmNW!H{``nZ!$Z{yez
zh=e%m93E9kmZdZ+1vR<c{6+_D-1N2aYv3dPdL#476yPUZ9ag!wQ#HdV@mxz{d5WdA
zTq~AZH8`jDb#Yf5_4#X+{3`f6&#_H{1<=}hcs1eXsNlDEe6DL5Mvrg_>H4qt`k?W5
zwINm0H#*aY^ze8Q9db}E`WnWYxclh!Q0q#*=M%9gBI$AkEBMStKt>ZI>TKc9tPd(M
zAE<!@F)T|(6^0SG*l+^rx{It5Sjh@hg=qtkgS+u2C5cnu8H=$c44kw;maTG3^Q{PF
zKTFj=HhDNf<j%Bxy^MR?b#_nf|G$c%!@X(uA3d{5Ny^|hqJAf<|A5$WoY(CH`dYRN
zYWjEyJJ$1p;KuRiHNA3|{WOZK`HC-VIzvpsSXMW4MtMKqn4&y4CBFJRQcyxs2Gw+;
zC{=>^tDHWo=PbWkqmve!dC{b!RYHBTp{yys_Pj}X7FrFad8Dq-^ZN;f?^f4vc^|Kz
z@F-&RzmA=#eDc@->tDU`dA)OW+wd3l7twB6WHp#SpaqGR=?67Bw5!h3arbKr7pp#E
z6&|q>jP*^~Xj+Q(tIc-rXnJ?N+3-5d7_Y<q64iw5mc@E8w=%#=<*q1EwuN`iAlF*n
zAQ%up{lPp0>TCVCA_+Ir=vINwg9mnyNkOa=|DNo?Sz#QLhggEtC9-uC%;5fb{t0l_
z0|;W=?oP1-)gmO!nY-N0L00ZJ?m@!-gbUq?y;Tu^UpXUdQ0c_`<B6rREb}Akr;YQz
zriHYAHP8*|l4w`hwNf=-^zXCThiQo7j&xm_j^Ce+oiWz6wGD3@`~nIHrq8B2wrU*Y
zc3}AH{d4%U@OUb0=g7HPjiBEzgtLF{-IfI$zx~eoc4u?er=MRfyniOj5L9lgJ>c-H
z3>H1cZ+<9g(TOhm8tZIq9v!>ywINdaPu-4g-g<y%N@bltSi$>RM@6&`lq}M@^?<*6
zE=%_J#O4F3jkat^+Z#`YT&{x3d#M}Wyat=|<QvaFQ`4?g9K7~MM3?IM>zmEKlzrTO
zR)jiC8Mx#G2AIbAB-F2=WK0FC_9h2*3}uNJyeyfAeYI+LPS0Vi!aIRGgOI=<`p2?)
zUjX@{RN~5VB|M@VNBEgx4cqTjib?yn7k6nc85YEqbyTuBi0UL@>MoE@taE(jV<WV7
z#bvl+IIU%$oXSn<X^-1?GWI5vL-de1dUF-<ZFa#1t39e)ayng{Dh4aHG>bv!(eF?2
zJ;H)k#lGhyD#J-y@yhtJ*uXnA0_@ci13LF*-72rIb-eYsk^p=JYR!U=bu(++KF!Nc
zOZqv(=KA4Qr?78N+@doM%#Sdsk>F}jf@JWP8jgTaeMOg?*H*PP_m8NraL=^e3?Pl@
zP468-d=LQzKV8a|E%7v^4#ZOvmGnSVI_14n;?ejV<4bfC3-hU@u?v4YxvNui49i4r
z9x*xf^@SB?D>=5Q;``d}Igl>QY8|~BCtta+b=L{i?cw-bp|u!*=H^`|eOVs3Z{FHq
zA;B0mi`9u_@(p}lh=e$FXRZ8J0V*wKM|OGq!<gIbF4<?xYw&P>`=UFQST8$=ITmz7
z*gke%8wI;+q;_s`H*aDFWTYjeD^5L}ewYPQlTY76(W+j;UFm9;I&~4%zjtXd(2L6|
z_r-{^Rgdx(D(G7^?@&^@R`!YVSQO%FzM4<>6NmD1=3$<d;_18tU)zrXVs_Afp*S=t
zJl9b{bJ4j{!2X9q2{>qJx{4OY<ZV*~s&r@lxOZNm(m^B5Gtf0|TZo%Dg#)EvP}F0+
zu<BpnH`gJ;L%!q|e(G>03mDwIe`7ZZ)3m<B`;<?;*Ft08^EtOe(eoG9+=6T=ShxkC
zUl_jUNHpZwbg!O+0+FpLEgL*!{@gx%XZj}R`<R2bWOkm@P64f_PHSSp+WXJ`kODhK
z#nSE@^{88dC0o4NP2v8$i)KIQgR)(5frXC&o1%WsOPiq3>9_e%EaS9$d}$`$Jw0O6
ze}tKO&?rx7cUYugKkWu12?L(S+?g!ur?v@p0{mGwGZ8T4CU&BnwrG_1M<Bu1aEDp9
zn@>&(?@jS(O{NSoLto%;o?G9|cbK!`SuE8FARUaE$9RmWaSN8Dqay>2VjK4aW|X3N
zImOAg<~^2BRk@@cLf|9z`M_<qaYxIu0fBdMRN;x^R`=dMYtB^o%J!F7rcezSqJ-3|
zNW4kjBO$;HoctE1$M!?wu0&@kYr!HVOC4x=N@Jp}PC+)Xl>N;427%<Gk+7_4Etu6I
znI*YOjTM<Vs5eIZ_ZEN5%m{EURc*>8OH3P90*qViv!7i>FgnVPEOdW6Plh-sK(@91
zZFcWV)AVZLOF7n|ujpGr&eO2M^U#ZmU%JA=W7<Y^HQ(?C>8lpq^E@7Jcv+cq7!>hp
z{f8i3EHxwv#Q5z+>_&d>ZJKqydC-}Zq)+`xvIQu~M<>o;reVE~Sr4e%;`L}5YgB!Y
z(Q?3#*)f{9G(|;YZGC(|B5LczQlv-d<l~Ls>+Lv&Z%UQXc3JR{uIRSyTx^Vjh$9It
z71G2=!)Jx+z%KAsquJe2S=~A?SAeszP;zKc2Kcw{AS`EpP56^fValzQFjv=aDOm(6
zK6T=i>(s>ep26{iy)Kc?+J`4z)DQmQi%F|(#8X}fTdxgLzfd?Ys=C~zh)>DZJa$KX
z7)X-ruOy>WD<R|sGeONPE+56LIPX~7j_btn=ilGdIpqm{?)6`num%nDd09orV4i@?
zlYqHDDk!m|NuppcE8M8MMER-rF5K5Z#ln<@s5pN1=A@NIas1zNy<A$85>W^?`_{5m
ziyreZ0@B{I?Tpi1s=JI4QLm5t;jHT%nsO`MI)MY)=#g$@;g~E(RbR{3w<wD0keWLv
z^PO4|<%r@Q7c}ik4ozSob-G~pB9O+sWe&OMUTn@_CU=*=O@GpOwxpWszD^b=3x{6w
z<1vkWZTJ58+4%C}1!-Oe)QnSFVTCE_c>~F@gzA^~W~F_j9-hlB$&U-GOkCdn3G(A|
zn+`knrN4RD7q7){7O&1$sbZ~}gNe+c=aB~K_fe4O7f5cw`oB{Pbx7_Ufclr8{P@|G
z%2#y0dB=6`-Mmb$;nUGCW#J&;A5tOkHpGTwE?AxFCd@i@lTEZOHQ+qhvLXzWz^C!%
zkeFn*EB4=%&{j8oX14fW8ezx<W8ZX@AdI}O^$#9^!Z#KOP-Wbc&RLh`yFzO#WuM_>
z7~)djmh5-z9KW(iBlc!iye>@QpiIErm7DB%r-`-AdfYzrVpfqZOk2LRR@jW-F5=a-
z$rW8YaU_89-~Y+1``j_fgO(G`bsM@cC^zt1?lC`l=lq|<2`4euXFfZY_`Cx9wA|#b
zD0fMK^HYD7AOh@z6{bP1jA$z>^w%!w7Ng<87F)xKEO@wpv~qX@{>%JN@e$0o*b8$7
zL#g8yCDL+hR3vx3-fI{3g{S$fZwiRD`yE_;d`RMi-X{^$8+w+rXtIFkCuylgLsuYn
z^7?b#>pL19(BM35UT)}*%b1UyjS881>r#`<JP%zPMg(iY;0UwOI}>skR1aE<U0bPY
zIT}eu{?e`LZ91kILc8pgsAOuPdC>5A*9%CmqUv#(>ytlf6$Vb2gq%LuPD8CCCW>vK
zoX4gRcA(Q+xfdDsp}PzA(<}8uQmo5QT#C{a>p$|T*M5W<#cEzox7L(Q*VfeTHg++k
zU3*C195Vr@Ig~$m#p?LAHX9<kPhbi7bBH;``isJOLgg<QddKoT?rpq%VltPc+2>KI
z+h1S`IgfK4*%;4qG=OOV4B>y=0oQ&a!)slhA1n)p?xl%e4L31s_0Xw0tekPcn2|@N
zt?E%$6zfA{X^&7ngG<Tc-2-7odAO8Gv7_B$VR@`kVI=l@*n5z>A}^N8<(q$2Iyf`*
zE>?Dota}^V5~c}}+<%C~u7wGtBb&wgv0#d4Ti8FEqX$$KLjq*Gqi~b@$(RS%XD3b8
zQk3Z;Lq_p(jjvC($o~*3NP%PFVdJ!kNAREfjkj$WHZ}!%F3Bsg8kH??$7f=Jjm_I}
zbxvr;>!j|6@^d-c@M*Q%?f$=-?pijUhw9!A^~hTfX!mEm-i+I=lKK25bmXANxEZE%
zewknY=a`;%$Rc2Wd@r?rxnaNM%Sobsx!4hk1JrcW>f56Um>O7nWS&$xZ~yOmbI&8;
zOZU~Rx@GM<(+&PRn}>JH+kbA&w?DVQHSO~IF`x10{Folpr!ZBC?DVjxq7xUVVcC*}
zgo77MpS;W}lq}LVt%(a{^b4VP`d+TIDLN7)Dy(_!=JcNeW?H?iAd)xc2wk4+<MW0k
zV}!r>A+g)`4F_AzDnozoZycz1>}~k%d9o>PK;6=(;gQJN2JZ!d!axS^i<r5`8%W=U
zLi{jn*Yegssh1#k(%Yi6p2ELK!ONZ#Q-{d^z`yXius_8oy>Ejcylr;Q*Pd1Ln-AO3
zv154c^g;lwNsxjnN{Y4Jc`Y(!+x(*S+Sd2(9_wiy*Uj%zJKvf!2=`m8TB3F|B^sg4
zl?wm1OZY<o#BgAZ<>D!*X<@UmVkoQRSbOBITddQXe|y+#M1Ii}^=`KU5D+iEy$MI(
z$LnmI-exozf?)7ILD4c~Q4kz<b-YP5Yt>IiK=eIYYLI*LAq%N(Z$nNr1HRMytJQc1
zH8h7mJ88~xB)&+{383q;bT`<DdYaxAN1#I@0$QC)Jx7~yIk6O9xjG`u`qR8de7F&*
z(W>)#_?^QIC4!8dq$Kg-X%zph9}dHOrta=u{xSM0m!@GED+}U!9OLWfqjMG$s6aX(
ztp>?WpyKImjrlFv4aB#pR9l7pTZXcvz==b7mG00;pV6nSnPd+rb2L?GaY-g^<=KbZ
zXTAK#`xVj_RNv~<^S_A!gA<m-xTBU;lfmpj-iaUlroo0%6Q}U^Zq5YC--*+?ae&lq
za>UtP+p@Xv+zjJ2KM!cDAIEo0HAw)*7tJAV&o06U!gYHDq$(mgVO+B1evo^j>C}kE
zsIALQ_%4P2cqW)XWv~Yi|L^uTQ69{M5vlVS4m67`C%ooI_{DgF2LWQfY=6pZ1l29#
z_<z3OO;@9+>)vD2scS}H07FQIC1O`Ty3HMlYL7CLU)5xi(qQIOW-2H^;R$jTG>6!}
z5gS5_s;{waD|h)uM)a5@Dhg4IHaS<&qhpt|Wq+6kaz@)lJ4&8wOX-i8h!p3Yj<gtu
z#J9oO@~zu!tk*yWyuV=a9N1NmQp&ZR_wm77xg$2t1=|9k#ph!Z#(qH9j3d2@%C_H?
zS>^W**K+e_%a@|Q0xrtQn2jK-0O<aB!NrJuqV@X48f#`)6Fv6A!Bs>k`A>~KVlJQ=
zUko?3cwwDc_W+Bgv^QQmJI-?i?0~c0jDC&vI+06XV_!hQLNC}-Rvt1*gT}bI+()89
z+gg8TqT5h1x8GcP*_FytV+8PW{v|t8Ua3Re3hA>=zfA^!MYB27=dMQwjm9|#iMM`R
z5o+57O`-s*^_E9I?n505r#gD$omHR?Njb1D`2hYms<KH02fc}q=Qer<8LJL7<rAGy
zyrX9J0yyEPjaJo$m@z`no~u_8jZQNlPp>=wpSN0i%riQoZbeB2)kk0g)+*T4_>{f_
zF4E3<z|Awtk<G8+{e+YQlpk6Uy*ltb|Bmu9TsD;C-g8*mu<>VtHAeI6Ve-Tg+cswB
zLGG?C8_4SM!5lY+Q6cEGS8yb37Q$^+z^@lId~e(8CY8Cf*q{F?o6dN5lYDBndLI+k
zuJ~zb=d8{y7(lS8o5m78fZ|c6!0A1*c|h<)=W!OPMF|~x#1XU5@gHYT8#Or_1-@Y~
z_!h^H;^@?3c$wwH4QM^fdQS;$?pe?i^<7NA{6!rzB}w-E^&ozl|0gB%a|>mnZNJFo
zc^p|#Z9%2tqgekxC3TdXvSWGQwiYfkEsG?J-&X=-Ds$W+T6UbwE(h^qw5JZx+eq0#
ztf)y54;zq?*0fQ>QNipWeH{AIl4vDWSAw{wh#H%HopgZWQ<w7K0Pxq-T(SAo*?Z$|
zsP16zEN2U#FA-Ma);0+G<|zMF*?1~^K8v1|5+g=6JpOkrA3Tz~Bj*X|z9~dSolSin
zK5H-$V!esJN*1~G#?r$J2b5D3Yi$JY{5`8l({}cbpDnCjP$D;MOoQ*uPZyaKeI&?m
zdRn1AeZ7nX@wURWYb1<HHJ!WO$@gXno!8Ix`geuZ=oh${iXaqv0X#OB&Oq&(;H9BL
zgJEOC1~(X@B-YtZD}tIOc`C{}IfQX|04{eAM-Oxr>sXW|E&_IBq0l>a=hEK!3!}d@
zPgq-o*B<y!nk;Pc?nZCBKfhOBa_k$o?=YVR$s)nA3pz5P@SinF`tPq}Ta=XUe(ic9
zMMMXuv2JbeiYm|R_08%hA2k--ghg6ExtH+7|EG@|spXa__+_hx)lXcAdg;{Lt_wzw
zIfJ1tJT>lcGz_@b{DrnR<d(Nx_!(L|w6K1PLZmsS#n$#*O$Zx9g*1di2R6XzZg#eY
zH-VY?va2w>&1xT)(IqC@j{f5W)u@vP!+*>QdD4ZLp*B+bE18eF4g+m2!^zI?kGtJI
z7<elp_Im#sl&xyHQiiSYb3ar4=5_31z-#tm3kZHS0&rV*DMc*|bkU1n`TdWLeTy(Z
ziJ-$cjcE6!N$SnJaptt2BHmLAwb%*1nrjVy?yV!_$@scHu_cyp(--8b85Mi@*9{bU
z#GA=gzl{3sQ$v!ScoNJ!6#Zo*=FH2z)U9992{QxVSBHF1h#hfM-q0rWZUZVirg&N=
zdW2yH{K5J*Z$h0?Lc9DpIlo@N?(xZ_fN4lP>;b{)uPGQ@BOUU*Hj9VW>Z#P!qcH-f
zb0rlBi9`rD$>RgJKzo^FeYKT4q#4L!B-Hgi>#`D^pZvKL(EY4d$jv9voDq@I?K;Oz
zc$*fU)E~9h<d4&3Ih$MvBVmM=<@4h`H-Qhv>?bnZ#>^#z*ytgpP?Aa7$rh^V)TU8z
zRKWrAg=nd8{=}ol_D@^w5kKo5o4EdX&IFgN-+kLEUse@CZ$0FwKF5!y4~P<@k9nJZ
z^6|D~R~^&nC=zw0*;`J(D$KkIiz@tT{-IO3oqP};hrenio3ga;@Gr0H_q7vhhBd%h
zapn5Lm2^eN7J=O!=TnF^i#<-PZ!a~FA@ndkx+&Y5XAHRaXZvUeLM;5QRe@{^0&G{O
z^U3)s&uOcRS~N?pK2byQZoIVsS8p_BL?fuh+*r3iAz3-nydE>qM(4RTOwJeK$bQKA
z9pyO^r4S!#O^k_8!z^BVENJ+p2;km$dp<wL!W(nD&9~doq2N@^l&?Aclym^dUm@mX
z6m=~$(&QDMUkL#>p%1cGTsDzy*H0HdCBwbHJ){xtZtqy~>~d0%Y&35OifxTZN&Zs7
zpUk^Y_|1`>Wq+QGa!uK3Sy}tczt^K9f2mIA!X<yiFBqlz?xpx@i!gSt3Xmc8f?{}@
zZ``5$P7H1I_bovBLwbtsdnwuhGxY5K5grAe>St8j?g?5@G`^Po6KXU^oiQP#dCD%v
zM-tR7gB@gl{BH<R{J=YM;f@Yr2^RTt@W|eJjmSnFRjb0GxmIy+eG_UVNb#0fl#^Di
z__NLBvABR74aVa`YG1<vcHOa4jH(fwD4iw!cExB7?klXPPuHhViLi8D&wtV<+AjN?
z^onkj$8HQBL56<e>PgPJy2$K&(47!SD{u;rgF97XC4DbbDF;n0ZdOG3#O21I#1=W(
zYQ8<xEa6gnPp|zBh}i>N-t#Vkg?G5Q!N=6&{`qo_=I^Ui?F0ajVZgV%5PI8+JAuXD
zXh^@#7{ygYMQuh4%?6WNbe0}hX+hBRA(8Or8_hJ$mdp8;gewWi@TXXeW9gyE+m}~k
zo0*7as!K9W>LqA!8u-@jfiE}(1vPC9lbHEu>w!@-Guy{NT=Y(<s4vbY@c`q(TmTi)
zq$*E*P;)^OU_r2$kLAXSdsP~)<*3OkZULu0gI3(P@~sf}=s3RO5!lrg>OzvatBG=W
zt(@22CVJeVwanV(i8wetUQTc^UQTmv)npIsl39(5{vfuoAl_b468g_e?)LmDCWwpD
z-B}$LicYtyyBo7(FV1b|odp%W2sIUK&`-bGaLx>vRH`@?Vn1JeW-SS0)>l3hR*5K1
zkAJrya@DCJW4CJIUW4}_@A!a+VYq%+Lc33J9Xz*h?8uwC5h%ZA@u|QDIeOKI8OvS-
z)QQz^i-deH8yRZuQ@TE8-Ws8WjkU$&bSepWuvO$;t(f|7m?Nxgy?{r0NzEe?f@r?2
z_6eA_BOlP6_I~QV2mcj|nyeEi`cAaVY&`@^H0)^#Ry89nYRV8=s*}L<C;2nVK#aYj
z;G+0jEh%ozS?4(m<qQiY^j)HMsKzq3ohQ6-$*SS)u7U}pn&;V=pOk)ipXRSQ)MS`Z
z&c9*%=JZqza1|5czM1IV6X$RN`Vwv!0yZC9oaWh{J<HE%p8dk*Su0~w+L<a&6eaGA
z%b3&6XzFbFdX!k_yq!$-<~H9ov3!Xia)h^rB6C0Ptiz=J4FgFbV}$(WKbeQT%KQdO
z>0BmY^v>{?vi|7Ffxp|oL1sF&E|hffvg^u*nT)9lYEtFB(2dU>gQpU#<Ntl!1`->&
zrPHGbBlkWqN3{EsM{^b4gqh5Yje9CMZEf-Cg3VWkWy&vC%0XDwH?Vn}lg{rJt-Yx<
zo-4-P6-VFjD97cEd=@y)YP2UW1`bO&Zq#O5uf9_!|6B4RQ>f9PY)Tnu`z}YG*|pZ3
zp4-Hd7hj!8V=cOorwlHT|0)3<sL3JEd<+&-5J_EDLSIzqZA#S9ZN%G7v!ZS%2sG)*
zuMWATg_lesA_Z&0&w$~Ox~qsl!{h@%UwL=j;1yin_ZoP(Juc7KWt!FDNM6udwQ9`r
zO7Lv7SA>{d$y_$4BB+Dr_pCQFRuITX;m}UY2j53*!Pq>u>v7%{AWf=RVd5(IpU(L4
z7Zf_=)jEJ|F;<)Y_L7(6D|k-nrFdq4igF-NvVP=9uX>|tcgm-gdQThochTq)N9Y>Z
z(<-SE)I_VgDl#6wM@surJ~+rT9(w<P1ld+|xP)S>1~KSO3kUmjSB+GtGmX@PV(Q~A
z>!g)zg|&4vD=ffg`W$XOqnyG^Q{y{X2#UBC#My*fF5cUtjZb`yGw88(>AFklkCo`-
zj~{{9k2HGLYI+^S$G^_fO9#!{D5nvD??}og1yQsqJ8zcpS+a=ztH=6o@z0v-tP5Ck
zufe?6DJueQ<*ZgD^*($WoEakp%{bNt)|-;QlxSyxWl$+)1ns_ECGNNnb^VtJ6dJXN
z+acz7Uw+q*d(d_KlTLJ#-Xyb)FYH>T!lsYnS4!F%=_uZX@#R{$IlK4D#yeTW+(WhA
zMzu=?@dhf8XXotAk=I;}3|Df*iSTj|*7mHQL?8LFO~x9(5S2#Kc;ew3l9EGSb~A@D
zAO8Z+e7_DJaC=tDi(C?Ojgj2_7QouOiW1vkje7mfq_0Zuf0Ns}n#O7rBBza8?oSGL
zMsr3{WL_|Ad-gxAu_`YfN$HvxoDGXGYeVMA3{N`t8tWau<?ILvc%{t9H=TM#1A7iG
z>x{yiLHHo|Y$uw>p@0Qm-5E(LkES1K0W&47;JtZ4{r~zZVS2oysBijfOwM>@yytKJ
zko-?fXL|L)_Q{bWHL<Cl|GH*{l3Q+3tq2eQ?w?fp1=HdbO(|9XabGi#MIDoID2Tsb
ztpD_)(i*#{kq(tB&hxPtgv?Frcy!nl^xp3p3BQ&{c>U$ABB#`a+!cm*H;_*Uu3q4a
z2Wp?i<eB~4Xf3AtaUOK3NDOR0!M*Hy2CfwS&SqM=t+z`fh{!WvVT+Vm$@M5JDJ|By
zx3><22aB5S^{22rjZtG=^5Jbv(9Lm2FNx(QPZnuBiOLJ;4{E&{(WK{v^XJ*k8$kM6
z)>!!5LS!^Kh7qbS7qozlt$ctMrp+oDXI$nUuU&!SsU8_dQ&!b0ThV25#$S`28IN52
z4DQ5jZ+h^$2)2h=i25mf9=1_qD2Gz6>?Af5FEET<SomS@n;f&@sf4CiHebHh;k=%`
zrE~2+C#AWS4N@Q%)m$hr_1@on;;XI$X{--7$=tnosB*-2Z;5KrXrRpr`T<CG#yf1&
zMaKQu>~ee5qb-Y}>XL8Va<vjq-KvMhvM>s$reXgm$i0p90m<;QZVi*%;@n*h3+aA}
zko8a`cU1m)sq19H0cR0V5WFkI%jHPOd<c(7!D7tk^5d`ff8xSdle&Tr68{l@UR(3%
zk>}pG&-#c1%3?Qx@js{xua^ndL{>4_3}E)#t&rCzzfi~%0;cMNGD()dQG9B?zc7ky
z_9^5NV(zYTCFY9_EiSDjYrf&OmO9HTAyhpJD`36msfk^SN?86<Qim+Fh$mbk&o4ZX
z=eDa^fPZqI6EiI|-(~vzmU>2cr3$Ii9bDiBEspqm^#B}IT@v$-{qSfSVb6OP)o)E5
zl>6Asi>MIBd6fbDZyu$Hzn{foWOh_C2+h^ex8K9rkeGVbGZVzq2O4Wct@ZE6aHpqZ
zN7+xHoK~the!DV8!kC|gdHun9OU@j7cy6_MI2u%=fSo?_k-=K5j+104bDnwu|DTqw
zqApmCoJA|NN+&drE2R7z1gcLJU3K<mcibWiq08~+A(F=4rDX`Oj8-KY&nG3;U~<>2
zSpIg%JqpzrfHo#79i8yGl6t7MR|dg(WInY`C*7(TKK-^r<n%v<Csv%674zP;s<}bB
zdBuX&UQ6Fe-Rq;ISnOdoSeWscKjW(N+))6v_DZ5mkid-sI|J`-=cc5!{LnwV;c%HM
z!q-@T+A*R!m13mu`Sg8lLK=w7r6)rQ-|28(?_!jJa~^XX%h;1k3+cZpOM=})F89Fw
zB3)<e!_97{EuFRVk=KZWE5eHTP0ntkIzmKUk-IfKZo^VKqt$o;5RX#-=*n^g_$HRJ
ztYsd@h}1u#*xH2))^{_#=h8v>feKD?G!_gc3Xa}AF{zh;QpeBd@kjH%Q7WVIk>;ao
zVpq2O`IXs&eBt)ZTNcjep`EqW;S?E~r<6c=?cS@xWggQx{C>*~4kzx=_d!E%y^e#J
zsz+M9tEJKc)&=b0Y98W{=K_%0zNQxb1^I$wwWFT>-y|N4)+YhM)IcbicRTV8S6zop
z-i%5D+&6w-i5lz8SO@d!{@EV0arfSiNk`qksAfna{xEGlWwO2|p`F&fVo|TR&Tam=
zVEiMh;8Sw!Pu2f3j^X?_@<#`fA{I~wH*+Lzh?Na4zd-9L)cIa~3Vc$8M0ew~=K`HF
zL}*q}`)i--v>;srg1&7|-;7J`ksGz!tI>xC1s-ed;9I!vy_C@0QHZSnm}skuV7xB7
zqxeQ`s=<;fk#9JrG->0mPQ19DVR1#sk=R)MB#)<#F0x$zgpYN~^FwCy?8W8DA5l%1
zopD!b4KJMbFWnMoH}Lr)n~=I;PL$9RTN881xBm9_n7NIzWoOshY4^dg1L&O#n&HJR
zOSHlSdfcr?7t|x`595AkK0HBih=1s06Q(y8Xrj-ES?Z3iJ46<A{bQ0iIkEMyCu~ci
z-79H@6`%Z8E}ywRtZs?*y0$mc+5OwKAa3*BdhD(3oI>h10u(4%(?;0uYAd&g5$oY7
zAGVnWME-$?()5Rwvk}}4H}K6`x|2J%ip48V{o9Bx#`%72UY4sE&?4#t7!v*sC?;2o
zh%DBRdC9LSrTe{$f-d<nA$#d#CZj41pGyDUwd9l8y(hdfdym`y&aqd!ZU0P~Ne398
za>3Z}?Kb|c(IfzNGtd6pL3MQTshF4WOGvfdx#B)p)NFwrrJXlq?U9=A5w%bKrsQC+
zv9qwTnaA^*bexwcRallC_E0}C{GVHN8aG)uXtp02K9;t8I{d^~DptT*AorplPUbr1
zg}g#gSV-J&{_3&Dg7oWr+0BJ6Mcd(W=5tZTyS?_Cb12{(e>6{rZ8X?Ein<eX+nT0|
zzqg}ksCv@e@CN<KhgEQ$N_CEp`jdFzkInipmZ2wo!yyIKD+ey;8I2yO(#D?f^+Tl(
zDM85P?vPR2I&r#XW5Sv-bt9aUn-*%F!0(xc*;pHZ@AbH+P@bqmoGRtH9d^rPDz|Q%
zfx|X85ZD%QtnPAwg#jx-?&zzL7|?2jx%jly>%9!VStz^~-J*@)PU~1gVheBteZTu$
zAa;_Gbs;U>!*zN#Y()vJtGpJW(;3sM%<&%?89@zFJO6)F2_KJ?IaR>f3ru6N=9R`j
zDMM-`P<>&i4;p!{j76G_rztzmYZWX9{{zD4={S|Obx-E*muq0u)K`~*_DiutJgA}B
z9g==z@O74NJ$+;RkKE1GHM@eS=BM;o%^d8T6@zj?&G5i+rSS%osNdJR<tWZlw_p>+
zldv<3pa2RGS3cBND;iEg&Hrn9;{zbTKH4)C-k~8zJo6G|&N4fWX?91@H%86_`;$>q
zvt}wLs!yHO9UH+Wh1n{eu*b(|295vz-F$0Bn0P5?v(6z}Geg@S8iY^u#k`uH)BF7O
z64r|^<NxDnR+o(UkSB6Xw8As^VDX*(HomcEpl{=9TKL2VdcoL8Z5>*Eb(NR=r(`p1
zk$c*Aama@!spTleics^;nn?rZcHKCw*$c37zn35^{&OUDzimHPTw!+BKx2#6nS-)D
zE1Be}6eDam!47KfMeBI<`d{KHV4ME5!US9Y$2!}Z<G2wCqO}(v!o5qALRMTP*%8$H
z^nyZrQ13|`rse2%REVF+3owP4D61jMKxxltlDAt9p^54LJ!tDczzsB=7=WhLyMB32
z1(1PldxJhU**Vdm=<_T{^aA@<ja}~3FlHQl)V$b!7<M;G=ONsAaj`!SA8Wl9J%$rl
zS{P6&b&qEi)1mIFo73R<PXMpGe<mrUmc~;Eh7&2>xs%&;6UA#DJlS#GLp5eBXlJKU
zggjF<8J2kt*ih57ac*HMlh#O>CtaQW*W93ysVU;!xk2T$??oW`qB2XNQpBY5@~02{
zEJgMCcqU<-_jrFd8d?_gm}qqeV>R5kvY)=T5G~Go<dY=#%cQc!OGv}nIqBAf;vWPh
zs<G-p*roW3gtC{`%&|ahj90>6J~;^PVOP+I>$4mnOT6A?_(C+B6iZKND7eREjN@<j
z5!@F8D$mmM5LsNCf(>nx4Y{UMdIRzxsM<CM7TgGo+M*P+Y#PyTS@7_Vv&m@HJ-DxB
z<e<VI8&2!4Q*UgY*7FvorMS@~HzC@4^2f|Su*uT=eNT$GHu#n>6^36=miLOpNoaNZ
z0T&85mpYX|q5-OUbx$lZZboTw`|fMv&th(nO_%b10$*ZfwhRzZ@>>@1^GSizNzAX!
z1^meKVn*m|0a^0%{b6e_B)@q}&MRE8p`bG=@#orrG*a5Pl$IBY<YmnjTw)j4*R8FH
z>h`hsj#qV7vfYtT)F}$F=e&MLe8T|}D|?u>45R+=|0{uTu1;Kmo_3$RJ{<I;$)76N
zE<f!4<#Ad?iZjJP8=`jvo7i<$QqkP;rvyBz8ryx_za0%W@y41Kz25dT-mbY3E1|^`
zLUWT0e~l%Kf64p?6FqafC@ljG$*>HB={L9u=F|&F_oc-mYX^{RKQ_K%)_@Cd!IyG8
zBrP^k2g0!k=>BS}%(;AyysX!V<!hqU&h;A>6%E%|{a+iIlOf;QfRB2OAFf+@RP!9$
zxNoMlL)iSGg!ZwAuE~NKXtrAExxY{wzEVW+T@v5+&1i~AOODO@@ES@>J>4e;K8wS^
zy%@i^n>u7(&$!Imm8ZZuJlw_Y16rfmJ<&cFC9|`KQm}GQqS>q{O*RC8B{a#H!6-8S
ziN0X(*5@VsAZu`mxAK?*r`L_~GaS`j=v_DX5eg46I#@i3lvo|I13N8*YWLV~n&kPY
zRC90}TR4ZS#AnH&1W~;nQeZ%jnPi24^3ngx>XQ2Z#2NyFe_$UjrHL&k3?7Hk)OaVB
z9H1IIIMLl+tga_?t5d1Mxa-V;g81bLm6AFByn!&LtilPEsghWn?lkYls6yiCI*i7J
zNEi1w{UOVGw?>*5!hVI4nF3=Vl9)jj)ZYC2wy*d4Efl6@7A4-eB*`-9^nPgZsQX6r
zU##dQBcSR0<3hu={xG>KEPuJb#3lTZyia=&?}}vE_c;JU^;NPy-c_wvacl2U>6$JU
zx~3lUA8SZsJ1eg1t$4m^Ii3UZV}gx|&G$+kX@gq&JN|n?TiKN;iRx$g<;mjx8Ic{u
zX46)%ip!CmCE}g2dlniv4vxtncH9+8fbIZ_&1>w3UMQIIXR2xi3(y@Qh0v`yKZVwo
zTy*A|ng=XCCbNveb#2fm$(9iSn)#NO@cX{acb{T5;NfaOj;XHFx!QI(|C;(~dOG)U
zZy&ag%oR*x796a<R&7InvH8YZ`CI=X6J?rI@Y|*&?$=k6T#z^4w>E@1X)g^9+zm*>
z6!>H3x;NnTaq=9WtU7TY1Kk)WA70AYV9yg!k7ths@g?c&Zp29u+oi!+%Nqw<SETP4
zK5U!|$g-VqdP^~`zuyt?OA%xpriv&lE=mxIRo!*3N|51}0kcNsTNIr{w(Pgr^G&Ev
zOrvOM2pUl#aN3>bc^YbL3as=zqK-sEbrnwQOg6}6H0?#v&M9_nr|`FPMKEwh8e)no
zpsqq3uM7$>)uy|;vb-Wm;HkL(NSlw+&OMci@5~d5iHObFoZ&g2OND|!ZHA)5rY3UK
z7Ekn4*e)3wOeTp>e0M7#F0cI*9+v<O1T>C?eN*Y_3d`mJ1;BsDKWFRif`VA{)_U|O
z=3sl;#@*ziq!z!ONTTnenv+P3AD-5i4)4$tt)FC46yqkYCkl&(CL+B<I*wV;ei(j)
z(i66QzHo;z9w4V9y24wACAgPD-yGOHq$g2?cv>8PKlH34EBaVp?2#^8Wl-$!xTD%)
zJ$D|olf4l<0rYhcw=CX7o9SKevOR+i|99C}DCqwCjt(ABus0e1BKv4Xw$Ak3&WyaP
zK9X}R?KK|c-ccm$ijV964Y?-a1JJ{(g!q%OA*8I1NWc5lb3V6|lq3ITg5y_+wp4LC
z<>Uxdr0*8&cGZv!YvRUAUh|JV3$3ojV?3eMgDr1k0G@mEdSfm@$vccFNzmajNY2Wn
zz9Qp$B$Vb}q4La{$jy5><t!c{;5F2^SBY^;{Zvj{M$kgw>=X1v;rVHhF5DJE-Rqi$
zW?a2KcTLrjU4hz&TH6|XwWT5P5-zwqSR$D|7}zLS>b^_F)&GkvYMe%35f@b`V>gGG
zr=tO|Kh$vG)AyWBOAm~yFL#L+oh42*ef-m_^DR9g`5+>FwM4B*#iJ?S!FQ8?VELa^
zWk>-J1{--F9yuj^Y!~Bgc_6|D#g-Toeae6y|LCzV(7JPYF<f!qORZPdlRv>+#l#M;
zsRP*+1H)o^ViI`wc1f{m;j>WmN{$TuJb<VGLdPpgJT+8za(f9A?M?pooGShAlZMOV
zLobX?DvfW(4xYO=kk@*kz-o?JEi5mMU{o&g<($E(PRA_Owv<l<#7lgLN^I0ebW`z1
zL@yQ^x-AG-1><5s5vIbcJ4_)aXkyu!MtMo+SP_-O<HR)P2(T!<7snLOMvf~0dJe~g
z!>72|s*nf+wU4E)Wc~<0wd)K`%Dt}4Bf;2bx@HzH8%IF-i5F_p$!tDuH+Um>k~#xf
z<4g-@vf7c}JaXC37`Gzg<b&2;dZ<8}3$b*Dal!}43rB^AF=yj1b*h_BT(bj>&62BQ
z)y!h<N;~`1>ad%y3|(NXdk=x5XHT|j#o3x_ozm_k3t3YXR|e;^#~t>GKSMcufj#pu
zP|JY*6=>+)O{I#%IR3mEKLZa{l-!PO$83M${^@CGKe2N0UgyW3DeJQvFEc}(_+0Py
ziQL+FuKGe0LBb@(!v4GwYQo$$_EN;p6BuUM8u>Vd<mh#gkmToAf<6Fr7<hSw><=;p
zL#^}6xm+s0eK$lfj6fdvkTncBt;M6kx-FC!G4V`6YNOW~wD9hGl96p2*cMCL6OGH|
z)$ZZ7ClvOAM__k%>RK<CMB_iIq#qdH-F9q`JvwX>&6dTe`+fG=OU8U2;YaAUW4)S_
zoC9Y_`YQoo^2&*HK%73g4#~Ya@4swMZ>3BQa{Z)G_6Lf#cEM0u4aD3im>>vRmlEra
zIHbRZl{nQ7ieMj7X4XkvpRzO#5dyWq@zJu=|30L-`UG;-lnO{hcQI<BQQf<L@9{a+
z;*z^wOZsg<%Oa42O5;wb-ua{Ybh1(Ysha7Cu?nxWEdbe=aOfo{BL;Og(&VXNeI#*L
zk|!hu#I5Cnu6Cf^{_{LvW$t=hm}*3dg`M+`z_%o+_pY4zJaPjyvQ5HiFSc1D49Jwh
z6WB!1;Ndj|8l8_fttYyF{2)OJPi_%^06;X|J`DcBO$wvm@lo!hM5CEU$cqcv@~elF
zt=9q2YxoJ2a-t{VHxX1^O3OW>RpVDzQ0hUVuS%?{)k4B9J{~3h^y9f*^{ci{+5WNo
zl#1ixj|G!7mc1<Moc|_i#abP7ch0^gnw;F4*{)5-)H2TG$Nx?L2y0eL;MIH<;>Wpb
z^BbJhA1P;)-cZve<mS+4qtnw`{uP{T0u_pg?vjiR4VGCkF)>3l<ZvWt<;CUQc7Yys
zK|K9=G^oh~a8xxretpiCLt=6zi|^Qi3eQ=Clj|9%8Oq871>I=qzk<OJARUB7Q1(@O
zeUXoa3q9Mu&SGQGRkH;Gv`#Bu=GmdVt^*a<9dpnldyN*xjO^0`RyVs9J-h*8OwRCD
z_TRLYRPy1FV&h_$Zm`Nb!@7xxe#Qa2@Zce5OpRi*><U;<@4+`_qY5}Roz(v{(|=#{
z0v`YbC3b6AHXkG+thyw1tY=;*Jhz8_a6m3@t5nW2r>EDi;Lyr_vqwpn!76cVdPV&X
zhBK@C|4FleZV<U&V-4!BFhN~Dg631Zdunxapdd$jv3^iZaQuZWCj8$PTh@)t>u4**
zw`ag`&mC3rWz~biXDfpIbi{3eUxRWt;!`eP>Jb%~t3ciP{+PjDd}Bq>znF57@TD3a
zk{~&BCopTURFpoeB!cR*>X+xHTAdNxaz8WTLwu1DpQh8uB@Vp!f+BKhJ}X(t)da2c
zUzE^ps}4Kaj}AjloL2uFR(_D!-i7M>D%I&P$pe&qbPgj!Yi|^P2A8Oii9df4EBO*c
z(0A{>`hTmgFbHGoTErjKik0*@>`ZDps4@!jJ?2N2V7a1AjuTh@K1LH>oSEiv$b4)c
zDJT~?M>#wPfa~?XyHEj=)Z$*meZwm0z>fUWKvnQZWtzt>Ow^Ln&L?BW%dC0)T$P48
zFC^=jx(|D0^@k8thl(BEK{5!5@8JEvNa<EpH|A25bTLe7M6cT{Z_GP4udB8Hmq*%{
zef|0av?m1Ds7&x!DsL~^k(bx`aH=o9ynTg%ssbjt!|{|Bjf`lp%$0|7bLrUc^=eG$
zHNkF`^9AfuCGzqDfJNi+Izg?^ZW^cV{ir##dpUv(9ga-->GP0;DSJt0tbXZp<O-;m
z=bkKf;+ycY5jx6`!2bq)=ha$56-RI_NYvCi1ax07fgZ%y1A6gGqiYbmni;!=Oc53!
z`^@<gD>sOM+wn7quhoVa<4b;ki#y+(c3H;w$iKBAOkoX|ylS@X<a*{=Zuu(J>)jKF
zK(O-WQnyaPeVWx@5qf$q#yzn2)MOMV0c>_emB=+|n(a&qFJIy8!B8@&3A5Ah0pqg;
z^7+UtYgqf>pG`0{+<|{*&+Gz26ll!i>dO4E0w@Hv+@A=B?0%Sq^dxJO;SM>^yLcst
z4|l+!ts@3{`;DiuuVs@h-8>xkh7l#;{2<p;ki}@|=lTWPv}&mHj5OE(VLu%=frAoc
zn$*%WQF$(RcNKmqat=09##>i<x1nRN!-__Yo6jkzt_r4isa0AO9$aMutFuxiD?2Di
zp9l5>)kk!QCcxTpOxWFe6vO3U_k;<GfAUf=(F}IG2BQHanoF>gotzouTxPSW!v17d
z*2+pyj7p2g8a#%tV2#>>Jw!(C_UbAI{&A5UEkw~0Pff=z@0TD<&CR9tHnNf42Y3I$
zYKN%T-vd~MNH&;v#yGumcqVd3nG70G-&BnWhIa$I>rGt`t35CE5YhJ$^!w&lAte~-
z&a{#Ws4adReWH#Y+pvmYz5YK{qNkgaL4LJ@L}j!%>4wzK>pgyMsN3I{d)|60xy49s
zrdv#W?mlo@Na0kXty{QxjitU{Ki~0+{3w}mOSyE1<NpAdXlOrwgu$saKL2jiQd=g$
zAlI5tX#Xx(9J4Qv=aaOVrT!V{4(!x0U#V|bNNp5fyP@&)$&_=KPv6_JzYkhC#6nl3
zAmLAGTzW>hnDq0kgx?LPUg&(B)*}l<(xrfdXV_C*sh6P2Y$=`a4p5jadfhNuzZvVz
zHs&WUMFAN}m{|d`;C}}Y=@%%|MdrcyhYDAkM}NPF$w*?{Qi=V!)w6h+e751Oe!2;s
zO_bm~;d%DSMQqwV#^99r?~-%TN|vq)_yiha*1t9Eq*y!c8WGI$Hlv;RUsu;1&({04
z2@<o!EGi<3Rw<>Wr6g8S1l5I_rTEq=HH%OyRz+J<n^JpKX=+n@YwZ|OqmrVw*xvK~
z{qg=j@0&kP&N<JxKgs7g_j&H;xvuMmI~?U)7DMm_2wI&TvjSTe4pd+j*^0ywn4BFv
z9(Xr`<Di*v6$2WjlSi?4%Zu)~a)$gBLI|x6^CPiQO@B3m)*YAk%ghmwl6d~|%q)**
z2&;bm#qtz4vM7|4ubHd(_{R=i(~>@&b6&FeG`XCGQ>6u@XZu^zFkRk`zwb$N{B_S=
zlSxhMfpma*8<NaU=0?C1@+{eJt2t#AMx*rIk{B8>1EyE)L<Y>@;;QCh(}|Ou<fJcZ
zfn%c9sohEpx~fw&Lb~+zBVSyw#%M}Fh0h_obL=D%7XT>9b6R8{DXApTLPL0XSKD|L
z><qBM0|bQrWvK34vFWv=0UiX~vF#Pt3PJp$AdXwDvGje`9c!^3S2LzVwrC@vi5}T^
zCb^jJWp8s|U!wGc&KknzysY?Z1!FD@eSFmDes&vpaSPaqT%|nNNv0)CO0ukWJh-~X
zA>@L2{m7+itXiYWZB#ft)8EnSPxytr(bs@N#62}C9}pDm(*N}624B2KX7Z<Hv)^S<
z9R4F)KDTNFmuzZf2auTe;|O&6!aTR!W$H0M_D-qwv#201%_NdC0=EK?r^Ev3?M4OL
z&Sm;(j79}#d^5=-y^||d1Gw^)!}M{!`E1EA5IE%;&ux`%P4GmVGN#H$PCw&T9lbv?
zT$0w?2`=rE9pC}&3Y})C#UTNvyTv&G#Vs~Cct)0o3qQV9clFyf@0i9%6MmstnEJ)3
zQq<UNUet$pz?6C9rRltZGhlaIpQDDOcs3nCkn2Qy5p&8pCd*S9>odm<yUNK-f|Wwm
z$bj-B?dH~(mdqd=Kyj1+1DH5eoyxUCAev_I9!(Hh_34V*`D$2%TgTiBb?$s8-geH=
zOh=*Z3%WMTBrOg8&NPXCAESUSgOkvzjWGZpj?It>zF-(trNrRhkTFh+JtaoGQVW~?
zT6KM*;Rb4EL{B^XM#25d`<%)))G^wGuHh`M$gUl(<TIucbSkd!B@xsQEX07&Q*#U|
zYrr&!LV>J|akc58WpF=<yhcPec0-h9QgdX{Xm{`+x1QWprS!m=CWL1mCS^Zy6r1
z*L-e1f`&{d#y~?-nY(aTOc^lJCv#<dn=^}=mmnw5CT%yfET$QHPv}5DbSM3Yc_7%X
z3&%~{w6$Ws=yefFqUzBDaAcgpD?R$Ci=fZ(N71lg-1=S*G5q9V(A-I_eKXrWV~P0z
z3eoKikzN~P!R$@DFgcq+H-w*1qAEU?2BTk?(;*LBgz7s9E<+^E4XSh3vu`Q@(teNC
ze{^leP?Yx|WTz<*Wl3Da+a+QHzY3~vU6V~%1sKlor-x9a=i}U?qh;v2IiIQO@=nD0
z2C81}(V+>H>@E&^kvy(2*qQ9A(@!EJa6hifHE3agjqhkToIVgiE<QCu$cF1KF3YQ0
z38G|dfCFYQ>ORIHiWLx-nVi*@@~Hp4fhFt<ZzhW>F}CR0C#;@qBDlJ@#(o7hq*|AD
zpW_YD*9TO}wK+p9xe!A#DEh5p*rpH?HVI@wI<)pW-18ZW6Y)jX=<EvgAUBJ-L`N|0
zSXBO1C{g(ysoqB`*EhzB9#BgfJ|bx*jZ166gZy*QD!0wAs=hMGJ@h>ZoJqn;WilYN
zZu~UA)V=I-yLy~@!s(>SD=Al>aY?R08W5FJXpQq4HX*D6OP=kdf%yG{T|~cjCy<yh
z4-WcF)jqDjkU%-QgS}ABFg+1kj&bGdZv3l}+o~cl?-HA;PM?m@#qQtGv3W&`V6<CQ
zozG!Z)P?U8aJHK@e)0P^BXo8XzXEvAtfP*(LyN?D?Iww!*;BtZ?NZmfP;|v&7}W%e
z0$^ziClypSh0P=Q5}2UhgIkdD2+V62omF_%cBJH$$RJ$9Rp?&%_FM%9*}}OwTQ0S=
zol)a}ZH-G{)@KHg;D|sI=`4xpeIhfId~X+<skMbAd;z8~fzZh;CJJZ=T99hC!S9RU
z2;sb(rIzO%U`K}HDheT)+W>`owh*KG9RNZvYr*qRpYh8FSuJn02w;89x1sCZoTfm(
z!My32D?hmkpKM@~KLm1{=lUH1yS}cNc;E70qO)auA=u^q%lok*+5zaOuH^-?;_@K%
zBc~}qPp`=cu-xev`)-iR0uTZYI1?xfQ`G8pV0dT8!~jyfHy?HVX*x*srt7XMCN6&|
z(6k2a_csIw1q%ghS_^Ep&!P~*j-fS%K=^0sPS+@c$p<{&hCC9dEEW(1`_#1~PtdgP
zAOlG$B|#<<!nDG7v%AuDa+-JrSBpNjXAkNF+B}T!INWrRI1w=ev-B#I-W8|T11PO|
zUhc~FH`wM0$hDgv9oS7L)$_J|{n86PN|stuWpk5B53T8gKMX3yZ4ND@pKwph#r@;9
zaU4eY81hMCJ)ZC2z%iV$56B2F)uufME}?y<7j0M3_WTln>bXF!_;^EK7!MVhsf<|I
zzTlczvqCbAn;H~(uMR`K9e+=mwA9dn4;y;%O<FsQUQiP)Q@Xejv<1n}{mqn7lEO#k
z2Idzx5&)GKGX!~jx7rINhcbL`jCyXQgQ*=cQzJpMP9ECatL&mFf0cMUSm>V2L`%5?
zJAe`S?e}jWm9)<1{ybh)LG0}>g<_#+PBE+iX%{r?1K=3USvWLKB~F=uoJxDO;2#cI
zoCdj}D&yaS8-~9}$V+Ij&x>)M&IvjvyzEWV`L59Oi8mxoI6-*-m3d0zA49uV49tZR
z+<B0QYEkZIaL!9gT^ALt3_x!Lru$jX$o(Y*G$V<*=e`SYzug4Hct)$XLI?#sKR3+#
z;U|KlkQmkqX3c;*BQB+{cigT6NiR;fZ;8b>tBfo4*#^*;yw;v`p$-Dz{(q;jg2!$u
zu?xuI*5&OC*Z6ja3RMtn8HTa34C#733Sb3C2ofLMMM1#3jpuJkMhd{vQLptdq+2vM
z<Z_M8cmn{R>7{s~8$)|9Tj(Y{H@0K{u1wMui>ha43z4>iZ@LUF+Wa|q?K-I!`DLb;
z9Z=-S>I8HYCkqgK04$)fY2qYj<%y4AXDQ9GWazQ0QNp~IzA_6I^KruU^B0TjhFNWB
z+gLYTV!9A_AQeX3w98wf4*rtp<kU@wT>j%?PVJJW9~iJ~yZ()i#jBLL@Kw#y+djyh
zDe;tXh4CD2kIP?}xBHX*S?h4HG2?YaL(92ioA6x!4v>hbBf0^?UO$hZdw6qQMTfTC
zQR`x`a5R-^Z?23RLj~FXDytJ8$r<;^f2kr~$uSh~UBr;Ck*}{3q{evPONM^TQUrd5
zC4{HYM5taYJ>ob?9sVj3N24@~UQ|3U%*J!yV_d;{iDHJ$;kH%Vx_T>d?i}%4Zc7>4
zZr?+7gm1`8u!kG7vDcZg2#!k}f2=c<b6?wl@%~C*`l!1Hh6lu(I^gdQV@ZaF#d={H
zr)$6gAC&UdiYRwMT5uS;EM4$V^>|opg}Arna0zENSCBR@?AvovD|+gHRlaf2rikR_
zn1#sWV6QWEgr9kdROCQL?83!clmt>(RRr#{C^TeA?aMMS@Fa($czZ8Yfs$n$yCy~3
zLM{kbMREZ-#;ch-Al04f)_V~TM@8$x9V;v&cqNf6FTBd_L|*nlL9YTGMeF0T4JypQ
z3#dQuJuMqR893J3DQl-Vc7>6JC*eC@R)5q{R$}yaz1F%ug3cg=hwh70lJtw|>edlA
zmE;+U;Fs-!AKdcY@@BORMLKU7vFkHB*&TMY*3DoPbA^GwV7f7C*YUOp)5Z6e;A~Z*
zy3};YYso5g_;3-2S+PG4MOTZlpggDH?ym3h^a}6ZTg$pyzIq&EN{uBrHOPU++&n&i
zCc|TjMZ`eYO<*IA97o%m&T%ufyF7StCe>&VE0qgje=1V(bSG)-xd^<Inq>ERcG5o}
ziUac~9>NW}Q5SBP0)fq|hT)#oQH6A78zhdL*0nd7ctN}&6PGI78AE=&i37(ylM0Xd
zqv^CU8rRcop`B<g6!#DPiv7q-|M*&l!EXg@eS`uCmL8`l7|>*^7j#1%wQ#|l;b$$!
z(5<$6nygq_Cb^HUDqlr)ihf+eS$w24OCm7z&+0igT;d{}BY&XmM=rR~edaN?Tsv&!
zC}lXQ<`}xkQn>t#gY!?dXE<txS%jMn>H;dZt*U`VrfGp`RxmG<W;r1>;$fP9o<k<1
z!gBrZd%l5fG4X&>-jdnKTsiD>y1aqx%Wuh@{v*z?6yAUs@D9uQyIri%4M*Yo>;f2~
z8k_a7U{$E`c}<T3YRiz)hX)!{<DS6|31w<^R*c7QvQ6G_;J^?rn?^S9KMPE{bBVVf
zV<FtO)?WOJrTOdhzJ`z<A9F_X7(`HwMd@Z2pLE-p=QJZ_uJ8Ic*voUFZq0fQ8v^y;
z!3hdjR(GNSTl-mVge~kISIf`D_h3xamCKdBd7QJf7h9YZDPXAt-RBeiVa+wyj88R1
zc%?vy;bzrGI)RN$v@+?rCMs^0=u%Le=SRu&>a43`>di%u8FvacE53`%quS7NOvmhR
z`BzLYRhDpNW%Gbwn%zbIah4)3D)<5pYzD8o!l~VhDrvtr7g-*NhQ;oZokTGFV}wP4
ziZ^28uND9C01vcudtJ{v=RslA<#>Usr)3g5pxH|07`h@MY-6Pp6y`ZLf!n#RK)cxX
znbM-H!Kg#~lxxL!Cp<VzE@{XJu`v)=6mm1k_@N$|?iRUa{Ste|hn#4*OYA>kGtMJM
zAqBvgOhjz%UX%Vbm#wW=!2`uqoW`d(m*qqh-E#mL5+x7cr1j6~nzPiR`_S*Tx#miR
z4d^pVek}SoraXy08TFC(Ll1q5<2_<qAZR<ZJhy!*l-7I)l4u;RiA@y{TAY`rch1gn
zL-AQeySCJ=2~AwmCbc%CYpjRhWYDglwg<}**O8{xS?!nA+IJIrqvtVsl^=N&ifO%L
zOL$6@O*K^!v{I|48l_?)biS<bpD68C7lm<+Iy^{{=Q>6!2WuHHl0SN;!tbgvmIO>b
z{;)ZzzC?VI_LZXCXvua@ClBUDmo^)Y>a*|`eQZ_vXe1|N06D3>vEMh24MkcFQSzJq
z{`DT-+L{VDdCz{d)cO~*6=y?K=}V=wN;t`oh%+jEnG4S_OYAH8t@WoF(@m;Reyy!B
zq<If}r!G@AnFrWw>&#uyomD|fnAxC|-^<jqBk%SnE*7Pkr<-dSl`?9t|LIhKuZkrZ
z@27NHZ&gYXR!@%Ebo3!i$}~B9kM9kyw4UsmA32>K3i;2JpJEP<0{-Sx{~mY{>Pvi9
zej=OnkCva8tM(HPtndRt(*xA_r4+;A&2P7?wl+#sXPp%XWnyX*kbDQvZ}evlVs77F
zQOq0Ex0*PJr+mJh{N!ZsSA&7oRAXIP!`JF{m*2!^vs0_p{f&g0=c!j*WUS&>{OTq#
zr@q^3Pn|GI&IC-P%*rT<bZ~aVLdnK!f45?%yw~RD>YMGumxnIf-|BlJ<y9K=JF8<4
zcWN!{wuG*-6%!)Zn@FW+k5y+&9;Gh*Zl!!PUCTq-(UYQ&>Ma`58)24s;WFuxCjrNs
z$dWw!x8<3Ttw@<ntKhR5zM9?Lg*)OaC*w1%Hm`(8NDCv_^SWEo)Rl$nhP^h8l}%~M
za$7_3$lh1tnFDLYWXH1I;>44PHHNb()5%W#`-D=j(=p#0(;Z`93B*`4&CWc-+LOVl
z=l*6gF@2wg4bK?x*|89A{Ae14gYen|vdYKsA-+9k9`6|qF~qdWvPP_jHirI~seboN
z_u>3|Z{6(i&)>_+7A{6iock)T&@mf$PT8w7y-h+S5Lt9i*k679r=LArsGT0IuBdZ<
zU-R6de}@=c_#0*4F3!$c>cXFXjnRX^PC9Cfk{{E(oC%}PmI7d_e7P&b0poKDmj`Yt
zCvgjZ$>+|5We(SyesW3RUQ^5}P}q!hkd2UCXWF!VpK?^1{wb<;R;gb@Q;}nedbg`1
zT1$^h=dKgFjdS<Vz|g4NFNk1ex4!-&?(5B6t6qZ{aeGMT@+Oy#MFO%Yaoy7RjOIT-
znot@8VWee92~b-LBGkr0#Az{8=mlABEmFdMWY>D(+79S(VFFDE&dfL}i@!hBUN~B7
zUUTpACrtc1h0BWlsqY#Fes+6r^-_!UXPY7Bb+>+r-r;nwRTwK>UM&1(FsRFEbStNa
z*pjvi$Laa`zg@XQCbssBv-wtWymj<YBR@l1atNw94apc4%0`F9{PM)CT?ya(QSOwp
zQ*1`pe?IjGvk`xI^w;lI{IQE!*ZqwY1g*S5IhWrXg|wm$u*0G;bHy7gD0~3oN=zL%
zNbATwixeu<yymS;e4~8*?;wS=H2?W%WTf`%6*kS^cUY<(zwV=Ko;M_hekb-GVa=wk
zJsLK7l?vjCu6jJ~{N=weldo%t-O3F1Z>84xVj>aw$ZhYo)>M~r<E@y@s?Z3<h8Ihv
z8nSXvk8j&MTpnar<yuI#UL|F%WgibM1<A<juA){2SKc8PR#NAaG_5-BC4F*!xDoXs
zZ7q=WJO2Bhiv6=VV@2J`sxG;fJgc|{l(!z{gnbD@Lql_RMn_|WFNlS(5uD;7Aq-4T
z1uv5}X=!No3ES@>muN9$<oQO$%HfM%bXr;owqzS~x;B@=6n%;BcDlHctjY2CO2J2_
zJ}HI)@jJs@x<94}XnTmer9eYZuWo`vR=88{Z0T11;fj5tgR1+CyZ!Z&Ez$ksB#VG-
zK?kXC@D_(l*(Z{uNNaqU@8`%jM@)G)o|!Eu!8mG#uDM)cm=?rr5<||}nQ^m#+-P6Y
z^?wp#*j9Vt$9Q&ycpiFkuzcuf9REW=so0lhPxQ1<JdR07OjPeV*S8{j$m;K++KxUf
zC1Sn#3#BU8j*$}cBF^E#0}Cr-Ihxv8&rYr8JC%N)KRkMzZnQICCdR-%VZUd=>`?Ed
zu{U8p_>o@BiK%@B<E&P%#Q&xw>tgs$>Im6(Oj*EohrB=VqK;Z3RV=9_#8k`kq(xw=
z#jNbM(U#f}WbVa?;MJU7jmIz){eqkfW}zbAZ|~hJtih8d$lr_OBEmi3q0XMexyRC>
z8f#FhjagEMM@LU1P3ZP_*bYoq)hANo(nh@Wtn_iKb8^a>y$b6?F6F{6!~B@SG6RPi
z)gV@;<$ZK}$Jv(QR|V<|huKs=wD_wHL~GFVn&yK(w`arch#BlfT|$?7s3>e%o^yxR
zZQ=fH(AlW6Bl<=r*Ne`K$jH8)_T1)jt0JyQy=#rKdl38y9ztX`sUL|N|G_|0l868G
z0b&U1nZnB@L)1Yl=keaj5H-4a7#ME}2lL{Gk|C0ydTIPIa2EuV#UoRI($L3X{Fy0G
znw$cZmRaN5Q-HQMPw`BtK&^&XczxiTt{nr$tA&82@eQd!Ex|%O<=?ya%fa}e3ZSh<
z8bp##tQw5Z{t8@00N(N9qtPSf-6|TIWj}fvC@^CG`{}I%)A+b~*vWX}dov({|2k6v
z(7HQ*GYz5)VtkFinhsF~Ek)p6(;>1#|8cKC+XukdKL6i7fU^}W1Rod;;m5ZF=J1Q@
zKqptClQJM;7yjd-fchgKn*V27J_fIr0Z{>w<M6>5fE805zBvOTe&K(g9y%Htj{jrD
R=^Y-F2|<8yN&h}}{{uoZmRtY;
diff --git a/fun_gg_boxplot.docx b/fun_gg_boxplot.docx
index 275bf32f0d8b4eac93608dc15dca16a2b8625644..3cab248e103f7784c258232ae7f8ca5dca6167cc 100755
GIT binary patch
delta 102999
zcmV)EK)}DBi3hKW2e3g23hN0rs26qt0F_viO9>f&>66>YmFM@3*#9Ay?eUOn3dOTn
zMUSe3b=ivW$URc`^aqDWkpz-tx<Diw0E@-me}3P~On?9nkSdiVBq|-@mV}E)yx%+X
z9slv;)7&SIER47a-qzYqt43Jh3XccVx3yn>zJGI3BT-BPkNP5DZ)<B7)qcGC-~a1B
zRzpvJxXU>UVgio{qTy=czOBvTcriRXi`*HT)5w`~Hx!YW#EvWGXJRtpE<0O^&^zn2
z+AaCFMJQYrMeut!G<c*@O+U_kI_+s*NLO$(KCpM@&S)63r_D3kpZ$>j+2z^A_JiiT
zOA`wgz{!aS=QM_YhSRe-4Ih?^H}KpAjro{=`#fI5BU|UmW5nCqG7N_L5pU90!CQvv
zRrDXp9pNdzt8Tf~<dDC37BU~+K?Ko^FVb&1f9})Z<Sco}W6>fW=YDeCYSHd}86a*I
zRW=Vi<@Gfw=e~OL;wQFSCr|_*k=}92E98GOd6PNi!RCjb^=UF6x847|2XuBGa2~OL
z&wE5)KjJJ}&o`{GT1>x;%6|#Pa<O?D|FWmO3m(#Ep}hKxXKA!#5gdKl%Y2;C1+enm
z9lo0eBBWy<-Uw(<fP6$!fFPIDu0WcOg}25(7i2XAY3zM?TWhs$E(X_^Ac6IvpF%u%
zd)aB<-lYd`*@Q0rc=yatc&K;L>kQg|@)s6Ah4=>#Jh;GT6S$>+KESJIG?+?(kN-Zy
zi~i-p;YaGft>K00Sg7s~??n(taGFLg2Ws7jWyo1b-m{f_)+mT}kGfI*kbK}+{r4u4
zf4agKVe+f@_bvFJx-9xVIoiHRj@;l2<c|5UhU6Fv-u&en51;9`K1)}Rzp(p%Ee&7Z
zY8YSrK<>g&ghS%`G>Qn|5Z)yaF_{Vh1fQ{hP<@U^BxG~($UNdL8HojXKf-6l>WS)e
z?O<^WcIx}}?g#z;2A~#;Uw}pZEvtk@p9hSzyXS-l1O%x2KsqgulkLtqS@C#AsOJ$H
zPNk*s*$B*l?XK59@AeG?+ZR!P;q{;imGLlIP#1LPB4iN@A6X3`_EGqcAiQQg8j?uN
zS<L4wA`F2?;BOI~!arRakTD~35#purk9Y?EqEE!*(ZZ)|Qom>qS`CN%JX=OhjZp^v
z(3<4%_#Hm)i@*byu%J=QTJ_r9hYx@MK;FG4wI-P?0~g%`C!llIB;enFkvN2759$M{
z&qw$KmBW}%9r7+fX61NKQ?YF<FC8;rAYdUPCi34a8p?krkAROJ6CR_^iG?8GL{8<e
zMGVg|@&~+uh-JK3#srLLg5HzQ;`AjBkpYQt(d&1*?fojr1j;yMgkRE+gv2pON%-7A
znop5S<CulVMdD?*b$;D{J6S2`FSkm0pBOa&IEO?c@}(zYCqlo2{760mlPNv>NV$*h
z$3dxvP#BzdZd*oh&i)TTSp*^_=I+ia4Or3@Xd@Oz&$Pow?M~;WvybKTXDrJfx197B
zGkkf<4i8cHhdg3Kf`N=^cy{K7G;qQ<i%|TFxv>gZvGEx&(*v4+OlQ+6-V~la@@c?8
z3q@xi<U>A9nV3h;ES~$vLAr%U^<gjW&hPpXT-zYcpRpi44pRNaDgjbdTQd>z-(dmN
zPaQ&OCwdHz1Z%JhgNe`GSSd8H&_T{;E_TSrn1=dn=3#Hi`(OU{H$o$Ll7t(WvJ>)u
zz#esirSI1p<jo&{r}3Zh@%YTmgTeL<*>6i8GVeItw~(z!)amp`nc!Up?t{_!UHjg`
z{dBnNZ&?YrqoXQa9rolhcZn_)84?*-9<#um;WiS{;@FWkob(WQKt6nommU}D2@6<*
zjIV&_<q9wFDD6dQ9(I)h<O||%?`LaWbXxnvPWdyh_p|MPuxtOOciq4K7VP>$T!~Ki
zbo%KV`VB9<fU$U5_#z)fHJ&W{-8}L~KmPNZG=x<@F8E%MgWK({jGtH*HUcXijcL1?
zj|W(YScqgGvbONB@IZK6xin@|V@-hl=za#c74`1=t<gpIbt1l^&~}QgLrw9L(cmDm
zWiE@O@?2DZl*eNIYhc|pSV*L+D9-7y=nQ%ngV8<&<j=fr{cgq!Yj@c3*^~vI%y1|c
zd=+tv722)+Vr6>OWIA&5g!{Ogb3gjq$2$uO3knH}75Cy9prYmUEha1`BqroRiHYX~
z{wyY{hzZ;$bt3*d;^L4kh)cU{^)Cmdu>><qyJ|UqgZ`m6H<B!tH!W{U-V8+Sq7Ju~
zokhh471UqMD4@dXaEl0qh&iJH0OIElzgQ_@F`@mx8IK-V=sX~0-jZ4rdo>FV3yvH(
z(jb9FhDF8|)<-3~LpeBP7FKRoa2m38z2$WJjYIEm9YV{U26HFw;My(y7X?L0-t2^(
zTSP#Av0SssH{YL6ze(~<yVV)BuPvv3Z-WC_@`PlHgA}fGe#M^E8@^DhbFCpJc|PXe
z+6Lw=D0V>MjCtS$W57oL`9YOI@BHq1)4az5W6#Gi2^jz4*s|#KWSslU!Cmii8x*5X
zx3@t`1{C?rB`EZv*8>XmaZCb5#25zVJI|JX61~Cs{q5ewY%3*_k7E=dR1!|t0>T19
z0wUSy|B@6=i-~GsVk`qoiL7`YGAsxTcQiBzKDUzVHN&8am0VAbPscg3?||FZZ<Tjd
zr?BkpU3YLX*wX2jt?pf>vu)NBJRh9*&vSgNpCF%HO>u4v41i(b3cqldporilBS8^=
zyFMrOW%ZHa@LpSeRCRr1_c1E&UH}jCxy>-LfXDzsV9Hr75H^3@0s;XMbN4~*aat+y
za-b+5{ejdGb4B1uiWeJphJjeY_t26g;4ZtpZW(tJuiBNOgcToFr+}d0f$J|l=KRh=
zVH2w?G_vM~SUBt{4yiMzArL|Cd$nwTR~w5CM8{afv6z1&fZUw7d!IEi${MC$#RQ6C
z8aoS-9|@Mh@-fny0`DU>YQNJvzi5?kHCHtd+O?`v#|oHg3m9yx@d_l2wY6Um%SMAi
zqF$$#^}cKe_k7SFjryh$zkCAhS%S;sQH&KRB@|xYO!4MZVU7x`J()8WuiuM*z7oVT
zRmS$F`9fJ=71fjuAANS4Gn-yiX5;&W^rDzODeL5rT2sL8Tz5y8=Vbx*7o-@qJKGC2
zd`-_5<cAYpEhit6@8!xFeJ*5T9!n#j_Waxv3!YaE9w${B0HX_wB8#HWM3HXq6V1fx
z`NpROQ8hqxDq8IHm6!R>2VqfvR`=DXQTzcAz(VHzM5pXJWb{GabpS9F{aR_`*!JD8
z4J-#BF%3C(203|O#<q+3TtFDsE{1xD2V3U-0+csQ6of2ZhM3Uw+)Y}WRbMtjYnnRK
z>B1MdO{YkEoU#|4-bFjj4LPd!^OwI$X4jvf#2AR_c()$BeDC|L?6m8Des^$D$_fg!
zw9Eg|uTGE&))%O_0ibn+^$jf^Bpy~2)W#y@0ivQv`d;4m+P9_I2F;2!Y403*_1BUJ
zY-7J=HrG$cM)3LAS<rY^4CKqs`RMYdEI|H(WQ3i5C#S0){S96M`3UrC0^e-dTs8jR
zbHP!aEiwH?`N`k)I1Ijj=HAz{Ul-~80~x*h&k~S)(2{NPEy=$+$@go99|~>IyViMU
z^tEizFT(Fng|FT33~s+x;Ns<JfF9xzE2Ok?L<K=RWGtTWFGf=3>R=hFkt|)z(WNEA
zpQ=R!PsD}Q&4$S+@+X4Tl=&sSgvp>)j6U;9OosAN9Sni{BDfZRQMH7t5g|!5Q?y_%
zpKzDPTm(^*M2ro|ERGk^@a#-qBqu*TvM^$f2&ZQ{r6)>Gzrklj&Mcn$FOGlfC6(Gs
z`gQc-{k!*n`B(zFTUY5D(*1*2MZ|@-p*(=EP2P_P4~R!&+WtqAxb-#s2S?vF{z+Ec
z_sN)%kj=#-^Wd+46A?192!+d{2ySk|CGm_<g0X*mSRG^#T;_{ae$4slxa!=sZZ7&3
zS9@`#KTaue^;h`BEFu&D5`f$1iuqz01898yzz|{aj1f72Sk_8vud3c4lV#wd0wZEV
zJ`^eH+fM&_aIqgQ<@+KBnmZP{Fk9F!JIy}=uNRveA)By&kOeL~n=&!~q={Y@bHl2g
z?;WB0S&$ihWS~G$oA?aH!q$%E7K<3>TrPY@BW7w0Z2y|vkb!*wG@DR226x02{&F59
zmVhRtl=OrGyD2YaE#*HBB1btN&p6s3^1ehhpv?i=W4fHPAa=+-JUM+@#A3W{8j9rt
zR7w#*TBG)V`MEjn8~~nGD@X_C<8uL$14SMXghph-!zd=qXYxRuJIra~ad~CCNge{R
zLL)Su(O4bqkogjBp?)OLj@1q`AV`7^elFtiQo7F#hoCxBKUyIHlzu-$F?J=L32Q98
zE+6%0PzakbqbY`68UQcjylT`K9*-72T`PUUP;!8Otda7RtY$1AOO!NZ;Z2BG#^8$_
z2hm;+<1mQwms=2(UmE^EK4P_f0;q`80WXfv1Lj~Eep7;O^S8`I)5BJ&`Vja1?rr~T
z<@4u0gu=OA_x(M{vR3jvj(;KeT|S-p@c%eDv1UF<*liw6`|X*kH=1PAT>-6(t&R%~
z8K?k%GhjYfn*XoV^Vu517E;u}nZ&h)PvJfYaB3&;ogBE}?6)P0WX%1nJ?OUEm+e6s
zPX|f#W*PAKjhMW_-%$^j1_(Y6A4eC=;lacK`Q*4Wco=sd-ef#w^`tZTmJay^9!A9Y
zU(Aiu*a~%BwuqV6B;zGUS_1UAF@%Bw66~CRWdtTpojFsmkFa8MI8@if_^`S<f-hNb
zIh`h{@xYz2sMdhL&gZozfiJpbe-He%&hBoU5X-y1==QGn=8)#k*ln<q3B)2>_wu@Z
zckxYx+KVjaRvQEnWAli_(&YC<z(~B}F8G}R`4w=6;ct&U!VwVvGB`=e7*xOmFxy{$
z#fkx+9252$(&hUuSAl>PFUR0t9m`oMEp3nGtd#cSA4o3%*ntgNAvR06#{n~a88xNx
zEg1;+i6{>uGG9kvq;s)6A*tf}ygj<xnAbTPCkM0IBzJigwaNFcNcc)|-XaFLZuffE
zBV!mqZHH=C6K%_)wT&*@*P~Ke$TISOF3`-wjIJhHjYlM};yqKygapzNgtw%1ToT#z
zmg9Eq-j7p4Z2v$;AO7;of8M?S+2*&{u+I-Fy=7RXjU!wUnS^w1YJ}#1(9UEgq3fpi
zgT$necD;+i&8^};6BhZ~k_Iu4Y0Muhwm1BUp`%EpfmMiL=;Bn0PV3F4ipti1BWTAW
z!-<f@E1~%=LdKJOq5@vY(zQejy4u<wS7da^?Q-E`N;vZp*@;OPo4h=V(SGJxtH@eK
zWIZJAjE2<3E!8LvrP5WqsnTG|B8W^tGOXXwCk|pjAJ6FO08kimSM41{fW!B>18FL8
z1n86-7$Wp!u7=nohB7H`^j_|N2Dtzl4d}h8LimU&3s{IbCq)cxppjR->>#OUBp8yU
z9JyHqG?eXxB=olH%dB{+ytgh{dPp9bOc+C>V@ocjqG7*a%GxL!mmz&fgl3J(Ml;fS
zsaPE^sE7Z=ygcB3#(}8euH~s@zeC<lGAV-@qSq(OP~KS~X)kjC!Y$i>%ye3#XisxA
z8*;ZOkmLs4Mi%hnDiJjCGuo2m+c>b~Hj>;@*{ej{<CO0-$!f;k8OrlHFuhFnJ{I`V
zBrhR#D`>ib`OPR0Xoq8qDAc`cb$hpEq>wIvtrU%h;NufEU&QOwrP2<+bD-GfTU2w9
zI*4x8#COwa`8)ajjW4c$_^>kI8{q5f6?YU-xP95{Uzf63Q*lWOC_Pv!^*k;MTy?C2
z`tS+8wQ_v5<FY38<gT63ibexwPo+(6eP`CUe-*yHBSnHE6Y7V=$0^`CuBR%t(EWo`
zX#j8&58N>&NysF=0t^WvTVej0sw^gT{D)#nEI_AWEEzKm5vb>X$b9e_Dj8S6U(>Ul
zhV(vmUYNX1Mx3NdHN>t6o7dP0Xv(~kF;jzqRT%;KDCX$Irms~b@-8pC*WE!G30g6u
zR;SD(c_l6cx|x<TC7?`DoWNY(z6h%3V~-iUC5E&(^H}7De1W>rs>au&8VknpZImX~
zdga36viNIf(++`ug;ABld>})GSE+gZfasEW63-C;B`}UVopV{!Y9+we2mKF4pug*4
zeX%*9H`MyL+F1NjxTdzgr<h2|T<MDF(o76g!wrvY4>KR<-}6ql)7li7TUIsLM*Sh<
zE=ut5;y{*RJ5+%P-ULfO->b`Xnn`ArWiOg}^Wn%M@@dw8Eg_}AxX^7IH4wF1tG3sN
zUm}(AXHJny7hT!I&em^#3jtJw*G|9JxhSQzUvD>Zc^x8gWgIdp6re^mYu-;Y-#;}?
zv4n&x+|sHFg;au~+$>G@twfly`VNoQDmT|LOw%1R6-n5Ruv;DRDs_Zm?eFUyR2-Xt
zm=1s`A1yzBmDUutJlHovrMB)*xj8tc(m<NZ1*)!Ax=H{RD_l;ucGk=kx;p12x1VT*
zt`)kb8%gzURuU)a^}8}zH-MiI9+MS~RH}DzR;OX2I#aO(3sVIOoAT_6rbim6XNKD@
z=brR)BJ)G?d2Gqa?N3p{CAQU?<mzPWV>!JeT9~PSIzhIaQmK01Ilpg9&9<!~^JlDz
z{9c6of!s()CbK>gtuPJQy54A#S@ej1u-L`)$xt`UOxj!?N5E(9iCW9aSj*|U)4uJO
zQds>YA}$+x@CvES0Ms6qPwwy};XxEL>g}!7lqt}@02>hgq+@KB7A}2En;eT}Oneqh
zF<V@J=1^jSc#P!<?~>n2oIpKh3g=F(QF45mPCqTE&*GTX8@jtUnqG7IRP@EmoAScv
z^x)Q7RlEeii6pD)13>6pqC$tj^*Jo@&u(5BNzI~6pl$HlET0$AXht)JO&u^$n5B6>
zx%|Bf$t%>!l_FU>@~4?<9c7BBq{}fzn5BV#qI!}n7XBJbI5cImMvfvoc9+0xu*695
z7}z~?9xzo#asmgJkXb8Ps|@dsWs0_p{QJH933g6r(7PCnw$$3;8L;`w?F`uRbK(!O
z4YLn+e&9J+F@Ct`n7T+$YC-d8>(1gmgVL4*@gw|eT5*HQj+Wg#wh8n{KmPNZG_P`h
zl25aYh_07xO5>^E-VWZoYiFTxq90L&gzjax*KX|>DaL%{y?qo6#nF{RehqodNUes1
z4Mf7}A^UB~3p(JGeO0#1uRLn8%oXq`>L%bY42dtMSP|}!d-NZ4<@%4P>u}JPnx&Er
z3Z$4y{0-SOBHQs2pNk<<RRCYaNjJBD{nq(q8NX1)d=J@#AT7s?fNW^wjD6}pG%zB*
zV(4n(JX=XRv5b9uC+vHv?uEY*HG5FsTSQjF0z=p{HWaw<Y7u>`dLm`UM&>y<?+Y2j
znqY?;N~fzy8u5{wC9Ovgk+^wPPvt7oOA$&oet^qqAT3<gux_0SUK{|K2yESdnDTz~
zw~wDAv(@P>;gIwq$w&ds`-N%t7BSxxF<W35gZ8EIY=ej67G1mD9Sx+-WsNs;AW;<4
zSaC0&70;9@ZFb**V7bbAtCV8SHKr%^Nhs!|h0?g)DDF9@__ch?yb=>4p@(5J-9A~3
zq66iUM!QdxN`t%p{a|l)d0t+By^<YeQ$SsuBc+LI%3^-SNa6|WNe(G7R9gPrzsRi+
zk&)!Pn$bd=v<h^p5@T)B*i|qxX;WGq8eY}SWF*xmHew4ZV|d!7e35Xn+4L-ol9ZR8
zdOK`isV44wWjlqA*SQAV+v_di;xIKA<03JvC}Pcx@3VoZ$0j^DmZxBUnkJCc#7K=?
zZ21Ybe;LhcxpUfoGnE?kKx`m&*a0njpbXt-Q%OMTbNY0cuT3GrK}}h96%PtziDM`u
z)hJuihm2XgVk)5or~RmkT2F?~Bhn^sGD-){p(KVbZ^mhZf5)_f7?jh~96RoVne>EI
z)`>+r|6cEr)*BdXU%*R$fL<1%8ajth8;fP&(NGQ%qapr-JgI?)%b>8Rr@V$qOxjJ-
zX_9VIT$gr~fCayN8A9ILv_Yd%3BDglVWg#sVU_YNqbP<_jev<MFHY5zxgEz^M@bG>
z;}{ncDd9V+07x?3NoxH<*MDpG)JRS#@|=z3BD9_Kl6+qv3mq?igfWYlSJGVCnn!lD
z4dQvP)wwNWMakzz2{Mbx6RaA61A}RXu>`f>dnB4*9trmb7z`$riaBJ6RUuLp>g`I2
zLBsS`L6{|{#!d^je8W)7;h@E?7Piv_UnGWMh_ES2$CHO`I$n9?Pf_OG*6c?a%T>GK
zI4G3|MN}+rv=!ce@Zj_<Gpia-_IC_RjlpK_HtWm`uCiHYWu?rq5UQz~%K<$#XEZ1t
zbWzrxAqOB92<-$Qq1tMF+5w1W;WBurmgWLG1G`g)Y67#*18gZV4Md2&3yZtKSjlfC
zzm@z)B>&#j;@o1+tTX;<opC8|BmTQ}$jiQ_yq7K}QdJgz673ep(r%00@)~m{XZan_
z8A*K~K~ruu<I%&BYH_njc3{;!AoKiqPOD*FuZ9Uk?9MU`b>8E@H;r(9B=2I~Q^(`c
z!l!H8wa8x4)CX|9UkDFhh(L~+qKic+p7<OL8`TRG1C*Eje}l_{3HMmws>anJymPYa
zF=rm9f!auakx$)R!6U++s2jtx+D`w1Jjq7qA^EQ*1>Xd;$p^rZsXE1{QT%}g9s{rH
zCpu-<A)^oSuK23{?!P`X$Qg-#3*&mwXp$9=XGFC%+l(QUY9~_VC3EVj0w~<qksUVj
z-x53NsQoJpJ17b03A{*l=kV#Gi9?3;=v#RgphgaV5d$BWPXj&ytbjM9$>s6M1l}A7
zdP1eN6)UeotWbY3qt<9G&6)~#k(3=o;*bWgY~<uMT(QI+F6p5dua(Edg8sH-O@gwX
zxj3rBUpJg0N{=qv*P{)4t52m5nnHf^vw#)P#R=;yA@v`nGC*}5H8DG3!=vOaX<4cL
zilnxG7SI^C&X|28&y%5SB_E1F-ns4H-<2|$hW#H670-ucjx@Y!p&Bi3&vh^k=im&y
ziDq<BwKjzwIzENEpwn)3%o$iAoeK0BW+LRj!;6B;1@xPW0RB=b(&fEW?U!yuU-Q@i
z8N&#xy>{%6ik%k?C8a7J&?byES6-jFqNB5a&I6EPNJ1Iw(LtjmaO;pe)y1tAh#JAH
z{^U%fi9OwFQS8BSJv9%Tz9FXcHH7`P1WAYfkkdaDT#}&}{jfQG$~`RmCIoxOVGGh#
zkO~+P#TY@0F>V6N!{*!CNJ$>qC$zyVt?B(*xA}I$C>bfB3RHBsR^>sV!I2S5WRpRE
zBm?HU05FwA_%I{kfYX*LMfE+!Mpk53-tI_ynh(h%b7SmnKGDs`-5ITrpw3T`3YUIT
zPk>F-^`O3_zd9b$KE<Zf9($tm1&I0x<G}*AjeM~>TEBeP#qIPm(w@->pSHvuIW*a%
ztEk2M_Si$PIP>wc?;;uO9NKRw8Qx8Q^z<&!=CZ}TZf|dK@ak|-i|Cxy)bt7o1v%Sb
zMjw@A#Ivag#r^qoRy&TkPxfS<iJSIKuhkt`+?a)%L(bseF?g}!*c#MGt+{_)R8FN&
z6z3@`Gn$O`cB4t^od(8#>fPdA5{6N7WAbQ%R)?0V%wucrP!r5rmo2c^p2o?4ninF9
z_?Y`VUL$>#+?U9%ZF2eLV;1ms`5wjuZRpKa`E&Aeyb*MXC&0qQbS$Mt1g2qWt${<A
z(5}Xb+B9%MR>&x(G}LLJZlGyKmbTstTT7JJ3U|oOp%*t&He@EAS4qHQYm7?K$zuzg
zswx!6%P2su3QVR{ZVvM(JIZ%|Ypy+AEPSrUmKA>^ZOfVRjq)wp@jGD-E3SSy4~QH@
zjPi<4aPc5Jm4My(a7BEucOYYFKhsqpf;Wf*L&<#AEP(DVwGR?HC9jGO>B!J?J)Dbv
zcW~cH58Bx2o{v*LEFJQ2Qfv%~kP|nkpISqEX4w}Bhf%xTySv_x7E?Zd$l<&$BbSpE
z3H8Y`lzZ`?E#OPL=&O){O<gc$-AJOz`u<3%=Wd~&Pz8=lw3kO?+D`4cqE(-Fx}DZ`
z#b0?b>X4sj94oOTE5d5kO1CEHRtv%D2zN?+Hbu)F9b}BfR*0}ow5_P;ak;;ylLIo#
zl~r6$wFp}s^6g(>vU|FJnVp@RRpzmnaz7$;ES9R{W6=lMvXsk?=M`0R-Zg|lW{sEJ
zk3m8ud(ZKzu-2vu0LaD`;3Mqf;7I4EGV7u27O5t(8#39~JV_<c{w8Xvt)$*<Y<JtE
zcGnnen@Y8Vxldp`Y+_;>rA748=`W_x3mGnHy0&3GQaBT{b<^p8s(6ACOQcvmzwjv!
zWR#{l#5?5G6llw`(<)U1m8a4%2dwOrdTtW`#eH|wyG;+;sOFxJQ$liQ0+@~rhC3qk
zSeOZBX3|8r;+dEQk!#f~Z%w1=_%ycJ4el0NLwWR+{Hq;c{d$yjesY@k$V5iNR{8>a
zUghP8)Aqbf!)EG#i*0lIW~(=0SQ@e<8+wx$qum|qx^P#f$W-=Q7&FyLRX#A+N_A&P
z`I*|q)*BRP7vr)5i=;)n)fu(7L-OTKvO<6>-f(b}s90!}cNrOdzi-zvBLQL_`%K32
zj%#(pvVMwVkw#f4SI@Ua43j3(7O?`w3<i7qR@)k%&uQg<4s9TVRf|-QvKSbLZsGoF
zU1mV!^Q15_3U^`~4*50YF(b7antYqsc*uTRaxetHibdGL@o_ebYQIIOKaMG2Q$dKB
zd+Th|R_?0G%m^-|=yPB%EZsf|Kq;X(Wn%sbF46<#Cd)t-YodKZNq~<Cnf#$3mh+>W
zQ`$Eg@-eS}ejR;y|L*-?iUp5hSQw%Kij7I+erdQo(k>t`1&mzrZ%Z10RZV5}T%~B~
z3K=W_X<r;6ssa_&j&DTjYJB-xO{^?h^<}ryZS{>=pfm`oM8_*s66q8m9r0?AF>@)Z
z7i=KQ1MFgAC{N17xn*I(SZN`6)vS~tRJID`(rBiCEP{Um>UcCqBbY{!aJkGPk$rV2
z*g%o2Fog%eMKU=K%k^~ga)9ib@Hx)RrMhtEe)PAG*v%dV5Qc9<8hB#vueI$|b=jMg
zHDQyXM3JKStrUjP=*{#nasI_)4AQ0Pbu!{3o0K%j17nLc=d4;bRNh?5SQxn_o}Iq4
zphRGQa<DRfq)NKezuJ=j!#e>Hb9RI0-qvFF6t|O4emW8G+wS!`znwPudW$>3M|Hp~
zaDwp=>!!IxA&!%nGAv-uZHI3rS->5sWf+X{$Y!E(lr~vgJ3gPQK9CN%OG=<+s+ugr
z1lf<{(m^qZAPvUAqjf*F6BPX`q8Ap7(>X(b5m``n^$I=&f<!gp+aLyv1^`iv#Y`&@
zpb=)`V<xMPs%YR^Ky5jAl8s*ll&)f-`YuFs>ibNUp2$H;IO+(Ni&K4)a4rf(G-Ra}
z`dl?^H~0$_Ysni5DpPcba)T8#5>;M_RBE}7?9qaS=;fRsZ_4)x4&{P<w(vA8PI5$l
z44SpuwXdV?2A4$jv^8_Z1!LIue~aW36ajc|N?(ch)Q^f7Gxun9ZeD)+15UkC_Fgg@
z@0i}kEw$rnyLUb4U2m#GG8#I6MrmkuNIt}B=<lzgw{c&>7mZlO=IQMAy)WTc*P)AQ
z$wg&<8u0aHSU+N_5?Y(b$OjJjC5`oe%5JOLBbTwDv^PL{7A<OJgTqZpHV)?2pzST)
zxJ~oC^g7i5DdzLcgRlc%xKt?=pcuj>Mhfz|JD;aCtglxYO;QvVjri|u+x=4;;)Q0U
zVbkK%fa317adXIc^)l9=W$db%h<dv5kr)oL^q0)kwaBt6L&9yWip36S4t2JFY6>KA
z%gxmuhiWjC$@x1Ag-pFwBP28x(`wPIhAA!qXy5h**V`!x<)y2FZHwS6Z9OAF0~jNR
za-XeI^sXH3j${g%MjoFfqdGJUF}jTnw@})m=@*B>Cv}}XpXAOn8ZSd;HMG^xISs9+
zh3)jTRr}AnM$}^yE_V%deUNH@7@<?u5){@qgJV>IhDvOad&^C-n$eiyIA1u#b#<wL
zrACcZfn?ukQ%jb?x!x1kogve(h8|LbuOhE-$S7G)nz|Da+4P;dI37XPy+<*l9xj7s
zR$V0d4BHh^q$^s`px^6Wm$JO|V0y55bU2tXOpo*sGHpc(TP{N3vIwhx3dd`#*!nb`
zep*nU#WAZ}?eoea4^NIy$HSb6P1+8G*xn7`Be&pDUp%Rr8eJW|>BhLPV}Vp3?IJ1B
zxoh2An3PeHRP^Vo`8TZ?+9d7x*U_^aWlx6_c<T19JGdB}%v1TxB~R%?mZ!>U(G{`6
z{rTX$e_lob@#Ii6(<-HZ2#2ZBDSgkxP{w3UB3CO}8V_lKx4n8(7OFNaqio@dG{@Dn
zSv5p&Iz{)hbGR(K>?-W1qzFI~Gg9JMEHU-Q^@aO@!w*R)RQ36gwB!Yw^v5Q-RELrT
zyC!L0s3Xe-tRKCWRX|wmI`GC*RrQknsgr)=&CZX!IrJ;m0DE132H5&D3@}~Pe3Ahs
z3tIMfL$#=S4XEPtA?f6Ic6J(u%37xsEfQTtc1my1YhT=#am5b#y9_|hDQ4FcE~3*p
zzwVAoxQJ)O`t@4!xNP-W-BB4Ac%`R|q87xK{eZT@=|Ekuj9b^nQ0ho)8?_iiNVR{3
zslurc)ll+Poo=arn74E@if&J45-c~hPfV&#CDnzmLadb=?N)W_ZE5W@>l?kw@GLk?
zVmc&@%cA^RAZkrggF~Gfsi(fvmQ$S>i72<X<yb`gFPVfW`Vuz6$EXZVuz%-zSe5qW
zbJs*J^_kZ|&mVug-FlfHZ?{hPb+}0X{7Eeu%SYMMx-lAm7po3LjEznrKK5B!PLe6#
zE~1s*h%X_n<Y>uCQEJmWY$`q5Ch?Y%V3drP(@dycZ6wa(co7ZH&ZcnJa_qQbeun9p
z;2KBlOka7%qX>Ba?Bb$*!XDyvr`x`jEhwy#s;rWVx%(h{AJN4k6i<APb!eMDM;?o?
z;d2u0E0PU=mzVvU+nX}V2A_^u;V3|jbW`Y-1Y3`^BrSVTS+~!cV;4n)VPTNfFcMWN
zqLt*Sm<qIDW8)@U^O$wBjM;FA>1xtU%`_V?j3FBaM05Ayb!?cRcflvhpTewO<x0ia
zKM%>c(N#yqe0eogLE5gFwyU-mfLvG%KVEOrLJpFDRr$snPc=j1BBE9yH+X|#GvRBl
zfmOtLRY6z6jGL}P6?2rqCpHs&NxckvT8M}S*a=qV5hl^OYLQ8G!!?sqYQwY6oGExL
z?biRa<$pU;i@{&I&9u>6H{;N#94oo(x5TYq!cCo4%hXPLTa%>s)6Qs!76+=!i_Rl~
z>Z*`1XSe^B8RU^O)YJ@ODKL294t5nNRcO{1E#BI=-!h1(=Z{n!jo2%VyGvhovcd5w
zlUOhzfAc>iel7hrh5%Xac1Fw<fw$T8+}-F@SI}^+J-9K8SBwQ2r!Ln(GqrQn?2Q-*
zDFf7`j>qN30>Qv%EYo0qiP&Z<B4qJ048SqO?nwmgLfmKh177`y{F^?i_)U_lfe{0J
zC^Kmc&ThRsyy(47uMT!jTI|0z?DN3&mmYJlf2UVwD(|fT9AyR-;oum&j<kN<3-ld>
z-XiH0A*mpdL@XTk6o=HAgNWln+F)oA3Y=_+C5QDynzp38Yd5mxV*oOFKT-EbWWJ1I
zbWf<~DfuQ*s(J^d-H%l}%O+m5TYz?aVr#qIDrcSid|C9~{r>sgO)2y6iZ@_O`?+Jj
zfAG0%U~}mGO^w5B^JU`u(s5Lm4E2xBKgk=#TmNYPvqApvo|_uTell@FERebORCZd)
z{|an~F<7!<z9r|m$L84lVGq{EOBr)f4c+io3m$7WWXy&lWpQ)f?%kAfd{z|G*jb2b
z?V=A;6Kc*p4sNXWjp;b)fV(r5$RT&Cf4W8PY}CQY)b~U)u?iYZQUieit}s>OFtqMC
zn$XkhUf*1nk?!c4SIayI*%b2~r8g%Djnt9^m$O8G#x_W1<+Raq8uKzLCaBZcO`#Yy
zt9R|KnnGC!w3{AC3Uh>O<&ak_2FgmEq-$Na0d&s9G1ZQ{`mu-}{ch*H*Ea`Sf0<ab
z!P=(4D=?8LJd(D3Zb@g>2CT9T_}Sxye;^Y12j>3B9-(l)kg40<PWR?^KYC2zl?l9?
zskK%I%KmV+$Y+!7zTds=e-i<%XG5qp-QR;$X(ektzIymwKArjS|2R3ZX1*`vFJ;Ag
z;czDdWqK}s!D?EoX>lP|(|#8qe{0q0@hC1*op_!G?!j^(Tx9pUHMmzqH3RPfS+LM$
zK}@$B#_B!%`U%T5h0rOPt3<-Lq%UI}5lM&S>G;2=f;re4DklAZ+ehBQS|Y1btxA0{
zl`1zka3kutZq;Km_DYrHU`Ds*FPFMCGkC-5RugtI%J=4q7hlbk6g42FfBa<GzbIG)
zlFLp+u@h0OlVy}5NN|$bPW^9v=#+ic60&LaT44TI+(pecj{SzT7^b~_=`t2`Dmlep
z%l=NWP(@5>ok&{bv|AYk&Qjnp=3!BiZA5DV<-dC)TC0sKuZ?k5V$l|2ku0dp?%~l5
zx9h?A`F+=zMQPgA+YMETf0d4ptalpK+A2L-XjWRvslF!Qs)9VMcMT=VRYMApbg02J
zq>EW~M8*|JLp7q2)a7=x)tRqsn=t0F532FdSWby|b&*AewWbsjoMSeXysAQ$RmO<g
zP1>=n@H$yRk2c1Nv`nP$Bm7gJ`jNaBIQs}*Q|3Hk5W}T0%Lg%3e;buCbLld|t!_9(
zeQYJRCh!mz%BdMbKe9>Klgk#^l?5KA-(bZ(^rAD*EaE)FsyCAMhKAaY2e<XnoTMOA
z7R==nke@CE>em%KQdDFhh6mGjJ~_f$dY9?_8%>}PVe>`2PAqFlek79%Yz?gIke`vu
zb&Evc`pMEKE9^)we_KVN^(D=&2-PWj0<+9tZ}$Hr(VK8RJnE5Q_>-+wfK&Zkgjor}
zRynEFQI$93vohhcJLJgzHq2KWq)M!4Q!c?Ub7R0&T_`s=l-bSt5{$bxHn;1p47sP^
zB@X#-X(P6EYL}FM>dL?a0q3zt?k4x)c!!10v2Qu)Gy&iKe?)G$CJ&Qq$Q{+|*FVzI
zKLVOF6k@)0Rgt_LjJlWIQuc{$R!Rqg1mo^ZZT$!kWIk0wPEuD@k2~bokc0UnbFBbY
zJOEk9AJHP(fu>Zc(Gs4BnzWg29LVR-X@I$)61j+mko~rlJ^AvmNU{qzE!Q&`-9ocu
z8}*t}l0Rbqf4hXml=yA}q!&!}G|+s5j#Cm!+5<%^reCKkK`}c3!!W5E@oplmJ<R^8
z(vX^~s?{xP${|C8ZB|eI>GKEHkW{=M{q3VBBPFg+qX>+BsVhPJC?Tb8Ry{mr-kRra
zyt%?$_0sC9{a;J92Q$^ON3y&nwLfXiG8{^>ebO|oe>ZBWHjy@aO6)9=t+an#M)@@4
zUbZ7@NLL^mg3XZZF?AoNp;!j84JFtQ7=o9L<mFlMq|+7m#gY<X^Y*QhG8K{(q>f*X
zMC&MKn}i^FK8R1)qv~A`de;Yvc=m8n{&K}d>X7B45^xbn-aR{16l-dwkuIdloQsg9
zX8(lIe|QOSA$3p0V2MknlXL&(_~!)h7B^+eE~@EGjm>G5l3yEMozD4%h1YjA$+v}R
z2^queT+pkbOy7}lO~uf%D{~ZsT2py}wLjHVLeKU!6CUsgLkD;f*j*~aq@g;?C8KuJ
zCeJ`smtmjh@tV|9<E`e9yC*tdpx0IVQzAEpe?wP@(EBv;`J&z+nd8dX8?V65v-06}
zsh@W$>a(MOz#5AwwiCy*8)QkO)bvG}QL~A;mYo=lGoq){PhjWa&44(W=PLUoHZTUt
zwPw<JvMU;5SgOd-8(p-zqcRQvzj97^?K~*@t=?rRQ?U;<nhMYm;V6BYBogxtt<&8e
ze~fglLJnqrfJVs!BTuRzuQ15-?TGqColmmVuDdhlKA@4N!LL6=35<6uJ-?;&{3~c@
z6(Gw8wxmdN=~C+IhzD*mjv-Otif7cGDau;V$u2Gl?~_sZkg+EosR<8QF&I#PWB(@)
zv2c3{xRnxF+Ws9X?TnHJG^!Z?iy=(7e~gsKQ@<@m%;aV|0uVvv%K!_wjR9d{2m+5Q
z6(8URyV~T5So$amex;P=dizH!?LvWl(`vT|Wn@TgYHYuJ3Evc~p;T*ZO1m@xl!-4l
z_f*;xX69)bu*rnlwB^JLroI@%X%;;4Pz1^ZlO#@Efz(F(qaXkI4XM#u5~o{<fA9eX
zP4;GVJ5*5k-!qQFbSkqSldXWpzhk-w?wCgpWX3|72$Gbfsf3<LXH}A)1>nq=4NtHZ
zO70mPeEuQ$Dfu<(cSeI!wmP4(xSn^R$_YPBKT|(5xXu!*fF9>B68Lw)<4$Xw625rj
z-;b+oyGc?d3sH?L%xsf<Q1c6=f6SKk2Mh4L=M2Pqly6LI!SOnSvkEbe(uoIvqjp`D
zf={%#SjLNG41RYM=SC3fftf`XUBA=0?ceN2l_?)CHH?atnqjK~drEG977s0`0{v2l
zr)Z(q=Y#H@;Wf+DU(vOkLCdF#b9VxzsSl70ozph^_JxgOo%4+PUZ34Ue>*2Bb2TC=
zy-c>yzz#^oGq<b2J)T%%JCHPavpP(KB5=XNi3sPcFt@o1WHh(g*+(40tvOK}Bp>d6
z{^i4anZk^+Y|MPIl4)cznZaW*<^BmF%!}UOw$t5D*W_Tv<jr4hP2N(WB(-k0VP3*_
zoyK%*Y%j^|xT_t<q5L2Jf50B3c6ew+6mc;4ISs|QVvVDYICO=&z26;Nl(NCsKH!;6
zrmVqvf|FKS((GSec27bDs?wYzG&0p_`6+g>lNSIJ$_gXd^Hc<p>6VNE3d&^tR4n;k
z$a1v7(kfYwTj9{3blr&(PTc)5El47`cd4VEyFmecIrTGvcX<bme<I;>S-PZC(bVqv
zA{37tb2l(K19WogWLp5fE&w7boJGik91QW_8Net^V!f^2m}gJSm3crvk_|+g$sJ_Q
zqKHnh={Z0xkWCNdPoU}@1K2R5BZ_qk*^s%IV~1v-E>)K(hkv7y=#Y<@M2?*wr~`T<
zl%76U&lszk<iqJjf0(0y{aL``UI4Sm#3oGm+%#=Z*-QHU`$}~zUsRqid_Hy-6dN<_
zWD#$5VW8tk_{$h%D%}*Er}gS!4Lm5MVxF*6`VA+kTPaG8`GBtwVvV||Km)dd?`==0
z{XR%+cjbdD62Ao!mEHK_RKIPty6ybHn(E7+QWAKSqiqJ8f5)g=zSSJ?fb`xTioLzN
zS{B?E+{S_%t<-$;RUS-4CP`%*?ZfY{!EHgsx8#98cr>Oe>755W=HQm#5mgwWU3UqZ
z5G&_}tV!D17Pcs<K1y(Sm_kW{2|R-JGH&54Ml8F#QF8{35SaD3h_FE-x=3&%N=A&j
zOGo|Fc*%XQf8LNR^N)XY{`t=m^f$MB@J;z3{QvBIi*uXE*5|*{SD8B{*LoN|EXmGP
zQl)3&T}RH^&fNReWHyCGTV{el3&^o%zx%&$cLNe2fj|P$jn1uGlNe#8PoLj;^y%}!
zbcU=VnZ^%Ee|+n<og+V4P7TH*ZP@0%<c*&h$NCwUe}i_H&j~@lUh7X$D;LupQn2mb
zaHu+3A{vBtdgM+FZ86g=RAtQpiF?F4J^t`G337hIa&Em{2{e;)zzAT3VZ^x~^ZP`!
z!xmL43hS<An2M?s^}<0^v~k&N0WywvrZQ}zH2?*tje?o1>Jm{2Z~Z8s(U{HAivWic
zfCE-ke_4NvoIgP`oG=uu#atpMN9#K_H<W{I%;~9;enx!FCCkEkq~koOW3iNIq62+|
zUv1)es<w)gG{FJ<lnOtp{uGIyN5i&_89)di6bm7r3X-rgr9=B7^$?;j5F|N=n^%G#
zz{?4%ob7lC>Kz9%fS40SOjN6dpXex*_=<DEe@0LcUw&pp&j1oLPENTLkngIYYfBpQ
zBu7_?eHu<20Qo*Oix4No{dWDL-{5iK^x9gQ8?l$x&0eiDa$v^U>Nl(|CC8grUoIWA
zQaR<snV3vTIu@2+38KXnLQez*I}3g^T1pn45VupZ)ya3NW}F(}P4mFsqJ<jtMV}uR
zfBXXd5e7bsq=uay!%Imsg44eh_2~(-vau(VoW(1;Q2~7c`Xa0^Uf2+@x9%{oEqX$7
zhNL>v;Xz%1x;U!3fGnoM!*p|5$C{Jd^dg%oYyCsp*;}*(bUo;py%x-fa~_?=cV=l0
zYt4%Rj|xEeDOlsKHynmBls1;1Oqfr|e@5GRCFhIz)RgH&0|UJ@rKVH^xj(V+o%~8&
z04l|=zsRqu(~@K9{FT=G03g5+pr2?{3kN0P_7={B>50tR=Nhz$hA}7WafjZ!9;9Ov
z++PKoP>@a7=uF~oNfO7Ci%zS3#bbCi!qJWIt|BM{BSW!5c8UR|epD;(4K)Z3e=xA0
zWv9^*aT8Y#y@^>LBvhqcUf5qFrvg#3dqbigxV(CiB&k%Q?lSR=p5qsDg$}S|JOb`;
z#2x>1Uqa&9+5w7gn(y{7y`@f)jp@{ZIF>>vjRUJ2>bUOf7)%*T=la<y4U@>aZY-%O
zrt{dyi!)hQ=L^kgp}-n)kk{F-e=$w#ytL7=cdLcIY+PK{8;LD+@O7}zTnP!a&|v}L
zaGB+6e7l*0$l<mN-4H*%qx^0UG@fi%tMSM*Q&onqwkf}TOaCg#rb!g3uwxB2&Hx;)
zQCyeO@7u?dP;WI`1A*k6^^fxA*Rya8>sTLqq;zQ%gz)#YX{y(YA1QcZf6Tl@9M3((
z&LH3&ktcpBJQG!}B0!n&NEV0D%Vx7v6Nh98^~vg3t6tlPJzlzm3Bq7cah8Zx@iX(b
zvWHs=-^<85|8F_r2*ITILyveYi5~z`8ah+Yv<!uC#KfweNE3>U)Z`&GF?pyM*kU(&
z>=0afb|w#a4r}d3chxXEf9RQb4}Ag8;Oi{l={{oL?5qk#1{4EwIx^sYf4ROJ4$$Xr
z|9^Eq4DUGr08B4AA^_ZLvOU>eZ`a%X^>A8X`<Bh&lkHwy6c~S6jMwmtIA|iok;z$T
zLfa$}XL4cfh}d(x<MHP_Sy5K;t^~CnV0p`Ad98^Y4n`Nlb#S<gf7J#)FMRZ6+<&;f
z9p9tT-R&n4Zp9xy4$*|pv~v#-pv6v0i#;v$Z<A*{`CbJBc8aVwX7g`#1(<)b%x?hm
zPm=l5>9@ZYmJQ55QRdId>BY7&@c$I~&&v9~vF;=QfKvnjZ9hT$ews@Q060Ye%w;(8
zohIi^W1)l3f0C?ce<AB6tPj<#A5x+Y;pt;?dHC<{r_VPDqPYE0do<`L7VhBdjP35D
zD7KeCxVhx8g>A{gSgF>}&~M0iwCd<*&Hn8dHfUUWGKpgsSr+1O*~d7(sM~(M@}SNz
zdWR-eQ#Z_6wNX1q^&0!n**RK2YtUz{;Ir^Ph(dCU($Wsnf5K=c^qoa#wsX$j_kB7Y
zxK4O!;<>1-sZOeue)pkrjw*xO4;8`<r$qf7x}f|Lnl%lFHI&4SA={$xhcc!||I=qI
zq71_F<e5;`5Sl-RW|*$%Q2tDeQ0kXJEYc`vG<EuHKWYnH9ucA}@`=!!t5Ks~yQIXv
zY1Z0Af3v^Ve}c^+1>#%hqI1PaxFq(eUki(tWt+s&v%1zup4f&-`N>Av5HDrrv(OSI
z`ao13yRq{}<CSLJR;kNHg_3$<=V(d(R-xVs*`x{8r}SsX<|XL_+`sJfu2`=S-&XYM
z<aD+5AQpVx@yMM=?3>eg<T-EhgbIeYr=t-$L)N^#e@?#pbj#djZrveK=<u9)jGk>f
z=ml*>#nujTt;<HMS=)vRIG%f(!`X5jz7K}IFCRXH+g~PUeDnc5J>yFkedCt#QE7uo
zj`E??ukX-H{)+wwYQ%U<Og&y+47hAM>YfJUJpsX3x_16PP)F&b&qX5Kiepo=*Y0yU
zdTXERe|ndgAS=10@O^~iDkw7FZ*{n-a`}nsl0WKi>$N{!Ju~i-kNi$7`El*Vk{>)b
zmLM9G3@5Tzi90|=9wdBc0}+U1z>_ALY*~_FD&${#=k%L#F)2CD`062@q^jX5))k4S
zHUvL-QZ>!%+3Y_0p(-|2%QdOxK4D8>PpA~IfA3RuELrN`&eM+fqHfqsqx-*3*5=PL
z)$;DoY^bZFM!bK}o%Z~l`d~!DEL))h=KpB5CMuMMy&`dH;$^iaHCmMM>;qRST@qK1
zw(l4Hc<a{C4{TMT0`Jk{J(3<(oem~?hg#lh{!(q5eXX-)^byrC^z`rj=xkNf{n6W#
zfA#y1Y-Q^$4LuLBlrTW1D_H$V9950|BGppC56O?}EbybjPT#sgH+i`lwB&muG=j<D
z-BMEZiMCL%^fxvQ-^P%yFy<1b6_?UXHi^f~D5y~q$ZoM|-4P$himK3)UCW?oC`t%x
zDNF=r6xBB;z$s4}8O^R5x$u5gXtNTcf3Uot{p#Wcn?%rMY6V=KY!hMizM@Vwn629E
z^$LXo`$6)*b4i}MTGV<reyqR9eHb(-GX3Y+*ST-edTk#8v~YB#i=6&xzE6V<)i$5A
zUP|llcHeS;qT?wfr}O9fH#PN9n#*i~8(pM-g|JiE!%p7N4mNdcPv$oUozbWqf6Nc;
zX(r5fH;1$3Qa*dEcYHtR*f*Cg8`r9`CD8^?)tO8jTH--!S{1*}WunZ8Qx<FZX}93P
z8Vj87L4-MpB9M2Ez5~v1${A*D&(8k%qyFcg=crOWOLFbk9@KiZaIh>|*#=){%GQ0v
zzWEd^Ex!ODCqJfVEmzU9zgM3Oe{&{l<j)0l@~ovLc}lrTe{cp)Rq1DqGxEOz(Vv6U
z{>GhjYCIugU;?wW67<ffHQSfntHi?5yJ+{FTdkFFxQ|i`2MEHUg0Qe?tOwgR?K0`e
z19%+ZaR6sB<c#NW{8In3Y-WS?o@I?_ebX0Rbz&i+H)x!esuso3|A>75e`3}1)p$`A
zHjtD3K2=5g&SEdoxmPd?bYRc}lbz2Z@BvY;7y$4b^)m7QU8r;{?+OL1(Gl5rKtf=H
z`A5Ru%cXoKI9Q#yTt{>gERjkA0g-d<Qg0va{7+1)`wZ*9eCQ<tL2q<%Im{~=H)|>=
z05;NE#96wpgHyck2TVEIf6^0W%Yn8zQEd~E&*kgJRRO2~RCYtf?cD`v0krHwi?7iW
zzy;v43of)R6yOAKvI{3Hk^(>hpg13=oD#?&#QKl{gffGWC}^=A_Oo7OmGjixcO>Mp
z+wAq%z3BiFqC+BpgAIjPA2o)pD{(9*3(onP?1rfml`+W$6z)o5f9yNIcO?mK$y?2f
zQGJzU<>zehlM~@AhfK8bY;JfZzEJLQVj}!xm{qk-N7VT*KF|JNY80H5QzbCN)u$Z|
zKZUyh)UZCapQ}#<QAe6|nFkf>j=lrJvE>}i429K5dRnD&c81=)LmMsaJVGg%{U&h>
zdIFIdw3h-Yf+RXVf0g&<K4M?8spA`sHY+Ix?=4d<`n|VFmHl*%)Ovc)e3uuO-d!(&
zQRk51|JlviEIl)D#d+fF;W8Vqq=DUqI^vwHa^Jf?^be$3lD^P1#}Mr(PArvcv{MwX
z>aE7AHZh4P@?bvsGn;$)N3`Kpu$we}V|3@<_jPUC*3`D`cBZy%yPw*&Z5vbDn%cJQ
z_UZS(e$Sg^B`=bd<nDXUK4;$@Arp5uH?Po|kut~b@fU#EV%u@qor+%ciRnO17uV*2
z$|JNDM$KOddC0?)<ML13ibTCmiOmk;%fqmmYYUDMm^Wu10D$jH@SB*M=N$olq(^Ui
zGe!V@1ktYN>_Ep&sOFfRiB&E~>taQj?_D}p7Ec}g=Aa>{$OQMkB6OhkP3=?X1n__@
zGz{|>E&+gCn>}~6n?d1m>E(K|;akXW*>X46zN-4r<s}iAU|l;MXOkxprA|tEr?!ST
z*3Yn##Kt<1LatCR#@`CiuS(pW(AIwoTvD-mgx*j_Gf9~z^vR>JgEeclMKJproURef
z`SgghJ9&N02*&U5O7&^P?o7(yeDvnw&lL_(mjK1dzxhgLmOnH{bGBO1$WWki2}b-~
z)qQ}^?Dx&8S9{V7fjb!_uWoPCUQ53Vs&W^hg`LLRentJ&8RT30^MI_c&ZkN-ftHCS
zP9)do#POB8d!uK=*O$ToJ)gL;YBxUqGs;K%?#IV?cP_h%&Ue+>*31Q=Z*s~5;<3Ov
z0ATcbImT+*Bo)!xU;IvKv-CctVU(09+0Kd8^!Kw<(}!SCOn9KMqVX!7$FVm^J_EQz
zg`a_9+V#>Z528@qiFFMke20<}1e(5`Ayer=q&YX<HYj7};>a#A?8$_t2(mzH5=<|2
zsY1K_GrfwYX{L!w$BS@8(C?iJ2Do^l6M#~ct{x1pETy~xqE5wwQAUZ#$w`dY>4uO6
zJx<Y!PK7v+>i*Q_`FUqpsBolWw6Kv8jp?0TzwVQiu}M<bqEoqCyVif=0*E1?=uyv}
zC<(4)EeYJJwk<NNFhN#TUUuu}Pi3-a!SkD;xWZ4e4|^ZSezR*LN!*IfRq3az0wl@|
z<Sn1FFICTq#fMxN5q>ah`&5j-{gX?CB42d>-Zcv(JxpyOH`7^dCg&T9?YIRw0a3y_
z%YkzuPsTk3?N=nGQrN<U_3EzUx`B$}ywM@sOwEQwN)4t+kW%;q*`2QJhBuTzjJ(eE
zszSKAwXyLPN%0sy*pwaQa?S6Z56}qTMFRUKdFD?``bDMo0c_ZZRfn&5-l;w<Dv(!p
zH8*l~O>6dS^z1v5%?ub7Na#rEYRPY(D}XtB!GI4byh_c4eDHccnm}?H0X;B{DqO!E
zrgBG+a(`FBgg@)#LV_lur|xwN%B9wGxkV#KWLh@Lt!scUbsCu`Ij6eY0Khw(QSGD@
zsE4nr^-)q4K_N&yPz7j2NVIA3GVVr|;1VP_FA?Z!zh#(kgGo^4{)Cb&H^Sm0Ub0-%
z*pNlW&i3+YV<!*u$4QlHt_!BGV1LE0yHI?l0ZH#RFA50s_-?T5f8)LT=jd(wVTm-m
zq0kV$ggdz~UsWKGH;69+fZrc1Cq}^`fgnofLl~ebA9WS9syF-l0RgwwN-S?}qDLHe
z_DscyJ*y%4<k%vJ0ZthK_XI`^6W7TwF=C7=@EvlAkEk51kMkR>1Z=!KA`e2N<Fh%w
zTcp=MZ$aI7<Ozvq!<1LG;$y!yFR$t}uEQ4eA1L>48F=fl{lTg80KAeiC^h`yXAiQ|
z$FGe7iTc$7a_($w<<jSYvph?SdGeyViFA=$eY3gbOGkN*G0qwpNCCzwl!Sa6KG;`(
zy~caExvR-3z*vMRz`&S8r)TZtr&))IP_f6s+@MxODE64>wFfSW&6&@YBp3#f8tJih
zXRgN@t;bPu=gs5|0T>>dW)}WL$3(}aw8vEkwN(uWciFI2oN@O)44I3i6}3!t>-yn_
zQp~d8nY(}U4%|;#GopPdae1LOte;mnsWi9mBOl<cVM`7Q@37P|^pxm$%J%nvIg=+5
z!kjX(C}k!UBTz@rj1jH}LMwEG&mX$-R1fPfW@KyOm`BTi0F<yLrZCceJ}aXOp;%v&
zx>CBqntTrJG-GHY24SrYSwp=^p_%obQ_<RBwJ?Bg6x1_XC68X-2Rij24Xt~C)8isl
zL(J@o;z+^Z%9ez@kjebX=j1TEWr|S<WCmt1g|L2#dkMXAP)tQ;9hVcN)t#?5B}z^n
zLvrr|@+*cJAW;u(zjn0W)Ro*ti7Y$D+B;J@BeB8vj&MF0Jv46S5#dFV($lyN6Z%!y
zgjQUmYvp|2b?LqnbO09dI_I9u%1hqS7p%Fp93D!8&1pr+YZZ9mXy_E{C6N0bj^Ek%
z(aF*#sV{>ZIAR}fUuPN`31)9A?GzuEO(|j#xDFl%F!7H^Xoc;qE@%gDh3&+*-}%V!
z-G*s`w(qd#rIbdRzczDg-~szBx97RhB$))>M5m3Bq_xcq$RluZeIn+xfbx`S_Ec8p
zmZmz9Uu3}jtbDvN^6x^9L+C3mjF;!mF;?RiL=jGyJ!@LKG&Pih>PAQaD*gW0fvv|0
zTeKemP>j4K@v4Jf#kXA{b9M0?L)MPrXG;k3;XgF;W_}+z7h44hY`jaR8$^bQ7y3Y#
zjp|ljD*wA`L$LJ2o6%?@xxX|n<|Z!V{;|lIxNW*$hZSPN5G^2IP0-dP)0mGURCiF|
z7Mo9w7l<#oCn!FzSge5`09D!u{XlxxrHu0itm0wTXJc)p2W^D&AD)I|ji7O0^dh74
zQ2e@Wz0BcgC`H_RBE2VoxehdTHx76H9dy1c{-ODs?!E+Y=(%|d20cPTkAR#yrRont
zJ04XiVvm<oPGE-DaVcN9dKexyDwJ_y%woQ9?vs<!k)hUO()(T$-(dK3D(lF_FlrGD
z*yq}~B|Q~$+Gtg}Kb`mlJ%j(9iYM`gC>k~T{Z827b`ZLY|K@YT7da0VfY;8yy`?Ac
z!H;&;>%!-Y<*Uv_nxyT!@My*Zx&+^s*KVLe&9#;%!5fc>Ar6zmhC3EJEOVnMwF&f=
zTzWuT916EdZje>H>10yIFeGGW>L=t2U<|_-AGFo|b5D4s{$lK6J)!g^Jl#ZvU9O;_
zp*Kvh+af;u+VL!F?}{{<8~G^pK6NxyHZp(I;}vSjVc3_uNtVv?jPmQZ{G+N=Sj#})
za=PP_4d`G^%=lGMJxz?Nw*PD_(rm!2Lb$L%=RCMW4)d^w^p$#1Xp~}%n5jb!013`-
z{}BEZkN);B`rdx<%!g*X5ibie@Lz<nU6~ztQyG$fqW7<Ok-q|_&^`a00S|Lyeaxh>
zkk~GO$wbYX4uKTNcgS*GdDVj~mr(->g6e{xV^>T_u{o*ojkTfpZLoujznlTGwte0;
zINWUejkRnux!trTo@!=5I~bT7fHf673cSoet_}w=a$ar@L8J{x7l*_t;K*_N)T^Kw
zb@%5}|7_0{DoW71z6QE(Tw3|3(kf6jNEKMC6fVu~s&W=j(h}p&WW5?4b)|90XO$U!
z%OXC#V~$Fg^d2REB!P<%{LmVyfqwZZyT<)f#+ONX<A`Yte=`h`b3thc2b|0QI%g;K
z6TI<N7<*rY_qv0-VhDQ_GN&#yh;M#vgzJV;hp9uG{4-fs6_^iH4pB~@sQm4&T7A8e
z)Q##MPPsXp1w)m5E!A9rE1E)+QavkRdS`VACyMLcJ0P4_<19#pmY0m6X+e<-t!7Y(
z#ZLIx{O)M>-EO84L7HTAAa#T@IK!v4+nQExhIfgIs0uwuq{kG7XWt?R=ZIrRfpH*`
zIJp;^(jF@vdsV6Fw9+w}m^ehy0fQ1QnU(>!u@|7#@h36ZXnrZtMTrRD6Y2t<ss2i<
z;4m|I=k}z94R44#m{Y<z){Dh1#9NePv~t$GbQ3XD|LL2QXSAYi=x&A*z+{c*;R|BT
z6J;=Qj)r>rw->8rU&9n@z;$o{ys3e|rXKORHHIbmQr^cNy=L!d2CvV({Ug=u58rV6
znBVAdSLm&2Ckw)wZf=7RY61cC9w#y57M#A|YlaxZg`)NB$k`5M)MEDv?uG?k@?vM}
z%A;vIG6Tcb<Q+$cb%)yc49EmeVN~JcB~zUBd7_H49$=q?F$QiUA1tHlV}I1ul=e4{
z0oJO8%)ye}ZqV)Oa@+tT^N#A#Mu34aW0dtgMK5E+#tQ7f=)cHMdG`DBdvIJD=Gqj5
zwcr@v!<fKqMXe+Bl1Pf3n;#Pj+lW=SX>cVvFhQb}DX#}0!&Rh-kkM6wm1ASEO3)7d
zZjrz0u7J70!=u%?LlL;ICbqHvyD3s2Lk7-{Du}HR9D(6YCOfd<O?y42;mzrgFJ)$M
zWS`6=Y7(4MAI0Fl48cSse3z2sU?!DsUl;IU0rVIR9Af@6DvBCZ*F6EN-z-ni7O_4=
z&L3z#a%qU5W+Ca=qDCbag8nrzCD0A&UOFg%m5936JcXb-566?RI#1WiwQ(br!6lWB
zs72W_&ZcJ#-7iEbrnwuma~Mog`e$;`8pLrQs%ggTZ_YcaHoPO3P@wXU(qBFHxp6?g
zKbr|=(kaLd%3dGgR~0*m8dQNnvd2U!${h&cgF8vhA5iW><eOXiX+gJvb3oY#)&VR~
zcOZy4p8Da1LoR|#$VL`>e2J6<HdT{N=FVrB<I1s>Tu7|igP27Az^nNg_IjJu5F`Vi
z{w9xe-+-X=;u;x7^*NYUfcBSR!idF5F_v)@NHvyY@zo#7UUZFd=1m^m!$|YJE%R~`
zi1Ig1V*MxAZN0z{a#2UwKP{0({|^wRUF!6^_OUfw`zA4AT~=OIw7r;LWnw;Rue99)
zX*E0FO&f9{Z5(MznzH>;fqd(V^od++Ff){D#UCS?^a`1Y0mnY{Z*z^5K!LWI73Hbu
z8rU>YC?4k#-ey;0e88DhQde}=<*{cG^Y8F|dPk9LDTz0Dh-%y1=pAXaS|~twjvyB%
z?+z>lxRZ&Nt1L`R;zZm4QQNOQJWWQLG!e_C#sJh&=<C5KyOEx#o{f#XD|evz=t#0*
zTn}2bvaRo}u2+(cKCC0D+@$+-0y5{+AP0VyOiKI!VEZhE7&9h?H~VlmLa`e{W|jMW
zK)^+4(gpfGMd-HTb$6)Bs1;x=s8Nd}RK)eV6XRQ^*~U7ju-#%(O>K46Ohx;F8?);s
z+qFl<O7Ee?fcoWc;ICt3{R(?o^2!$1vXf|f`cM(wGRe49X=4)Ec+SXF+9-7S)N4Jn
z0+QQmYSJ~hVl`~?HW>t4d4T%$(x4Q`vhhUsz~2D5D({riHZ#Llo_D|yRR?;l)eFgX
z=<{WM@_wp2zGENpza`4qq;<67S#xe-`9dJc;!2F9>g#;Ofu~R2a3q=Ty(Cwg*-(Ev
za-J6<!jMUzOG~wr0iuXA^l6TuMo57Pf#ya?a;P;Ezc=0lrYX>^S$fbm71q`~9f+oh
zl_uy}deWcKt|{bMzYQ2O^`D2KeYC?{?i13RtSgrn20m3Le+YGU0lG=X_HJMYq&O*9
z083^g6%nka(JET`^=&XhS*Uc<ZzX>v)Qe_dW&D1LI?z@c)t`nEuw+VXLuj0T{M3B#
zp9ag;K*;uZfuH(fF(@!EO<MjhWkOa+y5A8FjHc=pXYCcg0ra@nA-^Ig2S=}@2z2t8
zy^vK)R{n?%GS}h6Z%|DcHzY1(#;qA^ET;^vg^A|mJh17d8A^@IVSDN5+nQ-yggDu4
zY{+P*?{LQ&>0cVpplO%uw*KO{wrJD{btu`jf~1@)$8bOD7783t_mPySFzg#%7c3$i
zPv1xcqxNWV01o5@Q6wtec(wnRgYj9uctzSh4M8g%gyBe_k%#4nK8OHmY>DLlgX*46
z3W>d-lP=`7$>KzBovNu%Ce_?;lG&=Yq7}7^>3Nxt-m<4WK|bzQQP8f<wr=(f`FpZ|
zv%T3@2ly=a3Vdtj5Z$;Uys8-?jG*A2hbhG`1yFrCKYoflU-{buKYqb>aY+QxEJ7Tn
z4CFgJ&m;M?efe;f<g!F*F0%ZXX?WN~b#B%4xN+dzdlj)5@mgyyX!0Tx=+rALueX;W
zdTLj@4S3$gRFQ=WLZS;v7cle%)bhx0;iwh6-brs}vvFh8MekGYEu}W{Six0tDA*su
z1K#Zi7r$p~aLtb;DZT=^ZDX9Qv&dvgB@W1>Xl2Ry4povq>}}qOd!@=T<Z_H9zmF_$
zkxoWq%dDs0zLg&VH3&l|J#!0#PSxD07FI^tN3|ch9LYM*6+}uzN|e8p@?QkI8KpZ~
zoQftAxSSFv^vz&hoEm%HJFFYhKsG*205f`R@1XbD>wA6P#}dly^h_I{&<B*go7+o|
z{hQl;ut&RBwXE{G?blxcj6gB>W4DFdpM94@$ZZ*UJ_On;Fr%y|x?`%lb2F8Hf;I3u
zb|D)%9W_i3h%rnZdTC=ORce1KzlPl2ZC`f0Z{X#x!Sv)v=ltv{_4T&WC{e_~fC&dO
z0TM*N0W6dq)wzuTCNjnl<DW2xmGBehj3k+j82}D{S?T`0J}T>x%lXoJ-j<`A6?pe$
z)h6VpTKdAp*i}z4p7R*)s+}FGSL9P(;7$sWfM|P}KqiIgztKFgq6a^T+B6Q}_92%}
zPxKXeFTk!4OH;Cb9!MbzMXH(u<}|@F^TezSpw>x6FljN1e}s6|9dDVAO>1ORvd?^X
zBS}y>w#QUd$dmeGp-E78rSGlcAg>yF{r?(XbTCWwZ3oKtp{Smh_YpZn7QR=y&YuWF
z+zwanl&JU)phQI-5WYWOZQTI?AD);sm%qn~Zf-XLsP2zyg{s%g9&J2;7w-G4ACP5E
zTyQ1-y`l#L-Qmo|n<=`YpW-5rZj0$vq1vU6^cIF*wtW$TR->PY;)N50#!RwL)_c6F
znUmJoA~QcC(BWQ~^eZEfaws+BU6@3ZgSV#vZp|}=ury`20a=(l7~TS7itDReZI80I
zYN_LR;Wny$4>NnZ+7|_gHKc3hGKkn^|KqZBTdQUdbw<(4_LHLF#i!+2{=i$E;9}1?
zAysL;MVmK^sCQqeBhyNFmrY=sdhthZHSw(D78y|r{|JPY?OA!{WG=+{3}N(WjYuR@
zTdvWpexs@n`9Upf(0iDVQ#w45-LYE$^AFvGVXwc_$l1#YKLlJ~w7M>?J5aGr{7YFn
zq}N9xnuzRUqTp!Qx(aJ>MPW{HIwg@bYjSoB3P7dO0UL_HEXapRLhU>*EblWZa2D1f
zR^Z035X^^Zf?^j!H#dNiaU20jv4W+{hk2-5G*t%w`}l#Y1D=6XrwE=w2qvutK0ckU
zSnTfIBA9IG+XN6yE&wc_=7=Qk{fu>@26#3r56!2^3N#j8Z=jL;eRm9%2>WT{%AK%E
zslS?LqpXt6Tqu-3u4DbwL)8dE&K0%h+@KX``O>$_^?Ig@nL=AmF@sRq*X|2vB1l36
zlM0^0B98`h3qY+X_OKO*;8pySn-qeN2DTT$;X2q2sem3o2`&m(WDc!t^sbkM2$tV_
z;#&>ilxzFkjKvonq(vV#l{=3xkPky}nP-D~ehZ@&Xl}C>OSl&y@I`7s)~K-9!PbNv
zT>Z^X1vli-F@HIp#~Q%qu``L{ve{A2H+LRQUwhe6@R&RynfKeq^0UU?m{YR0@-{6T
zSI2~ly8!q($~w&s=~(?yp0hc@P?j#43qTHSPb(&{!=Z6U7-F<2rtAdKDGO>zz67&8
z>OU=vwFKSam=5SNsD7*2b7x$51gS`Ur41kE?W#MA{X6?pA2!uOhl%LVWIw7XCX6R<
z14yVhwp_t!gzdM6u_IK$$d&+Ftu|Rog#Di^8(@!aJ2I*lG+50VQCPHo)s&%=gMpuZ
zppd#w$a?n3uP`CU!5E8bAZcGaCML<Kwd|r&zjPfl(s~fPek+n{f0b(W3H_TQLBRKV
z%Nu)(QaTA5C#dP~b_yZgvqCIF$<`R6Q{cs@Shdq`!eJM6Q~L@qS1vlTgm^>MHxO$!
zHz1>-j1GEJ?gANjtFw+1O$&*YpCrs(^4U+Ox)UESt#X)70&(BC`!B`intlzAmbB|l
zU_|j=fbXj^ZSmk&*_OHP(@>vpx(dUATeSh4(_#>Z5_Saj!nl2YL|%gwnXQmXqF=IH
z`#M;e^>}ob;%XAHSAMQwd_~I3@i41y7vN4EU@&=V!{IVOxfo%qJpDU(cFz90YMI&$
z8nU+tHyA>gu0k=jM0-w^a+|!XC~<bj_arEqVpN|gbi{aG@^eX9dHOP=j5>!<Sy0Eq
zo0+NXM<kD@)HOo$r&TRQ{5Ns{HUmA}Lf=)aR&JWZ;6KO3z8XB#BZvGo2W?v_dcbbb
zcPM7+joSW>T$PmCIZ;k;h{ReRJ<fK_+{?ri{F}EI;M4x|JudE>@q@855XFBo9fTaj
zk3vuhFC$OX?}x;}lTabsw=^HzhJ#rnm=<1*D*aTq9MI^e95s$8DDM~9ES6L1M2Ui$
zbUfHkcOl6^+G$zt1YyZR!ezju0WqJ<mAY*Td|k}INNNPKDnEEh7yKVy#=*d7QAdC9
zk}mjz7iNFBRT@okhZxw+?YWkl!CCYpyODEgUp}i*pO9V-G&uhjs>&8WTAR^-TeieR
zGFntvA}Ov+(-_7U*Z+HXV~w>e8xuoJOH3lJq6rx$%tQdq_4q(|SRJqNmst^5a7f=T
z)35uH%zs&h`QrKNu7c4ZdC;?Hp%i7mcLG&F0wqaxNAQ=>tql(wd58E=C$?7^gdh9)
zbFvUYv^auU1}ThLlo`<a=LR^_YcZrd*voYFWyTPE6}#fF^YT4`m9GuX;F(E8C&Th!
zL%CrUX(a&6EaljUE*SsgLuHMA+>vsbKrR?Rl<SQ4C;TZ36RLf``FAEM`xMZj(SAnD
z6`E5cJP*yxO)8o%(cq8g$s5xp<-;jq%WJpn54rzZ9oFvuk*dhB?2%Tyab;u<Z%lzg
z(gLUa(LbH5voV6O-@CMYS*hbD$`$AX!Ujl-^yu`QaJU!sHJE5+;POnbe%t|3|5A`S
zVSe}&V!YP|n3Le%uAfk`zs?bTjGLh07U`n}#106vYyn=>e-7%3jUfdj`$pkr;{t#y
zFty8Mb*TQnRl2@cU=#d>Uxkm`>;>K@?)d+#>%S9q7N|rWKcq|8RRj=nLHDzE6X@y{
zZo;$E^N_n5xB&|WNWs6MlVQndC>1)O_$zw<&gB2*Ke8ZJd<F751$xF$@}*4JS;!Na
zSbnkQi-(hmlf9Bni`KR9_#gd&gIKnNUlU3Y1{WG31u+=!T1;S`q`|=Dl%X4&W;7H3
z^oYKf<1IJr0@zSCK-V4fXR+(rPrJ|aQ-cb0bR2>wH(g>3$F<J`Vpa|AO<5~aZ0kUS
z{V=?14xi8-f>S1vC0s^lxzqBoV;9xsvS8`Lwbg*}<xT2!=BmWLVN&$VX=U@k?DBe<
z8RZ%kw&C1cf(k|3*y;ePmL}oR`U5G7IC!i60is!zkOW=}z!6LuMn==ay>#~c$fS+^
znNR)XG{;(`^ZqrYrV*iT(SX=)WPjAhi!|VEZ89x3p{Usm4ymY_Y-*#ZIh2xs|B=^3
z%_1mbfzBztLuGP8^&2Ly4u<SW@{NaWKfkWxe_S_MMNz%k0yVFW`sb}1sQggfg~9zZ
zJ%OK^Y(~1e_umG|R9~V2q5<u(n+0Wbq@8LI@40jBhm|qTcpykUH?S3t!A785$rzm*
z3_aB((bVyFQXp<m0(jv1wM?LbHf`OuXId~x`)lauocNnSFNL<}pp=Ay)f=iLrHY*r
zBt~Dd=+ul`dnI*Nw%%&6jetzywqF}Rg<mB=8m`xF38TnUKiz=CV=dt*MGIz$?!(z`
z0g@=EQW%o+ikrGxU3l`$9on-jEx@b?ynkv}xcJn$2AjPszuK~F%d!PR_BYudk-lXi
zcz;UZaf&i)>sRE2!ro(G6Vj>%V7uKWaL)OKw5^jpN6rKABdNkdToSqp+K!5oQ0p_m
zzYS#<0O?&J?0A_5E`nC+FDjlm%wGamIpL%NeOV%MK1oLnpKyhPPIEQ=`Gv}BY$|3v
zeb~p)4P_yW76p|=z9dCDhcQ@&qY)5ZYveo}?@sgX3pa3GPCVOp`kuJaPbTqxa@jFs
z!1gTl;tTkwDrVoHV}rrU%1ZsU6((8$D?8>05{N<HQm0)qTYIxWXAS0X{RUIV0!Q+T
zaYDCKFIdN-pgfMzey2TPMXgwC&@qOP3Zi%H`6YmCB8~$1xME0=Z7ZTAMTDkeO3lX~
z>8Z8}|NNiy9tskz6@PV>$yx?=HOBn#3Q`C#Tej@PhForR)nKx(+05M8(a{AESvESr
z6I(itby9w<fVE_9VX?ON;kXdk*!skL+wggjG$+u0?m!#>F8<oxM?#ndxk37`nfA(x
zCy>{7c|#CwM>u=4z5I;${L>N!>w{b5k^kM#e1|6oCzp7P5lM{XN!s=2ijSKyv;0{O
z&*?QMdd@+bysJO)vCyfBISYWNeGYZZmp$2xP<;w*X#RMRg>UAeNCFO-{-oM}=s(uo
zjH{swAL%;LT8pr|oG<AIV*dRb>L(@_JrnXTL0k%~{AmazZ2G36FY&PVc>HB3AhHOX
z&iWyiN|YEPqAC=d;*E~%{K-jyTR^%#TPQ|kD-?@yGS~0}1_>G43&1AFv|7vUOuD$&
zE52+C3ua1fF=hMAZ0b2~6iWtQ6Axvq-j`_U0oS>$TZRq7GXAb7y7+`px;ZPY$+WSY
z*1wm9t#)(rjIC^+e{4M{=BvY<XQmG5)nd%?{o`t%m0$D^0&i6f6RSrlQtQwz2}w)X
z$bhQJ{@l!8{6q6O7=W}nZ?vVAq18~^p2f>YcS9gxWF{Dty9x)CD*n;%|9a#{XuFf_
zu{rL)D`L{9c?;7YVzK6I8EMOPf5!#u)%~(E522IEF^u%K{q9w#>`^1axG$sH>3teF
zJu!!|*dDk6kG~ksL%TqLLT1W<Fd=wX<7qan*1B(TcpkpW1i<gxmh2pGK|)8;>IYON
zf355FAY~hVz6lMl)q)x|q!|eRt976#+U#k=iyp5-BvgLKWx`6aJXzvce`%?+r#%#f
zz=@t<R;<|>8A=*wB3_@ZvQr3oo%$%*N!8Om`*W=Jw%iMQQSXdtCptk&l&r$^rOi$^
z;zoIuY?8T605Fls_9i9hbRv|0zMy@(eDo4fet%s*!`b|7=J-*iokd!sA|7cp$CaE7
z<`ZkC62&h<dZKfeNds~T-p4q&nIp<t#2SK!^=xg^SPqOUtMuxuS9R4d-WMp}Tx1?-
zd-V6vkYh)?N{GKv3>vO|2|bk$kyLhC|K7s)8z1vc0`mT^sd}0hc=2h=*mQQFI0xsS
z&vX(cuGNR+o$cS}lY|3D!nm~wMUQQ_8!p7Y#)@&!Gdv%0ENf6ViRHL+GoAtrMGh{k
z3}Ku_$cnWHDW?iW47Om_K(D|wZwGdi-`0MTrMdpTN#e<6zT9z+1C_c}Ib7CGFnoqO
zw{s4V72W({f05{Oy}qWT1=Hg8E!yKsrG7DtYg5>U7Hk)e@h{;e#s=#Y>s>!<)z$9q
zpd5(r-axdPB!(j^kklR>#=H%s?=1}YAe{zjmi9Fn%tI)F42te0M!c+;O3z!r$TPY7
zLYw!e<rzcN@4Rd1Dm8avlQXy_3K@(XL2>|w?^#(4x~?Va#|l1{Pjqxr0Os%l`0TKu
z-HmsrqO+M5WedjscI-PWk<8NZ%3Ncgb|KQxgRt??vIjj6_HTH9u<&0(@*-9#ODeGi
zG5E>ckDl)V5^J_F#?tgx1q_^j7tKUbHA*put2`N`p3v}8H?ol@f{WUVSYI6vS!)0Z
z3>=AAuUbC+QA6GszWSm}X*rU{HYD?ra7Q-EGi6BBRWT``bCI&z2fdycRF2p=OH0vr
z?3y#-2MA>CbQ>2%YOKK)CePK9w5!m)V84n!>IHoYE-5OO3|m^X`+U32{!iF~pZ|xm
z`-yD|kudamKt-X)Yq)HDVC4L#3Bp1Nxuk+!c|S{7Nw=DaY)t`~I!Bla*VB;5Z!HIC
z|4)qgD_5jlH2+CtsIJFZiKX2HAWF_Y^;8{dWa<~3IM|P355!5tnKsIv1A+kr3kDlI
zT=`IB>72mR3%nz^dn2N9Vb%+}(>Da_1>+9H+vc8EJeg4Z1)s)ACu`ndvTdN#N1@6;
zp|rtL|B?Nvil^y3kmJY3MOQSqD+KsY9=HR1v$JOqau}UO?>BHtz@MY?qnj_j4q^j>
z^%H}_uxEc{=YFR-)pl^04Zg%x{tc&P*p&};zfGj_{Fmler6yP_gruH1P9A6z1b9?6
zwf!-${mp=or?QI_V1p37TE22qR{%`edKE8U@ix_RJuM@7)2|I{0jJPD5^yL~>bq(v
zMQ<R;`hLLi7X+BZE4P0WZ#o4RI_!nDmGdc5m^iRA_;z?#8gHp<X5yR2WRajs$&(p?
zy$M}*>!w&QLcJrsKf1F!5zY(63mMWZ^jIXJ5UxB^Xh7YxNobzO=cdP~(uj**i?Zqo
zKMQ28pWe~L&yS2DTso&C0a)0fgoEYpA&e;#3++lxn4i%4PgzfC3qEJ3hy{%nQlZck
z98la-Z{ddytjMUi^8a(+RWfZLeIQ0GMgsVCcGJ^(;4AU+jcA|&ADL3v>(+D3Pl#C9
z()qZec4S&N4D<cIa!@vWVNrts_t}PYqJKft?owz|ya9;-F!_J)FrXZw=D>))yi^q?
zpaejO1O!Cph<-Ofw*him*`tb@i1vB*d5wX603hC~o4hrBY9LrETFE?OAaq}4bN2`r
zv4)&zLcg%bqa5k6%n9Gx17zXBJZ9GlxO4)p-fAFLMpjul^oSE+7&Mrl<cH$G3kYjT
zKdxlH-Bb|pg2T8w5+o;a14IFY3Wl0TAdd)hO(S4j0C`kWvEIj;bdV)5o`m_EEBBl=
zK9Dn4AIYF~%R)Qkp09AE0$k7R!j-Vtv=yle-74+O@~(tuYB}N9$wp%_OF(sBbb9V&
z0;@|z$vT&`swb`o!tYNndy!a6a}(L;Sdv+r=FS6PaG>ZOul$XLUB!zH@}Ci0c7N9_
zE7EpDgx#&vHDZ)W=<vw5;-wj4GqFD{L`(b?IpkZVV4JDAiN0y4;1P=HFy=H!8g_2q
zq&qkL*TmEmG(r?Z_7K|SoQ#j4zyIy|=1R@569E>YM4=(MsU9u}=suze9b6Q{gEQQG
znOYXWs%0&;%RX@Nt8S}W#|Cv_=Ui6tzxfgzbpby{uju3NJ1GRZJg2!K)u6e+1baqi
zNHnE!J%;PrKa>G^+0M_XIb{}taHf&BNJ|`UF(e=;paY;VBJi;cle`%)0SGQ&62m+I
z)2$4=Pf9p!nStD=RtYgyL<x@+6Gg2b!7X4GHIJP1p-e&Lqe8PqrCQRhp$3bkmXKYE
z?O>M7N58_XtS8H6czGx#d$ch5E$E}eA1yE|*LgsES~r_?QM{=!@bJWDQ;@<77u(74
z$S~oFb{-Xz%Ql`;A%u%`yJ3(WT~I@>fM=SOYX&$fF<y)^YnN!OFWX6CCv?xk037h(
zilz#uL-KdDW;~@>{t^DL3z(cit*>@UVFD7U3pKd-dv+o@&jOOh`7Xo61)^9eEv|u+
zK3U@NLscdCLA@9Wt3YpMsvr?rKJXmXh<%ySqwjYucZ)o28fXJ*xs)4dKD%{`cf4!p
z?5qeOQb2APLeQOqcQN_E+iyISlTko4NV6BecHek-dn*!AG)SP6M2c2Y*OrAdPBl`P
zFrp6;jBmf7vxCRn3Fq=yxgeQ!F0>ger<QGz+6%-S4$&Djfy4-*Goa$EqPhXf802x;
zMwh+S>TfS4h#za6lUG1eWAg7Q(^WT7Hv?3f+{0g2>FQt8zLB>cp>y4MY&d|aPHt~P
z!Q{^^bV{j#w*}v$1&6hAjE$WtrqX^>l^I`ZVxE?YXw_0xyI=*z{LnA3f?i6>0-1nJ
zk*du)zMc4I0?$A!!RuL*E+v^Ap=Em9=03)t;Qy#Vo9~Ih6of7VDMo_Vq9N^y64LGX
z)L|PV*^uQ!bWtPOC?>%{I)Li$l1bs>WOIfW;XZ|2RPYtL_M&U!(+z6&br2{eH-Un)
zfeXd>_<n;=u(6m_?9!uV$Gm2$G-p-2IS0h$l`1-l`{r-rY0{@D3*|7Vv;Ive1dk60
z>?fAjzhf*p-jujBK(wgZ{I-w(bhNs_qj!4k8)QpdvP>bz4Qt9#*8mqS(i3IX6(sA9
zd~cdLDD!x%p<SJ_eY%5SR<8T-*nldi;~orBD_+)b>hdZvHZ#r`^cAbI%5G#W96E1!
zZfX!l{hP!4DVLsiM!dYQ%c)caBi1WloUOujWWzUG73kRER>Ys)q|<xt?haigxM?TM
z$*@3yvuT+ufl5M=FrXyqe5s5J4?$kEhD%tnq<H^$)P{H0<v71IndJ_bU~9BTNVJv*
zOItvDJErIC`)n(^c7$-L=LAu}{qxtAg&w44imu1`7tMv$zKuug_SxTXh>evyW0S}3
zMabaN?XuVmu6D{j<+VUN`kJ0NvQ5Mln!7pJPf(TAliZ3g1pt~o@eY}y?b3i$p1x&@
zmD=}ws@GCy#kl}#RSX#V%_d@kXpB2T_j|o7j<?&*rou))C7tIa^9(3ni#I;|vIgna
z42P@#NT*A|L0aL>2$$xU$Mo0q1_{oamM8ttcqUQC?;pjP?@It>i$MY<izc@q!AT@G
z`=V6vrlBlUml*RL?X(JnQ+sDEyu04%W};FsmV`M8r2s@(?oo+Q)DUe!pF6R5{+xNE
zt8Jc`<3AGn3!12mOWlo>Mxqfo?8M|PooBwpK4)rx76<qW>;>PN!_u#d%74~n<vcXS
zz^Q46raW1uuq7c<>u)z_)4Q)LpFUw<0$$sDS?mc{@3jwct8{JWjJ#2Vs|}T4KbB#P
zFF^9H@_X3{LMzIGchV#T+l}u%+gGZvWN|@p<R=*03(6>gy_=u!Z(ahnUOw3Hc183R
zFQwc7VQm(caa1gmcWw*DaiSWJ1L53Zw)pJ7A=d%bw>C5SS{-=Njg5|l{d3(QS3*I2
z?!X;!GYrxeF4QX$#dlkZo@&2I@;gDYtt%O5RUb^Ae=h=z@mVG<CB{MYgo*R-w-cwt
z8UpdQ;(i=$^~+b26}SG>@39VpD1T#<>LZE+>ILtRqYQtXLnsT*t-vtnqN6Sko{*X$
zub2hi@D+zrFQyK|#Cr|5mZbuR;4tpvt-z22^bSpnx3qK9ap7v89XhgOKT&!++}zdr
z7-Cd-{{?R2XQI;U&`ST<32A|uDp|gmPFrhkarW72ib_yPa9W20B8JaEfWq(x!E2}h
z0a9nNe@9)<pnG^pt#5=_ozSh77<C)4rViC7&1CZ<P?(N=QFj%g2w{)y4(ma-S0fxF
z8~n+8P##k{8A9MqL)$cgIgMU0I<8@DouUfLIO7?Qr`th#i37RFBxWNn1A*Nj+E{)h
zqurtXb2`D?^JpcfMKJd~rKgM~ezAH0j^QsQmzV??X791jU=UoReb=&8`W6wZ1P97e
z?j#+uq|5BSgEjC~(5PK0*T-3Jo_$S_sLp^62z<n6P{yPIJa-NW-a8o2FLGfaMcPII
z4TK@)*?R_+?maImJRM*HUU%MJ+z}gl!L)r(u8T<)35===<zCv+C$^C`NN5hgwA7Q&
z82<;cP<GK%@&4|#j{f7Jy_<YenG|hN7lDasZ0y}uYN}>i-m;SmYCY49L?g$>4eJep
zQr%p_S486CEuw<dbp*YmsMIaHVs8X`eaQk7Q)QLeA}&3-cyhL(3~l|;33j@YRX&p2
zGW4Igl5UjTX~UZQ<x^{MI@V)AhxJVSW0m@W8j`{+)+|5k8lI%xY?=}(tUeZKCs1wK
z&04pEk3ko0mLyg>i6q%EA*tie7U#s+pFiuGTn})QO(yJf(H~913w3vtOESZ9j6L$r
z7Lo!wbCV$8Q_i2kT%S|HLfKy_cOu2qXNfhj@ipaJNtB9a%;I-BWet=7tLw7}F%-4I
zZyQYycYW*m`>jY@%>%i-WZHR{iDge%l4!cF;r5Buh&e0XMZSxUYWHeQ{+=c~6ur#$
zVPgt1nC1G6{KiZ7T0o;Oy~>~k-qnu<y|c$G?%rRo7k?wxbPN}}ZIEKTYZxR&F-zrL
zd5=QY9nW1mEBGs*k0^hD-L_l6izbXqPBGV7kbMG8T$af)?Y*~qx83ewa`UMP8zoh!
zZ4q$Agd^tTVMokDTxQu9&)AQyTYz=Hk)EUI-9PH>vJViir`Gcwb(0F`*~+XH=;rf>
zP=E&U;0yEPP|xd=qCdB!$JCwlom{VQ$){DQL0BnG2a;?CyXPK&(gFB)b7o09k8we0
z{&kz6{^{tCD8jNsCK+wY36*CLd@Y%K3P$_3E?w+duId=00JVXt0gxRLRiRf=$RxSA
z%19OiyA)?_su5W2LW<|_qYg0Sr8}ZlwbgTm{bo}!ZR9X6V_=Co!n*QES>?QcTx)OI
zxt23Q+_`osgZ}_@?OgMHq*$9m@^6!(d!T*U<)a{a_ejX#@4WLF-niX=*2wW6{`>A|
z5x)TjJ%zmI``Jxh2(#>fw#>kL_d0738|C!!&{nvus1o&xcsI%BoTm`yA(e#lT&}fW
zGy%`~D?b6v0f+7paRbfy57=1@-L%CZDJ#gJBl(m|h5=OUBJ8w^36JAJ@)91(g5D*8
z{c;IjI>ca9t%I)3{b#s`Vz7GAB~!mmm<@sCgBQ#pNFab1z=GOF@FPG-;=Mu!_-W+g
z?q?C5e^0RTK)TNdrg}j_{C5o=PR6%8v-0i1fN7q&h<1Xo)SV!1(36Q8Db@4o$mn#H
z1+dO|K=&wacT}DoF?}>&R5yt9VU(Q7+zb$FBEi&(6pakb@8Nxrw`b;Xlg3rb$(}J@
zK3?QZ@^8u%>@*yx&^lUlmXu1#{<Cc^d2=xD#9vmTjNFg<9Twvv>Ebbx-s1wiA+6G^
za?B0(*!ehc@9G@n)Mc!Lgy+wPeLA)4q61sh0DBVw7C}01?QXNfR~KdSA;Q{;G6zx@
zfr+VWoY!WzPu(ukZRr8mfewBYvch=QPQ{NMztjG!`kJ{UgO;8W)+EJBhSlv+l0bKv
z3mY$v1G;CS9n_O=+v|+8IcuB8ZP3)+N;wvv@GRwy$ZxXsOBWmXZB(eDbyvXin=$D#
z0CRt&4eK`z<M+WyNq^XyPc~~*nVhkx&H3F$)0*^e2GC8uW*VeE*lhMr{R%>~Ew<_%
z9m$l+QokZ9%*A)k;M6X#n=5=>!dy+VT*<RE*vYKdfYb#qkDHCW@o|lE&F+IR>d$5=
zB@d7EuETHt#iPehd*{7~TTy(Z6=mB^fZMHRdE^}!6JfEwWEK3NGhJE+G3KHvUv)*J
z`UIarv!<H)N>3^mMIz`lWlZ}Hl&}zgHTyww_1WI{mK)W#8N(dI>cO}6imt=3*|h}1
zApXVY)kPdzn2jY}c1lG8SesL1M~~npv6k^^{m8ZbqzWIgA72g2l?z!iTTCec;CG=I
z-Wdgj1euPH4u$#i+c<|d0XHIlqXvyL>+OC-95fTfBvKk@yy_{f#!iQ{k2^%Wws=6u
z6@XAGEopc>i+QT{lOX(I>6^jy`NWsVCxvN(uqo3~A0GT#5IPdRj0O6SP>?oI9U;hc
zY!-$VeCQ|vwGQUSBqT8^eV`~30Q3NBYvKM35zg)+M>*90RVRCfp9np-Gw~%<a*q&)
zud<NvWv5jsz*(1zp*pC&YgM(KR)zkPs_*^f_a6`*ye}$Z`C<;!jpR4%SmRG6wvt3B
z8BClkIt;Sbj`k%=T-{u%3o&E)J+9`96_~B9e%3-V;fq-{YcLw5jzIWTfbYAYZ}#U+
z>$tji;yTOo_7nr;_H<2=5$0X>*|X#K$JgeEz;@3!otyB1u%y7*gIkidIPYT7e=*|)
zadA4J6>c)&MlZr3e}v<qb)ZpZ)JGw93DgoL5Gu>MV6G}@0zQX5enK_9p!DVD$ybmX
zsrU=Y#|F`E_IYEw`<NdPAH9};5zPL%EL>b6m(`8Vv$eT-z6;jfu@NUax`A7kw7%XP
zB;77bec0QEvfcKE*nSQrODpf12#iYe{qVp%Up?MqcBZ`yGr05i@XP9HX*ZoT?a%2P
zYucX=1L2IQ(SLb#kF>Xs}vTn!c$Y@VC>L2loCBUt}R6}FuKP=JYNyhGUsi4Wh(
zSVFQ_r-p?#p<ID&7tew5J2yWcSLG*cb%S>vi@N?^n2EFkVAYkkLvR{vhEC@Jm>*l-
z+7Z+v7<7VW&0>%7?<X+OQ%~l_lA}g$ioT%(bp5(8E)5FxL2ttbi@TP!&8M9ZxQ+K1
zaG*hqK3DK{w(LXz#5x=Nx{5o?W41-BF3RqT_9jC^0l9J_E8^vthuk8+Z_TxyD_Pr)
zd)&Gx962{JhxQ8!{%=)z>Y0Agro2Nj;XlaO*x2P1Bdss@vix)$-kE(ux1~;;cnP<V
ztZJ-P1piL+*n#y~lpB-E#T^$Y)5K}>poFN1F~AkWR<qOr>!Ax>*Umuqv_=HArVK8F
zEU2>wPbu?M(LNI#+(#fs*aKOfsMu?;HtcNGofg<f6V{g1rYCcRr_PieK(QZ(nZKmJ
zw4bG}ACaE2C^H<!F~vH3I1V!-C8RGO*q$dxHw1nuee5qP>N-+}ft}sXC?gNs(L15)
zAP(XCLMSH!_$oY@OZKakMF>GEfGZrzAQC4i=IW}B7f}V7Iz)((hU$LNRT>Ou9_*3&
zSes0WQTQ^CHG4I<%~81C#Nc6?I=cQV=QKGwB<BDSQv|-MFom%!Ojf=qy<!eAj=^mS
zKwp>40Ax-m1Vqo`VGhXYJ#5}IEo8jC3Y=cmWHrYD8z0I1CLq6zbLY{1;nx2mZZ$+I
z!m^F#0%F1g*0c98vv^wD6@p|T;+suv7NdFe_cRVnhTM~1Q6{ja1T`<oOYfFpfhVP(
zm2H&Q&CEEZx-QMcd8^bQ68|{TBK0u^NyjT)l3oc?vnq}P4cmoshcK^}C~hX;>r9U^
zHE<dP_*B=b<D<V-G6P<D-oSq~rr!fCqJJ?je)Tu`*Bkyi>|e@oHTWYcj5=TGSnj!Y
zV2-(5HLg%M(DQ}nUo^W=v`g~vR%O`|)0Z=3XKh?rL%hhzL&HJ+F|YN*73w?x;|h+_
zlSbQ4#{Wc5eX#F>cwtXmEcwo1BgGg5{xtwJJ|&izL}Je*HPR~5W?!$zruZ8Iw;v_v
zST{+;dUc=;CN`QDnEntv=2E8}#PZ})aghtLw)i0sz2nVwoF{D0Uj_02eY8(DfKpiK
zzh*a?s<+FJ(mR<mUKqJVXZ7ERKedciZ6nU<^r&5rwjS*375QrQ3*;@P9^rOPf7$>r
zw;#1U1as=yihtrA{X0{8`^OFtp!p-5-KDN0rp<jvd>J_c_!Z|ZdGtPVFo2?}^n1;}
zZBgzUwf%e)r6us~ZrRt1ih;?9N_vyRw-T@|N9=}}(SjEuBQ&4#6U|s{q>3zH+f;@Y
z_uTC(h5IgvcWqKEi<P#AdHabyv;pvFl8C701whKX#V}C<7>5)s*3e=rR}|K2v_~0I
ztU&8+8J6l-Mt61215r{$eNzOn>OeQksLlnrufKrtl5Z`n6qcNF{)v8cc9Hpq%HLT4
zU)w>sC%jU?ULxYy6n{ZlgZ<vRxipS!E#Bhpiko*7^HSGOqLVmLmN@0}Rsei#uG`Ju
z4GcgB2YFI4@d_cR`h_O02mgI;b!k`VoKV^$*Laf%FEwUWl1=1%@~sfP*fTqNOY?ja
zpNa4ntC!ZDPa;v;Q7<NOQRYG0xz&Kg$L6ZL6DxLD-3!mz|4S*f7Uo4*)>Q7-1Z)9%
zgD1VVB2FNQ5l3q%jgf$LmIUy^dfS)HPE>@D8*fw+u_12!qCi}ti^`RXrH-L}4H5v-
z-Zh#<LO_s!gC{mqfU1~~6s9X@8f*&xw%d^O`2hWz#Z2$S;Hd^P1)fu}cQzufHrAtr
zw;(?)PBxaXQeY$)%k`|gV5%?gSa%~40hLUD##oKngmU)r@YH?E_5<wuK#J~nm!Z#}
z7JixiPVeHSIZ35aUW<Jl-9hMo0*M3?s(<7|ive<@Azbiv?z-VnhPYvY^OdYBkQ~*Y
zDvUGJER(~hYzJZ5MWpP!prhp8;O!)fUZ&}^T{l!Nv+)v^Ayajwtt4%ahgexk?sz%A
z!faQlP!8cUeaXqIy#(ySpN;ArOan<pi`Q<G4WYjP-|750z7nsP7;qIi)5mT9HBukp
zk9|V}-N~`FU8i#!TI_H2EogI4yn10)EB+g#6$WV*4$)}jLImA!2c;KQH_XSvQ+^9t
z#mjszFfm_P<j|F=<oL56IJvORVrL%uk7-Cn)OK9ItE;d#`WhfHf_2k#xJX>Pp5{5&
zBf#`bf%Ywq8Rv934Gjs2fuKH|P{oS-k1147mAHewU^l0|Xx}dWDNDAW7(lCBnPE|S
z`WD*Xi{!jjL+ZN6yq3(EuhQy~>@8?9kt*aYDKRfRe`c=H(tSQGCLtlAmRCC29;Hk|
zCjU*+GMmj>%M_4J(0Lu!ud(-Hi|6SF8UQN2>xWZXTK%?)L)e{Ek&4;v(ziA8D$oq}
z%D{srpJH*ZV+VgD3lOGIT@q<@Kb&ZkCNHMlA{(fLp(to!7YgBCzJ+Xsh1HX;NuFV_
z>TaHt5F0D68a<L^0;v%?un<|IRyiw9TFV;NVouK*2?3-kl&mfMNsS3;4Hfn$C*_%+
zLZ8XRg%-rKv9clw@p0qKKhl@KH2{ZWSCz$&R*}Kqb#X0LK%ni|YxzLX$?QVrM;rb#
zkLS_W>RQ_wZF69Qc-<7ox|wDq^9KJN=sZEzrK9GoF^*Z**?7t*3^k^<1pa)vqO&Bl
zUc9h!)d4_N1oC$1a}yr=r~v=+<`A}>)X^gt-3*bHhUZ6_%1-Ls>FG~~6Y{L|jpu&8
z>pjiFp$IxVUUd2>hTKvNGQ@=2S%+Tx(?0;oyY7aXN4J74n5EOQ($R-$wMgc++RlO3
z>)7=#?(%^|UY+PvYyISrskQCg#iBu_=h?;Y?G9jb9l0U;>Q7s!SH4&q6B+WQxj<l1
ztMzPUm1y~TOuhX3Zg>fdXVl&oy$#F5=p*9e+^w|fwW4UT(ebb;1C9$8@97xo7z)`J
zuwUSyYt_==M^>x5Almr1JU}taQfau@I@lm`u!&gSEYd>uiHXjLyTR1XoLiNRM`G;X
zsw!YnK;kSixXNs3ZKZ{ZM>}K^eCYBmT6Ua-_!ke8-((1Jx<wkMVlMU9vsQfP=LGt=
zg=bKEc6U{|GGDFs%hB3<9mNyl=aDru_I3VE>4RPLTT#zJ8toHznlw1)Gu2{LpXo@s
zc2_4Fy}>6R7Yag+cYI!ln-voH-?#%iNMHa$wBPK8uc#}``>SAy7!LJRI9W%AVQ~@7
z@AOsz;f-60h*O}^EjjJqzvio_hG2<-?rGSQZ@DsGDHl<+7H$FWHLyAAK3Eb}(u?Eb
z)jH*lnl{;`pn{Vx_Cb6%jT=B$rKfwO$aQQg@8LCZucjEy;rWw`-n(Ed?vv-GDf56c
zN^fTtJ0z6*1pA79$Bi@DWk6oIEiLXse34DYv%*{WpZ7JqQwHQe?~Nr25m(Q!=~7rb
zo@obC<LekRprDDFHk!*lal}o^N?&e*)2UDmbWyQIwL3lYz1muKm6{z@MT6T$)g0og
z8WmoWES1(~^4abbKcZ#%>ph2DaR|W9b$aCG?1*`+qrKYSs|Wr@aXwIHm%p3%PhL~+
zAscYZ;$VGLtSm9lL;EIf$u{_PFZ)S38(oZ2Gk=~3$*Q1md2R@{auY|nB<oS%McaNO
z&jwn{jJ6yYEqhJ3fT?isQ4y~@;vQ1u+Cb|rli6?at>39D8FwX3<|0LBa!deR6TsaQ
zug#LECoPP#t|W>&CrCfOLa>jE?}uO83CBFQ{gtnMm<Y|&`X*ZP48(=uSuCqHyORE#
zxa+~Q{M!&29teo>F_6v8xs4^HflNZ_a0(C%7DrOJ&G>&*eN%W?Vb^SIr*RrP4I4FP
zW81dPX2-Uj#*J;;wr$(V$@l&LdCqfg_Vs(U*Q}XYGqr<k<-ZPEDko>B9sk0RLF>j%
zPUhe1<i3gbr(`6W`eMf?D_H>h7O5vesXOGbxS~7PXz_qdO<6dC0Y%#jf2X_q1FQ$k
zCTOc}28w$KNeh<2a`11ex#n414DV>(6CRAf5fRdxz~ZP{fup3Tti)^@X+tc;Kz64<
zk7XZU3+OuRPiM~bSb=V-EV^*yJ05bO{>PFi1Z84n)MY*KkpQW3a%W&XDaY&};0d11
z<x$h<&xaCyqGg98UX*IKtg%RR^-=Q3tX&TV`KUAko={b7$H#+xsc$!raX|L3UAf{j
zlF&HWt$OvHdaxJlfP+TC7bl-XxGu-Q$srF@_$XFk+WPQtC0qHm5_Q`km6k7W1+MtQ
z$;EjJ(F<-QTkW?$&hbF6db5|#J`G-t=jL%A+3u}|I!&Wb#JbTBlzyoGr?sY01$Fvs
zr5jo0zTEW&a;xooiTk#-E%WXC{^Ew|=;8JQi{crpE0+30kGw0tOVXq<%yjM1m8TGq
z6?=5X6q3U>w=)fTQ`%J}vpmvOv&4M~I(eh#*s9AktM&qvIS~M<gOcW`0j&Q8dd{KH
zIUB+$4bydWF@K*T@iiZwGZNI2Gwn18w-nX%AcaEdHbib*sIa87GnZ0G#LO?o@<&ze
z3UOvODH5E0%-X3C;U8wr>e)eY`L?cnmP~l40cYah$?tpaj!?MX;erKa0yN|*CupsL
z1<oXP*1=Z4Im>}e9hkYaPd!g}(__5u0)E<!yTCw}0<*E{!A@RvK=dUV#`8u&??2og
z4ZaNv2uqR-HJJ;jX3gc&e<5mWo^~k9GwU!5@xJ_L@xZ+g`H*|<?7Jum&!A26Plb$$
zqFfp-8DQf6-myukd6QC_ww21Y#tKDuE}>Z>9orsg@rCgwp}!k9B6rm*<&<}sifLKj
zKv@^lHM-+-=oDQkaM_=nsfyIq8|M~jztLUo#amICYDi61L);_AeHc1NGlqiic*+(Q
z@nzG_TU%V%mYovI9#6nh6j7xLcBoFT(oQOISZ}OqUMLxHxD~5|`?@3}VxfB{zl714
zvAF~+&iQ2FdCL^5D!f3WpC1Xve(ePS&`GfhwaWwzp6iz@s>Peuu6&ieM;9wH5pm2P
zS45&6zFy;$51MbBpe+?x^SgMg#S_Ycv>n1DRw_<l$MPzd?c9+P&C1f$Q%P2@8Ktz$
z_`z{1zEYX}E}=IZio<+~*Vw5ivh#8#W)(nM;Z1(p)1OmiW!AUmzGt|n(2U+~t7eG9
z&ls-g413Ot3&N+b2(_;KgkB(7vbpDTrqeo9=`SI9rBVhK+?KwC2?IWxs>GGI=^$S0
zM?0?%yPvMfJ{4nY>I?TJJ<e;kW?#ed=O1^s&C9rMta3&jPwH$*R(k52#)<?wGZH}K
z^Db8$oW)G+I}WQyqwAwkDE7C1JVeM$ng-gw9M>OtJ1O5JHQ9q^K$Gli(4yioe2v>C
z<Do{ef9$eAMum4i5&S1(B#8;y&XgiGmP?45t&Q&aPOxV51!SXMXd7nZobFgr*P+Hm
zj~|}RRfqO;ns%mRJ*FL_XvM4nyA<gEEW+iCo$dYjqgRht8Gyw#%`84Za^*y}aV>57
zkI9Dt>`8xPUQ8$wJd<WQPPsu1EyD5DKOE7M(@H@u2REP9jm(<DN260o*j3Mb-Y1i4
z>E1Yq;9fANWi967dXt>1DC&^H_6KPd{Vva1thH0{>OnU=7*c3$>{l7cJRx9cK}W^0
zL&ae)x?K2^0n_Y&h$TMe-9~ZRWH7CF3%R0U&J8LeOBzLoxa31oYyAUp+T-+2Fv(s2
zpZ)i4u4kWTI?tzK+y$rnv5-ghx@DZoM#z)1{+*y}80R46jgFE?Wufnz)Up=qa*2d@
ziu7m=7=I(Np<;vwAdEOsK|ugrh^x)-sCHUf?U-?L?;k&}8@PGe$DB$X{8AK)=gl{-
z&t6L&KZubb(C0uk=A<h(KeBaPshl}GSDUNQUfP?S01*c&c$LWZ1l8PPjdUGD5=%o*
zD&K85Urtp52|IllfuE?{N;muJ1v47{j|wF^I%3{e9)yR48xX7iYJ`Aoqf`jhYZDg3
zSthL&2iGA`yi{6THsFwk&G`;CbHUs|*&weS>+zLnJjM0IkXDs3Uh@#Ed#CG_qfYn0
zq&Wc~k`^$WS1>20E`a<EO~tpsI+OC6s2r98?;iz5v708^qQ4dWH?@REeR6L;)}UAx
z)d?2ZF;}r9-)!++>H|m@=KAO=F8O<Cw6l`;{iDx9PMZjiYx}imqq63wqhV?-=}Qx5
zL=r#8br|Z7J2AU=tpXh2BW`T-`H^%l%2U${ZdHDjyJU#>)NcOMDT%EVidS1xQ=L;;
z%7vt=9Bfg}pZzH;QixKp+d`%K$CA_q4r^Ykr}a@Q$<p*D5CmwmZqA9gI{71mBprXy
z04q!HsH=r#?DtsPLQ|3bfv&+Y89Z{ZZ5!N6ysYuDFYWpg7SFa;mrJM|$z+b;=WJ<^
zI(;m!PsECAIu4)pf;E+d;AYxDCgP-#K&m=@>cmF6YGgfN`Xo?U^-fo_Of9)@a-K3S
zb2v5}TdjO{$qDqCNakx3ZcD=1TC|Cx1g~@E{OGb>HA<$Efr;)EV6lHgS<sv5olsDB
zs;RMLc>Fp#3bnHp`DI#$9&rpI9yPfN94jJ@x)`_3=^SN<b@O+pcWFe3YagV>g$K
zx09?ayG#$sVdQ@DbWBgJiY<DVxDB^Z;FjmL1v$IDg#v*VK`RQ88p`zWdTIC;$kfhA
z3vJ1}jsC*I#Kkm;#%NlbnLP+w;qV_1R{M-LF6x}+62-qMUDlDwHRCiRanuwEMj~_8
z$js*>bq+ro(qo19b;6~EDaXZe+r4!O*6v%zT^@g)Ja!<->2{p|>3*F+Ho|$An8lg1
zpPteUUjazpBVJi+QOGSZFPfGRK2X$aUvqH?w4zcezF^>95ys4~#<yV^w-vN{sNf|6
zxlopcU*}vfy&|*V*xc{03uRrili)mm(p3lq+ux0ri#ZkI@~G^f3Wyx(OITxm9j~?X
z16?_(Vrx#kQmu3NNu2-O8R7-yS26f;D_c^@^a0NN%_2k1@BtQg#+_}0!eCn#QeLT*
zIm!9o1z$eY9&Mzab^Z%{GQEs~V3&j%gDc=W`-16>d2=;x5~SMf?JktcNO^Aof>2Dh
z2(()2swMQK#D3h`-m!`Z4cZ<ONZZ-+T;0vj{e9zygEv;RHNL@3CIDT(JCt+hsdp>K
z4)D`>)OIrV^iJYy=am(m+*YK2He|QIw^za6X-mN%&eDqvAQE(-L+;TvauBXcXAZnI
z)aBBTicV5L*<LQ(@VwNADz+kDhVy5&aIFcX%WN0o-00g2Nz`c8kojkUz(GQYKOLbb
zIP?W$G9CNv2#V6j>o_#Bhw#_-(g221z(AO58?iMguRin$rMvD5WuWdP@x=Xk3kPol
zGCj$6UvsIXv_PV^d;V`SqD$8!v80J4j(UrJu8%qzez~)xC{&?MLCuztv&@7fowewa
zOAhgf@2%&|2Ylgvli&AuR<@KZ3@iHB!sxV$SS_j&dd<nOdaM5Kxk6)7v_F2?fFXIB
zZ{P2J^Z%f*;{0om@V7%`B1CF*cmR>wuO30<CncF<B+Ko$bMGd2ebfy|66LKhHig|5
zdc_bvKkKZBw@oI8t};j29^9Ay+n=dv_n80SiaH<rJlccIvBzBRf4`~mDcwW9Tjq=v
zp5eFAoG|IcAKg(7vD$A2ACnORa$bGYB!MhB@J2p*)GoiKi?wJLB|_H-DGjCF07lWH
zFzk;_>T$@4mp|bj52)A{HQ^7};y1M|YMFI7(Xz%^_L2!567TfK?)>6?F1)cascAOl
zqI}<5GqA6J_LCg!z*Op_T9{SrMw^E}=UK?*>C_%6_Ju^OtDr-Wyq@C&@WP;mqJr=u
z#D>m9*aD0M(Rv6gRLc|qT}@+%eYYI^-*(qh64;jpHw6)w&dt4{B^oc;ka%OX81vT5
z6Qpw&A{=i*nVO}KYO+f`6Phx}=?ZmZf;)JFeTfT%=JIocEdv6$M%&qwULnwT<C6D&
zfv7ZM5FF?jCMQ1?>3eBlW93zERzxvu0AG7r{FbQQK&<k_95!Jf0Ys)K{O5nh?+1yU
zr=_XH_k4ASR!_TP?8eUc^>J!+PIz;1$Prkqk<eE3C0N{zp?~wIP9D0Us^!zMP~h%9
zHu((977frd>zP09_bK8ugqovUW?q7Rt*qLfDg!+m7)8aJ9qWJtXUjVBna(9dMXQXh
zn2#m0Rom2}OOby|J8_BP)CK3vr9<si=tF&B1QRk;mKzj-xF5t?7bEmgT>Ky8K<E`*
z$HKN08a7u)Dy-Jj?@P#nT@&*GFm}oC0zdIt`bT8|Ngu2*Y7iYCc4{}#zFO7%pL?lV
z4A<ZsWHI?+9|!=g$sc9B(jd+b#jLrO92&l(A_FN0uu@Cp#>D73G`T_ra5qk0OBc6q
zLm|kk*TRsbdye1Jk@ma;!&5k}*6y~TKZOgoA!A3t;gs@qw3or|)5AM?g<~tqP43rN
z;f+_{Z+Ves5<9mD0y{W*F#oB~AzJx*i1~@@dwTi3C}skwC@;JCa-QD~+4&`j>WCU9
zD$Bg6^t_U+%fy_sZ_G;4M0U>Ww(<U0-<4sHge0(kD2b!bp2vMVn~M-vXlnG@6=L|Q
zXhu;tYp25$J{k`1m5ld>y~!icfaYI;KiyPZ=215fO^>G5<sXBqEo_HUVDpy%b7ME%
zjX85Ku-plF4wTU+U3McmC-SqyIZZi4dskI5f;*}ldE38C{Yvf#8>6CUQD`_ATRq)C
zBMfIHFGf8=*RVm3{1$mP;NG`@ZAl)%Vl1=Tw2Dl%f@jSb@A|SYCnVi%lI{nn{Xu=7
zn<InDc*u{sHcX0g=r_2IHnC0F<=^nZNlxRm#|H&qy^*+==t)6gyWOhk!1GMV^T#Rv
z&KXQ9wF+-qM{#bWU9~$N%c;SKg$z*n#&AGF`Ek5$z6Y}ya1hPcQp|6!+)SDi2a$q!
z`!eC{_WzRDg5sv@Vb+9M!{=$kL2fnqMGqais`a<3B(Ntf`sC7c?UXGJ@wV+`02y7H
z@?sr8ZM1vrnC(7To)dPQ_gEMl@>;;Et>%%8OR;V^l+NZz8#t{kSuP|v03Vt=r11;}
zr+}pSt+7a2mP0{`U^fwFax8$`+C(5%HR|3l3cryfLxdPDfZId2hevp212-U|y&>IL
z2&;aieo)TMYT)0Y6+GB7$C)@$VyRZ?Ub_hdIyZa3sSysho2U`O=_E-rgt2oyIKWO2
z#z;_<(U_HKoD8L%;F6H479Yx`=DevbD&-$n4$X*XH8(SwdHfpxHjM}lKrFvMC5a!}
z?|bJdC(+J*8zHK#QnXsKrA<63&=?jl<<ilt8{#i9{O_uX7=ZlG_{foy;HfXm38<MQ
z-l4XAPFv%xp=Nf)=b_~rKCdl;a3v1_9954iE*JjP&t$lI@sEkStgIG}Q!-8@sG8so
zpQ5V<%6oHG1G$~ISX{aFOyW$AMv^wQv<Z#cEE46lP<X2C278t_Gvv7K(eQsYmjMdZ
z;A5Chs$h4eLn!EkKF~yZnu5YMz~#FoR-#7zo~y;iVD$E!&SIjB3r7ZBa+S7%P$v!A
z!*EB0Tz3*DkC90O|BDj}_@`L;?sLHwVj9ogZ<lp`Uo&q6h)uY6f323Y0N5v?wI~?<
zvPF~U3PpNR$)wkJo$RsQvCO~2f_XC4B*!uQgMpzg`)s3*1s{j6n$29oK;t7gr;3tb
zCfkQ<Wgw@+<LwpH@aG=}=c%g0w}{Og-COO}*jHu?WO>i7jaOkA*rgNSWg48<SCI?k
z3ut-{Y=7yD&aVo2bsGX$U?0YKc1HIG>66S6zH;H?c8t02sP;POYp;}*82e=yczxOR
zL(YP2wb!NcRr{6ZzV-Y9IGl%8jbED9%`WTA`~Q~9-!v^Y-&k4yyIdliLDaS`mxWZw
zNuD6=QI$UGWc@B!D=FM-m1Ns~I|*x2p*F`*ix73sDBjB+af^_;FS4Ne$)YN)=&m6B
z)^T4_g8Q+yeb2_2G}8+9N$iU6bSq^j6<nh13SgH&(P(+l;9~{ZnK;BY-|>oA`okvw
zU`6OyHr#P<vRF2^=ph`l84cjf)M60E&4&8If*scHXnl*^QSK&@|Ml$r-Hx<+t>?a|
z{(bo%bAntP9HG{#yrX4~Y}O-)DxTSjS4@w*idgLL@m{zls=Jzr<eay@?!4mgGr0{I
zgoKH}1B1s?XcKq(;7!5^rT;_sfIO#Nd$;w}EE5%*hBnr_AEh0b{BL>oIAj7Xhrr_A
zE~l!1gJ#??Bt3{blgK!TVfG1;+A5o}jF5H4=@bW@z@De#Wj1xRh18g(r!4*M7m-fs
z@@VbNKaJT~=Gs=0E=aUvu<y<GJDb@+w{s(8>#}du;XQA>i|)&RY*}7Y@QUB_a5h+>
z8?o9*j#2+evHxT=^sU1w8Rvu>6MXeg;#6D+YPfI35?fn3(`3mp#G0Bq!$b%&q_Iuu
zGRu>0BI4*jA*LT{ABk<t7+cFW)`&Ax4aw)4`m+U}Q0t82^&bUk?cbWhxH@pK|JNAG
zs47nNZ>s4;{T*2j<agv>8UG_79osSIRWJ`NRMGH#n-WcX7aU1n*c7S^3;m+((`ag!
zSAzyT1xw6WS!U|9m(mIn8!2kNVws62V`)|^_P*`L9b!@Bv05_$eX>V<VgK6SRk<Em
z=*aP9DVI1JB>#|prk&FO8qzeZc&Fu-r{%NZ8ivogT;XH4Px8+!%Q<#V(nayy^E2Lm
znC=EsIVsHrdOzCn$rTR5J*%GUxh|4T@9Y^n;|Bi}CG<F}Oic57*U$0MO@9lnx|0%1
z*;7Up`>8*Zo#>$pfjUx5Y01^L^$)@gREK`MAFe95eSa`K8K*iDm}4S4{?3MkSt7_{
zq1@~>Jeyl>Uj7Zyie0`|ypA+DVkn}P{a+8ym5o<CTxN~N=gC^o-1$S$(eD?Myn}b`
zqV<>{2Xt1>89gFovvs;aRI+(}@_E?Q>|csHQYV+9zX)h-L=Fn@K@l(-UuUzsd=oVK
zt&5e4vtO?h4cP?F0io?9xG5znBlhAE+SSoMh0U&xJx*7#Dk-T>@dH#ers1rU`=4IX
zPX<+7*#FvycP1s_|7SRctR?EDSYdC|f4`&e)?VH`f!`VjwMEjV5S8Thu#K9vjBK_Y
znspF;_zc=FT6<sMmEWCsc!D4!v~?pn+HSb&zO34^aRhE3Ate{9beEeu3)~n8O*<33
zrO#-Vzch=hY-(^Fw5vFuWQaC0@!Lewhq$W$JhrWrLER7Av?wfsD`I!P&$i<-O%kU<
zuMhz9Wd4NojWc$`M@ejwhC`TNh!gJ*kaiWF4Be5$iy2psAOGtS3Sq02BfQ3TP>3wZ
zIPIhj5GjNlqevF`g2PQn^=j(L@m`q}nX-1BLNg%WwsqYF(~iLogf^TnMW`&zXtIp=
z(tl1KnkdfRn8OP+;&6q3gVBb;t@0-hE?nDdX<TBDy&HmT8E0OfNkvWkZt~m5k(Kqw
z4&{r{xQ?`wA6?})yakUH2GabR4QVBo7ajOxK&hdAaW{G1?(bMP4<z#O1;NZnT5ug0
z!OZ&Pjn6j(6IgPcZDHL3J2m5b<sk^WxH!?;D>E4#p9?Ug7%9gzn28s~Vy))C+iT8b
z?Ux~bM!74<TwoJPl0=y@H_<QIBAg-qYY#J7V{`MqJ};hfU}B~pxG4THeODLmR^aX&
zfR!uXC9@x5ELeiaCE-KsQ}2ZVrX1uWY-cph(}fGQ-F<&7cyLN^V<Ec@I-zK>l!F+3
zOh;|24+oJdh}&)^9wxvE8vfCsP^?pun6Zlv_9(m{OSSDET~l<7Ie!zA+dl3rB;p=N
z*6U9j0zLB!tzDl}-2#SSOI!0V9+h%F;81ek+-l|1XIf8r_$87&HFtC{>|mr8%a*Vq
zW$uxxFghY8%<Lb#x0Ji1OH&)s1(2K#g(Eet<_kACw8Kgo#vFPo{m6z+CPo3D%<vH9
zBcVd_+iE%9JeO!>wvI7k7jp1weFSelKJk#=lq!2<wu95}A0XzA;F>`(<mj0KrdnIV
z?F1=?tLOQ<hVez!F-4;wQu!+&Ix{DI`1taQ!f5Jlk2cs7tisEVA&zlVX(e4WOlFO&
zJ$^emRPm~J1ZUr8NdEJl<7P;nl=;Q!cUnrTbd8A#3%l<OBZh&*f)(POKafcDqMwZ?
z%-7TO5)8gt9@_81QL8p9GEQj$)TiCsXh3yX;fqv79MAszxF)S*kP;sP_#H*Q5P5Tn
zy5zM^J{gQ)*T=2dPRbEh9bE!-A>A?AgSA{e|C60OA6-qUDZoFt-<N0VeCOYAI`_>|
zf9U{|GUKVs+rY(K_?Qgevzd)pigm2w3QOTi`X(<bE9=8e-lW9X{eBk6%ZC#9*ChlG
zJAgr*T~&~3cAo#%*zaM;P{(()VCHG%w7(!Ciz3}d>ig5>mjh)F+X%sF_+YAr(iq9A
zq<rf5ovJgwc?ddiLBi>05wQ4IinRfflOtGT_6c$lA@lF}bh)+?ZZcg~17-^E04-)`
z2R0`~g}W4SH;d<Ce;yn_XIo7QW*lO-fWme!Y@C@6b`>_B47(X?hb_=p<|;R$UvV-a
z0YET!{m!X)S3(@EhD~AB(}W5(z3GV^nr3uQzOkZ|hAA9|5M~I8t@@$|9QS3flyAV6
zSAMJrZl_x^;NR_lHHvsl_15{3-pa>i5M3z}Tn0Ao2D&p(Zj=KSQ=kuQJbUm*n@7WP
zx+c$9khA^*8A8o~w_V}cN;Z{M1jy#;LHxLb6A4FH;8k~(6I^_sy~ztL7xAn@73b$u
z5HG}eS_vIBQktyYWVejkB1M~-&d?zixs6p-ITymDg9LivqrI5})Bp9o92*lJN9g~r
z?FIPqy_{V14Krk+j)7IMqkJVailLI`$F8-AVmOm0G%C$142rqW0VPxNp03}8h?LEK
z(R!CJIc7He(x_Rxe0=%0PEH<{b%kHf4BwK_2DYWb|9_!f<D);Zb#1WC9v~*&Jl&DV
z`EUAkD-wvs>(%+3T9s;GQD&Kf0)&cXEzM6m4ae0_*nr-yrK=^8`s`KRJ$m;=CdwWo
zk@H+HbxFw!@({F*Oko#yV6|YCte`g^A|ooA-&KLIlPzIU{8oMu7{k~d^jSelsg1tQ
z8ul1YY#({QR<ESZBebeNQ%`jOUd7XkZ)H#`{Eq))w)2~p{>f#Jygv)nt0*U>i;^h+
z0^6v&OKu}&6_350UjeduOr^e+VRCLUeb$_02x&dJ>UC4Hrl0G#F1}a%Nqo^B3sxa?
zdVC~;LO%xKv%~E=<gMP^Q3cuu67qjbmlSD$*R}?G2O##67zo_{KfpLQ%%!e7Hjjbx
zCL&(y^CR;J*dN)IBL7qxFf=ORQx;0iX|_lkiMQB!ELT;YzdA%aq}^)TJA&iSCz-vH
zp>!AV%p23aK0=1RfJjErtmW#uo@cXG4e(okSL&K8seuv0w?e**ST%298l>sM`+&Yy
zM1^K5e^Ef1#OxML+$h-XF#Y~tZ*X1St#iSb7ZAvqyFg;>9r9)1y7dzDp)mgI;3)i4
zPu{qVF?G*F!2TxIuM=#;W(2AW9?FZm(1z9z`(|IwMOBm+`R+~e+D02zo$VCDva9Y$
ztpL_*a4I84m7?x@3JfCbB>jX1R)^Q%D-u;a^d1+sN@?^w;=J6T>qxu?yV5@N`?veE
znsh+ae}7XAmx-3S+m?s>o7h}14~rsT&`Z9zY!`gB{OsP|hbk1sht{_NyV>N)wapy;
z?m;CZ7g7T*bEVChAspJ6L$yC|31*o7a6&4wBzc$4UI5cnJ=6QReqjTmt97MXG3eaT
zH%TJ2x*9ASe!Q`{u+J-wZ{?KV%}=o_3IXhPqLPze!c&-4X-!ChY$+Wy3%|mS*dxG;
z=(!X6F`|g8pFMWW*bDZiKln_mEH%*VkMtK<7df=k6|b&w9+xQ5kD|#g$9;I(SL?==
z@xrXf6FpUtIc;uL&Lzjr9TPowUKYqOg|YTd+Ekvl100Oi(^@^n5Hog5wu_g+4uE1=
z?c-+DoY~!rw8Db`Oa&<r%ayPxvXpj<R%ubxdH3zKBO@eQ$cq8t((~wnkzHp!TjwZC
zrnJ8BOM@ov=0T<3I^s6aPMk`o8#)edwAD-Bdh1w?mB5-sTtRE>5S#P^JYCP=@MVMV
zI0;DH=~K8%A{}mT)`SR_5v{)p+hJqSlK1zp0sWRR*-A*JJkfmkPfvbPgEfe1(|J3G
z-Zrl6G_fytYHDnpWE>w`mJG#{`nceV!AA&CAKz~OH0q#NoO1oGhd<1;$00Z3$2yGU
zDUq`zCNtwNX+-C?vrD5mVWkxt#97=AU431Q_`fHnMR9AK^nPKhD5-qsz^F}*zT&rM
zS9=30$N54U*--%}45?Uwv*Z%>#9Il&^-p_>>?FKmkMiAz5kC7DWC@e-4LD|Ak-le=
zoKTIkiJRIrj>w5NuV5xpDcgbs7}{Zid*9*uv?vygpQng4`EcOFJ1lhlG!{=1yu|%R
zWvhK!-ZD_bRfL8$H!95<fF+$~vyV;J&&iMZ@pN`pu*X$OvqEkI@)fS{GzkJ_@_xbG
z9H2kWFk?TOpOj1Z8KCcH`VRHVPmBc)M3xBcztep<GA)XoF5kOWZB)cUQCAMk^|ZX*
zK|lXdoI&qjs9l@fNmdOcY7j?nNu%lqTpl4iM^D@j|5*q%TZ#)H09v+(q3;OA_TK{t
zx3K0l*gz)rFv)!w=%2V)5UXB8UJ9b_H6BJso#N1OA8qeBGdEPx`s?C<^2%2>uD?u@
zo-b;A$Rj^T-iiEnR=;RbBn{%Z=?7mP4s`-ZZ(FVJnaQRNQ2wRrs<^~-K<Uhz&_Z)l
z{0Sz#@o|$bwEYBz0A*24a2DcRfjU%j`CaC;oX(Y3`l@S`#uqnEV%sk^;+>C>?w|h#
zA!4{YwPk@qfp^wW!C2I<VWhx3$>#9RjeEr@5wdh{BxD{yu19+OPTE;o+vG9v-n225
zy{B>h;6~?k(%IZGP|TbEFRH$Iz_@?jOF+2J8ro)2G2qMkmA!mbjZyw=2B#G<IuY?a
z@!k1ZRMhX_A^^G3o)&Z4A&HO6L_NWo6j2RS`imS0YFf6kF4=q?IU(+}yniLOK)(9L
z;_=?GhV9}lI-@sbSx0`r7+)|DdCs8ueEjzAW@C%n8s+#O{y)if-8Hl_q9c&dp8j~?
zyLS6NDm)$_!~PcDgndSQrF{^hqhB@HA%hF$3B$AmUc&ITtIrKXdl2oRImSvN1O>uK
z_392aNe4&+!CZvr={6Dbj250SHYg*P-_dCyV)Sm*Oz3gRqD4lU{w>Iu|M_XRkPnJ%
zxi|7h11Cnr!)vR()-2bpLv1iyIalpVni-DNe+>uTkSMV~^qAF+=Y~<KH?2FLP5B>=
zWB-`kPIk$sLAcw+08kahQoZoUfz&;%<n2?XGc9bw4e<vNF>b5Bw!C88;15V%kd<?b
zf84sYcMwG@8e+$Iq)i{7$lK6pN$|$5MYLd_gfH^M8LSbxk7CnIE0SXT6Xs*y07*O&
zMeqjDh`}-GyH?iAf5evYv$VwqIaPs%Qn0L+C#@CLUL^l?2I~PC4g@(82E)xKlEfBk
zbV`$|oHVurf#KG^fa*W?nnn=a1Krq_+Vr-$QR!}wRu9CnPM~p@I}AS!P?axd$Z!Sm
z!suOn?M9h5PGG>&f+N{*?l6ef(2`|7j*|z-1vDw~Dib9Ri<G0-2w0R#SQB3V(Y`M%
zQ2b-qeCQDC5&2`4Uveroof9DyIyl#wVTqYAZ-~51n~PWWna4(#ny1lpo7h~iI{bZs
zH2aO@@$>6Su_vP;uKZQiBcA7w|GO>QmMQbT5-bX=GK3c>Wi9Qs^zhUYCunHw9!D0?
ze7k8|nFN6bbImBYZjhN4Z~b+q*uD|{p>mnTQ+_v(JStm>1ntHk-+J+9>YPX8#qDU}
zH0<wJ1(CGsdGUD~9G5MvLiK^C9`A69O(4fSXJa|18QNAux=qoF0^h!H(?AH)6E8MK
zB@=;6(Jt$6*a{{Cj+6|T5R~&|*=kgPgZc&z?R?km-<}z_<7~SbA2weZi@6`0Zh4*V
zOGwJVQ-7pX3j7Xt&HK&P&_VBaD8umv>cOB#!f8-OUb}DALQHXI;cMlHk?Lm>Ve&UN
z;X1M@a5SX95>VM>M|$K|(<8r?uU60njJ|8+QSQF>_C6M~6(TW2L$}t?C}y+*IM<0g
zF%Q383KlPXTBuTP?0z71!8h>^&hFx~?_2zAMvUkaL_WWA@Qfig+p+QprViy88aIvZ
zMn!Ny?~z~XzbRAV9&Vhzw4my@D#FTfX;y_oVP`?<Y$r(FmK#F}u5Xt|_I_>n%8=`G
zg&uP|;?|eSnj&iip_rrz=Kso>w%M--Jrda|WIg|WUP9WCHr?nZBFM&U!tfmbaF43A
z1)Y1(x!3+fhh<o>iU^-npLT?PfmGy=@80_Pnij+S>(@W2-bv>sS{AP5()hg#)>hDl
zE5WOop%RAyeKvQ9PF)3R^w@Nqe&Qw}rxke!<{z6%DdGk+vmt&Z1oW~kFhq`x^tR(a
z@LDb=Iw_~N0hipkX3$s9=R*9?i+PsM{3Au#*r^hoLB`LGkZ^mKVo3N^KzrG@mGA;N
z5nV7-V4AO{=WG-)9wEEX@cReY>(2F(SD}#7_P?H98ZUJTaKt}*>cPy!rXB9pQ__b}
zyd+4f>My;dulLb#H21=F0InhSNYtNPoE7lQt|P3@bnIW-M;ydskV3Plk<g6C8D|c@
z6Blo$e9`!q`O0{vswFfd^hCxJerT7h7H`)%Eltg4SxF(Bf(Szc1U+Q+RlgtuFs;Yh
zYN4yfD`+0*Kk#|3qt&n+_aQ4%Q6xa*3**!&idu89Oy?fHq@_+mfakb2!zg@$x!HqX
zA)DMo((ysYQ*#yWaMNkm7UF7_d+SET+G~pLj6KmjrSGp3%oKTy{ang)>LKADj<Qq?
zl5lYQ8TsEKrs>$Fq$|WFQ~o&3<Se57{&%F-{BsZ0c>W-O4Q*+@aOSKm8$1;l5kXYx
z8Z<d7%B37-q>q~d0J5WRmYxAKR%n@vMFGL%MsynFAd26ODZai(S98^uk#gJnjO)kQ
zdfP(&&lr|T3oUQ6_eM=PagAZ_D33+ir)c(}Ng0)8)kdVQ`6}*B>&aKFVHFgT&aHth
zhdWx7B|Pqv)!F|T@t}#DTWYWn9cq^A2lb<952Vp=C3zf2pcta*GP+NeovU?P*_&6A
zlqB_c+PMe%fa=^kDxPBj-{@fX;<%29y_+xcPHBlny3>4ml7eA&#Rnv;sx7E_>4Ylf
z-V#OUBG-EXcjFO$H)Y%%m+<qN`NA+|{i|<mVS%^YYK9;o?^+$_y95|Gog4#9A7rsM
z(j?oY(WlKMATc!DES~}{God4tFTO3w0<C@Jjn3(~Fl(JvIPEpZP0;ep;Y!BmmhVsa
z!*BLtJbNzkrtMN6x34aU&7w+5v%TN@*PTeM3fb>q0sNcar>A<VYl~Y9a_D2@YMh72
zKVsN>pMHb9-=sxUioF;&B7U19=D<X(FAXVN>1bNbg$uOS?Yo&tJ0F{gXR5&fxF2yR
zS_?PUn%G}&#?2Z|OEi!v&`bvB&thdIv;*t*MfYnZc-I?V#?zA|Qsi%lO}4pPEy53r
zmmTd?<|p`wQjaHeE}YUWWO?C(9w=8IOe@zDmoBL+!r)%NBM)g7Q~9h5^Mk0#kl8xy
z?Q5)7rgrsl6WHAZh#IBt^?g<GTv6SxM2iG8gV!u(E!|!2SxJs%T$a4*aISIKfi72w
zUr+{Wpw!NxI}YhDn*OypV-=!BTH0YX0bkQVjb8KO4EAf{AXV!qncH9!=0rz*Q=HQs
zKB*jmd&fAf$jnwffkkxMFsId=AVP0}ilz|4bm&h<z_LXh2Ty|*ANS`4Mp_9jE)_?O
zw$C7Tw6=6_bXSaCT-nX#)x8v9zQVjM$onwtNzJ1WXmjT{Xy?(t_oYGH5zDr>nn*G`
zw)PCB8;Y<bXk^RBzgIrUAJ^#4(La~<Se7alDPW`wrMH)=1ZaziUK5@5<jCNTqNoLl
zsZ|OSgz=O37>Ddy_G!rg<RW7hu<f08icf#>HtRboqC?nBh=;C>*|v5d4$85S8R`1n
z_hwb}D8jSL8%;^g>$<z@f;h(*4Ng!G3Vu5O`>DB&G{_%0^x0oc^HV21SXQU;xNKUf
z@qL?ybTlJIl%^+u2Y=SH%Ikq9;euom604aP6}%eHmXQ3wh>2hdh)903;$Mup74V?A
zcgAn;>om9BbFL5V@^cq+^)B%MpMX@SuS8QvSHDaP`@1GHDPb-XG*agvgng9Yihis(
zrg_wk2i^9k!`pNA21VO~u3%Bo^U`GKhgrOWj7ZvUHCw$qG|n(NNO83@udYcsHHtY?
zHo#BZ<JAH2?1jJzpriZZL@+-`?tyM$mV0>2sS&s1`^-tRitew#>mXyu+1@K&G~z=#
z>*LLzK+aN=*a$sbX@xw!okM<-)WN(KF%BFW3?6#Nr(wn123;3~Uu{P8H~+p(`E%Tw
zY#{D!h3Jw`n>16w0Nh%P)^;zxj$eWx;o+(dgBl#tU3)G9+}OJY5#dbq15L%AEZDmg
zlwjNg{Ur(y#KDax^6AdD+JvG?Ka?kg2A~APTP~O){Oy~O#wC)Jks1U|)7$*q5Cuc3
zjK-5!XK-;Z-fSc{_TgcA8E0q@DHp$b^tM8mI7+<L>2*T_%&)V~i`L@Ce!6vf#E_M-
zKO=m4@5~?q4=5CUi5~9C3g|0KAF-;NRy9J`Y_}Vk2e4uj2l2b}Mr-C72hnA-R&0Gv
zr;%Zjlc^S8VT(U2xz+w!fONg^kXr}teYfD^&1eBxVpzIreP-6>2_RFFa;RmI9Z+k`
zoI5OS<9x1R+`#$e>6#=S+spLaMEfa8(UsyQJRgq;?9#lQnHUB6GYQqHtU7c#)-^6D
zzP_8hf{!hmZ8m7MVVGsH8diGmn%CMklN>vpT3~s&GV7x!MGLB=zZVQ2K@pj<v(sT?
zfk&!sC>?s#y3Q(mXf*BCcjzLlujSRUa-iY5AF*-=@ci->;@;_tw$_N+<oc1Chl2Vr
zdV>BpzP$k&=#4bYwb)$k62{>4UUlThKj12?ijg)6BMG%qj-@J3kCsqAXSb9Xm6X%x
zgiybCE(<$G*~?Ka5`rLNe~6>3ks`Z%TSEa`F$@8La2hi7Mf#<qHXD%~YAzp_-5rU(
zT8+I8=Oe(|$x*plJ4qC6tRA?90f}0S28akyYisEMRAA+<!Ssf<$ohgtCt-rQN)r2F
zf|zCv(-@HY;*AC|BU0|B!5#*W_q*?h!aQgxN=*4s+0OC0AT~Rn>Sw{Cc8^xK5PpLN
zZRrez^Dci$9fNSO%*MI5aU7BD%genK^lxuN9AM2V)J;3DBiuWC`BoH`@ZhtNfd(<b
zE5}-Z0XcsLt6l^`cx^{Cmrz8mN!9$DuN#)YHj?uU3|#PEyTS8f;oDJsL>}wb&c#x~
z&hllM^P4hzcbYAiji141uU#2Zoyo!Omr++MT#ddD^;2_go2EH%vnN5l-bv%ly+_ou
zoUnP6IJrf(L0F$&E%GgYK3s~n2l%Fa)@!){hf>086Ps$i(hVyXwWGO<&ef|+h@HG~
z3?rpFjAmg_k8DQVUNeL|ricNBM=wTPNg0SLfe{{dD_3qB2UDdq(Za+<S%KN)%|I_%
ze64Kjp3}TJhU<%xxC7_2I@hZ6k#)gVZ;wVsE^!r!r}v2N4ZpE1aAxLm3AquO%3t-s
zQ^AK+Y{YbgEsbV?CXTrQ7=GnKXjhIKId-c400Ng5Y|_lr*eICq>nOW+{U9x>b;qyN
z?`;^v-Tm6xF(=kQ6v@z67)xxCU3Tm2A;W!w><QN~oHU*RLPtENkzF+fc>xi<yKjwx
zp>sF#{%^VF*aZ(`BF3<T8Z%-87cpZXNDNku=6D$D0mI6&s61{pD>xL4f$CH5(n)y|
zOa89{ZTVkfvC#`?m?CoZD<2}dA)(5B;xY}A=GZlL@<!gYWj|#Z#7+L<+yr!M!c}K*
z6y}IyaMqwUP*s)5AE6YW(Po-3IxQy7uZ68XU9f_RAg?S0Z;St~2ZJqA=Iqsgzj!w{
z5-J?!8S@sQNWtr;@?|5TEl8&etyl~L5c?KKBA~7{i}>Adn9fjU5A5qtnk->LKaS)n
z_UvVb5gTYGE7pqfa_DoalGE9WF6vjkE^Q!wA1=}Qj1iSPZ`Ak_ebG@EkTzzMd&o$t
zW-tB40rbBKNKcqz<+36(V%87<m+Zylge>IdP1^@ens3u5nclX?8?|IwF9fGxt=ZZ6
zS)vq{l5Qe2vBo!I2*abFBEoLT#4e!pG8_KOcpL3{7p65zs}TKsf36vHlSclnqt9IU
zNj3JrTnn5I+8IwfKIeW-w2xc+Yls7~Se%_df$d|z5Oxykx&B;uHZ7Wd!~9MMYTReZ
zmX$2?zMg+hcF~e!8$Z>8r!1)%fMl$MoFXz3fE=ctFBOkfD@GvHMm$ex8o2MiFlY#}
zhL<uRYtfxRFz7V<?a|4KH=aA64&fHZ5&hp!X6X%8=S703w${j94<RZpb4NJ-v6Yl~
z{#FbC<uDda$T8~&+}vS;X#d6YK^$Y(#i%om0DB5@SH~2F=hHkHGV*58N(!e;6O6x!
zj@{2}{)N!fP*!ki$a{m_#{J)uUM4VBbbNF8ky?|W@3JtD!P_4ll(<Yh2DyjuK?k$%
zDVVISN9P6qB2)1u*Iq1VE;mZz<nP{L5t;&l?jLD3Zkk)9<Zj(5?I>yV)fSFkma2&>
zt3hI#@kpkZkHyB!2We+!Bqfoxw9X&7cN7NE@6oh$A+9D4XJAc&24$PZx31^oIw6~}
zPAq7-8yU?60m93AmdOA(o7?W;+gnIiIHXP@(o$Yra+FBa@8L-;(?(|URKnU^9%L{e
zz#Tgy+pKdzjoE#e0;hY8s4LCw1IW2@ZisKaqr)krjU(k&A&>gO4yz11CQDG+ReL3W
z`)c%#5)r*L_IyjmJr%l)UhQ~&lDvP@pDKoD^h(`mRNHwyMIdp!b0er8(@1-NQ$i-i
z@@;91Ue%)Qdst)YBhe4Yn=-EGudxQWi`!)9Jd>WXkUl#7<ylsend#?;>`&6!%iDPM
zHQ`5}{@=kSlpO0~`;osCLKUvth`x<=1M!h^;v_Wn9#sLk*>|s$lMU8FyHseMyZm<O
zYhE2(RLEphCUP;Zo9m>d_iF8+#qe=YK;~`Gx4CmLp~W>}bK<4skzWFoSa($rzau>l
z7-rix+y4ICwZ074&2D-W;HAmd@A2U}37X?9muZAv{QbnAA=6m;V8eON*|n*}_1R?Z
z{357@FEAjfbZT^+3uB@gw~OOVoY|zBfOSJpKSnv(RSk0c84DdQGx%J5z%l3dJ7ewh
zPy8)K(#;Pk{cFryfyPciH+$kTVIIE+slV5xs=E^}eRAV@Daxce=D-+KmbvOQKW~!f
z;uirL;w@LPHPkIv%4YW)(o2RQG#B_dveUboX>22;M$O8(OA8|d>)>k0VSDcysLZ7Q
zvNby@@m|rY4&fBo%|4j?kmx<3gmyVAA2l7j7+LnbM(3=wXPzg}-bl1=)vlv^>|nNX
zqPsT^tFJVy3P<o~<ww?KEb{oI1>VMkw%ud#ESTYX$BV%E!xKC==+tFRWf&`E;>dHV
znDSIajD1$-^FJ5m8HoHu*S~6S#Ema}*=wnpHW76!v<{yK*KPJCCHJIE(O}hn=P9#o
zwjV^X4((S!&K?|4o1)WU+;2EtoSreMac0X1UpRLvjot7({*>#qDg`~phb<)_R3&FZ
zc(K2Qw`IBE4v^Cf*8bzPIFw+OXB@VcO@l>+4-4zF9N#p$*WWHy9*La2HEO8a<nFs?
zokr|YF9rFBFxI_rG5F1UE=b5=HhXSnz(~+q)AKKz=cYSQIH#Ar+N#rCm-&UowPRB$
z81A}o7_^=(K`)6Zx<}?$o{Ke-{*Lq3^R(6coM$YD{mIR61})~uFM*`TBifP^K9vw=
z9<>L72LB=|^|pv{XPN1uYu?}~WvmCv`>Fv3e>L>;XqXa@sYq}`yKSIJeBeTwGGLEW
zumQ%fx_J>ujkyxdxll#$M==C}*zgyV+tLkfbN6s@e`brD*mmW}Y)|gXcTC#n;@s^I
z*yyY+`K9rE>QI}>!KKExrB#yGNz0Ge&-KpqKHS4ZCg=y>sYy=}HbJsc?yxMgvTI(Q
zs@5QM4@Vb%uZYvcfqsHJzDLY*2UR`b*V;_Ku(Sa{W9a*!V3x)_sgr)RIT2+-t{P2f
zzdk4`pU27g&uK_h7CkIQr&f7^t+pINwk}3!HfFCZ+wqW~Z1$@=4CLOR$Hm8O9<CIv
zK7*A6ALpA)Vu2TE>(0qv6nK4Pun#k@;s13~&aKlBe&=hHD1}LWeV=?HB~0}0TrJyp
z&?f_o$`_7r;NP#T>F#S+b~Yn-f_EA%wxwnuco)!F)?u^l(#;cmHjeD(w`mzZUpgWp
z+66SncqJ*YtI2*3*R>Qj7$CO^I)lh`71yg{-qqJnFY`cO4y>g0AXwGdduw!{NTk32
zR+&J6`JrZ~Fn)7|C-4JaGyub9<=$B(%nJ?3>HsiSO3%jLCR)MtBy)zuD4B0!f71)0
z5mVdaH8pWEbn;-BgU=wcf6IkN2OsY{@?=pH$rw)<t#P29x)#+;`9gYZ=Edj8@6q0A
zTFmv;Nc+}Fr?w9U*2QEHuXD#>Csi|0!m$VVR$;1)ZxFj}U<&kZ%LDYq-|`(h4_Sev
zRE&pk9=WWn_M!ahtBZ6dQV$R?vxgP5J7lcw=f@U+_G&oLX5NR8h%79oFG!LoO6XgT
zP$o+*%sHgU(ay=;`SS6A2l*SvBJGA`=^dPogv3|b`PQZ-bCsnc-p(2IZ84v(3U>U=
zDp`VzmZKh3Cid~K@peXBZ~CYom<d2j8A=LlSHvl^YajehpodBY`3cn-4EHELlr4gf
z0ehV<xWD+id0&lr5@K$k*s_+JyutF=!w_ps)8--bk;HLGZ5eIOExYWHTC1rJ2+YY!
zkH`()M&&9O3#`vf@0-sueKLxBc>XT8!l`;AHM|iOJRHK!zxRB`ivV2XM?L`K=56S0
z`#JT54QJ2sjVB~%jnE~L_AaR)=IaAV5e3nxPr7Dw7uU!qx%}hWM5pG{v288K@rR#d
zn=#Gz$R_?h#KsgT%(kd1uyg$aLbS%l7vp9x;u(mT$FWhW#yv{WI%MT1SF7sBEzMPo
z`Vpq#7z;zJvlWvSv$OG*C}4Gc-?-VE+}k)`tivp+`{9m0NoB*Prw+{~{4&`e*>&zk
zt@b?E9P#*v%rMAthzR9@EN2EH49lSacXrnV4+Ca_5SY*H#oDUDC$BUDd+sF2x*>HI
zq=J7Y%sKcP;uHT6b!%Djh6?F><rUYGKV??+8n;Z??qR8fx-wSy9RLOuh<ySp2rIlh
zSyR%z8Zc-kj#;9Urj5^SfvKoCmnLn9klfYbAWK~m4q8D`P&Dfjx%6X%eXph4v1pMG
z@+RJIg4bht$SH|-cbK?XNkLfIPFas2o+XY)C?P7Wo1dNuRJ*PYGZX5Pv#*AWb0gIR
z(W&jHw9e%TY_{(12;lt>H6<gaTkAqr%%%{-Auy13Y;fr7BGr|?M%V|Qs0@x(s#`dt
zO+R(=8N?Fukge>M`GaoNOW>a}k4Z_zU!BE~sKIiiO@_=PQUlywtcU2srtWbne_b>O
ze92D0;RvOTy9AK{mVI~QEDdIh!|ArB_^cK%)tO;Yy>fA{1+-6{S}n>qU&5){b&3|0
zK-d~KiAxk-kHAcVFPRy%4IXURJ9ig3_c=?N)Wa_VEoA9r>W-b>;%wI%wjWpk_$d8l
zwlQ4<+NrC1OBW@#&lhJ2Iih{v$4X@98MjXlhQ~^#xFimFFy^Sgim8RyrNP%k*T&IV
zNuU*^L92{>fZj#&WSfVZ4EvqIkT~Bqwq&GC>HIU;zl&RH^n01k!<P#@U*tbZyruai
zCFffSuvk6CMK9nkew)cOWT*}edKPquQ!Y-fWG=2TZEw2lj=S&m3dI_n49|m<kn4{%
z-dhk&xtEkRPU&X8^yH2G-%3VM#Bh+_dbrZ=riY{j;S{%LriL^Pt{1@Khx@++_4c7P
z@oYtM-=_v5APJr>wep&b8M*#IQGMa*;?i$Y9e-maB@7$yt)$-N`b(p4lYfFc?k!$E
zSB$)U<ZYWXGW#uDHaG6^<@%4q<BOcNmyo6VyUwmJF6PG>I9~m)Mv#X<n@2aGJC2u(
ziDr`_-7!(O@?CT#q000Ph}Ceyp@ipldYPm=i2QZ%W64k!7l|p0;xXa%WXX{%_U~5(
z0PQmm*chY2j}-NQ^iNDf%5}Idh$Y#G>kQFxP#t)PfG;+EwUi}vt?PW+BAJ6`LZIcJ
zmWo9I<U-f$YG^!$q||cqO8fTG@0!8Y(c0NkcU^fO^qxz63`#dm04oGNCEm;SgT`X_
zIPbT2D@8%Clm$G#FhsZPqy)ydcOw{j{#gVxKDr&-MV5F?mjxLyq%{4Wa+Yc7pU`GW
z>UgDRd9G0I>RKX6zthM33~OaRzb7`5qIl|wN);CHnHY3&_v0C28?`34sGB6~&h59-
zyU}4~kLB?CH9mcLRQ6?UlfF?@j_1P@MWo1xNnt<23?Sq6<!@e7yiXYw>M`#Ipk9wY
zR3E9m)og|g^d$PE8Re9We>0;WkoMK4DF`2h7kSgRsS0L%u9zDwjZ2#v;HK*w{GG#|
zekRv8-UeMkd*!Wt8`|upF`ee7llLFwo1R!+K(@|#6m}*DT)y-F{JXxD71tnmgH#X7
zEZWKrC?L>ODMM;1KhTyqFwQ-U_EBNvDJ3XS=R4(4ljFW&Dn^M;P=vhO_w{Y&M(S6{
zm~;^mCKCM5O!9fY7gq+sin!yW`SUe}+mds>R{J@6?lV?G{#hX>u0EX>w5LZe^P0{H
zO=3M*fM-Ac#6`hsPSp2=z89Y`xHH6zcu1N{84Lyg9L8!7WIAvb*~lbo@C#UGD=Eh@
zV1W_9tU7imZ2jK!MD+=9ugP6M_c|_L;jnsbQ);;@%nYwS$5?hLm{x7gFMnHR`JG?t
zed48JM7!PYO6}YtM*TGYh=z07Q)x3f7WSYHU_rHr@iIflv;1mdalL)391Emuy8@&0
zx~N?_U#~Gu?06yGSx(%b@KP~fY)5Vpo0lX+?p~?9Qcp$LaGK9rsVJ_xY#_lf?0{Z~
zkJwj%Q7m^{E%7G+)oDF;DWSVpM6$1G<lFK2g%u)?O8IJs``x|<m6A)o9nGBzEEJ>x
z7$yY$md+)Iz}cql8vW?ZFwZXE@RMRSyr6EX@4))Pfe;eCiE{s)H6Y5l(+2uTjfE<8
zfqIo#wCSjFv6d<6XF)tp2>-39-O48J@HW3Q0msFcYSf+z{K$6z{$-$?AcW;NzeyiF
znH$)lC^zbcl5V&kzLrG~ZcQWqDEZ|CxG|p|&qw<5<`4MPT)wmZRWxn*ot0Bl<n%Y%
ziIdDdUEu-I@U~g*E>t<idW>=S%A9jyTak^WG=0)3!~}eNz;D?NlBN!OgUSz#()1TZ
z2#X>GgP(WN1jzpE&xFf>z12>glq}Ojdap$!b3P%0&JT;-GYRRF!m)+Z;0+>7z~8-6
zem5F<zTdjs+5{_;{^Q%t=1^HktN<FFjd<4lt(u_A!ObsAOni+0!_+rMchW{r#+cZ)
zZB1<3wr%GZ+qP}n$;7s88xzdt{qOGC?f%lI`&0MR=iKL3)vfXrQ9Gi)JP@v8pKQ~r
zq%<Y6_wQfPX`0HiDlXB8tN?XC)~?g=SL2ShNi3?nb4QAQi$LQ4&Cq@g4xSrx>7tCw
zaYvnl>x}0+pFDDnJZ-L3ER}N)XeARnV@42%zuqoyF)ONAk$i3?aJ?>DC%N8Ki#poG
z!#t6;)5`JFufae*%N7QwL##4+4um~v{As<AoR*2>F`~5bg~7au%WrrF^A1M9q>4&~
zON|#D&*}vXUe=7dMEtCN&(Rjdj526PZ;Rpu$+V9JPok;i;h-7xXm+px2-NLMluw31
zJ~G4zKY?bBSNVhvVQ%I2;84CpIj=0v5#p{^`4P2wi9X8W%y#B=@FJ2&gy1F8K?{T2
zS7T06Z)Nkj$B3@_gbT44W;b9eBZ>@>1GY>YEn;pWNAYrJyNg8vzn#nR5<FR0$~G<u
zur6L)s(VaUo6rO7<MJ5*mj9a%v;fUR0z2m%rkx?`UPCNU*7$2332!&ly&4EK^jB=|
z#`juR>sc=ha+}C+uShpZsX>KH-g{MpQak{50R!gvrxT&GI@os%^aHOG?*%fElD|d)
zF9n`iG+1q#$c5QgzpW1|fL*g}`zGed;rnn=woI6@?ORsq0^EE&o86X{e3q=h>_&H}
zJttG6LgP%US+mc?*OW<D=6s94%<?bci>pL@9|-+b5Up!quSM#3VNCDUjwTn*1v{|M
zj%UGu9+4%h<=l|}NjLZaD_2w<KQ*h<v>osn&2*ZKRkQTHc=}tstGdyH11^AW%A%~n
zG}D_g7|HZ71z_h1LOYa6dR-G5_s*@P+4$Q>*{z#6En-ZXVe2X4vY6~1M5INn<>OlB
zDq)<2%%5vjrmep1oDkht_{vpQs<Rm}WRLJ*TAJRtM**E)dFGfiXI?lloJa*F5987+
z(AAoT0Te@=g|1W3_C^}!@K+ejM=b*nIUqdttvc+tC1BV*1tbp*1z|j*pWL&=>)5ua
zmCI8&PFi3mXLPhw_(WZ{%Xg^RUzh+rv*rsBtaT~J0nK}Ya(4vEq_SZ_1+5U`m?DXf
z-z(q3<eGhVynNV#)iumbwGo>W5F_UKu&H&2?2f@~bc!pcPTt@le`pd-nm8m&rodgU
zmy&XE4q(4P5`JE-g_0N(zEh2EkBvRJG7eFEtxwB~zQx_k8)9hE@Vxvo6Fe2SQOf<Q
zW&&L&(&;vh7bhfQ`2*Zl_a94!(-wd0ouT9m#^A|>>uqbnw;gAx*c7#6EsDUK#k=wY
zK79kRzgTCafqRO^f1y_*j^Js$S1NgLBA(kwxlOw;95EGh<asCKG1bsS(J&hs?Ed3x
zks&Wn(qxb+y^^*$;K4~HyJ{-y{ZpgNdD+(9=)%YtleZLTh_E%dX4iPC8o#--{>G4$
zKZ6DE+~MM%_cH`R{&PlXM4AEAkiIY%xiTCD3lmaD`9H^Gpc97HU7n>!3zi(C#LR*R
zlC^h%mZa}h*q~qeQYws#+q`!@1$EBiXl{ADQH5R(Te5BB_rR!P15~L7Yz%CLuxCn)
zMKmmWPWqK)B0(lZz80Ui@7ha}C@>o=fbbZi)U+Y)O_lrQqSW7MeoC32m!X5x;ll{O
z1L)YzRa)AfVbV(Jx3hkgeudYT;f&}`%f72)ak<uOnG)oCHsmYR3b8x#W~7s+Vnwnt
zg`omP@CWJX>oErQ;3r^M<7pwqe1OaRnMVo>UR7oU4pJ2PP9T!x^TpfNTD+?cKz;o8
zooVx{1wxM}*K|dYMemJQ*0v&)Z0`4aG}Jh%P_vJ(?zO<!F?c90-@41QF3}+HFNO_e
z^9b7Cg1$M)TQ})8nhb4?*(`Yw_Y#_Gv^SpO47waPVOGNM$nZ~;7oojk(9q(jt<cPK
zHn~lVXDpU5Q%{*T=9}c8QH_}H0Erv@u{EE&&Rd&pR+B^5M2<61!#s~dH=8hv>rWLr
z>Al#pmN0d5Vp;1!3Jpy1sr!SXuJ$afjk&yv;$oyB`UF!DUHzEQ*x|IaH|_iamJ|i`
z538EG6=1&l70Ycut5+bgcz~HKKiP-`Wc~?s^JZL21)O@DQ!oibg51zEfQSpxux9bP
zvL{Qf3*IASdnOJ4yfW{19+vDQuPRyJ;{^&HI^AQ3#n$62n0^ySrg99npq+C(#Kl{I
z2SnLD*^~gu;>0DIKN3yfqgi{Qr9|IS19Y~?0yE_>8ErL;fiBP`NQR<~{_}*cjjcaJ
zmzyH>YRGB%+LB_S_n6%UZn=}IS1daP*1C2EuY`8Q)vwM!9%*4iQc+9xs!0m-Sb!|>
zd&iYht=~1O!Gn(x@>~)W0fS@`jGWT38#p=`L!YY&i+_c9X;x2?K;u_FKZA*^X@@n!
z0}p6_l5<&R{#H$_uZJkGaOT2(<;h=0Mh_zS(|K2K%K25xmK|dJ;|CpobUwg)#EwR0
zLhoJ`$@XAvr^${VvCY!y%E(){p9RGI-DC1Qiu1TIQwQM2w7;%`is?9v#IPsA_#)1e
zD-(w5wo(NWd<;Uv_%!Pk_f|=jXQ#<edUu+T#Y?RnlF_`Y1oi|V=+<QGMY{4YcA8I-
z*5{8`4-O)>8RcA#lZ*a>&h`peAvb3CzZb-8q^~z&LR*1uz>8XMZCA}pr3O4_S~G!b
zJ||Kme#E176%3p~r4u7Xa52njt?I13>ePs!ckjaF2kuI3k;0Hp8p8NYd<I_d;#pil
zVhoHoKLxPAc){7Ad#u63b=d$)Got#RhQ3`h`6lWtlF%&@l>TyqKWws5U2nbll%`Ic
zmh_Cp>1(I)Ibei$GLNuoqXCZYs`cB>>ZF@!z_fJKKqMU1+EC7F$_`Q1|I|(|=_lLa
z*qJ+_JCcC+gR{wu{(;C*C-YfFA`~3WJgo2IGpJg136l9S_d2w6LP$!-HM7F~UK(pA
zhk+WgKx`{b>W4szH=3h7@GLUqIVzlI#BVX&GA<P*9ct%o1BWd-BL+;?Rpi+@%Hm~>
z!6!7Xb7?opV@*v>lS=yV-roFvkMXeW9T@z(VZB0`ZaIHl<>%$DP6$|x>3dH;lw(Si
z0ZQ~-(rmHPXmVIRzzCHueBK!F%E?Wcvy=(%A2%Q)zBeWGYlhwe5olPK+oryWjM^(C
zG#NApri65n3wDqTgaTkw6#r8s``<k$SFV=y`STL~d%2IAWY4;?ot;Wa)k?&|S(WaB
z8afB<+|IWRc9G3kBtmGSJ-ZKS@@8(-zh6h4KREPH!pUq2#`}Uzi$vEVg{oY0R_a5V
z7DM-;M?Ae0EriDdxQ9Ly<7hvwWUKn7yL!8}kTv?=0@VlrVLU;X(!^9IO?rR@#DClb
z5<`mhAvVLoy@IJ29dE}1q1j6D8uDMV0;Q0|o6H2;fi!OTiZ<v$5sDvBW%#_Me=8dk
z$R!DC(^CbI#Q0I>pSDl0B5p_CFj>9~^q#Y96@TDvN)H)8;~NHxav;h~PW>HP1sPHc
z<1<YyS2qn{D}h6A^!F|A>W!@xXFQx)WkLw_-Nv3`Q!_AY?Z0CGnE(jq%g5AF0|-VS
z5H;pj*Sh~$bkgA7hE4qRBy~Gep^(IX3*?J^D1t`My54R)XNK#n#S}n4TDhoFw&Io5
zd=u20U*He=2s68q9e4JnqIgNL73^*l@ar;kF%^I+@x|3N4JqktNsxwM(K5kp&%gzm
z@+YGK=EX!Rp#kJ^Zw&zSp>)IfP>|fLTrB5`v0hVD6BNoL=-}M!B!d4d$T+*xb(uq;
zVF466>ZMvnYre$J@K}_oBvLsejWoP0WK`lPp$CnQdful-V)9JUpG4AB(gYK4rVx$9
ztz$rU6AuWLLSpv11g*p`#rUHqjj1NcN~3dIPj!R76xD8G%UFpczX4DC9}1py_EZ{q
zoBZX+H7(+#A1vfPs;ZGS3z_#0?R)eW#W17iyv4|N(+{)sURrlupP!yxs}T*_x({1z
zVVcQSUzlS0U?eIn_SZmE8vSogwZ(MNuWEp~spi0es(P_2Y>mF?%0mUWI?i~eFVc%e
z4!UO#yPT$yA2-9Q?3e;(Rel1I>skN2Y*2qTcrO0RNN|Q-5b}^uX!kTP*-y2N;BrEt
z|F7110JC&ou>0MN2^>#On1+P$VX2_o^$vVk+lZGuCd%hLKWQDZ)gE<f6;(H)08(7E
zX+Q%xiVZVR_)vzk-2BGI1)=TVc_j#Jeucs%h+k7k<Z+J`*C4*6ZEAV<_%H9|rb*jl
zfL&b`D)lK8g3*3*Vldv8uz|GFDl=Oc-x7jp+rjn`g!a7lIWf{jS*Hb~-pZOC&<+kc
z_<B<J_`|@*74GWLsic$Duqx2K@k0_3S{CTcibmqKSi%)`IH5!mpg4*a7qPdNv-=Qj
zRjvH!u24hRyM$m)?9~V8bXBFEOAsrxDk>5daQ!|hld@N_0i~HGf{3xO4^|KZ%0tS+
zrl<wc8p$uh-v0QrvN4K>-COOEQ-{*x7jw{&@)tR2Fr;|ASi1ph*7QXGN7e`7HvLQM
zDNi)7sczK-Z<uVIP^N0QL=kN5lq6wfYiP*D=6A{#^$zTuTKBOy%(l?#bm|qh->Jl!
zD*D0+K7iJ@{M(7v)-ha*HDx&}%E{!N;}gED@l`3b;wmrbPRddd3%Y%TDTfwp*rUIE
zLzGJZD`W@X!q_TIlO?fV6X7#1{9;~7Pva&F-Q*mnSt|P38~uSR?!J`>XWv#tCW}od
zqKc8j!g2_;LLt)$;W_hqv3LyB+QT48c?eXC2SC^|$Vu6_Lf{99V8PPy^95{M9xHNG
zXX<K+A9eQKV~2}jTA5e9<XN`~L7m$LomGnwt9vH`?Xu?7HT5u;Wv$C`GKFo8<^$f7
zv=~VF>nXiM&O-INIRI$y!W8rhP@fX{pyo#K&+}2<c?RkLe&CUW%ihge4%On0oF2w*
zfG9|EF&sM?oW}x$_d3`*i{rX1SDLSm)IUvv-*Gbc^5lyYVI1;y4k*Nzw1~eo?==lh
zk3n(<ik->JJt&EqFyawnV-P%cwv2-kxmhATgFwaqLBkVnX|)p2ZNClpE#Vfzg$;iV
z^Sb6zv5f_p^2wefYJ_i4Tpb6zn1mgM1NtNSJt<F5Q@je*{OZ8-mJFTTN@@*5BZhAO
z75WpT)r58PY<v`%gT5~CGfcIoqYVFn#>+q#d3H0w2Ffro|DwrZVO+5(n@k30K^nrj
z4p5`3{MyTN(&^@(72OknA(77Vpgg9a=k7SOKH4yCuG^{7pLBP3%<UIFZd9TX1vJGu
zP#ksh6SWTM6hLY>;NYqu0B!!=)>c(RFg&IH5`8api!>8pcF&tUxfS$uHPvd0>KxE}
zDj*7qtDag7;S^-tsD3&2^f|oA8D){ah`j6i(fWV5-w2>g8!5)~&5vff3Wdt&gN_Aj
z_19AhLO*qcPpIVb$IL7-m2aps116Gvv0%Cqh>#i0kT>+g<Vwx%_X{i-SJ#`mxL#<A
z!oHiJR<|O%+|o$I$EJ}$AIBX&VQC-wnLLj&Z7&hT{3ZhJwI;$`hM^-&;zKb0g(bsK
z1UJ^1>l>{n`;Ib%iI(he{N=FN?QsT-VK`BU5FER9S@6V+dfIZ*qJHO00=~jxZU5T8
zYyy5ov78}zXGXMtDl{AXlMmwog+~Gsj|kov3Fo~wV1G(SwP($`l0-gz5St!<tr$PB
z!dL#tXh30}FI5gFs5_mtTKc{*OMU2DyY!>=64G0<XQ24u)8BExbHkYdKsKPkm9IRX
zls`@Bn%4eqivzq7l6eq00Bw!@N~Y8nd=((neE*ct1iT6TFC*bcGzL;beJoqbBT|{)
z>7@9(PoOJ~+Cmfjp1W!{tdqCCT)0o3^)R9XT4tSb==2O!Lhp2Ie}*DhoUEz^BHOG(
zMUe*zWFQ!319Rjl09)8k<KY78WdsGh6HKR>Gtpy#VNb8UA~6LK@O_j67|QNyWyW!6
zW6W_+Wig{?anBF1*(s7}X)Z_ieu#L*xb5^H!j5#OCvg%KOqGu~=2(&67qcgw-<NpG
zTb4|LV&FU4plB)+d7#x>PZzI}XUF>s&cxy0^0lK-$YpoG?A8o<vcaePg2=STMcHii
zdXn@nVXojSVqIYh5NekbWH<mk>(l!ihGk*&&;i-ZU?JhM`tkc+eEVCl<^aKfz--nz
zG=t`|i}ewfB(EWd$h%yJCk|j!_+=Ew7NZzVoWRCCIlHV52fxWBLj!>L>i?cxh9dB}
zL0Poj&(Exh(?9Xl`N5=fzH}$j`?Fe(+xSM$w-c3m+a-P(ot>yY8}HP|Y}lK-66d|3
zP*8w@L{Jd{D+>*HH)CFxws>%Cz#$qwzR}zjhziHf)Vy7MsiJQOjTIiahY-aET2$$u
z<I|Sr=!S=AWYM9F2!S)9E38jeT;0!<e9|InJ<%t{Y-0d;-x<Y&t%AgFkk29BpTk~Y
z68S=JvJq{Km}@40p(tlB!04MWBsX2wFk;T-W57H9LW%L2(;brVR{ZJr^+HokB~ew`
z5=Sa8_~P?o@3-~~d(I?vdDt@f2oD1Gu**%Wboc!|!wB=Odwg?a^1j2AjB;v!+I#wG
z?)LWU4eK3nm;S~GUtr7)->D7u<dCsR!`sm(F!(`%C#*UcWe|8f($<-bU83bXV{m|B
zfcOB47IYN3uNqHqc7`w79W+DFJfS5fC<DVv4;(5W19KFPhRY-%6Gd_yAo79v3hR<^
zLietoX*Ap{P{18OhX=G+7?_UkWJA;LH{2{CH(mt)OGpN0xEax!VKm56Q81}~2gfXR
ziBZVCxWosk6SROEUWSsB&YrDtEK+0+k7qS_4)0?Q8)_Sg=oi%X{qO%)M))k&HxoJd
zrazyr;Ez4@Hd1sSVyCe@IWyk#_k<O<koy&7g*RnN=^4UR#T-t;fMA+}-OQSO64Ndw
zfQ&*uQ?^gNYg;ajbu-wrx`@IUD9wiVmGgBv+&ONkDQZJg36wZ#jk-*wN%ykZds<K5
zcJ-UsG$Xqlt~GtLPuXRMQ+4?_HK&G2Ak5ARUL&6KMOcfBLLb&D%s<Gw<4?@Aa|ZfZ
z8iGY}1AOxy;Mq*m`g)GhoeUbD!llb;fM3(%x0#P(=e60#&}YycFn_e4!?)bT)ld(v
z52(*H_if!>U5xxOO^b6$W+y)qk0q4I$RRNd5sp+<nO)=GdYfhH<B@YuX6+4o>V!Wj
zor0X!XoS)p2+P4t_BJ7E>RHSP+BiwjEw<MILV|;y|H%?i`s;++v?bcSdCju{De^d1
zAH?Z|pYY~^lUAA?<#om&8ILuAHL&GnaSulvjgp5NKNGWRRixu-T{i=6tYw^RX>&PX
zI#z}##e>UAhY=DAR&qFH_RBKrR3gXi|9l&BR6$5gMgK$vnQkmXm_>&4?h@elN=sF|
z_L~~^e0t^V^G@}dXqu}~V|+#d>Kii>aT}cp#FsI^kK}5h-o02#pfSAg6%UG=7pBOL
zmJM@9IB5mz<Yp7{!xEPBmYjZ?!%$R6Lh)<DuG-iQY7Mlk?~ii)DDi)<iiMxiiWi1Q
zutzjEN1*38SglFUQ<iq`c0sI@eH?hB*#h|Z>`EH$(j@tHU`ibI2OPtI-O1|Z=Zr4|
zhf$)Jd2yKz;q%0{DQA^mJ46UycmL_V9#hq#VxjuD<Z}t02S}bq2PdyAPGsvgo*E?n
zUWq$T3@iMEri^5}**AiUW(BYdcmFXTF!P99Io5sA*ZVVGGZc!XrH>biTc<^q-C9i1
zJ-BCUi$|J|tj)1T{sW9RDQ6U_XKSYVGwVnrVUSU1yGV`^3`RHMJ_nbOW8ANVbFK5<
z7yDnG?rfUK5wR@eP-ZZ$$M&M$<OS(SnV~4^uCj)H2^lH41(nu4ZoF@{IcK<1b=?`&
z!*D&#pkMp1cVr&p1sF`a25pN^ht4Rb^MVKi8CC+bjL*<wKi&Xl6AL75F)9WNu^-9k
z7PB3~5_jbwUGsJYXqGNc(@t9`*`fPLfU`_gfbVJ_5s5&rU{yRnT2*SGB2tx)ioD4;
z>i!%@wXLi{(7mGD;=<8eA1kS*!|(`V-|+}2`haws+c0<4ehyM_*c!acNlTF_d3y@&
z_XKUws9v`yA&3Cxl)8DVRpE9Vn4wK*o0d3<jIVP+$QK~89`v1j2O6EUD#$w*Z7YVe
z+apev;tSxLAnoj#``IZ>%K6V2g$oD$nx`zs8(4j(qBT2R*5f*-fdU3OKH}jm=Lc-H
zTv})$Gql<-dg`mH<=;@Xx$WbJJbb@t;t>Wv)idAgEJ^^6Vu2uTPZPlAmML2ixsg0J
z#LN$J-tSjWPmM@q!`!=;;yw*WuB~AoV>tLO#?8q+s1D;cIbPlQ6P|4@hS6KE9~8bk
zQw(uAQO@u@RH*S*xC*AGRTXDO2eo|I@-3hAu(=aNxueFmRzsqHaNh}#Z<}+M85kyL
z2!%`x_m=^5FZCH!i&0jZ5)$*Y!aQSw0isKg6(NUT02a%<^Lx0Lls)uT-M>@dQ;1wf
zWLiAP0#0iEi@R}2D7Wq5+E}|h_Vg^C2XO5kg~<F7BvmSb&?JuDkJtfVg^N<<ibU%d
zXpKZ<`~%V09dA_8F0r4<*hJmLeh_0w48Um^F&lsq3CKQ|A?9|SeW@h+Jd?9vGQ$hZ
zInan?85=oaWT~@%fvkk`3u{;Axwx!!W3W4MEdy~|(3&BZZfnywYy_O6e3P}0Pf%&q
zrevF5>2lDEynW7zCIaRr>K1;5bK&Hy*`$s*$Q&FP+Uh86iWhx{sdRP~<UW0lhU*7v
zxnuy-L^c%RLLAN{(!p=-+A$o=dcm-x1TCK}2!bhYYpDg;As5a-*)ziLt#%5k04|)h
zq-35`1lh8&Fx*h{MWeWd09=rASKlwK-pxHV{o=Ld$`YrH!)Ln4?ajxD^%Cy76OUmz
z0pwZIt5^NpXv;YemV#8iG02_h76JT>KbwFZddnTg$<Q|q#+g9bZfKz1a~XN~k<6Zf
zt=GTr|IBZt+ZV-Jga1jfNzV3L4;tSim&Nfy24y>p1VpFZJe|HhCKUX{H6znStaZzL
ze=N~S`qvMe1yl}SFH%5A3aD=T?q=N(?%g<@9H6#9+)A)VTfGORm?7?T4)IWceU$(<
zRA$n!O%De2-th8*ndy6HB~Du2@kT;@9#zcqz->hFHa=r9QEw|uStOGQtwuVgQF8ps
z{%#?TAT_PMa4ba;uhgK9ozD^Zx(<htBnFu7Ts9-{(saQ_+sP|NJ;cUOWq;KDDTF#D
z?v+Cg!Ovc+f=qUh?ch^|)4#oXrriK2<7I|yl|n-j9FOywl$~W>%P_s<n!-cASDfm|
z<x!Zg%D%ArIObwn4P1DqB;U(Jq|UhReRqXY=MZ+*Sg#3)hF+st05Kog>FMnIw&k(3
z;e(j*(JY@JZ=)kgyCF#PfwWO2lear<{EaLREJu!@K8CJz7ziF~R32l#Y>x$8aJKLe
ztH{xqi>G)X67A#naib1s?1gshyDjk4xPkFLpHKu$8HfStf`R5=He^2wxt3RM-lP))
zx0AB>%cXLPg}i{;@-tV-w!q*yd(<h{{V5iwz58pMg%!#BI^Ml{v3Myk{!fr+1CM7x
zb+aQM!kyDs(fets@Pg!@3RoyWb|2mYTdlulYN=+5R$Gj9Q&;om#L;@YdRM@}hb!%4
zzWO`(+=^dqu=yUt#12p!U{OvWWkgx+kJ85RxwF-prnk!t?4B8?^@1+YoDtk|Te@-&
z;elmEU=^RXY<S8_0ZX#ie<;x}RPfB~i<2}D%P8+?Em2Q6Ai%eCf@27Xg*MdlhLY@H
zWI}Go!%+K<7t+2S3N}2AP;!?#gc3V-Y!Fo$GL5?-+dabwV?L6SlS<}UaeZ#5e_*=u
z&3iuranF#J;?;xAqA@K2<7n79_<94=|CY7fM%9N^kf8bzM&5Apg5XcR7)V1P%#{%d
z<y-koatNojD2=W)N0$JoCk1P^3h-Q4rX{w~t{fdjGIhqoK<=7U8$2T-vY;-L`+t{r
z-8CBnw_r5Wlc^b&#{Wve1lAM?oRIF8f>CMf3PAx~POm{)_xw(=Vmk4P!%d$#V6HuR
ze&-qlnD5O>@K>c6YRO07^~^it2QQjV90$Ed1nD!&%g`q7gWCfnv;^v^96f@RNEaE<
zfq&wd)Ezv6?P-<A-ipwy*Xx-tvW_^qncZ{h7gPo=#AtQ{V;+iKvmr}+aIPVjCkAod
zz_L;OVU~&;Uv+gu_Q5lZ$kU{eHR@z8?6XGPt+;$12%9WfhnOP5Vqw*>Dlb&71!9-_
zDomZ4%ndR?TR{NQZ20?saR%{k1apD$9t&|v2|ZALsT$2u++M;QMe-hh6`Nt-VMhPH
zYPfA8m@3#ILVxCi++Ag-d)XoC?gmM9m0x%F3Dh%2sSHRCEs*5~dq6i}CGI)~P$@5L
zRLT-17=E(mHJ%kwkN=g$+%U`AD7Mb<wXkIWX1*~u(DwuI6K@wdo_64^q2o%oi23B{
zTubYn3xgU+XUl=*u?=Fke2&+yEI&|WBg<%v<eMtLl?f2hz?x+S6aPdvbKeUa;8}5M
zX)g=DrlHyMju7DtdCVKr-nvqN<2NQh+SPd}C=1ba2DzKuFV;v}feg50&z?sZ&w8@C
z|C-`SrJez-3_iz+JxX<0sLe{&sH&f=q<eRvLa9ys5xop#%>>g?7cPP`%jXE?dYq0Y
zFv?ekZjXX<^o-bRba9q^&qqZnsb#;MWQ3dtZqxW94>!!GudkP&9Kw1b^;oG`1u-<?
z6j-+>RUyAD?X(;??Q6G;V-8os*ToUAk{z*2zR3@$`bin;3+mi&6|m$!A-tps-i@{&
zxt{_!&>i9c%Y+?yNgNq#Vf-Y7Zi4evqynq{yNdmXBhyl`av4)4R`nz2ZStD8Vda<O
zgCa0cvan#(Z8`mvle!v8!+L{w%S71u**@p+^A`UM2Il?}J@Xu-)#?$59oKyRbSkws
zf4cw#4J*^6%~0c9tl-<ftll#PBXL^YD?m0c^{l=Fp@!qFJ#n|F<UCc_R`LGgP23Wi
zJltzu=6w70*Naiczv8-a@Nc<>qTJ}A3gQZb;dM^DK>bSRR<cX;3=|mfETjg<DTCZw
zS?yn^$9#;^UmRjQ5*v2meSq+Gk9r+w7>7szH%EwRORH-ehn}=zwK)<l68ch$jmy3f
z#xb$EbYxQxGn2z$%Z&~?WKIxBsU60`3GVmJd4Ogck0XK-2(azeWjw=L0PXKLh-m{K
zn`fyCU4~#p8&hTK`YgfsDzncEGQrx~@9tmiw=^G(psW0l^JFiTnKRWqs1DNf+F1;M
zc2gKaSvZ44AP~FFPfDm_%{nxP>#DGB1x6T1kv7BUiTLR4`{+U}(MOurWV!iI=1sop
z33eo}8eBYFGReC8td+!qQ5CehIFiLb`F>5aW4{(e2{M&?(k&{^X>w5OYV<@Vv~f}v
zg*>~%{Y-h*T+L`;Tw3f_F6R%t%1_7v!z&2VC7?uyFr%ZzpTt-y{jM_}K*#%VlhAuy
zbhzlg7O_C+1hxCkY}~sRU300j&TFn@g6P#^5dp}?6)7x#6aydO%*ZF4yZ48w2G7Kv
zQfy$L)p8L#1tOCCbw&-JRT8eHw3=iTv=p1tu*1u*D5zXsA|uG3ZsUXeT1jRBOA}mY
z^{zgRa|y|CqNN*h8x(fp4?U4-snEu;DJ}s9IJz{e$%Kw(@)TK#M5??JQ#_VQFtLDY
za0Z*8;_Cpr-<R;M67n1W<WQL-6#A%zzKo>a7poRPlvM_r*kdFl7Oa=ZXF~Z3i?MBV
z&c?*48XXhQ6i^BSIbFxQ1_!wTI!M}nu)!XMBd$@QP`DA{j5i|Nn3gQQegfkb%6lTZ
z2~`m&5B+7~4=ZBF%?dkS5uIZJ$A3T*5}DgPOt@#QlGp{@B8C~)B78%Tq1{s5$!<dT
zqe{k2O#9d&c9gDa(qv%!lx$H-chQ{$pqJZKi+J&4WuIX-w#|4N`#pZ#dQdTo<-w+C
zXntrK^$Gj~1<JwVeD@gaTfSVWU=SEJ223+TX3`i~yW9IKlh8Afz_@O=`OlbYCA*{6
zK{>B?CP5nUZ{ceMzdY>s-$f23z$VIgQ;Q(2w(@}~nhnS%qCj-nU|j}jAOqCTNd`gZ
zLJwDl(BGUT&UZ{jQKlMzKySYi@<ftl>4!2WKX`!q&rc3S*5Z((r0&G&q#Dl_#pr~V
zQm-KpL89K~y6|%=Hld39SS>|tAE`vMHyg!MISzj(ezf6m22I_!5_{*ZcRZx6tfsD)
z=kT^qC2N99=Fa&R7Vi3XUxjof?;pnXucyBJoasAn!n(H>nHvE6DPONG1{!>EoTyn*
z)Gu9?K^w+DH_;(m8Qh$u9`*j>NY-D`6vhiM_gLo{_#V0WSyIVtQG2Dxg~&+TDrcoy
zFU9k+WMNrSQv4!NNzlAR0cVD?qsq8kZ<fzY5j}Gfu>+=2r%>SskxrKT`yW0rpwVI0
zf-+_({DrR1ulj)SI`pyTpb}a7ZP*n}DKIPsewOKGNI4sX56*TE>$WWfG5N!`BMb%Y
z=E0-yxckYx@mNnSn9AJColVy?yfKqq<y+*u7-SSQX7=?4nX|po=gUo3+oevW^{Qc3
zAT2^)V^281c6zG2gHav6!T}ucZOe6Nt<wP1@DrYyqZ2^=9AXQvyi>}crdulSIQ_8^
z-nHiKTy1n<7|Syp`Dg7Mf`b(fSUmB=s0f9OMmj^%(z!#^kz%pT%|A+%*wm&F3yH|w
ze+K`xXminDqrT7S`EElj_H#06KY%_XvUSE>9F~b@KtD#^*68X;h#Uc$ra*2u?RV}g
z?Nup!MgRh1qc^{^Irw2rfhOwX@t*f9w(mQ>i(Zs4$Nk~PY5G-Bs!;1;TYR?VzmOY6
z8M_s1#F}-o*qojtm@LP3$772e+*<BO#Hmof69f|NAL)Z-iPfj*gA;0>3p6d_h!gX@
zP*b?Cz4gVxL+PcDsYfwyP$zN|Wsv`bL+;BF#{!UG0(0@hqaZ)_yN!di^;cyNI_tbr
ze|u0`h`HB7j%{Z8cNs{pDuE@pb^c}!o7oJoQuwmV*RtB}4N!!!7Y|7zw&@CPO@Qdq
zh2Sqe>wy1V3iOfzVu7N%>X|m?4p%hk%KIjrx9PHnSR*YHlz}k=YP^0GYMs@rnw3E=
ze+3{brUmEwlsC)hm;TvRCsADaLkfrO{=>7o3o_MTF7RzT^k|v@7&Smajn-c5fB4P}
zm>&z>IqK8DC6gJ1)I}E>=q?!l?)N!BPJ^x-p}Z80*Y$d6!F=fBsjU8cd+In*)HJm5
z5tG_9tcBm-dIAT~Hd5Xk8AxAJf@h;LbOR&?;l?-xE?l@)=`+Gi8Q*7VzKb1Qk%Z59
zlRv|r8P&18Tc#J=Yb8gC?ecTzCsF6&ub6H;KB;9$oEG{{5kCh1z8*9};y(n|KSF}y
zn-Tp?=FnJTUdCL!B2D<Lo5b|cgy-(FQOT68(-flas-f9SJjH5V&~CVDc_R#ur}=S9
zd#v2B&<m6RR$}UWEH1UtV%tPQ<;<m-j#JN*C>A25kk+rnWnL<ciwz2c`g~{i#SV?M
zcQCDbz|NQ?PNQRWaYU^8B|OFge~S%qn5c@OCNDRLXfwgG^a&8rbK?8}rHW7782dwi
z&tiJ(yIjJ`h1#zu$l>{}F~UfqtpSjPD8aA_dXv^`i!waU_wmw-7KaU=<5AaP^U>Ah
z%b}ewngRJ%{7tu-OrrQU3M1AaI;h31v>)<PIFq3)+oD7t;W2>(F^udDU8va5t<J^0
zH3hLc<skbD-x1qsWGkDHix8=S&C17It&P{ntXKPb)W&hzBR}=zZRL&Gf)ao@!W4w_
zW*dvOXp2T(cTxI223h^r&ZmC-$MHe8#nu8z`l^*ngO>_Drq1F}`YNQb0gt2mi$!SY
zM;WnJ(5!;|7nmi|Y!StbUJk`!FBZ&9=@Tn_2MMZGr<PTYS>GrFp%PRyl0^BJ><F
zsCB^5E+*V6n$XqURPF|Az67Aup%WF>Q`R32&x<>ddU<a%_6RHe7%ivnEZ9CUT(fFp
z3HK<!M%rw704L2MovUV=8&+?N9Rr?VuKef3n(TlXpa1SFbBVFR?R8_xdVmOhK)nis
zppKtrnntOLJ*$T1q2h0t(088Gd|+{2=|h9pvH`|-d9FT9+R^ODG6lftS(2#Mi1+)d
z#0h+R7VFm?j-6ak9^U>$5swS#+;RY%DKFL_3PCTH{>+c{bs|NJywr7{WRI8XGl2Tz
zSg1oKlP#oqX5`Y6b+-GXqb18rlw~%>JN=L`gY%8_g>iR7(iq`!`k7H2R7wieY8zkT
z_%-ZL<K-V8wsz5r08&8ADYx0OTt2<!Igxk4rPuX75NYTaMbgn6CUe!_7{0N(%zGM3
za~v^_S(riRsBWs|gMs~#t+2ZFGwdBydj>&?UUm3g3=Y+XJ8TT;H;7%@rl0Y8XAsKt
zLpfrUO6KLjWs2K&bSu76JhMa0mgM}bRWfqkcy7xo2obz=_H96QVzscX#ZlQ#*6um;
z3ukD_pgg;h|BaFc^$1m;5T?Qdx0##mMBU|NZd;esSH55UJZ!iIJMv9znyb}CqLQ0V
zAWRR<kNy3pV8bBr<B{_-!@Hr26ZX2YJ~1-!qdO5arPcD~q~p`cRpImc^%-;QmQJJk
zM6p>;bLOctlL7#Jui-kI`sPI2{HXgmL1(4%)S1Z3tyP_8Tjezuv(2sGbJq1x2>5K;
zPMmNU#c-I>Dxkq1dB}P5w<YZCIFk|P*AsEbsnzXdN>NkE0G<!-03B-k(2rmHlGBQA
zs}<NLh_uTAGj82xjoJ#z%VGl==NWiU=rFIbj4K8eAtFG=m1F4-GcJj6)r3PdY)^L(
zUw~~jRDO$`iA8QM{mjZr46J7tiZ6+^NT^BZri_UMHBfH5-FxYV42?@_<U#~w8o1(b
zF2Y7aDGF$AZN!o!FjPnJa}!`akz{&NqFOZHB+P?BM2ygOGc^KId|1bt0d>Zcgi5hh
z%=u)DEoZ>hQlO>l+HE^N+V_U_aiYoTnR+J(R@0@z$gz%j{eVX9^x9sN8Og#e{KBQo
z<qv)!Y_ah^XtG@)x-9-(VPI{r*_>TJ)50Bre;)>GaumT)qK&{cV!zA8HsLmZR0?bV
z#m)7ApdWN&v0eKSKB?kdORO6VLH1gDnEnOg>jt20nNz2O&~2Z3YVK#t=&)OQ$!J`k
zR~dpE_Sk9<y23oS6|z4j>}hE0>q*=(gQf|DNZ3g%;6Tc}qX*plq~IR?B(qpt4#CrX
zW@(YX@z>JVl6*ZT4w&xWox<6XvOlDs)-uZ&DU9T#|5@?NzZS0>&06L0)XCmuf}7ZJ
zSq0p(;!q$+>lTp?EMoF*`^Q;Nw|>p|{n6RERNYVX$!))u?P-%VL+cf*7f+Nja33!8
zedKP5sJ!$^1YwOPq8?KEuevcB>s}%+-XRzMx`Oa7_+4DH!787-CVdAE`BT~!5Asg?
ztT{c6;ByA%_Fn;EIJjU}Rll2Wf2(Jz0=`koHEcw{#_BYnj?Bb<?444&Mz)f#k+TP4
zJOXh|CRhXX9_`pd31v?>Dp4s~Tfk866^-?wJxTY!FB1h+a@~tOOLSPPPK#G6`?4d*
zr%QXfZ&bThL48twKxREr{I^^O3AW6iD%0rbS;A@!NN`$$I2JLq`P~qy0=sVj2&{$@
z2*>jL`JJc6k}A)^Hbl-;2L$vMw#=ZXJcKDh7od3Pw$y3h3nh{a8%SgK3~P&S_bdM2
zdA}TtANvEZqgBP0@HHrl#aOlV@36ydD#C@qs!R&*LC&c7?A7(?_%~giTt2tk??j23
ztZ73aTzeblXUP!(*1RE@jMNzmuf0sUq)?F{-Jl^=Hdnkt3CN-Dp!k^?g9uqL03;<k
zxe71OSJJdujkg|dWA~htbC<x@RDid=-mw!85d&N?1@0W38*%bS6eBSHwXQryN_zjG
zP2TAk6YfB3ohC`~CX3B^aj*qMg7QDxCrKM}FUb(|$}iu5E-TBeIy*N{k|dt_7ICRu
z{Bpric9LHUrhNxn&$KnlgtAGnv`M@>VYHIBqbd@iV3d58o$!8MZ$y~fC0X(khZ2W0
zi*M<Pd9KfV+@eFX;BEnmj@qskx1}zZ3(S3gokLTRRuNa#23zJ~Yh!qHa5^MLc8L>-
z5_^7)d+~|`oLx#g`}@LWx=v8{v1zz1OYNRjg|C@v#>02HoIJED@vD;|Qb1yf*Cse0
zJ<J-9^o6;h_Njt(!L;vRPz~MNtK}bT=@ZE3@UY!`c62xPp*gc$*O%^f%5(8>@vjJ=
z+Ut@|LG%F|L~MOHw>y^i`J}~qY>c*{T6(=FKll*f2Rr#9e&QK#r=uV=M>&%7%IC~_
zriJT&I6`6#HBJPsAU{ypcXbbydW)snk3Vn26UCMFPLTbOE8BaSacEh^xR#GcE5-$F
zwoCmSw5FrDFQ6PCx8|`xF0s_h%R41fGG?Tr7rIhC*r17E>^7eV29v=H(*pA3m-($8
zRFwjlBttT<&++*%`$r;(?7O_d@6*q+`?BF_keQLtzvcVr6go~Z`1v1dHuia-Pu}m^
z`I%(xri|f7QnD5RP+6}_*($7k0@Wwv>`Gt<x{>J%s#f2=9OTIxPq0sbhrwuX`7%0J
zNdE>}%)%9i=c35k4)K72R~P9s9XnnYEdsnJETzztxRopw#0)K$)wmUfn^NS<d?e-e
zV%>W9zPI;{jK=BDXses9t(k{p>vC8t6!@K&ikmtYZ&Wj!Wog&qFG)v+hmXBFtga9d
z@1&w3G%<XjN5wkiRRSoFR1`B|rz>IezRQ^}#*?bhyCiBEu*h=0|4>Er^xiQs{{oDo
zOL73;%Z04@%Z-YNf7@i|4ej6%WYxfw(Es51SSaDb6oMftgm1knii6zQGN_r3Whk$3
z_NmM_dB`j6V2)xA*%6b{)0@4d2wpBbwz0Z<;SJA{VnV;f9d1$E3m7ib%e?U=9|)`f
zcii`?x&fcbYfS)cqHZqZ)j_x+4uEq_ccfDF74-W3iuEc+i^u$<IN#m90O(>XTY2VV
z=e#FAejim%(Q}0^;dJV5ZtNBU9v?GMzl;HO*m$l_BWu{kXXp5Lg9s*TFjyb&Hs8rX
zqJI><_YXffv+@nc)g6DM;=d}CoJ|;MtGwwys0-^$9OelgcHRWD^F`PM4UprPhWN4I
zd_l@Vj;`>~#T*G2bj_9VhFoo8IGU9i7?dlVf<M*>ErPWN9+o!1Acp5r2^Yz`#viV_
z3gsAfqg^~x#KbGb&L)jVb<6d>$V8F?bJ(T`G%_T3KZ&*cF%?iOQuZh=V!ZM{;ahL5
z_>p2g@$p)~lX7{TDIJ_W07iVR-8#GrtOaS<Z(}~T!{xad!aM_hJ7f-)13}`AhXXB#
z;;!DfwtPW3vmT`k&BZ78x7sYQZ42g-2#)xWgrS8J=k_$ba!aGkN}zO!p-_M>=X(XT
z{k^RE$6PbWQnvt5EXex+UN*5au6DQxCk9^wbo2rcLzqDDCxYj^0qA%=yLsB5H`lWL
zw$aZ+M*9x@f^yH5vPXU=2_)C7;vn9X$s|Mszt*$-HzlA64jSX62go=Vcw`<K<Q$9l
z4#$4-$t0TAX!fi=G7YsFmDVe#_{NCH{AU#F5rRbWJ&An4bPkh?scs{S1-Cia63t@6
zA|XXcY#;B!FPOh@0X;JfAc+HmdX-8bk93eyxjg+di0noKp`c{)y(Bh?Tm9O8-<05r
zze1%4Bv1@v!?2GF@(n~i!ygb0o054%sM^1|CMdw=ZAKR$LcCkP)|#e>Zin%5(N;pR
zY(|SUr;{4Et2KS+U3I$Y#9A3cg2DKLVr5zxfF6voCSQCE02(c=%6D$dZpDtbZ*GNH
z9I2H=FcAnj^(t^&3bd3#B~}`hp<5`@%ZQr9>)<~R`E0Y_ht(=W&yw7PAK1ur!l}nB
z83%rWPC#YR^+SM85Q8T(4g4x0B-be55n0RU*)m)+2wawv8t_1=P8>j=lGwm*H4s~{
z8^8%`zCv#Y3{(SGnMWwn`4D5wSgq56t%umKsZb=gK$1<V4f(8_C_*$X2al@>-*k{E
zX?Qwhs(9Lji)k;OyAj91-BTGt*9g6~A8$u@C#}!-5?b1X^MZ(&9#02~JRyPm>TPf1
zBT*i2#M6|UX$>$ex;YCPA9#88=}<?uA6B>}JhCeQey?$tL#=DRE6g~jwXm6U#)*n`
zTt+i5`G|;7%|XN~5*PhkfT7Dc%Uh@t)S5Bl40}uwj}VRo8^Dfjdcykl`PBD~Dw`M2
zQLJFuGWKj2ef=2sqd1z+FT#_8E$VDfZ$|k~rs}!VOnTlVbY=yY|0~QWCWS#rqbOIo
zN>&2gUx)yZE`$(u!Jvu_;{08pfT|-<G1)axs0Ja1uaNwe6i}&tUP=L<*hT^WCX+`-
z56WnM7yV0*7}eJp?QUf+Ms)Xkd|XMf?ZpXr#$F<ZZdj=X$#F_^oOH*F<b=#MY(%%t
z-Mvl=k-@$y^v7OG#a=HgF~8pz&E{s>ycYl|eaXu?%$WCZl-s&oAS)S!ne6s<^P60o
zhUc^%Mt3?>kug3{d8XdDa)KbMlGp0xmS<_wO-0-ko8yz}SW+Jk7L&6q2$H(xmayI6
zsc{d@uIgvVE!pl7tojh?Lu(QjeuTrt>M1*n7Le3s+V`8$c6wgM_%F?uwL4zZVi91S
zqUMlC05wo^Szt?#4O@irQ`@Em{QI~3$bA%6k0^d)94_=Jj|IEoKW_4dipll=un|Jd
z8UYy3LgX<_EB-;p_U>7e&%pB6-yv!J?78dr8EWDzm_TnB_otkY9c8+Ug?E}(MH6kt
zzky6CcK_QMVew;@qesTCk^tpm19Gu{POS~OYe^^Bi)o#I_a8$GmWmQo)2_TVvaChc
z$Wg7+x5-%7n*X!QwK)F4_{uwb1~a7~hF10LNJG6}@b;twj3;H9*k1o^e<>_-YfkiL
zV8x}%HtF5c%_bfJh^HnN!t&wHH%vQ!*Avxd7?P7M$^E1F=pt%bU;#WDHBk%GN)EYf
ze3N}+zesf1&86{=UNwkYq2U9+$E~-*u_S#c;<hQm`i{8$zA8{M1biF&K4<W*Fn01g
zFNRsPspjmsvUb@jhV4A78jt0V!bf2R#jdpO-|bs4+q}<(oBCB(tI(=T3;ku?lj)ve
zt&c004Zq18M{zTj00z*)&MjJK?b$_3V1V0H`Dsw&L1nB=BR0FxQ^^odJost2Fv4Z<
z+~>2gHur@kvs5p@wWsPcXK(7vNF!(d4b1EKXOiZO%#rR5g6=Kvs~Y4MKGSNB)KX>z
zahQ&<L`b508AkvH892h7SODM19yw7K<Z&@_)?rW~)9*DAS_fc?PKp2jpsiRc+#hJm
zkofE6I^1{eY(DFO4(p>-{Y~SXUHrRa;&wo2^^NzR@zpD0pF33F5E+C8<yIPgbcM_v
z6zK@uP=mIns6g`bRr-<Omz?M40RGDd;_p^GvA^uA0zEr`ZmLTcn~e1uEnqpg8-KU2
z;aUKrqvgE<7zSw`B^Y5hl({yW&iUi30_grCTTRD=&2@gWI!zR%)V{;7sZR+s{Pu@D
ziX*0c<Ygzp8(NZ2>*Jok>%j<}A9c7XuP<?f>Be`C_ZdYhil?Tj(AQ=eo;SKhBVBQ2
zO`LpYVNDYB-wV5Z-n2R`FeCNybTxQ3B5?o(Nib@dgID~%wK2KWvb{gt5wNEpfj3x2
ze`TyTKsQ!Ya++P2TMMc<4y!xY{Y>0c^Ue!0ZL#mx+*gYjn+J?Xrs#q&2x!eZ#s4hq
z_?d>Due`p(zgip0aJ;?!l1ZbW!86}UDcS3X+2i<=9zu};?sc)ruw3eN)6n@|d~65^
zx#Q-`7;5;V&l4mg)z27X0Mq_L{@Scs+k@t_`9uA`Zz)fjb;<|w5Y5=;o;`6d>#u1Y
z)O88Sg~6X!k>oasL!^4vE%Kc~=t8iBX=|XHx{T;H{A(%g&+AA>4<Nwd&O&}02uhMa
zeHHn7`+^(A!?$yJXOo{!k+(ac(0%}j$mMR)lZWNpjg^j_Gr~F5;E-jvTWoNvM23OX
zaOf<VjWamor9)2mT~j)`+W%c6*(Ow;%n`+#e_?Em?*T$0mF(fQ_02gU&PF=vCQ2(H
z*z-9dJ4}Gdj&Ntx_8j2UXI-}0CaZ6~SS9o9FmqB=HIbGPl;_pqWXn{a5N!f@$ywEO
zHsT$+^&G0v(`#WA&y!~GsQ9)LmCQ;@P5H2#6wLUw_^6#U(>C`1f}EXDlKC5y&a%h&
z$*P7rucSy((I(O4D$1w&Pho7CB1A`xpXMm;;=M&k3&ZsoMDqllt&H<2{;MgnaOsPi
zYE>94u=qM1l)o%zJnDaF#{YQ6Rauth;HZ?b(PsK1Uo*CF;WMun$(Cr`53sJoe-KG3
z%U(T>%NA^XSO7JK%$SqWamam8KTEgS_nhRlFb0-8xH~q2ZvB0I9Nk&iG3A<Tb@`A&
ztg)2kXAmxO^W{fr(~M=vh%z<$9Mu<&rV<aBrntBWGsIY}j*LQUK+7T#`<EP@Duu7r
zrc9z>Ma!c|#Q+}$x4vLoiU8>zc}N|!@$pr?lLBne{YTT#Vo(atbMBRdwFTeP9<tt-
zP%l#-tRucucrOb*HwGy-RPU(kP=#$9IoOf6T6Iq!eaJJ$K^14By;a1U(x}T8axUt?
ztTW6NCR-x(hEx6nz^UekV#xZBVwl_hzZ3(%d8x^mv(aKbq#7$1RIUsoUu1>*Krg^j
zD7Iz#^*?GMxN=>o%HY*8=Q{Zp`HK)QvqFp_y3U`M<XuK_zvXCs&z5H1@tAb$nwuVm
zJbb(6KMpe5p~QEA;D<8>4DVN|!}|Yx(nfg_(EtGZ%gm5@9!t~3T>`Rt9B7>dgV4#y
zLPq`KX_-2&K}A!I_X(+?CBD+5zPGM+O{E=hl(XUxtzf)*7cW7DI%$kkL?1vE6Uuu)
z$fu}S`LSfEJqt>oY=@f__JzGr0o07R<~JvXknTe*;bTXx2a>dfmeQG4v?qlgNFP?a
z2*BG#6nxyNQ{lc^qiWVnA$$vS*7n9i{hFoaIvu@v48~b00v^u>oD8i=AFW$@%&E@s
zu5k~JFZQc2MANSL$|IwkXlFv+z+cz~uF(4VV0}Ok^9ADe$|lZ(hmePGAKrqa@k4I*
zWrt?MELgl+64Ks7S>!(?<nGUy`hFS#U|(YPL9egUZU~CnEyCUOX;iiXmP&9qKGqDu
z#kD!O7@AJ>K>$|>C+zvl=orz`xXBVDYqW^t2y5d~*p26!R=JA&oDc3%vR6#-aq`;~
zcr3y=F?PIR>Q-&jy^2+(%pkpUJqdwvy!@sMNpG!K4{7s)1Pe}`ZBTb+@Ro7`aNg|*
zHkvJi<oNDf)L~#ZEoScWCt{uXbvFjYf090fSfLtE0w_>XpXH`AlWCDyUGE`1!hk4g
zOOpJZe|0&k3i&~jso+tKBd<IpIWl=-39DqMyikmUy5yxrPP}X~{=%@r<%O?V-R}-Q
zvGQ%KzdUjAs8}0SCtbWC>`#yqz_n-*Q_)i7b0=^?Lr{^loDv%zqMVhu#Hpm=FWNZk
zeMukQ^D;P#R`GpM0Rn7>$wc~5ULJPZKi^f0T%szOT-6!#q~)d|rw3k(v^B_Jn0QIz
zq$hMT{v;k-Ur}@_dJje=|Kf{7p(a<=kKkcy#s=)tML&VanDgvKp@y_ofPTD@XsL!v
z^@z*NgEDRTkVM_V0!^Zoj5sZSMduAKshVR1+8~KUl|)rW(#rF)r*KERh!PqK9|xB6
z+1Zu&_0%;pY8OC-r!piq0%1om67R50g3czj|4vf7BCEBc@9N-fl4)E(PKUhFnPfB+
zPA$kTmwF{$H=rB<<(V%4$dblyESCOXOuch(Z%xoO8r!yQ+qP}n_Aj<=+jeqdo}AdW
zb7Cjw=K0?5-uu>6?Vg%{rfS#TGpl>8?hZYa{i&qbE!7t(i#oF7!LI+)E2l$kDaG6e
zmSOhlg~s<I!h!s=Mj3Q7i7XZe_C(mX5#wWkv<p8=P}M1=n0iG3Q#P7HI}p9Ge<G!f
z8-YEto?k>01%7PsKTsKUg7mCl?rHaiBaw$Tp*a*(j$JKkzvFGr<OW#Vt++$@lJ3;6
z<`E5VXChuLADnT}<E!Yci&}#P7wQ?KXr-I<J_LQW#VX;Gr*hy;iT$esgAgZ4C(WHM
z_R{6t!zsOF<WC)^zKt-w%NH5B`8pSpY+)7p0(a!;-6PaNS{!$TN6HO3|9#f#b=yw|
zykWb2RS9GTvLi7J*9`E&mf9oh7{L>Kq%WQ7*n4QtNu9{|&=BeV*s>}=dS&%acJ6r_
zpzY9&3<WZ_RfHBA;t`e#l%VGJFTkM@f=9V8yI6GtBGK!b?aeJ9yHprkE+O_!PW2!w
z#cbe%2PQ&uCK4y>I4U;;4=0k6gOqN8%22Lxu|Bwem({TnkpXP4)Z|IS!>N$eDr-=-
zA2~<K=+yqjJZDZ`2qom-Iq+D)536yna!_SN03mjI?)VKR#;qtby7lh|I7#}EiC}gi
z#YkXscTaGJ>70=$VkFFoGYlyFJc_825)6JU$<O|UF@6!kfebEgNsj3hKWjk$vt3CH
zUp)YR@{~ldYx{oBNlpbb30h(6-f{doJDs|irzRs|3LTnCRTC1_e<TLl@;U2AjmLCT
z$IG@6Z-KMk&_xJ)z+ecdS6~ae$kC#Lu0cW_t17T138MV75Du1Og@f$#e>vYGIH>`<
z-e`&B<FNQ@2T|o7!)q@`0YtY&Vj6%i7LRSx{~R_9p28zK2u-(QDZKOWXg3t(kG-Yw
zVhFAkf8Y_J5UJ)t@4HE_=YaX`mmyo8UG&W_x##pM8*pqAo~s618RnYl$_=u<0%eZ>
zkuFJ)Y^4+u^T0)PrbB(~+Z^9#o80+#X=z%NI<*?0`r#?-GpPX<c>aI_L|+05w9#o^
zY16IJgQ<wnV8Af?aA3}=ul1NVlXY4j!koKQW2zHARsTRs#(}UcJM(*9W1bIg?wHE0
zJa!)MZi0hmh==Kb+5WAWC1cmGCIeqU;k~q#xG6_lzt9uQYVx3agjaUY<3dF9Q>|}o
zin37_`A1OA<89<uA*2%y5IC51W1sejv{pK(Ojr^UZskZwR6u$=D7V+au;<O6$ciik
z0D+)$1t$TLo4G23N+?DWBy6_n-pn0(OqeZX<x)gTo7vm8w6U&NlM8$W`XK9uV=e0H
zE-vjPQ24=XPk+P+6D*)n18JvxE9%`o>D+in6M)LE!M|qYTrjPGqDF-Zl%eD63>#JM
zf^dXiNb3d$az=D#;f#t3D=7M0FV&Fj<Zh+vw|`)u6^3dsh<QSV)=uMdY-;)~xRvZ9
zFbvZ-6hxi#VW|JC45{b;Zz)%`zzzd=aGHzrd^b@X#pKbrwi>wAPCjLpqXY~V&xLp;
zW|0Bl@xs2g1e1JZ(e&vdd3H5J^GJ<2u6v=^W~uu0UbZPEn6^qlopDkfmgA*M-7%Lv
z_5ErGOvjbt+v)sUTU`0#8-EG0V;p5hctsQD7xr%KVb;GroJGkoAF@?(62b?Re)b+C
zICuV!QXPqS|KROAKu8tm{{mEi{|`~cR=~rTH$3O=N=Ny>qD(EZ{9ja+lQaFdy6x+t
zN+~p-QqJfC+(qtB^=ZZwOAR85Vx1mO&6T5ttTTm9Z;%{dh6u0Cg(=`7ALYrQ;#TF_
zZ49;geW`t!AZ%an4{902hu)PR_}3<DFXiLjhyj1$h$6rUQPzuyBc7l+GH=NgOmWvG
z4+no)8CE|=drCNj3>l6(1968m94e4_9^Gdej#h$g-J0!f>an#-4Q~W%Bs1#l&FAzf
z28Ki$NE~HtyBnLzzr6O}$#JfiH<h%bc}i(qYiui@>~zMep^C^fA0~!1wzW5<b=SFP
zoh&%JBo|<9Ckai1Y@mLjO7_@=k;U^kOv_ABWKG;1@JnwuMLoTcD(X6RkyK`(7)zq?
z(~(L>DYZ}}K~GDKfzKf!0fyhj*txS&H~r<Ya6Ms(=C)@dEs6FohNi?eRu_xQvAqz|
ze@0yFdu)0>45E_LUNvHfn3uOJo%9#pu+c9Y0IO@xDLf5bwTDDvk~FVaALlPsF-kE?
z!aTci6QIVL=d2A6xoL~!?~4N+X##O3P*pTE<Ra_*5p|+ukvvSU6`j*MSZueNFqza0
zk*#Ar26~uZtxP6ArA>dS;*ZSC&3zLr*!|D}3Dj@h)-$P}MC0q$cZVLFD(-oTq~5q;
zfOEX5(Jzy0p<w>Fkoa>WF;^`H$6vgdR)1~FLhh5mgk+b6x~UTHhW@o8>)4FtNC?B*
zSS^ZMxe7t}R$(9_AXcMXkt@Qr89fdEwy|!y9dSeganSnl<`I=(;68NAz0uy&kDAdd
zfCcqC{(3Ow>Al9KBE4Ee^k14nM9#8d1EziYaP-6`EM1P8pR|efo@p{leTouQj~g^v
z6tFBQ3vgx<rp$xxVt%dFC9D>toh-9aI)6~u6}~``q!{IrMM~`Ri&IE(-q0WP?_W!g
z;mZ!`gpmcao^c|HOKY2K9gQUcL;4dvHuZdmXTUn$WGd_%{*)MSe5~l&<N7K63}AK4
zm1XLn<bTjt$?<nr?I!<6gz6*f<cga~@Di@2wp`J!33W6dmNy|+t&gSiAi8KL05+B<
zW#REv#SFdx6;GDKBg!=xhp}nD2Sz>`f%^WUWZGc_S8NJF!~?r$RgGlj09&1#BZ=*4
z#~qIqt9nBg0%6ODxpxIo>HiBnGGiYV7#F}p3KiCs2VlmaC<DD<XMn0F!KhW3d=Po8
zS}w_#GV(0<dE0Gm$*M2EoO{gfc#6REdk5+D`g7#4J`9%y<?C<~z@TWi=z`&qnZ5$z
zdg{^hyH$umcoM`ksvyUaTiiZ7MY_$+sj(`Fu>l1c?vrg{BL&qUd+>J|v=;@xk3T>v
zh>c9!f~^bje7IjYAGSo#m;sJfrm0yI6FJ;~)mOU6>z+8dj!AaRn<)YTxc^A>;{OE8
zpaE;LAA7UDh+c;d@UCmVMAh<gSRSyX4(qds+z5OhhU`W;PJ%XYA1_z%!%@tFbcr<)
zMR}Cr(r_}oMj;(p<Yn$_uBw;9fTZsB*ksi@7P<#(lAkN!9;9RMxJ<4ND`F*WSMwQl
zJc!+xaUvQ>hpW|L@aITfeiyww3S669rW{pWbv?T3)()9uFWspW#X;>-s;_6vE}bIs
zbS_$|_X;#BzFMu(RCn=E{;J)}r2M0-N$?Pg=k97JnY^wzr1BH2aUYBYDA~;a!EBRM
zq^}mZS7hE83{vN6gt?ASUB?F@OCav-fOOEkMs(Yt?v)~e?GQX=o_PO3)s6Ej6PFIj
zcl|0yMC$XzsUODh@PQCn>|fks_h88^@smAOBlu^9Q}s2oS3b8kfACpOMfhP~4nObt
zn%V%1l?}~jLY6MpbaqLA#Vfnm%1@VSeIVU?ycRz=T*H75J*=eN3E6!11bXz@=(hLT
z9f})fW*nVj`Xsqe+QBeKG6_r7am1U~-?PSWNAnQ+*)6KYECj6Xn<iIb!5IZOaJ^WI
zvgUG#W$PPuA|lpfZWMVq{HKHVJ0a1|Ai?6HWF?{g9X+K&-2yrQqltj%q*Vc1otV%b
z&RjA_nT`%`#<Ku9wVnYk-fQJ)UkE+Gj;d=E=JI%G6`%WiNrI;m`r-4Ds6abMwA93S
zc@KuGL28e;H}vO|DFpg&<Xgs5>GHY%_-^8NaUu|V_@I8(nDtDQJ+KYZ&7mkK<Keu%
zy-CDO8draKkxc-Nd8@idlsKg9OwUSB3oLY%+aSY<4#y;^ym^C`L@6?|b$P7;@Z8ba
zo)it6bWY_99-(S<;YYO}?6O)U5pO6FC9wcw!^<yAda?ur-;cBqvU%f(&+`4xVp|>m
z<tlU1uj1p)O}!o4TPL~L=MWq7kBf(5L^;pTnWi7v`Zb`bBa4JA2IAJChw`<W=@hqe
z+(e_}@a=MosEGMWRo3=lfw@?<Z5S!*Ka!tuA5V({&0o9vZla!a^co2ip`ugHm}4H7
z+i?o~PJ{w>up%w9_9KW}4ny0{r8KQIzDKr{{nr9mNISHsqzVTbcFXZ{65nABTy2D_
zyMe@FJAlgcw-`Bu|GgHb;^V`5){o*nb9^T1%Lw5Kp{eqPn=>Moa{=}d5-Bii^5iuo
zD~lY_LC#F%rqgO4mPKc(qqIi5U9|P)DB2>W==z!Mye5HQI3$NcHxr&bs&7Y#6CZ?n
zaWRXTQQ+xQ!G8tpj6Na@VY38HE=?G0g90RN7=T<E3)NZ!#g=Eg{Cr4BY-hd&Oaf-g
zgl=1*G)CVf&e3#F43R=0hBJ+J$UyRRRvB;fz-uYpa<1t%ukwsw<BRUS=Q~8aY_*~@
zGd*^kwov;ePNR&QiW=7Bykdd91%cUH&6QFbhT%~%<e*k5I}0932D8K_ZIqCicOR?h
z3Gl57okb43VeW6-zrl6X01{iH%W2QMGD{tB_zj1}iRL_<z`<7SWDD(u5VmUKm(Mg9
z{(!6V4Y@#sc;R{R(tpv}c|lCkn-$W2`29e^mP)+TEEnSzt|uj+O`%G*9oh83m<Vk)
z{^NWWl9(0Rx{^U-TWPKK%V@27-Twlk0C+Nse%9MRLwX)7+_E|#hUs=7uO88OM(mvI
z*!<}Gaa1+e`NIdpL+sBoBq@Y7z*el>JPqxXei>yP*g@egNpe;;7=^rR8^pk0%qNA%
zxBeG!g2cDB!YZ9+LK;X4fAH!!^>P|_J(M#j^(ghA9{7^6IbF;cEp&~|utU8CZ2eh=
zTQ0g`Q~590&WL*e+v%~9rC&-`K~NOq8Ll)WS(rH(gsGj=&3~Y;38rKfRkhpwMkXXc
z<#K9&X#(PWrc%|ML3Rsj8J(@QG1%wZieUPS(xmHue1rW&x>owz+UBOhz$wm72i|H?
zo^1X(b>wnssRRx$o1X#r@9{JI*oz`a*Ie|;4|e=c#P9Ms27XJF6FJ$s_(@+5sfPVr
zKgKgA9w^cvmIW@uLr*lfCKimM72f^?R|)a4BwsgzR-|D3;h+0%)g>sNdia@W>;2W(
z1C$evVio)rIY&?h@>qVgit3=GV!WdQtqvF?5ym{%kWwT7CscB=N}aLNz&c{!lCN5g
zAh6_xh=E8XJk2K8EuolNrSP&tsOdO512CRp3BYDZN6rQ})01+-xgDcLA)wbvAMiHy
zyUSx`GMM+30x?b0KoPsnC5Q5(OZV&*cb*X~2lX0_Cu+eOLR+etuGD|FaKDZ1;pPbQ
zF4248`>YaBd!=r9Tc*cn#1j}Pd7q5W(8PWBjGUyIOd<sC`Ujbzcz~S0ZcO=Dx@U?M
zsB>QQ32@~P?Yc3w@NAeRs?xD%5Ukxh!i^3H(?(gn+)KUQN>PiSCNBYg#38FE>)q;o
z`@8%28;T4Y`Zo`XKMf_Yy;XU^&G|X{WZaPfWY-s<&AW^w=fO>H&JY>CCo~*;B{{LE
zG;hZ57k}StfbWWp2u{tb%hR>(NXCHqt+<v(K(CigZt+XM1`i!!4VSlznzZ_5ARE^m
z*Ih%d*Vz7l6Dk`N<~x(oI{jQjy?!nCH)W!k--rj)b_`<ErdGw3ZTdErA&6#}1N=$<
z1^}im22XKK=<3Fjur|fGqSf<3Ce{X3Y4I!N3}{w^Qc-!LAc$G(h@rnLAs`>es9VJ2
zJSN+|8sDt1c8`kuE|VeE65xf*99C1GRb^_ePhAX7YmmyFyf?|03j6RU<AncJk$2GK
zO(u?V_81Rh0l+}UUL8-5=gAs%{o4Iv0f4Gd;qOSw<y4MT;>|ST2Dp1n5i*CqajJjz
zv7jZNNer7suo7=2e^QWKO0aQ$UYY`J7xNV1{9fPa6-MY~<UdD!y;XtnBZv`7m{$Vk
zFFXr0s@Hd{lbW*;%oxbg>}1|@<^)Rbi(hOPmdhFoQGsvZx+r}Ay`>(CJ5+%u*7L;I
zN*f31TO)~e6GDyD2#$Y(nN8suX;caWvFu{OLLe5k#CU#AO0=f-tdLBM2zP(?GxG3(
zaR9(Wj>c3Ccp_qn7h1x`HufHyq8s`bxku6k0&@aabQ0xv1<2^YZ(;4r9al<B0t_^~
z=0Pb!XSVUGjdpn(Hf#Gz6<f9V2&hfSP(fh`aJB{u8gu21VaP=s&9q|o)-=*(b1~>o
zmMD`YViu@2z*aUwE~H(nEC81ik5b}~0l-3Bys$e+KiSz-z3zgw*oFpxjY&EkqdN>6
z*rej`%jLK^x};V}QEGeV1jEyF$5YIuTs)jayzC9tW=z`$X&!RS^~^-x6Wf(812%W8
zY^43!tWb|p5^(A&uX*b_$!|k`7^yPT&b0B$x0^+4FDE^;AZixr!+S&`$?qcgZ~)+)
ziY>j~N@0^<A&;&O;AhsnMI|>ENYR!DGJF}lF-E>NB#{pkPC^fj2R~y{ComXw@qmi`
znZA|=^yZuK>T5H%v&4czD3~0(_cs=^lk>!C<o80Ra2pJ~M(3`$ON7jBkw1&HPqTv+
z=rkQ%C5NFpzVcEzI+fR<;EDP)G+?<08C>?wmBkQe8ACU57x`}&wpPJz(jDjS;ZHlV
z0u^>Z0L@yqtybYL1p?GhiSM@f2#J2{)G@k110Ei=rq1KZ*s4g}rG4YOPMS{5I$?3|
z_ERi5l~?xVp7s$Y#Uoo<!QC30`YBH}#(WmVfcIQ$Ph&c^7u>X|>~c43M1YFe#9~h8
zpd3AF?+VX^EtWADde^BO-6OY)<mSdiQ4@%k3I`YVRPx7218OOBW4I_={S$Ki6Z%Bh
zX32BO{@f#xclL5h+-9rZ5Sro4O+3Vs8z@w}+`W+DO8V}w{VhY6e_Bm?&~rl^v9V05
z0EXhaKG$z!CU0kHy=Azw-+<;%U^~Pxs3)Y0m*6)cuD0^9f1i)t80hF1I(se=Hw0vy
z$-e#%88$4l4pn=)kd1J`7dVvZB3cP}H0sYC)Db8r9tKlLFX2qiVjFS@UE$^;8nK%t
z0uQ7egXWMpo9?^XibsZUy!sxJ@m5#ubLImmG+L+aa0OxJav*IOC4dz3Mk=@10K$Wh
zR|-vFUO0$)JC$IocvH8Tt3+zfR;rhlZ3p1?4g=^b20BL4#k6Rlv%OzZ)ESi93}oJE
zf_`TzwobIJ{mpt~en-XSsg$7hT%mZ42`M$~7h5pCvd_?;g?=4+-IFPn&)3|DUH)<$
z-jpE6`~!Z&NxV;FzJRi#V=O&}_kdYmj~ZQ{3=2Z|?dzt8I>*z~D}=waPT>D?DdX=9
zX%ZgcQ!w^?3H<Qw|B8HL@1%OHAy29%@DF!Yc<d7d!Z)>)r%is*+~+SQt#Z}>VW2@r
zIAFxJQoL$rZF%5TZ2fM8WQD`LpY=A0+Y%6!8a%fa^?E9V4*=xWq)va?sEZ?BxtzZo
zaX5}y?m~>;c!dEuPrV>XXRgl%M!jrPTKe|yhR1xpbho<;{@cEv5k<T^aqYLftpS|6
z9KjoUy^({kc#CX%4C*EwFX9)k+j*Dk?2_}h$58cVK5C?OV$9Ru$Q^2@9O@GKE*(Xs
zd=l#v(7(U4{sN4Z`2?}St6sifk9(b`+ohP;CQwZBq$cxjbPzNttSHpo*dkxC5E6C5
zZo#C^Wp^=C$#cR^wDG5ch7YWH8~xUJU_qLKz%e+1{FrO2n@8%8MC!mQV=y56P5h4J
zPq~Qh0XBvJe0VfkPbC!alaY%hIi41B=yX)d*O}i`M+~^Rh4*f~@{Q24_?MOQAQXM?
zUmZE@tDI*-%NKa@@5Bpen&dA=-)6e`rJv{N#I+vpC2u-pPm(c4rqpjDmSl@VeF*GD
z&Db_ANW6`(Img*6*diZg4a0uK*L+IFy}^S5BSEAN1yjv^`Gb&oQ6-9avWoFEopT}a
z<j<$Ks|x@idCSqOnLd8)!iRVNx4xn6u4OSFI-}ZtqCw+2wRX%&OwIa`lx$i4VQhH*
zBAhzDn!D1nHt`F2Pl9)}dL9^OhNL>=f2$ZO?r<$3)qfm<Nzw2rL{`7&-T#q^|E`m$
z(?qU&@u<y{do=XEy4YM@&t}re<cPsx`zTYx1pGneqP5B~7Bs%VR!dmyxT1&a2eT1s
zXa>Wm+$p0uo2bMpGHC^kM#h0oqWW(h8l4m!oiyuE8EX<cHatTNcDa0o)>T%Qxw@`O
zHBZ<fg1ZQ}l6EH88vEwKzqt6z3-~OnIxl0=p#JtCmSmc@aP=n)de7wwXCR990!;-7
zl3qA3Tw97QFB$mlcLlwyu2^1qrONjz=Z$`b#jO@`-vlS?kkBo$nL)j~NI=D`#|3}>
zTs#Fo)e*KD3&!AVJ?k+()v($253G3`R6Gh9j~6liwNr7<EkG_6Z|6ZgPS1U0Cfo^C
zmB)>zxRmBYWJ{=mZC>aM0y*?gEvh0Q?1me0-6TI!ka79!?8N@!udX*ZU5ZazYp=zT
z(jrPb{E@>^vgGh}lyO5YW%kT)$U-OIK)jUt(;k3jCJz<-LfcHk(aZHoPmG-SI%jRI
zn|za4aD$0MA^=^+gQ!paeld6|p?`f_1pM8?(^|Jef~(jKY5J!onRJU!!14s(>pg(>
z!C!pR)Uf4Y_powwO}C-nAASc0GF1FZwj@BqK5qD;x1lET9%V;J4stZi_;d+;TS?4#
zfP;XyjLB%R0A5GTc=U1V^D><E%69fB-h<AHVUO3$?RA4v9~$W~Oh_Atip%zY58gt%
zH#Qoz0ua{hdw{*V8!QvJQ=mnL`~gtpU!3=r+5RqnxWWpD5-cbv5QhW}5HPH~CTrn6
zTRiA{p(q>Mv@a8E<3yu~yonN3AgS?)aj4QbAZm_TvC5pi&8DYl7~?>wSb9*?_*NaD
znUoBj4?f-o4CS_m%H*F&5gC<0Ik!tm-_4#h&WfB0G2j=;uY9jEB$;t2$k1V~FWN$e
ze*8@uWK!O+B<F7^AfcqsBzs*7l#^PW=+ARlC;i4p*&x4ijy${6-1e1<j_&<d`-_Fm
zAdC3c0FLozaSFb}_~?Sp68kv+TNe4z1XcO(vuAJUxW1f2a^&{-d)ExB_;h_R6dey_
zEL;c(9e@Y2dX40Ar<DctexMT<_NqL57!wf}5}qlm3Ds}up%GZ$SSFU}ykz+Z#+ymh
z#dZSov(jG1jE}fP(TSo%Wc-AQ2569=-z4Ias3Kta@{1>^`0}_uu5N5aCDiiq6yOta
zSz^4WnqnBccr(n-P{x5b=3ompfsYpkC}&<0Ie<S?c(dPmziv56Q;}q>{z$qC+}kF<
zqCZ4>pdz%Wgrfn0;@r<C3SQ34CnjhJjCw({N_ZrkSS^SW)hS9#4&#G>!h1rMxr;!<
zdqxqX-_zdHmb(+<G1b8iGGlW67~&3&iJ1u~WgyryNg=_ZT24tpkauLXLvnN!koKSK
zMSPUJMew=mFPO#AIjw4LBeEliAvo2WwS*C^P$Lf6QJ1z-vWLKLVshqK)0mnCmEj3g
z4->H<hJ1)WlPL)$GEaMjdb^5nW2ok?kEoIwx;VHCu-5{It023?$AOm;;RtwacNc>}
zy1>)5PeS%N{)|?+z5c__Ex_v8e;W;20PeVsD?^%%^xCcdd*U#T3>gCS3NA@wYAka#
zN1Em4QExK>&#@i}SO|i&<&ya(I{2PrB9O;g+FI6Vv>)!^l*dO67v`6>Pt-+hKV`6z
zf|6&rjP<VbVSKVm2JzeQl_nvthy=$PoC}}Y{t-wR#ksY%*9`Tewa|OYWQ=LV0dhz?
z8byN+PNmv9^PcdhT{}8cpcz~vqg!EtmEwg`K-9-=554{0?4ua+$J4$n=+SEiaoRUb
zT+n>=3m<S7Ko9Y(j?IpQfZ~{1`#vOBzfv+W_BYLzkUXThYRQ9BpmAQSh2SW096Icd
zp~jsUuE^=}F>>G;Fi;Sq_3vY600_%)-JZR_&{C3>%zNEf%YeKoKA)pbGz6;UxLBS?
zE6EsX#_&Z(fjbV?*r?G6W_u>!$&6~yI2tF_?Q@y|QPPA_q7m73Kfs5l(}y0ev9;Df
zbeb6mHsJ0U-&7}lr)*tP)YJEh4@>6lwCiAMdbwoLg*-IZzumP`ZtqAGpm2MiCl;TZ
zTS|2Lhl#&X>aU4lOXCeHJRBM|G+6kvnUh=~K~UtHw-{IbO8+Xnlc&YTl3UeJcN;x3
z1B@&KEcEeFnL{^1WoC4A7nkJVXs5M1nCLlDz6H>PG|CYot_9F|O^Ok^WVjYN61nKY
zb|2J+W+h%{pOq3g9`iLDfW~YCPLgvrF4i>5^YDaFcjeuH6$$}x1s$`3fL9fTQ7?Qk
z+z{rNhkl45aqt0K4=;|Am%r^14^m{cGGz{FHmkIWvsC(KSQyvR8mMs96#9g_<540j
z3l=v6??}Auk+Scz-!ZF;4hEW!6+=9-&&7M&4cubOr$qs*zv*oYz)y4s`L4+=h16FM
zG{h+fq^I3zhdw$1KfcX4fwQ|$;A8hmfp0Nj_W)6j&dUTQvw}Hk4io8MQt4iaS`B6t
zQEug#4Id=ao~Mz3>f<zM(q{MwBumzUxMVjG(nc^v(q=7@Zf%hO&SU99OrTGX>R7rx
z6}JI70pd_yiu)E1(4qZYvk_u`#^TDHae4{qi-DNvB7B64U>MMyKH5U0Z-Xdy%ctrF
zuj)!%iekW`oBxRf#+Mo?dOl!0yr0dIV^+6)^5SOPxOGXdKXug~Pae|zqZ1=ZJ!O4B
z%K8PWq-EPZjZ>JYWK_Tu46Hxwkx>w+uCOzCa%veMx0=HTn5TYC>$)L2q<19&(Y(rK
z;S2=N7oT=D05*dovl|LepDZE?dD{hZBeb2sPgY?~_G6VEm_ePs5`<N^%59}xv^hrI
z^M5FH%w$un++?UM>@phJ9b?Nl0*fN0;fR1KL5*E`{8qW3C)G$7M9Gmk?-+_I4y9om
z9r8aqK}x&_yxP)@^_uTjntZZrP;9VJoj^?mVBcMZkybWP+|pqAfk7|BmaD=KH={#O
zkBQq)=QdU~UYV0LE<&PkgI@%h`82wx)`k|b(MGgQb(0-m+|`SiE2nyIjt<xyj3FbR
z3y;tojv@c>y<v}PVXJ~q2LDRc<8(?<x$h?YC;U^0f$WY1f5xx4kbxLy$!F1L(bb`D
z15OmMDugsmPAPhxj2=2*Y=G`m#t>vVzR?D7IX+bx>yKq%=55s7lBl2@DE#0ng<<p6
zkr4w@teL;Eewu)C4>Xysw&Ojzv<1QUrI2?R3Crr~U+JHf`{kD;!9_wcGZ{s`dfj9T
zk?!CY+6Zev!$S;$(q#EZ!@w{ppF7hpDb{}WO0AQLl!_~oSN<0|;C&6F%cBaJD#<po
zde5Uza_!z#_(oY0R?h}&?Vi;dkGBTV%CM)zzwmU4*5V1;9lit6vPdJ(8jV7mn2?_~
z-Eeqzd!pO@R$Q$HcI$~((F7-f!YJ+UA|yvOCijN){!LRI4Y2AL$9l>ZqP^Bq?vXtY
zUb@enw=ZgmoN!h!fCb!qijOh=db)zw`8XbEGlMdre62s1P*rMfz;3c*ohUp#F~-@Q
z1>MRnf9TKX6if=|_Z}vJCXjvYAY%mA8u9d96w{jvofPTqh}VeV)=pe(e}383_mxr$
zezyC0%Jd46gJyi=&+!i?Kgr!cerYQTsn!7ZhB+DN@q}Rls9h*Gc{2^-E5B#mwEL#P
zPeV+QCH{uNf4tMYg}cvC`80QHUZ|c@zm!@8iY@lju}hC8?w$^TL715jAO3cI<~_73
zK$}cC>A>J!!p^dXa+(g9{aW~aO84@wnbFApkJtR-ma1AKq=DY{-%<xI(hJ$+WFx45
zq%0-mNh$fE0BS~Zw=u<JqcHsmRVvK+r&tSVdHo6o`yLlwK_i(=SZoD7>lbTB;l()P
zY_(J7t5Va5yfxLf_$nR$>#uq1W32zC5zU#}WJzu`Z>StYjm_ljsni?Lty+iGk!NiX
zRRl=y?+NuEmTR!b>PZaHIWBLWH=1%EG|OdCo11`GAVW}R{PvV$lBCo5=D&mGld5EN
zsumxOn0xz3i9l?@rFq5${Gybevtm6VSJB{HLNG$$)ZDP;hhpU@zY%6@SiFwhxA54W
zUvk{Z<6c-a!iox?u0L~y&b-!fvDP{kMc1LeMnlaA^S>Zqdqp?Xp{d;x4NH@H`Ym>R
z7nuR$YGf8megg^Zqb-%?n@Y(uV=R|=y<EAC_`@E0`K4=%dIwlHk(lRsb-%7P5pCnD
z(lkOWlTny|6-Jj&;NG=qoZ)9<8qnI8F^cS27JGhu45l$%YF~CcuUF@%Y>4+U&#KfQ
z)!}4hU}hZEinF{moy*{~umT{v7}?U>E_?trqzZ>u@2-dUPcTX%jW5Rs&BFD9n!0z)
zr`7W`8z`x6!DKNAi{k@iB9|m>bgDhC_iaAjUuE;%3RsJN_20Y{@sQk>B_ox|K2O~;
z*%`}>@b{Ry<^!)(Hiv2rnKVw{*RnCKYm-Y0G1l;Q_Ug(CwInoqpH~;zW;t>g1_gj+
zK;ru-JL$!!`B$xapM%~pjL7E4Z!3R)VFlcyY4NXG@0H4k3Q^*TdSd?jLTzidiyiDO
zQ~T)Dcw6{BBDC_#fA#@zUy$Q02K}U)G0y%1_ber#D=kgN*UByOl}|%|yng+3S=DJM
z&@a$${pK(iJ3o@>9F(mB({+j3_Dcf<Q4W|2Oc?Gb#33+1+^X3^E~}=Q5tdq8eFX&h
zk<DDZshj3^AhP_(8Eja7H<(M+!0*4kBEvD^>2OD$c9n?rT&liSp7S`Lfwr{?U@MK!
zcP|;AZcEC@d+bgf!$>UB6qqyRoRqzc4iDQzo)m=ytlI3sRdn&8i9alrZF~U6YUVCM
z@^<BDBJM8Y%Y<%vD+!)miC;+3aC?G)L@Fa&dTVUcGl+&-1-1#k)=W<`mwe-deRwWP
zhOv;JpeNvB$NX{y8V#v46jn$7$il;wuw+C{spYgs%^37{LI3LVVT@AtfE8cIUqY~*
zX=JlTlbxY3oF0lXxrM{vRgDJ7i^dec^ZDLCOV@AVU~xe#acoR7Y9ywv7cemI3iQdL
zKqVJ?=+LFDm~m=rSjE@4JOuCf$+GA|5MauFK&D8cGf%w_e$NZrXT!l6e8XT{;cgRr
zygTH`BcBv#NzIz;GJD5n{}sq9_S3tK(`4G6hx=&Tu*sF>#+mIxS)2kOK9YS+3yVOL
z4se)~w)~l|l?xDe#HOg3kRTg*mpoeDpUb76icm*TU1^iS^HzX4ayv>6dj74?WoD)O
zQ$02CS~p_;_p!TE?+yXa8=yrU^KnNkau2MUvZldHr3%e9;?S-u?{`#+YT??YMDztX
zOE3zDh;eUlwf;8=q1-Cqa^dRLDnC~4P>>YoDllcn=^Ath)2pQ{_9{B1XYzP(@zHsT
z{3_OTU5QXk4EncRn(Ufev;qG=Z0trOtCI&H9|7<8X|wdUetGt<tk!iUwqduCkuoSd
z%3{ZPJWv@Jb8oBhQYx0e@<Sdd9eAr9!L_~irIPl>)cI)2Am+sY)AbQ&_EqY{35}5!
zjPz2uOoSQTO0jlhg}H)-_rS@L&B+KU=9#{KD^Tyb&mT&;)@)4U6X_TT_Dr9u?dzzR
zP_e3uIPSiUJ=o)9d_u+|PHBAJ3Bgd{&NXZBVL&84hgovrN>d*YEW(o%srskB1z}Gu
z3+MZ4vLqTL4-}FB;o4yALPvg<LM!`KFC!Ij5KS;5<7Ibyh^BD@_9%rm0kGCvp!BGb
zl{kfnNy`-DjdlnThB`bE$-mRbi^KRKi9S}^SsV=vf>?QU2?(&%x7_y;aJV^mFzUxy
z(GO>jd#j>8JpvX_g}uyBvLw~4k?JR04GGAaQUylwu6`>3CNl7-I1hxO2Af1yp$kj*
zh3-y@jJQk340;uF#O_|Zr32Ml6pC&U652YgTk`&yJn%9OPR1mxHl8qOO-wy~G6K!)
z?XWiu8=3jKBi;RPhiwzs7tb6I!D4S(<~!-*Ktcwzp40*@89;jrc4K7l@pBj}!vovh
z#LRVK%t0YQtea?UV_#`U%%I!Et25xQIhk@`O?BC@DEgNQW$e{_+jX)u(Xd&}YbakL
zjq+X<3PD-nG1TO}k7utgG8bY#%xsqk9(2?A5=B&HIwrN5Ugm+U%*oL>n`O2_IyD0;
z1>)uMO2ydmk1&1z+a_srx*cT}EO#TeODes<-&h*}(H&S*W~<F3=WaDa^V3Jk<h1r<
zB2r%PC(CoUfiOeeYyqb>YqNIn7ld$*kxc5Mc$9VTsCLZVOlLrYAyG81hz!vL`<Sf~
zjM0eaP;MN1eo<vtQ4REmdpk5x1r-3}@6y@vTY@ZfYz_!d<M%!1p74~@gb85X6qkbo
zxCa(6Kzi(pI=hUO{DP`^_&SnYFHtMji?93hb>uV4gjuzDkspw$+FkcAgsz1x$n1*y
zmBBERYsAXU=tQEEa<VU;4UR-7UFr$=-<w|?Q9XAHor>oL#7zw<``fJl^c3_?Df#ce
zi{UIiU$@8#y_$dKSHy*(2}akpzL-A?VFEC(>6rE4jI>X~k(vl|OE7H*CQ+7Ab4D+|
z9EC7`6;?RY>LSKE$z5HhH})r-AIqY{B5DpM$#2mVW20+^dljkZHG(YW1S}TXRZ)Y$
zAB@O-RjTo9X*#3H@zdsw!(zKjGbK#AU(-djd2mWN7SyJl=rD2v@c4*jB>o-IK?BT7
z9u^87kzqctDR^b<z2jn$f@@5&I@ah`F81LwyHipvW2$Eo&(!kDS3Kx9la_5zr;n>*
ze&0J$PG2+O6yfP&kwB-K7Pq%5W*-Oi^MR$OGzG%=RP+veCsW_rFW|&BLJ@@lk8pHK
z{4}Q;_?jqZV)k#^_t%p0szI?uQ3m)lsEqYZF+`mi5f8_I)q7_NG2}5vujS1=3#+a(
zI<&~K1yi7~_3Jl)FmYo%;X7CDE~qfV6AFkc`O)kOA)tYnMF6?5jGmBR!5MgsKkqjJ
zl=de_tT?xmd(X6<g<NgdrFyL1AgZ&X1C|M+)^DsmJPP?|x{2qAV*LL;wgDJEvZ+PZ
z6o5Q)@&?Ot0~`zCe!AHAEErHWznED9CH2fzWFrv#`KPjZ&_7dTFPxrUm-TYR<+6nU
zy?2h6L2ag{6{&ph;6R*DZ8rG!NJC8P5Q*GOWzsROv$^8z8P#@R8Zd9&V}3!V1}FY*
z=M969KtmUDwy%;0$<|>%1qNW9kCY}XI}1i)rdl5u%`(n%%1S*MIoB@fSFoe`i#m#L
zhs$I4j{1&{WFgJ^Nv@q~U$BKW5xU^K5;GfVMZKFCZ_LZU%#%VIE-DdECCH-Ca!N`6
zY=RIv=h4H5ODN;?N@{Ab_4KIV%c3SVFfJ7l=L{1*MHI*ZU3l*Xlm<W+>VY8w)T(bj
z6v!dd9=5JfM-D30<^{tgbPstItSXu?oe@{%^;^c0s3gQ!i=!I0kTZuE6bHzf{%dUi
zRsMcyi%nCpq+R1N-_6t0?a*o?I1#V}lmcx`_t|6s+%bk3aP1QAF3iEcsD%l38C<K%
zEguPTYzb`#(I}dJmkE%WEpl;6%j5tycceZT!`J>t`03a#y6NDtKyEbt4g_^t3hF*0
zUJjhlDCKHuvTSM~kjT2QNU&BdiGOy11FGT?uU~v8fPgo5FBarCpKttRrwb}`)B4PB
zDDSU1LW2wmdHmj;+_w^VQi_sY#bZCSPyShDFB&><K4Ctx3J)-n2VKHd=E&DC%sR+6
z%JbH3U;>}?DeNSGI7@PQvh)GIm}&)CI{GJ-{t29(rb_`E-eJz~mq8WdoPwyi^ge07
z#o5&nt{}f^<cs8y9!0V^TD|fhDwD(sVq>g+iCv-=sbSq<^ECysf;wNiH;0*efP50r
zA14j`Q>(<?Y8@~PvSL*!A?kj19r8OQ+-9q)%(fyt{Gt@QI-~ilm)J<-DUIR8p10@_
z>+yO^07FJn+yEV48$hY`ORf5fv@2<g>U@>#-2DvGnrpcKIrhjf`}2Q7vQ&kdWQBig
zDd*=jp{$bGd7NqoJ^-?~;s$Q_{%m;JlsdcRIE%}GeFmpq*Vzla@PU%UslmIEWpWD&
zHzop(GKuq6IumN8xXpH3_h4jg8FK_j%co8^onx_VSr;(IQ=~+%s9+%&4u?zc7V<E#
z*05lrETAaEy{Aqs>)di85(+X~v=WI@N*!mCTt1y={-)pGtPO%AB^Q~WPze%)n24*A
zGJLlHVuU|FG6TOU)%S8IC@f72Ma%X<x{6aD^*qn~);2G3930&=ja1wb<q8mydOInn
zxKm%?#<7(x8~l(qySb@ZY^oU=KN(EIc{<JYiHv(92V@(>rv9VZ<Yz1BKF7A};byS`
z_OOJEh;1<VpP1Cg_sv7nZ^Mukbx-H?o8^H2Kgz37f?&$DY9+ed07plKO?Agvcx6lX
z7j1euIk}siYQ=w}to+X@jR8n?kN8ihG8mc}<?%RtqH%@jV%gLuM|eqAnRIC^)LCn{
zx#{o0iJ`#Dbh)S?^1T?!*lL2KN;BJ>@5u>8#&i@2bU!4RJ`FLAhD_QR;9J?`m9~4T
z&5=o1g_d_H<z`~e7#pX-=yRYh@mhO&WnGb(%8}LDhMQ{E>!D4>0=}o1Q*J$y5?gh_
z{<*Jkuvy|g{LrE-t5uN@%TR{i((0+8C;hG6!#TitQVr`*d13z66T86zMFy#3a&`i%
z|NW6p!+P;n?&s8VmdTYiVBT}f>2;kq$ln<zdQI8Z3DySDV!{6KNide3z3D8{Jgjqr
zAN>(UT|LhTTcmRgTaz~Rg$cwqru}$ihQQ(3pl$syad<o5q=z5m-G0`40^>pKbN+oe
ze@tOO>FspO)_sRNb*PjH#-<Zm#QlOzuQ@UrSH-Y`xpmIUk~Rn$pq^B&NNLmBQ#Q{M
zs!Q78nAy7ov_-<{Bjo%C_lF^g_MpwB2UCg{Zv=crI+)F^kMz6opZW#}cY*@5?-
zhW0z2VnMZQ7Fra-MZiw`g5}Zeb`+P*y-J}3_U<#MvU}iTmXu#?f8e7aJs?}MZpRDh
zFu(iC_q@{Rcx0FX_`rAR?Eabc|LQVgKPq?_Vg822f!cj%vepUwyJfiZ%y3jtL4|XB
z-#Q#W0n|F4zrUs0<Aexzn^Z;he~8g|-44aRF=uxV<tis6z`BsHNk(2)-mW+C%er($
z1^oW<I{y*kfq6%Cudh?(Q9xU@2tulC|7I_Cfj@2!2m%1sKUfj2SJv4j=x=S~_eX$4
zvcq!&ko9K?%44QKAzsM;$Bp&}y%8MDYx9w2=}78*wQ08ujsMHJoHLadYn0Ac>d?P^
zgy%LC8C|Jb@sv*ZGS~;7(@|HlcZ-26y1m`Gt4rXYvL%g*2o3gje%c_9GpC<kQZj`d
z13Wv2o^R@MxJk2<?;)!eh0%hTP7h?(HCgmqhBjxv4GMU{ud`NQI`Jv^!h23i1YVO2
zlshsaVeulyQg2F&js7VCF#|KJYC#XLV=J#0EAbW^J!uyP3jEgw*Jv#bW>@gt3BQw_
z@#Da|_{5-8^`;zmau)s7B9QwPu5F|f41oEfna@jc3dt1h0kw($;)sWBW36xue!J0p
zCoSb2sX=+W{Gi|bx0BzNJ!JR!)$Z=B|M45~Xfj2~$i%Vl{h8g177yv;iZSk8yKPh;
zX57oLFVAyCWAmpgW~5l{gWD>?8zr?OJPtI|s)ccL`W_A%MrSNoI&w8`c2vxS0)U<*
zCn~|f%KS~hfT|z1i;z6fICyteCgRu>9b1vXLRXJpBO&2#16MM6@ld9&wR~uHI_aok
z)>89dicxWHt=zh>I3#*ro~m2!{9H_dh&^N*{*s5rgulyb!bUs~8S;jLOH}X%f(>uw
zHC^pBmIY*6-Zq#dm8Q@U$1Q~VD8Tz-x18Sw2;GnzZFjRNm_<40xT(g`(2{o8%5sCw
zX6`k^z@1CcU?ig`Jhwm1(fE^G3EC%(zftIkL0wp0VjIf61b;4W?x*sG@XwrdDGqHp
z&TmE3+>Mlt*a3l<#!X1v#c{)guMnrTt)NF0=O#Qgcz`8)u=_$7p5C+{5I6g{^v)P(
zGo7JNShVd8{*&1AyiWVXO5P0!$H*irLy&oE<WBt`kxORrz>ME5w7Q1kn0dd>autJL
zc?XP9EMB#!_ptM^xuJeh{<Bcj{*X150$B}K%Is2k3k>%u){dOl7(Tk7v*aeu#~wNl
zil74fTL~2-ra=fwC%~+{n{NJN3i!Y(=DAczn`C<qB7HKAl|1<y8E+F@(vOY0QeKb(
zXR%^B#?f%Ok}I@xoxy?c`Xx;5EsGMn9^TSn1t`_1BM=`r+9#7M%i;zBX*1%lw%L!T
ztb9|nrP^AEQEbvvsnqhHHN7l~25*}SY~g;dQBs0E@1q9zjHCA#R7k*RNUhWN7Q*_m
zxMIO~PlgxmJq!gD(QKj-e9DUra=vv76m{u|<cy2CC+n2!mo&s|Z+&4ml@(PJ_=;z{
z9g4)l-rd78j}zCA$iNMEoZ;lv0_4+dCGp%ZB+C}|`=f{7Uz1Y*sFvX$2L!1cmHf7C
zBUS!deF+5Y##CbV8~FD&<-*yJ|C@st7L;BQO;jL+`SdCGjs$_ehv~mNt=J^R{Ug#a
zD%nn^m7qW&`JEu=D}o1Mn>H1oD!mf?C+>WxuylX9MV0AyeA>*{Z?0K{L^ju9=(h~X
z3B4OlX&{Fyy!W5%1d*qPiN}0%Hn$n-R^RLyHf9?Dq!`1+Wo1K2fBep}H4EhGc2;l4
z;iZF$y%D9O(7p8u>zkSFFoCfU+eQk;=t^StRqY=#@|4JXvJW;0=lsCNE2Kpj0U(D`
z9`AC7UtdZ0=*uioy-AH#X1+T2@ei;$kB~O^1gCao=o|G~)05kkm%r?fglefUA4*&O
zCu6Ar`Z;1Wy6tU{({t+!c%h?^2qp-)KWd*s9`HgnP9A)l;f#2Dc&R8iEG=TCA!6;-
z6}hqdE4vTw3`qHb)C!==M2f|ByPWpzZ&u&<|9EN=TFYb)(hVP2v7vd{`}0lg$)TF3
z{{<$--J%EcZO>-q7~RkE)pP(qXu^1E%E0?yVJ>ZA`Y7bFZ@EOc4<vIwk{x5{x#SE7
zsE5FA5TGbfmi?~67Jq0CmMJkZQQ|-@y{-+_?I`R0^}PB0=|MYg3HP5s4f{k#jYOVa
zt5n*j`%1Opr@MH4fSx{Qe<b^}<HLLNM)$i;pWS~LSv*A#;E?#x0?;^&&TSH`v$=l1
z+FiDNj%`WjT76xzTAq9YfAd*?^(=9}_I2sIOUnx}t=JVvx>7Gm(gHtgEB4vsxv?G3
zFirhMg!#`wZ-t?s4a*$64O@nc`~e78#a^E&+R#U8s@eUA%bJOi4G*iiw5eH{03U|#
zV7dtdJyeP;*NVJH2e465$01}2wS=oONqz-jFaUu&(rG^IgdND9I6gSC9{wV-W8Q|b
zm^*3VM!tJr>nx+v{lT5)%SCqQg)6y#T7p<}o9`2gQ}z4U_s%+9ob;3vo0B{?noeQm
z+(xs)yV?+(q+KKAAy+`Tvhv5qn6#UQTlc8kBOPoLPHLD509Zaafxlpq%E`dk^s^1x
z;-~XrH7JlTA08jssxCn+=))n`s<~hSiQThhhf4RRY`X#-w<o7A#Fg*SD5^E6@j29U
zD^p^vv)slAIE>Bm(SSsg?Y^(Q8zX-thpS$5ddM<F8t35TPQQ1YF*6QJ_k_ZFtC#F~
z)jUL8`1PoC0r{qhW^o7&<3t$~7EhnmNt75V&_D9CTb8Y%0tC-4HOdYA#Vhqj7IJ)^
z|5xf@ozpf>MjxdVxTXv1EWb&=A*z>3Kcm_`x^wPugmNjMcw>cFIh3GHVh7VoI3g+H
z)PzJ3tC0RF%(rSAxYFMd;n{$n{%mWWwu4^^&@#3Hr>}(0eAwdagWvh@;yLWfqVe)p
z*;=MHuPzG#OM0Co@5f9>Ntc1?JGfFB4UJlg^h@kKu65}&tXfJX6usQcSR6Z6zf!{a
zE1Py$F2koJ*z@jV@HJmq5eQQ1)3vO%9tMSUUM4*^5&=`JKp1)v$`t%YRIPddM#2nG
zWxAH$niNVJKT`A$Mf#%3qnkVk=W_$9DNxtd#QsJWu=AzXlGpP0x=@T+PR@4t7_j|2
z)1*5Tm(<;J01oo+(8yPhpALOhNBjJ#M^H|6O*<49zqQceps&O|Hr5uN)wM02K0SUn
zL)31?ia;G9Ms=Oesp}ovjP}tyE#VoE6LCos(8_ZG%RG&rV-gp2I5O;0<&718s_IZk
zIXE#ghoyiEIkR{xQnZ8BS_sOV73yXDTb_s*245F<ZUS6pc@-jFVesm767fEX5A%I&
z&+?7JL1@&XNh2YW-QX|~W4quke$_6_wOS!X(=f!uLCKXCTr6grWqK7pUf2eJWQ#p)
zBg>MXlwU3yl;18-jAuaFdrpgq!Y3G&A>nqe${~pt)5+i7=&@$%l}A7|XE|2i7?qiq
zSx$$19!;UfX4BD?K^O2Dw=h5G8y1M<h$0V>sYn5DS;WW_LJwNupd66_Sg2Z|ObFjh
z;d(kYaIq5AW^M4S({!U*=`IA|!IR1mU`~K9xw={##P$ySXx0mGSToh&n55jF5_y%c
zQ%0S=?~SY$zDc_+hdsAVz=CfSf(EZ<nxy4#10J6)8Ra7VVu%FhuUZ~qXWcl-7dz0J
zKSC#*=D2DJs4QsXlWnUN-*orUbBwK#sc6(`k5$WhedK!ZtsJhw7aag^=)#2N)lqyB
zGoA$s6)g4EPjwDMT&tpT3x&y)-Sg~v?Pc^W3GM83-R%~S>oD1D>GF6P0VCK34{#xA
z6o|h56!6}szeYy=^CM5Nl^x@zzip7eF-VM@<_(W&%YElHG508ej!TWP+K`tdlyH@`
zaG`r|ci2*PtV{Ixa~%K+!z7~@N-pf^>&3UQru77Psh2e8Ww24mPgI;9$fNQBb>3&{
zhbqTyWgYAbqeT~fkmwhhQ+(2p-WCK|Fa0iwYFQOX^8<u9<Alz_6Blr#Q}{@R22Flm
zO1vsjAtO`B0hVC~PU)xYlUC5;&V`>cp`Z6dTz+3>>^Qj7#Q1;{n7Dn88~22qW0|s|
zg-Z%M4{y2RAr^NDJO`1|)5X|P4Su~|sa>_R?fyfxdi1(^<eVsz?o(_z%8(j$k*NeU
z<I(Zh=;71ShV6L)Z{zNK7L?`5BnX?43=n0-{#hFE<V0&lQ|d%Bxi6G5p{2ucXKJJd
z6scBU7>GiHHEF;Eep83TU$tO*{HP<^@mEIkY~l>UTW{`Zj0IhHAK$)E7Ek{=|AMTD
ztu%|{&cPZ$=p;Vh@FYEAS>jf;YP!5Np%!;<f9+kL=EiCC&vy_BE<~-0C@wv%FL|*f
zt&HE__Ya>NX%MxHz}|)_#-i2rbqH(cKFQ7(>ypWmh!k)GLMl)ij{GKNLF%>Lwz#0W
zB+bt%U&*l}vFp{A#$G%&$f@`DrDjz0SQ0q*<(U_oTaA+6X3?1*T(qLxwzW+_4Izt+
z;+BAge`a6av>)>TmXk9&?TbClTboYt-~W}+7{Btq0j1H)O-Y%#N19(QrRSA(t?yzl
z!o36V?t?)c3nE4J6-WcTgV@%WiFD!+N~&zNEb{e*rDxqE{zUZgOYA-OJqcYa?E09^
zVqJY%is5(uS?8r8)Y$n;_?Q2g?Iq8l@34D;cV=nV*%&^8z<0@ybM7Xq(yyW~ge%UI
z6tX|EFfK3SG^Qb5H}x`x;RG*;FOY{L+3X7d@s{S`Zhq^@dx>k1Z>CmN)8$lQNvB3j
z!BjSdp_oIiRqw(27hknLy#qbJ!HE9G9?htu49DS~eTa#jvlZ;7{-#CtJKLy2Tp7I9
zalLD+fj+#9KH754LZ}^O_x}NsKyAN0i00^IpgloxGX<cHuE@15!3@O!UDwc`qv~#L
zh31w0B7gT|Bh6P&^GU&yf?*nR%QG%xSg`d9@_d@<Ki)0)IUmyXk?~(Kqrplq9s6DL
zGC_El@%D3w2>^})f?~^fNwW`UR$~r;#oMRCbD&~^@5ECA6=AmrWtJVG2fG%zqf8s_
zQ}_tB_bG!1)p=xcF$zjF{L7fcAz7Oa=-`gM6o0R^9%3ARt04+qcP&d7hI+1eD`2a`
z)pw1nFOoOS<mIw~GkQtNMXlSlX%~vyLwaRUg8a-pXKwQ34syVLJr$_=Aa7#YS)ZC+
zx0mI`-(zFb>qn*1p;EEwOrI>BZ%_Cy_R_gueZz5MP3V==O7nQFZ8g`iM0ag#H1x_G
zjen1Rt+pEHDM_Jxxh#i=`!i2CBmFGh+~md^pa6@-MmZ?SV`bzq<gqnS>2>TBd9Dns
zM*bw+i=O4HTMm~urBIyv)y`%(cEbrLAtLOT4@))mLu50Y<6_2=DC%Sw=Vd0GhU?ZA
z*R3(1#LafKcF?SbL+%p-ws`)RBU6YZl7H2chNL_RarI=i<clR=CV%Hi^IMDU2?vIm
zXX|Id9>^|?XdHU|w8JdPw@@7&LRIqF5OP0X9^#Qk&zT#kOJ%^c$bb@L1%yV*pfN!}
z84XLg>BUX&76y!lCFC*WvBjx`hNV<4Z*%R9E5-fe{cua_RP2rA$P^+;GWW(&bAOL!
zFl2RP^;EI?Yt3NV6Pm%6opzWd`4+09BW&|3d+MER;Pmo5!_#fYwgv@m4AY)e$aqDB
zY!V?a45ejhC=KEwXf<UAb{;3GE<-bcL&eMOzhiTs9T(B~DJpGDifY6y|K-}zsZxE$
zrt2`$W-!M{R3~3T$#TM#vZU!`Y=04DYM9T{BzYqRIPwi@2N+yUD)gD|c1cZ!D5`(e
zy7bhh+C9H3NFTN@2qUpPKN6J$T{#6TgnXj{kUZ{)BgXM@xv^gj0jgX)ES1(&sZNY)
z(c(nF=+NEEsX9%q?OJwhK`A*n^~SM4qmzUYDz@<q(q(IO0fSi5L|QtlgMaSLZ9X^#
zOY+h9Aes-<gsCX<qxJjQai!EO(;-aVXJ@Lseae$#wkV4eae}2>IXbQ%t296j>K6^D
zZM2it!P0Nq0j(%m^S(v-x7Z5<#ABl=s>Q>j!?h{^uYYB~%P4UxWT+ZTXu3lSNMKc2
z%P3a6Z2II*Zd}<oz|l}0p?}@i9EW6}Jv7yCb;D`f)TJ4ck@wDFyx*@L)*6**3vKXB
z%4<?71Z)x1)z;jgPcJs2AkGD&&d{c#{j7D}xNKF=N$sNEh;Mu;MyU|38kd(Bm*lKP
zGI`P&nQf>Q&Z{ElN#<`e`4l^8e!DumXpxKNq^x?t1Wu~0y5A5<s(-cG#by1h^@&_x
z_)D<|?&8&5LU;0GXIG?kam|o$dD*C47X&0eSFhk<rw#JuqJGvqYt+LNtI&k4!LfeO
ztTY<qa3K`Ln<Ay-TEna-eg8K9Ha~{`gJT&l3YQ6N&h(L&86Akr%X$@2A2r*Auncy_
z`wb>Xs`dKWHRHIyO@G;?VR9qfz<6b3vW21R{z*#g<}2vVnSsZAkiWtoJ3|w#9aK*h
zQ=Zwzv0orwfz3JIKFB<^a;eF7`D+={Y8AVWABkZS4y|0MfR?7~#Gwz=*7{^X9Y^a@
z=xX6UQfjjMRxVdCU`HVqSh(!fex><-5=-hMZ}7}+8Op@aReu);@x7Dc$gR%EPM|Ed
zP~qf$#kWg^jFfEV2^J}@h{iY@*ZT;uh(UK9f*--35b*zh_P&HUaU@&!uVhmdci<I8
zv3r@YBUnTg@oZer*we3~y01e(0-6#MR}!|Fe*NEHW=aA{XhUq7B0FYWSOTn*=Q~TD
zJd2rAxfTIC%YV{wg(O+ppkHNIk~g3$c)UmA6ibeN2GD=|;oNZU%px{%F23)CbMC?V
z;GEuEREzh)Da7w#5^+fICnGHHh+|n9QtJ!MX0mdllVYQjUY9GVDS&4avc6zxpY?^0
z?->5zgB5+z7aEDaOvP77W`!DWDo_XihC@W9q>xKGGJkgF3{ld~{5{shSIQSMY78tx
z-pMr>iVE%&k0WU7L=;gkNQwa`ZtfEeju7Cu>rwBy^@;17WK03nG*!zJAPiLvVQ#J}
z=#LEXm|!zF_YOp)*T{$Mru8;#qSdTNqel5WK5Ev%5uX2aR71L1KIM#ln3~6RYVVac
zdT80U+<%(^maw+iws*;w{18u>5?B4G&%EEWuYEdP^pG3nNKwiq#=RyUDPq@aiWtOQ
zX;&_rS02`0p>0+3<Q<CKW8t9k`D-(t^7{=qGUT@=p5kC%Y8e?`&}orIKNX>|))Y!w
zM1Lwaq$|Xs=JK^CDw0+crdObDbKt+k+5DdlXMd1eDlSRKqJnG$xV2(W!nn1f*IK1p
zsDjHxS<-v~TSn6sr<Zr!%itRD_*8@*0&p*bdl_8DxQs<O9n3YS@U%+!a*6{LDvDCK
zaib%UfEUqv$YZmAk5ph(Y};^vQj{m@E<n%ICu}7^tHyC8Wa2viNcWgc(FxeOwI4_q
zgMV+0e+l5}n*jQadjZ6W8#6;Vaw*Mod(2!a@@kI4*w(|{^L-EfgOp0|iMM(<s8v1S
z18=|g;GH0jv(mg11n&fK2y8j@itnOOBm&!ag4l_Z<@zP(<S{tu(O$J;{i5BW>2a}b
zi4iEU9^40f?mGm04l21CIkL~fY#oFZhkpo?9=6Vac9)^_DsYOJ7sKjaw(IS>ztt5g
zojTEEoM|W59-B3JVP?2htfGAcO=@H$8e(Cri8JUv7~c|;7GX|IT5k@*B*Vs-AxwG>
zV>wsZ<?0V`!)rAU@{~VUv96;E3vZlzIAO7)ow!BRP?c3Jx0@z(-$^-hZTd~E;(wr%
z5_y%z6e|;~M}pKeUe0p8ep$PuX*shaX*|v@3$irJnIZlnz3ZDK=ZrqDJG0yvc1;(G
z^8h`fq<9V;X<ye^;NY5{YyP8aeo7bAUtnAYf#L<nwu&iJ8xVk|*xZ_iKnzSLvW~$o
zYmIi9CPncwmK0H@cwKP<1-7t!VSo3%urczac*Kh~u6dgBI|P9oZ!{{GW%GE0bw|@8
z#Rq!sJ1K{rO+SqZn@(JMhoSq8uk_`!`;g97=PvEyWd23%;;QOqKGkP>d+f7mkxmIl
zi82b)MIIiy7)!}g2YZ?y0CwhZ|1(2HaGgr}FR10oMbog`w??WFjq$Lh@PEO$k8O3V
zTdFZm9X@V9ID5xhC)bX%VAl~MK6!eDSpQ{LvtIK+hRVd4qFailzBxY-o&|g<prvSd
zKx18`XnpVm$!OKj0Hb5L!*1h~M2^qJR?)^@S=5b3sEnWnN8iN>Y-&*aMLw0NDSsFc
zD1!8G3WvW6KP;^!>IGSykbl+=&gR!Ya>&{ZRpk|G2Ef_vOJj<_hqK^+pdW^6=KO=K
z-AD8XwteG;ECY26{sT32{{=U!<x4ZtO+BdvM1n)CARv%I?vDX=&hvEhobY=j>2q0J
za$cvK*K??iRc!`Gz<85}@EVLIO*cr7?<GK=T0otgci$9A=V--oRe#2(;Q3XLX;qPz
zKm<DCkqpf@0a2FKH}u{3>!$x+pf47cF?hwVwg%R@*3?k)M3`F<ZK7NCQnf`@B%U}s
zGbk!Sk4<S(U_fZ%#xBB)Q54i^*NRsJ^j_(4-yzatT?&#O9=|^rPptxFs2ETyXmA`S
zL7ut;FPgJ-X3}URhks_9(Y~Gs7jIs0dN5A@9kYa!Yisfp^l*-A>ytd&z0l}!U7b!{
zO%c;#)RtOEvoDzjMkv>7jdhzu2PfTk%!9i%R#P3;ut=xmF#~mXcvwG--Skj~ci<ak
zJ7Hwi8}+h7_0URIcs2nWtckNJ-%_dgWdnpOmDEg@<x1sv$bU7xT3LN_xU^<z3k7D9
zBbp;RA)=?!Wa@^rx1j7akmjBrhxGPamY9(w?qYECGUQ^IDYe|u;J8kiqd^a_xx2xE
zy}i2;4cv0Mc2TbE0^Cx$ToukB@hcU;4P!S&z_mS%Nf<4j6B$}*A%=y0o`m+4mBv23
zb#i~j94=;Q5P#Yk`ExaiV0CmHTFlEfUvau9DEe$z*<_vris!{glHeP|5eVa>eUUo?
zsc{4t5!;m`fWr>4jg6m#+mb*OTI72z?H4UJzJ(K8iH&u-%f*HTv0=o2cGU;|3?qmU
zCwi?1(&0_tn)>Lovp7AJPt8(IsClO{vCJn%l3Hx*V1Gvk(*V6_s}{x_`y#h$sj+H|
z$fDb-xm?54t5ZZ{inOV9h8RN=RqEyT#agF2l#crj{b|=bDIJP<_<_t(97#yjl%cyO
zD-H)uXr_m)gZ<eOi*!mJ)1z^{+J~a2{1~bkFKUg(fx5aRd)y!NE@h&@ad?$mm0EqH
zLmE?f_J5&<cD{W40X{`c%At=1&d}ZeGW|7dOTZUUD!Tha@9qycsOSE`;rs#IUT8a?
zz}NluTy(J3eJACvp6{Yi#qXi2X>e$O85~KX0#ps)?i|Rqe|I;8p$ew~rVaPa5o7eR
zNe?);SKEiyJIM@*M9hdf0l(dVjG|`Zlx}UsMSq9r(U)qP5-vv0bqqYpiyCBHsn=Uo
zs@h0cqxVeE-XPGTf6JLXDCd#ull$+Du1pj?dcy1pS~(G;;U(Aiv|%vDVLb2A;3Ixa
z4NG-$^N``|WDg!k_tEpXiGn#MW=f4px!Iy==}kYq71+9vWBmkJryVqTFp?&c0dBC5
zHGk67dC|~BNk;mdIjLoB>2LT7A9#iFjE1}wzd#7jjN=Rh+dc3Wu`A|*w}U)pAZ#}i
zJ7|Am94!u)HTq(n=NR$aJ+t{-TEeFy2P}-;6iK2q&Wg%1>&`RH;cyzZhjC)SwPO9E
z-Pr}#j#c5frXAN|?4}6UCZNgCj?GL)H-FmmG1BICpXWg}$OroaIej4mQP-u3lAVjX
z0gINVn@kFdID__{9Q`^}7mAG)&&5xml*@z3WG<^l`T9CkI!a|EA2d-vf$HofhCoyF
z$bN1h?I3dp(&89Ohq`dziK8y~Vd*}YhUZo?2DG3dD4{mn<Bc&K+hvL?amIRcrGG%4
zf+7^o#Q7X89}=mR#TTLeVor5s?YzVvnD0LNg7#<5yk|V}#OmrVBeb=n)Q=L8$o!yq
zJVq?tOFvd_WpMyyke+iMI-%@6gQa)5Z*zIq!(nTQZyYx+!LU!^yC}4alP>)SBKtsG
zVf;Fd56MB0G7yjnCio0`AdACaUVp#G7Xz(Qt>T!#?rdi5y2JMoC*`b2<^Uhn`yRCP
zJ*p?Hmk1Gw6;r8Rs<(VA&dyjdS&KqDe9dvy;oFyHqu#h|`!k(Bn45^kWPK#W1yRNn
zn_|gkqu6ZM<IAS&?L;=MOI$YTU^*EfpH|r<#BFVG7(S8RW7+|Ip^jYOXn(uIKk+_O
zzrr&mDnhYRZBs<W?eC4pUbpv^vBkf=28s$42uDC2163PAEt|t>cYuV<bTEKJaiK#n
zP~XktqmS=EhXydF(0qHe;%L{ZrCP(I7=yRa?zi*&4l1k#VLeIf(5*a<Vb;9n1w4fh
zPaZoDGQ{MCX_Z#JRBat}*ne|U66UJTz(Q}FWUvD77}W?o#W=1O<0;0+Gv7z)`ptpM
zfxB<uZsf+x>f2VS@$Gu4Qf=)5YwH_=V9m$#(FE%-c2flFn5ppy>~S*VImS80>BIO5
za{g&wlE^{;X4Kca5hL6F(0Zv-Es}wiTrY9Gv?FD>3<2C&8VDQXIDhAN9N!$@^eN}?
z<awNTPV%^M3UuZNdSNBKZnVaUll#aefJ*=)lqJdP<nZS3W`wc?`KpJG_;Pepg>E(?
zzTruaR=?p_s2S2wl<h!p9ux~T@&kNY&tJ%KAxN_j$f9mMLS+OsIQlM5U{izQFY>80
zABfl~hGCN66asSLD1YbHJci7T!BHs7<>F=i(JsmnselYArBh`Ga{c5&Oq#Z0A?!W~
zWp+wfG64wAd!!?&a3|t?k_s7SRY8Fh(U1@Xc{U%{I%k#5w^cp*xd4~GS_^afAW05F
zJGi+lE>F+z(nZm%Uza9R^y}wS#xw@8R}}{dl!<)}DxbeLPk)Yyy*)kA1ULu~BMZ_O
z9?eemQixlrN3}XI*L0h#{-nsT){G4*Fe<E5-BqHYm5z1HW>R~msM~uI9(#>D(0XVK
zKzMw7cm&-Z$mGFzp$rX4RjeGd%pXdSv1gQ-gyL?m_0a2wxiLbu-l|elO&o`%{U@Q<
z`T=U}qB8MrB7ay5*E$tJScS39TMHlXfwy(fbVHrnS*=LL{Gr@<{=?TYt5H)pl`QOS
zRv=%}3(7*4vu~oN<j^P39s6$}<<@AoyHDMR-XDq;G+ciWf56V7of!WFCG_Eb27d$o
znrrK?kgpA{`Nr64v}%neRW<5)Fcf9E0B%61an~2(Jb%4gwp!&<r&2o%NqgLsf)Duk
zMp6KQRt^{{7(gHnWQbqlG2}XYg_<aX8Ps)g0&zsuc!N+upF6$tcV2sg!#fT8+8ZQ9
zMyXkD3BKX2?U7NcRGBhJk=(ZyhqA}z>~^l)9)m*%W%o{D>?A>=R;yYQy#1!GiWQ+r
z*yiQG`+x0c$$x8csN}y4Cl~@VQ9&NT3@R~tGo>8>i52K+0Fd}`F`uPzL?e8<9zBM`
zVgkCLPt`Z<4-=oCB@!L91SGEfg*^Qk&fh}r{S<0&l^9`P9Kf-PE(cLfB(QdIS-J3+
zVG3c2C@PF4MU)G;79~~e^cy>V8ntU}z_{H#5q}c)UFtkpNF%6c$C8W-v%<7Nt5T|W
zsB9dWQB(y+bb4Zp&A|gLFfo_SYO_pYAM8?SSCOO<a%tgF)L0ePbf-L@;?irb<&OXK
zQB7c}-Tnr~(UGS)G^AJ2fF!wCQY1qXWoxhOA{P4|_`CQ9Ih~$boPa0~h*8DvJp@D(
z1b@YRqvF9JIKQZP{8i&;34GP~s!^b(X+F1F0(~gT(4nJ2b|@-}>WqMn&~G>x;&|8g
z30CKh2`br--E!oyBM9IoMWQ~3Ljy=U7^<_mEDe$8J^~_Q1jd7;7z+v4>J&g9LU#v1
z4geMb5LA@|-{-Q3Hg7n6_c|iLQ7>K8>VF3s>aiQTr-m`tufKemNXB$AC?H6GF{IJ*
z6>2*CVnuhqNV>j&`j^U8^^yZRwSXq2&FTP%+GIg&oH{n`-plf3h@M<h7V1LJ<|44(
zY1I)|sm8hq>%t1?1y!4T(cl<rP#MB6Cedq>3sYkzbKs{I_&S6@lyx-%*urJ8Tz@Vy
zqQo`bSd1j~%Ww)sL!Uxu=wAl1I`}ffCw|%&+&%R3tE4ZGQ}Z8U!#<Y_rNVsVo^2fm
z=$47AmP_rp13EeSPk<5(bfnZqqP7HvH2mc*k&n;mTJ@sbsQGc)o#??Dr_GZ@>#ue?
zi29<vtQ0}O_2An=AYN)ql#J6ySbu1?%Q{Daym%QbBzcsD=$o>05n2B%?hPJsoM$Zu
zg)Ai*9D#Xt?6J}!$UiZqbnVVU$>q@|T$aVTjvJ+Bz$f4-YKyfL)DG{;gMTmy$^dcP
znls4ElrB0na^@3|#v7gNJi~oF6t=6iYOUy@uzFP~7pa`fx6+zETR`|e$A9sz8!*fk
z^Pq^NP?lyAdg>dXtD2Er7SQkLD&9@rn0rY5`q~Vk!55r=u)UPhiv_cz!D_uo6Q!52
zEX}h^v<zqg%;j(&?y@4vw{bL(n^W|fmhfs)dU=q*rqRoTFj}hBYNciF7MQna2c_i@
zVGxqu-vuL;9AuE*h&Uwp{(o6~mvn#?{?dG@qK8oo5;O?UFSVAy4NYtNc!YJ?%Z3h9
z*9D}7?P^Js#cY87tU3lWQBus!G+7{g(<gDuvn-wsG_%W6Kxk=qLGR%SJonndWB;kq
zYlEk60%$$l-wVAb@J)CWKvQTnp8Ep$T>w2nXrr4CkL{Pvo%G$cNPk~a9#c)Me(=Ir
zPveFRy-MC{B!N94uOQI(wDC)2yz0#icTu?h+mtZTi=vORxKtO$$GNCL`DHF@;!OYf
zv-D?-`I0<DO_ee^gu8Ma#j;bxeOWLg?gi{26y~ZfVV^0mH7%2jxCTVc<3tFklOtpI
zAp^*&io68Um^p?q%73QYg5o&CNA9|!Dxmj#cZd9T%!VbN)9ZS}NE(x)Zv%`qbp|qu
znz%Jsid3wN6DjxCq%n@!Az1u&W3oRmlF6iwB2_j9+Lvk04IAn9uTIJU@7&OfQjr4C
zpSUjioIr0)P9>SWrn8?F^@0VI>`YS3ICdT=VnM@kToU8Yfq&1Sl)HAlReTS4D=w1e
zn)IANwOf1RTCq`D9}#gy;<Gpu@|-9VONAgyYWKwy4MUR#3*;7C$Jpi^ndQd;9{Z)|
za=E|!uA2y2D6*s*(ciIs54Ceg6@>sopb-&ZFmS|(7D<$wQ}4r5_?%eog4P(XaHCki
zsC$RQTnZPL#eYMc4IU#p;zJ>Q#N1*j$bmP>n5qlIa;83$9>Hqhc04USTXoiAl6C$x
znHnv)Vt_r~-LC8YdX24ZuPfoqzR0Io#PRs4O45sN4d^cqL8C3)10pCjqDhGkpLg-l
zyZeDe3EvO*d?@$-J)U_uEv%t=hglJUTUtIV;wXEPZGXYcXGI)qrTMIgL(Gcc;SH*u
z40A8>@W=F=t1y<Yu>-qhq}CsyGJ+Z$eMiFF#O@IJRH#qvJcdC|;tTmb>&Tza1EdbT
z=45GRRwx5pf&i6?F*U=aKf!On+<TH=bAO`y>-R=gu+w|AkIyl!dNxogR0_qF9X6fM
z!c6?Ps(-!c=&ufO&yu-}7JCE+OZVFXd{p`f_D4(sbjM&U4vna`8waf!*@=<+!!oN(
zbfJ?`Hu_T*!CU$nc3Wjttb!%uG+qbK5<M#?GX;4T6mpi9q|BX&D4^>Lw6irmT%3-{
z@yhW^F<$2?2kXAU$_n!%(t%IIf7W7iiBCGIf`7$MC!LOX4$hU;UP$7#71d!?Gm>Z&
zOAnuvg?XG5&o-mfNxB&2>7Kw9G4~p2A|fDM`|8~Kq)Z@*o%?7~$YqnuCTp^}-sYuJ
zFdenYTsj#G2k&}P>erwTe6af)X>g{#;&E$WDC*|uwZUS7%b~~8JB%@SjiZGswCrCu
z?tgnkE=e3mnkYMgD&7ERbU48!_?bLd`fRi4Tc(uTLBs|;h>dgNm@(!&ct<Pl%uabP
zXh`G1pQS(1J(D5+w~>6|&K0i0Nb7D6d;yn5Dvp4;7C)lRH*05+k;fMPP1SzkflXFj
zRgjcmni{oQxy{hj*b_?pP8?Fl#L==$B7b(cPA{RSE-N@A2@NV%J))<Dt<7=l`%Z$A
zhDKCjo|a|UIVX-ogyr6PT&bnQ)3RJ&|J#0^mUWkVT4(BMafZ9rG2H=@TO#wsAlu>M
zVb*ai4>HG5<vS5UUpvm7$l*;I&r1f$<`##wvsOh2t}I|p>{t~xvu^Ca#M%6xR(}W8
z8=f0?o~{p)z~e|A-?!ja#-!}Oh>z|hMI0K^E19=$H%k&_4T_^B7*65v7r8!GHFvDf
zS0o$dTBla_isWUzx^79~6iN5nS#wTZ6F-W@X0f)##E-;ztma}eP9DY)5_qd1GeVBo
zpx8;8qxd=%w8crGBpVn|j$^oMoqtNHK^)vzTEhKyq9v?L<eN9OL~!hx1pu8;^!FtI
z5pB9k@p7Zsllx&dv#ONE-E`brV{|RX-jWMP%7tSs2W3kRba6hHm+3Quq}TFFAaE7A
zTO4wE(0T5G8>?xRXX@}-3JVt!myqai?v|06`A-ISJ;Vb7l;Kp=(&|8BQ-67C()LCY
z8S%i`#SFW^Cs48Vvj*pR(C^-Vzw35>tUBw3Y4147d-q_0b%}iQHp=^m!2+((O+ane
zxz}uN8zfAG@{3}p7GU|AU5=n`zIga+@A<wdJc5S~_$u^-$HtS;KKTY}Z`$Um+xp-A
zvtu{tAdq9;{`lPZzui7S|9|T}Jb~83<D<}ey2e-tU6We5Aslp2>i%{D;eoysa5zrP
zk1?i&w$5dpT|}=w_q)BXSOWV3Xgq#>CeyY1!qXGF9IlR-;bMU1$@U&VyW9VM*Z2YM
zg+{Nx*^kCUa=B9~Gfncp8h6hE7fnXlcTGByhmjo?3pya#d*4jD@_)7Qp3+X%*6}WE
z9XxbzxX|*XUtfQ}6B>OX51xen6L>&+`_Yu@KwZronS9@T2(iM;Xgwyg$Y6PbMBJGS
zhi8hJFmqr}<FJz%c#ch6PVpR@o#y6ll}gEwNjWHV7;EUl+YF|Zg{f#Wm<Q%a4d6D5
z_7AF_n7*3w{Zr$JaetbFW9Vl~YU(7?rL^L--1IoQsqtwO?p!H^B`;S^D`$D%R3^15
z@0-eHlV#cDeN(w~ZX=zHp#5-7FHa96=0bby!pL7~Zg%5J2#UsKsdm9Z;SUOHaaaKb
zZc4S$m?beXxVlJ>y5QY-PtSvd!pP;$$@5aP+!C&s;ra1eWPk1?biV!l!-bDpM%d*g
zo(|UxaKY!?p3!Arx4UFv*%@8&pd(;zVb03JFyq;sfsyex_ho=98<QT=XobKh`Z*Dh
zN)p~Idi%3j9z13P$Q$X>)IVb!^to1@Ly|2>o^Ru7dU=w#arHiBFm;Wq>E(CQ<f|01
z+giCjc1S)byMGfuZlF*20pLAfkjBCU)mx=@`B1eiyATvh{l-W2Rk0#8*Y&L&)hwV|
z7IothAsA|K^j)05CeHRJp9;kLSk=74&cK^wOwIQ3d2~<G!AM1Zi9M|D8bI{NvvX0?
zA=*B@8{oz+Vsr)}NM+(iD|+;Asff!OSE&~y`~Btp1%E-;kd%eG&_n9dUD%{h2EAl0
zon++_&mT&WFuE3bFT!|GG$ejp%;$Y)DQ#jZX};1v0hPLKBDH|>*Wq?am)+IEtR~{X
zID25sBt_L;R`A$g`Vt2TljcYk2gpk!-dfh{k>&m(dPBdyHZ)PuUyxfk6OC*}8bM@h
zj3x5P?0+|)>r?xFsrcW8_gro@^U3wK<rv=@<mlm*Jm|RN{fpZvoz|Uj;m6o=5MRkG
zsg+0aCQ8WX&zf(HIN73W0pive_%u`P_MU{tUgHk59@+vhTdTS{jfR0*sMcFmmMSeA
zZBq%o-CpaV*YEb9gkI|hsHcnP{*s-^fqzD|N`Dg$+u&qtnP#}VQK(I5)f&wvO)DL@
z7ZZsa)k>XKhejLlceo`sSKTe!t#YYTsU3!xJ%)4-+&wDo4)Xqkr}R2P@O>_ll&@j7
zm<PpNpiEF6h71%04bIe82*@oUjt%q|t)K6b4rFO288n<1@IB#acQPgrkR0f$h6fIz
z-+y!L#^gVMUIXd>k%YV)r1Pi9pSK8{Ep!76RSgk7R}lb_Js63GXx|<z46sD6k<qZ{
zcP?}|;sSE^0-~w|EFIRq8C@oUc|}!t;PuR01`^j=AApjSQd41?^E;6rk<qY!sYATo
zSp~i#TDG$1TEPEVNa!PoJgi1Hj;yRTK!2(fYt<prrbHJ*FxOP%iqC*L2FuT;8-l}>
zU?kxQWl%)xi+mOPL|U4G5yYR?5wHv3X}Zw!V5E|>X{4JA;8};Z&PLy&z5(penLZ9a
z$)-R!`T6V7gp;4QpBKNQ%V4-Y<|!9jrCg9C?3baa8Bmu*C5vqfIZMSRK5a%kh<_YR
zOsP?CHx3l-I|MPQVvWN|S|QW&a*)Fm#<r_wwcIIENa!uvh%?HgTLVZ2(BMcD-<TAc
z+lZPB))bcl-|WWLA0h<*csy~9BItCG-;i)nNw^8Jw>~4Z-x-SN1<Bz3OitOh-B6_q
z6=9;;X<l4Z*5f@)0)<Ycd1;PZV1ET0UC@3YGcKQdZQ-&1)abRr)3<II>CLNBt8<hB
zhP_c4Iyx{VtPBA0z0mJBz6#*?Z{0fqbis390Ii3I$M#FB@%SY4yNzDh;v8=s&f06c
z{Q!otG{;>7)$urkBVbbzAGDE#OHo?_GV24V3j<LD0e0b#12K_Q#a&zBqJRGjk5{C7
zv%1qXAs(P$lIMpBP>hlGNq1~wPYO0T$YHKay;kcSXTN-4$6-H5EMdL{drNg&0Nu85
z4;syff4L_YbG4g9?Q#bt?G6g}M;Y@?cHG6_?gvXVu?5UTMVvs~0v%+Bq(MGToE^Zi
zszGE_4b#Z3-^CAGch?Pk)qnV^amD1Lm}EG?x@oO#xU$(<Y>o?n#E*;lERD0^H(-|^
znP4cqNh4#L;~5@j1%-f&brTU|NfBkooaM>nWde;C^KRCRnY;`Zl8k%SxM1U00K)Ft
z#bxEfA21Y(2Xe|x+6$P%Ai|jBEGeob;W5NAtG?DQNtLBRVJ;fe$bYfSbkQ86=+>s}
zlu^9>^AdajNE|I&9<8-2m1?CPAcXcsht_t#ork^Y$HTs3;gf|lf_io=$#R5#Z&gb5
z4waMbGPsIlquVRfY|aBSfXi7z5lcjQJ{37jPV)5WJDfwlCvwbQ?Gq<SkqilW;&Cl~
zt9(&wG-6wN_uILpr+*6EXshnor<0vY&T0ipg2pjtl&~iU@D(aR8Z(v06k|IgD$DFb
zktN-5H=JaLV|IjKB<b^%F(*fg0I)orUeTs+W)=H^upqc&>YAnEWx3<QgLRQUl5f{L
zJ@O)(`?R`(+>RN1c~zxRHg!Q0wCOdLV4{ihDRQmQW?3BJWPcu7yele%o)KZ+7V*x;
zX+)YG7e0X!XMT#A@2?Nc5EzDmr!8gH>Y!($BcI8EJK?L)YlBDO|2zx*C!r0py@wv?
zJ>T5{+zV#{8joL}?}gqIcz8@*1l=@*`c!>eli3%Zz|+H~uf!@Aa(^hz=O@s`lnM8J
z@bs;L2>d31?|*j>PvBdl4|)$a3e1u);r{#65BpO8d!r?Q)}zqC;Ajf11}-mo=#Z~i
zL4{FkY9oN060ab@7~60EP~wcC;!2)v3-8wC{>@g`dVke))%=C@ZC$l;_3NQ}63kGM
z6OmHP@VwmQrs0HaPG-qMhB|Q|Fc(|jDTrw%X!Bgi{eNo%zG{5cxWe&KIJn7;w-AsE
zOUlo1$C5%aNv(Z}zL?{fgMM!{$j{mren*=P#0x#UIT4a<Q36uCO4sf@t{lk_lEt}>
zlRPuv6Tq_)<7{G@m2#=EKEH6_%YEl8@ntOmh4`9i>Y$*Duh7L|`v+Ce8gM9M3I>jk
zam{#6ntxFmXG1brPQ!kl+;l{LXp9V%iajDAOA6E-nQ!Y3Z0dN`b3szfGZ_sI)mMO<
zQCm}H%qwS)9!5};6fIVUdj=}6FQgy|8ZQ<AGI`v$uz>DlspwP9ac~42xE$eC%zot<
zBQQufPu>SE93tliks;n1IWa;maFB42WHaJugny8MesVZ)=X4DZ<4r|Q#XBBwrNC?C
ztKz9A5X@A3N{|O#VD)jC<8dr_QVY6?-Baw&&NWFfl1KVV&2me);+SCzGyc(y56(S{
zqnduLS<Jwm^XEp8&xNCn6|vyL%8JB^SmI%~tE3I2nsd3oaD2#Y@`--`8@=Nf#yR?c
zC4Y2Chfkq6f>?i(>9KiJ3=4|7zJe5cWP$r3mou3%@-at@BH}g&@U~wAL>{1qm8Hc$
z-DDTu`Y)|A<ZN!$r6B3FDXvOGR4HSM<AXci6QE81L0}Yqin7C5O0;G||NUdFoz@Ww
zuANxaPOCxIJ8Gv!tJV=J9_@75sn@PZHh-vBJ2h*itIA4mouziN7Kc8V1Zk%r)JL(O
zLP0Qlgk?2oEHpg3ogC0uRmbMVLKwC6R;7ID&4;Cz+DT|{X2w|OI#LU+3MZ6ZRNI{k
zRwxTB>9dcDOkg6ljF$j?&lD=>2Uez7YJ%0T;@U$Nd!Qs$!}4-HQp&6u(jg4B6Mr6$
zz~o=;vRtcBIe5+#FD5y6TQJ=?FoOC}ljhiua}k0a6a3ETO=hqg$J{2<wXXELafu(-
za&q?D1vLZ*om#C$7#kA@X=wgHAItr<^(uodMw5pyzEcSC=xoy<Vri#rqDdp+-O$AA
zYy7exY8n!iz{mrGyvy2|!?Ch=xqo3wk73$rmKirflp7(hy*Vp$6QH(Nx-M^3<e6QU
z@<17?@=ZAph-aqX46~)xJ1l37E5|caj?wJqv3Ae;Mf7+%4GMS&epj6Ps<lq3RIjed
zJgXD$q&T<Mj14Lhz@PSW*J!_if<yhpj>bN7P*PW(c2H0ze%zgRVLb2XRDV3*NfA*}
zla?OzoC#|D64Q1lXy*K9jQ<-s-kkY(RT*J^BBt-$O_(#EUFL7mIGD7L?V<nozdiik
z1C9Q=hj5&Bq#Pw|tB6yyBjvkL`ml4raCh3-eO?<-+>Ua$MsZxy*R1EO##fD{Rcn^&
zovYR9&6y}?%AU;B!|KAAFMpD@1LoG3QR|z!@GnQt5z%9n&AGOBR@I#Inh6f?9T#$P
z$+OV;(G#QS*$fxBqW}n9GE^qU)D0qBR#XK44_NoyFtTpv{<6N&`o{&Wqb|}O(|t1c
zLcibmD)d3;@!{ScT@O0Xz1CCrq1V3#WYI{(1&QV9Xg)xylaD&dgMWo0{m%k=5Ws5p
zn6ObDnM^L93J9|g&lM8As0#oiW}PbmhbT9E!+h~Kk8hrt*&}&2Te?mWz7uzyxa$<#
zbvj?)h*gtgnHHDt!%YfT2iuzzkKG6{INYRglM-N3_>MbGR>iwv;f94Pg|%Vf40o*~
z{R~h1xFohQHhln6eShR>gZG=uO-SM(I8boWskNGAhJ1ltxKAOYRiyymB_q&EM+r8C
zBw<t{WZ3>khC<Tz;*$_h1i2k{FIKDUFLQ1AX(?l=+vV`k(`eNi%_c=GyG8z!D9Z)A
z$MMF~Q}n3ZnM;mN25ox`SFS1)tJRu^-JXjMcDwJS(RYtuUw=;t!;_ZM+x`H2pZj!V
zrx495>0cS$&aGUoP&o_K@>U9FNQM3d4>-=f0g@sWI1`Taz&SFv3z+*dN}o4c1#!>g
zygUzS>}jme4;S_W01A0+R_^)alCy)?(5da#5y17e`Ql};kmS+2b7?|c*PErHux`nJ
z?uTYf<J#gSD}T8DI1&vppb<3Z!$D{uOaAk7>CY1qu*<nn;0y>2&eT^(CWDS8O*a6+
zmR&|NR5kSGxvGqGV5sC>X!Br5q6`pIjU^bU?^ua)9hw+j4jsyhFs#!Gxb``o_Q`?Y
zZtMbzp{i;la*CkGI@PN->MciKQD;EVbti_PQPFUlK7S`A87zmQ-I*ux5!m5TyBAZ_
z?L7&Py~Z7AJ+y^52VNBMN_c#Hcm&;^-x1y)3WXd?s%HYU`f<5)NJY;ePfXpvH9^fz
z^-?&w6eT^|0F7RI`%`i11<Pr+!Todp2{Z-JZ414pZmZ$$#YYFd+tOJu)%&+bA3Qx^
z;_e^X-G5G3XnSd+=$R9DI1Apr)H)SGXxn$VQ*|s>>V`UBB@^M?nj6o5u3lxx`Lfbv
z1K#G}2`!UZbwJA?pP<ohcX49zAIk5I$6mMhb^Sp{w?&e0T|yu3XRP|dRq~Ht%QTPl
zvzIgD^Ym`vX_ZTzO6@Qt?KeVlgAYsrfmRL&L4O7ihyxkom-t{$hp$i*<@FHm0W{tq
z#0Muy^-8r|YNum5g86f=Q4w=-D0^JY=Y+l3cVTZDhF`J5UgEV`gB#3Eh82}D_wNqZ
z?(|~MadDb$CqabQ6w>$1p{J!mmYnlp|IhrB)x{3db}YgDw;PZdh@%(t1T5n|**l5a
z=6~MW5xq6e?sH-{u@07nb6E;mT%uPW=b{4Tm$|5kGyUh!(w|%#oE1DwFj)AICEZhH
zcM|?{C;oW+CrK|iB=^NV6~#dVrl3)&UR4enh)YUAK~kikC#QV%{rxjfRU{X|bc>)r
z2y{cl6Ewp_vV2vm2=zl!w4Z}cLQ6qWWq;K@Gl9!#O5~Ig`P>WB%{}v|ps8;KWq>5q
zeSu@*yfLA>5utNDa9e(h^TZJ@hMQ+7n+-7Nqv>{8(BV*3Mg>C}{$kXE-xkpI1;paH
zxyj|req4F5c#-pVKj+%<EVZMkKpDy97+p7TY7(4qT$~jyZpHTi&3qcpUsh^Naev;u
z1s`jlJC6(kB7%xQ{qbJV(C2UmP~u*Fw0@^ThnyHm*6Q!E{U>9#<-H(_11MwVB8^w+
ztfFCP(qLiOSxq0`KMwHx5*^QVa$hl1+w5!lP?X`v2{c~Jug&KEj_Rm;&;?7(U;M-$
zRxL6^)&t3uZlJXT68fnFWicC|Cx6J2z8FB&z;G&R;?Tg?e9~dL`aVygo2$A+I$rH4
zcMFBM1R<aE${$A@cNn7Zx%4;;*6$oiTz4UC#Bvyx8;Z-m;&d36F&qXCvy|^`m$9v3
z6sUr%;reqFr1tT!$%!=>52|t$WJt46q;A#~p;o!&YC1{+l&}nlvyX(B9e)7XXqUWd
z!CA1M+U~?@w=OgdV(<Cx&TaA>T*5rMtX{OsRC)y)19DlxQAD^)XqSt^q1q?!81Ayc
z12+1$9c=w~dg6@&BSw}w064gX-b54UQyflox7dv~MV2JHSHN@8VV~T0xP78XuY^qw
z-^dT8?&*zN1Ghj3mrTZ71Aiu*0{=&McnX9sYfoXKse^(pzCwqqK*7d{mo^<wQe&J&
z)oaDeHc|Xu&bse#&Qir$Q|7Z~B4ua?o?5xwuC>@v#W84cGl?-T@IxF5=_BUWm`Z`}
zcv^VlyDwscJ?ZC-&+ds$R#`Ef)H94t?wo;^^P5J+1akuD5s?#$bbo@P0#Vnci86yq
z;;96C+~}lkWEH*c@b<^P4$d}9_0H8cp|Uj0w5ROJF!w$Hh!)wWiMKC5>!piIaYKCF
z7oEFV*B#E}(=vG`N(vaNBX<~aOSz4uk(tOp44YKz0XW6^hrXQ;Vrb?ag_dm&^A+JM
zvX2#sa$nDNnGWIQr+;IabT~A=ugpu-S`qRmef$`2=nt05)gmDXy{=uN*W~ZdC-)-$
zMnj#W>($Z)`7UWP#r~A_>p(ROb+-Q9^0utsrcfL~^mi|c<m6$js)pm&$-*GNT7Pb+
z%6JK^4IzFHxja&bUp0w%a*_hSOU4jAWu;C&^u?x|yd0>bB@20hKC-~O#@qid0GE)P
z0Ts9SegQcK2`&_%r$`n60Na;3g#jaffyL)&)V=%8Sf1)@;wd|m3OyeHl8}Tl0k8m2
zvL=;(&pISw$@1zcPp0xD3Ja%CpMAge!8d>Wd3%5HVSPAm_PaNi`VZ>zV!gZF-)(m9
z-dz6W$MGs&UK~%W-QDVbzgxe#{Iot^{_)#?{-<v~ULDt`6TWu5z!JOT)%Ny(&E@;k
z>EY`2>*MYF^>%gq!~S8t1Ne4-*sf0a@8R9+?dtH;<HM`l{q|vX+T3jJH>XdpMJa#z
zvdsR?<>O&@^|HjP?dJBdKknb2mP@YoZ{Kci*DwE^uQ>dk`#xXN_P3AQ_3rd^$JdAT
zJ@&BQ9p7&rj_1|3|JT(3zCSO2^5Jhi#E0$u`FkI=`W=t^aeuh`<%-|4z2$<3!~S-C
zJc2>n`|}1jyU#bazgg&)d;bCVetE>FrEs}cPk(*6bNBZxAir6_AJ>2167FS**T+xW
z_0Q*pj`zQZB+t<Io14Sx@GNjI^w{2BefMs+Kdf%<u^rIm0*`j_w7biHZz1J>-S4*-
zAFm$Phg+~3Dx%fp>jk*m93SpipUUd?r+0__<L>VI{pw-;46Wa;9`8>-u5PYR`v+Y8
zVTDZwdAaiT{pzr~J*^Md535@+sP1>C!~XvKy}SK?>`xU6asdBcu6R<H&wpJ%s~)b{
zt+v?quhi*>{T+1U<JIGTVe|XxyIk;ecWuAC<<~db!&DqLck3S)a$TQ3-LD5Wf4%wZ
zI`8iO^msgNu;P<a{R2DrTN_yKmOK9$4F2)c!+NY&r$><K|LJB=oEZ0;haWbF!~XEy
z?hYdGkKXL{m+XYyy*r-&_^<u`bpEPRDkvNM^Y#|>3q<J@g(KpBN9Kr0=7qIiBTiL*
zAFX>X9C3AUM8t-`5#NOGz^GUUM<h7I5v#^zHCtQch%22_$M~2NM}ovo5o=81G_WR5
z_GV3~{6045v+8|?IH`*7+L>(lYR>7<IO2zJ{&-M<-^UwB97nA0oI2iB?>G`dd9j3F
zA#;fj+6>Oq5PIQ%Pa4(m6GT+wyH2VoexIb1=6g%JO=L8Ishn;J4B`@>f@^%WtV-iM
z$l81UJX!2Svx*MBazuRKa!^b!Tz0DJ6Rod0Pb9vIXwIK15IBw)v0QeV8ktk4NgDh~
zyK#fB*3FnmZx23mS`Sqlw+F+D<vblW`w8M)r8Lk=_eghtt(9a=cc7ISiE=?Jm$~%^
z(i62v8Y=Ag-n0>+`U50~$u&+Q-<$TrX|3U9I$uqs3#F(Q(;MZf7H9ed@u5)t!GoGg
zr58wdRDZMyX~H=RTTW$E!^;G|18YU|U0WL`{Bf#S8aZg#BW@|c9#uI{os2ZO#z^5j
zb;dO=RW1~NOKphuYIME>uYKj*_F}j%S%j)`sq)5qP969;Gr1;o;d0=^Fq}F;OW`yK
z;uGgwkRY7fQF-pSb<`tpx<#oI&OvfIr$JKIasDLh3a2$Z;Do2i_RM#XT-K9oe5u^N
zWyl-XpKNU5bjzkrIG1f2oD11y;nc~#WqzOHV&YnVU$vd^r&<J_9Px^KP2IG!oT5z%
zk12Gsq44KvUK`HSuFHgOU3#T*Er_`yrH*K=dE_lH%bChGkjS-ULHLsj0!VHX1?tHC
zkBDTMASUu?5Xh*Mb8uO<Q~rp!z<m~woT!9|_+mJ9BqZTaY7p4*`@l0b{-h?fz^`#?
zM9yu0rwz|#fJ95}s5mQ;(+yIsCaWQG;1u;%aXl69P2tbueRQ1GKKMfIjs)X8-&?Se
zN6|ngd5J{Rgi|LWLgDh31<sKp$N@QZvJ=anG{YQl>J&B3m9MW6>5)HAaR}u(f+Ud3
zTSdUdY26eYDn~?HF7coz_e|1Eo#<(Eo@a-DK*GdQG&hM*r1=uKgtRbtM7$t8FOiN=
zjN8$!8n@)qb(qL?52Nzc1|k$Ly+Asr+t6Xch2c7<fmyO9%54KG@chRpD?D}q;yRTO
zAih$H0!fxiuTjB78dDgp%HTWD$Ui36Xv6KB(Wdhl$sq7eRC<vGy9q@lC|;EUVl#h#
zQlqM;a%~LU4CjJ@iRbshMQ(B*3lBsQ3v<AC;8Y8oqVO`j)(M*&xxF@CPcm`irSPg0
zkckI09?2QsoA-4>ouHcI_X)-)PMzR0kNHf3AaJ@RW4XOH$yHt{F$ogFDVpIhCWsMU
z;{p<Sb<$+3CbTZF@m$`D%9?Ybn2{!bh)<S3Y4MeNCR2j&*ab-9(V(fyHcp+YdY@dQ
z+l1Ry4<1`0U(vjRVH)I`Tcl>dya^I{jt->roXIpO1DEUOkugxDL%q554wHBl$BY1a
zGsG@YAEk~}zH_S#MHpNXElh}}zR`*m-mL)QcmxS#!Z|C>aNh_dM*gJYHLu8jSP6mm
z7c7*K`)zAn%adzDv($=NYwF<7vq1cLtam&s0FnxgtE>%)+iUB@a|;Xq7kP$ky|%oD
z52WP@0`JJ5$9p^3bx?s@WgB!a{62_~c*bIb^NH%I4PFf{y}=J|cWl%MmnTTz)gl|E
z@SYKn!7ITwLhGod05YR!w1HQDm~3=Lam|XMHohAC9IwnFuIq}^4IZ#_sY;<JF7a6l
zkGyRb&8s*T9*}3^HbYZ6f3mf_qG_{B6S`$zdE9Gr(3xvV4%PA1Kmp!GutlnyAi?rD
z5f9^eo@Ehp@*bG2Axtb%Lo&g4&@>DwM@--mv2E7#Xx%pFc%KPK<h?e3+k)3KWN;r0
z3DrCr1Y#|hw=R9+OS(yk?*M*@SHFOi2@M7!6s|u5DInkVu#>(fkZ!n)j^KEH>Xt$Q
zl^r0iQ;P&7@CYBsB;s<=d(Km*oFvKxr+jGiNu5@TSJ{AcUd?k_6wi#DcACqz(;;zf
zbfN;+IVallX$B|K9F+qJhe(Lq10bIFvzO700WE*m{hf2EQGbd6w(|~@3p0D*5FhYP
zF_6TqzYEBJI1+W@*54(Sxc_lUJD&fzWTN6UNH+40n#-#3xd0$z@YS;Jd@jIc3A`2n
zWN;tsGIDW_RNRNtpvt6r?BM5k71ULD4E{XTr8K$58$Q(nr1ILJ>qw2b{s0-A3-B@%
zN$h_<JFdY!)QQ9c;gc3_7|pW}AieV^9j;6)#qgbXQh*Gh@tapLzVtp`>%g-^uYKYX
zq{k(EZ$KQc?t9UeS0lV27^HRwNay*Fmms{NxwJlb{o2Ej^X?cB!+ZSRn!)ATTbsFO
z@=irQf#>0XcwP@A^SUyS$}Jy|!Rtugxk-No-FuZX)p`$hYfkI1G+-vk;9BCtEPC?6
zWv)NVU>wg&d^Co~NI)hw-bbf-T?~lfb$=hdooESc81H>RM=bAjdE{o?9(d$gyn^Ag
z7ak+|93uBwz93efAlC4TCXkW&-l}fAzwN6iUOxcRczoci5neg<)jA$M0g1elg;ama
z@u<@`RTQT|)0ulfAj5NQY<BRjqDN-L=TLm}gZoC`b>XvyzKbYMgFc(m@yLGo6bO*W
zXW4vrHBEL96tBPep(2mP{0N=zI>7#OKNl3@M9R6a1nboD1+5~Nkf0H$(&q`<G;Ya*
zwuN_ngLZ~TTtN_vQRxlhCY=W$iT8iAfOPH~m$4@vaRuW$=TERIa%%}B^A2IKU=P&=
zAdT0&gLQ$=Mg;3SuX+R|o0HX$Y)+oX>CAKIfb@ggQ6Q5z5eSmX2@(_Msq?i??i134
zZeE0YO(2a&HNhj_o*>{imw5PK9`6KrT3(q8!35s%3}Cn7yAIy-Xgx#`K1+W9#PLi#
zL=$;#2Bh+{2q9v<$u%Kzn;4)y6P_kDneYn9I36E_thM4AoW=0@osdo9k!r}W|6H!i
zGBw8$iGlCB$lz7bP?ky(MXFZZGll9Vl~dT{$^1cRD)GKqXh!(7aA?->PFHAdGU*2-
z@y>T>UU)}f@k_i;9=e;%f`)${#>7&L<v7s<X`O51P<=wT;VPHJi13>8G%5pU%MsWR
z>S-hLEKX}6y;D7f!Wo`PL|J+%6Qno3H>2|88l=Izi;Vz3@F}-weBmAGXd(1`2iA$<
zG_av@>w=^p@G3!cw(-f|=%w>{{{`Xm@kk^x_uFv^UihwqE__ZXA|QX}(ME)YwbUlY
z=p)bGV_a+nMUqUW_F^)^WAd0)&3v_Dgm;R84Bl0Y#TGu(7>ldCG8YScAm3YEgq|bX
zPgbiYHNJ!DEU)**raI4qBeIW)ZZ}hTza%y*6D?_oys7mEl6iGME;9=p>AFtzr`ue)
zRCQz-{CS4Te5xf5llgyaL>%7nJUEUe6q+EKcddXFUPDPrTYh>CAw}n>6w-3Qk<u+G
z_$;cY2_Xfa+W?Xy-+?xcOGwf-^FCnG&htr3AcM!s2?EcvoFpQ=^N_?SK7j{BxRwB!
z%wMNv43R&nSxOZg5jCNk!RDwgAmq!FCpE5eFO%#tA2>mx=RAM4vrIfWts7s>iJef?
zdBvj*Ae#3VlJ|*Mb(3E*Ac_P$5SOapE1ytFP#`{O0i^I+bV|~>XG#kjsJ#YKc$}D$
zo#{_ndeR(mGND^?b;6(I2Jg8hL}L7eOUlY}i<Gi1yuy@_N$`qU%3gEppNjJQL_Cl*
z(Yd1g>>4Y)dYpd>qIFKusx?1f55#bf4rK5=I8{G6*^rQZ@Ct2ew(*K)>RKk!+mSo-
z)w%?pQKoJbk8Xijp8uq7CjpBMbn;c*)LrLucBzL+2PzGndHpR71oB*}hSmJ^bQ-Sk
z`T>y2Cr8o<mG{xJ(t&4fS^3H2RYn-kV;3aimS-$^88oL{$T|prpTNta3eSwPXvN<O
zkVPh){4BHnX%;Iy2h8FIui<ASmCw3oWAtQi#wOlP%Eou@%`&#etxH~f7GKS}$$JU1
zGf22O;)I`b&G0|`<W>$6_?sSbfJNd`6>Q@%QV!Sw^?*6}$n9H>%J4US<ft8=f(J4=
zOPXU4?icbBcJe2GUBXVj>!b^>t>=VHd~zSBc<&=;sl1nyGeQTxx4d)<C{hrraxNfu
z=M%ZPD$UQi=BkBzS|FDD?Oa77f70qQxBfr|?}aW1uZ!hoCldoe3ZE0o-RQut8NPB4
zSe9vbDphdx!fOEq5jo#m(a3iBY8vWI<?RWfoGV%jKS2S1B=UF%Vla_|qMhb5r$92#
zgMoBjaW2|VPD2zyjKrUGIaWX=v51*Ow?G1~`<Lb5{p1>}xwS0fI-eLQ%W2lhYN1SM
zjhu<w_@(T^yKo4u2d^XHCX;FVV)f*N7!b=V3uQToMs*IMW#+Y~vNZBJ=bQ;V4k^w>
zUM(ta>4s8&#OuuMPVq40oH|}Q_h!Z0%%^mU4~ECEK)m5{5Y%M;x&%n#<Qnh!3}A_B
zCZkaY-sJ(3c^9EXGdUwrU=4&*C)zyG`e++J2>@j9x=V?^abHrBu7lG$nabZEQjpK_
zYIw;i@)HLo>o&1f*=L?(6y#<+ek;XHdIhCeH_=9aD9}WMi;sLhq%`GtR}o0#{gTpV
z^P{CL?MsSuY>n@rtHyi$2-PR=l`0+XXR?}u4o{HCPj>(r6X_i~@C>;OnOK*qZ055?
z5Vpw~C?J;C{i~?T-zZT<55w;xh={3FRdI%QUw~v@tEu80pQWy{402J#=!xB_#@NZ;
z478Dd-^VyVp{T`TeAm`R-ubRhd44jrIz2h7SRIlxzBgw&KmA!f(hN=muPx79fpi{I
zRBtBT*y_EWtQPd->`M(ca4i8cCc6$U@!Sl^;1;O{U-)@VAd|O*RODHF`V2_r)s-5>
z@C>X*)A)qS6EeAGHeXev%e=B(V<<dASrA@-`KbxnJGFcZ!mV;mh?V($67-bQErp2>
zuFK2<f6^QhKg(0g^1?ETRI7Mh8HnXE5)jAV>rty~JhrUWPv%|fayp80yQ#q|^|gtf
z=v;%T;Iu}zz#~W?iO<Z{7BcUa*KQN<JJ+Em?<lN8cRqz!hjd;!t;jVews-i!Yu-(N
zEiYl<2%-(jQxJ6W#=|B~c!Uq6@MxnM*f&ZyAckvWv(9oZG-M^b2H9W|d4vz7PY|a%
zkJg*hleZT(XElF29*~JtHRlwMgqD|7@solG#WbJeYVZquas)`^73UU==5r1}8t>S&
zFzXbzXdQnuPmA91K443F;?a>TXg=?M-w@&RUT90pi<mePET123*-tvSt=Pe9PpvKo
zA~+HX_dl(<JlVlQZ2qKuHfhnivE0T33H*&LZRj?U>mi0uX18Gr*T$~&<UCVXeo{y3
zS~S<wuASl?b|8_*E<h4LThT=-&(VQ&o}YF^#QcqI-RQ|wWjCVuJ7N0LtEN1E1>*R0
z6Ij}L{kj|Hc$E!E;5kNLUgI*^fw#Om*{vAvWx9pVasF6qxr_oC{G?#F&hsp%TVMIP
z4<MaK4t;qc80VbRna}EWCxPo+cWyFe-MyN;QKEZ2c;wxKoxCNn2RE5u?NJG@4fd!f
z5pj<ubGy@%7XH?To^0fk2|YP~&nF~%^1>_5JtME@t1agnxx{B{`ALmlAR?TnRVHsZ
z>(v*28oW1AyaL#pi@bW=yAoc(=v^$IL+Ksi0jG8MiD&$Mn8K?OeaxoD$MU*8y6dsL
z(T?ieP;PR*aV%$@_<giYD#2r!VdD4EvnXw7-?%*(I%xj>odL#ii67#B2hUi>^4@)Z
zAF~`@o*;>LV~6pTd(B~kaGs9kz%{il!y)kIoO7n}9`f)=tT_$5G(P`^v}nwbAiM_#
z#PdjK1mq>0r@;w-Hv>W^$F1^+K5>r@r1Q?*NHTf%=19iyj^{{-$vL-^oA{Ce`^IZe
zKqhZR17i6!>PVsUF5Jj}R&x(Hmf2Rm>#7p>7^8a2<KEHqFTd>a8>I3ns?j9z8QRfI
z;I*`Yq>0x<MmL%FHAnZ8H?ja}yyh@^7`#d_#_}dHj#xFJ8x)oMZ6Lz4sWH5JuC~5@
zhK}EUv%Ok=CFws8=f9R8kGa@BUr|@vo5N;x@x$^vO0So1+#G&?Dx2LIzgfTCAJ$(Y
z*N-=6@YSnl;J97g-;cxU_6$6+bo=yUJ#GE=>F@g=R)=?=*M0fMfqr_oe*33i)>?i5
zXnpv%!~XH%8U1)zJv{#?%=s(&W!!Chwb`A%-)zsHA0Kb7&sXhMhfiNXkGs1+e>gnR
z=`(pgUY*{rw@*KRuJirs=@)69zPsMN`pfk*v2X7W*UJyjtbbTNJUsts&CNUg=JI~?
z?)^zGKRI*4|L#_YpPv46^G;qcdBWrwd-~7nc6l;<`{j?%pUL^NFTQ8apPA2}+4E=i
z^Jnh-nfv^iKY!+z&)$D}SRd{;yPw`%{_^MY`P=>d{r=;B`tIL9<G=annFz<_Czsmu
zuPs5m_Rk-@{QT1K;=|SY=M$)Qw>e#293M7!+tttDk$NumzkK_C^=bcj`qg(AZ29uT
zuUEKRomS@`@%h!oPa^)+UY1{hy4^q=uRm>XKL5DWzdUdJesersuOC*2)oFh?WB>Yu
zX?JzEzy0nkZuuRm=g<FJMZeS#!>eqxd1ZSXucE2#RYx3Gk&{Lv{eNCG?fk23-~O)v
zw~(6w@E8w-Gs$P+000100RR9P04SF*!vPtWsI&nye?!pSwcVy^QsNYXJLq+6u&e7h
zi6ORw?Tn4lcdx@v-Ek-JNl4$1>eH8*vQOoMc9sF%5Xh93uG4m4@c#CKO2|SKwiA=H
zih*@-YM!6!E*v`%MTpL;OH?LH4V~(RGx*7ARaX15K<B)oC@1FwD(PW|b}i>cMfQ}p
z?G3V|QMTz!U@*owA2AMcR7Ql2o0Uz&-gzTpLO)>F_gX6M<YJVKcuPou<&yN~!%Wbu
zd49KFRUaO4jkb9GU&~crC+!9==J6YFQ{z9Xl^<Tw+yR%7n*kIA*`eO!x5TvpH35J1
z&kSt`DidKt!HVKrNba^^la?f|+qbvvwr<1WW9~Wkmve8Dj@P;jR$!f~OonJgPyj|$
zQkg}D-it{%LV@$#NM2S3GPD7Q#`)z%DrzPw3$M1S!FmNQXi3H~QD<oBea$d-VhNhN
zsCAfDoL5$J-yZD(*Iaz_1>l&F8?1l9bICo&9Wbm1BHAIP7?AakEqj0z7)sD!yu&o2
z_~7+mwL9C;D@RVP{HBJp*1b5m*G>(_rfH%k>8)*kia%%5w|>-6nQm7CP@YP`gaz&^
zn?G?VQ+!nDmREA;XYHQn3grE!_XNT0c{+V5Qv6r%S}d$o_f9z_$*4CCm2^Ni_x1{q
zL7SItr`szZ$-`qYL3vDK5|XQsCIyYzb;1bw>e8L`AAow7;eTQrQX10Rf)bY8of7we
zyia`kXYyZ{kedM&w@kPJQw0UyGF4{nx1_xRB>{iIZrd;ny${$w7`ivI<0dr%Ma7U}
z=wVv}#BI--Oeex*Ns!c_+1HP@8&6q#nooR>^wSebvD?>9XGicsS+gd^EGHQlX&Yts
zYw}osy{bqSqA-on*1(#aAdp-B`4iiE>%d0^Awvlh*5nYQTTvS10Xh+8m^PSct?xuc
z==XorwylzIXXQ%=CenGHFDbl6Fby<UZW1J;;_4XxSG2LR_ZOb(lS4huYUrF65%`b(
zPd#gF>=>O`R@*3atvcX~>qUu~3A3%(L*UDb(J^Ar);EFAt74Aa2-xNzypR$5$ctrO
zmJ3D`ihXfTD=8v2nSWQ(+hE(6Jq*sXUV?wow1nYpb|7E8iYJ~knnLzX8FW#U<&x1c
zMvGqTy>N%X%bR6ybi!;Wg@z3_gSSElVDvA+?m_f}-U`*5JjT@#WVAjD>Nk#hPO_gO
zK;ObOISQ|YiDWDr!4PW~qUW`W9<?VtglSnCt7~2i1sGfw4q)(r@ufRFK|^?G@os4G
z1G7bHcJMG-P3)x3&S(6iy0M)T2AfQ$2wVIsJi6N6^^5Y>!%H%m{<DgQofER3P*vsC
zO}Yn3lI@U%1_wKhWg^%;Uc&4C2>jRVp*h#QPxPDhG`?P3yjhOluu5kRcVT)~{0{(^
zufPG?3hN0rs26qt0F_vmlfeN`0(pOz<iP<aBrX)7r$`n60NZT<01*HH0000000000
z00028g#iF}Z*pWWbC*EG0UMV}!T}lrDV&#P!T~!0=$E3x0ZReym+ry=B^-n^$!FmJ
z00311000;O0000000000000001htnh!vQKB*`eO!RRI71fC2yj5dZ)H0000000000
z002X_mvzGdJps>`v%>*E0oRxG!vQ}6NW7Od!~rG+-ZE8Y?U!K00TluwzL$5z0U`!P
I!2tjO03*+(NB{r;
delta 102881
zcmV)RK(oKEiU*&G2e3g23KCEG!iaSN0LM|2O9>f&>vP*iw&(Y)+W(=KnL9)7&=8-Z
z#JN^V&y1@wo?JV*`7kxt)dZSkM<AL3KvB&8^LtKr0|fX0X<3%hP*2swA_#zJoZor&
zdHm=1k29Y<urT5xcw1{btr}s0D?A=d-qwEk`R>hGjYKgGJnD;py{)ZSRQvwofB&!l
zTn;^d;Vxz@hzZ;xh=$9#`?fZX<N5I9Byy*0Mk8m&-B3hg96PR<orv+6yX<5sLhq#0
zYPaNH^H8`fis1LIY4AX!n!cU;c--B*kS^h3ykYOeozgI7kL!E1Kl>*Alk=0a%^S_O
zmnP;cfG@`)oY5FQhLe*S4euB8H*nuMjroXw`#fI3EnBC_ZN%H!A`FK57H`r=!Apkf
zQS?8_72z?zt1h|G<dDC35;7m2K?Kp1&(n80d+Ocb%V~0xhoVJ1%>3lM<-FbdGC<rY
zs;qB#%;RfP&V2Rc#dmDCj-UwMBE906N67zX@+33LgY^$T>D6RjZoB_^H|T8L;51@?
zpLdJCzQswjnyp!5IiGwPm467uV!pl`|FXNi3+~f<p}hKxdug;}5gdKl!+f05Ik57~
z9lo0cBBUc9o(O19fP6$!fFPIDE<l=>x+f=VUnqY3zM?TWhth&jwfLAc6IvpF%u%
zbKYs++@=R_*qAQ-c>9~5@KEop*BP{b<uA;C3h^HvcyNNx#&Ajhbbx11X)uujA3vYq
zNx!*p_<{OwYj~nM7OLyRI}yYYd`%;l1GTQjBIGP2@7Yq`>oSP8kGfI*ki6kY{r5VO
z|GL5#Ve+eYcP;pzIxYGmIodu;j$Gpd<c|5UhU6Fv-u!TdhfnlVpQNkDU)X+smxeEI
zIgBs9Be!8F!Xa^e8bt(f2+tCTm`sEKf=^jMsQ!*eBxE!3z&zqD7>NaWe~I^s)g9IA
z+QQ-n?9}_~-3|KvH9##EzW|H+T~-N;J`WgacTWiq2nbO3o^)CuC)=G<vgGlUP|qVY
zoJdRKvk{m9+ikCZ+U*+#wlAW8!s|g3D&t`^r!MHudB`FbKCl`<?4$4>KzL1gG$fIj
zv6#<TL>K~(z+Vxaz(<z`WW>l!gm@}^h^O$0J`#^dbDyqA{j5D`H5~HubP+W*Mj7}+
zE0V+GH+a3z0uNlmoJKKg)oZsOKK%0odH0^wnq<5PTyzhdfX-NxfPY7S;t-DAs~4nR
zAK_zE4kJEs$h!cUmE$!{#kR3Lb;N*yfQ5({%g;+1%8$t{;HAfe$Eb5+AqY5;6ZvZq
z!|xdR6J9{XBAzc|0!B1J@5yIz`VjlbfJ8X!^*i16Zk1#LWgK#eZ_<y1#4$)oc-=sn
zPmxRGn1zQ$;(51qdeu9BS}EsGw@P`J7&QPm`$Qu0r6*!1LcfE2Pd);ZDLwl@xsUG0
zUa5vq7@T%)T1Ifr?iWB=1R^Eo?$#;|SkeV(BNj$aw8Jmkoz8V<7t80ru`GYsa?)?i
z@Z~8xJVf0e@`w!y1~Q`I$%z}%zzN^XL-8APV->JsqZ44Jdo-DUP9_t)C_H)KlYoI1
zicUVrn|zuuF^imOJo68ObOVp-#h%@s-u5N9HbI*I#)9-PNc9`51V~YBO-0E6fCW%L
zbqJ-M=rKGJtiUb|#y)prrO?1a2RWZP*&!ce8tQK|H+xIo|MKIHghp^D2{$lh$K?Nj
zJ?aDt->)~wn?Fx~;=keT@t*4&gY6r#-xoY&-eI^`6xq)R;GM6nb=GO^cIflpyx!Ng
z#rOT|-c|qVCHVfC*c;uM$>h@uT4c|>fRT8d`ywBnGM+5E4QTRbKmOaBG&oQ{Ecl+0
zgWK({^h_-a8-W#%#<bncds-GE79tsltSme%JP;liE{)lL#8?wxH@cqyZbiM@e(UnA
z`#KR{QD{5G)}f|&$!M?_*)mJOQ5givf3{fv8dx_C782<y9CSP^I)mQX;BprN^548}
z{cgqsYj@c3*@Ok2+&eB7d=+tv722)+Vr6>OWIA&5nEMzfzq|bL<E;gS1%(8~l6&zK
zP|<Sw784eK6A}~hpv1&e0)G}0Rm22l_&O2)197p>7R05UxBBOU(pZ9-rCqd~L4V(q
z8%Y++o0c~vZw4ZEQHNX0&Z1(C3hK|N6i{JxxJ86Q#Ej7Z0P*vOU#yg{n9%;;lt=e0
zbncNdZ%HkRy_yAw1xF4XX^_Ao!y@AXD^HT$p&T54G7BrWD>w<+s@`%s{l>oMw+^A@
zPJ_7<cW~{N{<DIjBu{q4&MhLKSgzURn{UsjFOq!IZgno(SC&)1wZVZbc|<bBUJ6$_
zzhcMg4PU6$xz-SqJRfs!WdrjT6kDKhMm%tW5n!YL^q|V1cY1rZZi`}pvE${K1dQJ}
zHiJ2TJ{jlkd~n-4-vq^Fr`ua2B?F55=@JzB(CYz(dO0S6B4P{!^PMM4iQeG!?q+9V
zwv`ge%P|TNDha1+0bv0l0g-I<e@P0b#YD9*F_M9$L{>Zv85RVFI~p1UpIgcGnqg4I
zO0Gx8r=uL%x4>=dx5~S!V_5d)wmUc*Z0PiV^H%pZ(=#;d37!s4`=>cR)_0J1uBJFQ
z1_r>eaE4#FOHf4cl98Z@U7r&Bviityc(1HJs=7Y1`xq5>FMx;n%w`x_Kx6<RFy*Wk
z2%A4{0fB&sxqGknIIWa;K2Vg8{y^%8xgzi+#fuF)!$7RyyKhMnaOd4#w~RZA7wt-a
zQNoH3t5ZPG@WAyK9&`R+p|FWn78+S|L(Coa7>CrE(GZ9r_q|%StBpkmqGKfDSj=7s
zAlIku-e=9?v4-haF@d6(#?D;ihb?8We2BEB!25uW+VAvE&srs1%|#7_cCG5vu>z*r
z0tVY^yaEYhZS7~ove96WsMo1wy)WB;!95-HFE9J15x;x_>`8*l<57$iC?ynLFQ$0&
zsW1nH)sD;=i`Q>OUkPHFDr0-oe4(tbifT&7IzGA0nN2S$v+;dIdQr?Cm36XDttnu4
zuDX}!r)2^57o-@qJDUqNd`(Xl<ogp|Ehq1j@8!xFeJW&P9!n#j_VnBn3!Ybh4IU>|
z8vvs-iz17n&qR@K?-NbM^6AE>1yMCXbSzrz^p)p%&wF8EnEOi6dzF@qZPINaAt5md
zIrjfJdiTXPh51}Sl(i|0Poww)XbBcF?<YE8S0SVK@~VB(!$YLlGVf==gkeG=Wbq=z
zgr=u%(%P*0vJqO-)R|1?zQApNIz`&Uls)V8&e~~i$U%!BfBLIrcKr%UjDeVrw(G&m
z_q<QaPP^)N2WO?Mpg<eE{4f3D2$^8Lgo+ygT1Qx~(c(ekVM#%4%tIa^DvG4<`CYGl
zQ<`nitXPxw&b~)~Es4N3_FHCi{fKM?pO2h5ji<#xKJT1fo?n**$X}3ujIh=3<aqU?
z-@$VrAAw#?;F}Gbi^l(ZDmbdMC8pmfKl;5MhQZg|`+D-@BAve@m+$^j0+J6}vQ54v
z`Bx|TZq4vrp$&T5I_+G3EgSTU@cUEYWA{6Qo39nPczzn7hj_pWDXkn)LC_8viAVgy
zNUB`yEkiYurHeVbv_$xSYqf~riMX)3*)SPJ{z|Z#GQXtfFd4Lp(ciohlc9W62Sebl
z2(Cp`E#YcJND|Ey&6&%`+@&!WLDVD>V?#2H<9ReZIngJ{$xja~jF=<B$%#((iIT71
z;5{N|8qfS^$G`QGO6?{6dimk~yZ1kQECJoEtMr0&{~(qTap7ryC=cLmllPZ|2gIW>
zZU3uD-1-XsgQIU7|0YZB`((sO$Y$b!dGP00gpABX;j$=#i<@vtJSCK1>>qDd2N?vX
z`C^$LbACLoI=8Lsv%baEPF(4iQ%YR@9bPeu2nB!y;P$y<HebX58lT@YL|8m!L=GU9
zwUXMasyE1Z5xA&-z=#-=4@HXlrqjO~ob857`QFH$=8lCf%m(($PV<kz>&50q$i^&W
zfy+)NOw2xMqL;<ouxjU9N9cYQWJ(_xC=k>pK0~pvv17TxB8EAabDz<OnHmGzy(Twg
zU>^X@#?+0$9dU)fm<5R?pb04@J)yvE%1c>E`HzFhQO?JIQ;s%>ye?4<Xmfz}m@H;2
zh#hhVcTVpXu^2C#gkmuVl~M$d)@A$j)EsvX0MDuwqyzKusQ}4=A`b{cBQoY;6cgq%
zd7#c6<}~rRJhR;-_kmcV5gJcvtPXa_Yyp>0KN4uiYKIvRBtZv17x8!@-RFiwP@SnC
zt&jjp-=Cp>7`u?pgf$jkmyh~0D1`Nx(G<fj4S<(%UNveAk4JN#u9QAuC^<mZNO?+@
zQx=c~N*Xfv#zZV)@I?-TXs3s<A4K`nEr`l54ZkBFvD!WXR7C247suxTbFd7*DM7dX
zD>Kn_zg4PU#9hC8)Bjre{J9sQaIV*VcL%bpmAsFC!*2+FmrtfX{69{<STP?Y>?RMU
z{q{`N8%?tAu7Fm?M#qJQ3{-#_Fq<jO|99&7Y=vP9DQe)G#I=Q2;ob{yYRB-N95~_R
z_XUe&%>AT2=(gME?Lit(2TAm15%Bnp7{9^aQ8$+c2;L8GN9WAp!Po)$<hWC~8F%mB
zWISbmb*B^gln(g?ZbroDH|EA^Y=ycmo5##+lF<SqEdhGm7(zh-33kpR0u!gsoC(-R
zSj#yas&itzSzR5$hpe}pPLtGl;7(aoYrvngS*=OnjV{<f1AnEnyBkNu@~+Ohy{nx$
zr1@{`Hdx67Vv((Te$~D`dl8}bEQ`6(20_Gs*eoKkH2FOdFcL4h3w~!neg&Li_}gO-
za0JA^3{Fxq0u^u%%=UM&WWXnfguRAz`QFP#AYjFd5%^b!a#l)9+e0}krG5E((hC4~
zV1rhO%@Xc$z)W96O=)~f2Eu(J%7ci^RuLHKTr7@As<=9BU*4|G>l}@fgIR5oJH3j3
z+T>eTBz&bfZxI7rwR^p*OJf*7ZHH=C6K%_)wOyXKuP#ezA<M|SKr;_Bx|(P;9+13>
z_e>!Z5=ci7-jdc~No3Po4%@XmUrq_J{T;dd@WU_vxPAY#&2O<`pYK$9%dkotN4O#~
z4(ZI)2+ba$oykl>*G=ySiE$zAdS`=w>l?*?CM@#z1r1^z)0jV0Y;X7hLr0NH1FI0h
z(8Z|~omT5j6_u?=(2hlh6CsJ0Li1gOj3@U*1w4|aYl#+gv9Uj{$mo!p#oWh~aONel
z6O%61d3hA0{mipgk+q7*YDnBE4XKM;s!<$DrK@&RrNNX%5E+AHSiPZ-9K?WsKAzFl
z0iZDCuG%|@0Eh2$2hvpH2+%1vFhuCdTn({K3}sT>=)K%Easf0N&^uFw@D>vmun==j
ziWu5JBd>bdK~m31FeFJia<d9(DBB51=xx@QS@BePZ(XwVkUTJ%Fos6QhFnTT!*0Qp
zwNchCLwb`4%^H=BW~B8}u{vIVP!IpWygcB3#(}8ew&kg0zeC=QGbw`^qSwcZP+nOf
zX)kjC!VTNZbXucmPjfUIa<?du<ObbF=J4Yx5j62L+LGj(II!e4lH5|+t3=$xl<zdj
za?0H)%JUg8y-fB#6!_63FCld+Xu5&<%_tCPi(`u@)V*zWdpBjIkS>6KtrU%h;NufE
zo5!ovrP2<+v!~eRC8{|{9Yi;4;<M?r{H^@{#urz7SQ+pF`1*Rq9YqvwpZEG#rEJzz
zT#^Dx57tUOkIMpA9qXVzyh3lK9AE9YtVlh%YHPHj(SX@eX_H&uiS_MYg>Uajk>JRL
z`XTXg3b>BzsfsOh|KO{CGyu4c2X2{?BxDj_0fq#TtuX&gRTh&veo#z_1?V)4C1a)`
z0(Bpm4?aaD;|lmQ{kGGP-pAGhlefu;lT@jO*cD;@7+V2NnRhZ`YA~=WBOo8e44v5Y
zv5G|A`FZ!MJ18STD`wQ{lzAkt#DzdN(^94clnIIxn9JK2LDhVJ>@b5j#E=$e9*f+N
z&rvs8)%Yq^BUW_fg5R>lYinx`;e)Z0!mJ-dMO3LddXIpT*$_`*^AecAozAJOT(x}u
zwb4^BQt#rsB6hg#VtuhW>|m(%aj~}erPxhveNVBMlE%^%(WRLfsD=w3*bX*7%-N@%
zZl|>_GPg`@u#NhEUB+FM;NiuAEW>uF0uj6k7Jj~0m+3T<%qq)XH1p=ekwxUwtXo1#
zf!U$kHfkVhw^nVh55Ghz<-a*bDxGy@4?A1G{SpGG2(O)fuX9#PKcL=j<nlU1;^a8w
zvQU5;)vS3p&3ymZG{q7Uu5e4MDil%)igL3w*|!p5#_HRDKU%BYT*oj?cf?eTVKc&R
zb;PUG5r(zDuXj*!tV3lwG^TvC{8U;~*m7^*2$kBpJLP8Ylu83>Di^4_TInhQSgde4
z-Pl<(Q|RiPm)w4$6}ndFnr<Z3yIDz`q}OlDWZeLMOn6L|G*YSF#aW$(iRw(n5-dy=
zD6Gq~E1DjEX`r4NZn~U1($9&^56S1TC0}m7iV`ldt=1%0CtDxO=^fGBOw|dp<(Nv<
zyUyudTWYpV6`B9Ws>p9e$nVItgk&=7Bhd<zkge*CCYeSL_z#O+OrH#O!_1`3<zWPT
z=5DFAoQ$=ct~%|Tekq03Pa@*7p$D&!$_zm5Vfo~L7Ecl$L@}e@&RR{G0__X10pVXd
z!e(jV!pF49kyymUXTb!s#bpj9CWyyap71XDt;7k`W2SKK)CwiXr^)2gocb(|S-qjV
zd!y+!qmM;zyu8UUd`9<ftyRTK0Gvp&sy+aO&Lk>y2wZ=MCH~pXD<i2{lq0kWUYq~(
zEE>&!XvVOq0|pASG|wlO*;gTXg*v%XBuhvBG*hjkOc9liIl>6DG*DDelKI?UVF`z(
zY}Uw8WXJ9jm<^T~DINp6XT}4j%1Dmj;1V)xC2N)8*|AK~mXUwEhd;v3=?r>jgUb!I
zwtog}{&YJ7w)~v<oovJGgPk9E4pxls?>VM_PSTTF&^+3>vUty+wB<nj0RNg+_Moz(
zWjBva0{z*K|Mn)$s~qRkEF+@pC7aTCD!8_T*KXTcXq@O<6d|E|-tDzpyG4pIFL`Gl
z1w(Ok;gDZL9y3y_Az=fNaC*poU+{tsIAvdzE%PgnS}b#!ABws$I1EGLiwRbQJLC?3
z{RdsS{ypkC9JHlosbqr!DW(#CLpF`bcD%&<Vu(}~z!!1S^-aHZdS1piR59N}HX%q0
zH6tJ!8aX4My7vu?h%XtsnmEr^l1?mQAD;>PUaD*1FQR4->T8S0YFJ<hd&Y(W7hWx*
zk5o^j%-G002j_huV^|aHa6{>IF-{|YK9aMf^#~#oH?QicTt#{*LdnJta5@d7g{vCY
ztx~~@10WNDts7I`UH<s-b7Z#qdP6uQy+|@rK=XcKn!QELiy~$N3?tCKG@h>UaM+@2
zx4V}EX>(cQ%^XM+#Wa@Oi>JjiWlEdfw<lPxvfe7Cm~)NkNqroO8EK(3Za0d5d(J6-
zE#ERP#F$9vVc1N!PgbMoK)IyR?jxnr;I@A^*qL3PmshW3N7)om7w1T6qMEXpUonz+
z#CnoLN(_~jKX)&3BSd5*`7Wlk&?c<{ovOrG>oj&1j7-{;R)>Zcbu$@B^@)wxoXQxU
zb}64FoNP8d3!^0ErKjEw+f}N6iThsJPNAb!t^xPvdP}%COwGl(NDM2ASaa?BtRd>L
zF%J&qDVU}SBsDQoBNtnKLhW8gvs&()_TNmUMm-Q~NbPq(%N{60_t{hukot^1?&oV$
zNU&E^mR-ez0$Jb~%1AZJmh>Sb7B87fD8Xqzs-o7Dq4S8e$(xMQL31d7iJ{AjaoXTN
zF|8m5<@7kij{9IHJtCELVv)|j*L$S(1_s;b@DQMvMW}|(;oU}J5qLC|L&Ru^AIPH`
zc(@1(i+ajyn8c*rB%LPdCdGAWM+sQ)^OqsytxX#=DwW{-jub{(su)%&-!h6~DAfp<
znDXFMO_|$rtaX%Re>IMOaWRn+KBEeNB-5Rw)-QDZH+D~r<dh=M*;p<@+et6U_XV=h
z;X)X*h<PE+rHy%H2iqW?_FA2rGFFs)Zj>Oim^{I%5jZfIW*AFQ`@ILE3FeV-XMn+A
zQmL3jhFBFMRiWOllo&KjZxw`Da%$|jaLWsZTJ{GmwzaUGCipCWF$_b5O;I|YJZ#hP
z$|HY@GVeBKKgw9H+6~7+sXQp6VtJ#j@PvEka+z7xaJ0W;SZWM5bGKP%W^k3wIx8z>
z4uw!n)m#qfsTrd|@t}*c_6*qru|Q}i014Gr>(dTEGz}NQeYG?f*csTJI#d&weI8&-
ziAf+r>|I#g4aQ1;ek=K{<Ub(!ccvES7IR{q@mK4NOL-gdKdeJu_BG|bbTO8yvXE#u
zIF@!(?3UM<Gdas|kIqQy`v{tH%PEiU4^)erMY27s<^h@I$8%Z@^LjN*AYyl#X{hra
zKd&3%d{5rRx~GoEqq$F4xNDI;qp1(zc)t*CJ{N%;Gezfr^H4nU85lOI7bpfO&-?!Y
zrv+o~vA|W0t3!C^WY=THJWd0(ks|N9K7&VuJyJJ@d$pbZ8F`eA&O`EF3ktpoXp;|s
zBSz}){_8`7oRH}EFs=uUCRy@$N>mfD_0T>kq$5>~F{7R;7s6c-*^k0ZAy+gXJ^wDT
zlaAWI!mxvXl7ODT7cRSV_;lXHAwzoft-K0QBZr8AkIN?k9|Kmv8`9+R_;?Jz3Isi(
zQre1@S0PrYKc7--w3cQ~g}X?~4kB?#16Ve4@(RvaU=Nq{P>kowV`5H!U$7=YS<hS?
z)#0xiP7$Rq&)Zj*YxY*JN*^?Z{N!f=E1roX)>%S->OV;JhUz+MVs^rsN6A~#vQqmM
zNo_5lF>akP`z}3ChO(7>C<1xsrhj)^%3vDye>7A)?~^&w@TP@ow7fmn!8n|OGw>#w
z(s|X|6ng0R6zYObyVWsgV1;y(&<mQ1kpBS>3N9DWZz2Ntqf(^Hd#TzlU5mcvp#d_6
z5mtMD?bsg`J1-haN>x0dO&DvgygqY9M`xXTAj6P^GS;JmMoHk-A-AfFTP+YZf?55^
zH;pFtbgM<N2gmi)JZ$=gn9}DE_WJ@P9r{B~zc9F@Z#JWkxtnF5gkbMDY(csVQUN2P
z7$ayg#!WzZ*nC?XDaj-Ih&Gs|HN9W!Hs4l%7$qYGRDp^PK&w0`G&nM1iL5h-WWZb(
z0H%@%Z)PMMaN1I(sJ^4v$cpUB+Z}07^C5X)Zj8On$GZ8rJEav8)cGk=;lfYq39yN}
z9@LlgSI0xzr&xE|V~=z;2T>nkJXqkikuR19>z8l4xSd``+A|vA-4?i;h9;YI6}5PO
z-yC}g7H2+Q_Dv*%okRPQlHuK0PwxV4E?eB|_VyMBuMP*bh|X9|O|Otpkh2Y@^g&5R
z{5BP#m~oKKYR3`x$)3y;anru;wYmd~8?$h;&l&tD1}~NzTZ0;@HFvLz%Bl2;;yh+$
zMw7ALZZt`~)4=#oy<6N%!Z1p%O&(2u(CW}qm3eH;9cqGE>#{i(+tWB%^ISv`A90_@
zE2OWI`x4o;buPcW%^aRC-@}-o4ZS`q|D8M?F9coU39v9R9ZRVZfoWJ;Yv9l&w5xHV
zHVs^m6*7t`4Rso*3uu~=rLFhE))M8l!X0wG@4=0f4Vj7ORTA*n8lzHl^4J`Ir>Y9Y
z@iGdKs{)frqU*yv%8v5gm}^hxbDyiRWySAE+j6G7qkM~Yd{3BvimPAD0wM<yqrBo{
zTs+85C1AH+ToE7a9LQMO*K`qx;0@xyP%>XM3!uAA?Y)Ff$*ZD6Ix_TB59h4k9o%)&
zgEn@$<K>hOOZz;W6dOY#<irhs>ZjJwj#>6a!r`*r?%iJPMvEygWPe_lk;}=7g!*_9
z%Dwo{7VxEA^hL<PrY@MWZY0rUeSf6XbGOh>r~=0&+RLLcZKw8J(W*~7-A-$>;;+0I
zb;!?Cj+I!F6=Ah%r5h7;tA*fnggYfZo1o>64l>4KD@52P+E&!_xZGcV)5!su<;p59
zr&@%q4*7O3Fxfp_&(6-xD)U%OxgQZa5)0MwvFL?tSjuI`^OCAL?;64&vqlT<#~`7S
zz2|sYSZh-S0Ayne@DjFhaG>*3ne|Y1i&PWY4Vmm~o}?0Je-pLTMpEx4w!7`icGnne
zn@Y94xldp`tYcytrA73A(&;az&<hzZX}Y#%JyJLmvvu9+s(6ACOQcvmKldpQWR#{l
z#M|f96llw`(<)U1m8a4n2dwOrdTJ8?*<JUtcat8pQOzAMr-bCr1TY;L40lB6u`m<N
z%%q8K#4|AqBG;-}UYbVJ@osFr8{93lhVtks`CHq=`t>O5{Nyx$?}3SohOP7&_Pol=
z5y$O$nTE~O7u)3Y^;U1fury>zHuO3#M!P%Ib>XT^k*VyrFk-5cs=Q&YmFm`x@)Na-
ztv4vpF2-dA7D<bCt8>}j49S-_$qE6kc*DU>qGF*@-eqL;{jOchj0A{z>@yk5JFL|W
z%lavfMH*$HTs+->7BNhk6n3|>0;OCG_V%r|H9nuy${pH32CEjS9%L~v4&B22)w;}p
z$mdC6VifMgHXQP6$YVxoH8lCwvGI`ozTjX8eie(bgX80D7S(QxP`?~gz@~x_G51#4
zrmfsnm6;KoNYUrOURb()6o67fal*vx6P%<6%8eI+D%M1Q`-GAJZxJ&2LqRO(M>)r|
zZ#3j(Ui^Cb;r+Y!KNJfd!>}+!0~8yR$o<lAaiCp5UJ4kw;@=lE0IQnH>bXkM(iJjT
z0MfoVLR1ARsvX~m)YbU%m6}*twCeM2r`zfqvp{JOREdrks3g)UKsw^pAS32dR4>>-
zmIv6y#894pl!<eT!i2HXLhzzlDM6@g70RX2Oj!gU0d+i@p%F}@NVr^Pk;uL}6l|bK
zR+z#K;3SzGhvj;@c{xCKMfeP7=2Bg_b9edUN9<;g0tmymAq_k+^H<t-s=Dk=%9^mr
zP@+gt{8kD>X!K@!m^i=j7=v_adYz2;$R;HXa?jX*Jk2?)mJO9RmogScZiy$y?<^=0
zm>jH(AE}b=^slz$|L{sc#Ef0zcW-MkdyL!3D?gqH_-=Q4oiC?NzTV=F@KGHw3!Gpy
z#JXuNQHbLtCJYOhbKBwT*&J|1Y8eJ2JhGl>9HmXxR*uhSst=??Zj%ydnW`qsFhTa?
zxO7l|3?fK_F>q_$kL?IW|BC2^1><zaP(&7#UA=-Afgn*$_|}L4qX9q^V=>bb1Zaer
z_?XG6qbeG>7El`wo@C=!0i}yrsJ;u)jQT!Pr6+Qb5{^28<>Ex1B%F&v5e->sg+5gc
z+YSB##ai-)g31&fqTFBwjYO4KB9&UMBYQM|XCZnyN64G<y@GwYV4p2K4U3Z;5QApz
zcJ1nDyT&ES_p~jd?9N=b5^pvi<UKPtFLlydek=lxgHnc2GBoaxLc_hL!}_duHRxTf
z%O*0aEB}pBUFwj$iB(<S4mm}W1(j0U<cRWt{~}@)8$q+(D!znoU3>f0l9<Y5GvMui
z%Sd>{RPnI3D3K2w@=F>Xl?_U@6Dz|fX*qrREgGT9Y=w)G1Qtw3L334lKI?vS>3OPW
zP|RnU8)4JBaG|meKxKprbguL1E}y4Ptb0@$_oyh681X;YrpKiAkP8h=!v?IUF}Upk
z-R6+-;(6>u%h;7+Y|X=ZAn<{(2eR~kmyDsc$g(O!!fmALtM+IP^=;J@NaC=Yt3CFW
zEhv-o4;BiUt*WLjXey=!mRSu`oHgIR=?$(nck#<hR|VS^!SU02tbhhE2HWJ;RHf)$
zIocgat}%@~K1-%sXc%HZ7<*Kqv_;b|4uwzZI^R4=H)k|ngv@GatD$omS`XNN+G+!<
zwrq7-rN_ox?kDJy8r5S#XLltitgmOSsG0_q3nI5;n`AkqF~bSCaER-IOaV)c8mR)w
zKGC}TDuZ*qjjWp~reO^|*99L%p5c(oWIbuNOGIR!bn4<n0a+>@#f*Bm44PSWk>oRM
zzeJI)XhDO1uX|O>iqO3owdT=(VQ-QxJ<xW?v=t?6IS+-)A}qukt+2@H(`52#PJI@~
ztZuc>D~milIzAl@b0RjW)DN*E8o)>HTBE*rRAnf-pm^QDa95GPR3GgkDbcxYU7wkh
zgpm}%=L_K1-3!_zZIai~;}2yUgd=$B=C(UH8ywA3`O_s&=|h&M%4*Sn6|utI>EN_~
zT1El!=ukA%Dy0aAsnIEY&%{v1{6->ID_I&3X@R%7dQ%pvHZ7xUt%o$n)qqwtcWyF4
z_p`IVlDX_E?58AeUJ^4>;#e#&DaZANdyn%1Nhnmo_K>vX3G0l(COKD!lAN?AX`iVh
zi#e<xy_RJ_SnN9R#1mD2D3blDqkiMf){nf|_bb)_dtC<D>N5;5T?c%W0VZoeb~g~U
zsCo^k;<F*?<ac(qdTh$_pcE|<T}5_EZ_sO>-Ia004*91HK+Py-*Ay<I(>cBBUY2kX
zPl)yFwd8Tx>b1I;WnAEe9?Xea5WCz1+6o8ibip!iT^U2E1KnVM)M5-F)$SFh3a3I;
zL&;Zlx}{>?(#<HkJ(@|d+@L!#t~!-e<GKp5R&E+v)v33kwNI>X^eV%%;4q2FkT5Rm
z*lU5PHAxK)b!w!Z`c4~8b!Lj7+}@Ux1o6LQ9HQt;*a&Z<GBm;do$2{b+LzCr6S>r9
zUIRUU{Oxw@d49Zq-8$me;UxX@M|CwUA7zhgU}#*dIuJ4T28sB{XK6V}rsBDXR(d18
zgtU^=7%N4ob-S&J^lY2NTS$UYGFnVBp?0;AIE~|ZG(0((z*UQp<BHh{re}g{9I+F9
z<_V7?;Qf=cv-S~th*zC%`$l$cuu7`3N-E~=y=)gm=krj1Jn|XVp{-j3c`U*n$w{=Y
zNH(0G_pfiR%P1RsI%0*>^EA>;p<5DcJ<!Fo>_KJSacYK*2N8yaK~}>^RH=wolBZ%S
z(1Nv%n{3Tv*3BYj!y%@tNi#LmpSv)IY#0#D-G|q)VS?TTA1i+fvwD>)6=TaeB;#6F
z9TfBB)ldb0X}f0HuG(Gza$zz2XthoY*-KXC8?Qap42_G3T7}%;HHOWEuek<R5$9C}
zT?sR;`wdmhQ3jvbOz<W3GVF0KA{t=xRGCMZMCYnSCe=XJOiHN@&pLA^;IXt@|I?QL
z?MN*Ke{`E^|Fv$$p;0+ja@lW*TR(@3I<1ze&FMBmhA;1?ozW024pf&Hod*KdRUyxI
z|1C4fBgc-ZvAt4Y@WLJJDp0D>tS?%;wQ*lEh^R+)R2_}jD~-DgUv{#=i6fK6Fd=`_
zUnIUQ{WOLES?+d5%oTyR-t^qw=u~IWaHT!C5sR0M1sSI<*FZD1bJVzs7zil?)TEBb
z<;DWRz-KJeV19|%dMhGi@gfYsF~sgk1nolHXZaJJ{ipnsKB)LjlB<Cc1AQnnoD06)
zcy@TuJDpw~Y<+35|Jtz61J_@8%)x)2UYS9*Hv({!neBtKR`5L1`hF+Sw+woVq*sKb
zf<zKAci3YbQfCGtjt6Ojp?N597$BA$))Q&klJc(I$d-=*$mI1zT_2IzB8t&Hp`NGY
zn?$MV9h7!IR_!dCc+z$O+TkIm?RKl2b@H=C(R26vr?=Oo%)=|*fGzFoj`@Gw=dyv#
zzUMbJ4ztOZiSJ8CQC%|BzdHXWZxnC+tNrf=`M*0ZY9RZ`#0fD+=Gsx&X(j(FupvfZ
z$%@&AoaY{!Ve^L_SQ{;5%tbYH!%HoAtl5w;8;X?0^=Z3zUCN15QA}fJE~>SQK1@xh
z8S^-}vD!DLqof1w*1RHz+^T=-7P+%g2Pacs6HUc3Xf#O;1Om9iRE@*Xy5nF%Pp^A*
zeO^YoqibF*^B`ms%y*RDoFp_-OA=hp5&;_9AeoiZM$2i;%cz*3PGdKPV$`hOwKr;T
zV<FJ4dmt&y5w4X(Ua=S`D|M2tb=d~cITOcJJMQB9B6{?@ozq_59BhAOV$B9?n*y)E
zM56FW+V;66omd<2RoZ|O-+QL}$WEZ}eIe_%yPfX!&2EgC@*2kQY^GLJ?J4fV2_m0O
zw!41!rvD;hSWkdZYr4Aw+0jbYdU*BlyL>YB;s0^+#ftg9kiV1_-i0Ha43yltBn7Ka
ztv<ztSbh2ppin%@TY`Vw<566sRPlQnxO>ZiaQWS<*5FQ&+YC+!$ee{P3u3z2Fjnv3
z*N<4P84%F249Qp}622vU8RLjZIwViW|2-AV!PZbQ>Ho(r@)p(-S!HXL?XxLcxw(NG
zQHQmz9-FaKsw4+9S~q{X)Vi6$8&>O@u#-`~Cs(}qYNn*90V#jwC(G_d!6J}cb|Q+M
zh+^F@qZC1clgxJNf9pfX?5mcLO{>=e^ZVj1YPNCg3({hk_V%TVSj?#86n`cAJH<j3
zF{O1PX_3=*WfV9|fybDKMM<_1tqGL>=8b5rHm<xj##xC)n~O!VpfdZ42e;y`2B)WY
zU1JudX;*JIR3(2_IzF=AX;f<y^<be{X(^}rnt-bc^03}DlqeSsDL~Sp29uD^r_~V|
z7a$GQh(=PE8`V~4z9yZC>XFAjsK$L`IVIlJNfsH_no>w`4%t-ltO{9H86#>toyW4m
z>tqEz+88U+GLgQI@KJyDJ$Wy1_7Oa$%+bIghD&3X4`P3)HYy|L(nW+@-EfHd*h*|g
z;3h1TQ!|9VWs|T+m(8&&3*1aU!IFFEMQ83=#Ce8QZzSyv4YeQlZtJ5tNkOK}najr@
zKV1scuPb<@sK`JJ52o#Wa)h_^F4OZjnm{4KX7hNJSk{vKNG2E98d%pMKO>jx7Ky<1
z<AqO_*pYu;wu(aQOPXB~s#Ep|W|_ZQ@Bc}nH{pD^)dR!uCtIlir~0`Fvl4=ha#E|K
zDsRYVWx{87$dUbRn6EZSl~~cHT!LZd#(=B3P;PK2^Pcr77<X-~FV|fea(BT~9P;1N
zMr`ZUE-C-inSloa&SMYUO|HZ74s)Ml-*VDv0=|F!vD|P?ZYI}|JF3^Of25^<1T<$T
z#BAfNB6&Hu?4EZ^*(bJHDIEwBjJs2{^&>!#`BVuxNnKSv?vP(Y4(5-{wE|f20AwM5
zK#OP#no^}k3%Da{(q_7GAn!k;0p@~A<RThE_WMHi<jcb%$u3;CT+d*11I?0c)MH9X
z{(yh|?-CYM;=2iuo-@_cK=TbcPDv<f2NbQCex0ra#q0nK!=!G+yRo$PF#D%ULuxLn
zR=2DvhYSt2Sv~pa^Ly5iRJ^<V@uMaqC9Y4S2#kHHD?$7yA*C)>Jv?RJn&+*(xx!ra
z((0<+Z%efYGu5(3vb-g=ziQ1g97?i%(lnFVHWGg%<X*NTYDkwL8-n$a>=AYEC!tsb
zvJEBJ4;X@%jpXIG;!&q7?usQP!shK;C1olkDM%gP9Eny@%+?7(^7|k@VvnkKHRxUK
zDdO3|MfuYe7pX&*i%P&nAbEG}P*JS5l}5UdDsv`6mYV%zM&kv*h15L}gC#DRPR{+C
z;lF=lz+2puDZ8kq*EKfB)lhzIcy&6bXBJ-H)Fj^&rX^$yvvWZ&hBAFe#x)f~%dX5(
z3~Ei~1=jvrQwcqr=Zty4BMcqjNnm%W43mcHESHShO`ALeRb7UCo<}QEOO3ahLvA1G
zY>r-6?XQX47!F+_LhsYq=kt1lWR5FiZ?t~|JI~68*QI{msi@DE0s?C+rr1s#zg;6s
zBBiEJ%8Z&#%(d*qXq*u}nS25}4=)D9$vjutC$WJsP_8wT&XZlz5W`YMhTi2_t9x0-
z0pJ(T5s#e*MZeWMFJ&tBzD8348X_E}Pm@Gqwx)Hu+k=tLWyrzI56~#NXXH^8<Q0Dg
zd0vjFU(^#NOYOQlW$ry1X&U_cMU=pJx6<<^rRU#4JF5U$KCmT4noAc_S4TW>lW`1*
z0#`hv_DoUMf=+gENqC)%!iS7K@<>g1z>2|u`fK|?xrv3_bHJ^X$kO)jP-$nBG@wz%
z=r@Ki;WAPpzxsV4VkS4!5r7CPUj~0zz-<f&3qufiT&ef~H`vuCPsGw&QSd9JG}qfd
zT4@&w?CVy$Jt!kXYExtL;Y;|WU=5{OV_n*%37|}Txw)s(rZ6*4%Ybzz)VeJvo-pyn
z2)<^)0}n-@OfX5})D=i=v_Jdt-`<cKttD}~r3mj)&}1)0w?hSm|B`VOCKG>|{g`Y8
zH2#9=9=Ky3-IFN`Wg<vYlBN=RBAr!9einc;Up73!S}3_^u=oD^;HTu*WxsPdC}pej
z35)A_7pk1_)ATd-HG}Icu?pyMevrVw4IZ{y<CO5h8-F>hvh5~Gl`KRxt}wGr@<Gin
zlrme^AI!n?o-q*bQNA&;1;>Bu49+UVI7%n(0gl>rRSG`Q;$jic7cuzVQJfn=s0U^i
zS#<qQ=ca$X8&#&fxYRHzR%(W=3hXJl{aHM;pbGR$9iF0vUY!oQw}#g&Q-4WUat1A*
zD9+sql%_sFGIUPc?3*Vx*8HvIMXpjpC0WTv64(c*c=B`=Fyx5^HpPEPQy8lQB@_g6
z7LG+YV}%*MRUo6eEx$hEC~D1#S}lLL{rQ&<?`0w{a`cG#VkwiU<d(k2V#@s^0(ob>
z!A+;To36>hjNz9*-5P!+LP>FFV=c`$8K8t%pTu-zY_G-aIH)bdy8JJH&mNt22xvqU
zaWFVI3B{;ljg<~KJcWN+zuz64m9hcWF5sC>rmQJ?gcDO*((GPdHfFB^IcZJ~nxAU)
z`UJbY$rFGHWd)Ax2`YlfbQ{M21!b~+DwcfDWK-H;VpVKPv>Z3Wg+J-K3?-bn`+Zul
zLvU|TM+LXT1^RUAX9Cmm3K~Vi<+8L#r)sI)<9R3^IOaBB@&bS8<kbDP0DN5lL{vEQ
zkOw&!;^z!t6eh7ARc~CgN9M{rneWLOqRr$AGGkE$PLhIJAe#cnpFq_+0<dAmLlo;4
zt08kS2Mx_YUCJ#{4xiCTbjZg{qQlk?)B(K_N>87wdyG`g?*8-}+{4HIC17#SfmvkY
z4<<yeo0_NWB_n?VyGmIsUsRqid_HpK6dM<8W!Y|YDWIcB_=^~1DqR<3r}f-m4csfF
zVjitj`VA+k>nKVd`GAiQVhy*aKm)dd?`=+S{WeH!cjb*N5?=y|%5Hpis^7F)-FAMU
zOZDYXDG7X;qiqJ8hp1Y<)$H+r^xhYWy`B4A7Tgxx#)5wvt<-#TQXY&&CW&Jc?Za=c
z!EM38OY*><JQ`D#6wU)4b8t)Wh$<Y=uDb+Hh?V6+*6eI-7+aK7A0;?kOQ9se1RlY9
z61Q+xA(o|Gt2ul|2+aCSMA$$OT_m^=B_l@NrKA37wBWv1Z%CH;*S|Xd{&xxbe_K9y
zQ9cM6&RBo-oKP&CU^LWi+2B*8Y7|D;&`h6{jAwaYXlI<=oL-i5g3igiV7%|i%`}HR
z7V8}j9*`BmAhhZs>k5A{3nEgN&GW!KlAREj2_-F`zhXX*hABf!KDQXL7?BuJ`(s&~
z2s==rQ1N8<DG_=+0P#|SsA%V`+qcMgIiE_z|DS)oZ*guD+4}rf`YLm$<XR7-hb7sW
zN~-isyz9tW+nIa6n#`uK)Rvhb&;oL-+3)`E+ueW!2oOjhy3x6HYZ4=@^y%|Ek3M}K
zL~8&FP8$VtS<xh-65e`IK%+6+qX_{HCjbXLRb=fgvi}6laKcb9mkWuU9L?|8*isIb
zzMy}nO8Ob`HCHSP>yeJTppL~#qKOXl5q`CZ<EdB*PSPX>@KY-MDB4pbejatp(&qpn
zfKV)iJSs@S(-|FV7paF3eSsj!KHR(#`~Y4~SmlJ{#jkg4$N*wa6fse)2!5iYLgFj3
z1sl;oeEGQ^JtIfVI638#Prj?VrmkqrlN^6tCH84BaRB7|)GR`r5ck{li++R0h0|+m
zX`#nnS~q*O&d7!tXRF_`x|AGuo_o1;&|2k`6K7&FCF$5yekF(&R|q{Z6YMPb(P$-^
zctYGx$x<fYshY8CfVa&9dy5uo&=);^oc{~7M;M4Kk{Y&q@UA3P_fN;3Do;<Cm5qNr
z8RRTp(ew)F3(yxqeeuGEWW9BUWrgSo$sQtVPfrJR0qWwY>H@Nu3J=pQWDTo!Zqtiw
zD6I7lZTD`{641>cVD_3nBhG$w7T=krIjl7=20SVN;ip86yYBE5#!%XLeKKJ_AshYX
zmYgpaGef2w|8?}zl$ufv<o?9Mck+KLbpfapzy2b>Dt1ebmHk&*?*o7ULx6swO)YGc
z1lwEK6PPD5YcH$QCK~#Jtj8G&?|6{5O>lk{Y(hacVXHHVw<Sp&PcAyG_7#ue)d)v7
zzPpN`42%rL3fU<Jl=@Mvyxa8O-{9YRmhDDI#7$f~^af^qkWiI&d0~HzoSJ__$?gq_
zdf@WvMUtdaiMqqYGkT6+E~a$A8{-jhha>KIr{5A1&o&NFbUS^ghv_YKk}OQS^2f0h
zLTMaW-9RsPPsd=&P&)U{QfQb&)--)ZO);HUMqZrDnzC4`b_)g8kb}I=evN5b`=zap
zy`dKRvT<=)ZzQ(R{@1}mb0vQy)ItXZgu`W)ukr164kCx!E_6fu_>S_sJ<xcv9j(SA
z(+p)ge6<Yu?OXa+Nj40kNQE71uyqFDaE;=+lwRLHo`ia<*%}BW=d6E}JKvp!V_3)f
z*dwJwqacL8r%hAcUi@f^C;Hq?#Btq2Y!Axa5qaX5!ZT6jDgu-Vk7R#w7`<#ZJ2i1g
zmQbIpj<xEw^w{I2Lzo~8b`@udSQS4_FO)spQutm*?)g9Dgd+r#;txIIu_W>6qg^Ro
z{Blf7qz}aASMtD;m^_^7*kt!w>?l}zb|w$|4QuU2cilGHZ=c<Vp0J<)br$w>9<gt>
zR|Sppi@-wsyvDoX|Neh+eK#DS&)xq2>V6pBa{vvP?s3G3uyc*HC)?}odb__F>k4cS
z*&IIE?#2y)@u$Uj70>Bl_Mof7;abUAVM5yo5jS&b?uck}yW{caAbC1n$Ac2oLV)Eh
zljSuhaxhq14A;RGFIF3PyztSNasT1^c6^UUcekHJxHW$GI7ELFI`hprK!6rIEiLx6
z)V@ug@#K3I4A?2MUY{?%)u+JxlVyGbn17PYpUuAgwKOeY{)sYwK~680g@OO4$bVM$
z_KkJ#005jK0BCy#;`h^BQvkpz0)S3)2hN`&=M8<SfzN-EtY;zXB&-kAbsti~4dLly
za(Vdg?x)W;38H_v{ZV@~=qDC#|LctH&Z8){!x;LUuAQ(YnHW#2^)vJv(jUz_`dPJp
z`-KhSmYxja*hQ92*j)B7jxXxMuU8(FIY#f$q-tooF|Rgi=cry||2aEHn`aIBtmS_e
z+y_xeZc$p=ep(pKgub)r%x})w`@T=7=hg`?4ZIMQHPwGfwbJiCRL)UlaQmS`xM7!Y
zze5+4UqZ8{VY7ykxHW`Y6#h`g^yq*3j75||Se`r+${IrRM^|;j5gp2(i5^J(5{N}A
z<&3J#o~=i9iOVBGltms9dUG{u)N7ZN*f-5uo9J)$*P6e1qCkA>Ty(A&375n^^=m=V
zGA)BRdS-vu9LW<)Hz+^Zs2Ad;tbG<*;zJ*Z%3~*X{;0pwtlKJexu{T5FYFwx$lof|
zTOpfd4~i#0+cqyrr`-N!r+3AAX?V7xS0|^d(1V!&b=xDiAF*#v<B{jQ$rCCV?w*cD
z<P2GJ_d5CR(=BtB8M;HF(BV4q7(H8--wO&wMQDGAwbo^$)vSe~0*>eI=5V%Lo9}~R
z@5_e|!S;~J86SN>PtW+$Mc=q(d{o+ClB0a!^y@kFlE0$=ff_L$6H||u7XvPvayqBM
zcuznumTsKC57bfm=y8z<x8m5;?6vz`j;7kD`rIML&q{77d>`Ss3X074p$<1yE<aIS
z@<)IDZN2uVqi4ol@{!kxB|mPwSn`AG#u7w>lHo-5Dscy>$b*FEY#;)W40zH+l}%I9
zjVbxp-r4<F985~KGroRECylCjhPA0gQycsrJSnQ`_B3`L{ZOVBRm(N0<vd|aU{9zN
zu<uheESbvR_S260qNZC+qkF$jRu|7Q)$)JN&ur+atw!8`(3yt(o%&!z!c1#Q1<d==
zdQDU)b!$!H%)qN^O=`3#<Jkv}R5~QCABFE1{Wx@M=m(ZEr2_BL;y#id6^#z=d52o=
zYW`9zi+!!VW%Lo%F!c2A{%CJi)A`Zelg;~&Y-O7*4LlF9lrTW1D_H+X995P5BGrFV
z{twBI%G~#(&Q9N&PB(eA9z^7MBQT=K=G{tCw28W$V(D*e=$?%sU!gA~Oe-#>xoi-R
znNd)qB#_-=)2t&tjulm*C!dx<(NL5S)>4=V%qXgFPJmOcG%}hUHFDwpJf&?+h{AGz
z_Nt2)Y!E@0sTFW^vPFc|{fau(V3vPkvDYgU3hW2T`_3hK>S$5(S^u&5BIlvspvd%}
zV_)aIMeDUa1kl3KwJx&zi+MivHyqo3&Uy`PzT0`r`H7BiknGNwo8MHGM`<Cm1#Wd&
z{uRPbWe+>KLoL`;u05IG7<5LXaxmYwr<pL{*&NQ6OZn`w-uC_MW8XqHEnI)A%BDnH
zKUL>4acGGLsj5}{x{!%7BTiYY;iuhn2OBJKz6TNJB#J=ZIr<Jb!zpK&wLLrg<B$5E
zf1aaC^(@J?UwcsN)q=sYXl3hvohe)A5&Py-u(bRFfSml8o;4js%l=+{(v7*Sl0TQ!
z$un1`<SOMV{lOl{RHdIa&d7iN3PgVnPWu~Y5~%)!h=K9Vno7_+qt<L+cCQi(NAIHD
zw{Nx9!r?qhEgT>ShYG^dq_G}sS+xJ8Ef3&vfX4xx$&fRy$MH-3&$5{f)|HmEsP#-I
zbkvE7h~A)aTB=&6w*GhL)7#x(^%rHz23E4)r>bb*S>+`<_X=i#4h(;KV6yXB1U?|@
z6$1dCqh1F7ze|OVtzDylH98_24@mHBF#kx{d%2X)1PALAm*a>|{3TLJARuzCUFz+l
zZKsK8b)I4Emk+%}An1)QE{Ay~<7Q0-1;AEXi#SX7ba0CI{eUS)TY92wInXvIs%;|j
zxjfytDgYIL%5JDQy}N$^Er6C?Xz?_90=NKNcEN?Vg#w%aPIlphMN$AL0F+%oVco_6
zD47987);pgXC22X=c&1G%fV&0+3RmQ&;cZbhD6*XyU6k~8wmALW7xV9M<KG{oTtfd
zkUDV)@iQj5=)zq|jD6?#t|Y-Nd8>Ibs;`r*yqxuaaw44Nkcod5o-cH_#23mvPE3TK
zbfc=&>4-Y-#pl`oOO1k)a<>F#xbn24;iqsHfGXBz)^qiVAnHhyF7u#5+0l1EIJTUl
zxjtnzlCD;%oSmU}@6c9DJC{&OX1__?f}TKR`t7AaiXe%OPvyNikJy)N>-a{a%}R>F
zd&`uIUhi#EWj}wNBc-0+GvDRKrFYj$VAMHe_<weDHcQV8TydT_d$`QTD`{YNp^i8w
ztK9br2mJ#nrlc)Z)iy*siW5ua8toLtt9q-ku1!oLiaeN4{><iH{t<0><?m)t8Q$I9
z-l6L;s+_Mg{#iBdIjVfs)+4&vuxoctH+fZg!m5d@&XIr9MmhR<X8nfh<QMNz_ijjg
zIil-(+Rd?oK3(76Uynb~e~8Y`zC*+D@7H&?<4?o!!`rVKdDmHx<GD?ARh1VeJ2q9%
zzS1KT2UY<*xhx@RTKb~u;_$2XReKlsAALrUv4$>#WW;Msy<>1?T^F{yW83W5b~@>x
zW81dbu~yJAJL=d@IyO4$*tTt3C(nDnbL#uEch&x}s@AMM$GFG5uQ`TAy?1k&wEasK
z(buMplGwCAJ}&l+SiZo;`cw|ZN;$jf(IFc7{9xE?(!B`V6#|zj)>w_zFuQI1#Gfy?
zHoFgOI4SQru5;P|I5(D9>fWuymy#ibE!)HYQpT`F8x6+sH3*3dpYTMe(SW;%iXD=<
zl}#*GMti%%p71*#W}W@vY^wG+X9Bj1LMdxa7S<1>d|&$ZEg-9o=JB-S^AR}w@kqqn
zhT2ZrTVLVI&SUGF7~Y)LG~M4I-d5j?m^v#z$`oV&u(99n1-|e7aOZE=5?u26C^q^{
zOipLB5fkHm^|4wF_?c}eq?g(IsdGG7+@o|#(?5s4Wsh~%Jl~rh{#d5y)7u(;O5i$q
zGp}MDmMhRD1Q4~?{Ph0_plp-nZYfP^K8zCa8h)o9^IxOO#=)>?yXRC4Qp)GW4;JqE
z7z!5(lsO;iM|@MPuqZeR*k^BhS|Q2vU4lOhIK*auU&w8s)364ZY*dylmQ(E7V@*i9
zJTZg)r_J=?=4C5-pvx%eJi!<ei4bNJU<fl%k=IvXfN@AOj7|8&&?7kh7yS-D&zd>v
zn=)4G5c8cU1B+8P;zIK#*b2La<Gb=k7O%DR{Lmq0b(~tGZMx{A9QPA~qZZ?|pICEa
znbiYkw`+(mn|0CN1eAo^f1N|#3H}TfJnUE`86uqQPO}`kwnod*yC@ZqzAVCYB5Gun
z)iyd$0~Fv=oaeyf(x&87q1}*{hMTLH!qtv_JLqc{W=Jdbnx@R2`m;ZaD(rB=!}GK@
z10ynjAUU!Rj?{zM0oXsY+;G$BgxLTG>Nz5b8dFJ89_mi({FU_d6=14B;8#jxLiPDt
zC^H8=D{kPX!S4D?0vps`1h)_aE<_Jl!%*ZNAbaZ@aZK!c`abULCg^tIzDj!&`byqn
zvt&b%M!;B>6XP&t$aC!1;wjYW?(i2;$(>(FzMrR+$RRrLAns?A0q0cCcsj%{&~_fU
zkHll?wHk(k(lzRjZ})KYD~?~?`59|t$R^fMi(|m;X(N(Se6f52zr%kxivA``DfIZ+
z2J}H-j9a#DkxE&dZS@rvM;FOCx?Gojxb}M<C|Q5<qE`2onkc<GxWl?!>Y?frx^rp0
z#5)G4uzSqFA{u+tztB1hV@0$hnMv9|i2Pal0emq$E$DtJhM-rl_tMP?I!Dll{|82&
zV30u8v}2rub;qJ{SNxaxq&}6`$L0b<z{h4GN}w|CDw<z+{IJ&{xw2fUe==w@qiC~z
zU~%l)yD-oMT~;l~e#Z3XADrA<J<F{G*2Dy=FKRVgeR1~7pl`N7F0(}VZb9o^_)WQz
zv$T5~pY1#!87J<W6!40mz#wiI08fK59sjdjA@A$Agd5NI(+LXA6FCfh1-Y_m;F;t$
z#+9=K<ClinOo2PgtwnePZ%GEZTr$;960NC02^4pOkhb%z>#7)en{c>p**G|`zoWVA
zx9OSe(X4}_A#B}5;P-Li9#IGjEZ&r9qh7O92(14qWI|SMyssGuuK)Gire>yyWzB18
zlMM0>KGBZw(m&CcR|NSSfYT%I0zMlf=6j{bR*0t@nt7V>E+!GthI)$5{I@z|;++Xd
z#E|xl-nN98e=eTHp0OLDRIZggqin{i=!*6huC6e@q7S14-=?IN$_*>SB=sNc#-6?Z
zq@D;oa&F61H>tKC8)rnS@23nXjmoEjEP7wSlg*S+Vqu(DHUIo;Qb)3n0$T-r0|W%&
z47S>bcm$7xwAfDNlLj)^{`w|ZC6OuyJ(RpuP!}xn#4l#*Q>ks?Pbw*G>5@a1h-tnk
zUY^Bve6n#}b>KC761H_Qu6=Y)N^{B{5E7f%RWOV3HACazQW<Yu8%p~$TwwAmJ{gT7
zOWx?+iLwx{ug^a3ARv(m$m5u`iMH>+8F!V3mC5n@a;G8H$6g`VqWJfv7Kuq;`cKFZ
z>u4)EZXa}(U$RBBh2az5%O8e`Ws<p!#nqkli32}lm<tBM3q&ULMdg2+thyA06-!%W
zTZy+iL(J&4H>Napj%}b?2A1h9H?U?kj;2kwfF=m!41<{qod7N{z#b>&KorjSx#Lvc
zt&OBiYBeeyJDe@MG1%s(^NG$r%547lI58wOl5nVMxrA6@{?EX3)wkNI&SgYF>_q~w
zBUIh5_(ughQf>th6r?0@f8@*8?5UC=KTb&C9G@pp{-Tx#Xd=1W*8Wi#3gX-(4nHCO
zHPhzn=a{}a<q0`E0|*O_-1IrPX|#%8_IAW3;x&d@axDM$k2q~umZv<6I>~_^yw4D$
zXG5W;pMw*g8dP!9j496C(1%s=m7>s&mmA6zi&s<o2q+$xDz<E8yQWrlS<iRTYgleD
zax{#%iTh%_VFMY@ys|^q6TX!xHJEy+l9d(aB<o}hMyOO{0>q;OE9YM1-p?fRf@}}D
zmjaR@63LP{V$H<%|EjR&*c@IoVit%L;v5;aw^5yzY_Vv%@<w^SLJoAN9y|5dten@U
z>1Bn55f%)|a|MZatRGmloOs@mLae(-Ct}qpSj3Nb3d_?IPHe?|kz!8UA(kI2W_~~^
z;TF|%COi^E1z5>GE`ZbZUtdDp4<B-0805)eop+yoyn(mN7sY0Xg}X<iN29SQA?GKh
zh3qp#O=`#>>{ctfUCSWiD^dM4T)CPGZArgbgV#{$W9Z2$eo714Z>EP#a61fVS@a^l
zw8&oU5Qx`7!lx``Za491AC3b()a52RS=TtuMPfsefM?n8GTl8a4!-cg#X5WUk}wIt
zG*5lTosN@FtdRb3PE@uQ1_ih!n2u)36o~T9<;J7hg}zm>@0~6|hv^X|t-+aS5dLWf
z&;3H01C^VN9_;5)jIhmb7Il-jS|f>qmrEEqy3G%U3mO<9nC8{p9w9i`3Y&XT|7*d^
zg92R$U=6JcZNOV9k%Ch{DU*e|0Hf%Bg-7p)xd}mL;>0Ly$L}fB4Z7|C1Ij&$az{qR
zpukC7uPnc|7~I4e6LLgef->H6->(hG-P2Z{K-myO0S%<gw5z3sv;zf$WVViSc93Hv
zg8dmCGDeIXZ5xmCKv72<PY;?F?EN+BNFXdQ=#+E{Cb#%>XKvMWM=honssFL1|5&7x
zT{e*zUJ+0IpXy}yRG+YR8I_WBE$YzXWu}Zqx1_@rOf0?giNOx}tN9=Sk_0P2_Q$NF
z1JU}cy2UADA;}b7bH~($FFVH|nW0yQr_0l(bCLTCUHT~uz0A4TTqB+{husNV(gFEK
zzvlYt5dXq!!dGLCDv#Dw1muEBp-Y((R6jh_M=uuR{$hB9Q?ES-!qX&ONHydgi6+w*
zaTWrjdke#OVO)>?p}|5#NB)X5Li9xS)3O2>l|61{IzF=t=OdYK2U%vcF^Vz2(rr)$
zA8eyHjEl0I>J|7kDN%!b`beC+v{+JbcRAK&83tjfQu|OCe-I}TuW2;wSG)hBrjC?z
z!=ga{#lnf#+y~X<R!D^~P&Aovq(TC?hBnZ0UEbF5_kM#I*;sNg;|P;O^-kNSfG}Bv
zyoytfF)Cls?xn{bxPMj(j4}EV_PO(5`>UOHpWv#`SV9=9RAdWIz)cooEK-dSSI|&!
zCWoJ!`vn#^zK*Ll^4)c-RY#-?{(#)syEZtNy2Cx7f*>P!Ans4U8t1X}_8k_Mm-HG8
z+xgk$!R`;XXM^2Q4N)VEQvR0BqV8?%;8^UWs(!O>ypU-o-rnxPKC*FI>sXbT4VCOJ
zjwf-lVnU*Im@uSj0Bm!!#2~KbLZ(%`?hLw_GbrI3;VxOpm>w26_B@{23>x9$Inw{I
z?`C1c6>qe|FvQieKMgiAJ?itLPJg<4dX2_P)sTTX`!3PxJkTv)!gzx}BZlGl=)};T
zZhW8lqgxdL3^mN`hWMMmCfC1OQs+jPdtR-n=Pss{i#VoaJ%Gch#na8@2{lO2P7b}5
zHiL!u^Mbqz#Hw@`V_3HcqTdRY5TXG0{dYuxj>QYpI!9KIaL>Sad)XSWtH=0(Yh#fM
zO!O<N2m?n4;TFs9*YLd}hi^asVST@7he0B_Mb34HO$jJg{jLxm4`yT+H%)$YIvW*Q
z?KU+FYC?BJ=mC~d=hqwIQ0K>^J5cB4neX|Q7^F4%q;Zt+j~m3BvTS6AMU-V#`FocW
zBhh}p#Dgus><M66q>KZEo|wAGH?1Q6s_!$22F(hRAi-vZL9^6f`w-_Cn*zD7W!w-8
z;a`m7pQREp9D_j4o)nCqo6<2zk%P1A59Def0c5lABG3ru7!*g~y#-Awa{pZ=e&^VL
zIj%I@o_0oU>STf~uJoAQjI6xnCo6<9l7_!=mybmyaT3^(5^0<VVo-8}w&7d&r)5<P
z*~K~cJC-ilRK;1~x2f73uvyApenx2JLG`PTk@am`>0#dw7h;IUsvzBGsl)(r$w=X6
zW-O0h0f5#be~_nt5*nWIkr2NkE3YovQpm41G84639Gr`>l$r0Y3p1NCj5;Pw-EyWt
zxjs&QPf5BxCW3h?6giCc5SD_2)GV^5Jy7sPsj{vi!G+8Y+`?Z3nea!<`TCNi-yOy9
zy7>704}Rs;l7CJ(yt(5A<ZsYH>P-{kj+7A^kAQ(fP6h(;B~(ISOD!WG71Y3xS(g#&
zDy9=MV+z(NDQCJ?H{9+|cjNx19fN)YdwWH9e*dF>G4!1ner$Nvi_g-{2kqPt@uomw
z_(>K8y;Y`%Cp&d40l5prGfq8FmqP8^{Esh|>=7lk?CDBS_+CW%KEgAh*P+CHqnB>C
z0Wj@ZnjG7TTjQ(h`%q>;AULozI*hzibGz5V!s!?nWA-RsbC_Gk>0pt4GH7Bo69kvV
ziYYano9DvnnCjbe97Mo^yH*aDg``EdxQY+IN}SpAUPvvG<X1HlP0zih%ZGFydLKJN
zIay2Nk=iV?)pplRH-=CU9ARH&uYS#Z1B?^bA{ID(eqD5Xdsyy0QtF55I)>HX!Eu<S
z<YD5=Neo(Hf=XTH0&DNS-5SAeOeDb8?9$6yEkOI?E-KdseYrInM4XIq3l>1kE79Nb
z5#jN^tb0}A^HvPZ(St6TAjg~G-$XiN^Vbpkfv{{Q)D~rz4mx9VHeoKnD!_Tk1Z=T+
z{&cCn_d5|0*6$G#3{#-S$*X+_BsE|_2C3cD<&hS=4~f|Rvv9)4zoLi08=!GVk%HpX
z`%borj+pUe3c@RogC``%IfD`Rqb~EY*~3p*`1Cfa#J(KLPNwcnriN}zo(f`6MYu_O
z{->DmuS9uKm^c>0z->h#Y#~1kB4G%AeNoa{46A<KTI2nsx5czG)!Wt(*>T?$Qo0?9
zYnW;4x|_>5Lpu-xM7j6OnrS*R;tQETP0gG0m9tP=i<rxjni+fC@fLcQ7IPSyl|~(|
zoTopvDuvrr9GZL*KD4lb{qakQV4}_yj$L^nbcyn&Z@QmR+u+sQ{hPIiccjBZADpsj
zB)$YDWmqoEZ4ipqxk&8vW#@QeNa!hQp;8u?d`85I+rr{7+8?K7%8La~>^$xPgV!T*
zhh~i1upb+()RimpjoN}DdbduWNBV|~o$s?9!fz`CWcCjz?h88riWo+wRq#xV3MSC{
z{GRLmAk^&d^%1<DLEe*O9ppT+KiT|Z6(*qjAwW1YfhRt8ocgAn(9R`p@364ZrW5_7
zH=W8#%*ABQm<1bKxmbL1raKGst#ZE2?3<^Adb~(ZG`t|?)NjN01v_eopmJqC&%*O@
zd?EzcoyYLzyS-Cj9KZMY3`5tmSAfvzPO$z2dYLyq{39nOVTX`_8g6<JjaLdCK}AZz
z6N^$%7eNqmk#;U`C;{<kLzj^loahNHX-w-QJK}!gE*JNWI}OhxR~?7V&aPtWHR<D~
zsP;r3%M{-9n8J(>^~)fRxYdGN42KrmzPvNusco*k{k=N?{FcL3@o(!qeL`>t4FCSn
z;ZM1E4IA#K3~3V#c!g|!LC|AfeIsf&BtCza-q}rhIiEm**yPD36ntL$vZ_xQP!6|z
z&>j58UpDAG$rg+0t({ISrLTs-k0p})_oqNIHo}G(-5%7vi{PRsE$R*3`RLC#h)Xj#
zR0gj(H&Fqw2u_M9x^dB~2_>${uR%Ofq+s?~V~n^W;3krdcHmK@L(v^9ArvC_*Eq#8
zGh~?t3U_MCdjokm+1gH2!#JZ)ba-B+`vaSg^UH{JS>)J$h59k2huS@Vy9p`Cx*9ax
zgsfnTI*k=nlq?lyx%isV2%V1G31x>RI{b<9T)qP{3R#$F^_Icdf5fWcwCk}$s=u+h
zbUQ13ZLytdE$^h2D3_Zab=e>XWWzU)++qUJyAhLpB*RI7RzVK*!HH+|<mA4KRu-rq
zI`TvkLM@%dKr>$6`@p$g(+5jC{XlZ|_ajmX60(RNj~5f}y)Q2(D^9bOGv7j6+k$X1
zy2^o0{06VzL%nP73#kXY{Kgx)<`bo}uI~To(WhMxDpz-_@hin1SIElma2i^>7}Cvd
zy(rSn&RXWXMj+|toS)^dW$y0L>ciJm!t4)+)4Ar*qIZ%Je4IPE=0%p){yK2YP@L?F
z>~(e+Ip$Q{_f%m4h#_O+W<PZ>dh|aXEudqZJZq^f76WU@+}`&uVu!2rLCKm`(=y?-
zl#eSoRl}uA^{MX757R%`nm}GG=2#EAY_{CpUVa4SQeuWhd_~za3css%(Mvfy2Y<q#
zVDrD*$g@~-_OdAl*aY`-9w}$nv$DCeSaciMj}uuDp4dh1-%FrNH$Q5cy7TH!yypL|
z8ujkn2&Ko$>I+?45O5&}uZ<m}hN3H)`0^?;e@5-;w!%*1F=i6xZbC}q2s@d!S(bqB
zL0&vrT05rFWGZS5Cdh}<DVBgU2FfWyY2)_9uh#hQ!EaHh1`szcN0kYQOmt4E<&=UU
zsO3Bf=eVfloMynO0TBZh2&%OE5}zMl?wRMjs<g!IhR9?qign4wW)l_Cti8Q3j{Jf#
z7D|P%2eOUuG0kZO<*p&JL8|HGPVvl%loiT|uCRG3zG5?YiiNOh5ycU2Y~|~S*I7R(
z#1DN2^3}ft!d@HtO|ui1p+puzHz)b)DEiVE-Wi>9ft8^EKTMcqPlw|=UZMDhS!*Jj
zmxkL9`-;^w!qK^}lNouYh!+BA%{GWg_n0dJ7Z?dAlYiLKU^sv&lb0*63KwfRR1<K;
z>4R0vF_>c3ep}szUAPbLNuc04USFjiL6^0#eHCh4k+gLST1el6Sdn!6cStlCwL=x^
zOl4E9>FgB*xD^kXQ6mK3yLl*c^!1n-!?GXyd1I55%-JbJ{M%YK`OS#?1w2g$vqmyy
z$%8{%*+9w5Q|Q9*)2}F__rl{lr^@L;sS>W5vE$&O_*{9XhasQ3;r!^4-ol4D&FFqc
z?t2aJJQi^S?a0DC^1}rma9%#2lM$+P1rXyhe620e&6dd%)$~6}O+*f$p%+in#KhaT
z%~-p6YP*=m3+mYelO+o;{EXadUHtMtNxM6!a7ept^%h+Fv^)8TH{u2D2T=5SS_~>1
zs6|sH-yb`?Z_GThX_WdMP^XXmWc)ZX1@*qBcu@;7xYUy4G>iv}B)^nR?i<7G@~evy
z0(e8U?+}wN$Ckr+jTDaDH4=~>FFo6ew!)K-VI<q6YY<Eg*KUr6mH17B-=Y$pj??OG
zf?BL?X_vZD=#nE4p7&I06477sz4ASma9}>qU^?P98Uk2H)iBm&&1lLL>3f6e0*9%7
zHA5xIfva=v*<<9?(tb*2GnJ6jU_Wp%piFx^`=rj}ZrJ24_-{_r+ALDGDt($huLRb2
ziQcC={qhV-LQsM|CO&LK$#=apXEB*npWpTI^527Aaxu8lNP_wZi4)|=@3d8vc$;kU
z$~^oO!CVWkhW_Ok--J5J-Vs+xB2_U&=aGTf^>uMeg4Vt?5wq@u74DS<>G4sG0z2Ou
zcA8feb$j4GP?(vvs`|S#H4`f5h4}-JV(NwUIQ|mlU;R!exP7{Rf8Kh2ASM0yEcZQN
zD8W<Np`;-}a{fg;9KzwCFBRzcqKagO4mLwO7;uZH6XI*XijVmtwrO<J_G(A%<oC;M
zrYjkZ!i7LqZP&KbpFs@rzaRz#U`s=y|AtK@{BP|Xv%S2`z)33kQ_60mLfHcHO2O<Y
zH?b4pU^S`Y+T{G%0#+9h%)s$#HR_|b32|CG^X<07E9Fu<iK^(I?kY$h@W3||Ss<75
zDwe@(>q%w!R=*1+H!8A(Fq|1@Fb+<y|DTs1s;%c;8UDvK%LE8HOa4wbW}?zqPYT6A
zGx8jKosfisM+#CIeLaZfF65OEiT+-5>W>d4gposwET@>?1yc?Q{Y$z%hOa`dCZITb
z5APgL^L^Y8tqkY>YL_1sb`5hJH9c-ju|pA70%gbauQtuua*kI3BPHLX@ZEQF|Bz!S
z#g5QY?!horR&pSaiSU>~sD^N8w7{HyRtr-O$&ntcDQ6t~nBg=1&1&uT>zC9G#&syd
z`Rcx%KwbY|o~xLOQ!}$)%>HwtdB=1v;;^VX1l}-9($+sSKK&?-_(EoiS8Kdl4r+nC
zyv+RIT;>dZ$1Ze3TBzjz7nGu2uMCL=kjf#hpD+;Payf4Qhc-#z%gQts1pnR17-L4s
ziNCJ@$t1Gz1(JQ@4u+~0tus2tIN^(HPnE8CjWE18=_3yV$Ig3bY=ba7w2rjfWe{f6
zC<uubH^-|xcz8BAue{T`sVc2IcO`BOw1h;+LR?D_@&Cra|NXGS1%TBOd<tGuZqvba
zHVbap&8&%<ioSA<${Bo;$Xw7RCe6r;@i%p!RFah84IU-~ah>s+Tv>3*nM+x+k6rW`
z1LKYskgxyW+rxCaq9kHpJiXO=zB6?`*6G)(mCBMg*f{mTNwU!GUW%Xo5^GMVn6$RI
z>kMuh`E$5CCIDh1K#fyXCnTr1dKB!Q>(w11midvlkR8nCUtwE)lj@mRVT6-m0rKt3
zQK*^K+?C5R6UoxlS-S$W@UN@qNobyR9V=JwB;Z$CV6q;w=SB>AUr-nZ+_-`h%&4xj
zgU+c0ZGg^SzdW)v<Ma;rzW>@J)1j2`u%k<zct+{QRX!vFWE>Q2?ntLmTWJY}DYTyE
zYScTcVz&klQOvBr;mC<BH(VI_Bo&4wQN6_RL0b@yLa>ewsPF66BNnQW^7v1eao7Y+
z(={fZZo(9*rGnv0mv<?FOVEl)ge!xB)ktb2rZ-CQP=eL|PuWgW5!h4qS!D#PwbbCi
zm$R7y%mq}$8XsN(3^Swuwr7&WT5lonpz-#lLo7nrpumu^HY4>9@&@Gl!yy+xS{b|U
z$+MsyMV(Jyn;x%S(hik*-64e}Ch|#1)V(o;Fn^CAJY)rf;Lg8ot+S*1IbOg$WKAk7
za}W>*J4<7&VuJU_7F;Q_6Whc8pG@cosK9auuW*6c!+$2iRZR0i{bn<s%J)0)rLOtu
z$q4;#x|o5m7Vv0+$S24)Dz=+P7M>yA(8-;38{k{}j=9sejf$=avx+77Wld>+z*I29
zXX8Fx6-73s{A;i131s|!?$TU(3N68w+ji;j*ny36a29d5HWMOy26#>%9wtiyk#r+5
zbkC#&Y4t3aV=G~=hk<O?^C7+SR`A2fyJEw!br#!$v4RJRYQ%-Xw%>Ybu5|0m9S>=e
z+~nq#<Q@B40vXCkWz!Dn!m?qlG+M|*aBK6rv)qR6$KHD;zo_v`Tx5(l@u#ed=lg0=
z7^)PTID=MYE6`jAMCX=1gxH(_(&7lxEsehV(P(ZwF*FF{&Y6~K4DRMS;WP$3Q5G$B
zUZoavm%Z4w+YTsR<?oXOrrS_gppril$uYY)dZ}b5cr){WY{Ibu)UaYmLC6XW1%?<S
zjm&}{ZK3(<Ue}}^m)1{+h?N-IL)^x~a9go@=ZC-?LCe{bmb9>0R`(^K-tmw2|K|kP
z#*(Gq4E*N<LvrG6Ark0rAibb8xqk!;xPK8qPlH)_JMVQADg_>SenQlPKKn4Yp%@4t
z4WL*$^0iCstiW)*KHR<azJ2<?!Ym?pF)E4GQT!4ZdTM*YL&mT_*CneGy<?F_&5wOC
z26X>gn7EuOO94%nrZ<O#e~_qQ7QXg)Dp3FUBVyzAimNTiYcZfjoG95XT-1@|=xJE%
za3-Ft(VMq+eO4~==t_g016l?!d!~AMj{X;aUQqtG3C;u~BCUvZ^?f52Clv3#Qy~FU
zl!FCZacYzKlm=pudV(=SA%}xpx0YyOI+OWOUO>I=3gqEbJc0K^^i7i8>yM%-eLt9l
zv(ex^n<J@n_WHSf=$$BYl(X@no!^2}w4V2^E1Bl$y{;0?`oUAqid6VWGI&3<)HFB*
z8r+$*!;9Q4vR%fiAk&inmVDLH6|?yzTTp$i_EOe?@ER*ne(K!VVy4WK>lQyvMZO)|
zpanes;Oi^)Se^lAPmArrZ{vBHJZoz0AXY#mAt4hN{%+4HA2|2IJLNDdF_uA=v9v#h
z!7M$6suFC$k?@X2zKj$?Ij2xO_>XxpPK{%rh<2h{N}$Qlu;SMSJ!+(<DyF01%a)4^
zM{xax(PN;bn~5U4+sn@|T)9lP*thab?K)uoPs>WPx5>j;h!dBJ<MnCKPbqAc?m6j?
zvnqoyc~;<!#N=)vKtOApmGo$V2V?r&KsQF*Y%vri;w~B&QDutrCcO~QN#jX&CWDCX
zH^ZaJ-X7SC8fGQc(HL`3XuTSH<mzMYY@aP0uez=Ji?e8Q#IzokEFna9V*DCxn*uam
z$gk4Ru>JiuT*USy&hPOeka51Gb-R-O;72om({#t#dSl}d$6u3(vqVNbT4zkAHWel!
zQqLqzR*vFA@1mIWF7WL>!J*y^Q^_vU6eGHGb(O_pa9n+<-B7!>fn(``R^j$U{ZPxV
z`yVYC@vk3pBJY$vrW0=hw<W`*WnD53t~&%jhsN7dgb_qiNBW1wVU}Lj3@zt|a`W&&
zA|{h~(cil_fy{tz&qRDgGM4ovSSCD&?QnUn<;mZ>ozF9A2eL-hqqxqSjw1=3)>yzn
zE!70ej05C;Ky%As#`5CJ8krZb<w9FJK5XcEd4EX@UjH$g(q}CsnkjLb9*L#2;b%^s
zcQ5`QrJ&O0AH%XdglBBUdFuS@F+9Z7c!X@FBmXOh#`QI<6Y2FcdV0M?e<T}a<?&9$
z)4#<1nI13X{V<J+fd+j=R}E00BybQL^RVwMyx{OI!^IJZ@?#(~(-4t$ukX8_n%c$}
z8&204afAL{NdDe=US5OtQ`wsF?C052oz(byQ&cem9<<2bmWKn$sjQmZNyDGM?7M6s
zTnb4Gf<w#Jk>bchXh|qaN1fL`R|Md2=qiDyA@dA>%8A8@_~=}Z@9v>u>wz`Q!9?9v
zF$G7CSqo7Nt$dvR3NL1<drZRL8JTD!!38Y^98b<?9F+;oO$oS9I=($oeLmQJhN4s1
zS(2u9WHXV7`*x}mC8+XRM2s+bIH}Ep0e1|FyKMY_?C|&O8nQ5kh$NkLn%AVNEa7K{
z&vkxj)_&f5P?mf$3Ah$rk>gYy+w(DLd42Pln4Dlg;r{x)(9!42Chdy=@<r%Lmfu5^
zQc`H_!o!J$8*)WH+x$VU;JglcLWnIDOv2Qc--w-!33aw|0B?Hlk~3{KM2l#^y+~hm
zr;7@Eiy0tF$u)NW2gbzQKRDrE54<CYhlD3(5IYMD8-xIf5c^zyTTtQ>&({UEDYSJd
zlBzx757QMC1apsh3G#5f6p~4$RCyv`anepvG@EJdUB)Evc-G3=$_m)KIBnXV%sZNg
ztQfK)r9?&nYFj;$!BG5RU{HQAIPAHA5z0SYUR9)Q)R6j8SAc1P7{IIj>q^_EmJFm0
zA|ivBY&UX-Rhua02_jfNkd!%^+NNFfA~$D2Rx)Z(F8=d5{xylTtDYdjcE>nkEV*go
z2MMwA&!HFGC}3Em>ExJnpI9J*N?k1hDo&8L38pPj!+$u-tiv!w2^(gf{VJ|OB4m1@
z!t>vvkb|LR?_&@6Dbbk5akH*})eY}$eeiq&iYu1iT%{HIne+L;INy=!+$5oNtqAe>
zeRzMt1RQi#bCqP>9@73g_MDBTwR?cUx13nRk)1mg?<*H=)Q92#;cM<Q{fSlh8Qf*%
zvH+mXeZchw4qSwiTTdBXB$+;3Co`B`uf^yH-x6Sog9qnYVz~15O>Zntc1M6{aA_EP
zKa=Q4+<)vL6Nm+bi-;TBYIbyUy9jd6S3M7j-O4JQ3I6}teJjc7?+?HE5bCcsX9LoZ
zN%B~P`7M`ZQLVmPv{OWw6L$kr;M7OH!2twh_5Lu-#6=NSV8kF4CSnpx6n{sst3V`~
z)ImjU6vrILoVuWH0K{K$m7{4$2SOlVkjx<kVd<-u{D*v+3Y-Xn(9*}gCB<`xzS$3v
z{|pm0J5|MF`{wMg17>MxnSuw8IR$}9gNbO%h&1>BVzKQa88b&v141u34SOQtB4ake
zWI<H0R6^e}h%whS{02o)7!bwkUae?`c!5M(<~(PC1uHTbCxLE~9{Y}&X2gAO@mRH+
z&hf=Fe(5oLQce1KmbtrKIg7L^T4wCyva)5MVjw&tf9f}@SE%{oR|!pTQs54((BDon
z()Nd$^l}2}gi}M8$x(Qy)B&$@y@f5!yccr82=3Ql@6s}ozh+n)+Xw4ka7U1%Vn1><
zCzz}uz_k!<$=76%Z#4oO$0monMm+*XDMzE2GoV=51%gtqTy$T5r>A2O;u?I7q)o~H
z{p=48vBTYlmZ0lM04!wKG%o+{FbtY|uU`M77YtYS;YwBV>0^uq!W|<IFf03IzM%!d
z^AU1P75}po9KqJcZ4IV2t)XgMnMUuGmu<g`nv$LS63#lX9BF)rFNO+&1(`yGkw6WV
zmF7=C3V!7UBQwqcSTALee3QcwN{rQ>bc#rEBZ~N}$`tg$fN$Hm0}?Xghe}yxk7}%p
zb=tqqP3_7i*8CP_=i)>XZUaiQa_^lRi41Ub+{j|phcT|tq0^3cUki?~Yafn3VT`;?
zB3T^=o3Ct(E0P9r>^KSa7O7dygHDnBp3~91Psvz~C7|JD_%)}ZAx~t(j3|cXPhBO&
zb$-eR&Wx!5AkgPD<yYs~pihaZt8vMh71_~VFw(TN!0X%jX$yc3Giey!LiH;j<@hvn
z6BZaw*lCb@AQStXO8LtUnkwlE2FhpyNS}k(7?$z2IGP2f4iK;saBT0&jBfq~@{;iV
zN^{{xweB?^!LVwAL*l;$b$^5(3ekKSSkPuG`O-%U1jMY;D7GE+V)1`teZU4UP%^B(
z7<7)TUEqtB&&L8OZRWy=zANB{F)=aY6TN)p#R~g*#%|;QW-$_d3%^044I%oid*|>m
z=v)Cyt_Bl3KsM*Sb!^1Ex#-513?%8x*F-JJtAjUZTjVO?g7beHvaA9ZEX%)veDVBL
zjnE5-lG=5?ZHiLObghYuukw6zffyNcII4@I$4d&7rV=H~maf)!>X(oh9*1%y*S^kV
zt^YcC|1Be}KoEOg<SoR4-6HGF^EE&cyPVZ35g<Zn{tq=x4_%+BxVU38CT-DR><`p~
zLaSOOj6o1lt`Ogl{wfpbvjqreh3DDmacPg~mbW~x+E0rbSlf52*>T&T2u^v*^El@1
zT#4eSTI5ymf6j)GR8PFM)T?UX`y*HSb1W^=&1CXCqzrZAq!GU6Kd3$byQm`0xVYos
z4!2}e*3iO#R~7h{G{Fp0H?LV|W!!CDVez{IwoFrHPZr0Ji$;RW09nhr6*|Uzu5y*7
z_>-xH>%^4(9NyDCGx&S5euFv!hrve^F_w=8T&<nlCk-hx#xmYsr;09p8U2T7SgZQW
zxr-9mF8}V|6Wsrk=Kjx3+y(;CQ=g87N36X>*ioA+B|rX#K+gHt*jZ;tP*VkV%GGc&
z!AID1Y5<$3suU0ppQ>D`lfUoC!-Lkc4=bFS9Ft96eD!lBGc-$me9)oUgZu*;t>fI#
z@y*ef@4f7;+?)F^F|hcW$Njo_U))^mOJu3#wfE<ys=6w=X>ZN1{DcDE(PShcY0g{y
z0w>EaU%+~QH!DoV3gvH0^ZWOB7eT|SK?&wt6ji-kfDTv|{528mpCHttg+d(=SgK{G
z`=(%Uu60vZ`5;>@fUer*E+~o4yUqJ}KCo<cxz_D2=!B46f5<w*kK(v^?Yc;45$z$k
zu?R(cGanvr64UeL!1C^Z_l56Aq~)Q*UMr@JL!hPj=f-;8Ldeo=70w#9Is$p-Fdi%_
zQenA8vPzKN4vW?;YT%9TId{$<_!hb=$YU83=7wjy;^utb$@t2OeG>+~6O0@DiRCW0
z1Yi$IMRwIxu~_$XG{@DE3?SkqB(3W`=p}Hv(gC^Y;HOac#AiG<IQME+T(hI8Nz?u^
zqt10i@^w%uy#;nZJY4rLeNNv6M6>NLLJqe~+@tWfN@E<WwD1q$kj*~dkSDRbV?Uh0
z*R3l2be5A0p>d#zFNl$Z@Xf7l-N8{*<0m0SPWMFhh|xi9vPN;Bi7MTfXc-tRO6A?m
z8epjLdD=PR@2No9^F!@TRLOF9JH>Yzijy9z+6GBa9tqg#Lw*=z&H+8u1{SA`qJRF!
z%-*_-y}6nsFR-D;f06snqwNn#HNSuP5cJ;I&ftnaf2q2STViZJ@$kSEtToesvot>M
z-h|*;;th9~hBi_G8;acZ@R>OO@N3gyj01$wzbEJD_^7BzA%eR<TzEe`bp2DXM^Y@m
zV7f8c08{fBa;_BO04Khwl*i{czf3{-9M9$Wq-LFH6FryO(tYXmrD^49kr}L=FVx-h
ziLyFe!uzuAV!uEJC(`+M%2%9#l|=l%sXx-O390dl#SnkO|LhHUQCJey1A%IoorD3p
z;m!Exe5Sf=FN(@hSOOo^F+Hda%#7a*>-+E(`|Re5VlAvVei&Pdk^m0V`ml+fkHtx{
zPBIe1Zc>Hr^<O?)N50(#coJOl#%14)y~Zw@;RuY5{nf7;APUp!pO~=PEha3e<Nl7~
z&bCM1G~icsv#CWaa@1-m8@zI%2vC6YYroI#=S2aMQj%WT^D4|oTmpBYG31)FmG~&W
zPcG@yM0#l*PqtET@>>^fmZ=8HSQEQFW!E76Cg-6?Gl73Dq4`7`@8&O!$s1-g_(0>-
zwIgSm<bW)obQo(x1T73o^mC>K$RVS;o7gCZV6U2ef>5uzLK&=yWW`qA5d~1T4@@|H
ziTC{{5(p(_SA-F8FdqI?E0~qAH&1~*y&lH)Fc9i_F@iv^t9-BPOy&grZs`JJ^9y>)
z2;NP5+Q6xn>;0YfNUtdpTuHed2{X({tb^80bdByr>PG*!a`CWyO<~sJ$R%7V#)(+r
zRwa1>nVxqcYv%9N7=7v)+8bap<D!;aVm`ro%0QT0m|Vzv!xAB<quW?sj?v!)@e;hf
zw#R0mKN{!csjmetUj?#s|J6hiWz~DaxqfoCQ4RSU1p8TPbhM|Ra*?^_5yl1O4poW)
z^)p35cBQR-Ktxf<)o9TB{zwMY8{=JQzwE3;FD|LbLOZR3;kaqk?l+)mIIE&#M8+?n
znd03c;ri_xPI}fu7Sc?f1Na})RCSNP^E#+jwuhWdy5s6m`a7lPMhkX*<w?^^!DCX!
zcKZ==sfjeYebyo4O*>{E;%Bn7BQrd!;YMa*o|6RVt=^AW7nW>)j@SP}S@TXIk>o_o
zmh@y^_gi+|^=zo)uYv)Ocn}-=?h%jD@UB_Kyvsiw6X_>2%oZ4Kd_4X-Yz-zg9AR-v
z(1cbd36vSF3V-uiQ9o`Du<o3RvNX8rWfQPzu3WuiJ#KjZ5Cini&e_QfrZ3hF|3Y6p
zClY&NKR8_=JY83<=|lWds54-1?)Mv7DDhk<ovE0_G+zK4LZAXJ+W(zrbF`aAi!BGf
zL5v9Mqm~;mUHVx<2SS!kMS*Z=zIX9)6hT8pzBL-xs8?WT?j(Reh|YguXHn|z8<r9Y
zS<}0!4a->@44I#!Zo?sC>;EUJco56X9*x^H5hC`URx)!$kY+QX61Wfh$UkD7G2nHk
zs~z@GR0|XK@O2h|zS0kSI0s}JtAkQ50?PQIKbjPyVQMyl3X#L!h*=(2`usGgvf=SQ
z^t8x7N08kH?Gb-|nK$O*g@LUcK(iV)R>ZL?pb{l>ezhe)MI{x`BxQCPL1lrR`IqT>
zrt7lozY?H&5Bg+7`?}9StH`H-j7$LaEt2qFfhP3aZy+K>^S<b@QRxNFPX(kJPn-k`
zsDL($Vcy-ip=4Oe?(G65zV7{H1z}5SyqjJ3g9718rw~_hLG6yD6#VCkn5f~a>^^68
z^^j1V&jC!<sn`b~2d_)`FI}L^83-T}B#r{8dm-f=wqyM9I1CVni_v89?>J2}ZNP8;
z)Gd90ZkNp-f_|hKq%Q)J)v4hvmSUNqF<qn+bWZB6Wxu?(mSWgfe5}b%94UsAF#)4@
zx&5Q9U(XCrg5l#NvqF?V&8$SLb;Xs{KgQv{lcT1~E|CRJm3ptpL7=d-@#a0Kyf2uT
zmd{QXkR#x5tG)@xE*8x~szL9XbP%6!xbYzY9EVKIm8TZZB~mdX4;B2KF{uT+JDgF+
z3(^Pd_C5<oRUK!pHg-Pwh}y0f&xe$FZBpb2lqmNtjHkpW_+z;HFE%s~P@LWoZ_%&j
zakk-|b{8Rk_XJ>_!8<n2^w{~*s{Y)2Z?gIPv9f2LO0aev%ywy8i?Q4CW&&SoT-)dX
zzAg>E_3g5fiqA?j2tsYE;wrV3CxoCigh!Rw>1SlArqM2&8MYPogQMk5TJ<7SKI?Dk
z)xfU@o0d9*OAhyxtPkL|zf$_UA6!2Fn<eWLRUXha-eSORuU{VtI>1*;R)|cR4i-{9
za*#_qzPY(&{6I10sydg*!O%E}WdI2Sx_kS2p%qENG_k0$O41?!hJ|Do0#gZ`^{r;7
z%f3c7tdi@OyM{w|@G-EYo;nFQjY3H%!ao)$56BqIQGRq>uslr|7#q|weFR^2Y)4J3
zM-zwfeIDN|k<z2C%xd$~s$)Vq;$XW2!L4GAgJi}E>u8!~UX>uX=JDoJNqQR~LlXG1
zq(SN7L@oqjdjChLxpwoEdh>l12M#=8tjGKX+i%j4CMcJ1DF?&SZar<Up%+ub&!dW<
z59jdxRkBKINiS8U=&50$kh`PdL+0PN))V?{R<x%#bqJVw#W3W=`bB@<1QIOKK`77%
z>E`R9)zutq*y-PCSJKgT95yuIIv(~M1Qo)4*Z%yim!OS$&d<<K<Dd>jlAQX2;+-_@
zF*HPB`4WcraV#+>BWWT+#?0QKQHzdt4?$4+Qot`Ru?t0lUph?XNDk7y#f^?k>tl95
z!RcqA4AM9>it^%`F5XT<i2~*oSsleyhPImzl!Jp&vJ(2%`J{#&BmpY05`?tm*Cynr
z_8QhStm&+>$o{Y~#_YW@-dJFQbIo}2;QaCWzV<4((fPsXE_(M}Qt<fJJ<(R2zpmid
zU4iy4%Gh{d)4nLhug=+)VB+=3jeBM$(uZ%~v{6c=5Xl@{!@cr#>E^_iMVWNWBC@u7
z)7u382&sBO+%|AFi*Nh@fS0TnQEqDkv(Cc=pu=*sK76LhSZ00gFR{GwI#vk9_x*LD
z!qxFbbh<tC(nN>@X}^PG!Ndi$X3fA;DhMX^_p2*UYt21Cp&>OFr=<7pNaeFD$Qy!$
zW8rkZm1E(*-k)4}-O`=G-jStX{0*rMgjx^v$7|z4q-`HrJK#saOz{P;=-)l>7%8fI
za?j2sa+K*;-iq0(se_58v-6zxN21MaqC@~%fAL%%zn$f>ht1s5AE$A+eKYt>G`Z5#
zX!PRw&atxPdy+(hwtYre=eHrDR(yrq@TY?4!#(3`XcmuufeaY@#;mMb?w;wNAUhxj
zbM8wG4?4&0c83BsI5QKVH#m_t)V#RwGOe2Q(a{$5));GRDOG1UP~N6I<z@u^D$n%Y
zDVjGu;#NcvDZi2uGWmwbFwhq#Ss2l2PhXSKmWIVdMX4t2=)8JHC`h|*m^&)I9uda)
zo4$gLsb;2x+I}sO1N}pqV|2}%A-rSFMMI$ZjxHT028IY23sBF;kh-2P+ja4eqzsfV
z$q-$Q>RQ5&?42g}?_W2>wD^1Wvsg@t#@t5=WTvRE*YEG&xEM299M*hUa-nVm#lG*Q
z(@KBnMoV4XqPk^Jr#TJ(?rZhs-b-gjiCSK=IZyr70<h&i4xH%vjuVl3DJl4whp>45
z_?3R>MRN{Zvgf3oDvj;<uq{h+o|5L^=irB0!?sjlAWeAb3xHwPeZxdoLd>47B{XlB
zXA)Q*;Of^k-7ZehZ7cTvZD$;MI0&~zWkH=Ot!SjLEqP;c%$8y>$BPt_<{9ln_fat}
z_ciE2fv#=oW&?eD_eJuXazOhoW!0S~y-m1-f(sMyCM_wGnc7)fEK3r6DwAeHC2lJ{
z$Or|=7@~uQEQAO{f}=a7mE6$?Om-tlqTGtvO-=h4qAeNj1cYGBN5k+9f|6-9i!N$q
zd&3YIW-1aETFt(WzAP2}*qLKcr1pFJlgr8(Ad5U>U1;26)4<jZ_RkT-22n8|RI80b
zye<oDn4`J>8=L>#9s~cTmig_2=q+&eXx7tz2_8s21h(e8BpYUxjZEa)IB3ao!;i3Z
zD>67Bm;11)t`kl#s|3VIXgOGyT+zIIr(IV1Hz(q`=WEq8;>xnKl8?=<ju?B0+cmRx
z>Tby7r94i(Vr0{3I+#VW;w`{B<SVFm`nG|l-(R$rDz3;+QgUJJO8%y#WlTO@3^5AS
z25rnK&bjvHiFFY{p-rsT*P|(aiOR-r-bK;LrujxG#n9-FOwvX;)pi!YF>@}$19~kS
z&4Zw_GknTyG}jD~KV<q*B6x`QBRZ>R6W@Eqbp9OruyNr}!xhroBTnIh@*s-kQ{fZf
z)#U3wL)z{a^m2>0qWoMrb5Z-HH@XkzWw-o_dVaS1{OfjnfQ_C6F7&O<q6XOnb38&?
zb=EsQ%r+PIo5F5mG}boE-Q|M5jXzF`6K0ZpQ2SilNkLp2jGWFrd0~##$Y<S>)<`j3
z57kq4t?fkiQ!RJXE&GxyVs1O!!Gst%isP?<hOZBW<4^*XT`EI$XO}o9cr+%GojPJ%
z(d(PHN-=^~TG9?hFEFmFqgwD1gaXsQ5m$qwR#6`=rW{aXI;xTt{}I$y@D#_my19>s
zB>p;74AtIAzA3d&(0K?{uc%`KB=V>UNE<O+qB_=N2|qH_X_m>s^KE&*7k>rBFSfkV
zV;wr>q#qt65Ye%7DJ~(QG{K04n-N^k?X8<7@<%Z6DYgDdSQVJ^sL3Tj-}_msp2b_3
z_lJo+hbSCpe6}3=TU0)<>c5i|!Jv4i=ud4<<YIc;C4I2>ET%%?6+@5(X-Cd1$YQ4g
zZ3nRoy}9(eu_lv$k|9THca{O*V4bCXNFhK`u@kRi(^SsN#J8qTo#Pg}v^RQ{^-^^*
z9HQO^+iVqpMliZ?{8x?YeE%~CUmV_=?hY(h?izU0KKk`#F!ZnCv-W8{dJ5D(byC!U
z@7hD3t3E9~Eki#PGncjLTi~(Zoj35W_00Z8Gz2>C`jD{}q7Civd3zVA*yRHoctI}d
z*Hfa;oq^ezFmZi;Eyy!qQgY%xClBR=UHOp*PSH32pa==RYUji9G4-k372&!jSRP5;
zu?7+m)yd#J?Sq$Rof7F4ar}Vck#*<1-mzYpX;HH+wT<MQ9-kF*b+VdU6HlU{hH2`*
zOy_agd<KrO>ZORSt5?9c?*=+SHnaT(j%~4tUL7koliTpupf{#MU++J&mpN)0%l{<o
zH+l3~VIiD21Agzk5v>&ZX!11Qql`-EH;?(fTVs+t=7WQcI12iWGRcB9>w<fb!q?Et
zObqcEnWM<~II^iOs<r4v9eH+EOYG3t<3E~4%^vBhNH|FD@$mt9HSljrBCST>(3JCY
z7tJ1f_`j>1Q7idM5+Q$%q{c>_q#?cw|H5Izrdk}<ULgkODD;;{QgmpG2wx>Au+LAY
zMw4DWa{Njvz?<x(NlfaYQ?rWVv+pqd9Yuf2LPE)pF<NTjofcLRh2_<BahJHnM4vsS
zvY!b;-?ZW9B&2VEdixi)^4m>c3G5+sYPX$92trgsMR*uhL0+qQd5zu6VC5E`hd&7S
z0!;6vlr+FoE^Mu=&bgP3e{nxKTX+p|T3;ly{K#Hntw%z;z8V}_!!2u?4SSG}V_ofK
z;jZ=3Y2LBUh>v1o#&CpfIOa|dXBfZTU|OH^pA_Zv84nqtDB<Su^7=|NTXUIA@zTwJ
z=F-&EqP6AN^9LiCWR9Jk?7wPH2~=z3WXvzgU8oj)9FvlD^`u%B5>LBbwT$+W7c93&
z1W6)BWSnEaLiubS=$047r|vK1QKY!{*|()Se*u&Bsr5}K@l26V;4|)AMB>fgujgF5
zefb&}TqoTCTF#o6b4W~(f&OJ?xv|q2IM6~s%Dvwh`qHd2P(Qm#-mq#lgz!YaIv7{b
zbc|Ez8~P%DFv~)z%2%H8i+SxPgbcF{6|@mE8G=sCy1CSz4zlSU2cbtT8DE63%-2et
z=lyWt@h{YFi8;F^6BZ>o)6u!O536rJbJMI<t9k(%V!a-VUZA)Y|HiUDCwaatA9~og
z6Q7_|Ec|(CNRn#UmPsn7@id{2AaZGLhSrU0XLVA0pKDWD_K}+&wkq{emI}$B{CkfB
zdjQM*`aUl9R<Uep4ko3PUlMNa$_}yRk*Cm{Sc?>bgg+C)lgU`bx-?L~pODMxE0XZ5
zXoVM$PVR0hd*hwWVl@3$I=|d#wzRc9tgF0Qxqeu{;6-Uh4L!NU!oeOr2#&2n`QgV`
z<nZAf?%axIWM5omNvA$OeDJsBULY@AaP%*)j_%{JZ}frIa~fwL@{H<oN6dWLtR
zb-!WL4x|PXQ-O>opb+yjs9clc@O%wlG=K@9=vxsM5+dKUw&$kf3x4y-nISvC9|<pL
zOGC{s$Y;rH@(OOIq#nit4H8o`l_svt?~kxkf^O)Yl7@KGpCrcOb>_k+d^MmZ^n6o9
zD`=9gwg2eV_$XUt6+#G&zHGsIqgu3pO)GHkN{|`aq`hP4k==~(&ky`?E!snn5bXjm
zv;=E1)CITl)5U_~T7Ro5V`1;87l&LyL`Og62@}aJn|prLXVctNjpbeuk~Rq1+!vfW
zb;hOs1j`L|$|`EN3*&cW30^94y+a`SKEkjE!0h@A<*fAVN<Krs^AU|E-Eq;+MLJ7g
zf1CF*F-<CL4dtjCZcLsHxA<g|_iq7Qg+~rH5HO%@8z~^pHygAv?b`N|%6)t^Q#$i!
zJZHgfb<T-=75^7x{^Tgn6pljjni98b7rKJ7m-9mp$2WzwgoJ7`TiXe54?sR|lcZ1n
zG$A0LEC*&OInIu@ztTO!4z9@FOYwW-9@Vg%zzTX^rPOA-Z{uEW(V!Qt7C%6V&+U+W
zL4j~dwJ|}XmNGNQvlcDwsUiCKhQvM|VR2W-E*O9J7ov2%4P9W8MI_NqxHsx2zPW1p
zoHlD=`HmjvsN@e6yiRPF*T|>ctMJ#ynsy1snhpwB4oiw~%qmE4+XZdI(wAorJOQXO
zA+JlZ&o*uq_vly(^!<qyWc0wll-dGaMm6bCuF*sWsY%{=h2)fNnRqch(>NA1PUz<`
z`nuRj-$!H~0Pg!PX~>~`m>CXsuIj<Kwd%ib&m}%|2uUO2yUayg!_c`p&|b=bHAA!>
z7v}mT<*_*gv^i0=@$15ni!=<>58p-_0T#DJL@dFwuNgLUe<B7$B$a^y7i-2pHjVpb
z_AyrNhIa0qVvPq1jy0-xnJbRX_day=USVRLve_XRl?+<;&X2h9nRjzdu>?ejTeRe_
zqb)zB&2Xgt9{{94TfdZif~?Jwtd%gvAho=eGNuDER|1(UeMZ(hTX3u(TCSd-r^-cf
zt7g3!TIupm+%?jy)$Q9c!R>1_+JDT_?W>y{PD`b;icr~*ox65BCK;bqOG>E;Wb7tm
zi6}Pcj?axdR!X=iQprCv5s&S*Eny3<l`ChZx=0JBEl%~deoXE)SBm9EU8Jc!S8IhA
zVjr<$Hu>yY-o{5R(vF*e!3{a*#Y4Ap1zXv=K^q;}db5Pf)kKjMN5j5rvVZl4LQoeu
z_@1bw+X__cA5hbo)_oMvX<FVbU(uGEOz+b!=<E(yH_*(OYq1(}W=?2AHIEF{Janr1
z`zv}fRD>o8xyAxJs~ytCIpUxz4w`$t@1sv<D|Y%gEsB?zlqNIDz_=>J=(3-@Y|cQ#
zZ&|36H%Mi~%0gn7sJKc5u7B;&%_t|;lXHn$^Y3jh?<{?zxD;5$oAF{HT|J)9(8s?%
zT7O5zj|hA6QedNYrj*M9ZZrTio=zzxb^%+md2Fcax?$1G(T+l!{F-!{kosW!)Z#|K
z5|q5z>HZ^$onyDl8LCd_Q(282r$6WF`UPAn?)7NX$_~mMEgHlh>woG@V@-?%-9Usm
zkr4J~vYI=vRh=w5s4JaLcYv;yFBts~?saLq)DHfv3|iy`-po^0mjBusT(h=<w$Hr4
zs*eX~(52iZ%M2F6_@87&fi+!Y($}`(kI>MVPSs~qLzgIra<QART@)+0=3CfYZB{UD
z9EepR$ciY0re!?m9e+hj+wEPh{O4)GM{Ek-l`r5i|K5W+hn&doAUaZMG+KjJxAR^9
zs`NUo&gbO^-Cf(MQ&J)y?&s(?(69N#e~VuyKo@&4JHB33YC<Kn5++io(_`l%J<966
z%u_Fynw69;+LCRbb9qUJ<1o@H(5bV;FHs++8vUm_8X;5rsejGYDVk}fsXl5ZdEa`~
zG@X1dak&C<BIj@fTQ0@<Ka}T@Gfv0d{e!$_TL6TVa^&w)F<1Cs_P4E(3hblwm9q0f
z0U}Jv>_x!lN(u6JcA^pK^YtF05#kCf$AH<+)fahp$@u@mBDTr@@8Q8+uR{l4G{a==
z1wK=%LQ>>!9e*WT&_wk98KWtE)UPXTjOyK++q;269SYRy(}%b(v1&jZdH+#%T#ALS
z8^h=T&p25PC&-GxGv4J%gJ%q$aheLpC0~3GN3i8G&sYGsGn33HoTC&Q-X=adSM0Jr
zfOq9k1cG;SOZwE8oU9)h(1OjK*UEeJ%Cv7cWN*h4eSbK88EMD=Pn#Q5PY-%`fz@uJ
z5?<^?%OnMh#M6;P$2Xm!#sIzy;D<$5BMmhU460;HSJ`SwGRf@^E^+MHut`akYLM07
z(X^QWYL%&|?ZDHj0Xtc+>tT)_i$b9qt}A!s@1mVkCvuZFo)K8XWDmD=8=b+XdGL~e
zL+{z>#D9Ah>W?wC77QGD44k;5Y}1(+3pBGAnn}oix2b5oW6gf>hfrH;G6P1>CJ3(-
zRsSijZugF8t$A{eu_yvHZXkH87_hiy4?nfJH5sZ?8vPAzb|xyfK%Ko@(l8#OPKn9e
z)AW%x*G9MMqgI=1>JRoQ7YW-2(ApgakVM85WPdt00sy`Y;J^DuzzK16n@B;myU#S*
z^7!~XHLUUx;z6ZmyLBsa?5o!O2kuLzr~|C1>u#_0Ki$qiX}5p3b72+q`|%?D6)B)B
zSvB@TU&&@_tw)Wiu8kd`1c56B>Ow^*A%r0X%|wV$;fvF9xw!)Gg;KdVSN!1hJ}y4I
z34gXx<%c53GAXX|&1fdBW4N{z?vMjEGE;Tp{?0Vw>KJ1-t<xl?#9DhV;rBtCF>lq!
z7MF>TTa_2V>cIOrv|koxxlDc_T}30!(pVuWqXkX}C_gfE+L-&5SBjV~b2bW?j%ywo
z^Evqlnrix^^+aDCs+RT`!ItnRo<^lw^na}zh)q0lO-<Yo`VJ(l;>B4B?|o!P+`#?N
zn7+*P<KdGw{1IVx%|)eLJg<nq*{c_Q@~@ZRaW+zg1%~-IS|CT5dt;_%@vK=&s~u;O
z@a*<hFV&l`eUn(+yRN9Muw}cTyc!wSF`J$tD>8*B9oJ=H^^J|#8C=YA3Gd!>QGcqQ
z;&TaP+hRi*k=47L9GWK0ZBm{>=DxTXWlZ+7WEIy>{|T*^7sF)p$hZCBkZdjgg43P8
z;<ODntLR6S`G__KA+|O3%-Hm(Fp%o?)5f_-{R>!22C_3-lOi@P%F(OnbuyjOZSpqb
zQ}Uo!+5Q&U?Zir{Q9We0(>Zf`WPf}WyJLsgVu`lvqjE_V&5ZXb;xMk0$^-niNq5N-
zgz&^0WK+lYVGSld;w+U#C#$=5!ng5MPXxY|<(!F4Fk%xV1wL>8rB8IC+tty+dSP=S
zB?E<Qf><#o3z_TME9wYL^r~!U+d;#*rc(Rw;f-Cr4GsOLMwV>ove7!q>wm<CC!cuH
z%urvdUajDg1g^=x_K;HNVBO0czwPBl6q2e^StiBx-6_lfzP#XX&Q=bSW*5biijwN>
z#J2lkWJ`0ai2!Th0j>{lgQiv2yY0K1PDT=yyG=e3oi-<_cY8hDzwLG!t<GoE>wfK{
z9=_>*A^XnvF-E_#bOvp7JbyP{kM#%k8TV^uAxhRP@VMLpeji!0yP_EpLXMdxps_K}
zPs+4&Zt0Vq@U-F98UkWFKqb;x=d^hchKr^LFhu%yZCOF2Uj%ykHYwkSM*HcqQw^jf
zM^O+AN<rtGSE`_M(&4?pc)-4eNZfQjLc7czR@$g>hiRdx9SWe0SAX3?k_sK#@6++E
zWfPhrIPt)k>ERpAi4W5@wPK}SZi=+0@Q{Oa94K6X1e~+<(2i184DqkS#`+YWl<Lc2
zS%5V=KVe{H(mGtra0JC!bbyz+xpwfhSSsR%NL&@ENKiNc$g-ys)5=fP=jZ90^Vd-F
z_?&ph{7r@hCmM(o4}XLE!VzlTb<m|9`C0t*`ZO6ng-^K@Pu#?{$9Zvx^I}$%c!D(&
zk|@cQJP3c4DG<S~>lDF{edUA&K3eXnhi995`Q9T8QrSwf6$UHR?qmyF27iZkyoo6h
z%hgh)lu*?eNJKFtq8Y)3z)A@h&x8^<50mv~S^}+4W(jTzEq}ikRWY)i@S0yE(As1S
zh!9%14-;Coa^<YF_lynif=z2q^jp}fE5&l7F4UlxHE|&JAuDE69a_fRL;<rT1X}Bc
zjttz~N@c;x+MI%yTH1lI%}PU-0~hzC{TIy2aV0zYE8;Lcsl;8PO&@hbr@2tfT(+K6
zYtf2>n@x*GB!9-3JjY8k1Bs2P(^BcIB2r1?)du}99zZ%$#}@e-S_Es&RoxuZ%r)A~
zQ#Sr^7s4X43q-5HReh+%4PB;i{|LT$GWvhZn*;8hdlJ^rFP@+2)U3BaQ#k#-C)VpV
z%);QYt2Y|Q`k`sh=wC1Dbi&WUYoZ<^n?$ESN!Y~A4}X(!KA`wWphZgQLKo-&!D8%k
zq7z&!Rf~!6kCU+`eLPqobY;B$-zv=o&iZ|e{`L_BYO;&=iAjpHx>%jGxr|R`7ja9J
zzAC*A?RmgR#11~uPMNx4t=@h*emQmGr>aHW#UWX}I?>V0ARp6TW)Iq&IrX~5o}x}<
z?6#Vq5P$h^7g!-mcy}bVQk3$p4iqCO#=}vJK^vvp6t6}01~o1jY5}zOZc`+|*p^79
z$oBkAkBqOv_k;vlCIy+b+o^2PZU>U#cM&9u9Un`Q2QyP)Es1r_d+?xN5HDA9XUJ`A
z7XysBFWkOOD);vV^p}bL7Y8}!h5%#&K*rGtT7NaOH$`jjLtGbKJ)Y0dC7U!4+AD_k
zs+<(TXiN#<bh<bL_%eX+uL3u>c4pK;n<ck8d)JC_`#ag)oHAjv(3VtSoCsl@xI2Lz
zp}{pq&2GEh{Ys;1^|sRQ<374pzF^d7^_AOOtn_dLDa`@yp^mb;VS)Dc+iKOHw7lXp
zb$`pZnodWzyT0A)ox*k7?hYt$t-g&H`r2x@QKvgVH3r&kTlpO~3ZGbCYbF@;;NZ#U
z;L#*VzzrO8fGj!PcVD<?tud$Sv#AkxH-~buo7sW?AO-9~O^ASSxfC=bKvTTz^tS|f
zp_l^T1$dK!cSs|9$83f?*Xu=b0EYyURDaQ*62q*g=_75fjc(OPtv1)xAM8^xCOLuE
zwF7PyBeevKsnh_TjDgMej{+U>VF#X=VuXCIL$iV0uLyA`4l*}A_#r-#X?Q`Uihu%B
zsh5|wLjC5FulG^3YqL=~bIn58#DCt$LA#1~xyPVC9ug1KJT~UU1ND#u*TVMr1b^+P
zHZ|euPjEo(QJmiZEs4;AyLEAO<$ZzRvuI;QtNYXkQ`0Sa-_9DovDb!A>RcVt^2VPU
zZDP7vlY*a`F-;15D>goinYlqTW}Sxfs(HL%T#&oM;K$|e5weXKbm^$Zbz{|It=<cf
zzycA3)aW8YfGBkbb-(Z)-o%s1IDao~rOPjD3LDMFlW7MLl*`B<GMC!FvLaJJe^qDC
zf6s(rgO+5*aA7xPE5W>-?k(}LSQDBWfj*(d_Rvt(X>Xhd;*IfE!bn$Vtc|<<fU3W-
z1~Yl5Uk-EBRN}U&<cH-Ku}@7*>cHub+K4=687OyreEdhFJID#3Z@`U9&3~LYj0*{I
z4m{W|qZbmA$zodvM$Jk{TTDQy#g3_Vfxg=vb*@Vw_AX>VVdYXpHbDu5Bq%U&i6cJm
z-=}eh16n<6S{>L{d`Sy;I1=C&c!%R_I7hj-fk=zA!WJ%W4j$f&U%O`cUJeqsg+O6h
z3q>p+%|G_I(`_Vs>x^e}9;0GIsBdhJ`zK?KW}GCMKF`){`+p=9BR<9oe%asoQ2p
z-i)zhX}msZy6x%_Ie<)Krtx<K5K>@bq{SX#4K+DoKyX$zPY=@A<&A)XUoNs=Uhoce
z0&RE*+7P%<Y<wWPDVY>?MK|4+CS;*!S+<JQhz<I3?H}u6^F%qVtbbs=63u=W)}_h`
zV0|Eb&9UAWygZv;mS}eD<x-qKAdOB~RnbVZw4pUIbVNt4vdE!2Mfs7TYe+X<k!8?4
zbMs}+(#u(<h%uj&k084r-y-QeuHLUVOSsHjp$rNOJH>|owqmhXtQH+~HsiIyiVUPW
zexj0YE6_psfSPk-#(%~GAK8;mb~!{m<t<-P?xUNcbH~w>`jc{>JterKAo80#YG-w~
zZq`b3-titWr`c(5>pq{vI5Sz9X(X?XTop$7$=Lph{mx(P8?CiTikTx@%FV8;yv?dn
zX`a^kxay@KU(uH?)#!a(0^7>|Gr2lUNBKoL%!4{NIG`jBsDF4~E;jHAJ66xj)k@Lt
z)Vz-$!l}Kk8lC)b1Y0f@yR?RWStRU}eHo#GEp>Xx%xr=5=^zIFky^)#OTlY2yYG!`
zyMY3={5S}{75aN&y6jvNh=;g+BSf%XndQ4mxdJU2_j=tPoyVCwLgN>m1;vhOhXgh*
zOg5HVFV9tapMUReaA$yS?)q%3sDj!p@~>_awQ=Y3;2PoozEj$0&~<($@361=u3zIG
zMy&yA_4#|B=qZ6%-LqJwV}HNkUjMG|f`bd##N}(XS;4q*Ao5iplugTc&O7px&MmxL
z`Oh=Ukf6`U8wNb)KPfTij1?xogNQ(-(P#}?-OhLYtAEn#v^t-cA6!miM}K{^{*FrI
z!~Gon2KqIh_;2wWsp#S;=4#ffN=>Lcup=Vwrqg3|iJD5gkE5*K%RKdRsaZ)Wc9(3z
z`Q;@ItuoRo@H-y$VXD!8s-qDywV&EtouZj$n(CuwlJ~7wP1DKe_HVa=vy>hS{D<;5
z*yfqWn12mbt86h;4NSQ7OgJGN;mqVB3mJWi)&>)pEHw<i*<<2GV72TtH4zVHw`anF
zPOcXsjZXvc<ptmNektNB5yw3DMr}ek$`#Py8ryL#Pk5}}y}4C-0~R5p-fo2kS6Y3R
zuyBl&_U*M&!-G~`X`^nhfqUp{tKDW{wA<Tu3x79q$wyp(z=R>Nup!0kMCTu4S(4}?
z)?)v_!ZnKsj$i2H5UDA_JY3<#Rd|V-J)Mr)ED`FVmJb&S8;YaSJchVw$dfTrr-YeC
zZ%F0)JmUSRk#1N|gz#wSaoL~FT+|(@!yysfw9f&JT1}sy)CbL)4ArUK6#*DXCI&ii
z=YPb%KZA?6CFAVC!1HRKNxLuo>ts5mG3Xal8=)!LwE)};QIYhBK4zF1gt0`{46|k(
z1Lj5lgOxfn(S|nlyO&7a*y39@rDGS+u<RBBhT|%)xxi?xO}HdD#p)q0318)CUQTBC
zXOWg98w=5N2euEwKsno%Av`lPMwFv+3x6|bk(YxO8IZw7UV&wvIc(`kLJK+`dN`%?
za{Zzy(&1#rT;-6hU{$&=ppPv3g_bw1zK32_6HWAqHBqP2w>69Rh*r_+XlUp^i8sqC
zZ%yv`B6UokqgA#Y#VN+6&%c<Q*7TRz19`&uag17<f3@RJvr@v*WK5i6yVN)GuYcKW
zqE(e*qrTd@w$?m3uhthYy0geUj~`S_lOKveY&sd|nU(hI^W1otYBQP^H0a&oC^u!k
zxCgsm+v9$5${q*zD|y_vN-10w<pS;FS4;cTlFe`kOV5OaNl^`fG!rKMMXC2iP@#fG
zvNppDfl7cE;1v>HVgRf9TYmKNJb$IWkv4)Pz5U*Ug_;ocY6X`rL|Qa)t<~c>@tc_2
zB-@cgMr6jt8&u)$c{jdEg-%(me|2tM>2$gSbgg{BsBw4OZq=0mZlHQw>GyG;)!i}}
zK~wElQytOynN#)IWbuBK6QSZ(W?3o_z)*3!JNpR&NU2ra7K`QDnV(!aWPcabJ`{mi
zfTcnqyGd1E-CQP?`UZBT*M&oZXJYzIGLMjmN%vH%of1A0OTcd7po)fwzbq}FzpAt6
zzaz>k!7$k{0x#{Amw;DxxB3D^b0C_-mMPGZ4?ddX^jhQL8~QYEWB+ehe9GN#?GtnE
z?1)c5^6^?Ef*o#?9Q+mw?te7ae_yb64f~btyBmi0;&J{<+1JFbk}cKBb^n<%5G8>q
z2}DUCN|Ll(Ip8P>KNNvj;9Y&m62v$=Rb;|p9IqyR{e9P5%5VcS{S(-c`!5NUJi}v;
zsKjkjFxPA=nO02Lba(KQD89A<-JCVu45pV%@Wxe2G;Cw8#~>tAO@CJfs)e`>u`n70
zpy13jyafT}18;VERs+0#@QN_c3tWF-VL>oXv~9=wZH2A9WwSi&^8b8pIx#Juwh(Na
zv9m~l(50XO`!ZzTl$%`l7Ml}kk6fVJEYmHU8ByM1W&{LRQ^O*DqEFWR)kG)Qib}Cz
zt*8AmLdvV%L}jDzm47Rp@}~BB;))spj5**_h>&g{AuX>&NXrl*-To~1NWtKiFyfU?
z^<@l)WmaFAP5`GVc`OT$<>Up}E)W38w4{j;tUpmo<#TqOyNaau9a)e})cWB^AlPy#
zq?2*}!T9r-4dS5OUfhwJx;zjtfA>lN>C9gPVvv#;-aU`4kADY)R_F6JBiIlvK&f^A
zVZTd0>@H>jQ*f8?`<R07ZkGbAFOMGB$2`0hyK$5HI7nksKyrcfo$VFG9;E`n8}HnR
z`ZFk4?6(z*wPLmC9G?`pC$fG@2_JlnJ>9UFm4{UQZD!0hl&48DB7}lOAe{rT8z$q0
zh2R0+wFS0;Vt-jFrCOi6%?9ZGtp4b{rs*F)*H-sMqasu7QPW4ElYHd5%v=GbcyV%;
zZfo8|Qwn(tp$I$jq9g1!8IdhRZ_?5j<oO<y$QL1G2*e<=q~0=)2!?|sxUR={JIW1a
z4U+6;NKS0`u8$k8uGj3%#Ck7-q>xr^>?6XsB@J<bBYzDIy{o&3AY_u9SPcZ6&T+(8
z9z7D|0=W|Wwmt*#;*6dus?g4(JKs>UEwUzKWaxCREZu2|Pm{Ij8L`R1#(41_dDe3-
zq4rVZtW-KZ5$Fi6E-z8pTUaC|lvg1jl?AX|;PxXmsRXwl_PN3BN0?OquxGzDvAZ3?
zJl1qk^MARuTF!bmtvO*ffo`zolgB6W<40&;n39tJfM%+$K58>fw-)OCn2{hT0sGH`
zxyMAmpiNhqa{7`^V&587ON5UQ_0uwkdUgkNTuD59(uO}+yG~uTUgoyDC5R;6bDWqn
zjVzsnV=vbxgo#eX75PF2zDa*H(BtDD&+1gOEPt<kdXUs02^Dx~A;Eh^CMRrdCR;lx
z|84Q!*f07YY-_ANHoA1%qDP1&-qDLQ&K$WTX&YsajF*S0cC4Dmd@sl9bZWe6qew%X
z#0wX(5mZC626dwIEwI5k%$p`JcxLDXZH{g2yj;I%inO)B^f6Jf`1N^i5aOD}R1(~>
zcYj_ysH^1-eng-Sk<J?;uAZ!-b6o*m^7%61JCY6~=$7>kO4$#qZf*2OAOqqQ8v0Ld
zZaF>y_XcdwVE;m1Iubb;A@4Gt?W76GESZ4Y3+f~+UE%yp{S#u9qCmDXi?&es=Guz@
zy|^H3kskCgt>D07lr$#@;*^KW@b-OSo`24>cbgh)fW1u*`2zYkdYo&|)BLka=wVKS
zL%Dx`aHj0qpD3)20N?56lg)B;Zg5l!k%Qss6ehjLnHsPBgiK2%=KRAO`Vh2hX11g_
zXg<L`2M1wYoA9D}j?d1E!uVhtcNm`KlqO8?lhBkg3+Q!XJsB?+x63}V8N<eptbb#M
z9<-tQVrttHr?_}Va0`UqKmbzNAw`0aPIK!XG}KX0-qK7<8zJ?<_^Clh?d@$0iA<rz
ztoW5;&sEwMXvw(O>-JEqLrWAr*K@4OEQl#Z1;A~9D!tElH@GuEH+TI3YqN*0@jsLX
zuD5QKHmY~qWaD3OebDXgdz;CYrhm!?T7RJQSS@YscVam!oyKnj7#kkDsP=Q~3+~W&
z9e2W2Q2U}!{`JD^YPlb=gYw#Hvx0HsK%}lfq+(jebKa4vv<UWc<v&khEFlVBD;w~b
zf7xx$AujSeh+tG2jn<&m?R?k2D!mS&)8z-<{U4}7O*@45od)5#PyDy|)qm%@*pd>D
zQm-ntRBI%*kKhi8<8X}lrD~Mbdzq(RE;TDDb$U+LSHHZZgP|B{6_906yG?zVYV@D#
zXoO7dr#4roXr`H_`ly-Yed|@zbn?0VTXPWDN{<B~lAURIC*!=TzcJyjdhCZicM3~J
ztl1sH#A>UxEkeQJV5VTI6@M%P5_{XjJMVRLV?HA$z#I7ojbqRNBC85g_b(P9hLIlA
z3E$HUz)uhSE!vg&hFa>~&Om8(`b1#lAl8|AIUzw4D=@hQq8ae0GG4e!bAh0E^+3U2
z0&h3%0^lVBKAupw4M76=Vr;joW19zh@u_&bu5~=yaZD^=R+Or-1b?qXXkI6(5ZVk(
zgobse=cm<kU%BIb2aL%*${hl$)g$JHHrGbC>Z4YhYw8d7DHhq~oJZ@&OW$~!DKA1G
zZ{6P)=&cWbXsZ1rj!Ve1;Q`^(v$I-JAb}izWQ4sFv2piyO63vz)vGv3A<7Oo!MwCT
zjS_D|QOa*u6vig&)qh8e@R!&(uWJzqW2J}hyTjc8zP#WsLyxOsNt{K7wm%P_CepdP
zso@@-Nl|Yr{XVASDI}A>5jCag3#Mq)yY0K1PB!{9Ih9*3)qaF>0>i=(Snxv-1~M6o
zz3x{wZf0=Z>T}u(+g7<$KEJU2RH3(}bQ&8cAr(s{tca8?>3_c$m-wa0%>B?kNrDd}
zxW&AC@8z%3!$@hj5&rKxrF|U5n-OQXSGDdx9GLj%Ch1H9WiaTqYIg(Nkn8OIx1-eS
z-Cl#11yF)(<1e_)<XwcVIWLK7cKT~ix%lNsejMUQVMO0Ur%R;ql7G6PdpfC98%j!7
z4+-vxN&&~_k$;O-LOENK(+@B2=bI3m-|d@9m(UH`lCZHz&w_t@_uUl)p7t4a+(+_U
zFnLt!rN+gXhdi=-wJ8uKKuTPelVkjq>GFPh<q|XcjdbCE_ZUcwn;`b2&7IsZ@6IPC
z(*E_LPHD*{%^w?0#$)oVZXx|;_CS8MlfAKjX4$h=%zt<aVIakGWjR)dmN5@Y>!5W@
z7Mc@sn#{G)qW_Oc6H!?^E<$VMf0ch#OoFPNo+!9^SOgWfVXk=A7b71|QS{%(<GDK1
zY^w6Kq(hg%6d5FbnVD5SE7w=!AE)@FR8KW(9k-YIk<*_}F`JAl=y(ko=TC$tYH^+w
zXSs12VSjn$4i`p*-ij#^R+2r;1CwiY1L+zO)@xMA3<jFTX<DjvtP_@eM5@U}w@IY@
zj1C&%KgJa0%~GY>T(oil+dyQQ{i|CxuUiD}w~x?NeIQ1(cY08$fLNgf4|xazSQr=4
z{;K3vW<W0nxw6FPV5{o+h0YVOKjJ!C^5@G;r+=+USKlFayG|RVyv`>UvE?T}=<l1*
zo+!4R<$jjDyetH<U7FC<L@EBWyD#`LtLL4AstpycB?WbGy`3j^GsMJo5PpE4SE|_n
zJcC)30KUB7Z_zZs?lb9^I@WF4FE7$bj@Z3_v)gWWztUSwJ2~SMi+bCw|2*1$rR5)s
zkbhq*Uofg++;OKy47&7-``F2ebblj{yKRgzB++6Js99W5wwJyJUejiy1Uw{rM<DWm
z`<hI1|9>c}qm36dA*w;P5f)MJW%Rl6YVW(i#|PXPqsf@Q^`!pv|9rGd9rXU&u$W9)
zEf&kQvqYFOJMJVu6oHs>n|QtJryxe8n}0lv2)%ev0;gu0Q(aSeIluD$?6&Ymgj;PX
z@Adv%Oq(l2iUR_TNJ}GpUKq0{T=e4B_&!bj3%r3rgpVz0&Bsy@Vw`ej+;`!q%l~#H
zXf8|m@DsOT5w2&&^J5l8S=UOR7tnM*53XsL&9<g`-LJeJ^1EKU8=zL7oI}0a>wn?;
zp#3{i`u%RbrSO0mR_fPwylfe#^N-`XqiLKSXX^9w^ex}K>(E#kdy72cFzFhPF5mK~
zg|(m?`9(qu_K=vD1)^6N?6IqphQS_E4EBKPi}YB41^K`Zgj`}Fa*4sQkU?npD;c+o
zupaMl=c9cy<|8eP+U|oJgo?{!a(`U`B$zyw8AgNTu}B#gmFfwrZwASlG~i*4T!l|7
z1F}d@4!n03O_azN@A$J!6zk#sEv^q*U+_DEVB~|zWR<ZXN2GkW-R7SG*{4VLnaw|g
zKxq=>Awd2agr+hLFQgZy2=9jULa#~d+7R7NnlnP;ZD)`3LE<e~h}(at5`Rr)nTH$o
z1$~^Fz<|`#j5dde!l}TNVj{VI>oV7#z2T7*?+udojG+=eHBx~}^aGO(P>FuP;3siq
zz^@C4b$5dmniYr!*UEqmHlVi&jaIXXdu-PIVx~^OH9&G#0Bj-wyQH8?=qopDnoPGR
z<mCVmc)HRJiqplS8G*<HqJPJk=6+%FR>8MT*@PLcl`45q7fb4dS5d7lTyy|@4a1we
z{s7f5%9jc*1Ci`HU$#drt|pmu=%|vQt<#NHi^a2(ia;9=8Y<Ns;2!EIzqRx+>fxL2
z7e%^T9~-I4DpG7HBRAp+^)BX`1!eWvL-vl*HL?Fui<QhVE8O|Y+<zkM<i6IUpZtP*
z{X6osGQgkLpAL2hST0eg!9Qr{9<6=cQ}r6)R1iMA?cqM|)G^(8N0Ev&uPE_JMsxw}
zRPWxk8&r}|YP;=qYi)d!6QPU|4Dcif70MWQC)1&fQGzl?a0k<40TI0YQUwx$(}GtJ
zZy&}(rHAptOy}Od+kbxuRB3`HUj^!~$ru^>^v(86hzVjnk^M5~$YjGEhEvrvHP##6
zd1OKx#=DLpbSOred6Dtb(1^w^@6k(+x9H8lltCEDA&ek-QWk>RUTy4AyTwf*T+i1U
zzvzEFKK}8nPBn{NkGr$A_9r;wDe*vd*!M=_;uXK4E-<D%6n_f?@LZ<P|7Y)8nA29a
zbpJ|kvd;|Eu-Sm0$tnAm3?iY5fG^<W>`Hu7GPG??K;lTU9iKDv-*2y$1QH1GFhbp8
zO`XZM#lzjb-jDv);q*AiO}Sc6ZUZB+&_mn5-0o!vnmC)@VoD+oa`V|^Z~%kcgm4tL
zu{=u^Px5W_+kYBLG27`ys9o-q0mKar#BZ-hX4n@q#9?i()q~%@X|<vo@!bJ49y(7L
zlP5IEEr7|h%pnlfms@&m0D6(QX-#`a+v<poX7tTpk&umy*-QSUeSHKK_?~FJ3VAP!
z7frGdEC|O)*ov}^^+-`)AbGbtH`hz`{#v<wRX!>E$A2dU-FLEn;DmP}w;f8p+?Usk
z*Rg3)<d!TegT?f-W|yXU%Fx{hWCT<j%hd+&1DP_8SAv7=H{=ow2Y&HE<Na2ogJ&o#
zZ-vWSZZYD*AG;7#KAGvCG=EO3GP>7^7biiykB{(i6J)IYq^|71-4FMNYLQT9Q1Xwg
zF#U4FB!9!F$w-I4z`l4iGGFKrRXD2funJ$4(m+EWdC{Rv$eC|eEOgVkfscs8RhItn
z$bS8!ib_2Zt;QIJvSS}>hW0?mF#5&sCPq#;I8F$2!O(O>;{i-MC%asrvpfIU&m+e>
zX;+Qob5#{(7mzY%L5cT_8h?Den%`VBPfJ7{A%6yBV-I1HTG*TbLW?$H-q{1#W>*z^
z7k>_(_y??ux$*+nOb8`_C{xKSlhfVMOH*q57Xkd=51#Z{PO?R3nOr>}sm~I<r5Sst
zy{%1!^KZKGHv;%+9&q`*Hs0$NcALsHOkz+v1n3gTIBLx3%<_*u9M;qdk@<l*=_n(_
zvwuyD)e-FwRp{P~+ya`T;z9b6HXv&I*o{GeKev@tQlif2Cz}Nn0T2(zRqle42#B|@
zlSV*1DFE>>v`tO~rqeai3loAKJdkxmxdV=1OQde{7Fl@5jF_{-`4)mYyY;#ysAUN;
zToR+-{qk5umE14X==W)=08gjwaH}F{wtwx4Xfxp@`I+zok!-gE3R>`NZA^I{k7yPp
z|8439(;Lijd3czNOeYMcw_j53j)Zf_jHseA=28i8-yeyeZgzI9tNeVHMnp1=G~4wi
zjIXHsBQ|DAf0~bjubVz*ad?3b;=5ASHtDL@Ev3djef1Y4T)hRPtvcVFkYk(tQGf6B
z7xLVqa|EAA((e%EOQ%^UpVjV1Mw*uTQEhVyef)Uc>h!A#475PKAP^U%b~!xTJd^#q
zJxd`cXGUQLI5~ZWNRz^sX!g3`-6nKqtf$*NO{4j!c|3IUaqdWd@JI(gSp=uuna|XX
z^1Sb18pO1QHfpd;TfO;yIrK%dR(~W$hd~_jM(gWfSKPPYFHjTAO<6kU4=j*3!}v}B
z{qmNKuSoi3yA2f3AvhEbwXKMz+P=8mqJ?c=#PZUL?M&09HKpC|#56DxrPWoH`d<WK
zQ=5(EEcY7DyQbdUxHJ3m-DtM!s#*^w5~j!1?2*q(U4^veC*_MuKHs$Z&425bN|;gd
zA!gw`yK0N)soDGBtgbr$fMljSd%4qXsy+EE7}}p~+n=3wS7N`Ts3|J!cibayrJ?|~
z69EBWIgz}{k(X<2zh(jx^LW6?6CROelA>E!;Zf>^ABzZc2V|!`vFYE**_4<$-&Oa$
zbhvTg0<v9@+%tt*5VO={bAQNT*QcOv?3tmJ25j*fvjJ|ch(~TdokKxfoDr;?S8MMZ
zBAtl{{weCOWc62diT;Xp*G^8V^@IjGZ?%;TXOIb9gGZAg3%TM=hKsj|<^j!vq~<}t
z&$}M(?-R>p;p`aVT;N~wAKG~GcQ<KKoqtgXWFkyzCBzkRBWVr{D}SQgI4;){9mJXh
z^dV2ZyJkx=bGyXHn_)#Xlv<xrll`yuN$Ij*)Gqm=!n5V6Q>h9}<lW9yE5a_Mqj(T@
zVP7YWWe6-oB&=5=4@Dpmz%s<)HDWHNk|JicgkGP!DZJN>Z(Cp2?};{IA;|D`I^d!R
z^Hz)pn3m!^@@XqDlYhm}*4xvOG}>TbT0<RX7%}w-^U|3*)S0bthhv~d<p1JWl_{^y
z!Y8lWblPaD+x~Esd&?Y?M=HXEJsi`!w|w?8l<pGG1(&!!5d*OsZyzP*WHBDljMw~h
zK-ZA-L3$?jZ5+M8ziDseZMBKOf_X@Ha-@0#&IMU6uM$~fS%2@EilE$%j>~CIhRU>Z
zLm23kcN6JT%v$KZ8wb37zJ7Dv@(^8WEv45}dmpCAENGKbPDPrdkv6e;2N}sF0I9KY
zK3`_7DG+HfpT9>T!hF62`FspD6BB_cj57I+<N@Fot$KRYSR%a~fYePjN9EcJUJn&e
zZ~432Z#f9+!hbfdu1=Xw5pf%`hv!1X=kHJX7okPiJ%uqo7dt-po7B93oHvXwT!jB_
z%+EEN#q0~;y%0<CV{5P@U9yyRees)@Tb?z(-fw;faM;keN)8Y^Yl+{v=D`>jV_b}J
zF~&{Zqrn(AIT7G7E>@MY4>`P1kJ_|*>*6q?VySW-GJiG@TGVxqVH}Lhb!b+yta7jw
zI@{zG#A^^jge7af+?3YPaOZbh@-AOM7>1EvIX}4`oc0Irh>fG_XO+sSNSGFvY(JHj
zWS+t0@<EcY-sCg<yNOHJ*h82Q=jz#cSxsnTyGLGqx6Tg5+svgq%@dHE%_qmJ$+*)v
z&7WDP*?)A1>Bbusm!4DAEG0W;UU{%U)c}C?J4)IL$vf0`d^mLhuwSs@cl~9)c?e0U
znB-<@o{W)hJE_lCd2{k*ZoD&c_Vz53B8Ox0aa;bEOn@=@h^i+W##Q*)&BzRUV}@w0
zw&8&1Xm72j_WRBDN0_tGzG=01E6^-!#-4uXX@9ftxf1Dg>uMKzcQRoU7LP(!oBg%P
z=msEX*Vio=7Xyu0-c;v$vv=K6{+hevwYpyn-Q0t-H8>I)dFU-XggrTdLp<Qbk8N6x
z-#aXsqYx4H%jy%L1(#F%Vxnrzjbox}*W18E6*_*Kb3h#CJEpLZO{R(>iELGgD6Q*H
z%72yGZ`PESm!QIQ2(s$iSQXPD$U|>VhX_D!>?KcRnMmH*xgi9!#}EMs=#a+(9df*|
zrWvne)1o9lg&QZfV==-N0E2vXabCBlpFtkWvI@Z~3L;V^{)&Q9K0P~C1X}MFUB=!X
z!i31w6@^5pl*$!_EYNJND+<DrZpk2(sekg$4)w(D>&N@giFzb3sbWYCKI<kw2i(N}
zGv5!7;O^Px2?UQzc{g7_sqH{V<`gw9(3)A~8FLSDt*v}kVe_+yasg<rS!ciTjCsGq
zKiX2*3v-_jf3`$I@N?5I52VuPwM7I(-26wP--@7cZ_+3N@h*bG?dzlw6b?b*4u9KI
ziRPCX8hYAxf2UVJep|7f+SzgA>`1`vTr!93$G{v<epH5e?~d@wJGX)Mwb0bO->0Dp
z)9Qx9ewNL-!{{l~ItimE8KS4}h8O>QzE-+@{&J_`#y`=3-P1dA%irS7++x92z1e86
z(5c=3u;$+h#7FBZ$f#&5a)|=;K7U~}hS3;CV;GI4?)hLemYfJ+gOq!2*zR~x<j$xx
z-<wumKLSl(`4I9wad3UM39sL$eu;Nml{Bkkr@ezsSMC<t^`qOl*-5fe!tZAc_7-FY
zACQr0S%Ok_dd;ZCFZE}NWQ=BfQ`sWi(nu2%;CQu81P;Y3!N?3#X4v9Nuz!@sXI2&n
z++V)#sy$wbX!Fg<?BFd)4$kkoNUm=BGZaDIv*wLFh5Ou{K7Z6xKC2(3D>SW%G5A4R
zkiq$yl$3(F;04M`jpYSDlp8_iMq2BYfxsx5%c6yd*&&q~A|+s{q$eBLSFo*}wn}b#
z9wrJrQ2xPQ>8hSDqCllqT7N-9ke2Na5Kc?7-~mBekcZyNLs+#$awEC3N^bprY73S*
zHGBi*s!~a`eQeteK*%tn57Zd)23%P^vm}}XJfdwNc<Gwe&71B50w)Mf0qoOw@SjoR
zkB?VjQpaI|uF1<wS>dR9T&pEI9CoKiu`DO*I;<P-Uc49+l&hha1b^JgPICmaThtk#
z+23^IZ_y|Yv7kqpkjEJ%*Vw8&yK<*)gt~;nUT)f)y3=m+ox2ejonx;6GJ~|%dBX;R
zIK?Pd6u726lRr3IqUq<S61o_9jHri1LU}e)B)KW<KCdq+Js)zv-|V+kD5BLn{oca8
zXGys966*rnpP!_jM}H=3OL#>pk!A8o49Uy?MMQX)A+`Z7PTqsr8Q#5M?q{-8TL;BI
zeetiHl`E*-6VmR2Ot;1Q@m}l4aAnbR`Jy4n@c#2*@dV~1ks0EBg`Lbf4C0ivJALkZ
zxYuXq9W4Q6#A`O5l}`QLr0pbbyP@xfn}3X^70{>FY4@6SwUe5=4S#D}gbrm?512*D
zgb#{Yq)d|+vq&C8ydzgn&JxS>qK`%ZxHmA3XKKNe9>;Xk6nAuFzK}emEe!Joa}%gD
zGELq$b=U5XyxWcAdPgWbX>$TV%-V>!NmpX>nW1ol(X?!2fNV3s;E@h~66cW;Lvx-i
z3e@M6N5$^>=8`bL_J7ND89_Ig55T_A6?^#ABo7ZiCfbNPPPhnzW?pbJpFq}$>xmTu
zDFyXnqPOKn&wOSGY%M?39F1Je0R4-E`@?X5-xH0P9x>7L1w-@cZbVC(UGnx)(nidU
zXKwuk>345b$tZeRgaNlOm_^eKZwq{+1TPq~2chW_%QnZic7I;2y>E!LHR*8nexFz-
zW3KU*CgQ4be6G$ON^c@|*#EfHf_P^p$1Z>?B*yw(G=CS~?1lIpNk=0g>dl$!K}eJh
zHg@^fV1svl&d-ksyjb8(>3|(7dFfc>V3hpSb#^OLlKC-9+Vdbln1#XlvG!|)C6kwI
zk`c@e0<5v;|9=+>-=;gcP2ZAXx*|EkV^!k7a+6S%!2IOl3OoxRHNQ{R55f3|&740^
z!ULU{z9V)nUj3|6IXxC>gKR&QmTdMn9@;}GpPrp6bMkhso}HJ4iG~83+2%@(w?pwZ
za|yyyATPS}Xg)cfm*n;tnm@BnGjc}<+LKM?BwY+tP=9RcMLz0X4m1Ori0&w9D<tnw
z8@i0NJM)=hBcd#R-ByI3%M<uxs?!ZYcRRWkd=uAGg?<X?u3($CiDvVk2d2f|JTZ+S
zV=Z!L{t~6*0^4ts<9hM0_aN~TBmM?AcV^mgdQ<0<(j!-@+6L?z((jP(JofpX`)>3<
z-Smna|9`~25AsR*0uu(@Q8%R4X?40ixHiHE;&i%QwRhcV*8%?Ks`HQQR;Mq-#Svt@
zM72&9FN$EcB{`nq7t)D)MR_jo;=+9L3>*tE|A#zSbTZ0MG{>P9G~(p`-@BYQv6N?F
z3iG#^_qzSv`gDG$tmY7QC>}P04gTH8eDTi5c7HUBg*2~oc#?r#Cmnt6aL@6OXqewg
zZ8MnP!Tb*9chKh)I@lR#exU2eI`h0O(2~_|x6_4rsX5+8JnJ3i4sd!N%yzqp2q8b;
z^!k7nkgpVoLO!YgQ0i)}`B`bXK_dC0*7}`p+8f>Jy#?uJ>-G9dmqXKD{YBmt38PBB
zT7U59(`NB}k9=%?QQPpc?}EGUIw~;M?W-TvZjV3U34nf=eO0VEsa8*pMXG{kgp9N7
z%qO+0k|rRnd;rDb54qL<`Q*;*bYrB2S&_Rpk0*^IRjnV0c@&5hY{#7Bee(y2l#99l
z3nf>W#q{zmOWqcpy2<&h#r_7fU`oB-gnwb1-;J+Ix6O3G?1RA+Gz1G3_TgR*z5%}G
zgZEdWay;{@qjsWPB?>0F=}9voVnI=fc&+*CtUfF}wQ8ktbdt=LcAJN|xFEVgIBOLE
zm4P|f+TDo4kJ@lZZ2C+sZA8Y@wzUUpv)7#$N)7h8o$D1$+X*BB*cF*+@<I=s@qfyd
zzZUQwViK%et(_dH6$#qxQ`^HEJ5vqQ8Z*Q5%6@<}2c+Zx*7IIBbP<|TG0l;jKw?4v
zfQkRMf_|;j?kml9kEskSz-M1eU;!R^20i&?Bj!}dyN%ZvH_6z&>4cD62gnv-EH?K#
z6B7O))culS5Mm5~z<Ur03j7<9H-C?g&(7XQ(mDJJr?dA+Ogd*-zN0f<Oz8bPtD2)#
zzAZM}iL-;%1J0#7`r883P`uM0(PtKpArTT>`0ZjbQFIpZZk5Y+(jHHX&BU@Ds#j>4
zub;}m#hFmoXeRfeq1T8t=G@IyvBu^;s0aguuvmq}&l>S(r32|K*gA9W+J85nuhcH-
zG)S$b^m=NK{g?^MC$O!HUa))3xOMZnoekF`0`Xm(C1)%uFOr;;{65n@-Og7~JxnJx
zd%hSb?fSfqaxdlc_j!{n!_5IoS0zfTMb!U$qqItJx!?bcKk`xO_q)xjn?8($tkpVQ
zexV<}t)+fZTl~9G{80RsEq|Ozn!v|eMAMz{3AdR>3FwUuR~^3c@~`j9$k=PN#q6?V
zf)Pt%luWB@Y9sc*D^z>n1e{u%WkFwA8=c;B@NTTFD9pen8z)DVQ=zokZ)EArfWU$U
zC)G3$zZS%@a70+*rd?cQrN}4Y>HAlqy~@ti@4PJlwiJqtbRr`kX@BH0BoUGlw}I31
zlgj%ekv1URa6=(_Ffq_fjJ^khRm9m$IjbmV0v&&}h57nPZIhPT{@DNIr^zK#8>>*>
zrn&xRFV>byLa@)BE}#x8Z9o)h&cLR41;bA;&MF|4f|qUVl4?OZ_j^`jnOb4L*aPM1
z=|d)sg%972DbWKEK!4L`3(!AYglFID{;j|#=ekt`et?iDLJ}J;fjACPg!XmPh$6HT
z6rt1eqxZ*0YwMNp+qqux9xc-=cqu6{5txF1ugJizOl^p>r_87R1X^KCN^hM|I&sZ`
zp>$%T=QQ`iX!yBwZ?T}%Wre^@0A^elxg%Z>Pu1f}Em22dOMl>5<}i3;02I5S4ICIu
z>P{n;O|z^KEkw%Z+(U_FGutEAK;nL%$)GEx_nc*237x%IBqQyPc>y4(D+?FNZ+N$E
z@6aRwN)Ph`rn_7~<Smho^GK~1-KH#%p)4pc$fNd5VcZEdD(-)(<mHk1!?QEhZ8FH`
z_-haixj$?xSAXa&OCeY2EelK>32wtixXI`xZ+@1tS>pA)WdG-QxkKkz1m}3eR5}`Q
zfyf(fF%2wu6=4P1!%{yaZg2IoO663f2_Tou%c!*ECr7joRrq&ov3KuV+ZMITG0Bi>
zBNy!2X4+`v&AN`;-Pu{CelC7xq(r`ImcjP)A(mF&ZhykG5^jr^i@}FGYJH^HcFDAc
zbQ-5*fON#7WAm8~rvn1bSrE@BuyfCgMd^?L;JiEYw%ET<W5bQ;GeGxPif4h;U?)QR
z0CxjE5OT`S>gD%Wk^Wcs73{$i&seS>C&lsed9|Xd^F?|AgF+1T1fc|?@n>1p;BCS~
zv?Oy?FMrK04IVX%sVQnrr!j_0VBff>6xGE5SCi7yY1>9@7Udgq-Qbzi>z@L7rl7MA
zUN$nF<;r++NGg>SQ59Skn8zfIWX}+#qhc_SOH4rLEv-c-!MhHz?2e9irdb0W?@W`|
zyL!Us*SNADY6K=lGMLGIk>sXYAy=5Y!hDFVaep9w=R8XUxB$YcP{R~V#K{|$`H~N?
zR&Z{^sxRZ;vdI^OTGeBw>I&+0E$w*R=<xm0iGEK^WAw`Y@f<g}8D;QGLXKnUcTWx-
zf|p0N!6V^j+?!n!0UU>JUN)6|_V3I~n&t}LE^oGLoo-jH^;>@trPu4!nhL<i_B%vX
zYJZ<x!0hl{=7mel-tFcO{9316M-?|~(Y;XZsl~B2nT%fZjhi;yp5`yXF)Z4K*~72`
z4PU=zM}_yV-=+iwQXY8!dS54v_pf(y{~9A{i4!Ft;|MyHX&S$I7!ML(_ULI5;5uTs
z4f6wLe)-g^hxh4*UJ5zsMUJu5=(Jj$uYZ8*KASR8@AP}F@4goJ&?n`KO0HD39h@@Z
zLOAtQf51-g`%;A@Le3UUt)&6lj_!`c?hcZd|LbOMyuVg1UzJbF{_#mc_noXCI03Hc
z`-v>wCR(6N$!#%(sxkyph6BIwjwb#7-c3SN;nL8^TVnCn91aHOFc${HZB}vn<bN0w
zxo!>lgQr-KqkM`;QmWjxafz~%HYZRW&_;{}9{BHEqqt=-JADSIoRur7s}xMWppZ*S
zW4HBeMEy|3tB>=?nSnX->-T(D6@LHG@H7~?ht}_zx8%P?2b&i-ir9NqNj3*>r@JpM
zNHxXg;O!Q@nlM#VMj2K{8;rl2HGeM6CEPfhe4A>uTaL6TVbAy@wdf*F7I^6NpHIvi
zbIfV5$KGi;#CRItQSPf_&yhrevMqls<@X3M!$`4UW)W`wTa2A_D*zK<KQYF09yJGe
zEf5eMJVR}+(j;;NH%F^o;HWqU>9l%Qt(E6k1MhDqPF5^2WnXG{hGQGa9)Iu8j*re~
zeMJFI2)}YroZxL;AWmRT@}2psq>Uzz8u{&ll=m~KEpQMK1V1`1S5fqnA-DidGf4cj
zh`(9WOm5LxE0P>E!-QLDqMK&+&@8huR@p`G4DzzLiY$#bnK{R1mHs;@Z$}{Zz>b(q
zOMlG7fLV$+#xTfsFLk}lE`PtM%XeMyS}s1DL;W6J;)6F9KfPX3n7lQUE`Ts%Iw3J(
z+#wi0Y!H)vhnw_6lbM3aj1xx0FWbeUWAV#=BN_`=0&0?lsy0)gxfnMEcBzz4&rTKb
zce%zM!sI(w&(6zgLL2k6{0<w1-mSAk@iub_Tvy=zy5<ve`LvQX_J6!a^Jms+HZ5CA
zNxD_g71^QQ<v=r_-QbRrwnFj_wE;3myEC6D1{K(<f~_jM6;(MHYV&1cl!{R*N~si5
zI@gpE4StI$#{*{*#Qr#7;&0H#FjFT};EzvAkF?bG$Nne4{Q9gafD1<YpOijPS}oqo
zP;WLGY8T$@H6K9JLw|HT#7?>r`1gEmvcJw-Az%qT<+JK`QmoZSf$A>993O|jh-e|m
zO<Pq05FG6QY~%pZh0_mYrk=;=n_iz>sU%-11UwM7=X|xmjlTB~fP{InqTTI$?Je6l
zZ2P2KKJ%uXi?Qt`*><JTSG%OG{Mj__Nmu>c`J%{H!n<@Sb$>KZDfrFz5*hEpuWrxG
zlJ;2_2D(H(F$a@0Il44=^Zc*Og(tl50xf**i`wnou=gr`^<%_)mwKD<S*2a~KIkG<
zFMK=%Yrn*2Wioi(ReNf?roz74id2kw39Kg>%PkS6TIZ%!2R#D3cHQk<wbai!#L3?L
zhy&(aAVx);?0@}fbi~O<oa{vPO5~vkBm&qbe*jQy?m#i<@OIrj-q*cQ-$dEYm~79N
z(n1rPH@VAw5KQ2nl=s&VI@bNVEudr3pk!L7Y=#`+qQ72D$5>94PG%qi&7Z+(e=8)1
z1J!xpxMS@n^=AIDkjnj`S|rpNlt^>&o6OPE*x(?lZhyZLN4)dS_eCF#z@r~nRCDOC
zIPK}xje$O-#(=W#{WYKyXSzn7p|U4*px^6sNb)wN>>EbrOQGZwt1#V|VUpp~WTeBV
zY-m<DU+56c2AYjIv+)L0XfeD|1O-mcatIJ<rc7Q0{Z5?$Do4jC-wDaLf4wk0&(2-=
zwjI)bhJQSM{a20Sb2X`(G8?ZSdV^RD<~^gvA0Mygx3#jFU39wmQ^<cK(;`L&Ni@GD
zc~;M33E7#;=d5XPRN@g?iMy=xZ(6olzSVM7sR(%k1h(dyW!_XHEyYqy9k?ssc*O#=
z+F@jq9DRkqi-5#WjQG)c!A&nH@N8VCo8q{EH-G8mnCs~CrZjGsXg<>-RqBklDKSEq
zc{p-t?^>-j8%?!Nu5S8K*M6Ek(!Oc6NWR^HVZP+L)#>-udLe<%IaJUBu}W4i$Y3rj
zIi3+AS1kg0vAE?iM6f4<>dqtlDQ--lG-^bBV|XUf)@*Fs=ESyb+qTU&wr$(CJ+Y06
z?TM4g&H2uE@9q7k`{|#%pI)`9YOmrpWoE_?M}P%0kDqCmIif5Bok7{O{M$9St_N6V
znh1H6?a(xO>Qn+=%}&#Kbb3<fHj<}p$hQI;Lj0rae@)8Tch?5Rka5Q`Vk990@HFlJ
z7r1@_y&>@Be}O-<oG58UdlV4DMa1ptv14ifwhP`Bj5qX;hb-&<y{NY+ixZ0ic|r~&
zeD$o(b9?`78djR@Odxy{Ig^9MtqRclFR{>GL44E3ozDjO^%>|rX)LRdZ=NZJ{SI3w
zAx>oa;A%J;9+^9|w7>N15Bsza9Ta2DEx~3lLyt`Lu8U_e&j_M#3#>AIYBi<REe`B4
zRf5pFap7k`WWSIcFR%Q3{hy}^jR4v6O2483)&VV?D`V{q7vj$g{AY72P<Oy)ZU2N@
z`5ga_s&<LXk^|dH-G_JYMz=jRpz>+R!=y}&uM|dXNW6I^D~`qV@;s?p{u}k<l+Ze!
zv`Q4<MC?~l)UJuU7NP5jB`ZKHo=mI&a*r8xv@RCpq!K9;_w4d`Sk4b@ty_5{)4bMW
zJ?J~0<vbjxZsm9L_q}9KZ4=OI11^YR#-gIhG~1Us6vgx~O~D(Cekhy#GB5m_C$EZT
z<Gr7%M>lOo)PyY4&P&vFDa9k0So@c@uUEO7q)9TfKwh<cdtKKl2{u6N)<s;dvkf(T
zkKkxZmdR{T0fkX*;gGv<PBb-=Tm>l)<IFeI-Hw3)2uF;au0zz}UJ8J)|2G`!wT798
z3<MefUhCxtZ_oV#=cl0{NJ0&Cdysy*_z<`AdkEX52li?wAk0R{++)3d#$N(K@XN4i
zd;UhYuGQHg`<tOKkc>E@XjNQFCWt<!M&af8%CsQ9>@<|Ffw*RFi+WpP$<qeG-ZhuJ
zryi8v(Vc+IV<FJR90?fp!J#xEOT@GP6}UOL;w9F}y(!9PT=8!e9T?gRsxJmvV?im$
zkCy3z;Q1l>J{a`V<<$&>#-V4A_Uto4Zn8Na-JFOgB)T2$0gBADjMMN1{5PLUhZF2u
z-x@UA?S3NWaHGZlBc}3FdH>ozxB0(_KS4G$8qoC6>Q2xEU9~HQGTsdQI$hcP2$QsE
zwkz|Yk0<VZ+b+C{u+r1gO+*IudHWb)&Mb&4?`6zvVrYzfwAD;&n9F;|;E><1**#cX
z92{WHs{|h==m>4xF`dgMYVU2nCrs9z!{*Z3<q}x+Hv&N!wPY|R%LHo7Se%z$U5|!?
z4RxUUpDQ}Z8CUx*-^#NUTb@yJcF_~r#-&hOI=OBR6VS;mU{95e_7&_PF-nqQ%*4uE
zf^e+HUc|(#>1x~Lk^DK&3bgVtdO6&dPYY9F1VqijNEhulQoq?)ehD%u#-i|bR&sSP
za2@4!1oAULhmNjapt?fpd|I#8zw*K<oGaaPEo^NxvB-KSR~rAwg<zXhGg_Ca4gL74
zMZ2~{9xzlO5A`B7dowl17V`ET(Rdb0GZ*AC`{0$rj8~NrgNqPF_9qZc;`RD@b2ZJ~
zK<(F4vPjzp4SbI$?^I2&dC!AS)}}DLOo7`2HgXbmh~?XR+jd~W6f7K<Z`1k3mT(9t
zntn_9EQ*$E&^tGE=MM8mleWDXmjyrkL1I&b7GV1^PQS~39bqL5j|}fjX%XHp6a^)Y
z$`;kMV1w7fXdZJBBjcQHW428W0>g~yCSkibDe>)(+ulZ()x_8(nZrEf5dXdC^*X}B
z=I=!9!~uK-8$^d$(X7=Bg$731jKfh04<`<mmO=p)DJjx0J(4M~u0Cvd{743ddk#Sn
z3&5|!y5|*bgIZ8u!`h{;x9uC?7^3g#D1Ygg#Am@7l9sLbmPmMwcIOaMhD7<{Ct@xn
zquQnGs@_ZaZUoQJo!K>l^QxbbV_0&Jd}`$Vj~A%;X|#`}rrY;(VEG$1GnL}8MD1LX
zpl-efyunMaDCPw|OOuys0m#+9cjg^MRx*<P%Z&&*qKnK_BjmI-u!ed-mmrx+wtIgs
z>}>4<7`olH>y|@LE7q5lihL&QFDktKU4z&uu{U(G_#|~AZDG4cJ=4R7rK6V}RFmNr
zu>qIh_fD%PRlcg#FrlZY`OXRQAfYk|hK^}?t!o|hVQ=-I<xiE~x|K5&kVLg_Z_pyE
znh}ji;6JC!X$>uMo-|Sq4^;}y9eJ_e1@hI=u|r9Lbe<*Kb3Y}sWyS0N0U(l2Pe*u8
z(6PvkX+6s$IqxkUb*PfY<g#@;GV{+J7y$9_f0+D_;yo|SbpUv=9WQI3;<}DwvFu4M
zen<-xDnwy=E!07TAKTEdzAgGCeYMgRd8%>0c)Co<6QtMo$!XqHgL=ad{nYrlQ7`?=
z+{QDc426=lLPLq|N4eJyxW!<gbG!prDNNV{?uD?L=o?I#(A5*y<3??DwyWYNR{-8K
ztr<Zz-a_aReiKnU^M<ZqvWXERIhdBUH1t<Kb?PLryVu}~gAU|%$>2%H4dDD{{z6Z2
z2&~V+(SJ;}-vx2KdLh_hd2PeP^x6cLro{~3jb6Cr`sWy|k})h3mEU>5AGW}$9d=%Q
zYtm>ak8>yD4Rq4^9Wf)in8#Rg(E&$yHG6Gm4zSHMVA{IsAd*h%?dWH<<%g&nqir+G
z2C4SA_7=_<PNd)i;B2zvqY$}j<i2ajL_!i-hYkJwhSh7X!7*~rOovv!kjWW%=GJ)s
z>U@@P=7_P1g|;)~L2yL{qlAhhPGiHI@}tFvJ?F!m!ZR@v;SNu>F&JZtq5-u1W&SOL
zY;LwBTN0Dzm$ozXW{fNx$z=CWJ>8c_90#p$5U`gt^;*S-Wn%3P?+e#<fuC|5|4vlm
zct$0efhA9rEH<i)W`^{l^KtnjS8ZS~-CT9W%bAElD1%cYM$!Xc7bqM-fkw@F+^Re1
z=!2qz5}@<J3-IQ-kcYUSNdT5;Qs2qaIJxT8kLb?4t!aywm3%+^b5W9abGfcTwX{Yp
z>VJ{)w1Z|r@7E7pZYHMT$CB>bxJy*D_on|xlW_Tien9uL_|o-{Y?@a{uI2M}or?6@
zQ~H;^_d&;lU6qWP29g;DKEeXIAWkHkayR?7hOR-hn}O1Q!m<1%<F15>X-v9wVe*iH
zWGQ$$6pIo)Mnbw}Vu@SN?zDXhG%!~D9~7WNNkT&AN_0v4Mt<yEOho+L`#uFxSh*0E
zI;f-#{_)0X{OCZ`nYs<HX$B6ily#>KzbTiBFmsb(7y~fD<Ih;40Zc69ocli?@?L^A
ziW00!|3TL97GQ|Z>(s60ao6Q~wk!;E!>|lTD+6c_B0j4WzQE}#6NDQQXF7IHhaebx
z00>I+_xft@yH=@l7Sj8eyN{u~erYN+y7WI0T!nw~V38~B*Bkc*;buz-c|b9&TlAT$
z3EF!>NlG1bXcGWlP5@=<P=6+dn=Et5=0q;1DN8GB5ulDh&cMK!n$C#=dISX{7tGB9
zxl}{(d^E_cj9euoh&;)o{~yX&y75vtWNuz=j-*9^TJsq8c&l!yR>0MITh+15m8
z{t$FnD2b+GjgHx75Q!HmDMK2$Siw*WEf*_3tu#T<A37&JzXua3W$t(knIz35(WL7U
zG-D~}Y{28}15}Bil(RKyBN4TzP|}hK{~SR@RB8LMUEo)m`n`0CmjcNP?5xj6*$Rml
zzd_O&d!vC}cpMuf1yzi`UTWz^%2{Cjoq#$O%J7j`F|~(+jQGN7c3<0Ykg;=@F+ak3
z2unjzs@3X1lzQ6lXmmR4^I&v3t+V!)BDUyX4S>8<Gw^5)y~I_f7JpPN8i^ZSJ2IPR
zvGrm%?W3nnPBW=DyIx%mViBX7An~x}LQn|~cn}v1Pj5|JWb-a~=^xqcf!!lzT>*pq
zq&M4<=|8m=He<hX>$Q^{RBg7(LXF_vR8gzr4dS-G5&wrA;dfJCvI$-5NV&9*jhd7P
zs4m*k!9(pPhMC2^$s##yzVTB+lJ^|TB=E1J!Qqm`Z?Z_{^GuY~B0Z;XY5TMWtnB8i
zPTrn^UA?ww3#b)=(SEwq8}1gPgS1hpFk7135`pR1!}Z^W_j>d@GtxcUq=%rTQ?NK;
z>>G9Tai;1IfQL&g)YoHFO(&`2R$zPrf{_wh6zfh&M&mYH!WMKp5k?UsJ4hClaJ5u%
z`%rGwY&=d_#0qEX=7Zk0)E%czSDgKbMZ2n2mnE$b`hSw;Y;PSyj={1KWDTJnBfnq=
zo5o>BUS6SHawxJ&g^=Rl9?T&IRfbhW%+LyBG*g~cd;$pO++vjuOBOlfrw?TS5*PE(
zQ3@Be>9AzI0A_!nMh#uzE^EB}>MAPSb#@xd1wfkbY@cSL8&bMyFk>xBx;V07KB6d&
z3nF52;<>8JlqZXz&V3poovVu;2X2$iYbvObhL&WuPs88B_o>E~H)87rRdodgn#AM1
zJDQxy-C{VE4FSmg)WsfV6vtQqDR&-h#C?ELYqWC!GfXevy!bLgvpHE{Gr=n%(n@h@
zPs2LLPwecPWh&OW2kj3aI<ymEAKHq@V7CcD(J-o8SPY|9E@VAMxM0gJmr8_GzZ>8#
z4})X{2wJB)DVo*_e8LdSdpn1hhizJ4s_@k1<>X78bPm<yMMz=Vn$-cG2yD89AWj@(
zPpTw|HT_e;ci0Q-n|he5vR4&3Si?6*ia;MTwd=_Q8mN4PT}2pnGW(@J<-{<l!2HSP
z16kO^-+uyK@(pzW0>ER5mwj7p9BL(<xqKWwVjw9cZ0zLlo{Km>T96woP8)Ju>3*8h
z-(rN^@v`^v<H?la914IsCsdM4S|o0*do9DuW02gz5*LaJZz|#?$O*{sF-TsAn?|9H
zT<p=F!N6jHU{I*`v}*|&4*w=Xws6Z~VkZG7iT?;m4~CfvsN7%}Med1w1eOZ!oN-V&
zP&whw&dRlv*2f;&cT)Hi7CIL=<{6xy{}^CijW)a|riyz7JOS*E3}25!G0|K<9=Ca)
zrJ#!4I9X7lWZPOmsSAX7*0pP9lOct%XD#16O|-SZTR9%)?VJlT=loD_X)IsLL(!Bx
z{X0&lBldOn%U_-2XZBfC&e}XRn`6so<eQM3DUwF~6BxersGUCz-azZCx4RWh-cULk
z=L$6Ct)#4=Nr27|U3LKCWHrrFmfj-#^L%(1T%Va$AN>e)(1Lj<!`vaR$1QK3mZC<;
z9^KR9cE1Z`mL*1X(BDS>#TpW9{y+pG6Kmk%A~5p#Q<7gQUjJCRRl*Cb_a>-NpUr5m
z#KR;9GNdeh@VWCcO2fmG26ZjwZcf(gV+h_|pzAtQe=MK$5^)LHq!1^``>)uVM}Z;_
z6Kq?ngfX8X;Cn4Oi02U`k){cq;ooAvLP-j*E;rIO+0FLur;L;;Ut<5kZz2+I6F!9M
zCZ4@`?a*T@6f5cDBEW_MPMyDHB-p<=KCS_<k1S^|K3S0+pNcJp-wNTppa{r75|JUB
zV|+gC!GMi<5#^o*+a^+}oZq-?;{p^E(bZnE*A`<@rZsXDNCNuPyVXh`;n`Y~-^O!a
znk|z08xHki4zt@=p&$yPnF7FVz``p$1pX<uUk0|dcem1<kc|<|;>j?!sO}U?Z9$g|
zD0Zw~N45j*Ljdy<?_~<XYHPDxYQ)4}6?Hjj3IO?qn_?JkRiV$>i}*r&MB0maLX#O!
zV|x%4_USvWcc7#-E{_hE#JdX9&eedWnw07&NFkyOM<Lwd2!2JA2}Buv{6T%sBEj^4
zXbcGyg?I36SPr!&l%f2dM~4iTHT5#ko0GCfd6%(ivl9f?gc%<iWjea5(LCRSuhA`f
zeLzImLG4t8u42Nmit$_PlS;<|u2hQv%Gd2fiF61iq4Ra}h6;%nM!oHH@d{-&!dFlR
z?!czM1Jwdv`>Qq2Ca8;@0hJFb(;q)Y%axmH%5Ty_kq^}7!VLHwPRQXP(7ZQa7mO2q
z!q_3Hj6kVdJ&W9zIf0931?yCQkPk+mJOInI?*86!r3Ka-!4`HChnh6yy#uaI|N2(e
zm|qM@A{)=t+=>P~!WNe-4FLFS;NR2=6rt}8>XO}lK~`<NzJ-_W`y6HNm)>N?Kz8eK
zJOB9kPScMYK@R5ZMDEFCw;^`Z!N!dw{~48%5)34gniyC`c+jT>>$<GflVcMe31IZ~
zN^@7}B@#DV`+D&uhp`hpQFP!DN*ottS#5AZFkPOj7ZIwNO@}%rjLw9ixG`0Ebw6AB
zNsFZYmmOJ>bqwT<Wz`6-(%F$fWjQdC%$K$*miLYE7J3CDew%y|{^1hw8k5WqZigDh
z&X&1$5*UhV?gEUyIbCwgbsaPI4A49lr29o6&Tm0?NXl0^I^gGxs)j+Trn(J@Tv7PN
z@5w%3<NxuLMdtdjZTi6n33|89%b<4q?wVtC^4WcHZRd2q%anp@!Eow-^!Z|YkABDU
zuupt%geN?1j_1mQc(%{nuHo$r2n=~)!52~+iZ%?YA8YSQ!9CXY8#6q>6gNbA07VZz
zirSY;AUr!G5bX(`rDvX$78jC*Wu*rW6O@HLiXg{h5|oVwJq{H8!191|O+2A{<H#}|
zX%Q^kN}j(4UMh-8R@%b=4-d9d5=*w-!!^fPW)$`)DfLC}0xjfumZjpPb6{(lh!UO0
z=Uoe#$M2rUf!aYPhJo6-XaCRMh@8dgWub)J3>5rKvT^u%71NK@Wg<brjQ{j+%$i5o
z!<M>|hbpz~4DqUB9yf7NC|%KhcHJ)ddJhW#PN|S3XHd0cUoMknH^jU8L1_Y%Zp-(=
z`LYw?62E-7wCSV_N|L<(t6a5N@3O^bM&H10?H>txW=;iLTgFzeit8e$@(TAaPEFGw
z*xhCPCVZEJ@K#yHeq2nKZ^$`k|Cp)2nCYjb4okO#yz)Mv`m9riMh<bk%qpIum2266
zFZ1%Ro!2VYt(oWX4aj{^f2^-RpSh~*;XYii5U*K?y9WCPxcQSB)|8WK7ym@>N{ErL
zW2I?BY^iGVx@Iqio7Ae+aU>{z+gi0(iehM7K<wA31<>x}tD=hcG@@(k*-i4={;z2P
z1|0e{R5{pB9;hE?$DVlSW+lf7pvz_5aQm~XKeZitw(i%Yl^;6lM2-Wxt(hR-7r%l(
z=_PC2iagDDt96oZ=QV+?go`w)3-FHAKq<zm&@Wwjw!8?i(vu3Ozxp$-sluvpAGefS
zS`>`R>Z(1Qnq_>oMrE33ABX$<;Uy_uVDzw;$L*JJw=<tI7;fUN-jn?RD4J~Q-^4sG
zN~(TJf;UG|3H2N*S&nt5$mH_UG&8!w`uHbSF<&TuKhMCRy9l-X%F>DvTaZgQfw-(;
zH{9Ho#Rsd_|L%!BY4W}sh(|uaidTLlvLw`ZBqNqMn6AsuGEsJpcEc~ye$Tq%xxjgO
zu1gxN(8UJyp-AowMC~C101Qj&Ynaah1Nh<V68#pF_|oY;W_iWRcL-i*;r+#=kD=`?
z+JE!zvLBC*l?Zu-0kpj45-c{}D-`kMixvV|s1#vcB5xcgEj-EiRGV-u*avj{u<XMs
zE%did-w(g!?a|p|=HIOajy#us?YEHg?UO?|m|PM8r>=JmvI92)GIA%;dJm?Xz*6?L
zqdKU#Hi~3~{~?*@99K}%vUWzai%<W)iJ$~G6?oaDaDmCJ)ED_3^4tBc)bShdNSXOE
z;-adI=?W7ixXFgqFLzmfw?F@Pjm$=8_#pk!JPmFOsQIaNng>`A-5R_*Asa5YgwZn;
z9B@b_#55VtlI3(aKtLvpqE)X1Cfa&By+zh#8eiI<2Y=bmAHG(xIzK0OxoDm2GYZT$
zNeO(QdsZX_u7O4I`0L|AO--^T2ZeBlWz~O%!JwygjNP)c-SW`VQN22)KBRL1wFGz$
zG-ZTxfZQbc-0fI{ztakS%0pGXDsgiP73LOwS-;Y{zSNrt5R`UsZB%M~7nUqVWRh7b
zp7`U{g2|Ayv7H1Qy$7KmS9qtM$#ayVxEzdSsfdJa&JlF<{s!7Y`>hC=HwqI?{j$tV
zZZffYkHBtrJ#WHw%LWDsb$BMmUn~mVYB{skLSSa|Sn;vf(8|A~ZgbZs4ZnZ>)F2`Z
zd8=oAG+2-ZoaTaoU!A3b&Mq={qVl14?unWn<vvf|FDz)0N=LN!EycYW|Gu<Ed`aZs
zJsbNHepVhJZ+5%55Bhbwv)IYxEO3+ecY(2=)17{n>#jsmu*yj=wY0XZC_1wH-I8}@
zGMv%#SGXrZ><_DjiSBaLFWIx?tt-G&OB)E47#f-mXkQ;TpN*HV|0OIOWsZFyga9*L
z%r8lYm;U{`#II_6ZCxWsXWjK&1(Q+aC?eJJUJ~fH7O(<<Lteh;ki_21_p7IK`Y4j)
z&q1X0Cs|yLGBicZ$n}g37)pe^OrdzFX*I`0Sjs0Di}le#IpZSNgNk*?UDPW!kwgzz
zPA<DOz=8z-a|nLwk8>!ET$gWBA!=GkrHLRSnKE-D7rYcz;WzAJ0I!rubH11Jd<zka
zC&wa$fQwiY+>&j7_Kp?5W0+r>-sw3qz1Ea;+X`bYN|}%UDcSFU*@?P&-s1TP3f3Gl
zCtMT`4on>l)OMu{fx|R9`$`Jmekc2l1NA&|fLRhcsz?zoXENE)jZWPJ4_1Rv_)(&^
z?=}SCG>@J1BHXYmXOP?((Ldd8O6ou^-1X!X-cv-m@{w@7FpDMQ_{BUtkP0`y5A{yH
zLv@3a^_9BGRG9J0Y|QQ#%9TSF5B-__h};0OJn8k@eqQXQ0&sI#I==+O9xR(cUgqc@
z?aFSk$0!Z{zR@rTJlhi!$ag6-4=;w*C${tUDfy5nD4=_)GaT3ozMo`+vFZs?Jxj=Q
z3*j*b`fux^x{ggO^Svd*XB-15I_BP~@^R`Xg2$MzQ(bdC$nyaLc|WQ6D3A5VG)n12
zR`>SFaRmR8k}dQhU-Y^>%23hzxw`Zb;Nmy2$#p8C1R3g75AFh!q53Whb5}+Y#&JXf
zZO(^4wH)%Sm~S?96fb9^1|sc#n6pV0cHiFHo#6uHOrh7AU6$3|8cApHP0EjtA9(cN
zPalzN{Kv1ZjNv~dVJ~i9ha_CBX>4zHN_RdSt=`4Jmj|o{x0w4DdqW^D_IEy@*Fi}h
zcLU=qmEd3l@)?O85qB%I$-yWA^@20ua9s9;yN@0lr^b$$DJf!R_ILlDJ{>etnkTkm
zyQYc?lO#x&Pw!rvMgj0dLZ_cw=*y13)Q%Ta#(Rw_4y>z=DD4Ix?HklKg$CB)q}C%%
z5loRZSryuN#Sl0u$&@_tRLdcNyXtWHAVQIwCYLaMH5iKN<#neJY7CEKnyx+B+^SXN
ztdfHV+7yNb<ctIhNH6}l1+z*{ZO45a4SImQ5!k(FglW2!!0Iwm%dRTdCOa51*PD)l
z&p2?+ht8bJeTTAHzgnRll>8g=-qPbvQp4^v31@r%uh{h>OLUdOrwTFvi#?F%kCi4!
zyE1CK@{%p~f~|*HXZCol+oC^c;KH%&iA3=OYFYJ{Cd`t!K|<F@CrDADKUHL51DO2w
z&VvujmcF;&E!egdyY+?+(1JP2YG1bY7VaOj3cd;<ebw;Xg&fA10MK|+VA#;r=)ZO`
z5-bC}qqW3?Ibfh~p4kq75H5-kuWMkk<Dn7BKkmAkZz3>ut&pf;xfaqp)UhI1x#Oc4
zO7Q7?UFp8X26&SRG+fl`51K10hg}mA4KIEthp^jb^wdwDOty6?=*HV4*3mDAsJ?IQ
zp=v6=OoGIvBQR~<hkv0w8Fzx|2zUka!=M7I7U=fi^p>UZjh09NX;rk4t*+tj%UT?y
zZs`r<qsUgy<b*H-b7`X|;1u>GHK-t$1vf)Wap24P^Ia*Lp_zieRE@z*fth|WT&igo
zbnVdPoK=sx#!k#W){90nt@#`cxMHL`vDwqpBmLt61&F^})j}<K!(5%k);&_?(5uj<
z#T!HRo91P-N``3wdWh|Td#lFw;iVI0`!gU@&O&qt_mTS=Rq2o8)oaRzCo3Ed_wE+A
z-KM2c!K-y@Jt5gAn^)ZlGN0{R2-IlcUDlBum4WCLlP7mQT+zcYOCBm^DdY`%nJW9u
zaP}&X?|uhQmo31JlM!$;>zY+mskFj#DSsEH{hrC|Gke|uz|gLE23_7l`E<Z~LAZ`Y
zGbaT6k)u+L6w2?dpG&lPPI*brvKla9yl9x}oAYOd^v>2?{=oA!m}*(~3BP-Q(Od=n
z=_jGRV`D2JlyM-~?NBFl0u~~!(!NV&rEDtMLq)?DmV76R;;YE6IgBhb`K*)U%$^HM
zH_oOy@&o+<Z^H?$5M#Zk?uI&!v@1A2$-b=&{-ucU0d!8>$iAykHp>s=-P&qXgf7yI
zM%ck=a{Ku}p*4&d)+kX|WP?{Dh=Hw@`<8TSkUIwIJx^HSws41P@r^ZWxoDmf!n569
z(w$Wy>aLKt69?rwiOXO?=T!MisN;E$)+cBgTC{3_oZ8S!!l$uT=Zy;81ncIq=^CaN
zcWR`Xl&Pr`@=6Y<jyec&*l9rrIG2lTVu2CCaui2Q+=B;{L9_jnqz55t5@|iFWeg*P
zB2fFjsC=vlzrM~c3JaLZwUjZpUTx^G)MGG%*;M8HUm8bU(Agi`H7j$M(stfIfs|aW
z{@GW6KRqQ>Y~KhMVOt=j*BM_e^+^6Keb|&L--E3w4q%M<NteWl@mBbj(2UlYcQx7&
zs$b1K{~Q=s%9N{D>Mv);I1c9~&5CzlIsXy)21peK4!W+?J@QaDAR3hZ!Tn>y@7X`p
z@BaKZLHen{nF6iQ0=n7q36G2TG&Q=qvM+ZZP>qP9(5lJFU{|5@;oaOXFHaORpcEa>
zT~K-xJPTPMcr{SAS6O;dB4MX<t$dbvgrS`WR$f~dTwhQRISiJuNl#M~|44pp%PJ3U
z^o6s!3k?lsQ#&jtC1HaogmmW{pyu*#75aNb&A^`E?~v+#9WvrU8}90!fT%97W=#k<
zfOE7rd#FiwtE*e5zLaXU87dCS<wBdi%ds)qIk9<kbYnLIi}OVLg+4lXRtQ+79n{JN
z0r2uUQKOUB0ZAPgz<%R4foUs(f$tk^&dA^9Te@6_DGb%t1dYBvQ}C(A;{TLHl&%5T
z1=}uO47?h6niF!3?5!zdp+f}UPMTT;n8Vs{fJmZ(W{?F0Y_SW#fJm*{fa`ud5Hp}n
z4*@UOY4);`t=)KvFT@nR!RttuS^DSN5vrTv!uF=cO^rJ(UcSrO{GK-{idq{_{B5k*
zzhPkte%^yPQ=uQzy8Jg*KB`@vf#kS0W>Tt<S6{TBmEelIITgH1oAUa}(uq$Mz=CRQ
z6CNt8HSs$18=AUL8Uo!hSA{^(%X8Qm#8VtL+>0=aIAAo~hSN@Nj$@O)#cT=BT~8_j
zyjsb)U_{gMSk@`!k8f~h<O7}qr&APT7qTDumT*uhWyq}30U6Pnk|v*$X?MyhEow^I
z@>MAW$*&J5nA|Vl8t5;cqk{Y!zywR9GXl3QJ|TlEnVCppW!vF%<W{1u!$GOoP=*P$
z7SX!++9W#p1b1`!q1^9;Dgx6PUaMqS{clZJ{dEYbEx#RCb$r{zoR;rgDpQm~f3=YJ
z(U6C7j#6-niXaO!oWz6@jS9IO1V2$Jw)Kwr#CT!TL$ZZJ3Q=&UgCvhoK&VFtN!u$H
z)V*lzB?de)Ckpa@7pmKuWcKezP-*dO5Xy@vT{*mDNPeM|cIID{p%3aZ%VG$$3<MD=
zDV?LFTVCq8?Z8bkWMPByS2Soc4b2@iP8@Isw6YX3AFWJxX@(BHrpB+a*4d0-Z3&>#
z6=QU`*Y}RLMF#UX?1zcqfU#<X(g7krT0?#9Ok|)Q<!7XDVPp#MUYljl?|WGiw2Wr&
zK}OI-VtZ3ZpZM}|7SVAO-?P@}69>~&XOVVv_w6YI)E4F=LhJhPkCBNr>QEYJqU<xn
zJhTQoAC#Kus8k{v5X*hCb*^qSDD}vEFkD&8Bnw2?yK(5;;CSp@0K>5H_9HG^B3z|z
zqCjdi>}xkFqZ=qyWawo`SJGf&bx@5~NMdzGNG*5@r?Z>R&v5{aW30|D6?-)+Gy<ip
z<0-C0j{CEHcrQGz8(pWB;;Q}FTU>*69vlCsV{E{8>7pXNVNkx0g`p(?k%FZ0-)Fz2
zM!`3Dt!vaCsMlycpp@AI`{mxQyTXy^OWTyat72Y^d>R7vYBpmkw=>uz2oly4&89a0
z-c;6^n4g?48FWs?RVA{?9C=$+Se2wG8)io}I*+xh^2kv|>O36Eh`ErG42iGi%u`ze
z!C%==)|i1c$;5HvfD5FkqexfV^Ob+^0O)XE8!7pGJ>F6c@b!gxPz}y}ZD^D%!x8M3
zh9WdJH7Cn_9f-O$%#T37yGi#Lys&ge_XfU<LC4YeZSwcQqQ@jpJ*eum>&1P?2$FgC
zWu$xEqPQj!20PnUhr;#l@cn70irrd{`Z|4;6-b-N&%_BHu#%DH;b>e>pm=}}eA{{*
zR>wRDHFAUxFn4m+IEUEAujrCCtnHD?KhAjULU1d6Jy#zetYLXTBG=}G{sRSFIAZb1
zi=ZMHG7_T(%fRFT$3%+9GCPL|6FRBQBoYypvJL&`+nOJcCNS|)&hy=KS`_MG-E;+d
zN9pWLyfmm5&KZA2xMkGd6Bjb_WtIxOZue(sPph~Np#GlV7ZbT39w@SfWCAo(l|YVo
z62101{88~Ck2e;II>p|ti&=_QhurS9EP6?38E@`izM4?d%Wn1O2-|2nq&t;VVb0bj
zw=i0h<V+Mqs&A$ni9J%+#+?|>trl`dxEn1IY^Atj+iLHFdz#5ri%P>Qc3-~miuyO%
z0?s`VU|VX0$yk1V3Dl|27ipG<z2ix3hpg_S%BJ#bJjdMC+&iGB7P<qr&6Q3_aARqu
zkjwG;y-MimqtCjM?*l`e?h4+eUTobR+Kv?7od?cecEJVTRTlV)34ETSs_v0K@%B<Y
z{L=d_y|2Y`k61fA1A>JiM`EmL1AdL&s*;@rAeXlxBdG=B`)V-F9GL#;0Z%Ej4*UX%
z(ErV|{ReWYuS(9xe*9iM^n1h?87WRLb+{OGcCh?d_|D0I{xyZ%Ftk3V$WU$3gnPjD
z03{u|YK-w(EJ4rvp%v?)pSP;!{no;1^6+u^VnKXLYgL<IpzQ=6pkt`AH8#jrS&C;3
zP#(Hs1?9pS#VTFdQWvoJF@PU%G(W^nY)~TQBFbK1&yVU`JS;Iv&9zWr#`Xd1+bGm{
zd1`*Qoc^w5PMj6{h!8V}cGK>BBk`XE<)0%(^392Ug?4DDFsrhWZB7<_=%+LJ<HmJ$
z)ud#_-fIk9dDT`IAd_ddF6}Vdv~~~-AQG-q&5pg4?sWnu0hb&-n@-AVvRbi~Q2gUr
z$jYw|AW4J?%%Sxwbe>Sm=I2HrAi1;H`YVO?t8XxsZpy}#K3cbBetuM_<`*3CoSWww
zBNk?TSA(xJN|+&PUg60%=|j@+2BBnN$g=pXM`S+DhXi{1;*layIM>e6u?h7sAjZ~+
z96Lq768?a}YlS|n)cO93wx58BNci#RZ*r$*_Ga27n$GR=DzRUqsTL=;PgsvNiUDnN
zEE$Tq8O?1W&z7p(NOV9gfrSKnMIIskcHn>h>Bxes%QVbJ?^&~yz+!#^_8T}dqE_jU
zr@8BznB{TDTEINoW{k9zb-aBHfGH#r$QXycJ>bAHonc?e=%vOnL&<Nr;&B)(H+w37
zS!E{@t!dmo{qI4l(b888*T9$oE%w@NcuRnk^=iIBhbIL?T9|>Q*pi#i=;DA381hEa
zkfK%qbb$wCFl{Pz%xMgUtUU=sC(Z1FQEP`0-*_H_bGnvTDVDh8hIZyRfRD6GkqV>w
z(&!eNu}I`U*)QI_xDUL;i@e24E4oPbj82zyI@*JusO->oNuI-tGY92rnWx0I+2U(K
z6HArG&8;X-*zts~{xa4Xn%zD&7p;GgqW@H<PQ`QNr;f2zs^i3-u)V3ii4cElblM6p
zDu}#kcGfk+0#xbkQ)nIl7GLkEo&QRZG#Ux~o~1j(ZcgFAU*S2*2j}7(kCiaGfX}S=
z!kY5X_9GGZVi+)`kEMSnT~L<3@0F-`*1Za({&TL>rH-u<&N?lATE;!y{hFSX>n+MW
zUF4f~%9!SUC3|Mv+mtdve4Ke=lmL~M2DM%xkQ^-@_FZ@~?!*2W8T2HWH1^nQz9L^h
zZ}pqlr|{DIT8NM=?29t_Xda8X_C1z=qCV@M#;WvEoN@|wh~=?|I@R~h(d1^h-PRS{
zA*wx{h%7#3xcIZ|7sl>^G4$S_ZZUVNU2}P0aF%T;Q(?8TtY+^czBS>Q@{(a&wBj@r
z*XFJfGDzogx|M(m0O(Q~E_#xC1$^x1GUOwV-Uz{XgX-&VahQ2M7@1n9$$BJ+l+U<K
zz3hj}_a=+$I#rsrN!*zL%_;Gro%<$vIX%V6IBCa1kI+l<JU%LQPQyI^x=(U}82LF8
zp9>jLVbf=NGGMV*XBW1p-5djUfo^;~@x*MJXKC!Unq>1sfH4KcBnWR?F3aeS?_-WG
zt3Xn<^|3L9=-k}Pqeb?`Zu3~J-mh;Lm3~zq@1-r|ADn{46r=q|TQG9J(=L}a?DXj8
z+21>H$YrJ2`25w^$PtoI!VwnyrlDaPqcuk@y%uM%Z2%>=18)4%y*l+N<jd6-8o^t>
zp4e#)OJxTj1_K4AvIFnd2RA-3-^7eZDpG%2pkRpmgBog4u(L|bqnlgRNJVyUME53O
z5{WbS+0n6+U;@hPbovU}lVEZ~i(C%|&E!+$5hA7|7o~<~PejkjfJeKEU7LUyNTkyX
zkk*p=q~M&4B49;yResPuVpxaz9~!JjN!4PT7)!t;><zZN^=!-Lm4|LT4B)!eIkMU1
zFRor-%tkBqp>rL}x*_eHx$T`;GxE6y<hfhfldf_wF{>TGsOe_9=mPaywSld{PFr5{
zbTd~p{wW;J^eD1Dln0SZ++mM}ZPI<=s13^DiwFBb!64Y)CO7UQVoH^zjwCl6f*i5z
zFayww=jTMrwx&V{q1Q3}w`G7Wv(tY0IkRbFRdpDC#B;kN_6qCVPS#<RsJF4bzc+EW
z44NhgB60VikOMjEj^6)fJw)S3lyb)MJeYmvD_oo6RiKW+f#mBkY0&KE?iAkM)b5CB
zO3y4~q#%ZeDZ2JaQb$2Qo}=FDp@OsD4hN9ZdtS3^<(|t+>lB*=reN~d6_me@ZRdgK
z<CsOOUL#cILF=H6<7sy+Q|%RFkU;8J;0|*1^YrcVf6AK_%sgXQ1)R)2gP+x4j9aaU
z?66G4WFziX=tV-B^%AX@Il~|(-s@ezY&0P6yKdh!lFRWOG<@$H-`)|aq278a4{)h#
zrTVH+VBCm|Ow?~ooRE#pB!Z`YgX4nQCgTY4GwsVc0eT6{d!~B@D~L67uR<AbYt>kG
zL?YIY`Y18vqh1<B$!quj_pi_16j`j<&ktSc*ZPQeYR^+EUrD@oe8&k@991)8YF`)H
zR;u45qZROzkwfdQ#@K*Ltf4_LE#v?smWMXC#f(HswI%45$ocA^z@CoI8IX+oI7NsO
zVo&XkdUZU}M2cZ!A-ut1U9r94A5+HvU5C}se0YOtm%r5QWc<v_V{lgN1AEPI+KO=y
z&a2ac2M}{AUk41lI)N<KX4bBq_j)j+B^%q}2sU4I0R|kox(zRI#v^rxqEk<k&M6b5
z2v;Zw6>YU{@B*?(e~|r64Pk`J=>LhsWYYeB0s9~!SxK>A3w%}i^KQ^DarQ1^fGVb-
zZ7Xv+UiE@<1j@g`L%>K;7YMT5CkywtJJ?!>d2*Zi!jE_YT|&&S9Nj%L-kN(w4G$<P
zzxwi6S?n;_xw;dl&?>S~NZ}Pw4u-Q8M=KfjAML!=*5;AGp*Yu}@acinO4^ODOn`${
z^j&qu|NMI=!ssr|k)Jr0IF?>^k4)Zwb|K;&6lUOj1yX#Fb}@frw?A8I=KpkxMn_sl
zTt^pbGfb_E>)pZbni$hBMkYe$g9=D;>yA4)lXVLOAZNPIP!F-GyD!S_TvUc_nrSCD
z`kzkhYZWT%|APMoj3?HR>U45HWcs@w(j93C6^aM3?ckbv=*Cei|7c5>NUnf|@uqv{
zqhuJ>k^Q=<GC5O`kAW}^aT4U_@-O-LTE45~%*|LgAdLYtWV!miY({ZvY7OuyPbXnh
z*<}!tM^4>-40dE2v@L~zR^FUf3apvN8#v}DtW2~biubGMVpb7)kBJ<Q`KzrsqS%U|
z$!gF3bbA-O?gtNkK{o}3m;|g!B^!9w9&_@ek*}M^LdF)e#%4!H(~4TdQH+X3^h)dc
zkRykA&~1hXR30kA65WPNSpcjZSB!rE1YFya!`>BOOtQ#9`y(9D^JGh(2T_)V1tp`~
zL0_KfO*6xwQIlWzr2U^6h%V)nK&Lr}(>{l?CeK~G4CG3X&T^{nA6PJpA9j8q_rfFL
zxynZ`!$Jl7fo{=gAEezofdLO@iO|qFxp;#3)O4vE$zgCm>Z~2D6VTOH!@-}F<y@;;
zpNiAs)bX!n#U9OO1Da~-UrD*cn2){z|3&@%{b73ai5oKevlj6wN9-@v3WLsbrE|DS
zR%@5{6AYSTZ6Rc2<TrvF9d6MPou&h$ceg#`1>_r*W5m!LYU$uYt<e1?LUeIHE$1Cm
z%-=KU#U;)CoaBvb=>UFlapDc$a|)u}N+jI*^N6#F{HJKxb)PSjVz0_&u<Z+66OhY!
zDMey*sJ?o&lsg1UlTh>J^AW${?9!O`@(?z9p&Vp9i`C=O6FY+gNIvdc_Hla0k@X8Q
zQsPk34iA{^`E<9DCBAsFxA4u&Z+Txe4I%;Z=c_$+D2K82R)FL1-Ht3xeaVr`hp>Ck
z)8>;Ho&Mj?L_}X+#l03g_^Oj0d)K`1aR=#BOFrvU$>Q_&a#BRl@OU^NM%0W+qbACH
zx|u@ue)}c>Eiwc=@t}i3yMc?7)PPvJzv2(gW%C_~r!VO-IcP&HEsr$XR%s`=Panpg
zINTFB{C6Gv7T}+-0al(r8Z0kS3MyKTW%5fOGYr{SF_&^_clhZq%^QnaLqjrEvj|t(
zKGm=e&^^j}X9OtXG0B2OH{?A<_db3j{uFB$ns|661SK?CXzpoVmw5;Zu<lDFQ9h4g
zU#Ce{wtV-p<!Rm-BOFgbr$U$6MHx7D?_7Kqs8k$57XXd3tH)(83;#y{OmjYJ#^=zV
z<6voa%1FNeA%}uvdQeD$nMAM^c)S%-&yH^p58msTk(G?RfiBxs)+6CmF@b4s!YH&b
z0@LB@Yj$x2c5$R`VZ_RSts?KR_Mi7fzZ}iWQ1xE}wKEHULAJ~-ESendB1zz9gBtvl
zVn*=+5db!-_gc|#ad777?d}grH6|%AeTsl+e5C5IDSBdo!L=ywLe>y2%!Tq$n9vK|
z5gT$xO#jh=rW-&yGsOCR044#$ABQKU8pj&j3sTosj$KAh^V1ky1>-4$oZ&%!FiL`D
zX$(mWu*pki9n?wUEc>C7TA6!RcnY#EvLnV<0O%Lz7B2B<h_I)xOMj6wzHllw9U;4i
zH6X*X<56in;2WWyLOj2|6{Cq#W6^_?%7tJPSf_*s{V0>flOiEA!VgHLe=bg`^3ix(
z@ntSg36@t?)|erCP(3`=H6hI#audYKWyEh9#E;e1obKDOHipnpP+sA<nRokuCgRO#
z0b7rOMk|~0gIj7_iDSKoTao75npI)6rhaE3rPdRHt3k{^k3V;OjjHg8<Vkvll{g1W
zW0`ZC(kn=QN%oR`r6I6N;9T$&nnr@yM$O>shJsop0n8Mdg=iySG%4XwxGCX0aJaG!
zn^M#ryI|Fx4d5(EZDDj7i7h&JW5fWuZg6`>n}G{0Lsi)PiO^<kcUU2JVjcLENY85_
z=;jqh19ttCL0dMXrj-OA`$)7jf*cAp0^Fj7RoAZE2@?_DC@taYc|Ln54$@mwcV8k&
z99-jgASErYW+G%iQK5Vek9G+#=??ZIn5s(kyXcmkoTZEobzUlr2a+4lB0a!S4LsW9
z$8l0ZZYtt2uZNb7OW32PX-fAjmB@^QX$puP5H>lpv}ua4;UL*di2BpMHAS;fOgXDD
zoUts_SU^wi=saT7aj;(=H)$w;<D_Nq?`r}=G?><^2%hLXcOv2p^9%c3eK6g&zg;>?
zvp<fG9Wv1`jTtg>Ezb^Kp;CZxc}f%q<$JUuX3!alAQ89^7?}u=p<Ky8!(f&qGCPrZ
zJWxtdk)+TSa=-(UQSgJ~+L6(xFjDltwzC(M!ZY0JfOZH6+V^J#`Lw96)O1)!e~Btv
z;xudcNE3M(iu+YuCWgxY)}^1{w5OL08{86cf%hYK3KsZHw&5edu<wpfQ~PCsIP5Kl
z9I>So!B+ZWq@ZQSV~u@7d!H*@@62*FB`6Rr$1$w8tm-qda<kO@c2}g=RmV-W%xO4Z
zIPLGUe}13~vuA<w2SE_5I0wA`{(oyPz03eNK~sI}v=G2mq+EQ>OZJf?ljphGZ)p{8
zhLxg!vEO|Uz`muI<-fW)ODam7p$0#4tyU})GSok+o_$!~Y>L*Ov8Z!I{4tT}I9qIv
z`rL1VlHHY4`@@Y$B&|BLrj!2)j4512#S5SMkR4-)crQ@#gtk2l#m~2DWeWxl03ZY6
zToNERndo5hOWCV@dCEW)-5=<d8t_x^gZr!tp}cqiAb%?N|7Ve3FXxf3mF&*^B_zbh
zn0*kIwyL|dqASA2Pu6TP^Dm#aRIm}3A#4$%xm8*bT)Bo{RprI%dvGRY+HSrDGH}Mz
zDa)$uWVj)O$l8@xd7hc}Y_ZLH_IB_}Bl|`*Qpey2vy)t-Us~ydcbY+Cp)PcP?_b}8
z2V|z^Wwx_dGmj*(%9bA+9Ml`dw;V$Fy}3*i_!Oz%Kqq{M*ljVYWD>q9;f~H**p|-_
z{5nD~7g$v>ST}Jieq>?sig1@%Irr0P*1>Q)^HX4hfb+g;$Y46Nq{BrZjH4J$LyXll
zAD@JIGcF}6Q2?DCxs(O{h}aA{tIMP{0&qZ}4GaPCMV4{hj77LA4YXPR4;$WMA65_w
zD`kYDl=d=Q^d&W68VOk&Sop<PkYj{yKiK-2wZCn~te-SLg84c9iP@c#*fF@x(y$JB
zQH|3g=i0^>U7@GOf;H}20)c&@?eLd~6Nz;<5zZ|kRY{c}e?)~NjDbtW3}h9c!tNVu
zR1$J^eld^7>EpLj@Xn{gxt!-WCHlUzo8UKpwzzeGkNsAb{;Ii8FY>Q-`gTxQ<CX8*
z<mv^f-vg?Dm>j~AYCD}ErZVOpifoK-xRFOoOfco?D&t7$OWx~qkl?R35_cQE_&fWW
zVDB!Vhx*dgHgls^8$llaCcqu=HC@N(WOc6yhDnxB1xC~ZWg)|+d;a($2)e(-RGT?z
z+p4=>k0FLu>d47u?oS4T_!Gl~`hcYvY28WWfr0#Q<-q^iclu`-e|lW>2jsZXG}CLx
z>$Fl;#Usg7h<o!i?@OK1;n#Sx6&_*ZuoW35;M|9S09NBVhv`g7#s(aKyFjXH95|Wt
z+B?p`-jqB_?fFlhWcaPGz$Y++KRL@izl~Lml1{JH?(AZQv)aBbFGFv|#Jj>g8-n+F
z=ZOl=_AZ0TW$Mo;<YVTIQ(pr>G0V8CHTY{d*cxjIp1%)l>G-!a<xgl1HRm1Ps~n^G
zzLa@fUf0WicxhgBj%+{+Zw-?FcsUDa+iSE%f)v!x*-l%8wcVq<w#nA^AcyQdWYT;{
z>kZkL$VAE-m+W%xFZm~pwzSvF*~I2R5$}tH)2qZjGyby{$S$Gq!RdXBG%@U4CylFb
zEu}vSda_Xi2Dm)gNUwvz$qJ`0qF-lUbmRC0_O5Si3Nxt+_U3?bod;3*wCx$n@Z7rz
ziivYZXy;l~@|+IKO##)1aFALK-Q}|13@rE=X2%8Ic->tbURR0s36aP0#c`HCSXvYM
z0Z>UL&$C$ul$}r(A=nNRW@nI``Cc=gM<5ng_|s|njI)?>tU4ahHKp%vQu=k7x~k|K
zC@S+Rh?w#}Z5!bK>W;SK9eURtssB|SC4a%H<<m)?KPI85?8RtRxF}NRVRF$$(>4qZ
zHn*T9|06Po<Af84B^PT^U6!(zTeQ?qR!9{<bN-Sd)<~U$c`J5yDo!{tlj{^s>lB-@
z!uOsS^paAt{+AnaNem*i=sw+-zchO^%5exVy%6X~d$=siv>qCjJ~GjUeBfWg86kG%
z`>N3v5AzG!xd#MGRY}*X_t$aY?!^euq{n(69=-<O0r#_ZUw+ENTn%eb|JSKke^(z{
zX900ksfzf^LU;l0X!624E+^q}ax}GR;v#rNg$`p@1~PkN(Hpi|K|aD9DNZ{RP+4Tt
zx<kSKB~PbD$!EPKn<P}(`Y2mD$lvbX7i3QpC^;YpV}Lw1`DRp^gBX0j@V2kzMXk_F
z$&-YWg~->Dp&^j&FikI#6N*xF8#6V_6$ue^>$Ku%Jyi=0{FaAdbL${u^cDL0EAA*;
zo2YM@9lt$<Qt0k+8-zzZ)-<LKfLGC@OY<W!ocu>(aOL=4iQ#mvk{bIVXPxXS5Iq(#
zh!Qm>&iFvD;npB2!Nlybw=CsZrqH@=`Pcez*Q~oRII2$}0d~b0BYd5)&yYg~aliFg
zeUJ7g!KwIki~37^ltnnM9aB#kod~k$Y>?CGe7di<%t1Y%ZOR0I@jI%&U2SH_GOwlO
z{5BqCJr<1KigDOXOeuq5(VRkq@1%mc#@DLU$e0lI!QfMGo1W@62-<Z~m~IF`le3qo
zVvRi71*)(AI3x0NK=>aC>2ed<XnUr1zg(xYHO_^Da3R#Jgw`kLsj%Kd9Z_><o+q-5
zrFPP#W{lUkK1e?RyF=8+RSf*MahKvhUV~aW3&n_StU0?ION}d+*5eF}mI+uF<wyj4
zTX1spW&?~K*a_!)qr0L%@ceOKMWI^uC0Cxg6~w!f3Pu6Ln>eBxHgAv4TyQ?Y9`0=t
zyaY%D36790xEp>JrymkgP1yU5t0lo6J=7#jkn(^#>!WJG-Bi3|Nx4_Ofm*u}7<%_e
zH}m^pnQB-Hk+Gys^G6rAuHa%w8i^MXLSgLircj@2L`&mFOZ3d)0?z};^%F5qUK?hm
zTC!_?*c+`ualwY!cN5UjGLzKSk>-g<^>wEjPE|6C?9SCBWQM2e%MKKSt#ZBQ)&(gx
zyawBl-t16%4Gb^>pqI(Qen#BFwHCG6?4}<V94K9nS&YblP6jB*NP*>sCyQx?L__}}
zBg&9Cd0UF&O@JeH>JsEZim9-<JbP|gSXxZl>>@_FT0^lIIaS$5ySsKtO!BEgky8NX
zI+d^|xa1c<Gh@{eE7uJMIHtMMnNgo^g{><N>95#`Jsm|HfDBa>7`j-nKVhnPiE16I
zdLK#VSl{1XBZXX+Vlf;6toMNgDT)}We{YrkPkS#<V_Z#>uUf~p>{?s`v}1e0?ty-S
z9YC36%u)_Y4=lf%Pc%#u4}u^`!6mEs=~=VZRum*_Kv1iVxN9X!*4c7hl&PLCPGFU5
zCeWheR@icz0iS5F7mSP59e<-ry2?n?@>TU-2vDj!Mqu`lOVmnMmaQI7-izvV^@%H^
zp!2s<xEx)di`_5mk)!xHX9+b##z!D;iALk@bV@MTVGUn9YM1S%mJK|dK9566h$!jN
zH#!qdh9RnjI26!nCh7#21wwcg3ued=wHC>T?n^g&0G9d``Xl7f#y<FRYf54}bm&j6
ziMv73EFWKs{{{<npuf$R2i(b+ipO%i<_T&>`R^p{CCum0wo5FvZSteR#gb_UU^ESm
zHWu;1aYWbh3TPrB4h?lK2FgTTpnS@kc-nm7i{)L8X!Zq{;8qIRZ+jLAIj?RXEyMqj
zZ6m1V0)RzyG7@Z-{+M&n=b`SdivFBUD9|%O)lM`0KUBSAaHUPuH5}VcCbn(cwrx8T
zp4hf++qP{xnHZB~;xG66JfEt6?5<O%s?Swdo$lRhuf10HJ@5xxv{Ei9T3e3Pqz4^%
z*eFRVDXvW6`*!<I7O8c8Z~SPbeb|WuzUZ)>&$*Bk3+tda#6wrNUj7#Hg1ECj2^ZwT
z*FRbVerjp|XKf0;Y2Zx%wgtu@T73u<&KX+9F+^WjK*R&xzwT>xqQ<h6)c70zwO`O&
zU#h9a+xGl-lh+u;M1t#^s37t6@<=H8O2YDcWuYjFLBT#14{X@}P#QNXb!QY4ok>bh
zmgBjkC3w((Mrag7fS|;1C6VvxIu0}fh@gy@gc4~7&6X>5Gut~~6VWo`mD`@IEl@y%
zf}w1Z0V@r3o~nh3XqWG5UeU#@2NH2@?Kx@^2iAC%Tgy>Ff>7C?bj-j?(}}|TkKg+6
zN5}*b^J4VFg$rV_^bc|c7@gxsBP6Z~u=S{}2{M$@$3>W~2~m<i>z@<v86o9I;p!b{
zl|cV|0qFm%w{lKwuoimkl90D+=jq5v4ycG42OG10?NRC~6^ksNwHY}wmI9evUhy*^
zcrYHU@HL-!g^%{Wgo{os$`F6EIUUPxpIZCTF_g=HGp9=qRtsl3TAo8s=ts4X5($)J
zje+9(k1a^~A9hJ<$gVF&BK0UNq1IMVzEkJ!!)Wl<`3EV54=$(Wf43MwN*AFqU8v^o
z!uy<a2pEqe6fb>c3EHp@m9tRDU`Vt}V9%Wd_jCW;wu>Mw&yL#0ClpKiWfk}q8SiC7
zZA?oo43)ZReZi84AMpYb<Qu7ZB)m`&U76rd#};QNIVOPW^jb0%6>iNM7=eT;CfrJ1
zOY7tKe9xMRdI-!{BzPvEgb)TQ1SCL=Aq=F`nkQ|#wHV_EVAuU0fIZn|Qi%jo&P{ht
z?^4a{uJxsz+{SD9!NFm+-%P0hV_?(!0#-^kjan+8xqsFxGogb{w3QoE$-EjD)>lGR
zk7AYtd@q^C*0xYPdA<*_QXU_BZz7?z0P&RtZ>AB?7$=z%aDmFEh*UF8Qj99v+isPS
zDy9=l;XF!wp64&f6>E5GP)vzny<ZM}UzDItA?KH|QL7offs*uNW@#%&>wzN<x$^HU
zSK&s;hNVc$rs9n{D?wBN_|o&X8e7Yp5V^fEd#qDh;|CS^g!>;1DmAc3@E6RbD>aes
zbFV7SBw&Qp^<T*N^Mf&ni9Cmor*k95`Cx|ul}5^Zg8BfD3@Imcop$g!5{OYAk9}py
zTc(*o3tlEb&zc8%M2otZQwLVV^CHIPq*1AZaK&6MqvGxNe{74%6SK#}<zJ*JcxFl`
z`JUdX#l6_jhUev&1^U5W3t4Rqk@{$;3I<hqfHa@~A8V^4-XJpPR9EV$!da{~tG;7y
zM*Fk7cDeM0d_Gg<5-|-wI}NSmbWiBh(VKSRs?4|KF6(}VGvf`hmdV$&Si)J4^nr=t
zK&Q!NRB&x+uF}kOonjsDn<L#zrDEk`?aBNo>)zt$$2$1n0+9Ioe{TZ%4d1NSUUSga
z!|L*7>A(#h2j0=YS~4;|YL)rQv!TV#wFbdTb4e(UyPcg%XdB-M?^UBh9v279TbpM^
zqdemO)L{n~sY8q^@F^DZpQGtlUO!t3tHo`$Zk8}+PkELu8|LZo7zlkoquoOP?_UFT
zeFy~qM|nGp)|ditk2SE=rduYi<F>3jW}NCgPZ$9%5;^jKY#a`Rg`!u`HePQmnv^4+
zR41#DB{?D#EfkF%m0S0s-&anU2&PCHx>J)GuxK**=Ab_tkrsyXqRuY0DLExMwbbn6
z_DDw5JndqXkeZO{k(m0+Z>>{Q(Iq(Onor754CIPBqf-??{z&bjp&u^xOd71GihvN=
z;;Ur_9humY<K!$VrHnQNmHZ7a1|5l_dY%9qH5npteXmp?K~IZ-=|1(4OThLXq0Qxo
z*}?35vU50D{)<FIytW%2iufn+#*OcJa&^uFJ++!$C~}&Xjjb^sr=en9X;>BCG<FMx
zh^;&$F$0*bvdR%pJtEJ8&yS0et~KF++(-B|bYiVGWETw#b}8?Rpv(;;1Br=QXOh}s
zM3u`QLB^~f3q7fV?AbIT9+bm*vZu(z4$?5hWy@CncvJ?pV`FXZfo9rQAY3z(8F188
zBoZ8R^3eHh--!Q|jjTw$mC|PmvIWL9rckb|&;YKo5G0uc&Ytf(8G<~O7?UF=(Eh3A
z0j^4fM<L%*B+bLo95I17%k$(>^XC3=&&m^Nh*Rj$&xvHAKebgO^Ok1K-xnNEz#P^8
z9r2Xlp`Ub%yfD5p&Kr?Q{(03~9z5x>3?Je!P#>hAdrwTD!)2M#B|ZA^^u(qr*v=~-
zfNCUq?-Uy44%P9D2aP5rlIW%+g{M)86Y8;lBL-KiV_9?3M_!3Z?Y^k&^WI<yQjGE_
zqD6PP#3seRo-myD?w(1DV#$nZgpCKV|NaUuET(0&cRm{T58_99`%wfMngZu+lcRWX
zm~mR?cw5qS%#Na*&tRJ+OW#V$ex;9?vFh!l+fD!^GWCqKcfiZWc}vh%UZ8GGfj^uN
z%9~QJ(8JZd7Fe+pfE>w_vhaK@r{h*YN+il<6J_s<M_sc!1|}X3QGI&nH|Q`zEHHs0
z<v=*HszI@~fve5V6h-s2;!42Ps@x0*LC~ZnpaGKsHbtO9I&uI^8000O7j*PcwZx~@
z?gNj)f6G@&awQEt$h}_oS{Ex~OE2gDX0|<rVfws*bpMtzyICKC%YgE>zwjfMv0ZXL
zbx%)S1#vlZ@0w<ns1=+5F^R~_a^MoP%S@6gQKWaUKAy@F1p^g`X?{KqPBUk8v4YEW
ziqj_mC=<d<Wn#<Sh`i-n#HVdDCS^enB_=XluZad1tWE1HQ|9`Go6<}tI^oX*i1;(S
z#(U+{xf!Yq5<D;R|9jj8-EG$y!gVExutr4<-5Ctmd42?q1!efsc-kP}j>8Gs1@#_s
zG#*BDIev38w1_Z530h=d-={N|va);L9`;fM6xY%g9jjKsP5I!%{_OkR{dB)QiBxQ2
zL@pyGN`ls%48_)A8;Sif5NfiTWC9FZ-sZWv2;IFtrS6xX;JEs%RE?bBD^_gcM@8vV
zF0-Z}Dj3K0^r=^D@CmVK-diT!Lb46i`Dr*$BV3kC#(0S&c5t^6OrMkL(F&XF^q5Hi
z$=c1I{|^N{+TGC*9idvdi^Rx13~0gx>gGCNOWoTyr`4q{aeVN4P7AKlU6V4IV-v@7
zMUHt_BOJGwnmd?TpgI6}o^YvDAWSk<P<a&EQy`%o6i;=-?I}bLQGcdg<W#r3?BI*L
zk5|U_T7dPUigt4?Ll*~H=NQ1s^NWPKH`f}oKb3ZjmS9Lc?M2^N<l&d+Op|#-$%kjJ
zcGl|FFdw<N2y8Q1;*X?I`(4cNx$VWj0ld6duIK>XjeHpwcWKp86LNc<=pMzp2bG`z
zOcSq&=syLlIvH}3lr!nG#VSRYyqz=NicGOW_2KcRXic~7X=svb60iWc4F#r$O$#~e
z1jm=uq+`2-8~RJ|9ffe~Gc1cTJ!{W7|I0aVYx4WTH!PME3VFcadxT2k?OdPnF?Dfd
z=+|}AHRIVR=C(S9fIh$B{6H>{uM=9S6YaR>CsNTqDAD{HqInN%re7)`*@E7tI-JP#
zw(U)rysyA2W~FsPmO6lv@YYHChI7LT&oM869~Du`?8b=f6uDn9)%96aI4d#nOz;W~
zP$q6#T#0Z}afH|0JyQWxXI-R+qFe}6qX?8U&f(*O*SGho@W=TN;zK<afxvUj0e^;x
zw{%e<&xZF0lWup%tWiHG4AuBQlKPH{!Ktrd<B>yX&+WN&8lXq+gh-3k-(!n!fFp!!
zPlykrsS|I`;0yvfMkr2OJ1Q*}Ij7Ptbo^Lgnb{NCVvo7td`dZn!X1ywJ`Br|bEL$1
zDPnir6O}&vj9CkcpKio-2XWf4aoqTbq^8|%TaUPABZVBRIR=|Mv9Vx0ovtqS7~4S1
zkT1LHkvTI}41jGyPR(@QG}igMFmG;_b;*B>*Ixcg!X>{gLab-5gUhr#K|f!>A81;Z
zeJpQ!n%2+Jgc#L%%o8>*X+gQ0<iE73&}~$MH4DzRtxQV6_W=ukn{{K*mQw1<{rxAm
zl+4JTj`n=2NPD@6A{TVvns!~sFEOlyBc5kRF2{-m2#Sg3>xt!0=vJi?2tXUr6d@oH
zVFgua@`NzkN6<A!lpr#t|0P`1EM>bPR*s9kfMV}*NoCFQ6Tzf2>(kv~m;l{VL8$O5
z+O*YOsjE#9e_~LM+|19WjE-qypW0_{-!~h{Af;=r|Ap>eB&_E}E=a1BPc8-GJ$#jH
z;nTPUkfHL41#*qM4}tgC``rXWIB34nd{R7C1m<pH-d|o2{>!hqKq*8=rr(9K8{?=c
z7@9qMZ~j3bp4(HkPo?igZEHjgk?CLtt{LwskE5Uft{~M+yO?iZ_+f)NgZPhl^bR0E
zmXIejVg(4D2vjeFL~2}csPj2xuyZu&n~bpo5QTmCLf5p`$Iu#-=j2Vf-4WA1s?ovo
zm%gd&KQh0pp4kf)#4m)_0t4KPA8vwX(ZY^zgV1NTUSt!Its=ores3`RMVTiZd%1)J
z4)<cIIsqxZYjJ^5JnfJmkOa=`(N6Bs0Nz{-i)X@S!e$B3%dd(Y1p|zrStk8PxqcS_
zj|PisSm%-?-6nHipM5yP{-%|;S3+X8S1|o9z92L~fH4H5ezD!fbGU~dvS0ytso&*k
z6VPk^cureQ6!dTbY{;Tt_7AIp%Gy{8^5tfMJNgqp>Lk~?TyL`|RAh0!wZCanJP?F>
zZ@=@p9Kp)Q-9SSuot%ho^M9~@!2baL_bb8pv4&G_DdCWhA($iHn?NkTG6MT+CKkwQ
zKow57Gi$@xqBf*Rp|gDl;bGUR`lB*>;BrQ1#OejOnIn(8zVl;{{$f3ZEmg!yZ#%ge
zDi?cKngit#_95rEXV@hY`LG+^HQ=kkM<ZRMz0Y%7ArXZBX~rjt=m*FpBNbR<CxNtv
zf<-?sX+xt(!VrLv2s`SI%sPS6H40;=#*)x-H2y|Dhx_$uhq!BRZLzl;rdQh3lNs=J
z-{|>xLc?;VZAAfpv%f{Fir~p-*1V{fifek4f8-$Ar)VqRta45&U50H+bl(`i3K|r;
zwm99Dlrtt`F7<wNR0wD^$QVpB?wxi8LXJCT<Iy#AIlLwzt|JxlhdIqA(G?Alan_D0
znoM#{kpOZm3WFJl+1~|gr|^FCinJYFa}olG*{+k>rXSH|I43tRR@A*%%+&pdiX=C5
zI@H?tigrWuLOAHoO&Zkox}_F+sL~}zjS=JLw$W&<a09{P19<U25+Ua(aS`zm@wH;m
zAaA9&|I6lHr7s?;fHFH6wjGQt4q|yji6X713e>s(g<JvulRYN*ml?h1YjP}x&N&~L
zdA`c1b}87z+P5MtVW$L%(`;BeJZl7^kg<VSJRKg2fE9x}DjI_@(L%m(s=n1XCH%h1
zp^wDJ0mQ@t(AU_k8EY84G5eU%P~i+Yv`edp_`HU)mwi*9Z5PXs4JGZ>3j%lfKt^U9
z6UZGYk}+&Mzq;o73i+PfZDA~nF}>_t6j9cx7pVspIF#LJV~iViL(Oi?$&oeFr4PWF
z*Osg%{aD}rc-boJ0#Aekd9;maV&ehz+P~r7<MMq2@J3<%C}t_kS&3jb>R~@j5(0=s
z51+lGANEFh87B$;G~#s{CJS8fLKAUh79?y--ZUrvD*8Z%|DrT+Bh;1n06adFzEP2L
z5b`S#-k~Pl8i@x44vjGyh++Bxz8(82k0Y#xkEHb97@ju8N-|uD4gCgE!4T<r=|5_v
zK7ttlzTBkzm6<vyyy;4<>tX081vB3S)0!exldSUn(QKlDf*|JAM7SW1iZmv6ESQW8
z2=})~1doSEY;_XwUgZhf>6l^1@*Eu=Qcu29G3-Qx4o1{<*@MZ1D+z&v%bQoNWzAj7
zTvMn?{EDdZ7$Q^u=Hes$KZLi~?I=*9<FY0J@>=_(NU}3-u&qTZ67W2@7s**ILq^2x
zAVkdz0Y)<lKNh!Wt&Akg;_e}mz;a|2XkHCPUy{V0$>X-dbU<?OlJ=x8+N%n++p(;s
zw_BYHOy%<ZX<&LAZaUmCl3VOyzEP%D1v=@-gzjHo`R>ZfPN8b)(oRuXSIR|wReHMs
zQ|Dp<SH8u<E811|Fm4hIo~(q_PgX{|I5@pgvSFsb7WsPZ<3N*_In0~3$ftF=;Khqf
z+mklSUN85&++6gL{b(7*p5PSmB;WIIK>{FXI8qv{W!Ly+v2Ck+@)@)oFea?|Qg&ti
zcb)mIO3<z42^7Hv6{%pcp`(5+?4uEYi%S-^e{>Bzzz;9<_b<cwJrQQ4@(NPocOIjN
z+rOMxjm@p5^R+`DXn1dlJ4i}y=lL^8`LsBid+pHQRI}<TV=1kd;Ly9F^&hKFL@##2
z14}+Um<=vc$hYG6kiR`x>IM2piW~y+`eqUU(kmaY75kRCMnPZPUm15C5AWOnKi3P%
zGOKBj^N}MG^eCO&8O_LcDLlOEJXNrlO128C+-olKBs4#{rwTfzX%@Hj$-0-y%&G^}
zl_-M@3IX5Q<}OMM9LOhUPdS;+Xh<vR@ujTx30Ydy?p5w_oAd)TYUjC(^>deEv8~Pd
zk_He>W%h2Y$;5Zsa@1nz`cN4FYwZ(q?GyTV$y(8D(ZS+hLeIpNq?xTozCkpDnLh~c
z%MU*tM05{=2CJ!iLne3h9lj}5sR7S*F+@hf$@~}!8+x46MvR{KlDaE!=hKZJ;I@b%
zP*2D>e*?h<IU7pF?!JyYG0e~}wD;Y^ZtzJt61{yd)2x~2>?@GgA?x7)ffqOwslu9Z
zxYVhLtyEzsr|$ZbNH3v`PNJK#2%e#KLh8|LMZk|FZ3AZSI6Q7`R}>FCL-J`o2_!h)
zbPrhd!7!^GbVC+K*hzvlBj={rv@m!>tVF$W1kUL7Lx%g4<YeKeoj+|m2<gruFDIU@
zTXg;Rs|*1<A=hS>p32C8aNUm_G#AgsYvDwVKSN>kE$<%8XEal0(^QyjBfNiXf`E_x
zb7J#W%e^VC=r$<sf*zKf@7Wb{H`>m4kDr+~j`yHaUVt9qC2oeZY$ZXr*xGgBKKQ;j
zbL=nmGJRQH>&9+4`h35pcklW@H16g+4{^Rr9wMzYn}PNo_b)90?}!g`jLMrv8l(z*
z&u}lf+#%jT?8CF#?&L2nc`w<yZ_$^3t_q9~7{3~NpNH|g>bd(KL|GvDG>v}817=aZ
z&okhNz}w2;4bP2!eNbvprM+)dhvTo89M;|2d_~Ur!PmC@4gqqaxg$wrYyyB{@ZaU?
zUcY{^^Wyrfxm>*iUiVT-;y%Lravwh-0B;)f61#UXX#&KANfgUAm>+mrU!5|nVRmz>
zEPRV5W#NNSVh-xz7N|R?v#hynTwYHN5_Gwo697?`jnX=SM+@FcOv#~o)+JNiqcWXF
zk;o!GZPH6+Y3=1^f(H3j`Rf}S<SS+ZLLZoInB@7)4kk)~96RiI4__i^X#cvW;k5cA
zGtwjkj{YgbbHU~M5U>w2kv)^NUN?Wb$X}wjS#p{ysAxQp(Tyk-rSOk{h<?^sg~>3A
zg>flQCq5S)k?n(D50*SH^bK^68Cj2lkq<7F;X~d^ImR@+{uke;zk#NRZ!vndTCdMP
zT+b$tba{RO+EXFB5{%H(C8r6Q6U~qGAh44>qFXc}bvHxie^_|11zt<ly&t%Dd<wb!
zK_dbKo`m*!qqT!s!;nQ``LeiDvkH?dhXs+Oua8S3Pv7Jndw2gdc~8nO^MYCERBZMv
zN<FFd%LkEr44Y&cF=jA5nnDs%rTND25PZt8E4j-77x|S<!Z$MR1dnJ{+|c&)ag_)^
zA79WGUtp<2s^#o}Nsw~O2G@QTU49BZGqvHg=!-S)b|ttouL4e5E*%UkYz8fjHfgQr
zuhNwDI2HD5^X)=Ep=IXsy|R1l9E|SB6hXaG$m#{3=+%TI#7Cgei|`bq=Ox2Mg?{EF
zs(@%!)o2j+A4xG1R(yK|BF#GUvT-`Qk?z@C;fSx_Xq)Ng^|hkk0d~_K7a7@KBd~&V
zs@ET~&&nB?7HC?#L_M(|Ny*|FCX8@O+Z4rM<Wglp#wskCu^=pu9Tch>d|CO83U~V~
zCu;d7=gN4KW6ZRkd>7c)f~7tR-uc6K`vCUhtte^^$~}x-7L@jp622`wnm+rrFVsnQ
zM07H7$17gW!y|s7rB|Ek<GWl2>*Goid-mY0^u>ZCWHPl4%tHw7k_ZUlws1mM^PE>|
z*E!$uo3n_2XC*j%J)T)_XX^UEQ-yv%%B<L2%gw<ipjPmF50=0An66VEMT?wVH37IO
zkn6}5mG~cX<Q*=h0bUpxD_Og`e3(j7lJ7+9t@ab{<8kcKF$o4CiFlF@DId-G4@C~{
zZHfWC8+zQU709uac*Bj=PgGW_{03_-0)2ZEJ_bq;oby*7Yn0YZ&nrK#XsB)w+z9zy
zxlT_tv#6f?`;xk_JD3A$Nyy1$9l+%##DS)SRUb1ZPX)LB*CGf*X`Shl`MdJ~LH0VM
z*>WuO6eK`KJnYtYF$GA-04lj8Ah0U*|1(-aC)EXugrBII3#kOu<s5)Zfe_}e8l`Z7
z?)gLj2<seealn6<?a)Apg!-=+r2T=4SlH>HKw?OQ@F|t-05v}K2EtgR>qZn0=V(BS
z6AREFRwROvGc3p#<p11pd#Z)82tiDy`qN2jFa#fu%U1d4W~;<huX!s?|A|6DN6uNr
zZd>;8{7x)oOHv)L8?UqQqY6HR9t1vMl>UaPh;oo<l^!0KH!@8Y00B}Y1(<H7Mg((J
zr4j?)Ky)!_0iVi*dF3KwkYsH&8Gq$Dwp&H`yw=vL3(U=3u5?WY9Gc7jq0()xu0ske
zseU+BP6%}R>I<9Pp8l2!t||K1Cd(``!wd*c)dL$14G2ip6J5DbVx!K~66GYsofUUe
zAt8c;LI4{ZkH&&_No2eq!Y{+Z9+8?T2*SKQjyPG512DTP_ohvK3Xc;V&)<i~N*u3&
z1qhla6c#}f1;&<2T}H%~!pGp^$(>tDClyHnJDZR%%)X*0jBrS}$l?lP<$GfXG~*EV
zbf$rE=Pr~sF^{?29W#8)K#>KbWIirr$9-gy`i<}w?gS52Cli79`-l3l9?#)$;XEu%
zmSHg%0Y=Af7I5Xfpo*WU{#dq)80H@!6g10I797ws6$|@``HeKw2ZNTn5pkY^K>mc|
zRfrC3Y$6y#;7_P?iF7vSza;qtLQML~L_~&D$Gb!eTv|Xin{|v@a52T1u;Rk@gwu>C
zv7ZT8QApM1PJOCW)Dyh<07)sh$MwV7=4DpEL+}oka(+}@fXl=2QMob~FYO8fO3=ON
z25!)B!b+<6kHz?w>4)|{MKP}9&is5p(5OCNyCa2RMhijtQT;7=Owt;wF2gV7l2_<U
zHi@<cBf2**o<zAe#BRxhnUlq`5h?n`bZH2$l#j_~+J@eZvD}3l<e?Lfk%w^|*{UPJ
ztmb(>E)DF|)8~Vb+<+L@dz14@x22yfrQ95Uv{}9>i#(P>S6QI`YW-eeEG)HWp5F1=
zKNkAtUWDEusOp5+)n`@CG~?D2`9F*6l};|l819scAG#G7ZA}3-o;MgCEC+T7N(t^n
zK$FOp%J~=n?06k&d+8QN9*ceE%S(*`9GXK)I<&FIL*U$fM0||10qXLh^kDN)B3oSm
z*B*gzGgLoWad@M&;2~{d&W!wkI+$U{!UOjy^;k=|vqAmE!v(*h*%9GGp$3`UhK+(u
zqnFrkhLK`YRcwY`c}l_Dh*<uG?MaK22=US0^ps<<k#>=C&V6;<4Dyh{qaQowjsP&*
zltBn~zshUHX9ME}iNnQ!)1MSkhyF$%7L=B*YpvfMzF%0;=XEVTwaCNjd11Z>zY8%`
z`OCAm{XzNUU$<&3;hBF9?xYDPcs#0~4@CkACI!{F6?Q5j+8;eu7T8PvrGNlpvVi{i
zT)0^1c@iXKo;awrrOlU4;G0>1Y?d~@8xm%C)bteb5^T|-X`ilufS{_ZXozlhaXvAQ
zdv_r5U#Pgh{U#^SOi;0Z{f|<hnW9LCXp^9lh|6nshpT9pXSZ`dDTCy(TX9n6)}AVy
zGJmB{F+UGg1h_7pf;LJ02Fq?7=6v5T%k;lzk7kB4z}j$w{|bhRT*|%y)GHYMd!Bb8
zLQ*TTXP030$R0jSq-}$Rbt-Ql04t%^Amr#NkfE0ZoO@lNbk!zETyMOj%#B-VDV`T}
zaEhN-T&`JTr#igO@MFpreHXdP>>}Sc{LUir)&&i6%mVRlHEh8c8ABM`tk2`<<V*V8
zdvxGcS!#YoB*)-o0-s(2q)TE!LEW2CdQxIehI}H*u0FHkfu!SYwh>Ui&w(bbgpPnS
zr%els_Yom3dy>a3I}&Qum+@g;RqkZGU6u$lDKw63TgLrn#i|pY47hEdY3!@yTMwmY
z!F2KL1=Q~y0xRQ(`wX3IDXg`6ViZf{^g`uYMAjNv;RmY@5;m0p?g;|@!-$=zuDfix
zbyduxs#~@`a_D7Ky?aVzGJ4h;%oH&`<Qkw)UMG2FXZC<8tJ!wU;1wk)>*ufr{A-oE
zrV#t7DQ^GWJ(36%UfSiJqkT;3dc@zSav=xM$jfKw^n=V5o^bsbJ3^qe8iH;eDkcnk
zI0$#f@$Dl>RbWd2e)JXgPb2Iy=7iC4$gU@wwO?Twe_Ze2qfVipHo?(ZF{A%&rJp8d
z6FP~GoGKc*4lZPV^(Et+B<Gy84VNlI)HoJW7Qo0g+~Ia`ffe(9;Y2YxXmM6z0b)`j
zTHzwO0vip)dEXD9E~=orVIcB^gr5Q@REO$q#ReOlRJND_6t>i~-WgC=e*;A61H1?`
zaBcbI)CW~@PKvBO^%9+*erpsoQciYX8ym7Y9fgPgB{WX?iBW&|g}o{Z-}AhN1Z>nx
z?G__+e;@N7_c{TwKM?$ky-}h4v38PoqW1zDBfX&<m_XIA$*Nz)?0l*}6mcx!cI*qo
zRuF6HEdYDG1EmfTpOPf>)@0fsWGSBECU7YpWeLkqv}fXR)ZP{^qv+3n;mwC(_12aa
z1yZP5xUza04|nx9-m0|aK|8kr!Sf!KvmXx1NUdnOWl171C`OJAmvnJ4o_g}S%IK~6
zg-3KX7#ArY?k{?k(UfvPv&aIO#E|H8m5~Djz%&B`r&w70%To|%mtd2Y8GzCeS`K5s
z<{eBYfO(AX*4$rG8bBwIG}_Sg=AWq@l_@wEp)h@cdV!q_?NOkfV}>@WMO4;{PE&qx
zW^cB`{g1eEC1m$4g@P$Y0=0fB>u<o!g;=b+a=W)_m2|&ypWHKXRsv{LW>mLa*pM<P
zAg_&sX%a1&ic&?fbehG{)_{K_f<^&sy_eUMb>#WvBU3mKLR}`-JP1o$n0|C+LN&j^
z6NU$qhDLT5EXd%uE^u=LL?7;<VHoY0#o+N&17S?R$r19m;~EZ5X>Q5Y=ary~ZsyZ@
z(%3Vhy+*FF9=#t_R-(&Cw1qV&rfMA=;5Elou*(aUN!?1NA+SM^Uj;nFjvas=CnG^I
z0tFDGhxkY11NkaL^~KDsWu<mf`J}o8G+pT0yGMB=aZb-b&B><m3vT-C%42i^m@1JX
z#zD@#f~9#AX+QNd_q{@EPV45|%%FcW`@X2GqpW=Y?`M0bffS#kd<8>_Od!z*psfO%
zE<*P&DXk|{0gYcO7|oS%u5d?0tp&S`^#%8n<yO+zE5FJh@3`Zkw6~x)Y?eLTR3Ueq
z5d|K_#c*!9o9yyM)ot9|6vcm2?&Oha;)r6@iM)Dj@f5m>0@-=_{1F@tG0{3U$tVB$
zwf~NkLVe1<NLQuxWSKr^0f5P-Mg_fG_pUkEKb-K(#w0ZK2;R6>7escsTw)^~=B+Sz
z|1c3jka?s;cL?%#_?+h4h}FQWNN`R;7(s9<F4)o{(Nf}Sgt;o_-^Xs-ILn9#IBaAw
zFU;y81^KVnj}{;WM`QqF!##OKv$Asr4AS?zD~|Pjz0FoS$%3E{Vn9r@>`doNnn`Sl
z<aG6GFachcg_2xr0clo@*(%p>Cr%@t&?7HisfIGoVYY4}hD9D708crt+9fJYP3SEt
znX#V`x?BSCeUJJ%UM7J)ja><Y@V-T%$LHBV3ge~NW2xUtwGa8ql(uhfjXII`4@6I2
zH!hZB`arUj!fI~u1we2yFr~Lyc~Y0iejAx7dt13g&kr`dojOs=+4iYqJu;b9D_U`t
zQCx&hq~(^)1x<pQnObg?eR`N$n_QVm^LivVl6|ars7vR>KhBOpDH66$$%3)cRqI`z
z$g@oYJ}N4YQRy(R>_2lb$Sj_)FHO~#wK4RrigY$cR|3vs0Mt52whovc%!pkxD0g)m
ziS~3?>ao&rJYDJvS$gMT)i}^kc;@(YRO_+QU$mS%Q-W=<6-(f=Pd<6g^aWW|%Oxy6
zT?G&X528Kw)hvyF?`x0Fd6-%-)y^sSPM`Zf*_vv4FXO^m_tf)z?HNw%SFF(gh#_SG
zbeIcWKZv{c0cEOzbRFaRy|euc{c}F?fROL2zx&2O3MF2eTpCi(LYblFnc&XS|L0`&
z6elpxwXRq<OgETHR{u5V;spbNg;0Y#{MbynpiLeUmsyLN#;+*BB9+%!yIMlub<3x0
z-SDDLufdq9{z@CP!OGCPC%|1PYKh;0<ysSZ<CqDUl3B9M?Noj5_d&;@pMRZccgCu_
z(S$phjk+)rz;G52ZfMYfA`Fh?oa*W{_k=D@U!~Y8BJjS#Px7i`l2j1?{L3gV>I=dQ
zT4us8Pryl;7I}7Qd{#mps+2J$YE~(?HEhnXyDKhHyGzJ4X<tCuHFTlWf}BGB^l4l)
z@X`o4FU9H&8ktQtUK%eg=by{%@lE7ugMf(TcZ+s^h(aedd!mq;Zc|`L0vjYP&%MQr
zvT4ksrD2s&>+lqE@FUHj166pAf($#7L}!}XJQ%n)a!Z7R6ZnF`KF8X@e}A*jl0!Z<
z(3G4rlVI|U(Z1!+E%MQI9H!B{{R8*jvRDlOq&l<ax{#Ho<eh81wFL#Ei2K`(3Yq6L
zX=Qw5`k7+g68}m-y39*32orH>qa;x0lG9qFb~@&wiGM7Rc-v(0IEW80X_MUyJJJbT
zm_BlI{Mg0ke$Cscg!OVL4c`Z=q^PR<txSKb5huRk%#(piktW!1AdkKRYYIkY8#)O1
z4X)M|D8`>&yH>KXu)>S2JqieA-3F$}I@p8^X1Ft#!QMcp@W`49D*NL)NOBcx!=p$b
zDhfR<n<BIB9I4Oujg3=pWVz}>;Kl#?HuXD&ZK)vhOLo(iI@74b_&^z~m1UvLG8URF
zq>;DTL>DdNh1`fDf+zNpM@W5_eX$JCT8|bNMH#@n(13U7m3f9SzN|jn#GYDAn~pG}
zQ!d(S6tj@G_~t)Rw6zgdVVvpxeNFV2i%pJlmi0=9#8d`0Y?a`HXzMBh5^S`>46%!k
zb0^vu1wWsWs9ieuBQFpN{JxSCUJQuX>qtvBRB`ehf>~$+Bbg5qYi{sO>;3~^SxAOV
zlkgf<M5sOhXSf*$N2$HVf}gG$(2ou*x$ednCUo_KF%_!J1p8y)_cN!icP{*4e6?kg
z<#9I{FG&$O|L0rb>ep<7q>M;Mvvj(00$G%FvS5S^&M)DYWMpzAlTgWv>cr2-_wOlw
zw>D0TkJ15dFfsf>);NW0_R<6}jJ`;beT19G1}7zSn4c4LS4Djc7r!YM!^+?=uJ(Dl
zrIw_+j;N=%Qgu)D8yZy)IUz-z77S(YbT&{qD=SShT03VjxC{1%9wm-u=4Q}`hOOjW
z{h{X0r_H`0<g81si(sKUCH-Hq@;D-9gx+KvU5UWv2Arnofb;txcA6&u$NTuyeO&G+
zU$m2OefL0lS5&{#_;<VCtr@9Ocop?X5efRIvPJaOlFb%bl3?fv`#qB<p;}?T3Ym|z
z&=hjw$<xj6HeyG7Ui4g-5Ka{1xC%KGB{~M>;daJ>wB+fLIICIK3|drOb6NaV>Y2Ew
zD#Ss^!Os=qs5CsXTxd=}3${xFZQuJ~Bb+O+hRj;4o94xMt{NdG{_yO^iwX&QkgV7L
z0QU(}RrMD3O%UHWV0M92O468Q<o6gBbfX+aZ#_Z*l)K@S0fcK<bwX7BL;}z*?5n>+
z3s1#z$*<~~M1VNOULEe`WBhWz%ai-dsq$gp83=$8l|yCN0q<!njYcM@_3BXgR03K5
zYDdGn3A%73I8t5NjL8nK*HIc!?xVG+MXN?=kSQ{ZZC+Byhv}2m-{=o%)WWG6j7_bd
zg{!eTDpJG|$<?EE&z`a48Bj4L8V*2mFx}o(rozTrBFrA=|C}-W-0^a#>38Ep9kR?&
z8YQ+~c<m^w0Z^%CB^44wU`EFEJ>&PR;dW`*;=`fN(T1JCBnr|h_Go3d8)1TPLJCLf
zt@v06>FaBDM!p0KH1X6JIkkZ-zq-^!S*UB_?uAR*jY3j6K1+o*lvTiRh9fdIof>|-
z<{#3gd#QimGMFxt4e*k$wKU<bZth_9fHtaU+I${=0F>^6DX~9;nn;slW@P}2RA|pE
zwC*X~4?mZDFohB3TR)RN?I`^IWxZnFEmgcw(Y8bO=2i=y+G_PIO?m}A)6_T~E&H$q
zsFU?UB&mE{)Rq#^McS966SDg3XfV}vo)d%FB-sFhq*cRUF(4{9Lic_)ZmE8-TnYA&
zh>0wC!0nK5p+t&iQuv}p3QDe;yOd}cEVg^LM@$u%^Pgy~;auBR#LS0*FZL->63fy+
za>olbAaaVrUK0B!oz07q1Co5GNQDxe@c?1}D<r{L^7ZcEoB_n!b$(m~gel&PA8~A*
zL`=M`KQ{VoJS+6togvnxg$7Sl4V!#36VEMW04;`K9MgtdJeAAFx(r{aE$9RLCYtu-
zrCa#q1Cug)h0ROR=iDneLT60b%0K%D9}I4`H(R)z?50*+nQn8nBxBd{K)KUS8N%Wr
zF<4}`r|gFr!Wh{KbtsL8Ay`CF#O~hJEp~VwX__?rr<ggdI;HZ$BlD_xbP59pN+D!O
z06yKyNB`4Skqkp}oQCrvG(99@2(n-8oGiAm_(_Y|?R?&r7sCY)d`jB`PWeOTj0Fb3
zMyaer28kgX&l52LvXpoh@IK<AGJjHC;l`fA7qAVObeUbnUK5lu`TDGy0y>a{x2_*-
zeJBTzqoY0VSC@FMClrZ*2}tnRN2#!=0YN~Ur2f)Lz^cMP*VO)j4eEA#0A``;=tXS`
zLI{xtcNk{g3z#Ec{l4&n{IFv8w_K*E3a`J-+N}3%Iw&lE$S&HJ5o0SMZ7TbEQ!AXM
z^;oOYW-g;Hv%H$#FuBs%2TX;rU;)({{WVVgbvv{5^y8)KIG2Hjdhu@5Wzb6bfI!-S
z86fv@W_*=DT*?TwxJ%^u#|~oK?If4qLEke@R$EiCU0_$51V9Jbt_88}<E;D07*_VO
z)n3z6kj_G0o1j}y<*<5sCW4bU4HB%oV*#MryOi+^n=Dtkc3lcmGB$gUJFIu2)hJHC
zH7jdqHf-Q6<*5rYDwkGu^#}8~0SKZ{*@%4NK083Gtb#qrW=9*KnLiuTT8778W|jCW
z$jEo2aC|Ba06Atpy#lqM+*h3GgoB}0he@NMqsC&9LqsE1+pDEm4Z-%PZ<e>s`SJst
zJA{!(24y2}wKbnWF&Z|qB5alRTzZLcv$VI`r3OYleiS))S#LgCp7G5YU@RRC2@6(J
zqLq)<FecQRRaN3A$--w<J(8l+qhMoIrI=CdNftYU2X3;?cV%+eg2o%mcQ|`)P~_b`
zQ(H)Q@(6s28hHRt{527i=qHz`ALf@3%EQ=VLHfOTJf=4EGem%Uh=F$+^F&1qi1MXY
z&-+_IZold5&^#9=UKRk3R=Jj?B5b8qLpuAv&vI9|-~z@dPKJ^is${BVkBxb4BcBdE
zrG}OBaoX(GF0e=%lP7U^XD6lZy9Xm!(~hA8W{C*znc+dGINNJ1Ol4r8En&fhn2AvO
z`Y$K^I{b12!_&j7m7;M}_!c5j53Uy6Z~nXNcYh(pRhm4Ia07mUL*m!OCI8+R!Oi|m
z_D@l$?B`07n-v!flO6(hmZSRPx!eC)-?qlqy<4wlsO*L)n}>+h-A*ydmHZ61hNtXN
z=Zdw}$wkF%T}WU5L1i4u-EO8wXjB$HpRYk|#=6GVnR=^&;df-O5qcKg+Zqy=7N#vm
z|0N&U(OvDR{06uiu(KiWF&Rt#e}P}N3ucJNXm-`l$^Ov902F}lWbf$1tY3w)s&X5|
z$Y!yg^&g*_L$dr%iC|tMj&vjvBLn+jI2-sveZJB*Y|$PxHex$jtgwvMAYV)lfq&Zy
zVF-bu<ET>Uf&#rx6N`$s8CzSr^b|0pV19&vE9D5q23^`U#8}iJgQY56NODy2mPQud
zO9Xhl)`A6uQHE{kjocQzoKGr<6FYJ_I>-}j5(65uRYz*Cq2!p~iA-cWMtiV0wl>2*
zI4I3bF}bgy4^>NcfE~Yw-KE{r!{Z6MiV<g4pw}WZoBkqs8i_+<W*m#>{jq)>SK&4w
z`(tu8!}vyP!DHL;btkpIw;f#Us(hypv;~soPb<?0B8i-wEw`}_QLeiJc#o)>8i#Hu
zQm&K43d{*lbpK|d9oKz*SXhq+-ZSU%U31yGRh+-ymJ31SyC?D2!nBZr35Q|jhs#Y1
z_bslpkunxI+csDMw;K+<mgq=Kd0i3!OWU%!DWyNSQZgwGm2Gop(Gq*`y1)VlgMkgG
zZafHIG5ZnXr)aWZAJQad`6{>+v|wcE!u@1yMxA!3WjXae_^a_^DZVeKU!#_mWq=%(
z^{HjRUx}-|psD6vzIk|p{}m;EO(%4aea^&XnbeMF&V;#iJD7W>)#k31HlPo1X3(`{
z*CW&9CsKRbHkSwb*<r|WocB1)_63UrCGf^*rR{&aZLs@HaGXa$iNpB&r&4MPv}rQu
zcw3>;4hg|&MjdHp;VkeXQL{zCf8@pOU9QO)4XCc;YZ#-io~`SSf4?f-T7qD7eMWeV
z`L|I+e2b4u-H|sF84p@wyw3&zj9B!I=K4<w=k3FSnrsjVAhpv}e+LMVNO8Q^0a*n$
zq5Mt%#Mcc93%d1+^%eNu3Wxyr8!l`57h`IT=k&B@UIUZ&%}^^6$BH~a?kU{kKCs4d
z>;;XkQ<nKB0sq}Q1XIvPU9e@r4nKHs$XlY0|CzEbhSZM~Wjd;Bjou1CYqL=bfwRTf
z8b>0#x7%1CQ%OY>mkvXxi;SfMF=`vb_Q*$)yfSTY6|@)6=i&#wcGSFPnPv;NP7h-H
za1RpD@YtX3m@TS$#Y-(HE||)_Wwb7D@<*CAQ$p4D1`7=6Uz?r3oD8wy``<B>yVdS)
z$=l>5on*<A99oGUXwWslgW(%nHOSHrom?a0COi1ou<quyip_DG6JL1_*DKPrLf;|*
z(GHPc%6Q)1?fUD>P2mQT`O&S%%v9C|g&&bZZZHVy`^8gR-GRnIAg2YOVvU|!?pf<2
z`g=vdQ-RC+ox&+SA>1wZYLn=jv?M}^gy?jWR(2IdH)s%~EnyzOF6<`K<cPXaA!o{R
zzW{qpyC?B$#DnOi4BAk0|7GP+>5r(`Xj0Y&hDY4WkxBE*Xu|PFu`XS0#oeRx|6O!V
zr?j9^aaUXr60<l%!}H+0J1SG$60(DA$w7C{<9#Y>EkckAZcWlH%x?|aM6l+8wc!a}
z5wf%37(fP0pd)p_cMGk2@jf{$;j{Ti)}ch+?nnw@SM@(^{ybJwicvF*9R{1_hfE!J
zHU-1sl%k0I-fSD=7fLx;&m_KPzE@ghK6#OSaOVQTxws{@Xkq@(9Qi^#@<JT=vVgf0
zITMK^3<;e(pSXj=j_F1PesgnfhYDUnL|VuY8!%(2_mmIO%Ct9N;cWex=2e0{UYl3A
z^9mj({Ub@t@WevY2?)hVCoMsceywLqUMqjPIr2;{?4R*@fL2qV6F2M4Oo^cPDQ$x>
zgwv}O@f=b*IT6$=$bA-!I2f>_lqapiN>W%ZXoBH7!`h|&7lVNgC=}bmdEdw6Mge47
z10LcEyLEpfO7Ev;Tr`T_PKAXR(QPIPn?<;aTy9}#%qB~|NjTbK;@&S5=d*vZt)}zq
z>-oJ`ijNQt&$v4s+4M0KPsZ_1INEW#bV1|?cY$9)DJ~BUZ;-lpgs%hp%n`dN6UI3s
zp5zvSw;-W!iG^hT7k_Obs?nB*Kx6}E77RiHwIy%m7I<p!K6%-c>xC!n55V>Ik5Yy#
z=jGVp!1FMlVbwND?~rVXVR|Rlo}gz=P&RqZwJ9phHrXIL*JpO}kach>FB&w8r8w+|
z+mRXikX>*O<*SETL9%KWCLmr^J3NM!%G`Bn@$t$zYnD`$KeDN>DPK^N3xNP;`c-%l
z7kvQ)|60Pmkk41feoal}t3M#t2>c5&^#%L$x1c}IA~TFKo-l7865!Xx=eke-SKG%_
zC-C#!-1+!>yx1Uo$goA1-;#1dd%lB74=b6q+s_>~bfuxGg8=(w13b~(o{Q-FFLcJ#
zCoN2WOtwg$AiP}p7aLoT1vWfD&jt26_mojCnwuyg>W%y2vb?Uy!!2=oz}JKP_WECM
zmg=wdjbHITnNPC(+?)A7+JZ69?P3jl`50YQEK(Dsh20bUF@o6?gtR=vjY9%J49YpY
zDsg>ZlCQDWI-<Rj>&lHRbe?0cA#`t`Ex+SjT9;t#G_FpH*{&cAesv-NXeLK^sq77y
zO{CYz6DC{jt_Rp%TiL>l?gN6-hs0evMat!YDOY0T!?^ECj@EsZN%0`mp^_b?R$W-5
z8hE?5{AN#tQ0tHBfT)Wnn(f94?B;yT`;LBU*IK40(M$HCIt~Gn?OtO1eqDrGSlV-s
z5I12xJ+%i<p>(Axe_wb3A^>x*3YnMU#ZHTMdKL9&X~?h=-2hfQ{)&DFYtwuWq$B8o
zXW?|eWn3E5Gg#e!c;LE~5dvD5NTK@v0iwaIC>D*jb$CT2bIW6K8Yy*o^p(Zx-)mwm
z^1XU%Hph)p)W>sC=GPB%HqM`GPja78>n7i=_O;YAmvt#sKRGf0v%@w%OTq2owNR^%
z9P=vAyu{1pzHSo!JtnHOUE^ERVRd|l?A1%{TcgJl_1}s1CruaR+H9XM>BglOUjxj%
zjo4GG+xFTl!x$w%ivC-zuTtNyf@PA~nk89M`V^SGc(MCOI^!WB+3a1OT>>Xf0}ghL
zBgMQW7d8yl-e?y93prYVB-em+!egnajDyD*U=H7A2rtJ1i|8N7kyz#VEp$)z*!I+x
z0?m)ql5P*qV)?j{9sb+bc3U}_E)_Ap@j|inx2B!!>Bc;syw*=+`(*7dj-!t<KQ9%H
z<VJ=2X2Kb)tm``qyo*iFe&Q)Y*z+}{BQsx2v{9pRnAs3O?E+<I0l!AWgqQV+6$pS%
zEGGqJ*~!{tjTiSxubClQJ-4*9)Lsjl-hzd{Uw%jp6uN9h51kZ9Q+EQ=>&8J-g(;U*
zR<F3?d|zI0D4b(PyH&_CTMte0*bao1X!=-mHb5L&(!zHn^e9RhuwH_h+|Oh&rneoN
z6bOp`P^k>?xKiAJo_KOBuzigcN@U~i#t+&i{`FrSk+&g3))5~*E$sl~%k}qE5%+?&
zY?H>oWU}k|ekI+gBi&Z-t<CIbOBolw_xk)s^8rI(>hH{scgsARYwy>W<0iLWutTZ)
zqV~T~n!0S;8fm|3*EI?v8~O?kN85Fp!}h5qHroNNSuquCf5UzADCjmpEAMPt!+Su`
zVeS)<bZLWspd_+B<UPQzLN~*|jL#QJ_xak&R#${0FrYlPp>qA1XQx*2cfH!Y)H%~L
zst039hmZ19Ds*^O;lm2__n+&j<*Le-;Y~+)sN~9Jq$$I87AHkP&K;E>xS4ys*=u#R
zCS?F+fB|hDtLkCE=E&39TuR?X3!0+8YHcfyV;2%|X{Q%T^Ibw^WfDaJH&L{sa>PY4
zXOQD88m%{q0<$6&^Q(-2Sa`0qf?Q)e5u)9&HD=1GTPPqHacdVw9W{$U@9ptZT1jpX
zV7cz?lAU^ISbS6QJK!pSbN$EoHS*ntZVvGGZHeHN`-*lj20>eqgX!iExA<riWJbr^
z+zqC}ZW^(zy2DTJL%YQncvc9$h5)=>NN@Gxz0i@R9(uC6gfEeKhl$D+lT-F@RT}o#
zLZN_3xxciyyI2`Ro}cwj_&b!#SwR@?gycy`4AC9Xq@o2!mcHC>LCS9H{R>o@pS*zA
zsO-eEK&!uiWcwrX=Dqd0)lx9pIw#?%+_~Rdwyi@Q=xqZs?LuOll-nsFg(LVGCyqld
z@i@cUr}dkiv6#hWe-uy3t`);8)}U^2r$WFKoDZMS`QDOekk`od<Ls*Oo>{ur5mjHC
zj6K#xtL0@a(x|xNFp|Zr+d9%{{k{NChxZ5GA^u1XC~^?#bEA4T1s&W$bf9JSDrsx+
zixuXH5<{0#*+2csT+IdSnR`HMb-L&_dWx~3Di`FK<78+dkCyt=-ThHEAbB_}>8i4f
zvTjfDJnGj;BTk+VhL($8B-)Nc9y{h>ffz~!6IU}%l5%&Rf1lnNXv1;Eu<`(w&stvn
zEIV=&f9b@o_>q}$SL3L|A~zz-CfHWeK5FhFXX=}w(9mg47CV%G?l(KnY;|veR&BhZ
z^AZ`>#C1xSJ7bR#(LYJ#jvUt6TeuXC>VqgpXXsdLGuiV(1LS*d5jOdjP||2)#L=RP
z?g3jJqo&n^<Eq7Cd>f*y?l#~o<Gi2AZn^!s4&%Uq2n2%TQw@P>C#$y;NC1?Q0)D7G
zCZ9cUGkkDU(z4$>N!IJ^N%_RGQ3{a;uS#TMCD|=2)F9pE<gQ*-MT$XBGP+!%BO!{?
z{=O=Pq^~%QdMI^wwSpc&ZH=c|u}}F_ihG>ralHHGWT<$OHzXY5YaYP)6%H$`Zh^D3
z4SdgrBO?*EVv@osd4;>cMDt@PpSwm{M20dF@IZt7Fok#o#agmrX-gaCnoh$S*42KH
z%PM{1tXFbCZtLbLTR6z<CXQ<_Tzs|^J)_R2+by}LYH7RYM70*V`Ug2H!npGcTb3fI
zN=<mufND_M9~(Vv?niT@rCdrsy_RBH)Xlj#0O!6$a5?F=NjiYkbYn>qh7?<-Pm~Jj
zsoh8wX1ETFd4~@iRgvL}^e}c^r_F<UAT?I>A;sn+jd>=13h`fewi)c78t$IYP|=K3
z=pE2gB99Iw{))#70ww3LTSw<987k7Yi)1qyYY-ZVmo&Cm?-IaFV>{Ggk|t0x^OIt7
z^R`>Dye`_x!PfiBd<ayo)x|<7d6T6$PZ-1W0bSliYgq$=3k35aJ>#K>HCDo$DSP0W
z9uFO#Sk5xblBo5Oo$GK_O*NAp{zw1QSAHX)_=DE}ao3MRB<;tv0FS$}#PA0niEcF)
zK8bfS2Dv3#S_0_qvYvVq^*Xxv_W9J8GBZS-{Q1`Eu-0c4Acq535ac(8h+#)uAvECS
za5wU?alChtWC}Hj>O1aVqdHZLSq&2~#6x)Fh*m8au|C0VO*T<b{x4!M2DRPWMo_-Q
zbS?4x>$B$@t^#(h!+Wfx=;1vmsb?(rv}ftsLl>aoRi^dELtlNcXcDJ=`?kr4nsVFh
zfmC*He?Ewb+}!Z^;B&v@6y@x@Fq2IZ52-jNo>;v{*tb|z3-@Jj0lGAm6iw~MhG~x(
z6Jk9AQmVRfXw?KZwPJ6+&2(bz>2UZ8vwFB84(zT|c|g8L0w?oGth0Y}!QXXU2koJa
zrUvMj-I_S@vuC2UQ{o+B0&PtB_eaQ8lfw5JIx>q2$j()))(+O365`w|^ozIVW>wu%
zpMU+#&$jsW!it|>Q<;%g1pog6l|X90c}8Un3yxk!o=-FVN2i4#=R>+aGX5)8G+gPW
z<GgEJB?u2QMn8v`0N^MfD7K83H2ZL7HRb?VynQM>e}^h21a3SfP!aZeP-fW`da!4c
z7G>J7N#P^d-X}~RROg|^#V9Dz@L$FxF3CEyPx~!4C|+wl!Z`d@LlnC1+O{qX^<42*
zz*dL5?;3YsByXC@%Vh&+^rDoDTDR-aE)=(i^vbXV`H^|f+~!FOa=?B*6{z(fUt-#s
zNv*Eef6MaX@3FDzwc}#(NU7L#rcV|>Z%_CyHt5_>({R056MFf)+&EclTg`QB(Ot(L
z4*W7lqpM#l&HCq*q)@)xmZQ`CktdvyeiUzSbE5-Lh{a-~9F*j-67m@G*czzxI(C|R
zuLP_{{vh0op5?n+4!1X@P@MbO&Sp4vqX{Pwe<B=|j*3<FLu50Y<6`ELDC%Sw=Viv6
zhWpkQ_pK41#I1Ivdf2E$L+%p-ws`)RGgF8plGT%jq&$gm^?0}Bn<d{Sf8|NzTa%p$
z`-YWg`)9%F%Px#)9D4n-!z#&9sE!VyDg|r^xgRet@kpcR%&pX=GGJO{Knb$~LL+6+
ze;6a6goY(N^x~m+3lm1e67m@G*y2<|!%`}@x4F*7<^6+`gJ?_YRGf|F%oHL?GS9}-
z)*j7Z$m+=Isbclln!&UuG=nX>?66956sn^mYV#`j>YZ%h^zuH#(_P20`vq<c)1Fkw
zctwP45+N@Pr6p-74dWtgHDwoe9w(_Te?v2dL)p*mZ?Uz{j*4jf5S2E@MK$8K|5Ek%
zT&X@|lXVzrGniv0s*^9FWI5wXS<-Yewuv$|EZ}L9ypaMN1tzrv46Y^>Os2bCQHvpp
z>R+`kJ$I;f&)*8uhn)+eNG#2dL?uC2N&yQY->3j2pF84+adJ|sA5<cMD(xQ?e~W9X
zR3~P&XmKK7bm;EoR@{cx_G~A%pp=}Pdh=MI(MiGx729|Q>9RGtfI%#2A}yWOLHFi1
zADn<C`FL~@%?E12R22EqOfx&Gl$vEagrx`UOy##vd2+%wWpN`;u$0QjC$$rmCa6LE
zqCRzucG5an22DGl6(wukw<!M>e;Y7BJT{u5vVU}Zv{nV+^+OK2j1sp)MyjzymOHY6
zgm#s+%wo0MW<c)bMwN|291YbG+NS2ZBm?cCseY#$Zrh<A&5(?IcMjwILFK4gFHhQN
z!*^0%lS(0Ai=eKy=7oKFu@MDvE*N#DHW}>~&71mFv+|i#FKhMq#+PE2e+tp6esy(u
zMJ}2olP8^_)rMN(yee{@Wd632PqDMcx9f|`Cb?{k%c_S=;H1*51r3p;QmtNI)h?PJ
z$<1Z36np3`Ufm_~Cck!ZO`4ZC3<+0P_3BMQK;l#78h-4&PQF~$E*cm0T6AF*TCg=Z
z)(;!ydVLfwgo1ceq;x%de~|U1@88zn)@kH_a4q9U;WB~EnSSbHMhD{ZvVKL>N6oPy
zEQ8(g{|1vIm0IoMhH>2Amh93nei2?^yfU=d#?bZtAVv1_HFW39z|VY;e}(_-3{A9p
zSUFcrd1ePkvq1a;n{$jl%skanvB6IHYnjq&6}!_@VpxPjD;FuCf2HX<ap?nfG?VnH
z>uOyJT`jyvN-g%@%H;|M>?p(n3zxk*C^z1ZV@U(#4d2-<Lz$Sm>f#{2cTya=)tT8D
zl*JY*oII}tPN|TQlFdB9Cgm5=7-i%7A0ZYo?5;!bBlr^nei2hSW&uCSGH`_lS^A(K
zMOX3%*cE8+v2chbf1*DD=s$b2Yd8-^5euAaPfOt(HP{I_S2q?_(tdgf@lzZ}tP}i;
zO)MX<W621q@rA@jvf`$b;!P*LD%Z540Fgt;_=1!^;|p(}0RH5Sw7$4Qhg)BI+5_di
zLcK?u*sXulHk+kk@k=h->_iVyUe5e!u8yyYFXXH?unE#me?E}5W<jNRu%L|tQ8c$;
zSRU<$sm~=iLqOneM7-z59@j_1>-|4_U&5R?lCAq!vZ;zYaEqeYy-e5<ETW2dHm+yv
z=~q$R*C8MQ-4YU461JIs{oik9N&-n}Lu{EMJ7!#10<4qgJ4>EC3xJxYYIy>Lp{gOw
z%~b{cks%%ve{2Tl-hqhp8u_r@wBCkIw3_v3)F_|FN6k7o!t<YwYDhQBr<~CbQ}eh^
z?Y+`Q4=vl4do#ci))w3LF8Pul;we+&svq^4_j~rWPlt;ha-$q6O1Z?i*Tf@5?0QWR
zgSad0%4PG)!`dsft!kdULy>zd98^AkZN^i6zX3;vfBe?OQylC|EhEDVIxW)Zry?}g
znnFp7=ugFlbcHz7T)y^1Mbc`*^a|8%4*Z8WoBz|{4021wCFxjHkc|MhR_sX_w^sC8
zt8@!haG5AenlE6>XxifR@{W5MTmv4TiqJy<?qzT<gUc9~u?VMwx#kp}RtaBDaiBs)
zQR+5se{=*A@FH3dd2IIYkqV58Z5s|yit;4g1?YMDgslW<)i{oXOkC$5=^nEwIsrSk
z_5<l+@U8JL0X%&ZK)-P>fH-kuW(Y?vrFm|TnM*}p%~2TJdboSO@1cK?Qt3VMRu2cY
zst0`F?e`wM6U1>=ns<WWogfZ@Er(w5T@;E$e_;Dg5Ib?QT)*U;JO(E{+N)NqU$i?k
zJubE_F#-kFgZqHbeTRV0K_yosNA_8mt%I=Q5JA$z))~<5GL&8gP7(8BSl!EZy<PXW
zx<aK>Cz^~i?c~~HvnDUh47Z9^w2z=kjf_M?EQ~dA2HgkaTVm28%!x_s%|V!C*cdZ}
ze@V|_Eaxh_T>SxVc&+9^p7Q4^)^${2;f-?-CoFcf6Ss&Os<Nu(cGHCJJ1J+bO~0vC
z9CT75uhN)eWrFockebHJS+3VFYnL=FXLclw$Ju2;mS#CK#9yR$eUs#z(dTt%mK(#a
z=|XWHphuJx&!Hpj>lzCjT=R3ye{{`Hf9Zny3yjMkP`u#SRxxF20|L+#n_JTmh=J)u
z)-m{Ht<f&iq$pm-k|N3!uPaWVz!sJ-?7kN^MxGRpc+ti+Pg8z}AdusYM&+_>9&fPj
zXj-KBK+k<A<<PU~r!isEiA(P=bieVHzI=8c(%I_VrCprNzo=bYRo%>|`b=++e|<JB
z(ka0xQAT09$iqVyV<}ncU{CV{z|I`*e`cr%u2V_>1+_f6Xc~6=)<`v?F&?%QJ{b40
zt&Vj|HO8sK$L$Aa?^x^P+Hn@_Izq%JPp=T`zwBz(YaYl@nHW=aOVQLf=Lf>GfG-8K
z6b%n(tcw(_51t?yt@;^YbPRXce{Fn{$nm+@D%#j9i@Na$l@Zk7=({+9O$~~_$fpuD
z<qrb_MUWm&;qX`Cho!Ydy&$U-(%Qk<{Q5@@S-YXCyh6<YIJ<pmOcD5S7W@zN!%)qf
ze~`8Ni2lH~Z@iFYppL=6p{DM?;D)t)X-2xKC$)e`aEKKI1Tx6|F`&+Qf1Ylh6Ml~*
zeJ+bj&g*pZdJeU*s?FdC7;n-LUW2iu=?3ZXy#(k}3#gOx?wca%9IZI6%J>vKzv?lq
zD)JJDKu0{1q4_2t%Ch=~z8int^uHJAi$!G&Uh%7~fpxAmHIzIN=2k?T=vKW{ZBZ48
zC(h0cib~L9Q<@YQ5SqBLe~U0<6a{tKwc-^4y;pkNcZl>@mx82+$L|luQ>#E3DhAXF
z8XU(-kf-jzi{>nynKT;7q1k4%ujj$Vn-`oOjFW%IEaBwZnmh$Poa5U1B+qs)G<sZD
zr&Cu`#IzW-r54icOQwMl%Jo`f-6qk&N%tM|;BJl8REISz(kXe&e?Z+G9@Y<IH$9Z$
z9r#AsP8eDBM!oD%J+zV)o=w08YvOFmw^S;A*#O~6B{h>}xl%bEa!s#RR^J>hty$Va
zftlop=7>&+=&3ZBx*_c?C_4?Lx#!0rz5SLYW+aKb7#zI}xfo_jEq63Hu2be{&;xAl
zZg604?`}i`w_L7Wf0Qe`0Jl^wSA{c3{7MCI!`Mv`aBWXx5=M*XM21#ch+$!$C!sxM
zrLj+Mo!nnBhl^Pngmy;$TumZa9UX@j^RmrXoGuEAJ{wjxndgAwdGV1X_{MMq!uV)k
z<c>gU905kecI61*utRKP<0s*^BoKub`Cd!=MT?DZ;lx&Ae`B5Qa<O4SY#8yMUG;%K
z!w6!;iC*i0ba>OZrat=YEKU#QQ?pbPYTl_#Ec1zxq!!yc*wMi>Krh;=g)zsz$gNsx
ztQsS-=(cJu*D&?!6w#O>ZEBq%#?VBSdbxeE)~ODq<Gw?G+VxIKhaw(+AafK)5)w6K
z=&s3%!$A|8f9YZCV1Ks6BAt@Q^k`hK_Mzx0KZa_?i&~>`psp^-9`^^mOPOeJ9A4#C
zrB>hQkj50AeW;<GFCTw^PZ5)H=wpF1boZa8zlLoI_yS5rcYo;J{Q(E{+#ficKY-f{
zZRZpCy5F9Q4%WKwq}<i>T@<SLJybOf4h=AaBS}<%f2sl8oddb{@9u^$RN*whwBf!v
zVvIgE=>g~VYWvW7Cz&CUh#65Q;I|u)QPfPF(ygty=ny^nQcY9B#mKpifk$~!gN!Tn
zdaFuR8wqRlo(b9;1X}cOIg<zFJaT<<|FzMTiK0hOm_0!&Ct@_b<ocdA48}N&=RF#H
z#E+?Af2mGx9x{BL?7`#cK6)NEQ835EOsP>RH(N9<z3Hd70$Ueyte*huw1XxOM$$ww
zzzz1XMw&V=8k#7{NS`w&wX7}u4PW5{uP~m`khkI&2;rG=oPl7w2i_uf#XRtKkjD&!
z?Pg*J?N5xO#o@9>U(E9yBc8iwHlIsN_*CS8e}%D|B1x3SSy5SL-Fc=t98SabFis4(
zR;*vNJG<c8u__$bwBtIA-4x;41T-1iv6;!}MteR++T8B*Jg5fwV1FQ|FJvI<x-?O;
zb5S>7(b9C2NkI{3(B6}yUx(^Kv9aR0_z9G9c`%vGWz{HOUx!LZsf^@<Ch8|poxQ{m
ze`tyx+0PB69c1o6S{y^^P#5kyanuDrEZqmw@Z3tqfEF|aCDdknyfKDjyG(H<&RB1*
z6v$Ihgu<CPpQGhNBDJ#kBGg~ZsjjS@m-qwo-A7;0{>+*8j7OeWUHxT*wsw^IQ6dtV
z9~6(rh^2e!$I7iN4xkLubIwC2l)Y!LfAlW*Z7%P6IBYHPjpN2881^ZA7ln3l(xrbx
zWFLqtj9<s`Avp+A1_Cm{1fM|<WO4Y*>-YF#pjE0>923}`&8%H__&(yKoE6C&;G=rq
zgO<KW^@Q~jAtJG2D%DH%mT$$`87n4hQD}#+Ij%Z<`_gRG8<%Z=rqc&=6VaHge~*N?
zAj+6xQ!LqR6r1gOeA#rpoyewjiOVJ(OeX{6(<+;UxUCHi!zZ$POgo@2)R7AuZFl%5
z-e>Apc&0=}C|0U%im15#z46%V_P#Q<__x<UQGo*C2&iMAY9pv+b2#k|kdT=U25=}Y
zbO;9OyLo){@g3;U0LB!WZ;w_Sf9+beRBLz?WAGN*{dS(;L4~y-tS4z5x|PQ<%$nD{
zfT!@`$z$h1hM2rCt<tKOs;z?#drnHiT-6y^=#7&MR^T0@8iA)6$JJsy#rSyU`zT$%
zIdC~}_YK^Q+;~}i+bT7_T`yIttzBSkeM1ne`FK8>U>(M8ieMcxH6DRIe@<pR$2iA0
zeHcGM&Ohx-5?KhqjQV;vVr1JNS}#?qMKZ9G>m{z2cBBlKA%Gi817Twv=lqW2o8y~4
z<s6<okMqt+9yd;b&ip_xtfbeC);Mu;AGri@31Eb>Bw3vt-W=YHP?jKH_0SPtj&7>Z
z%|^sGJn7NuH~b1ULmG;*e;o+UgJPjZet=Kw`3pHN1ZfrmS=5b3sEnWnN8iN>Y-&*a
zMLw100}(sLFiaAhLO?DY<=mRbkhw893T3%mysSUkML8lBkRhdXs_a0npInGZ(^f2m
z-3Ot}P6<mU0Ks{WbR-q-M4V4jA;YXHC~zVg5`rMl=HpuDtg`vGf2v167vR!YYhg|w
zB*{T&2RE0+<>~odx+t3U>(XS3e*JvPn8qOXs^TDlGO@2g<@49($uY6Frze^K2LWPa
zLHfd@*{NO%aVzzxRtM&qZj;rY6dBf<u|Wk!g>|aCN;I_6v5whHYR?pPdr!h+uW<)j
z4{ZSmkB<+JpxXnPe>@m3l%XN1ij`xQ`9ldZ_KY%<P~7dc9(w&SH%6$|TUBbRiQ}-e
z|0MKUKR}ILR3_d{1Z&}1ry>ZeFxGi%;R8PKw(gm3sB=536{(m%lpD`~_*!N)Y6_>4
zg}u!R<V$)%S;%tsP1KYe`UJXT{|%(v8tr!Xsr%6TL$QK}f9nt857=3>6XTztgg)HQ
z;BUZRb8Y<<^0mP=-xynsR;|&bszyBzhN3JNzzyg$?)pNUr+3R%t6b_-YKI|dkDF5P
z0UzH;3LwzR0Ye1?2*iO5@k>00T!*hv6J;=ix-L#2j;I=M5Gv?%r+5C&Yj1FPr(s`v
zgM`Q^HOno*e>c3fJu*s_DpLk2lKa-;Q1-Z--OiQUV{qu8?A|Giog_%qYE^53x8KxN
zu_81H+r0dDzx^!vZ!Hd${FmVbLtrK<$Rn6RB}Q+iv;!cq0zC}?5<f2HvowxqgiqI_
z$8cCoKo|6>`iA{s;`6gaqJx%z#Ff90r$58_Tgbhie?kqe5+m%312|UE<shnw1lBGt
zD;NGUOd(7WMTN1Xh;jkfqNIwQeq+Z^qjs$g7`MA8Lc+dFohJ)v1oiA#l5t^Hm^Nrt
zO7#wvjUzLPs=$a&PmHlSc%TI)=CWCBmPzb`T?*|gk~BgtEj)@EtHPS@l;=}idd;=m
z@t;1be+ewL+uy)AI`TA!hV&{LkR%sNieyNlZ0(g@#A4qAe;3~%r_)o56A<MAF{;?T
zhk$5;pqOt|JQxJ$7Zs1cYWysLuNq%93e+^s=T=Lg4@DU|bTr5gMMY7a5zrC(4F^LU
z@7g}W>fA9wCHt{kjy!e*0o<fW)aP($07(Z!e|0vOr6Ka%M?gf3z<7`pV<F*ModW1X
z=<Wc>0l*>vf~s=h`&<^$<_)LsUPlBt>ZOZX{Xj!Kc0>2nFy{L8moF2^m@Wnd1nDn^
zG+MqwO^08s=<XLu*B4O#Qn{*LazLjR(4@3k9RN|AEQpO$$EMwTS-uR>lS|4%UFg|d
ze+1S$tvcc=)mS%SU04CVplXvZ8XQ9nDnt0iBzjG9VQS1|4*b*tUxyHgvaUt|TevKi
z%SA?%xTYJ6k)(bZPN8V%QwR<H%Rp8KUuO8kPy2$qhkkyQ^aXNi{!MJy=W?M`n2+4E
zt>XaQGI7;%sU3GfCrAGYP=bMul-fwtf0n?IhQHh;^6@!ct6r2FH9t<f6Fpetw0V+f
z{nbtfQD3x|l_ChZ9(-E}#7k|7l5zS73(a;}=O~aDFN1|7kFpSbQ+6&Q>z~EF!6T0I
ztmUAPr6hwRFt3h1R$2u4C#ICH-B~EPJlcfIvN+dqqtp!e1UyA;v6h0`;az#~e-9=>
z86b{Za|W52(nW_x&U^yWc%ze@XSk1t!gjS*tra~KR<A1MB9(LbR$9|%3kcumINo&w
zhS_2s6p<9l(o8~6eFJn=GqTG9`W;=xyU81K52;^Yn;|s#g7Xiymr{DMV0JWEtruyc
z^fH#Ed3K4G0WE;J9PYziRz&$We~t!nbBbQm5?)P8FAoyfG<tauMoYC?t+dSD0`nH_
zptKw!3_{ZTyI`b}gACFe5r+idKa1~@4zR*snlDxKFp5Ef2I2Xo))Kg(X>A{mur7Pq
z&|&JjfV8k(Es3(24bY!e$6zK(in*C43xsd_ByM?@#j}BCc3BDtE$uGoe?2^b=U!WQ
z>_0VnZSeF>0Ii4nd!hFPz6p;4XbP>yb6)_z3!o<mZFKYDvHjAylfJta=}XFEs)^MP
zUKs0X+>oJH$y<#iuqWgd1p1yfeyNOCy_w-I3fF&|5+-_4^idX<>caRq7ZoVK%tcL{
z=|6v#{){nSl830NQYMFRe^-v9SayoIFAHYGy?{N0!d%rQ>@x+nre%^5*MO*boCpDR
za%Aj2WB^%Jk(WRkGsiGS*>qb_9B25*T~|~E^q%kTkl&8ku*7qEU2hmkV{-IufU%~|
zKt@p$w+2g*igj@!<^Gy9#xXkti{EZc_6J5Xnbc9F%EmzZGR?VRe<R)g)kzuPof~>l
zDpCOY6W2wb6X>nUsU)-4boR5NUa)|Yok@xr$Ib&qEND26OJe*v@EMeH*RHpU?*VVc
zMbcc8o)f5cYj0dDHcIOwBCbe$7KcKf6GdXF5M)X1zL=t6XwqPT++yn(+ngh_{5Zg4
zzw}%#_m|&w6G00_f0lG3`a8Dop?2=5q7Xm`G$H~F296lfB8hTy>V0?$pA*Yn&>G_v
zZWQYmb?<PPOX1?Oc&M|%V?;-MD5Q^=TPy`R@Fp2kbzxY})MwHoSPk5cr-f&$&RR^e
z&YvbzqXkzCu*bXGb=_aDv9;}WC7jt8`4o#d9zRt{deN-`fBoelXtafUKm?^mG%3;H
z^DZ8GcR!FQ;rrp959R*9$1@M7g*7zqFe@T(OUq|P9A!_kEtvVNh-0lZpA~V4SrI(E
zLDiFC?j;`nn4WVL#_~0GV7H9a`Xf|EP=llINSK@09U`9!^@*LwFvv-KA-`uG`SW>z
z)PdKWEX~Xce`SD65TG(Kre>J*C-@DRdr$If?oV`o{obewc6yKY@j0ed&ju=mN};&2
z!>03Dn2G<bYA-tat3%wgWG<t{9)ZEq{k8xfl|F*~5mNx&F&K+OBdYDjL2E{KV&wj?
z%qkOI=wy_Q{**=VmVSoaR#_FRV97X**TJ(y&&tV6e?guFg`A}&DRU<x3h4R*?QBgC
z7pG%#ymGu!jMur!!Mbm-vcmj`bl}tQpS9Rr;*(CQV6oFlrz4(&b7i#`l6Y-Jby(Gm
zBpSui!zX279w)`K%_wz}E=GB}CvZi~y+)dd2ng4{I=4P46G&p`KAIGA+2pdxnryDO
zd8rgke@AUHmrlmQ!MmQ6`ZeeSAME}{8l0)Gc-$Hoin=*^ZLpZ&a_F)24r2^n<7lA@
zE&JDv`yP=?633Ay%1)q)H^3PkPH+i+CJ&ZA+bsH)Ddlz$u>lWa<D58Vj5!b9(TY2>
zQ{D?2(s=M^=}&aeWQhN5Bwx65g{v^qx|;)Ef52stiX&jI#gAz7&DvRH<gtZ+Q?*}s
zV3So>6(l8?rbew+ZZkAB_Jk6@6Nl6>akMOxh#jueOX#V~3eHGEgNjv;=xJeVa~%7=
zlc1!b5mlI{Wf^wPiQ^DqxwjryYU%K_EZ5inwx6eE-Q}LvnR;5B;jVQ|cYx%U$UHH~
ze|ETdm~~vsgUoSM`A$U8*N$^1a(I)*^O8Zbxy51atW^<$D+^c?J646wtQ-3eaW?;_
z)j{=!=f<6<>w_flI8w*=Ex462Df=(tqdQ3vhlcb@=B?Y!l0;d9;%EtmQ#kxZu8&pC
z9qaQI$ws->sg=DVd0DTnTT(bh(*1VUf1Fd-#E)XJS*&d_@gs2_tGSqrlZSDH1l}sh
zjF2NXD0Y(OD85bwZE;d4$p!|L;~4H*r&4MV2RD|MaKD{s3F{L1=1naT9D8N~KqnOa
zeF;EBo32v4+$i?sewfXyDrIpu9rxB4U5l}|<ie41;aJN-*^&cYoX_QD`ph8df3>_4
z2wX+(7KdCObe?<Q#%fySnL2!y!or2bB_ukWyJci%{*%F75AlEiWjGbJv^tR3RGyl&
zy^%ylJaBd~!!Ga%RBZjM!Fe9^yZ7Jkx}6`Z&U#_mJC5?+Jy>8}BHz4?@;+j)fGczp
zP}_CxHJjT82@|3GqS&bgSbk=ge<P@yFCPBdd%kZ9kKmyLz6w3zvGF9dPriZLo3?rC
zw*F85?AQ%D2;`WzKR!49U$+m?|9THkp!M+hD72ogF&0ACq*iVS2OX5UznwsMpf3d+
zjuZ1^jA@~*b6IB>(W}q>Ztp9Wz`g()k6)k3bnU+I^n@;lt0QK(7@&Ewf4v9L?)Jam
zHGY75q0#Ga_M`ETT<(<0Oq2Yt#@(~PMUzqXU6YRFVPuEJf(}Ub-Zzu3d~Lj^w3D@U
zybD_g58WFsv^?q8*Wd4iMqkK-C!zlY9+2LCG^ILFS2IT@-!~sZtne~gkI5`DSe_se
zcP7K(nPMi)9N5!1>|_R>e`6DuQ#{9Jr@6UXrBX6vQVt3o#u~ctHiIc;VJg}T=7Bj<
z1Gvqi{e!9}rmv=a|I|2QoaW#d`q`42I*D{CtvD?=J&tZ_eA<LNR|;Xt%T?3LS>895
zNv+EJrgGV2SvGm!R4$#{NGBs`KU~wx)5D0l&>p)m@>iOh-MA8hf1+_&s$FnU_=Cb)
z99BSqn^J8wW=V_;t}fD}E_gTI)AJyqFmkzb^1Re6w}dNZcz(PVnL7!cZ-4)A;iHxj
zc6o`X!!-k3@Hw|<blKPKE?HQ1Mpr!O2$)-#v$8PEcy?!CWW3FN8Q{vsq=z(GA@GTQ
zP6VWqgg1-c{w$UUe~;M!@<zHe_0JdweXdpKkYo#z=i9iNUY;awT)j^jOkLw@dik9+
z`6@;1wpMPB9g+{q?!=E9=o5Ybc+VH4u`ogPR;gV+R4vOc1jSOn@lkzMtO(6@eJe*b
z3#gVw-FQR@h8i4w7bmcZv;E1Z0`WdpHSe%9@Fp2kvweIXf8CRGFjA3UVh^jk1`z%6
z>|E4zh_+Af2Dq_{7@a`~Qkl5XiXQ!2D&n%nRq6%Fet&s?LC`fMWuY$gkh*jiHYt=r
zFIh_`S-HgXhf*Yru0`I9Fdh^Qi60m9dEZ$|o0v+Pue47<rEZ%@Euj2$xLwj^ceOC9
zi8wIM9vCx8e^IrU6+HHrzQjSoq&bqs0rJv_x0dyKWVyeH-q5eF4NX+^7vvVsL?fG#
zMiAK=V~Kn+`wi&&)V^OT{#W5Wms`zza(!(%#<vDJdUz!dI_`M?;&w`>bthc-F}57U
zS29a#<&nIJ67u=8<{KkUw&+@bxHSeo%~ZR+C*iTzf4Bp!hqeIB)~c>fqhX*Hs`XZt
zrAiA&+f+htx7T{;^}GEiq1XBW>gnRSzhq}};Ga>g(uBh{IN4gJ8SZWrY7<(uMzcxN
zO2_TRMB+xZQm56S(FXh-Zi&rRcguFGT<TP6haqN<A>9Lak4n3Py#L@Sy^aukpNk~r
zYnUzOe?c)9C=-;2Ap-?LgERFN0&)w8V*~v~>*u?q16i6$1`X#0d{21Vos0<tBnP^x
z;ekWw_Z+)1`46DiK>B|qAuk8%{3-J1Edpl?-2g*XLxj&&1VCgDMxr6ww+9OYEYWLZ
zH0=4E3muNQfSkR6sOkVqhqZ4;mq}n=Q57C|e?2pofyA}e2cRUS)Kr+}{7&RYWHjtw
z>JV>tR)MdGmaXi$7Vv)-68Z=t53A9QBP(kSkSfJmb%?Yn(ZvwVH5IwyGoX&a^0VoN
z;4mc^Nq9mT6w&%3U&TI=mS$iC@uzhJ>;ib2F7!MYspM=L>E;4>)}gJl(YL5?06TQ1
ze~*JtvMCTwe*Sti;pFG-=f&^nG8k@;dCJ9BDHkLO`(-F<2Gk`{$zt0=&Qh_7Pn!`B
zA_o&wYSi0}14a7|K}@Px<8YEz$h5p1<S>P??W$QVcZw7edW$yVjPmH#0FnVTIMT#7
zCPn5pq9%hi#ihVEyRr3$2*E!dPh6u2e>xrHHzXWX5^jR*t<MPUcZMQ*K{9whlT)^B
zH&p3DMVM%Inim(9^>`1HK%rA<UYa8pSiwdYv>(Wf%jaHOc<etldTsFZts6#q^QzS9
z9HoF^Z&Zek4onFv13-K)^!tsk0{Hz~_f7y^@Z1+b>*3+C{nBbYJ_-G9qZhU~f5%&g
zv-a9<KY*bu&2iU2bv(}C2-sA_2W=$bQq-1!%=!T8!a&qOfL%D`Kujc6ao1M3=>Njw
z73tor?let^2Pl~2`C$SSW2Al39h=ybf(;IGnCnun)jG%7FCW-(*v}D5m~X+}Qr#9n
zw=LX*M)Toc?#abm?IuyX+(AjZe}lsPQO10e9d|Lf`@zyoYymS-5hoD0KnK|&X^@W-
zX9sYsY7iM!!!)w%ck#p4-E{+BHNI+GG5IJa8BVZnT5B7wY<3o#;{qV@<6=Hb<1F|M
z*yTqi7z%IF$e8AMhR0b!As}PjM8sH9MA<QCc`|vKK;y-{n>Ax5FN1|7f8(AtF4#C0
zfUvuEaap<W2MmScft)gv_5!9bh%hEOONweqcnq=3s;{+6Qe|mSn2W|Vax61lG{-2q
zwJAGg6mS2$1RnqrN6VH+Ywb#<TB!#Jp?%SzwcT&$VQ>2Ju<uy-WFd{9o*hfF9HHM^
zl~TP!<z%}It|Hm!_R2JyfAhc$;BuBw#1c`SPesm>lRSO;4(E{Xi5#<6`@~67Btt@;
zcw9^0DqqwZjo6mn{dR8YsRB3Js(be7WM`7IT7i<Fam*Pd?8yOqg$j_yOyx1f*p7(G
zGP_V@NjKaLC)wec9U&M=`aEUK$&n%eEKjFbwCS5!#eN_x2=17=e`cw8S?+l7U|pn-
z<lFU5kG#m{KCP}Gw_^rhUR7z7O<m9gZF-F*m}ugBid-wSSr$h)nTHndiVC4;MA)}Q
zyt8o{k!Ht*PoTt^pJL|w>jN_chGF1oOPRGg=$YuqXL8_9_$u_;;8FO$&qDu6XoGC;
zp$B@;cXt5y!kK`^f8*EZd!hFP9v)K{K{pMdK2_h=WcGz8@bs|hE3t}&+#gEw`3ZC}
zWx{<QJbh~*0>261``yD6_}1uy-h+(-vm{Ko|NivDzSRHTXbGV8C^RrQnnJ6A%S#?Q
z<SSNCVbq%12q34#D+n;g_S-*{IAf@|l4sk(yEVCgv(>fUe_u6SHGd&}TUV`I{d%aL
z1T$3RM5GimJTEu7X*eO9lUcHmp-vnK%*EDs3SycG+B_F>|Js1B8ecW8aC{UFZgS%-
z1mwb!@-y7Aq>xNfYhR)-<~Zh{-&+mxv-XAG(PjhjLeFkagd|&(fYh$iwL6b1M>2$D
zajxSe&kXnkfAFltIGdPerCe&P&o3PKa^E>id|69CA-*P>Iw<JkD|B(#{z28V1{}(m
zf`Q{>Tr-}NW|YR+kPMd7u%9P49nl{eBSWQPj|j+;0(D2`+qwgrI$rf$kQDPwMuS83
z72syn)|46Z%Gslb5!56_i<RM?fy(O(DM*6Giv@s8e;)TOETB7CD*9A&92@}$E=PD3
zvtK#J2n-U=llOrOhse1>WQey$PK=NX93&hh*^D?EA!MMR91h$$UBkn8Q;}2gjt5*R
z@EZB5c<KoRGZmi_<Uto$eO%^v91EV*f^K5>6uYx?O;U{Hk-k#1+!C%hX4t}ve{|!6
zbI;<af2Ln+7BjHt{J9b2bKz)XMJ%|mvLbOJmU!6hDrp0$=3MSC93L{9e4^j~M(_BA
zagIJ<2_4emQz(ug*571$Y~B>Zf}*akAjKY8;C{&EOs0%{%n_rAxXl5)?biU22dH6X
zY4J}t*@d_MOREeyn_G1$NIGqbtI`lv%9!H#f8dVy1ZdNL5EzA@qU>;%60MogfB#r(
zr*(vaYbO@9(`u0Qj@qfws&#~lM>}11>a{D94eHfS&06WIveH{;shzCFp${fO+9?S2
zQS7Hs5X>H7Sq&Ns4bN^T2Q*gIv3ao&Ms2-SDPMZ?Vd<rI655-YG1j?`)Pk$R31t`6
ze|G1B70Loj`s|}36PQRX<0U}fGlk0eft4winqc*-xb~369w<rGu)JK4lrn3EbO=N3
zgvTQ=`B%Fv*D6#Fo-@UZNzUCCOg9dUpgz>3IrigRgkZ-6zcYH18SKU}x5;#^EB$U<
z;>Wd|oc(q|4Z%UDR%;Q)#>7Dynm^FTe{z3qy~?1A(c~eF?-W8jI@>geSlTI@Xwpb{
zH#G738ow-vnubIrF!I14@3MC0aIEZIZkW<zn0A_F#*Gl=M#yV#&dS^bsO^=m%UczB
zX4j=WP==~}Q_ch8ndvvfY-#ll%UR>f@ywKCG`o4M-LrlXJzh?O0v>|j73aQcf2~t0
z)vIeV&+5cGDbB4mV}ptW@TdLUHQH~W;7~uYqp{B%l+=}{9Tb#_A9v?n7|%O870-84
zM3mH|r3XD{f*QZXv>ghXIsX~s|3;2CXFgt4Mwp+7={t85=FDf8`CBv&ChcQ;=>Pq1
z55M<7qrdJU9H$*AM+w_1;uP&jfB7zyKI|MY+?{rIpVtNyx1-#xQ5=`_HS77R@l|7K
z)taSx=W2C&b0*4}vL`e3u(~kji=^#<x%Fk#`lc@Y%h7X0^jKwcuI-&wHRrr$g2Q{q
zg`8aSEOdVK#3*_;!v*dr0791xm5DKRg9w)u6~X@p);%|jtlPQ2tZ%gbe{n(UsEf46
zbf3(<(C;_C3VqOde7Ltq*MrV;ul3Y@==HAwSu_%HL1KA2nh%ib<fBgVV4+C=vw$81
zu-ZK)Y*a@klgp<9!tBFyg+wpv0>Fq_=SsjK$_?KzU;NGEn`dVBNS@7>u2Y2X#9b%u
zI>mOK&eu0$)#O;F#pU~Of0M%1!S*J_V>dz!4mT;>qy(50zT-}lRq<|ExMAT+VQpAA
z!(Ho0Kf@D0E{Sc7O&@?%A9>o~{pNBLk~jzs6kK#_t!9}aUtkyRQ^;slDZqEh2(;2s
zf=wYw7?lVaw*Qf#khHz{B*YUzZin5A)hheTTw8ux%2?`lIXv_<e_FLhvq=%lZjt{a
z%5uT(alG;L6g?_;=8~h6LE9d~m8%NHYPIHJx96gR-R?VS^xfmv*Hgmqq^0z>KLFq7
zJ{{R9M6*izS4OvUE0-%&&H}Z(m4X>kp?|>xj&pB-q(}wMgd;t0j?C=>=Dv*5=Z#iD
z-19gu&qEq}8te1Je}(-3fI?oIm3uz9<m})zbZWbG1aN(AzIYidBzd&%T$&Kq^=7Fk
ztXuM*`=J@rxVCu73a&qnL_-W{1kL$y5E{sm|NLC~^MnNKaxN4&144r{^%auIpkqnX
z4M4DEmyrxr4ZV4;DkB{jDtQ;$JQ$KF1H@Eg2?pvrR-#;oe<ntkLx-{=4C}N4u6>TD
zeRANp8@qsFsH)nCoFeG4PW7sdddtyQ)EN+T-H9P+R5aYC&q+xJ%b{p@=1F`6c6ik8
z#ng0rPr_rbaR*utZ6VHq7e%}h9v>ebLAU33g!hL+A;*&HnE<VRT<#oF(KE;sQ}=I8
zP_t9L6izNhe@V|aK%>{*{#2ZL!E%~yaR1zY0!;yQ+d}WD+iJLb@zFu=wsaOu_5Q8V
z2Tu=}xci58x6>8cUfL*n=ENP&f_E>qPDK#f_8sn29gCH^q0U#yL^!wR#`B-6R~d4?
ztTfqxxA}KM%Vbs^&@#v;Xtdj1oLKya@_Xa4*X?~>e}B-?ZIL8gm(Yj%8LPf<mHgw^
zGR-6X?B&e(JiS|ZTIEuwQacPu`;Cy?-~&@Ypq0ZxkO2hZK!*4wKG@UYE7U}JJ%oDz
zjW-DK!AVlRQZ1L->6ngS{@iO+#9SQ89vAaDVK4Sw*qesoSFEs?cx~3;26K~PMP<zW
zyTi3Rf4$gqT%2axNf6;Rh4ejh=xJ$?CFgwD|1<w&b+LoA9ZPWk?FM89;^@UZ0n4~g
z_D-U<xwm#iZ;iA2oY+mQgJt1dmO>Vn=+(!$s6hE;E^6XT|M|1@C)WmN1rHMp7CvN2
z_f*-Pg#X-$KOX-{(#s9WeQ{4kanOJ%XjH0Kf0cs<;*wHOkQ6EC$thobfB(!=70E>~
z-6H4@0^Jbt1kEs!EML_sLj8~w?dPDA&{9xTS#{4$;BuN0Ib}pX_ri2@&paw<>RUk>
zAPIF};FvgXOz3Vz=o}B+mfzw$afFNE<{8Rn0}T3Tx?L7@I8>EU!H|Z(7`5QH1$2D@
zf3bLOZgTmuA6FhMUgW&p&$)IyOYJBsP)2e&M%N9Tngk~t7iWcwTk$<WGoOa@mz5e*
zoOf@*$J*!4BZGj5pdwIzycaa|Iotu1xK|&o->J|cCq|OB`g?5u$(U_<FUaBm%2>Hb
z<5fDVXc(F_SQvIz)5rIZ13bS($8(+Be^<=ZHv5`B6lM5v0*x2*YqPn(qdMvybioqy
z7eDcbRg27!^*}PE8))r-gnsHkS<D9L39_Uw22eFHoQj$_H1IW_bXcyw&r|5;sxFa^
zS3AnxLLn|e$mhKB#}UUJhA4b4Jr0BQJ4X`NT?iYo9ERnF;<B$e9foBLhk?T^f8~4I
zWo&C01*#xxxc(dkseL?ba$*g}gQ^?_8PY5ishf30s8w#cnvRkHB`gEt>?0v&2S7I3
zC9hg=7VM|CJ8{~r3r&OAd%nAKn>+`XFpn;)7ws~YUcttITvl)t5$+P&<)Uz?_Q^Yj
zyKL})jlOLMTmPM&c%#6Gk);j*e-3V;H_^oT6o(VtEq0?#ktNCQ74Tei*eCZLZl5U9
zD`AtvH}XTNdwS#6z%3BMC6h7NfC;C-|Ir<u0^!TrQ<!M#prDJd(BUdjurcDLO~;ee
z7-v!STJf??6u+0V?mL{bRB_gn`K*~p8QOuTRxY<|Ep}9K3|ibwV$2Kte-MX4`iQwT
zrc$6go)+Ht?u*!9Px^V|vwLEbRaQ(V^$cT^J7=Kf{H9Sc!JGhkMC61bouH^d)OBg1
z%%GBZD#0E%I;k63MXx)&{jsluv&~YybG1#VEDba5DSI-^y$=AQMfPdp?aR-4>7r8H
z5MTF2=Wf<@hco%KOrD97e*%W;$Q?%9Qf^~uWG3<t!zR^w08VlKp>OAd7@B!Up=F!H
zd`0+*>|;fu+}CqmrbBr7=~yNm4vp_C^AfdIg#1Y#KgJvSgXMCyNC-l&YnSLX`TO(9
zy@<cjQ0M4+wRAzgOPWlvKV|(oPz^(!t$(+?E$g=_6h{#K-HReQJb4(as^R!`vM|W6
z)}I@yGF}2}Lx|r)E|1jVS4|?GoTR|-k}*V2S*eo`eX;2#F9+&qNnW6jEby-J_LtwS
z0T-88!T}Jsmwy2%1_}S1A$mm?0069)$%O$Ue?i6P$m-sGXDm<kHSv@^GZlJ103;y^
zWddLUpkz%d|DJV7LXzdxQ=UxaNfZ`NpFaD3>w|Cq@XPk@#ryT)xY_SsU+8bt#f$as
zW`Dccy?uT0*Pq7Ac=6(RTJ3IEcl+J?^~J~a@!}8P{p-Jc`{DAqKArHj;|naYJ6>*Y
ze_mg_JDu(?U%fisyjyQq$8Yxc>m9&1`@?p1!ha8MUu{>1pC9gD-t4#ctJCIsbGJEt
zd?iZxi|1wbuP+`ByUXV#UT!xxhy8K?=CoXLxqtI!bF+T_=X}NC_uTjClD5Bj*sgb{
z$2-0{tnaXg{qFc~bALRqw*9}Z2Jqc^f06fp?;+lA@6O-*pw;hq+z<Q1?WZe#&-RuJ
z?hpH$_3;P>ZST$--0VKz*#2grPxt-?_kMoF$E9$&R*!#uxpViAEg-*Hz#rHD*b?q}
ziC4#u+x0K!g^qW>ha^wX51Z@5>hL6R&-B>dTz>y{w?C||@30-v<pmz?#pCWSf4+m1
z|82kDzW8u?zdqc6)ld<wE?zCb?dEuYxB6ICH$T5U>>qZwSMOH$>nCXaX7zA)`e}82
zb=u$K>h~*bGRX6lH}6)5)y-*rxVm56fI)S?J013S=kML_|Fl0<D98c)d%ogPT|WPH
z^`v^ZVz=62+rLt$ANRM=jSrU(e}~QQr|)vX<K4CW@|IuUXb)3y*xasvTF7;E`gpe<
z*!<PzZ|l6f{o}*&w84sxO7&0d;O}i<y<6`5XE6AukN4}bUY#C5rvIm#J#u2)ZSH^E
z91i=#_q$t&z(0GlS6{LdcK7yp{^P&)`_uWWN~xf1^iSJc&@T|BQxuMfe;=76CYfi}
zevLR)`F*tRxp2hQ!4VM~0!Mrkz5}CT9UPJ13`eXQlhtf(kt42jP95W8P8<mmJ4LK9
ziPONEJlUHyrSkjOpwFuJ72>2SzH4W);j1~PL*s}a!ujJt1%4lIAaNYAzH{n$SH0s%
z2<6!leuc~>K4>#IPebU1e?Mtd!%q-Vjqf_CqWFE1PMYs6={Awk1g3JjB`}Cfd<w4d
z)v_v$?;vaM`SWD46U{0*_{tISfy+TLy>Qv7s!z1O>O7J7Dxx`mszBg4V#IRUX=-Fn
zohE7UC+)@!zFIe9BE3EM%xOJTZQLFVE0*(g*z6~WbCuFSE8QdAf3;SUHQj+$W+ci5
ztz72TA4pHsB5A0w<9pLagz68F946N|iF|L`3#YY)m+5>pkuH>?T1;=0r&^rp6U2u?
z^#>1XDwSR!-BJC~CZq}HENnTIQ4KE>_ztWU&3A2Wobbn~Vrk@{VUM__0DDyBJasbC
z<QgM|^VAvFxKz1Ne=M~j+N;s|4!ri2bK8sIzGM-q%B9L1?>Tkg=gj1q(1pu^55sWk
z1TBTrAc#+#b3uY|Zb#+0-_}u&!08sHPB;h2>6`{hS;zU4tSg+>@PHGZCfhULL2_A7
zuJNUE`<5YZTz|5$h0`sYI^kTlX>cxNmxWU&`<D5Aii?SBe|^<<!k=mpcyh!m?lpDO
z&T@)2DLkgo&4$9Cr+IBSPrEJ?x^?N5%C#Woij+E{wdRqxz$|Af*FYlIk_F*UDhMFC
zO%$jj_dg<%WrCQ<qd_2}QqI9;*-rT*;sW<sBEA?-0|`m^^B9D0d^Px=#-GQ87Wg$z
zjmUZGwBZ>Ge~@UYwG?M1a_T^K)nqlq2b^x+Dz1&<y(#>8ypN9R99+FG)VfG8&hxzm
z8+oh@WRiVIG)*{l65<dpRaxKyIf6`&QztvI{7EyEo>Ql&ajtKDjYyCDd5S~8&Jm<~
zT;3|eDo*RB;5IoT+H#2pJ-PppX6i&6oAW#e1QI6pf1tTZ93st^z$K)G!Q<Wq;n|0D
z1YF!&cGbAOmafA@u6r1juQm{eaOnloIo*a16D|zbIStH`8&Pf>P=RMTMp@wz3J}++
zgaGlCQWQwCRC<jHCep~kXjKN^fku`wxkejq-;6e$M?wbSZKBePT-Qw~Dnaqu6A+vE
zlNwb$f0b)v;AS`%3`{(~4=!?(`&f7&iddKfz5}OP;1q?I;T28T<jC!{@p_VP8!v^|
zo`6g|pz%n-_};v)6Y2!j9KTO6K5^;<pLt|w5(I(MEg8%0wMnk>dWcDo5KhqyhcQ8n
z@Jbet$ZL=$TQ#9|fsN<#R#euU3&o5yL430Oe@Tn4+%uUHgvTyG5|8;zRkm^JRMq?B
z8r>$`u6pnY5?P7nbqdoU*W4mC1LjST$TM*uoo7a-K^eGQH;=r4A|2|@rFWRbYc*yB
z(3>H4iTWsYtn!^(Whlbnl4xN<H1&;Etni)$5XU1(AQR46afbUwATjbM6|Z^S#Yzag
ze~VzDjNETq<653v6Pl$~%vw_if1U;6&ttvgSpkq#Xk2A&NZejqC!Ska_`k^WVC%Kz
zm3km8PY`%V{yg5>$*zM6+$!6kd*Sy%gv2uz8=Oy6Pi^pOaOn+xaJyrpMz}md0<Q_#
zD1~>1fDB&mwGmoJEd`JnMWYS8j%1@Ve~N2X47Ksq;OBV#4RKvpoNn-dol8{;MRAGG
zT6pAbvuIwcvG9OA6So<f%K4M6<#kJ&U7FA>`^w{9n}g0=OLC}=uLcV6UV$xA-2@4i
z$BB3t&+{ycn3H$9Yz<*zks6W-zJsP=NI7BxkBDuvo=5ApImf$6KqBv`*%rK>Um=70
zU`VLu(I60Oxx97h6JOFzN_+?KOT5Mfq)ccq5TS7W8At*7u7{m;FM)K!Wpo6`^HaAJ
z3aIP=ah+NuAc05tKqe8FgWhwVI^`r$E;!{wqfhFzQoPCrq$mEQmp_gH6%d&PC(<01
z1BXb6+XEn;cdeIXjsY!ySNENBsZoE50Jif!lM6F@;1D11elU>4t-lM%e>jp<;$Frj
z?Rb{sl8K5xX|j=b$6Qv8_xOQ~!B@+=^B%v;5_rZBWN=UGGBR$CRNRNtpvt61?BLCK
zt<zQb1^ztMr8K$58$O`|r1A=&>qvmOmH-)?3-BcqdFwtquBSbJ)QMaJ;ZqZC7|nAG
zAieV^9j;95!SJ1TMSu*UQJPmUvGhJ(>%enBuYKZiqQ@nCZ$KQc;d{}R*Brbc<fGOF
zNatCOmms`uximO<mD<Bx^G+8K!~6B#n!)ATTbsH6@lHiPP3Pfec=iq?^Xf5>%IzDF
z!K+2yxk;VedzCVO)p`$hYfkI1v{@#|;9BCtEJpIdWv)NVkQ>iFd^CnfLO>>#+()N*
z^$UpM)qNkmooESc81HaEM=bAWd1PSR9(ZI^yiVb>7aj@u93sz4d_hD!L9F3*OCTfj
zy;a?Kui95pyh;G1@hHGoBfLK9t93j^0up(>3JH_rv88W+swhr_rZe||K!)eq*zDlF
zLy!E2&y@J)2ltJ>>%!*-eHT%j27NYh<B{X=i4Gu<&!ze9YMSgIC|+grLq#5O`4Kwb
zb%6cnel93PgOqb&3CXGD3tB}kAweS?rOy+zY21<rZ42-12JH-wxPl-SqtYA1P5KK!
z67O08>D)JeE+b4l;tIxh&Yxga<kk{M=6%3m!5*p$KpL-X2kQc#fC$!iUfT#rDJQET
zrJOvE)0t<;0Z9h8qd+EcA`qmG6C@_iQ|D`)+$W?7-Mk3*nm`(lYJx{rJwd>4F7fce
zJl+ZLw7mWnf(g8@8NhDEcOAUv(Rzp?d}08I<N0=fh$iw33`phY3PQwslWRicHZeeZ
zCOl1QGOZPoaXdZ<S!=~LIE&#^Iw70HBh^rrIX8|-418}z2Cry_vQ&jAQnlj#Csa47
zg2M7nW(h)5iTA)lGs5S4L$ii=twM8?**qYLcd<kB!mIm>_u)0~(A{KuGxRVf_Fyar
zfF?+P>s;rC>Jz#RSGl}J1ks$QQ5m>Yj=)k-e;ScZaasfEooXZ0%J6(4%F-8^AieRu
z8I>p3Alc;|X@v2CPpn1b3-2yR3!&#buucr8fenq@5QP4L*8-xmjnC>vFP%^5F9@G#
zN4k)?caBSl!gn2X;WIxGVK0v<BJ8N8mM=zsA9;=*<6<i)l4LT07n2bldB?13=BpJW
zyz2{O@D5=tw($AHSX|}xw^-om_}=Ow^c>NCvRXB%@f}oWdF?hf)p@oXkz-7ByP3*+
zAF)}PXh}nSO|3tW%q#VAnLpr2*L9*l-R8=rsw0=+&ofl!6Do0-%;y~9@Q!D@aV!CU
z&jiuD0|lh;N=Z`M^7Cd096CQ&kd`Bflx|7EXHh*(2psqf29O;24zzJxLXx(b_w<r>
zo=;T*89X9R5O|);BoX0Vgd|4sX*wXnwFJmymO3pXh5Sj)Qjp+?s0rN+Hb-><fnA<F
zsd1HinPit)zX=jO=c%3L+sSF&_-amn?1ZAuD;{kC(Y%Y0yidG#oBWdJP$b}ixKssS
z`Ls!b0`aK{Aca?)Q<Ba-Q(D+S?KP0X<HVHgOn=hSU*?FD3Eh&b6aFMOc>gpZ660r5
zQdX8*q?C2xb)<xRf!D=S_L^J&RFvmu*ny;p&K2Ef*I3~-;#3f=bBb22`H6UcAclK%
zAcJSSsrtz&g@o*b*I`q$jn^$x*D{gbjtrTv)+O+~F?FMObPL4t{3mrgsokdTI-i+K
zJxuycY3R(WY-u2n=TbGS=I5Q$aD`V1fK)yOl18Y!C!UoKJZsC!Po}6c!gwCLAl<e+
zW68_VIORgtLHIOX7FBp=ltn9leqCL8G}LYT_a!5xK}e5;5E4zYlx?OY+e4FWB$Dis
zC1aZ`U!p>`n#ROnO1!dU$(}(m_LzjGvW_v?#yT;0rt#|io%1{IdGCL(`+Tl*pL1WI
zYq{_1Txp0Wmhaciwv=yvQA0&QUvwODDW;gH5}hcWb3GF&RwVD|sVidy&%3Ekc+!?y
zOLi#syr747Mi81Th{(#Ofm^4_q!m}gHD`F9kPB76b-EWwSMn_Y=+&pA_0l-n%3Qa#
zK70`B5xk?>B{v3}Imw(szE8-FqMA#(TmmVASz9ykt7}lx0UJ!0l6L|$DiAX-c&$Lr
zj^TW9d7#J-ZJ$lF?6*J2uDPhj9&JoE<`#^22K<f_npkrdagprnzxtCmRvga4HVwwz
z(u`urRH|VY?MkK=9#VM^$OpI7=RaLqAJG~n^o}YR={nRLiIL(pn^}XBUF8L?fcpKj
zrg0v50(|qNa_1Ui&RvxUN0Siq!N#!`Zlan=0df0W#$kQd+dbMSw4Wn66fSu!6I+ZO
zG0OTZr7Kpy_V@RP>>nb|!Y-z5R!Ue_$3l-UOfJ*mNN@0)lm%&{zE6HgPhTo}Kvyh6
z+9u}->4x3d7gj+cHET3?`gXT27)lCR?oY1IqD+g+3Z>tUi)d%az3EX470e*_Xiu;(
zEr&-eO<qGucs@ne8KWexb1S!84<*<t|1cQb@i|6Z5I7exUT8eo06S$){pg!($q3}A
zbk9a;lRPB0=i{D*626r%H7Obgf;DAYac)Y@4{XW^J__PK`5!w>;rupWb}%V?f5BRe
zxo@@O=P#K`C$H$$b%uYor{_i`EMU)H!a>=Zj{TP&E1S(IYsaP}{em8kc;x)Zvg~hr
z9hRW{{?Qo&F_FGSuDu;&CH77JV;Wo3m~2;4a-qEmcJW-rh}~Km<%)x2XRV6cFY>aT
zJJTZ&@}qmjJ%@O0*iEAa^jnl)33MnjKzQEB&CAg*gZBJ2NzlwC&XmdcLW<(3ncn3W
z_0mt3oW};--Ecvo4&Dz=D})^T<HxZ#1@I985RafIcJVL{^JGR(q@n7%8r&UHu)biC
z>J|Y(z4ncQGDNeSH3|78gPtzm%QeJhUa9S`;0z})b8ocp(?C?C61kTH7rfs)pV>wh
zpn9H*L-<0Ih<(MLn2w`YO6*DHLge+Mf{FB%xi=%)GEQF?lx@{Rb?3>A<HJ+9ogc=?
zkn)dqK{phfrbHDp$jbCP2eZQa{OXg~xr^7Z^#{Jb(kE({UDm6}LCs`~jp}ozv{(L>
zDXAK?ZxF&b1@70wnBEpsHfR&DiKrPAVje3zCemJ8<>;*|DfRBKOZ?rlNqe#{%89nQ
zri=aaJ^w%m)!D5SXt!}mHagkVsL!u(s~C~smW}k&-r|Sr@+z(;qYs*}bOq#6!|&dB
z5t{5&lh*24bY&$%w(_Lm+ML-DE#g98;wlDKQ^+U;Eu;lAkx6w-<>O;0=2UGF3hyo~
zE%@*s;#cQ9rt1kE@C%$!<1>CefF8Gx8-~*fA#`tuw9i(IRzCD|ceo#Xr}#9YSRqtB
zd(oG=XT|y%-R_cdNyL|P^0?W#i+U8KzPuYZtv2A7N7%b#)|}sGqUY^x^LXz?y%$tl
zBE^8?sp!2ZQE^rJHLSuTi}}IHC6|d3VxJo~n!Ny))jFHR3B@c_IL%@*hu&6jLIg4f
z6dt#b#-5-D=07P1H(@`7mg}F-DqME4nt%>A%NVRLf4n;3_bs0Asd)2E0u>TDj7@AX
z#Ws%5n;!)EE+ci;dlK?;X2qTvVrUYxdEiqjYb4V)#d*5AmSbQ76A>L4QpMKeHWm&d
zzK;`BTmWYO6&J(hn}QXtzs_V*Uj0g2jU9Ub4_T87x1^0E!o8|#?wC-S@r~;fS4~3$
zRL<*mT*wr&;=pex;ijrSon8B9&hfcUP2gVIe_RX15Ae_%$ii#!-fHd`TX<_a=m?cZ
zW{{hz+P8X}e^y?)fb0s6$i9K6iA3yi#!!SXCPMLR4#Y)@2wuLS<z*%^!N)g^QPkpH
zKRhrrmC))9yu#8^7-M3CslXS^Y=B}G$rP>V9?^wXB+ulG_E<&r#75I=mF2MKbAYNS
zsnCMY7|8&S(nbO4zPqO<LqYqb$Q0{l#=RAvouR-g4x~6L`Hq%()T1yhu?Tg4Ip67x
zj6tiP(fxzJXO>j=f->C{8A!$9lq!-$iUP`N4swMNb1ZT7?j50(ypp#Sc`xbVlL0XN
zr1nl9jYC~W!(eqthvAXUol3{%t$W*y*A9K>%Y_m6{^gWg6_ZS1a1r*j^zowR);dEh
zXsmpBx$|}?8Y1KPKAMBct&3r8vB%R(+eayjaFBix(R{OYn_g6#Nn|?8a_N;|189Vh
zMzN6ydN4ORMkajJB_OOwR%eto+x2*{lx5#ELZeH5ZeQM{tO%!v>{#Xr<BP%<m*^B4
zr}94MP*&FpSRM}By}x11Bs8^;tlMfV$~v(3@HElCQLQZvqonBYFMa9coYmIR5!!6m
zu5&BS+fqgY(<ZZ9{oXN8XflRNXL$#6nNL=-Sl&Nb-kY4P-<&Lc@FyJnNg-}Cs>5Yh
zsE?camuIPxM6NR&tZr=@1+PzXXT3S+Wer6^S_8+-CUPfQF_FrhkDryb)N6?V8@Y^q
zCF22CC1Gi7!syb1<f@q=Zim<!UBA-+UiiJxbg(XJn@c@CR(`dVMOo%7Oq+`C-huV4
z=#1Z$OLLH|rxL-m`N_R5hkD(-@~`aD_JtZmP8A4}Ynn;plTpG}ZMTp{5|$O)>fa-M
z*n(lmRj_mpf7m@DD^&;PpHBDGs%>WOXQ|qJ^(ok;RLkdqUQ1285_sBMKmz~cPaWAb
z2k1M>57(HMn8r)i0w^$kaJ&{@?q@UoysY>YwW_Ur@8ig$Z_qkToz979sD5o%9$EQB
zG(weq%gdP9E3l*%UnId_P!{sG&M=;(-=+If<e`7ZT$!8eab{`P<9COeP{yM9tOar0
zrQ%n%hT3~9@aRatSJTIej%@m@0;8cEolNbzoafJ){-DV{sA660uP1g?NaL!$(F5t{
zqPH1Jo<o)BiNqJ5+2qjN%#wS-I=v>(taprNa(?}6Pr0U!;nq*uNtWMB(us_&VC2JI
zhc!zMtsW{+?Mb~uD|N5~|8v-VuChG__4ajuHUQ@C8jZ&F8h^W1nZI@9S(Ey=f|E&$
zn@j#pZ%2c+UPKQ3W{eRE=(T6!C9W5#VoykkipPrL?7W`7V!Iul$p3%|P=uZ4`|C&S
zWjbeFUP4<8_t3#<#I>1MvPO~q;*Dm_uNJSP`c?b-Zw~8Oo*p<|+8CG~7Uy@SWt%t1
zvRnHrIMq0lM~5E$$|t&%VI$PN)0Js>@qo}tpR*IUrL3b??^piF^`FBj6ro~l+Z`8O
zq{sY<Anrq2*4STPbr2oI>Uk%`oi~>CrvVjYp=%p28`n>~@ca~9HKgx!Bda=$g63b3
z8H#-1x@nNxB7TZn8m@i>RT5d0^UPzI@W|r*B|>B*fA`D&zPh0$T>1>H>cvX~B@MwI
zl>+li?qRsIMlJ6{lz$7nobcQt9(_EcR1nU8T;o34VvKJczBW(n0RjAABOqQs-T>J0
z^PiHFyS`kn-2$8hK&!yfvqH_n-@vm`fgbv@R(w}FGTi2FJMoeIDcP3FZ64+s!!N&A
z$a<d7d$(anoKToAQht_SUy<9JHbv69pNq%JQ!XAHGHbpyo4!FU^~~G;5ZaC#tQ<g2
zv79!y#m4pAKk^I0FAFYFw@>VAk|#gm3mHS6eXB7fMO}*>2A7h2olZ>+B;uhVa`x70
z`$>?`hqN5R#6JmijA%sJ)o94?u#g))`7*DyLr`lxOjC}nm6VvCMW@PJtErOA<CWWY
zU2c2Lg!HkezPp+C#LiXUOZ6Ukrxu{{*Zy(o2D-eftYf1}$sH5(!}+($>}^(Kt-K}+
zgE|R19>UEmHPGVPv&?Ixekl0~mPS5cVPny1o_eNtKICdo>|j1UvTNyM_56l`)N$Q$
z!E)F5epht|8I|p0FWDdHFYBF6uw9jU5B7I@x^O$K9Wa~N$V7GLS<_`j>IvCi$QO6d
z=%BL9kFPenBVS||ULUyrT~e+=jcL9DtQG1TT#5sI)_enK$E!W>)UnTAw_+9y40v;1
zzt_lRND(Ye64bd2S%3fK>3q`0>bVz|yN-|A2ZBYFYe+3wupvXUXCaTReS#|K*E@Nd
zyZ*i)k^TYPkjX~(?<O@kBHI@`Itb!h8>Jy>v@<D`<lg`u@e%lO;4pH#Ein|}>9WQv
zv;j!og&2HD8=$QC|Bq7sJUpU2|Gg9*0zBb?!5%O){xcak_J@I_c<}`MW*eZt5A_z{
z!9&Uc8T{TbQ4zd*J0Jx(<73)^lgIx<PkTW201wa6{|4j<t^x2TYJtD--`fEZ=|4vP
zXHP8e|7w0$i{~c;h5#IYnG9S6Oz;V00Kr?|gzq2&O8<9>@vCG&g7;}NUZ4Yzhtz)9
H!{PZiukqC5
diff --git a/fun_gg_scatter.docx b/fun_gg_scatter.docx
index f3605cccde00d915611f7fefbaadb229383748ad..5e322b4aceb287f5acf07e405e8826e32125aa73 100755
GIT binary patch
delta 49155
zcmZ6yV{l;K7quC5Y}>Zcu{*ZybZm8S<8;)qZQHi(j&0k?^zZ*p%}h<5`{8^&b#Cqb
zthLwase?+ahpIt`fh9S-IARY11&ODxCxn^U1Nyl1_$+(udnR?0VAl>Nj}W&g_vU-4
z)F!Mt)HNR#gn}h|(s#!hM2mxtX4yY~-K*X*3x+f&P}vMy)fZ)`3yqRYJ8e;IyQ6ra
zZ*4O;Ye?p?A!qvQhk-BQ{JuWK(KqqVNuAbp`nd^N_x4v<-FGk`>iH4O!e)TVfK1@L
z3kdx@mP?A*c#0`nX5b;RJ8&57-nCA$v@mkkbylAcZSLm_Ur${9>iK{m<ieK@CfG&U
zzuea^kZ5L8tm5&RQ>rmjiUc<={&p$i#}EM$7kZTol81ADao|mf&B5=$__r(I3#=oB
zCc5V;P|+PJXx`@IQc?9XZ`krl6Ph=p1Ncr=NPUAY3w2Zu5=2}S2=2mS%+nItV#LGE
z9?i!J{(8}z#~=$u72fRsj<LZ~-!`h29c^N;!I)!Pu>C&hGG~)B;G*mz<2gv9nOtAq
zw0>43nH!p7Q(Kb3OU{d_bstqfRFEH2YHsEJnjjb2<wNQ2j)XSO$dsP*Gd{H{3dkCj
z&DK_U{drYtLTA><d-|ISt3+(uSjZ)n`$$h1I0kM*AzX|u^Wo~18IsHnxP+kd;L!9^
zjvJ-Eak&`%+o$%fK(#Sb9F}k}PUxPIy5$S$1+_uDD%quwBryhXeR`^4c_tu7({@tl
z^8F-f;)smTr8BiqG|qs}r_i?qXc9$qAg9;a1Thh{tqjA4^LGDmJvnuK1fopOwNIsc
znQ96UarwGbPh+;88?VE#9EyFNDWxN??WSImneY|aD}}HU6*u2U*JV6uaLQAz$JIh_
z3EIIF0Xxv{LsP*wVL`ISk;(ola_2F2;V7B%vg2LF1;h79Ui%a8nM6uJ)~~#{;Gg}u
z9C))>*j$$Jp<Wm2%?M?#&^{IR`kbUJ%{KBO%m{?n7k<b`_uq({zyrKrCe?Uzn7L4S
z6w$O8o4`fV^qYU)2TJ#%@<mZF)nyxGD?a9*ZwWnKNu1k=&oYrxV(}2`La6FI+|s<m
z+{1Ows6xV8*5*R^p5QtGCvCq6AW;k9Xd0{MOR5%ao7)Xm$YYin&UAfUzc8(`6PRdQ
zRTmMJAOklS4<?0d<AjqxcDtl;Z40+F_`7FsY6vQxxVtV_Ac<!J<<rD|GBq=-)-Y0X
zq=jMM!o^e9AsFZpov>LfDLf@1jq(x-e%VAtvUu(2due~X{4^2)cE4LZ!N9=~;;M>E
zWm>4a2s0HtUngaCrjrkBo0PTcdE=VTB}P0bTP6|?^rJTuKAF2Tq<B<J=&UEC{`iaE
z;qy-9wbhr=6j%AC@hiT4+P|J4+7pLY<&gojuLv>_&Y_M?p0?1h#uXeD%E5Sg;`6Gx
zodOh03Yw(Xs<z0=)Xk=+NgX)bI72b+%rq3zBVi<VJZU1@?@!%x_=PmN`glYcqg8K3
z@KU2VU(vb{jbgQugx6q6-^ZX!dxx<?HhXUL=Tx$~eI0K3FG_q$=Ii^OOh#iaUu<pi
zeB!2i`OCpeVLZtV79gysby{+hPx*OyTO*{INWUw(gObN&&`*e&iwq`sb)(?d={_8)
zM_1&1M=hdw)GR;QB}~<<9BOdQxqm44lw=e!4?nXJ_Ut&jA#C)w9&y+b^4iR@yiXq#
zVxoBos}eywJ+mU}8lyKa9h)8%(n=alo18xB^ffP;sLe)S3Fy?aY+Q3HTzi%+kS|s%
zl~MnwrdP9@yfqjy+0c14x;$w)?<b_$ySZl2^ZHrGWAKUj@%v8M=0)D-f&c8&wbD4{
zlVa<xNdifD7ioC!lOX_Z|6-VK=fMD7a@T{R!pgx4zvMI{Sq)s1+sSY+{#cC-W#a3G
zwcsWmzszZK4{(B9P|PU^Q;b_I;1y&@^!g45nY`(-Ke0olErPg=TGjgvB4Rd`E~&s6
zBF;8IFd42l8sWXMRhLk-QvHDpAC(Uu;ikx;6;0W?`BL&CSIfwsLI$rYS@CrqCi+Nu
z*D>C(!X-0C$&Qq<wBJzxs<T5hYcpASUGZDo@q@HsB7l@{5?UsjTV6Ipnj<ch6=IFz
z&~nrz4i@T_cPVJK(M{AC*w0^UXMZJ&g;wt`CN;5>Ws`Q)uNUu~fQ#gG6k&(>L>}Rv
zY1`_|q|x>eeH(JhKaLe<6-mY93d?SUE^>mh)Gyu}Ek*DgmS}l%Ghq-KignaF1YJGN
zry=+R0`w>S^u~qe$ZmmVW%PDLDVRKlseTcIBIm?t)!+QCG(6=8VKYd2N%K1Z+`;l#
zX**Iz)Y0;ogzZLC!%sNvOw>dV6eKgu6{1HB><m`%iKy{%D|)rJ4XlFhFYo-|fDN${
zs;eb+HTkydrh1fn()uWBI=4t$vvRUz6l66>HLzsO(K^Y8Y>h;X8JNE#VAUVR*U6+4
zDvZY*`@6>NTqP41TYq?{LN@e;Wt|*GRV~EwjruR_vShyr(y}P%0Gseprg*C|LZOfQ
z{yV4sb{&TwCQ))_*l)b4z|fR)aU*2PI@$Ye7>RAqRu>XsWiY~sE^JG8q)bMdaf5po
zPJm58-FLP$RnBip@`$x%6ApnM_b;&v`9#m|FD9Xm;ImGLPAqM_22TG<gsLZ=ny_9c
z!ki-Z-m}i0!AUz&5!4(nONZ<!rXNGT?0Tv)g=z5m3zGy^w;g}Ne;IferiJN~8n!2{
zv6kLdI+^>VCN4=%rqd@wePdqwpl;?xx(DW!HM+FE8WglZ-;7GN8hz*Ro$DX(mMD?B
zPrr4mbXV-dB`5b3d_?$eCGk4dld^?aNmCNtoWc}+(^nY5%cZg(qQ6Hg;;d>Q#8p$~
zEN8e}t7L_9_EuxFTk6-@ZiyL~{H1BRyTENoCkN7#&U+M03vExjGnq712hBw9Ap^YV
z)+$PV>rh9ODq)c_LNiz;_PLf%u+4sv;*c<Bb~>g)@;oK%y<EhQVp;9tN8-?vi~WM;
z${@WNdEXx~9RX<aZ?;D52OhJ<!L4X=u+U4-CdSCSKLmdL#M_xPz4|?O6>s61hib;0
z4@6a>aL^@4gjeyElBpFdZA=jRr~x>3GYCoFLw>wJFxftK)R82@EGD3E65_Ju%|e~b
zv(0RIkPZdQAnhjl{if-IjwseZrHc;QML3Q6wj_Ha=yT6$n)9yM?Fg%qdlVxvI)_|G
z98l|0AiFbB$5bE6?>Q?2iCx>e>v@^+Lz*<Br*ZKqg}SsUz7WBxx677#y9`j9nx@5Y
z+-9XDGgFtc5IjcPTYeXz9%hR3we+zXto}V^K3udcASB3wO{-9Gb8j=xBiKT<?KMl0
zaveO`nh#Eq*Vs!>*hlc~(y7uxNjHT&XzjoOq;K-d<6`uI>9u*Ye)X*vfw{58Tf3bD
ztZ(afNWqXj7L6>@eO~#rsRZb7D`P!Qar*nMpU&STaXa&-2$9hA?k|aES-2D;`sG;C
zUOh=O{qGG;Mpz4!liV;^I@Y}~mZGHUHQ$&ETDe!-B!#q^$vl(eAfjk}-9x7E4nhU!
zRq&unI)QV_;iZ=;C5-eQm5Np3NuQN3QP1}`MS@!L?|XN|qmCNu5m-R0@h3g%dc2CL
zIRPyG`o@Fn46KF$TE-++gt#QV-R+aC4@Nujvf65g%<CV`@ql<vWQ_C`KKIC4<2&N(
zLn{lX+6J(zu?U%Ko~@m0Z6?rQ_p<<WP7MX1&OqhiiR(b<n27+~o0gB^qMZYemRp$i
zK%A+WSgHpB0T&iFVgq;xX$NMR+YUY5%dSu<<=h}P71O6Ar?I0FM<W*r!H(BN1W5%1
z10eU_j7*_;e~^cdZ%pIRf!!{1f`l%CXv+PmRI7zesUO{p5MJ5SV4SkP1yhNqJQEjU
z?>wH6E-+w=-sie)qjvI<Bj~6#w7#i-kbwTqw`8s+bnU;T6$Nnr@MpWMz_hpEz{|Pi
z$+i<OV`vpkt7I%~NYg9bB!8gwY1NO#eH2S7a-nh;pucELO~-kf<OaHuKP%iN&+^yp
zB(qmUng$V@qLtej6Y-pmzN%D7^HGs?R;V1o5!<$IZs7F;i94XHORkOGVCz+W67FFz
zTeywgwr+VxbO3{rDy>kO^kx`-oI=>06d2*+yU3s`TnH(p2v)w#9?{<fCKd=Grcz2V
z*i_yy#fu&%$yDBO*cyk6&skL9BKUQYAdsgGX_0pL!92(^)j{1Ai#a6~3?wW$1ry(^
zPP%pSRfj{0rvFl_VU8XIiaB8xFh28ceUG!G*v=tTt_L>MAG&yy-k!$dwX;gbkUmw^
z@$hwIcJ3j(-%gmO>Ko_2izOwkU8tvD1>Af23Q|BuQ&&eh{#kNblEyX-Wk<dR)1742
zOK?HP0@V|@s&lwhf2EFZS5Lp-j7tLpZHv@>YGN1<3%wkVYkZF*U8>rcqTK+w2_yN9
z)ZlGZ<PBtNf2|@7%YFYaw^3w;yRX)swgS-g0{^yxK4}=e@0uzv_k)hK0xTgU5oeA?
z(QNzty*Y>6>{ua;*lmW#PQ=Ju!Hj&+#8JOgB$l!DOkARZoug*qFbCi^BW+&zH0UcP
zlU%x5_{b)15WQR6SD|3`!}u=FUnPk(uy<k{KY(963lcKvvm@@VhO(-x>C~?-hA~(6
zLB0bHh!&0qYrCFJDSAPMl)WoXRip-OH<fpLKU|Y-2w-D|O<VeG;T*5ax*Z0A3L~Nn
z8043W@B$YT2K}IZ2*O-&gE`OR_~;PyE4loU3ni!aLmWj>HNSFeMI@6Q)aG{x#(3q3
zV&K@m1j={`oNvWTd`f6U!^u1r&e%c2`7_U{Zdnha=ZBuIYz5V2_puNRbWGuqV_D97
zJq;C7cI7&aW|z!!@wy$#n5W!}#Ocxr98uv=;5v%h{szpa>eVLd(t4m>f0jblj~QqG
zlM%f+TNaE^p@FAuYfHhDqj5VGG5+1cKY-x=KKFMmyh;vb$QDy)1j*xZr=^EOhM5Bv
z&L0x5Vp*7b$jiN-w?-_88G-Er>kkh}alxex{0<7oyteIi@e~Ua_-{h<99Hwk*6+75
z(}lTT=F`flz#Xu1wo2U(By`5r8;k@^ql=FSjE<uFtMDk9hHsAr*upo-^a~X+0%_~;
zngg5%=BrLmxf}OWCd|82GKIRfwTvo~;b)VpTWN7MPF&LyDUU-GDm`1<b5yR9#iSgj
z%9a}&f-V$?ps~L~tGPKdv(@rlci%}Z-h}S98G>^AJ@!i`He(Zu^MGvPFG7&@(Nr>U
zc@~n6u#n}#2IsA4eA||t%kH#GAX&}Ca)}=HY@#d6>X-fVoQA?zp*;MdirX4e78`C3
zD`60ye8AASf4}?DOFeW~%{COi^cnx`BAcv2QYGtToMGBpITD|Efw`J${^<b+H@z*X
zpEd^VA#cnR-^h>i9>Zq8QU)ENn{`;4af}oHPcL)znH7XLvsB;fET5(oAhG=1Ja@K@
z<t{~db$fOx{Ng8p@gv(UJbqRfbpm_qZ#rC%XOkr=&P)`yDL<|bMrX*RwiPr^3q#;K
z<ZKoAq55lX_TiF@>GuSjefrC`Z72?6$?ekw)Q)PGVc6%vkE8Vy4r0*f);Br1YaDK3
zWcJYNi`Cat&?`bWzYjkifQdd?dvLgjWWDu(PlU23yo{etpPoEKTF-9{6?urVNMx6I
za_){+eV)Xrm2l6#0zLayrQc~TU)fhvKD;X}rz_3AJ*D({sr<P0k4{xyvwn|Tts?HX
zPQLrwnem87g7|etbJq|Gxa#dcZZU4PYWW9CjFg8jFBf^bwtj0%fQ!k1lWGV)8D^xW
zSd=D6PZU;hEe&N5*X%W+6nMrSqTV%ROY@$cU-@V-T$s+2X$wp&HRhQj$PS(YIDP;V
z*fTFzk0&MJd!z0<6pBxsy$W~+@iwHjUm%1cNVQP}C9SX@>;_2a-`-w3QH^!lT8HDh
z;$2?6M^sRGcD>MJVEcCuXiXbptOa5p8_M8Sy9<G_IzrK}zvHBP`<#iOo}`euD>BcA
zSGj&Vy626Q<|<_Cw5fi6$Or)s5G++zIogcsjAeGHSKgsA9S^8q6l;>+_u8|pJ}6Cp
zFsGas?^>GXLVa}=w6wR(7{}t-d=8mp40BsJ&8P2^AF>uV0i>~=Ym}BOyGl@5Xs2LZ
z8V%7#L!xg&Kete13XYZV;=6b?@}}Ga;81>;aVsT$;+^2NNhe$lDPRP1wu+2T7>ET#
zl)weQs$Ss0d+!B*E@MwPQo(+7RZpOcWERx2Dt)`A(}Wu^_np)?WxA_6`F6w$W*XsU
zuECeX45doD58Ne+V`Qtzs&bFZ#I2Qn7xaZMZYqMP><4{pDZ(ts-9|c)+UVO&w0^Ov
z%788*pu&b(SlR*c(uF+P7XKRR*;HJEiSVmw%tU(t6l|xuR%2HE;IyX07L(?Vy@xvp
zHDh2O@~*fGW;3l-6woS}x~C||thClH!?ANiNf%3tf#RE)lmot_rdC5ba^9<<{*o@@
zyK6-PoL{E{(ZSPGGK=FfwOSrjh;D=+PR#U(S&s?1Jg&4x8h(0vdhh{pdv3P)oW$C)
zF8m8!^Vb%SvL-}tGO^UUloUds<dg+X7>b~%$|}lA4PnBXq*=k@rbY4=R@rLgwX#F=
z^Ubxnz)8J5ni-D_Sv)>Yd`)}$1zU1*qVY}LE@bmv=3eI;^?}o^_j|<NW&h38EFrQJ
zGU1FJRFE6Bl-tjOpFAfxA0P+9(B5%=8v3Q7Z6)bj@L!&liW7{|0kW_f`nmyI|M<%s
zrGgCPu?613mig+$8N~hzYH7Z!N+6$lc%qmPfJ}wiqt)3gbM;!0-aKyMRy$+NX}378
zSXpLzRm`9Ll*EY+wLe)Eji2oG_Lj}OR_;9ig^NV2Vjm+>)1#XmBmZ&PrBv;ch2%OB
zv&He@J}0yM&<o?K@P%j7<`HDBu>X=Tl=VBe@y@iUMU8nEA7Y9Z$%v=nHatk5bLdwz
zkR-wlhcZeMFI|4)uxyH^*y32U#nrCdxw7LHaB*w&rD^T`pIY-S<H9YlL=4C!x?CuC
zTGL&h0%^COabH!wqnyFVQ2o)Dk|`P~+v`Wd0xBBQiYpb$8@_-iWct!{DT>$sb9_`*
zkIM(Aez_>Sq$}^8RZf9=uLOHj1kN5%_+6DDFENl&<SjB$sk78f=Vsnra2eP6XgKLE
ze>);ZfFrm_qZA^_>ZLB%9HnMb11v6Q@noUc3c>q4OKcTGrgsT<E1DM+%c|jW&oC`!
z-<Ku>&-Oz>UwmKf>;=R_<2LWCdsknXoxK$PpJiP<q}>fGY?jjR3K_7>I%L317G-aX
z$jQ^9fTubYd;+MY06Q*v0}=cSSDtK^{_{ePg_-nmwT}|`MAPq`DU&>Y{(1YYyG_|^
z#+Q&+eg&*^T}D&7+NERh>-#f8Et@>t9WpQ3K_hFdk3lVLSzR}tf#(^N<a#>?+914}
z&JGV!lJ~|#_+DVY1}oy_P8VqH_$%RY?g?h7z=?!ksd*-qS;%!9ZS!3tmMQl*T1^~O
z)s{C%a8hBNjo3Jpe*ES1`v7Yoag@Pj19$5&y~fD0^bIn0o%D&KhZv6EamcfzM@-SF
zpk)oePkPpAU>K|Nd@5Sr3H#EVxf>!WG#=P4x7^Vyr)XC>|3UDtX;C0>C>$L3evLPc
z#-ER529I{WbV((|bd_xgbc~9{1|FZY&v)hl8s%ma!Tv3hnS3yYot|F`UP=_oW5tH?
zXth58tM~E0dLWZsL#lL(H(OMEs3!q7R=<<gtd)fzJQi0UB;eaX2BUX~cT`CpQ9^Hn
z#WQxnNhfJt=Nji$063M+$3$viQ!L5&L&y>{nazLDYU<yOfKDtwX~|sM6-KkZ{DYiY
z*wG?yGS{<vh;rFml}@km2$I=~3mmHDje28c?gF0b%2OZLyvM6VszyGgM56cNjv`db
z+-vAMJTpEewmbjjTBpDued~(h$|(7N;sP=8CIN%DP?EMEn0)sUP2#zQDP12+1Y*h-
z5^XxuIh*6uP3Ouc6P-J?>Nlr@swAHt{ST~^+WwkvLaG;By^F#3olIaC+cw8wBpXz|
zPb)dd^Ba~!ZtR&{9-PzY%$;oI$QP?tF10_X-$i^%S#h0BrVk=9KAF|JPH4UMXQIC$
zoG*LVe1@w5Px33;IEV^I4-v6!6h(p{bWJs`*R2tAT2m9zSMbe0*`S5iOU3ZXgpWge
zl*r#EepNN1Q3t>mN$LZkGJ@~Y0WRI-AxYotQMY0(%=pw@Z_iRcEd*_v;BcU9rNVaJ
z8%tN5d!=tcOI`=)L$L~~(>9!x9_${%`AdgJ+Uc@^es?WuRUrlu3!d8<Z9!Zm^s{x1
zrqLe3hck4OD|^Avw5nIeaS-<<|3{-1G*Ch`L0s6E0Ntq4Th{eGC&XK?edAWe(xC=&
zfpszk3+<~PbKfDde({;|+P6R<D*rO_w11Qtzw;otG7RxFsHt3%{~Kl~XnAwDyAyM0
z_wa57&^@<gzo{3iFmai>6>8hcWRLWnMjRNm&MSmr&6q-Rh|DN#J)MhVmIDjvYHkE8
zC0qx8m=Y~k*l38XkJQM&Nkys8cW;TCwwyK+7xd=#34W*J!HrPP`ZZj2$nrzS<@7wQ
z`$TUY9IaAt@W<(E=RK8pKJ9XeL^&#_*Ie5-fT-+s{(ObatCmQNlqmrP8fR;0Q~9yO
z!}HoV=3uz9ja19RN_WDf7?Jh%y|)f40+oP8I0lck*+DkCu8!o8zKfE0tiwE5zBCx3
z+w1r`wsyid9yh9~t=Z`ZCE^$MstYPqzPMmDdIS_wVk`gau5gF;p~B8DYCrA=rXk`K
zKmb#5*znbZ#>ZQ88eWCAiYAe_bPSIoElvyEV*1sd120>U8G9pL7rEVMrgrId_ecD}
z53pa}_leDdTq$nwK4$lcnYd6D)$kyj0}M|=Yl<i!mkb&&(oQ#|f<)MKxa$iB%-A09
z)mfQ|)8Q76EFrt#N`&8ImXC&ENKt-S0kW_6(<vkrU$4jrV!w6Bb5j7(o=+)v_Fl1&
z8(vm6NdvWz!z^UEuTuG+s`(vOph=1q_yHqc%lQ`c!if*OtXV%_C%9@2Q;lOsz!oWR
zYI{*uhQ9rNol#hSG=xy&AHJ~vt8)=K+IB0BuH@=ZgyLUzrm41U@>o6$tula24CKOB
zJ&C;ti1U(x<Pwr5n_!dr8voV_d5`uHUs{HQTKVhOB2MaVK9I>$%{47Qaz?Wi>QvVC
zk0MSaQxydxJI6X0ha-mL<k!i7VNo!l&p)AxRCMvNWydOccx=2d>qyBEL&O$)C-o)+
z#6N_h;R5j;I~ZY?iLm#fnVhKY01@PHVfnnkj7Enc{@Y$+s6c$ppds9S-G5%5x9g`d
z`U4YkQ@I3)@ZZMrrLduXquThvlo2zcGghL5h(YBYU2-(XnbGMRMVGRZ&K(&8{*PoH
zmXSq_e_p*QKs#tLA&0iGSHSEWwR<ui<F2}ardOEeYaa8TE3|V~FrWpjH6@!EHf?zD
z$X8S@j_<dxs^}Nd{s(#8IM`?$1!2*w(J+=#NQW#wE<G#@sdCg<WhW1ohtT9gA_^l~
zhe5|<C#GVu_6I>J?X8ve9~Kc5WJz}oSAXJ((Oga6Rg1-N5J%6oIMR<W%0gGVPpuUQ
z=C|z)?Y>g@t>;x<A%PE()YvRmHH8tT*FtF%(P5;oDCM56eLy2S6M4^!(>L|z>1%3p
zdshrlGayLKM3)Qu3X3tP%8f0gA;i`SxjpM^Qsp5?l&mt2&Pe`t0>EI>1=Zw)<!&vD
z-tu7X`<u>wQK!L0rq~s1$-V&ymMnM5oK)?QAICny^(jkX0IW2o32AaiSI+|VD_-_(
zK7Pye6Y(!a!(Xg;k91*77GC<p<m+}1+qjyMQ0AJUsY8}p#*V-5@p@nURkhWo@gx-l
zDU$hlr~Prh(0SI&xaca)i#JckCdZD*jV^`leWM|1j42UW&3(qb8Y+MM2vA%B+sqap
z02}6TLfsRM1dzbP#Q&bZkIjq^Ow7QRLg^tv{oT_(0nssg@ua77XPPPfwZ=yQci9iM
zy5C;tT-OO7j;UamAl$-cUJ75ww-h0;P*i{a|5BPM;9a2J{SUuSkveLFgWR8@eqG$H
z-&%}ZrG4_A6rR2I;MZis^b5Av!2S{1dI7?)wZ{>OrwQphuc1Z>^(ly_l(x?aWzvNq
zMZNnA83Ji^#tyAxLh{<@O^5ky+>Zj16_ht8)!vjB*22x7)G6%?=B)VEIkXNV&P-l?
z>gg8K35K=fqm}oBFJmq%7bvn%qae2|^^;aYc^}HwlgiJrY79KKa;_Q*52Nr!Spl%%
zHwWqU)OXiuIQo5`EF$La9hwl&rC!*lSpxPiFZ@g5phR$}kToTDIyXuu9IdSy7qqJD
z<D2ZSS61_CIYA{139X-LoNy2Q5Z^yJ_F^K2^1shLTlb?m2Z>tAt(9$RP+5KHdn%8k
zslD_9V|xznrQw1b^NP`Wny`8Q?q66e!Pnp--U60)(6Z-APz4`gX7L~!+dnZAi=~(z
zOPDcu%Y*!x$aE;3xc)9;zmmnXg>vggJ|z?5+!|Avu*a@hyY4zb*@+@h$9q=~Z6d13
z08c643?(aV<+LL*HY6oAfHi4f5xrt}A_pyP&4~5>12G2OVD@<_@zS$O7jU!XUEPEy
zCk}^vG<Aj=hULvty-F0M%=(|-BQi^59c~;Qfbkl@`VYYXt(uJLEYS}A2L~eXnO)QT
zdR-C&x<OGxZo#jO-0R<SQOgK7vDd_Mi@@Bo&nqYYA{3fgK@l63*u3RH-sHLL4`j}7
z1CD7^54Pq4#il=Jep>@V`Iha>DX1gP@cg(*3XNH;<nvi9h|B}u0-2y6ZfXa#em~iH
ztXTy46F|Df=A#$5jquBknLD>ivx8-{j7yw3`0-a!Bvw2|zDqQmlzNUDdOrD&jqI93
z#EzprPJ+dHl$JBtd|Dwe8mtZe@ZGR7aEUE}Gmqg+<PQ{|Ao~f#t#To`JJrk3D%Y=^
z#_62o%YM={>r`x)(QHX>@c8tWwY~l`lWtdO=UJzLCNg)a=0Q!rZllNfznR~Q+IUI$
zO}JHuRNRKsNIOmr|4k`k3jcVE&^9|A5Ean6QoJb?UadL(u=^iQWfyN&XkA$ZEV$T~
z@-^Er_or1+1NZx}9O1dn$@oWhZ!K#D)b2_=<Ebnzs=eG(6x=xm$Ufq^4*Wh!s_1Qm
zh{%kZA?ZUd){E}aHIRaS3*Q{9#LHt`ELT?97xg>B2mc=cI*R(KQjZ(EXa+hXft-OX
z+@P2eiK@n_o_iH<70GJDAs+7YaB-wKzo-i2EGP%{$1)>_e_J6K!UMI?cBeQ@=H2zt
zd#aQ#*0&p>Z&mycN&<#Q1y3hw#}U|1<y!Z=s?&LyFbg#0?8%EsIJD)y;H+Y6tCkU)
z<R~(jMJ_19xGyxW{;u%W=e4U?d~Y_7@b+c;klLyk<DfNE>r0cZ{5Gn%m~B4YwFy*d
z`fx5!oxF+ccIdO5=8$;lckn$w9N+FKM^l`+s1EYES(Paw3z5_%96uCNW|tod2M}|a
zR;8s5nT=W;9&}Ljz;u|O+%ELpuk*WjoC&b;6-e;+URJKhX8s{);PHUU2TgJO9jigE
zxLYAI@yWeNt4i>SWr93$30-2s+z!+yY}}CHI)i!*TK-dkWgY89G^xv!j}wLJ^*1lz
zfwL7y!sm!+KdaCcUuI2TV*Ln-d=4(A5sC;z4RCzSiibeY=)nEB9P^tTtJ0|*i0)3^
zj^npU$rf}Ce+DPKTjz2$62aMB=B7ODZ}!Jrmgq#l@41_r$8?p!Pb%4`eF7?(GbIw9
zi(v2wq6l3VvH6f#ifDgD1)@e@TL{EZh;wH{<N8Xgjw>O`7(xHX4L|JJjG<!`A(y@K
zytDH%QT!B<V#itFQ3P2A7(_pZ<@KK$d-azG$?RgsGE==rQ?QUMY+ifaO*q*kicimM
z@-qG;z_<?l9cME+_qfmmz&`kI-TBCcvM&f731`jCSd}q+q+LwRi!EMLvcpd9S6GZ=
z$a02*h5vBtV+Eff!8-+EeMjdafgU8LniZb1V&6k~HjD`rMx+-pL2zO$D^%{{0a(Jz
zSfJ6G$oz=Uk8RJy_I7IijvC80I8rwI4Sigw&U^JIcl~;0ay702X4!1a@&r?N$yzOa
zRWs0sm;$nfvODr$V~=UoqSylO|9M2wfe)AG2iWV$-w2?VmI^*znqs?h%kR_uSzu}u
z$sY%Z3R;U&p1}_ohD1v2jgnezT0@O0p@Xtw>Jjj&j50;&d#8Ij4X*0Gz89=W{iHSz
zhYB&yeQ>5N^E#<~*y8hh%cE7^D9r8yA378slFgv~hqTzy79%5h_V_{=}Vo-7mC}
z|I>u=GQy8@CI&$f!4fN@tYv8R*Nu<!UHJR8ZYuvi#xh1$&%G&M;9<eYN>=DGTyqOd
z!WR+BAuqfoTBv0As11>1q1r>o<lViB(@Omveo85EuD#GHFezQWQtIC3RDb(jD#dF?
z=Wo7ncuE+F>=s0B9YjY&_-_aYRGgtu6c4H@Dq3o-$S{ZV)H_Iq0zwyle8cE*2j3s6
znHY>^Dy+&b2T}y3uQqIOy6WMBGRukeQ|^C2ZTX1*t(*@c<g{G!)Z9%mHFH{}5uuk{
z<tMiyfm&q%rE7DL<;J!|fF*u$xo9sJwP}b<*k+g8=h+7J)v<CMj<;tQAN8l2<vBdD
z_Nkg7B=bW+s@_NGqxledrg6HB<xd+1%yT+6igu??KO0AQo133pi^VB0FGB{-MVqlm
zs8dkjWk?P9vo3KT+Y3-e7eN_<`BQG>ks!u_Vg-?aApxS0#CD0GfK_XClk%}ov1ZD;
zVV7?oznqF(O(AtxvrB9Joy}4p7f2i+QdAFfP+Q7gNi!iWUkPi=gqw?ozLb^!3`^66
z$7`?$+}jeRG9+_hMC7m@I*xjYtC8hjtlj;qbLcP87En~A&cvp^PNqUF+*qEQ=YBl5
zLb5>Z{uW;rHyBBEn92BgEM3;olxGcTP$IU&^5J$8>G|H8D<}eO=aw=RyXh>|P_-G>
zV!O1mP{T~{+NO6{agxX~k{Gn}Eh{55S04h{<pC|nrTqU;Cd3qSqKPL3l<@L$sz?iG
zl$+SkK9#}jj^nqKiTkR}5xp&jUK*b-5O+%`1I@>EDNot0&!v)cC%$>*`C%#d8hQ?Q
zEKXuTO&^5$y4aIkXm}i9U$smsiw+$>7?Y^JxAO-}dR%SJ7DcOUJ?@`um#k#ifY_Dv
zn^-QRfF<uUSwJJ@C@)aAXQx)pKHbo{rFfvg5Ve5d0d>d_w6I9`n(yypO5qX&Qns!?
z!#rhz7sqQj+6`-yuraxWex^)kVw6QF$xw<PXsa_Wu5Y{f>Y}&g+toVpGaOG(bMIv8
zEm{7NGw(Vo(?a)yM~W{ww$B&JXSAO4WVssK{P*LDrGt90hHMV49qo!&VPElu%|v(U
z5_qltXRvTI@r^ue!&;{en2!Uk;pb_kYgq;)&t|bd_0ANx9sSzfv`s=0VA6e1S{c0B
zliuU#+8^DcmrU2Fj!@!%4igM~kDklO{q-9`nr)wl9Xy1?TL0C0GFYrWiMh)RV?$qe
zJnMP4!$bd%Ys2{xqozDF#}	J=QFhg#B*|f6$hT(Q<Fh-zG%Mkd2IhC{gz-96=bN
zE<%XUt>^fN;zudsa<VVa$!|d{WT_I+;}^uq*>-2voX|DZ1;l86qLpjfvi%s*SWzRV
z;}ED2>((ZP<(%rD*+jeyz24%SF0O4^r-JNHc9V^L);h;jd{_Pw?OqLTPac(^=JzJv
zrzLE!XEG=#p`E+TM3UZsrkb~lCDm1Bn++;EQ~G{>4vR1}@xO1TPiFX8^_Sn6noN~$
zj7wMG;wjWT)25m?-$lYS1|9Jt+?wS>UOj3t@i%TW7(0%Z?@bR0QMMRpNe@Qw$V8~s
z5oEI#RiWU>SDA0Mu{fyx@Sy1poxJJo&9}cign68rb?i4|euV<bYniB}jsRWafn6sl
zhh({4;@ZAS=Pa>Vs^rsCRGThEqxxF63Yv|ovZZ9^pXK}r91*AA)B!{u$fG;M<m39O
zV-%xvY$2j%d}=HEN^z^sxg4|yAKJ&F?2}dBO48wZB`TT!QV5;)E}F?`YwX;ZCplD3
zHZhBCRbj4K;VlD?FNs*SKAXwz6>R>W?^&B4?FjrCH}e)Uk{vEp!-b(mymhtEGJD99
z+caP;1a>3~9zsTl&>xWl&p7%;y*9~ILw<#LDRr!w@SE(_vRa|p$*3Ho3~FPJp3}4O
z)~;tMUOU*6pO9cx^DUd*&Iz|?+)~fCc(0DFa5W9fF;W2SB?lktzoN!=$syL{P!k;X
zQHt}<>|4XDl6d3soYeG*llrwln0C&&a}&!N#gP;LwGDH!aCM3Iuxjz@CTHccKLNuc
z_?~C|^_Dl)P0(@NnQDG~vOLkv(B$Pq-i=a}j{ds4JYg{6m-jtsmUVk?!IU$io<6I+
z&G?{HW(_cBbsHZIN&OMfy0B01@=$TAON=@68boa>{oS!d+xOA$)8Ky!mcl1cS+v`N
zH?3%b!7T6Pxx}p7xA2ON?Un3RHn-~e_9ivJT9&w?S;DoO%Q78@&Fm*!I!;llGP@7X
z3WdKls?>^thCCISbw3&H@t!e`FI~sEfHK)9aB*_7rioKlpPHtBM6(~0zaJwWwT;8t
z!@|;2gknX}n!=DG3F>t##~_kdDfiof$up=fC4+L~`GrJPf4Eze$@d0L;hYqu`30Xz
zIB}%T{wbviV>s@J30oR(h&KoS{Bt1PoT2yN0sf`G%;ADXfurXfmOUaYTn^R93DwvE
zn7CwV;QDFbRL>=R-*8RqIa~o+LuGP2Siit!>9vLy%FFm5cdu;VOHos|&^%|=uu3qO
z#QWq_sPje5C4g4CX1hF&v{C3&-U0KtZBbW=YrQCGCD2JbvkTVqAY>Phn+smUNo=B}
zmQZ&zIL=z|pQE+X2oW?h{T$?RStS8v{rT+Jz@S%S^R3X2m083s2wVDGQnVXc7qiKc
zF7SXMhn01M;ZQaQ&#_I_@cP;r56VU~M$%1`X|;mkdwXMT7Y{^(q00je`sQr<ofksd
zRj7E#APd5O|4f6H5KjzuUBydMva_}F2Z5lOQD-y=9QO{3$}OLjhebzknwA|9q-fLS
zz=)a~`u;KJ-Q<vZG-&EagQslJ{{MG$Ik}J2O6g*?b=u;k?slr^o59;oW(udamf-wL
zshXXbh;rn^q>_Jwr;NjE94mCNY=JMML!@$zP;*!B*A*^$s$a)I>yJTYq5>Oq(D>=n
zncHK7tJlB}uWnIfB^pz>H;ww-vt}quGj3U(c$}8+pfp0W%f%5;Ng4<j`Q7cF#H6Ad
z{P@AQm|%$Z2Mjq$Z)qeN8aCe*4!6EO{xz3}EuhvrwZ7IQxG%z{)=3F9qz@sY2I-jN
zJc05%tqwUl-{-3^{fQ%21>VG|zob|_EjzvxYC3S1NMTJk0DVZteI&p%OR8G~Yxqm(
zb_omGyn}SjvW6A|1j2B<5ssVNNM|!tr{;J6D~cV}`jX^&jsvgYAGY{jzN%s~9in2U
zvH1Eyl=5jtcJwu*YRup|lXTT6YiT$2{81ehR5BV)l^;#HfnTHmiQHHghHiO)jwrp;
zwI9%juXsY5?kq@r#5*E9BJ3fwBL>ff^^(2H<S6pnJt%P~{y(p(c7@NR4FUZ&mneB>
zVrb4vD~C;qWwJ|9Vg}Njc0fD)dM(izqwjxVunPum#3^o&s9e$?@in4?zcf8q;Ze?>
zCw*A@at+nISkMWN5a!Q$9fe^EC>RI;i|&SVDV|NKXLebkjGuCvs^?PHZt2-rM0Rzq
zev<M3k|%3cIL%m6nWtltW(bZ5rSlYtI<2?SNWD5^NRr-~t@o4Yu;0w%aHk=79uYc^
zso_I~C>G#L>WxbkCq;HE7abKM+?*J!6)2r1;P11HN#jUO3O4u^&z$Le5~Q90bxMR*
z&-qTXd(Qo6r~Ljd=?1ym^!5-=7~46el>$>}QPxA$wkTp<?ED;)+kv>k$`C6UYQrw)
zrkT~8#+A|4f$BcD-8aL2CoxWzPs}1k-f2Db9<&`^U28kP0oFpDYa;Fh??cwG!nUsC
zW+xrW2gXwO38Oc+1}U8Jx(69RLxhj}|3m@*O9IxC%{3C$G3s|PvA!!@lO{9i_7IJZ
zE7UM-EvBore)M(=uNI7qqHT|KBuN%;c36&JvaPFNm@sPOybg-#TB(ojs!TmS55FMY
z!b@m1adC~P<<4|Hs{L?kq}JOPE+zY4LKf^>gv1#^hy5B>+&qxVBENQ+#gVE1Z;W9j
zmbo>q)mchN<ZSOOw^HAJyp=PX`Tw0`lo~SmI(W{bt_}C;M#+bNKt5UvXaa*)oMy1T
z+<X)o1(X~-ten|t^bYhJzax!Cq;28T*@7Bb9E5foS+wW}q3M16{vye+)S!c-#l{LU
z>H}W6%mX2?NygChN_2(T>VN2D1UAM-yHNYV;2XqvXZ#_sHT`OtWL(f8un7dv4cFRn
z1y5Omqddd^MH|sMi)u0_PJVIs5GL<L-l3t`Wvqi>UlLGp)x-v_$=dhyQo`lj_+9^R
z^$;MZ9h0XXXXSpWp48*=q_e#h;N}DN)1q=3VIm|uTc4&|c}Iub7eTzI+@3Hv{%ir9
z$W=chPv~>4hxmpr>&Ci*f62#w;X6@tlkSDUHz&|Q^RRlt#em?+$Z3QWPY7foQjSSB
zmO)@K6*!VDIkJ4a%epOkDo}_tAXVr`Q&lktyjFF_dfDv_X58sB*SD&&+!_VkVS{E&
zs2ks5UVimW{@$ZtO5+aI$DCqxLN?+dnO_MS>rqWxLT_rakgk$!{veKkMERzJC|qd>
z5^1#L-$%~|5@NW07hI1oy$hOZy!a}Z6{BIj>3;zIoGK&5sJ|~f74FGTU`5uz{$pS(
z!cD({tLGdcqk(WW-G>FDT-ypL|7DO_^5JzL#h{>i0QFV6opj6qlXQ&0Fze{pWfu0x
zTPwspIK&?GY1&<MdSyRGSQ2s56QQbI;t92je8Tmz4PM$%mW4y3z>z$jm3h=7d3(rq
zeeHhw|JQZ8g-ZC8`X`TwcCifVjRxz5o^WcMAQC{&5|aySX2;q8E|E7ieJAO*RX6D@
zMa?wx3fiy2RT5<hUFQeLxg$=&)uV}qyQc(FG_itak%e?ux#%9&tV{5(P{G91)#B>=
zgyhn;+<ioP%FFwu(PuLW^XY}1P^6W{+l^<OCio4VcxVf<Khp6vE-e?4^m4862J2`f
z02%)37eAu}^SGfre!twsK-on)XUXm7P{ypcrUB|hP|i7MPCb5L=m!K6A;q%!?AB?z
z+kLyo;CRGdiwVp|7j$3Ub3UpzG!O31Cf>SC@OaWMeYJ@ZB+^?^^3qW8$(fK{=wn?w
zb?6%~64N6SW&72hEKz)yZ>H~e_Wg6a0523)2Ub{={tpTdnO*ZxpHmMlXbp{Bg0xz9
zhYZ3n=g{mL);3mL>`Ie46aMy;jX{aJC`szvYjT%q-m-6(QIG@S!)^IimWnuQ(QU``
z#EI(h3=@^jX$5PfjlYsz?9Fa1@_mN|Us8R9zYMN!Im*uYoAz<I^ul^jTMYJz0hjF}
zIwMr;mJ?}s@xyhO!HSqh?Vvg9N{?cLohiBosh{zGdMW?yggRUkc&p<&jbe*!{wsZ9
zga3Ew3vQDx3%hA%4qo-V%mc}SUOTVxMlWqF<XvIFMIBGD>jkzI+vLL?x7=bW_Slm|
zeR8ss+HRu=Vyix4Tmhion`pNK@kUTsH#`iB%#kE4b2hA<7il!1DirLu3W<&QKq9Tz
z-H0O$WnUEBB1nIy2_5O0C+><ehpt4+HrW3w?~BP(i;4c(E&1%6?6@&SKv_kmrb@^_
zN(@Yd1ywH8gw)YuMWpx67}nCxz+KCq&`lnlITdpN{Xu1lz?`CV6#Zo3U-I-2N060i
zqv_F~=dWq^w<}U;LhzhT6bS~jXqt3d!`TYi$s01N!-@ifRR_tg@_n<i$I%JHFI2U3
z{`pX^f5D6>t;_z<V3$I+@!Lf%bJLOzr5EEoPLJ7ce*V<#x1s$%K7SLs3l;LnL(FYK
zDI9Zv&Mf(aRC7(^5jMXOu;BqP8PV#F1v;`EYOK0N&8zV~RKv2G`4zUsis_*mK!~gs
zjbP^}Qk=#!Y<w~xiEsQ}KQ-T#n7JqSO8*b|s~OVs>?OT?rRUEq4EQ{r4xR_L#@W9q
z6X>|G%I<DG>pBm`Jf7+NDsFGgPwogl&*nTpXLpubCR9uW!T0x|zCS&ZF*W${BNc6B
z*7YL+4n4Jj2*}F^+7O86b8rB%1qK7Gr?JD+n)M(zaSv{}4NZEra7V}kQ;V!lAs1zI
zb^$BikdnXJ=%r~>p}dA&{F_9%jNN|GdQzKHV#krIr1Ft|l?#kd>1N$q?hDk~ErK<W
z`zYG(;ik@@zA>nqP=7MALCoGqcGUKs8fZ4ST13!vW)uG>EhmPBzn?T5MvZNA{DfaP
zuCOl^-}>(I4-C;%kg<rfh*Jgfe6#L%l(>0YloOM3eYu2SGE^GDb>=q;F5HJV#KLIR
zQk3$hLWK%=tR4c`*NWOf+NoKt%fwXxj>}EI*4DQ_qU*?YTOIUqX9c3Z>uoPL=S^No
zoE*YA`(lbP%rH6K2V_Nl0o7{(6=Zuz<~g4YmU`E#V}_a1lHmt?rl}zJ`1%YlH|;+j
zva}WzzC2rAl#xfCh-vZP3oyf{4*f4y-toXc42j2Pa5E;z6~>{%G7l=u6Ab7?G1ip1
z1z8t$K@!yE9tsE$5k7!^^Qp;ALi$d2GP{vGz8D9>DWuz*TuyTN7BOC<YnwXaQ@U>}
z&Usy+b<8>FUO~SwuP_I%Rbr8G?_Vo4%|GGt_dk*jZ9pUYx7D31?hg<hwcx4nla`vO
z>khe3z3pBgQ4jNK`)BwA(f{7wSS=SyK#FV-@qE=n_()Z}_1|9;&iDU!&<~RUya-QZ
zyC7c7mPTRi*{wMyp7iZR{Vs^HYp|?`wQOcJy}zJfh#u$ag3{|OAJH49)*GkkZLK?Q
zGa)C3m~0kglmfjv%?!91#9>=)2+09*m21Y~VH?F!m6JD_0zUyLD4lfB`iQd~kbM#x
zZLt<dD?!5+Q9;K36<tpB4T;cil_D!zr~jvkBl-#z!446~1*}qS5Yt&-o`3cHQf^cT
z31KO^nK#Ak2`x)v9OK*9`!`Lv!Y_2sU^t?^HIqHZe*M=x`1@1t6<o~<<SSd+US=HB
zG28p2Q09^VJ`gSuR%a3gg$xAD%)VlDq>CdOYixmF=4F`g#$WieI&Go?YZKtXcghzP
z4SnE8P9!=s?Kgh^V!$UQnxWt^PzfihUUT#6j|;hE=l3VDffXIkP^0GZzoh~G4~qCw
zX<+y)Qg`L$Z=u$=_Xw8|ccDqPR_X5S(RbcNR=g$P#2qkGS806XpVZ5snN}l9V0;8l
zt@+E*5-)H))qUrHY&F)L>?}E@8^ZYAD9Pi)yayiDu43>@`S|@^aYECt#WNa;euLW`
z5$VLgXL}}(XBx?)c)xoD2tS#Sxx7ongBSDc!)EgA1FUDcb{b1&pmQ+A8QS78eSW1`
z){_7T;445Nv&v<p(+H`fVP^8(Acgr@W+E+w6tQ`@K)ND5ru9loS+je`<Myl||4FX&
z|NVXg$)S2%Y*DHh4^0&gE77y?B=#q|p@KOiyf5V{lB&nRbEDdOcXDavM-yd>E}ia(
zp{gJTG%eVpb=l`<Q3q_<B2$u!DNYe!siN|tz&NGm#F5hUhCOiQ>NX~Pn{8_H0tdo^
zi#N4z2PW(kOXMf?JoVRqTyQIDWdeKWW$58Al{v-<h7?WpstY&@>FDWLO?x-QpHA;}
zyW!#1wa#z!Z5wan%$~KDbw1~&JPV_DSvx4Pe6cr=5ujOJ(P)u`_BO?KX-j?pqC9Cw
z>P6h|6seF5DRTsUlKN4UKa3(soLM@>zF!Sb6_38>b+LW>JJOX7`a6LV!BD)gpit0F
zoRaZdzw3ABx8Su>z<AB<-M0y5Av;D?(r>52(bQK<RnqZt(s-oJAdI?%@}Awm(=n&(
z3cY1*+UD~G<rY}zM$Gd;aEKUzu^JkaKdPtbtSq{Ph=e5imroeY*u&gI$XOn?<LNh%
z419DM2NW8DM2-T`QzDTxkwHb<doZ9<JIM&CAU#c!Mq^@)Mh40S^R~<BTm`i52e9e)
zn)bI+zkN}aV&MvHP-4VtB<%#MZXytqcQ%xmL#>z6^;$O)aQ6>GCKo0E7VJw?Wv`#=
zZ4c`8{T(l{X;a9hS{SbT#~j5wI`$*50$zF2a<~ohorwZt1@z-z*`4<G@_1UWpR{)|
zPc}J8hL?Xg>&jc8tIIpMgEI^MJ$Ou@hnzu+p_cl46JNjek%@yqY$UaABylP@r9}pl
zm(}%je_t(1UbdMa9UYhqyxlyqWeTSkvf<#tzTOM{%NH+8e_r~s>3(PhDt9MAdlN1>
zx!49BQ6l)XcDCm?Y;C$+wo9Ki$ZjOj#v<H3T#&W-2<a7ieqm|dV4|>)+KP3|G%sAO
z-%49K2PR54QfWFr7wjD`^-BwEl{JVZl$7kWf{m(&1Zg#2;7Be322TvZW7?gg2s#bA
zHwntbPE9GEaADVBtrB32#Su}2O%7N2JUs|-<$`~kgK%8nSV_-^G~0cjYA&WUZ}K#$
zlOOxc{%ZY=_}KK8W;oY_E6!uK@?gq9(8UK`A*g|)-qZ;(#dWsm<C>VUKlMP(tWGW$
zFQnK<7R^cv?d(VZTDQa8SofgBW@uIM4n+MM&bv+a^$HNx?}ATz<;Tw#lAa1TFV@}c
z{+J*?b)P4XFcYw|*y;edI~axDM&hTdEL!!kQQ5VHe`50ITdTaIMZA}Ej}(sxV>Ozj
zI3u%)5Yq#H%$u6xVy6a`V9B`a#>Gf|POo|k5NiG?r7Q3PQl(HUH?c0X%V|x^+krq{
zoAD_HgxKp}jr61(bm33CbvsgxH|{!Lf|fe)GZ@x6(-5=75%(ZA8hP)alTA=bn`<OY
ztGsV}Ah%R-Z%_L#!~0;}SMEi=77c&6FUY?iQ7pZHK>Y3J<rQ$_Tb;%jKTuX1`M7UQ
zkwBYaT%^SUL?mQ91`v;YioIi+45i5^>=zm`#44@;+?S5L24f@frl|+f4~0@KOI>#H
z1Y^j+J_(I6+CD?@UdYRSNQpZkT?F&7bht?`vz@-NTMT#R&|Hh++e=<k-Xu%-5hk7i
z3l{sf)rdp7U<uwCqt`2P^7BWwbS8iLKHffXn|mif@cJzLbE!$$h6|HGI?tk*jsV@~
zqfVI~9kJ22Vg77-JwBL&sxp;qK94HUAf!zc>6>%A!HAm3PBOJ8MB!>1veTuny1GvE
zW^07~qn;Rb1lauzQWb3_XEqd5z~EGobQ2Zl970)^Ga9>!`7i!SS1tt?_p?Y2iN3n8
zNVO&aVIjLMGe0iYYjQRi5d;4eqG$qvFG}^|hn3l18rFTMX=MXsig5{_c*tz*PSQ5?
z9F#DwY`CcBROcVGRO(c|0rKQ>|3+hF{y)CA)u-W=e_3m{X6Nn_QJzH(7dn-;(4+1W
z9_szP?dhvNN*p=a%-{qaO*t}u(XDh6QE5N}P*S;0Qqd!b;%Y06_&>^aITE^f4XMRB
z62@<7x7Yr-_BWU|LhC#ex8nc#2IjIuq)0QX+<kKZ)|Ef(h-I?hGUzpW(aFy4C3RW2
z;SyrUF3f`_l<sFL!~=c;k1Z`}vDsC*eiHc&?wyh%_W$DRor7a(yRhHbwr$(CZEMH2
zC$??w*vU?^WAE6uZF47Ip6C7We&?Jz)m1Z9J^jaY^_qKr>$<L9ZRRAZi%;wq^59g(
z8UBDgH@sO4C=w-A7-CdQf;)+|!)-F(>op;_c|Z^qet?n5P1r<fwBW}R)sqRMuNV6w
z441@XF;X^)x4ba+9WAmjnwH+$7=Zp0L@wCX-2r<I1?)jt*8xjc{-<|y2!n^0;7UQ~
z>4k5d1-?;A6_4*u7dq+3<F>b%<>9t}rl@}7$bh0RpclzoaB~?jB>*uspgkwkutRck
z;`42chs7dt#{P&ZRw6UoUF&f*`eKZFJ0pba8qB`;Rp-XM9dtd(WLsmrVvz!31!K*K
z(`*P$w29nnwdoZae!U~#T_Bp!MxD}$pd@3mkQW0nn#zh-hR<lixB?hWtpTHw)dw1Z
zJ{y4oioJkbmJlXF2Pd_IgKkR=*!qHljHnDl^m>&9xQnpg-k4IZ#SZ$E1a6vdADex$
zn;`a*ZEsh-sI&`!Qmuznf`P$(mZ&~FgT9<jmQ%z6c055IY(RXTCJDIm9M|slFosb<
zHuXI_!LAVEC?hDI9QlA(Tr7QHhEYS3h_p2TgmIL3L*xUuA*6{A!>FB|f&0*oe8LEU
zVf)add#&5HxnY~3C{?IH9q|WsY5b|4*KJcHRFN9FU@ncdZENR;zlfLoCxVn7vU&18
zkTVI>zE?LlViHod@mcoFj-ooH@1?RJq@^mN78(REB=}@6L)xwyp?`=75m`}chbRXS
z0oE(bMOcjE!v}@g5xnAvbd^9%xFS>^*s<PgNzg0&8`K3Dq5qOQI+#DuRQe%;%t7Dw
zp+3}ya1q#EOY3i6H~A`Z+ln+CUScxolrg(97l{YX-A)9a=Ps7#E)sEOIa=U_^rAl;
zor???>W)XI$@AlE#2I_2Kzliqvj_s<=kZ{+k27IC*e$ZSFM>;N;~G^8ZS)Qx73h@^
z5*S4~V9Nw%7(Y({NJy*KBaT|9-jlISL2{l)0_h^x^+^GQm57tgaYH{63<ON6S@LcU
zNFCMyr}F{0^H+8%Se|cq^?~aXLL&<C<z`qjKT}Xb<;AJpp@KS@)(+=s7+Ev`cTEtw
zg7{~cZWs5@FeDm$ezhPwrB%I{ORYDvWP%uNheeI-q9Kjf&QSy=Zw)%%v+A!~u_#oz
zT=a>Jfd%dF-@po}s)gd)(YjnNysO9VEb4FG3?AC7m481hyrlj<QYOeu+mE?yN<?_y
zvIthy;zxCQ|H3L*266W<fKSf6vr^#Jn;Uihc41y5+`KSj=_9D$i`~vg-!#_|MQIJH
zI=f}&0uy`=iOF8ZNWbYlwhw|R$t0el#7bZBvMifxPNbig`_tWB(s@Okb;fji7O7&6
z@&1LC8NyW9(KzJr$Ij;j<TItf1HH?K@ZTK@Vv90NVJvlzL4GP^00iN%5lU@9btv{Q
zv<t-oKGtBN(S|7Ld`>~6bq#aVQIU3q9$WCsWlu7g{{V<xlaP_bo|yy(6Fw65ASFsp
zM*Hqd7|J{EiV%<q1CZ@c;2T5p@gCeSNbdQw#BG<huNd~>Uvs8SlFvk{q?_(4+HC!Q
zda;vW7gZ7Vw}%Eo0UEn{d*Hb2;4qFX;8+#OqbKbA9a-X)xvZt8wx>8F0wewxX508d
z8A|)tC_RNBXeSRQ5a{h`ATNZF;7IboX+Ly&TQ4#drU)vNIZ^w_wU3bg%9hV^$Ms&S
zA@s)`Ek_H9u2iez)I5>`f9)GV^�?K!GM8lWG5`4>p4A10V$9Z@kj)+d^W3V;K$h
zZ3`|P)~s-0m2(9Ib9d|em>e2Mv?>a*Wr#9`8zBsNje})~p7z_nz5>s-qFt3PGNj&P
z+H~N4?eEx!%yFGg;I0?O757N{zVH_hqZUO;!iybr&=Y5jh<1_N>4wdL4MzR*9SFok
z`;bJGNtAN{0nQ<v3>4et1}T`BsD~yWE(vc-;qD&v(*w&FOlr2(5wzty2$8E;$8JH-
zsMG|m&`*;iRF?*dZXofsbf>CCakaKpJdib-cv5Z)vu5R2l6HeM%NGS4#%5t}*sD~A
z^JV>+EeZ8x3m{OrEnez=i6Zq?kx@>d3)+0vLgF5PWG`s-Uy8nU214zDFzL1P^7DP6
zp%Zg2i}O$*6U2O3L??eAfU@`N3?KUizh<yIso!pSab=QGgt3hH?r%eOFR4hbQ@53Y
z2Z@ba`UbwrXG%57H5fv`)X6Yv95J8Hidco9-hB*AsHvznc3(G_WAUVuz1#U;-%{U9
z^Mu*}#TA?5{?VNo_)pV<M6}XLsjHr3bp;`#f6|hasg7xDROrY^jSn%Gnod&T`~x9g
zmHl&E^!SRF0tetr$(4SBqq4{h^8S2&Wg|mrQmThXkQ4gStLfcckN@N~8WNvh{DF;e
zL$9eISQM>imqH6eyCdCjH^CD&OX@PLfOjAb@ZI&=mj<nIF4410j~3Qg1wdA_XdPbu
zlKK*qqtmXwSmCowjevCGG~2Bp3Q0E_c`6Wd;$Od{*hOLLV{enK)>`HRBbb42#l|I^
z3B5yJiBZZ%?g4G@lNhLST)fXxK`H&qIu|E&svI@9&obOnDZ;V#8ZF@;=~$9dr-m^F
z=-Y?<h^J(6KZqucUYSn|NIaeBZJ+%{)jLS<CWq4={0f>Hou)24Zm15bb9Ir_d+Wl^
z6=n2Xiu4~}v$0<VA3qB5E~UwnH5md=AF4-@1&>d(gIy_tGxW#rDdF5=?A&#;bsiY(
zLVbu^hJ71Uu^$Z4ZeW#Jh+HM*{KX{!F*$Pnp3N@pvz+spb=2>RrFimf^s$EKz7uBt
zp*qE+3Ek1@2o$+(_KTk4oM}ARBqTZQVdUgaKi3p6amhgXy5i4(IO(cCjkcX@7m;6Q
ze8i8ya6`5ToWt7rNz*^rH5chf3SqTT2Ob!q|B_r4YF^aW^%(usIj9VL!9-33U_r)v
z{Sw7^!N!Gy+#)+4M__f_crH8beQ&Oi-$N2R$=q{WBSpj~O6<(Cqsd_tN+M%gTX5_y
zPusPLd(_ja;>5^cq`wcYVPwWaWtAK<S_%-C!<s5nwK56Em`fxF9+YqWyq#XBu%^y;
z$o}qd3VzM_K`zdwuPdB?j)QswUfy?=U&osTw_}FwyY6oqw$ynhduIJ``9CWPmj3R8
z$|QeDdOY@|GDwepS|@xZ0Ay4gt}ZzR*`a+B!|6zJy8VP&bG`3xpIrVXGQGA8WO$Tq
zP~~EHcS;g=j9HZ2pBVeezDixq+QceyQvEHUYJ{L(79Gx9Nw=^7-NaS^<U`pETf}Yo
z`29(XorVC-;>DPDUC%RluhyzXBR73_3oAe3Z~v-x&3*<@hctz(i^XzNHmTJqceyBX
zZRJ*Yd(WMhv6lO7lQ-VnXkD|V%E@TNJ|6S!9GYP5z369d4U^)Qxp|v;u*u9rvS|Nu
z^j%#Iz6xL9+|N$D6G+eoj3uK|v%YT&qAZI|OnLdN$7`~cOQpFMdyFtAzS*Af|70K!
zNv9*WfG)O%FEvdPpZoY#_WVv(M^5;b#w`K9Kxsu|H`A)T)}R*rfqPK<B;#fyrK0zY
zA+HQSFPvqit>q$77m=nVc}6>G<~9JD8D(&izM+?5a9NbC*h(c2Ftk&R&lxR@8;!X8
zD2<9h<?X@|ef{#Z{$Ol=x5*Kd4e3yXnhuuZ?5Q^(gu)8I!VXxxu*ZPa<6Sms{c`$a
z?+?A==j?7ny4d29b+*&@v2x0K3w+u=M&Z_QOH_;hk|ln@{X8#t4)zjw3_C!6*r`pE
zlC()gu}Xl4OQl)}P_9tK;xkh&s>B@f<{lJ;Dn|_+r!pxr(-AUx`D`x3xA9t;9zEri
zs+2&bS9t3mOr$$bBR6N$ZCqrQP}EW_<B({|m9UFwCd{yBt3>z$<|NMcGrg8@*cV+y
z!XZ@j*$BdL3u@~YKMZ~5_U<(K9sW}uieUG>QuA}P$nKaOFenv(a?KtobplE{DO+9}
z0}sKy*3ME@h;U-{Ih1!lH?*xxoGhO4k$oUZhnu|pa#}!ZhAQk*pZYGm!q56d7ee}+
zCe9uG=gdrGoKG_XGlYRr-4@V_AFU_<MYarR7fcebI~X{>5Y2dY^ip=fOuc_&_T1&D
zzAJH9W~lrDC<Px>iFZog>X50(3WG1>*<Uap+g_F|w$vd{=Is}Y_ZtgwnVX!Gl@ZS3
zqRcQ_YLYIGD@w*1kSQUD&5hPrSnPUq+9~MNM|dDYNA6OW6sJZ`)i~{(G&htH*8X)c
zBsW3D9O~PKp|PD3DzdjYNyr-L>$0S%!wyT4M))HVU}jBgsFr0PLzm%y`u8Cz)nol3
zI&|D5(}VvDG3#=YF@J~n)bECw&ckNdC6I#b)47h%w=9F(w&Rh;waO%N`%<OiN+ByP
zM8L<g``|_58l6O>P_yVH*%hlSq<ZwDs4Pd%&3gHo0L?0u>Vqk|`bQ5HbJY>-qRxJX
zNegTwVA=3&bIR^mLIpwJkOX~}Aqn~D{&Hb<#J7)s33Q<DMk7;deTIr%XO78kKy^Sx
z#Z&?R?h317#XW3gmVNdN>Yeh{>N_^va`WBeD7%B!jF2lurK-5R8vQntz3Jm8A+KV`
z$vh&vq9YOE#($ham1r-tj{B_U$;i`lnL*?OIQ`zlkDtuD>ZBOG=1jd4_rq3_HBxpv
z_Mg8IG|*$Ays-Q+ZMIp1L6ALt)m<s;gpv8@Bvn771H&6(e*Rh}d6VuuX_DojZ=`9Z
z#<y8JMWUq6<s?~tA+;^jEC1G$$N2&Jmh;#9a^tD<gf76%ZbT=fOsutQa<H478+OA8
zV7oYl3#Yh$IX1!I&57j~KLD}cZMghAmmL*rp7auIhriZTxNSAv{3;v^zM+Wm-O^s%
zO+5gwwE;;98MSE_PqSWKC?sTZSrYbl`t&X>Ab%>fPWHsBn^wlnJ(PLgCZ$(-xr#cN
zvGknyQnw}?wM~sLYI}4C$;);QD#5D*qQRP1?wT9L%kqXy+LjV5#%54+eT7FKGmcOt
z0(fUpTE|RMGN^~nyqf&vnAfRid^zW<v~@|pq)cQh+gA^mYW<Y0HRIarI#BoZ-jN8U
zV+4okE-)Y~(wK`CniWW;b>s0@*}BrU0KZtT^aQL1`hczXY!(6zE!I9PS)I@U!4+Er
z+g{8h8l4x$C_my>rEM$YM~&#~n&=U#r)3>mezAXvVXLK-imPjCL9@2CM$#qFd}oY3
z7alB!K~6#c_;;9^4io344At7-WsV<uUPb2JSXKo2yC=~|UKB(ZmV;kl1Z^@4S2B$8
zq<i@Mv|eFLP}%iG-Hw(Nsf5r5B&uj&NG-w&uM$-7Aepn_S&w{iU5H0gVEZCk5Mu*j
z?xR}ug~XsVkgB*9fP`nepQ3r%>?W!P@<XOz$l$qguq#oe1vqO$gBKjj5TRYj?Iy|D
z22ew}lem-Y^4guH$IuH=5~IXK2Z#~(yoevm|ArC@24efY#@$<>?361Cs9&iAW|NBT
zeG>{!3cRBaT@09lJn?mA<=#5+B^L4p(rwM_0GCT2=%hHL?Oh^If9=D-Had5|p9Ru|
ztanpsz&ASY9k=VWPYUG%(xpt0=P$gts2A@$hGd8u|Mw6dB~vO2Ld96%76vaqM3zk;
z-E=k3`mFDFOhR{90Jtt>umf0H3b=^=t}`hen|81tkN_xXN5>k_4K~ya1BmNPps5K+
zzB*2`N}Dg70Upjvj2ltz4C9sI%@$EUIeKnnx6fzMobC8LDke`tOt(8;LFP=$id@W6
zF(SiuWM;#l+Jh*K_<dFFg0;L=3gS*P0$3e7rgB1Su7&6epx(J^jgHVwVnJbnejuDV
zhelf)oJoBziHciJohh88p{mG}5;JC^qPPLH978B-R7xJ`=L!-oj$}CV<Y<(VDosJU
zN0XQIf+!&f+Hr&wOR9OvTqPvq--{xTfZ%H<%h2`b^EU(UeSUSJq(GT34Cf|^bMX?T
zZmPO%D^T1s0Lk$rByg<pVg`TlA1%+MxG3BR2jZ)f45tAFSVDGvX_D8`sddy&JFLXp
zO|Ke_GzbFUc^>Sh_{%%-Vv1NY5T(SCW^((lt)h1^`*RhYsv(nNEL|hyw72nymtX`?
zQdIO5=-(KPabM~DZ2XgzeQ17(cbrBZzAN~sGa)De016>LIT-)<(&bbjHefSyJq?Tk
zXjoz388E7jTCCJ@Su5Dgz{1}G#+Y&+RLDSO--!X^2ma>2p(vm^AW)$`a7NM;Knld8
z-^E6>mkt-8D&-i11(JnCMT|r!gG>lw3GzE&(WD2$5`-N@8c4pt^?OeM#mJ-?kUY=~
zSOyiq_h9*_FZfKz0SloJD#C6tFdMKC>Q58i0E&S0b<a&QSSKJhVWMnrk=510sfn^u
zy6>{e0xE$1-(|%{{GYPQO88G%Wg!u|gIe}W2gyN`gA1X4pB3E2A8XJ8Xd<BwD5L-i
zyaY^0NO>dC4aF0zXGVaRgA!_gx#=%y_6ULRzc=v~lX@?{GKrzk;Jd37O7uiBa}$bu
z5^|7a(kT7md`Lf_B-86GO``)Kicte5X(9-&5-cE!NVaMaITR=96ue|I9RCnQ;2c=!
zitm@+wt%Z4Rfk~EDRA2<TVl=tmqaR7C#;<4JbN1q_ah}t7l4K2@M+lw2_!SSq)0E(
zu2|OynG8%?HGvapqOV<Cu+tU|dKB@Lm=p*wrO1JAzHTH?PQ*y=#k|zsrzVj%tRxLg
zacM9}$e>6#u*e>p^0t)*JQKK>Of16Wl?eSb4@tnLq|tL0<gn1t{=ogv#_Q^8;!tR_
z#C`zNpd{9VKY#};yjK)#zi4D&9|njRw8Vn?<udzF8v+VDA=Ob^TWEw144lZ!NOAxc
zttkNx3+(DY>k+UJv(XwR%!F-g$LnDvC<Z1lI$4FtZs*+6EqS3MV{IgX8f^X>0TJ5g
zurO5!3QSUgWLvLyNl{MtNss3O*!zom0;C%1QaFIml0)Vg-Ro&-zVHHlOApu=*f3IT
z0CP|ZQ*jlm5kLS^9g<Ox5*{~YbaZEIkbLK6p#^Q!{cFMyRfA-V5W_lDh4<hl@+2f|
z;787yjeX^yO}~#AbEAIk5@VJ#%iN<wBQrgg`QcSdF`pRoS&@!S4q<bhQ+$k!>fI7V
z0yluTt)sPQWLgx0iZoLrbOgU(Zx9omGUM=E=uoWU<kjo7)AfVpz`!zJ7-;kedGm(6
zRu4MFBq>(hDJ(8X&fRMBgyyytNuNOx93R~yQ(6lVh0wlvr2^ulTuew(7$EGs;JJw^
z6zTXpYQ^Sl+Rhmi2Lg&j`tJgWk51?g0CipXPpK9Z1M$H8Q>rc~C*>j~bG8gVCre&!
zb2@r47(}1-1t2Lz9X@6!<xHds%jGLHw3x84WbH?{Ly<)zq=!LWzSaT;Oz6-~4L=tI
z<T2%*i+pR0i_FszKT7n)Q{Lz&^~APTsW>rhn&a|2%Nq$nSUo%gk9iRQ%#DK4%mc<^
z(x={*a(8)iPn+(pm&5ND7pMd?jhwHYM0|H_0eTl`h>2$F=H860N2}vOg;K1s7Pj1M
zq0D~~=Qm-`QcsM!x^yjixTK77Thl0Bt6l1E$~cD5ei#*v(?=8WN~h#<|GfCqxWnIV
zFvf|~9MYiAhfo)_sRw)r;Gx+$4~Mkg>$+%X&aCbv+Ukp6BPkt}j8Vk!4HXS!-)5s6
zySs9R?ZBfY^vB>~OHP&pb&nP!0!oeDt~-Y;@G6;mC~)>!x#|H0KB=fqzPeUPk>sf3
zH!_PRVi;SxBWVfv`Ma<TuKQv+s)m>vUPow#acAu-W-+V7aTxg&fKhVVsZ^<!HeD5E
zlj!+_B8tSo1Nr8#DFqvaETK?)1xB0ZSJZ1!0&S10i6uPA2aIyvQ8v9e+8EVC5_5+B
zz@PB!KP-I<s>d1A<NOuwWR+?4da&GoW!0!S=Y|bo$9I{TSOSck<X>_2^Fx~{X%U{I
zI!+Q(litMJG%6xq0BfJ!0Iwz7O70FiB7Lk+4wZXe6NIUTQzy#y`{jQTeM$Qq?iP#l
zw5uwm(rU%AlGXZINRXdPMp9&42L`d!y2Bh;$t+U)>eL@{iE<(=xifgcctmORt;HlO
z`hWm!f&MDZ-N|#BUlZ&ip7MP=E*Iqn-_%^bDWd4fb`@|5<4q2FELnN!+AuqSapk85
z`P<m|5%OI_EG|tqfIHBLY<&9;KB?gEXXF{&DR&kfF3_PpO<1O=jF&9+QnnkizWi_O
z=gw9#+s>8=E6?n*YTtzB*4@AFa9VQp8TT9>d|rZtBGR8{r><%Q;udWp+bSPOCUF93
zq^;%z&-{S4RA6nHBZxrWJY<o(!P_m?aeMrw<@*v9e`bpH0q-yTKi>Y(T8Qd#c0(?U
zOFuf)D%16lQEf*r3tvgjE015GHrf9IV}a)@BPn57z(=Gq*%jp=<FNg@fnZbXe1>B+
zMvLW)UVH3-{4u=3Bb2`*7Yp_2`iFKG_vXWafXW{bh>8*YXkAs0W!kAaLD&K8mBgVZ
z!K?$JW7L%_HojzydzLxdF_If%-K|uEPeZU4nsv_jXn)+|f=c+Q_uVllYsDfT*(sb^
znsx6G;8qRQIbI}CRj$XX6qA#Ti-0j*;h6td$IwiNV<PHu%pWi3g3^_mUdI85CU4Uu
zS{rnP0auf11xbLdG#QK)AIYhoOw1ci2&LoW$Zrd26dxg($LeZ?b>S}MdM2nYbe(qH
zfIp+_k4uYo&D~?+4;NS+JZdjCw{TQ2ML9l4HOZhJN&2BWb_^xk&S54cHy;_?nISK5
zHyxA|Bh#c})wOzsAkXrT?gSX6VvCX&FU63VKWivl=g4oYDdAuK$(1nOs$`K}Z7f6Y
zO~iP8(LyJ4s6l9m;=Y@{15K9jf=($L+AMrtP;alboP^+LasCMF#=U2~gd?RFP=|)=
z9{!8r0`k~YAY}EQj$muvZn7WYQ$QoOf=6ar)IZxkd}=BQrA+J#AK)x~J$>wblM}f$
zV&?BAu({b5k(hUGUpCZJh8NqNy_YzyDwp2hpL`-rfvj2y{xh1rKy^wcRz?V<l<1{F
zN<cs13!m`C-v=cBL>q)AFz-Z0!#%X=2+F}fA&93HkyVcKGjR#-_u)h#2#)DueWG35
zg8_outx&VKE0I7|J)mA<u<Vp5+J084BoymzjUQ3x7c1Fx66btwA-7us|00|1diqCp
zdmx0WhBIE~FC0<9<cC;5j7o@g!wOYPd<7Jw5fJh7sB4&Z3tyhN%PUbD{z`=!dn+gg
zM-9QNJ`LXliPGsjTXJ;DDLblvh$FDbh{<D44Gs}K-o`>y0)YA3+4*CSlGc=<P@L$W
zvWeP$Uh`6VE}!;j7}Uuh+YPMoH2d-`n{HSpkgnbzoG*_FeoX*3X`h`N<b8|p6E4@G
z-V7zWM6}H5=p>YM=Jzl%8Mf7FG1kR;%4wj>BHV^n^iL&Szooi|fhG0XMt4G^L(M#C
zu>Yry3TUzXtDR(!F9l0smJ;D1bv<3Gs!y!*EJo;$%LS+xY*pKFoM*@daiHgOrm43`
z38zH3nSKT;lqypCefUUT2vyMr2_xR5%u`oQ2=+z@S%@S<B^5+)2y&hp<;*bIB32(?
z_8zrBI(Kh)!s*8M3fd)|{Lwm~A?9jcI`6w)0SM2JtX)^&NfxF98*v%an#b&@k*D`g
zN6Pai9<%6PsepT9gn0?2dPQotO-^sNMMbG<r}O=P2$syi3VxZl%*n5AGT2-EQR=ap
zuBH{q#3at;Hm<m%{npp$nW(l29sKQo&8F>>dX89f#K}I$dyFzZ2>}I$gZvLs_)m|R
z6HkBSKyMjI8dDGm?jthk-dH)bcpF!bb#*$|pR9Op{Ox4oWq-f(jBICOo$$~$e+#iE
zE;@%if8%9m96FfC6;D_=QztMNS$zHI@?d*F?@cE^jG`|%Sd^J}P>>I4V4FoMx;CsC
zp$j|PXj3Vk+=w5wQ2@+2YXjfZ!nyf2Ea;G3UgFzrM{_bCJSXyuuO?||&s+1(h<LJl
z9WHrTM<vlRzW64tsvyz+DbG?0o_sgyd9lQcNO-PaGXH_Q*-F}c{SV+Ul5R+gobKnt
zw0B%mAyUbI1z(*3X!RFfA2fqMD2o`sLR_yGzJ_C>KPG;14^yz5JOOXIpG^+0Pd`eQ
zpu<j*f)P_BkHyYe6%PA6G*tO&@Wd0auBiwl`8??C&%?_IQbqm39!;NC2gU+)+P%uP
zUQ)MAM;*#2bEJO{SAQVb1ZcERzW1Q@Ym+n%k~=F%obLXIqt0fIyrnkyDm|_uXWT9F
zzT-@f^VkdB*C<Z@p4A>%V{>X4%n_UI_XCpS*axJyvJw_PPxg-8A@(waBNGTlGl=)f
zzy3?Rq<JI<sF<U)OimxiF>nNH<kX@s$%ec4Pb1DWxw|W#UL~R9Xj?gO%!g>8U{0?2
zjpG%A;QEOSsNj}#S)5nc!#DIf%$kto>If<BnoeSWyZfy|@n_E;n^PyD{-ZTGb5;7T
z_ov^0P-TZ>Vdh+J%2LGFQ$ez~tWfu-KGcJc-CYsT8hEOJ@~uy{T7<D-1$>WyTD%DB
z-h8QF_eru|+WIT!E7V?R9cNYIQ(26`Wo>}rH6xQAUqGRvl7zOCJi}h2X$hqB8;&R8
z0ud)pc11YI52w<F2UA95??4x0ZQgUebRQ(IzEt5-Tr-eka1Bu|8%8bst9S2$?(d(|
zMTj6km1%`Mm?gpXO$h|Ya%(wR+tad*(?AX5<*YW2c#619h;<+-p3d>1jE(*c<~Kza
zr4mJV)<M(Ey6ZLqvjpxNF};E}lBC6s27S1a%s_BYI`96Sz@a%(b<nWgXQ=3yhJY$f
z{azS;ebufWlG0g()N;6CVm5mxhlUfd(8dCsSyvWb3DbeXjR{K?_>l8=v!p5umth~e
zf4~g-tR~DX2lJHM2vT8`yh9^e>c9UTTaE3__UpWeViuVL4v|bAxV#e{SzWsHEb-4f
zqDKR@D*cXJg&%pBei^q<iuyH(e)L;Qg0Dm)@3z|h&aO6*AelJ4VE9+=^jNIL0PI7n
z+m-=9#4PRx502k%au!v(XUoAGHUW?c0}6)~#3twrCaI~IgUBy|7=tM<htfWG)s7iJ
zg)9B1QS=wi)xMp{%xC(i^GNQ4Y-_%umaGc1rQeY=g!X-V!*=9sus_@IG^{-@0<zV5
z!a~(_M3DYtG<UzGv!%Q-ZbJ+|0L=)|SX&)^I%4)i)@n5--CC%LsIrtgOGN#QrjOR$
zt0?A#Iqqp#!K>`HzQ3*LP~~cD;xcPDlC9r6;>>n+>Y2<VsE#DwUTloHe*#NDAko#j
z>6wC**FT9Z#VEmyxB*bx`Jy}g+TpW#uljFV6SKB~FEu4S-8{4y0rT#cfFhmmyW6Ia
z{#(H%wPnpUqYM(89S242*!8ubE<tn8MYYJ_gFonf3wb3=V!L&^VnOrm4pmT$6c+<J
z!|;&p;KDv8m;`<rPX4>ha$zt$k!6l7o*U3I%5|>Jbev8k>2BKdg-~zk`;=c=-&oO{
zTSziIc2sKzuv<qb@4*7!SLfWuKOy8!_4MWx8D5!Vx57=&<V@%#lCp&pX*n76nVreq
zQ;Ui#lbd~m1cPOs?CV9!k?P@q8X#r}_zg4xJP6Jl4O<XKmOZ`a&QOl&SHvUAb44-^
z!Q9c+VGJWf?La-`-NaDjEu(we_05?tyCPZ-q%<&q#gJzs0#h4;V!Fc1ozZvh5zyDF
zvU%kPw1GPqKiNds&o&BK9vucu+1IoPwvMg^&m>Z?$;R~OkODZ=C#}<K`>b^a%+xcy
z3<Ne!D(bjP<45-jHO5qNJ?C#dd8GY}8vzpqf0D!c(0x+fp>L@h|8Z?dr&}NWh%f%R
z`P>75D07hg9z?nuJB3oDT_6ZeW|hBdd_kcJJ8^BPN`GNwM6JLzWZ@M5viFht%zO03
z)!#T?WeBM7;BC#tmDc=xL#c?^i;QJ*5Z06=y$63@$&m984K5XL8+R{Wuw*|#uF;~7
z;&vP9F#Tz=j%^~j+-MmAKPldf^{pwlD`f%dIHF5R>V+E7J#DB#sL+(&H&z5OQ+ocx
zkSOZYW-pQdNXL32t0ZQLRxt*vp>24iP1oXTa?f(Z^j{Xp?!WZPJ5aMvxgSO;AeQhm
z1wm`@lFN1Tk!a-#$>_6H1cnnAnu!OpbH3eD@zRGx00!EfW>%_96#*SrQ#8QfhI4yA
z63mJIUy5v^D_t}NBH@sPo1M$EQF%DH6M3p$@Do|?M8#%7M3OlqF_IQDd?7C<hbHqE
z&!eb@*MYPVKq92yI0OPJ>CFQ7EQX9=A_Ki#!)u~LI<i5iGO{^?0$iWAm>WX^2&?|6
z;g=i}_sAnQ)jLtp@+bWUIOe!ILE(D?1m@x>y&ohf?yK%kucpB7pY%LDI=?1K?uu3A
zmizjXe?gve^M-dmi1>unAgk6`S5z(prK<M+7%!rP>64OvesR?fCCXu!^t=yNGi@`P
zx>0$R%&NFkb>mhOGL=J7Gs`CJN*C4VQ>q~V*D&_tPyO-0hHo|tpvS3X`%783Dazlq
zO*zWcfIgMLC%Yx*5q7j7{G>KtZ)bSta0Y=Rr(I}cHirZBS1t+SlUWA2@>`}`5hx|W
z3jhw;pq$4xU%m6hm^=*DayKdi8Qnn8$}np@@Tj<9iWVwSoR?!zDP5MEIn-W=0*g?M
zUc6U@V9;H!1skdW^x9Le%J0#*$Ak=}E4$bI<Gmv^(>9|*YigFeV&W4*bxvM{Ckw`Q
z3uEtCeq5bP*t3l!=Gfgm+C1h1JoAQKFrV)R7vL@OZs`|XDrjIDQCzf%o%>BV{`KIk
zN6R_J&`UbWutS}@=X5T@yp5H7<0*f(la(Uf;E1b#$Co%U4HO&5X%==ws0qk%IeNjU
zG8E6j=0X2;<t^@6>yT<{3K4tYa|BoNb=NW+uSs)bS9LrO9t6gN==XG|C`v}H@p}AI
zjZ%NO@%KSwzmP&N<xoO=N(UGPwHl~0^@X$b*-X3up>~eT20;r{K$qTi)k52Ui$<xB
z?_7ZI9%c><w&Na;Lx)_Rm&r2fVdd{nF^Ygk8F6mIr{Sd8_kV&>fm<mI-vb@JH3MZe
zcqVm)y<fOF=j<qRM~Twzj}qU(Y`%W9HdlyWqx7pXijhuY@}>%Zi39f_k)wz>lpJW<
zmPk>w=Y%5GxhrtKne*JU|L9puDW<0AtpG5x3YfF>C{T}fq}cT%+@1LS#Hl;^qe3Yv
z7NWXr*%KbRxJAoCxkBXO^Gk6sZ<7o#3C<fW7yTJzEYnU$y6E1%a$*eoEKmu$Q4ouC
zp*z*=Abp~;L*PAxpqqd>vF7|uq$NUPCU|^DsZ5Q%(<?^D4if^`rMttjd8MG&%Yf1c
zrh;{qppTlDKx5Ggbz8j<wxw?;ryt^LzcGh8olI}#f12eR4|*^78H6@Rkh+GgF6mLg
z=@I1-1xmCjYvUH<WW*^RtYve9rgT3AWI>8mAyKkvZvX#{^M8GFT0m6drE{}Cb3O$*
zCaL}x1@?T9^s70a`+077{A2B$)o1M0dg7Xgat-^`y)(<@ts?m5rh5uxO6iyIzoz<`
z?^~7dC+h4IK&0KHl9gqjM3D8AyASR=Q~bYNb!Dp|k$WiNh|!<U#jZLhw*-C{H2JAh
z#ltZzDe=e$1YJunuEBGNW74uvyqD>S>3p)^SUP>0bK1f`^rYROM?H3g!aA6=Rs23J
z)9Uq=F2W@%miAk0YLtqfPmnP`J<jlXO%E|LTk>&=A9?L+;=YWs%bzEB?OC5>3Nijs
zK~)<Eh618tU1yFxE>A>&#Dj&$2Gcdzh4HA*GNlyJ%$yS8H2f7z?^*OL`*u>wi*STa
z9l-XFseZ!5*Ya(KJWj;RGI%R$#v~_PBW?Mb!0}2SQ*Y`w&Qa4|PExqfuAS`VXl=6$
zm-j;W%+@bRSY0z=SL%w|-st|ykQ#oNX<+@=L$@!SS!Lt@5~h6+={eAVKV4*Ke|3$X
zdsyU%Wo6-BV8&L?+g3xqwJFmyT>*6S<>7Qt7WcVQMji{7uDd+o!wD6^^;)|R`&~#S
ztz7aC^7Cpx{a<vd$9uyby}_+`F@N_EEhZzA1mcU*Uit_FF9e$|+@$m&Jm3V)D^;&3
z>ZWikA}y8UJ1*jb<FM|!FoC+XP|<agJQJq%9%z*va&|)q(2BV%*yie56LNIpYj%<)
zS%^7$!$`$k9`nAm{vBHaFDZ-q)MyzSAg4U``f*ola^9(D1zrU2s76u;^pDfCI)oe8
zKPC-ssmR*P^y+m-O7@qs=XPFLvkAt)S^%F8o#)RmrBF}<+eS#ZGHL-_5US_jIQK>N
z4qyA*7@H~peBY3G&6XMQoKD{0*Ke>D4QLiH++g>$xZ1x3TA^2zd<E*#FiM;NckF?i
z^5}*d63>cENJH45M01?!6=zE@^EblMAx-NuJdz6B_Zdr%Og1pC(O+akcUpV*X~hk6
zI918eHExFVrDl~Y3STk3U@qxGEHeX#O9Tn8@;?j!EMZ?c@C^EsPuA4imJ31O>SQuf
zOZ4z=W_J1li9Vx9+<)-cY)@W`oR%?y^d>kxg8FvcMeI;}mkC&;0lW-dcjZv)HT5O%
zc3>r}o>XA=h*m7yDT?Dq^K)P*6Ld@uu9~_3ZI<tUb18#AzqVQ4O>_@n$ryRShf^zh
zq$Wd>YAGctVSp(u>!>(_>RZd)>xXP-(fCKjG?rkQ^@93i<PMKZXpb`vrh1-bT1xx;
zbFFZ{Gvw$SP9ZL0pr{<{+7bmmbs}<NX;=Ft&;9onC=!gE{ASOTC--z5wC}H585wqX
z-#VfWwKp!KaqCl$7yxhp+=^FxCX8LEx+?|sV~4d`eYNrG;d$&mX8SoGO=|V|`du}P
z5WCW38@N6JoNNTS^DuZ56`0e->{EQBbQAk@zMB4TT5TgsDs*0f$7We{D4b_1$b?yt
z2l<B8(*uJ3mC(Ph&NT~CIPXLI4{iq`c{1HU7eeUNLe}Si01eTi{`hi*0Xdn%PQ$r^
zzlzQAQMD{JQ+$b{V!jYEdR3*XJ;tqL==GCZhEm{t^!&Ymw_V^ja7V}9_qr@k@(gIi
zEmBJ??|!X$heA4j@*?#F6R+&?lXHOey*2w2M-49lXMsC?Q>zhznXak!9*7|nQp<`6
z<ySpnZndR3U|;kLbrE}x^-G&WcHt5}Ht8WvrJD1c1Sj=`1ry*Pz2a$?Ie&9++ju9u
zWdHDd?J@DyYw~BrWDOy&Gk=<u67cBC<CfXqtY7FrEx<%BNjRRU+eCNF|F2qFK$i$K
zm-_&Z&l$PjWkGkxQ_Z+^cWJ!?5%Gm!%HIQ4EeP-$L0u1i`_4ufIG6Fjv>ZIm_#y)q
z>s-ELxiH_@iYW774eM*x%eWiZGJc(7PPSN8BT^u|zL7p;A670VDF)I;GY~zn_h)~o
zb*WovJA4bj0@*!Xv*%Og+TEU)GpY9Dkz55i6avS1%da|4;t2MtVKwUmW>^r0G;xSH
z7NAn>R{r(|EJmvlW%B=$KOq<Xkw5jBrZlsMR2;R4#$h^}(SXrl%KhwbGGU`q_y3TX
zDQH30TCh;D!NyNoFcjJ%H;%AL)Lr!J#U-vj2Njq2l<wOHoBac?2XXno<)#Z2IhgCO
z4z963H!{RF-vA+6#!)a9t0{NfWI46)*sdF{dqj*QIq@G&z1=b24lRsuYQFwM$j9d}
zD)t>F&QK|d(TJdmU774)Ae8+7x#2_vkt;(ZJ`iRFPh3dWr&gL7AvpcULi`hz=Q4LB
z!7D_O1{0Y0x``5m;R9t321xI+KCci1HQ^cnG%W|^ehsZ@5r^~tL9l%l@~yC^Bj6@P
zr#->~XMbO-nFv<raWg0Crph{aF(VdTHaFegIHRHXRpkIv{mJNdnI<)#IK_!tt>z0x
zFq+{q<njHH43@K^iDH2F3p}uC*eA@PyfBqowlu}xk#ai#RQR|G5ibxR?N#)tEYi+b
z;i&eFjY%Z5Z>YVwn@{@}YVY*^8*1H*4}q7eZvTPWZ>IefC1lWX+-L3)cA$M{-p<Na
zhR|4+6=wII1W@IiF&aDjL7JR~a$=EWd=sx6r9_W`JA|ZVW`!j*;>VI(s0D!OQ5F7C
zdxJsoO8>2alFL0-WLRa5{uq8(18ql$UP}T>G->dj(tt+^4qjidLk+%}yo+7cFF%C6
zHEI3*h>5(SPn8<D87cj)IB7pR_nF)I#Vh6^y~1%U*8I`25IE6ei6SFXlnr$BPN)cS
zZ@l=_VtsgTa_j4L%;CHE5f2!s?5nh^x87OUJ6L$R0w}gplDfs#d}@#bYU|-x@_6=j
z7d(&4`a*F%nu4wgAKXOoZ}s8wZ9U6>vb9JT)P}GHL`c3ozBB2F0d8MQk*ny|KJ|0H
zAF*)(1GnegQx`9B^+Y&3{;kOtYrLOp`qMl%yi1RM@N|g@h2E+??EpG#n!)Q~X8fH^
z7_aP=^oUK!R`Mrgs$K5T$Vdg2`h5SSiC>5){IDbZ&O|TO!yiVQ@Dk3EBa;Dl50hIS
z$#>fA7!O7~kv1&Fu*-{XfQvQ`0x0K&xCKiPg20m564-Kc_|+&^Vky3galeFz`hPTB
zB*P&3rL@bm%`mXWjoMVf{zq;7i<rf}Fq`UW!Z5!Wr`j@fF}PczCS<*(?}Gfu5++?)
zJ>itiwm(F4+J<6F{K<dX7NTF(fn4ghg*}Wgm<jQ>fJaId{^8+k^&R8rS7P;L&zxzQ
zd!Y!HeYe7@I*6D^7s6D!p&>90Af`Mqz;uB&usH*-IG}H^1+J?c8dWVxFKrC?l<0Hm
zw;!Qeg}=YoklwPE(DN38&@t@E*_X?HPA5o4FrOm91ik3jDM-7i0O*$f`G!X)>W#k{
z5}l4c3V~x5(-kF(fdi6CFxkNZzgye~5+0hn!mG)Cr?Q>`c_1E?W=gyh0D}-^Pv{Ci
z375ts4o7kmeDZK@wM0KGg9PI^p_4MJb{C7@BCrJ2ey|R5*H5|LI0Qk72nPn^)(R0#
zFFFbY7;xUHUw24nr!n4kUBH%68y?)>(wyTn4CGlgjP6rs!!3JDq7@KxAQgUT%}s2T
z>9~4RnU;(?3)Gw=2_fMK2&80Z)>juw1W#D)@vFKXS>~#N*j$lSYF=0qxcsRDbKvC6
z)(B`;gPuDBdI$mDEYpE8LSRzFnHZ=n0lE%&X8R3aS5?wmD2L0!QJoe<W$H{tSnLa-
zqVVrYC7p$ndTdx%z*`|Ks`w^&C@!k>&J|(bD$#>eacK{+@&U9QYiMA=Kj<IS+tjef
zO6II5gxb(Xx-t$`eo&>$V6upmVu=!?OT}u1s^vhzQ~l&&VG4q+D8FqB7OA{D7oqW{
z<_=Y}c?SzsBZ}wuyk-1JBl-tFAdjiw9Zdyz0{BQe*h`BTeFy(052AI~jToIP#Yz+c
zO{FC5PDu*<APgW<`7;R@rovs)1z{96RIS4FloL(GZjJeSUkKC6vTQ6(b}G&#F1pbs
zJ_7oB8qOs?=*|qjreG(-_e2DX6au1P!&Ao$>TC;(lD)Zw5GF!RC?j=Tf}{?Dyu?Z+
z9*<ddDeYtGxXKU<CmD;OvBl7p^H|b9Rg2E2CabT4N$Kz=z|0^mKhuRwfcY_+Cg;}Q
z<Mabg#{k|w*H}OfsjUxB*y+h8z?O(EQkkWTDS;`}Qe7F;^DgNyroh^aoG_tDbg!sc
z1I-_-w)BT*Lq8gvMw)Hmi-i<%a>HtrXi{X!|8RlSF<A!Y)E%{?Cz%3sfobq`z@$%^
z0^<Q-sFFxEgp~^sx}@9LuNVz-4VPZ?<S@HN;6p+5O)9kbl0jmS6_zeb_pCpiPm0D4
zWwbHOmEHZMKeh3-6rqdM3+%|s!6H?`H7Ye3YBeg!WO_UuAT{G?Kf)~GgpsQ)?<eGZ
z^y^8Gl<~BsW!aJ`3__C94f-M!X2EE)BIN*nLE59Xzsff}u)yvxAVmXuHm!C?)u!X&
ze!Qh<HW$WyGPiubAXq>aZ`1;6b$fOK<b&mCsvkA9^hbBL!_em>}`3pWW8_^`V1
zjTqgbXI7YGaGCC6(jLG8ZATN(mM&6Mg5i4Jgd){h8y!-<tM$6ejY2vS4}6+_btC|s
zhy=!L#=wr1ZoRz<>!#Ugvqk;<_%uo>CO7j7o=-yb-Y8!ks8B~r?oEWzC~mf^0k!Dm
z5d*MsE6x`&g_YF%wj5fVuvE=r&8#RGcYNa0XPz<MSlGW6ZN_eaV$q{N0$5bk=)?2!
z>%)#r-5NU^>XI!?QD7&kY7$e>?#BT1g(evqn9i?zX9dtXoX{1!az1#m1d=PxafXqV
z3L>DXLC8ZtpmdDKd2CW}gc~K%o=}$R^|hF2hiQeOCWuL}l0dt3kGPP^6V)H-)t6Un
zl$|gQ;FRgvr_O`MLsZ4kQ0|na7Rn0<A1FY7igY1ugWYd21VFbmzsh>Z(xCz_mNs#A
z!cFbje|E%i&b-3hA6n~YAVBGw#oH{J6w=5|!a23U@e$Wzs6+_<;5QCWp5%MuW2HYc
zhIfT0{fOo=I74c)j8p3v6U$U+FhMdGt^CtnSck~;SM+;;hTC|t&t2<nb=JXN)0&0D
zP#Wg3N}D%^p@EdMb~}(4vEChkDQbbQ_78RYjo48Cl4gZQCHpM-wqVNW17gS8gcVpa
zQBtisYc+k?%1m=pK+?lsx+D4^{uv!zDd)cQ_UwgjPuJMwON+TuXMZQRnY}K*1g30q
z4T-%(oZhNiYltU%;=b9XU}~aPdgQRTahXDu{SFq+FluC+d`xF~g%AROV(HRHA~-Iq
zcBS*_xl*W)245W-qZsiCy)KT^vm7if=ubq`4j1EIeLYub+Uv_AC|co=g;Ftq2Ha%r
zuA|nn+b%y1$ZfKmFE2oZ`U+{g;)#1S=Ri+rB-d;<_e0r!@s(xGopLX}w%)xls^y~x
z(YcIvncD@lwcotlLo*F9bx?)#Y-73ED)8%j+$G95hu=^?DOt0ifNz@a=FPn(F{A_+
z$*Lc(8_QJ$?E0WuEV7m(AN{>Fz#y=Z)I<4MfTfZ;LTkc*8Ejs?bupa5T7_pMpSS#7
zn&jYxteDZ`S^q)$tYFD+E;btR3rib7^m?CurD;)7elK1*T8Ibm`svc1jncZ5MsDKt
zQ^gWSq^MFanSh6--9hP(EO|+<0T`+oBt=AgP2W@_V=Vlv+P%jmnFOT()NQpL{wDb`
zwP)GQ`A!_q&&d~-Xc<UvQI2qp!h(%GiT9yVkJy4izkxSmV#A4x4nZKyG8UAmOstws
zjtNC?hwvvvJu4OfW1BuA!O$>XZD>?W{UKSAo-DrkMzhZD;)#j2dSSDU%5F7@LUv!T
z4t+-;Sj~i=zUC#Z?SSR)-zO+7TkE?aq+bOqHzB5?3^|7NkzgV>4>~@fa!-Hnq8>vi
zUvATD5m=yktGmh(SKG@;(B<-t%FgA*(S>}CBMFyCs5D9d;u!~+WRYiN+!~KiR?<h&
z!ug(~AMYITN(WS?bSw=`+s(hNR(BpslHWzQz*Kcmm__c}cXBQV`fpK~L8xav+pz1=
zhTKu(60=u1&51k#)jM5`ZPBqoYH|5d_gi^`%1A=?M6E~%Pn=r|#uDYA?#vtY+eogq
z#~FMi9=7oSV}ww)3mrYzY|*@RV>x`yhpkGVRISO}AQS~-S~{7M=rv*eE&l482$mE2
zs<uamp;SDfE$2CPy3;U+tOG8@`5%EB!;da#^J<bhy<jlcAJCHXb^5;XQL8(*aLOd2
zDRxnUWg!f;3sf3o{cM@Hnaap5zpL(@yXfTHEIe@m#tj8+S0ryT7tu0poqgOW*Ox>M
zZC-jZ$EspnbIg-S?6-f$_;7$V*&A~L9_kQ5I$O(tf$MGF*VtExzPEx+`-D0XM(N|n
zJ!=YfD%bqU-3uqXQ2VD;W(ibrRSC}^_Uj#7!#7meW~WnG>ucT$GPlD@J(`p!G3#kj
zhhu=+quVAHSKNKoYo3#&Nw<^KPkvJ8S5vsac*c0YM#V~wc?)xh9yxT7WZZ>mH0N@m
zZqsT7wZ>@9kp+0#`y|S`QFZ<6S=k5JiF2Dqj3@PPT)H?n%y77ay+1L0p~TO5iPP$R
zoA8$GhL0y4WuQjAUvJ%uC*1wJ{~Vd*_Y;6UgUZ2VT@*M)(vI^76r;rNwV<<{qDYs!
z^=n~t-ZD+3?C-${<fpvp;P`s;>QBi$bwZRtQPV!_uQE=uD_TdI5)pA7O|Iiqm4<E@
zuy#}bymeLT(40wRaKo})$nDA!oxtItjqYkej_~OWPG)4EV7|lV4>eBpvCLh+6)(W%
zwa_lwFC5~`Kr2giRXST0VXI_C=A)KhV%Yk))1=`7K;st?KTyDHI+sMFn6s35^C>I_
zhz@dAcM|vpU*3dZOHjpeOgPCSD!_?{?EI_a`S6&P!bMF^lQ2a^^9xYr<6ZDbPa!HT
z%3e8@TK&H-Y{aO!&yRji{uk!V8+icYRK`s!ZuTmOZeIlTqGDpDwx^iC`Qc3wapI%d
zB)Qn8`JH4#Z(xM<#wor(umj}CMQ;^;<T#c27O;$Yku^~u0)60uG=i1PRt8a7hX#h9
z(9lJ5IMWy`QW$9e3<x6v7{!ViWN;tg^mPTxg2&g&fS*=s2!u-7FpD>PfEX~DndG`a
z+AxY4UW`7@vamJLUz+S`vfJnpnKJ%?a;!hyJEq8*J^2Z*#>IhddwPOb4EDJbCAgPk
zc|%k4>zwkcoLfV#Dz+Jz<7OwCrC(h}=ZPAFP9;L6XiPeit8b!>xBc@5Pk?ffWA@^}
zF#?g-Yg^hRd?P}1^TSCyMip=phoOp5=7&L#Ho%tEb4{rk)l9<(ht=i$6=Bk}n7U4X
zim-TfzAL|+Bah_U<=o-7Eis8v^F3YUmq3kyBh92jx6N}+M_oDj>@(*d(BoP~7I;qJ
zirp3G9GVd<6xVY&bH@X*+I(qT{3>wuF<>M)*?fjdcWu<R?B*{ooN)kh>iI(!?C;4Q
zyUo$jM*U4_<Z+d*tqYE8-R<U{pYBzwZ@=FPIt)23*>%VWuX1J_3JhH?KceQh8N%NN
z=cbBrI~s<M5+Hh_ct>#txEa_Otkn%|w%!qtZneb77;dm^NYANR=NG;v94SaF_h(`U
zQtP9>=v=-Z@w@hT-}VUDK`2uUci2a`A97~Kz`lC$T2KEv4}PdKhx|*u$v=YsApNE)
zfl{R3sjO@5mr{4?v2vWcQdrnLFEz37QtQ{2ToRWf`YWH?&V0&Vzd7uyYV~i3O!<~A
z+W|+{&WFx3f!o#f=g)jAcP?}+UT35wv<AtvcPLCpq+a2jO<^FwdXf*MaYuW6>u>Mj
zG|?%ANayc+9uMHNGY=Tz$S`_IeR2zBxI1Uj7j`Z``f_pj8gFgx@J<|k?Hl6VtognD
zH=p{kzkEe|h<T|e4a@;zu7jY?_!)YApxd5;EaHl6T)F5<+=f$_X-7Lw_v#KhDZe3C
zD8_o+_Gihq*M>&`z4Oe%8$gdoLMi+FFT$*04qkIW#GzTDY-(>yZXpRY(iTI}5b<a`
zqdKdUh>o(rPBp1&N#j2<B50cV#t;uMnEL$>FupFFbfp)zf@yX{(LcI_tCeW>&$_b*
zR`*qS9Q@VDoh+<X(R{Z-+wisN5ZXNYHt|$A6mHqW<MGP?=Q?OKUV0d90t2PV<KFBB
z1XzgQn*-x`)|t;C&j}WY{ssb-=nEq%)Wx1Vo;~WMTa2O72rF3&|M98)vk0ue<vjg<
z;^TO5WB&e46A$v~YAd)woVyX2C%&jY8LWX#Fuq6O1Bb^&0RlE&2YNjDTm!7s>|f__
zZ7@T-4HtmX^@Vuw25{TJylI)tRdtb^6*5mPFuSfy-VSw3f7@$^BK95MUu}GIf8*er
zW)=xf1<|9P*;5m`oIuJqMmgBQ_o&F#!PYHXv14nvMNGa330Y+ZgC*3Mw3@?dr<pqp
zC$FlEQl2@GQZa7}F&+@tOOng`TwQo4H8v3D|54OeM#T{{YqJnM!6o?O4#BgqXmAY@
z+=9Ei4H}%qH+TXBcS3Lou0eylZU_W-{dnJdzjNouOrP#EbLR9+S3TABR27{E&=z83
z^4ciJB$a(i6_0+o*}@fntJ+jfw|KIXkP!AKF*l{buf`}UDgM{Wi{Sj?eR1ZPlX`R3
zPcly0=oSm-q^!VdHZ<sWx#Cmtng3Wj?-KHzs=s9@*giJMBh*zv9*NM#-GTI@BQmpm
zjR%=jH2Pbsu<w$OE{fEN;(O`hvS&mT;xD7)Pi6E-ETg}BM61&N8r2h*YCTh@_2J6*
zk&c^LBmew2_{1+v{^PT!ieA#tqA|gZ-tmy5{=CQkT+(K-PnZuKj&K1{0qIZJ{ff#O
zNa80p{GE1ix;Jy-RgCzqp=*<tw308d&s-<arN6DSPi`!qw2Z_2X~w>7-azoOxXt`$
zPep#~GcJ9Loeo6(^NYCszk*uRLdo4@m~AiDbBlSw?RQV0rPbye$ZSb^`Ndz_op}Qi
za&Sc$$)L!>Ic>R{N6c@)reW)3LC}};0QKD&>!QIwb7_HP_Q`deMYnO~kmouIyqvQc
z8tOLBRfn|REJ~+JBT<~ZEBiAc?n2-~kjR!UwdlfvG_l9R(I7vJ#C>wS{sS3#Y-7J`
zC1Yg%h6g<H6*JZHqYs8v5!YwT7gq=7-Sx!E2X)7@`D9Y!TP^ADfUjqzxZKHW-{;L^
zwx9J&OLfsy^P*Mu-nc(vrnf<MhZL1#Tv`NTkzK3AQ#^{~e|Xhp-m5deJ_xX+?9pWV
zeqv24u5jO`Fjx(ZW-p1RsP6mDvX4JEevUP0$to)RLExKnX(v(aSa|(DOCqme4e#fI
zN5aD=i@C^_4bJLgz|yE$t4y2R{x`G}T0Q6~K6|k`raRZ%+mZ3y#LKUSc`5>z<q=X1
zF1zP<QSY)Umvoo1Sg%i65~569iVQxTrGnYLxPRf)WdvGM>L#c1+gWA2i&b<by6^ZV
z?P3YTKoX|gKZ*fK`OP8jVm`$EdBMuDcEwkX{FY8MNDB^loM6<w@m39xvX{sx(CD{^
z^+#8QEXp@6>e@yn7tktO3R8`zSc4NjO#3ykC6Oe)4IvVnynT;S8KP@o3v<8`pw|C`
zwve_Bkq)w{P*zd4|EbRAiT#HNrpj~jyIh0m62K64v&nC>`2Lt;#vni-8CYe`?P#qu
zY&D7gVw(W?WcWvvjeEO3q!{wsHt5l9&LM!YnqtaG4kv9j0Q(3<Hb?TyQgBQZCq!cv
zqTxl?xOC<C+d-X2Jli%<e|RICTjZw7Q+%l8;<%ZuaWW7kFx9}#;P0^KZ^zlk;xO{<
zl=B3X%pj4i7JyME=sM-@ziJ{0(7!U$YISbhd2VD*=}POnCZj^6FW-d^;5yvUQqq2o
z^Z&zdKvPP&KPYbp1YC18wv^C&w{E-1N!-4mTpap-sNTMC<Hj6F{2n7Ny2Z}iG58hq
zIzP^3=2`V$vhH8jb2JD$2#VZlw)iI!hN%IH#=x9Zf<3nVg8x{PCJNyCbES~!gzXsa
z;sL?}{;QCTnx|#1q&XTKwY+~$tf8O(5yWyjUdv1uGdhm~W87A)R6M|pMO=CCQS{dx
z|0Zf*T9GiekH>t)$2|{ktG|A=$t8c`SHINe960xwyvUi)rbAOmqkCTXMMf(;MNOE$
z%SWnnckzb!vO&)+;<Mz|{nU&oHQ$n@xj!)5UHpAaa|VxIk3>)S@5jRP{?Cu4FVO~j
zJ8QniO}@xtnsjIKUGwYLrO#VRewgF>RNpBf_i<I4wS=pFNm|C-?m2}+Ta_;OU#(6m
zhG3!qgT;1u>?gi|>N6Y;(5blO46^d?f$tM*iu~7qbGp}|J8jEZ-uX~b1}OOl#GW4j
zOUT@j^C1|)b6+>U!Q4vZMC}QEE93h*oBY-*^^vdyNtTJ(Mrk=A&@TFWbLA{VZbXA|
z9Z|1-&CS)dB(DG0*x2Y3KQH?qF=1hj6p}BtOt%!I6>MH~Eu=&&Dsg4HUC+DzmLX!h
zG0>JsV{Z2S;V?vdZD)=_VddFgPul(sSjH4f$Wd(GYK7c)i3h$5Xxy>wNT<pw*5jl7
zpjjdMM$jXD^K<Rhtir?fn^4RQx%5)}U*|A0ek7KkHuGdZ-)$yHP>j74#uEbv5TG6(
zcfFE$megP%0pqxL{6~U(%dAr0BALA~`-^5|wEnwS0OunH#mjw_L59Rb@SsE@@NF>h
znV#_Oh#pO^9+S|SJi-y^8T*FTn+E9z$1y07wttk0z`(^JWDC|YpNWg$Da36Ss$O}^
z+nkf5q`?00)u{reE;n9w5pI02=(VpPFe$;DZ!pbyUO^#VjnA}}p27AhI$Hh83?F*|
z(tnLjAsxVC?q>$eQSo51syQ+Pw#Wy8SX93n*9>B-h{b0sM*3R!zgEoerG>X>(o^c0
zOkcn6ONGyUj!{pp@s<dY@iR05Tw|;!-*t&eApevXeZy2%L;+LsX>&;F03|ZWQN^cG
z*Lr=PNzrR`r~~+#KN_1De9F&&^|ocQXzxl>ndaM9Kx!RHgcA4FVu7pJh^LWGi~5lp
zm)vU3zH4n<WFvpuuu{WHorhXxEL^Ot(aQoa)~I=h&kIw^zS>%5@2w9g+8xnnSq=u#
za<3SinrqcMWs>AO|K0qvDravrn4&6MnX@EX{g}Km<uG$-kYiQM-e_lnp9W<K;N9^3
zY6uk%5VkT8Qcy9c?gAK&fv!<}F2QpWqUB9^Y4=p#gEDO-)`A2@*cn*I)PNO5?cC&b
zZDS(u_vaj2PeJM`{MyWA#9O`Cx|f*VPYz%_u{WMQ`Td!~L$U%X+mOwrWV|l8q8794
zEMe`BY*baNqKMlTNfI%}L(YFyg6}yjPrRIEFC*`BhPar0#{gw05}JZ&*dITbTjbWC
z#1W3RvOV(faSbcpnm@g4RK3GN4k5N`rwm@b(UPqKnV%OYMvnV<ICt7VlU$4Qr3Hj@
zz+xtKqS38!qgcK@XNW9hP3e2B@~T~^pRf>jas++p<<~AI{`Y%PWEF`Ij6a6y#EVrY
zkfznLUd;Rc4$K5HAp8@0MG1quP`@`PQdTL>lX>_8e;g<xcQ4+Wst)E;#X)H8Gk+{U
zn-76$l{Nc*`Ck*y%W^#=KT@NhRiVlZAwSN(E{P$u)Nk^0eNWL<RBASgZ_g9U+aj1t
zxNjf7tdGmuZzGGlNN=VQj^ECaGo&M+9u=gN9f!+blf?ntsS7!fPkm3!a`~Oxj~2+o
ze;%E~g)DhU`>xyZZCcJUxaQh1iBp;6R1N>%XVGWQin`XXB)jj@Nkv`w7sk4oIVcnF
zCiYy^BR+_xr;v^Pe?#xmwuo{nwN0{h*kcKoH|vkysIkA(2V-QXZ>XVQh3OY%oYs)}
zk+NKXxX=vrB3hRpM!kSnrA$@kj$vg+!ofyK^dIXk?k&bD3jFBcw$H)qUrn+5idAHW
zptcJ8t%#s_O@<fA1`hwnX%J>M5i%ndE2CXFOJm>)ncEyqZV#f6lI)kNJ#i41*J(#e
zFB&GkfaksI0<|3#ib|4uvqr@S-;Y06QpJJ(`4?St6vpemYb|$*Bm9GR6k8!C^nK&0
z>sn$1rt6jT&Rrd0hT<t)ge!lUqLZPus3Wf6S5~6lW`dubwQUT4mv!_PUQvNJYop|$
zH9WJjvzk^#rB4|f%BIDf2BmL3wp<u1Zb!JDS|nvuFWT=yx+s+LaMt=Mwqke8ICz0&
zgpHTFT)MdT`>T}SGA82kPj+4tc~dK4%aVD8;+70wXM0wt?35~DrveW+KNMmE6isMg
zX{FJ&YCx}Y{Lvy@MDOn)<a{D&sBxJ8Yo%)9@5k$7JtxmelV69-T2dEQ1+<NOsEE0i
zs}gtMiupmgzNhr(d{S6VQIu%ivK-J99EOC0e1wBMpS~=D!NZp)Qn+2alUT1QQgK1#
z%~WSEEb9Mn^qa|VnF>{B4u;VHb!qv+kge33>bp-t@4RB7+D9M3N7aYFsNGA>XU~h&
zl<gC<=_yLGxe?e(lLz0ca#)^b<K<HB?P%yx)CMf-(MyD`R=p-L5C$ZaZwde;fk>N@
z^U8e;sr177O_?w`sY9HRe1gd^Rmq`0opl(Cr)o7zF(nSfqhzk5Q7kmfY1;WmyDpba
zqaJZgjX<)nOa5C*ip{2Qw83`5!`;NX+%w6Im?#q;Ke>wOKMd={$HP3vI`uuEidN)u
zmR{J~W>3o7SS#THHQeKuv5r72+6ptV&2<8`JumeF8H72Cxy-_%OD>5z$bew?&y7qi
z<~LZfOv#cXHe{M3Dv*H-ZbQ6g_cChi$*olk(_Fp?vAV$iV)e!9bMTdplQA9H^h_99
zlV**}aTbAiv4?Jme=n(h?y-=t(K)3DIa@jTU-qJviNpj#O%;AElTF}z4Cip)h;TL+
z9_NqR{Tss=*$L@wrMOZ<`>0xPyGTxmY36lO!yd*yf1Rt@Cca~sCqePIH^FopUq07|
zHgNX4fOwM=UM`-Uzga(C#aUmcq2#_C6;v~D(?>QrzU=iKq6n9|HaLkiV*7VxyeC)T
z`xdFRSK{ww9|<P~C`uRjp*a3seBE76Fg8dI9`WY$SKQhTTJjiwXJd4C5*)O?x1Mkb
z=c+#s{e!WbbQ61SSR(&xTbSXJ7?$VT3*R4%H*%zQF624WWr<nvykV}AH2*kYy|hi$
zx0&I1MMCCJMe?=361KFZ%PDsBo_W!~#`a@F;?;rLO{WSq@HY*z*=BD*|0P$WI5&7b
zH&D6hVaO*}p;o4-KF`%?<ZDW>Fa4Wpu#tq`*xT{9=p6sT{8l#@zKbUEW;|{vx9Q?Y
zp=Ts>F?K5!&h5@}bhH1_nn~^W_DI)1Zn|;Jfu3}k=kmndD)#$TwfJ@{#s47wCdmIF
z{$(G)*}BP<%Qc1q7jk>6Tg+nGsn4Q?Z5@Fsn$P;(v#+5&Ba|DqX)_b^D~w%T=fF6#
zoL!o3mZW%oj#^qjroiZulQGoD6(7k%dUlN%CHz8hk6oL`m^jZTU~N2`19m(;rnavy
zl#qrqn&h&pdm=fsfg&90169i<x8PGte<=wp8C2XbLw@_%e5EMvS}eDUWVIkMCBRMj
z?DKagDf%IgaFIDhm1_D{Ehb3|CAn<)zX;yUeJj|oI}y8GZYLr@@R&bR_}93}VP<^w
zNN(E4sVwJ>z@Mfd3=fL;_*`B3z@Ovv;qM67{SEgIP6c;pw=9j@KYpj}mn>C0r~&8#
zgUowWs^ylaYWp3rc31y?81C1hjqEG4A&a{autB7wq{9b*ndbMpQg0Ic{)5a}=RCGQ
z(3~VJ|4SZ;bRh09u?^#`^w)Cs>R47VK_x=|n1U)kqET)xlx)4S`ab}?-gt4qcx&vR
zycz^sfSI&`gGu10wkt|?rOc~Kx{sSSJnLWBy<7g|HXZ>WW)Nr&P8tGoLcVB#M}~kb
zfM41=PK|{kTCLS1Ecd1&t=|^PRvLwWsp>e*bkMQ+PSAuXOI<yOwsPg7+k6pqTJ672
z&Lq?iRan>EWBY@VMahT$ydj)k^Vb>+qBV%N_&OWC8F`#k_)KHLK%Pv0n>3AdM%vsn
zU;lkblc--WN~^f+GpfLO;~Dm&q~K<;1PDTg4)_@m&|PapPZdswMCa^qxbX6E{+@eP
zcg*VQum@TSF=LHw^ykC=HMEv-P`7n98po&h`ZdGG%lv)OVXB1KcU<*9%Zlvr-YR9(
z{@f%em&Pj>tVJzaImyt)V@4CLaW=CP<whYL_?~oSrnUYws&$HA2W68r`9_&73I_61
z6=%efij+yr`325bU#lgGNuS~VAd~7*w)rxZo32ZHr&N(DW2Kt&1yqzLh*@)1a>vi-
z9?NqqEr^&5izemG3W$A-D`B->ddpwX$mmW|{w?$64Ii-UAsG2?H!g+SF|n9C)g0UW
z=wIRtLO~E;38qbsrgbj;ZsVZNuoHmBzCu0vhlE<C3(S?eYx7^u>%7FyD$BPtK2Rm8
zNEaC3$2s*{bNxr0LJ<)w)vy8MeEh-vu^D3ttdu=O9Mf1I>!<e_RMGfQEBq~T=6vHg
zd=aLG_3AwjPt|Mp-pm1TpjA=>UP@981MhJZ6M7V8Qyp#e$>8!`Y1Pz+q&uKgZ-M?E
z`X<S2qDzIz_2ku#o<R^<y=1$IyyGy_-};`^-(}KeY5ac@Rm?;z^@pP{q?!8h1kxNS
zvbeH6N3_1*FwEMNKl>{TV~?i&oY7Z?i1`++h|ZQLq$e??{(z=zUK-8}hS~*fUv{LR
zabpT%2--YRSSTg`-Dx&FC$7h=v3HpJ>$X>>0e21u2@roM$*GnLY%)6hx6jAI(T|U}
z$71XcZw?1>A!kv*_x^J&tA&$AfNYV=;PCJW5DRig5xgn_^cr0e(#zlCI}QK!JP51=
zk~R!JiR2wV+0mAUEuKD!4dzrC!lrYkNEy4(^^h3vzx1xX40=WW#DvQp#hE2K&#l`T
zHQ2R&Cq7m$I_}{7ct~*V5*q7&CM3DN8ffKKy&a?^HBbr-G9V*$xs5rmv9Gb0`lo+U
z;iadR;j()$<N&D{uh`@0&%iZ}-3DG8{GD*R>z5iE$wY2VHv7PBv9!Tsre&R0ui@)i
zw;JfSc!u*X#GMrb75Ofcl0bj%^jf9?xqi0|B~2>vr|AOM$^8C&wEvpT8gRMq-~6Y6
zg}2t2nDY2t+|tEv8V}veDH$2N-XpTjcF{CN*xD-6bJvIF_Z|A;ONQ+nX}~^v&w!EA
zdUH<FQgbhBv$s*%VBTqLRprY+XS*zl_foBdl<%V)D;J6@=2XJ1Ivr9KU${|<brvj~
znODSDFh2@GzZ~47m1bt5(^38jsfQH4v>7n|TjO%u#xva7_{;Hy0rk3*yMr&D_tMYV
z(TUXn9D6uY@NC_{43*b9r!4So%d&a7$?Q7;^1oW8OJ9@0I>zjW0dwb~1oU#`UoVmz
zaBoW#jXs9-P9COIhhAV|AMujP2B~Y?U@v=DoV0R!8tHTdvOHq!eP`YdMa)}FtNGtz
zi=e7f@J`dGoM$fRa4fPz6f8PDca0`gfp;@+I9rFF{(j!hf9h1gRh$A^qt2{Khvy+|
z`v-*cR@52{Yi}D@*5Z`NEhe3X2AdBu>8RN$b`u4;-nKQyx`7q&-aBA@QbGStB`15R
z`mUE|*y~qDt`Fh(-J7;AM5}_U@rBx=Aw-@dcIP)=r)55n3GBO-FI6b4uF6Tdoj9)U
z{#MgK_dMF$6kD6iiZ3z*aP=vwu@`Pl-_<as?ORXnFfHgz6y6m$n-v=#BA2e?9UC$q
zr>A6y6#5d7<3>rjTk-f%Ni2a}$cVtG8jPs<LxHY(Np_$Vp@6}E?No!mh?{3Nrhd5G
zR{8z`!qXf`?<lulJwU225+2SCtn<wx<}ZKwuD_<lDyHycQbk<=H0fN7-n?l>2H6-r
z8giQ6Z%+i2P>?mmX}uz@&MD#aY>EkBc^u)TzyrkqS21f#am)5-0e3~|!2LQh@{Z!y
z7kKlyFYcobS)O@%9Kd@d1`TA)X%vhiqSLaXMV-!@kMxm2!DBT~oarZEQ+^x=X^al!
zW|)kVd?iEVGRF>ZRQZ#d-Tic~wKRX@WJ`-D6_W=oM%>KC8%fqJcmsleA&luxbQ`uP
z{`%p8omda^DgGZ$#f7W_H#pgL_YTFB^P@-oXZKx0B;`U1!Xi?9rgd5sMS2V6CULg|
z*OHPWu1hSZW*K;pubx<Wr5bE|F3N8xtEt*PtnJ$+usZ~(@G0wsXJ`f;S-RU$pW=63
zA>%4x*qT~w$;X)$=%#3Q(<+)hb7P{r#IcI7TtzYeL~ggp>`$K9*sExCu#qU0$9_==
z5R4QzTxR2DuN??^MXRsh<xk{j6JYZlm52zx93z#6+gGqWWj!*YaTQYCTq6sMW3axG
zv8_BIAtM07<7=mS%-IYLJ^N!U$?Tl1KJslz>>9!JtC=c8E{tpU3rXfiJ9zt#Was_9
zcM7IFIBQMMw>!<xo_|vdXsN)!T)^$tERwmzDcxRUnzBlv_6>b$|1H7;<sy&h-QKhp
z37N5~gQWKLA7!B#l{A5&0QxluI<k~l;}{V9_RVX<{9}j--|OtTV1`7a7ZkNGKi#l*
zH8gDszTV||i59Hatyz9HzxW1AOLEcW0)jIP0cSHwy(fO3ND1n(N?+w@b>&#H<@l8J
zNl=$<Btntabo3KT6b$t`_UyJ#1|Ag)`VGi@GpwqFvA<&KxVl8z#7<(zp^)~IusiZ2
z=CQmcUznEJ2t0@~SXaN1g|k`V_gO7zbCR;hU!=V1^51J06>OQM>FN#KA4ygz$PEj;
zD0%0-nNiWKRW~z7cyo7<eE;i*sqk&Zg!A`v)@EbY#luWtzkKph--T&TrMwDbkz3_o
zz8B$XAf`@#b<>T~uT=(B@B*@&H0j*o$L0>PA}tq6TtD1a=)@h(t29Z4nPbwKzbVBW
zdG)vt9lAny^~Jcm2e_w6WDV2KPn}3`(-aVKBQzcaLBk5j1^w=)MM6@qfdeVeq$~xm
zl~&LbN|$v#*>pn4p&hMkc?8$U8w2j3*3mW&CSgA3O{=>3)Q#t5L>V)4&=8S~s)XFz
z!afdhrMs-T#B&XL!Ufu!gD%ca0szmJhmD4Z9N&k(KK`yPZ7mObk#kR{hzq99AkQb4
zrgo@do)b*G?a{9dc)$g&$Dj6qE0fg~8XD?j>#J&)*rhvfA=2mZ^JAB*!+%S6v1$WZ
zJr75!-U9}~G-`;f9+bzk`{|{uo*VAGqgBIfr$ZcE9h$1DYU+*-&;H)I@kYqQ@$K2o
z<;of7>aw9xw%S!O4EHYCyBJ1O&6P0zbhHz5=Inbo&A8=pa<l7mcMM$jHX^nJ`dwPu
zpFBLhpWGU=p5|IX_vdaYc~6z_>zm!%oudxjGYaFYk-dPM5#mlupS#f3v%O6p0EXJq
z97MK0^RE#-va@xPd;F0q_u)PU*4Vu>3w_+5qlW`)ua?(Vz^gP*fsKg>gSPf3Kll6n
zk@&-EU?deP3AEhOx`<`xy*Tt0bK+WkI4NG5zB%3?@9Ao5>n?Biqg>e+mw-dkp8Y)m
z*T;_=9+BcVH^E5^S1*BEzpFi;I~J2C?98)JbY`{dzT%!6mlGFgXQzmvY1-}H<wihD
z&`s;eG$1bOVKYP)dmYeMFcxq(cgOtDL);ii5%=Ki@poXFk!X9QbsSN?CN^{zI#9fH
zye}8@^o8)C@8<96HA2+-h^E7sh7>@1di-kwySO~f6I&BE+(>{qKhnJJ>J&g4WzW4I
z*WD2Z+C{3iHa7T*_MWDDsCCSYOjaIzJjL?HfVTE<%6mDQ%gebD%q@T4r&HLIudry_
zeJM}7g8ja&55>c$bzr!JvUW^1p=mR3=g6NSb^a}#W^<On%vr78^qp2B86YC(Kstc0
zv$P&16g?iIV#!vZDnN=#76JQ(N_g#VSd%-Rh5hxk*~`o=KT4?lnC!wyK+KMNmCc)w
zJK%+};s)Pe^Vsp>C(&fH4fplw?*TZWm0j+=Gtc;Djou91q!}+~gC1k(Ir#>yuZ))z
zn^LQR&aS?L!M)S#?<{lIPC#Ag=#Y*<p?;M(L5gM~5nA44)YHzK^b#kHqz501_gUAe
zP`s-kmk1AIpSH{&%1#3;f+otsHzsX=uA-Vo8F>`418|a4ha1HD%4`$_Yh9bitYmh`
zTZuS_RE6P*2X2V{6eH(=-wUK8|7;Sb@)m}!<lS<Y5|Js!5#)o<{uYhVzX4t`=Y2X~
z8ANXGv`Cq0JUypfFhfj+?;!&Emc(ofNo$p4;y1Va_2J_N^<2lk_Fej7&YaRR`v%IN
zU@d+5(Bh@NYc?eH3SDpWV}yqZvG`n=&Vdb_2f`coVZOgcww9<Tn%DW=!6pH3ukVf%
zevDL-Z6IPigG)us8Q{~Bk!5BuZhrRBWN{&gOzPT5#hgcA{X3?p1=sz=$@yLE6@_s~
z9Z7#fBicDU93*xN&FlLwr7jToU+HW5D;0Idx?P1@ux91gvvDU4v9Zi7j}h~xSTmZ2
zn~)>aWKo(dm6Ro_{V%Tq9B_CQMJqkIu#FtSDrBI<(YfC;0I2{2ddDKykG_H3cuuX=
z3S8N1_kgJ3Tc4q%Y8bk|hHPRPo%8G9y2UP=eZnuyzpWS6D>n*olF{Iac%(3~ir_T%
zsK(gwtm`){?lp9u>dC?Fn!Bb2ljr;`kI5(3)F<(BG=iU?X{{geujkE9bJjTZ`kv;u
zm$&URzc2=<04d&3yL#5|V{`mVn>THih_r<ef!6^$b=BFJ3zSm31!Et%#0qNA()L;B
zzvRc0)$-9uF1VF}6A-KpuBwx!;FxbLT&Q9qs5pt^k|FtQ-mw{SsS<V{WaA<0^d(W=
zH*wkqB%KDH3Q0MCTQ&voA!qXo+R<UvYVm3ASW9^Gz}bSW<c7RNz~(qeb7%6aW35gV
z_N(n^gASaF-SNGy)B6M}D*>*P9<pLuMTU=&SGh1ppBf<H1$O`r+5rWSN4s#cKvv4l
zMLUfqBULoB?I)$!OeLdLJ$msn|GI+gVVzIdTsOeWiZWvSf)}HYxX@1wS$YgwggRC3
z7Jrov*uQD}vEyE2ajG0`q#nIWCC}4{p7?HLOA{P%rf!Xw$2RvfX7YmzlV@&@hgZE4
zD*jl0&MSzz5~$F{*WG0=1okVV-`!B6Wy>9DHLDaCjW<G(II)j$_2ayuT8J`ro-hIn
zma{QJF#ZZzFJe3^WvKb{S&^Jr0XXr4WHv${KxQBlLk4H-Y#4_ah#}rN2duWqe_6y}
zAn9;3X1cQCdOPASAQCdh7t9rfNTN=+KB0^u+~=gplBj8ZPFn)yb=xF%-j(JZuX-!V
z-)F6r+<jvx9Xx2sPd;Ao4%&Nah`>Rx(HNX7O|HL|o?K`7dio)XAnX%Mt}T7cih`9Z
z0IS6xi|V3TTIbC)L)_qjQ`|}6Lhsh2MzS@V4#Gakp9M6y6Igl)G9&6V<BkJ}6RDef
zQ8sSubt;biFgf!1hF46bQ|%d!1ENqms{$HmX|i2Gdc!5`O+qx15k;rXRjaIgiw7DW
zty#l-?h;JaUR6s!`HG{mBi!A<nZ)}*7vm1(tDF;Si?lI2QZi*;7<Hvtxu3^y++`C%
zBdw^yy0dTvO@J)=0b`>GeN?x%tUS~ngMnS2xBGXMWJxJh4sE;>qz7H~LIJ&Q{6Nn}
z+#RoF`P=n4YAs%i5*XwEgeVUtGA485^OaG9ck!|eZhyd+_dE`8uwNogxE=%SQJo@w
zmIpfZeZI-6;`eD=_vTF2Cv}!^Di!oAn5l=(3s-*#n@2USJ(-Pl6WUK>%SQJ+(UJ=0
zsABvgnY*NZ1O|Jk2z9p?c{{m~tQ);Qjg}}XXnB>N3Awsj<uHDd%q~9v3NC{|i!O{`
zUcHtuM4bKL8yj}l*n*&gW37R20W~=WVl;+3&HOH&8CsixiQUbyxRWO_uM+>hrQ7u3
zKw)rY-(~b2OAC21hE^+_gLnw^P;}~%JBgu`-b)#mjEgpK1J|FCsm!wL=Vc>;maLb2
zl=tg3ecHym&M4vIJZ?+-?seI%*L5SBSanV9mwo!V-!6|CVrEJ6*l7Sy%(Q__YL0m8
zx`2pAgm<}^Oh?fSA|ITH;<Q$#Y;1laK$5$EPD~wUP-s<`kIl&><E1~`w$<+~F2L~i
znbbuVaTR&}F~*EA5<4E|*wo)T4t+~79(kQq$2&)_(#NMl__ha~*T)dn%o<af`GbAd
ztvl7-H>*ZI0B5$U)*a|&TAe|b-b;KYAI9u2pJH8IVUH&4cPoC9BLF=OAUF>KUd+gH
zglwsMzM(zKjWA-^X_zo|+-m-A+P=Xfi{!mALkbQ<_zQK5fJ4)M?q^ES!ED3$xHSA<
zF&y6;xKz10lVodAqoeG#nIjFZlaclMsx+g8us3pn_=6Geg8{8Wfe__folnz0vSTCG
zq2roJ>bEz+%DJAnT{_Pv+U|1}c?|}h!Y^&Grd|;Slxv)7VkGr8t0>u_*a#c!qj;pc
zaC-%hlzK^JNv&N@sXBo|ml4wZpFA!MG5HrfthIcJme+>Ngg&x_@;h<t!M~6F6y%4_
z9)-Z}u4F3^KtwF`fb!@8OAiW54Ef#1sga;p%LQQ^IMP1Q;P7<~LpX(SP>+C-!>P99
zZQ2?Ig8v;qhbhUHKK6t%PzQ#`L`A^9pp6v#eQ$aETJ*OC@k$lz*|5;E$?nDU1o(rY
zpTQgG5pE$X?_xLa@~x0-HYEJ8VvArZ=h$guqJ66taR1|jLn8_QY47(j{HHhN&#aiu
zU$O@wON~h}n;AcEh{nISvfadu8GX~j8y9KwdJ?McM^-N@{AlKNJ)hpt7W1m`BxoM#
z;e*4yRMe;?Q54;Y1e!@igv94N8gSb?#tNb)mDCk{r5%;HO2ftw{#-r0L}szMI&&Ri
z!G&c5Um!vEkiybA@=MFBLPlX{2{Z(e*}C9ONu|zZ#{n|!9054C8h5q-^nOX!r?XDi
z(_x(%I+d~_E%Qw=`|YYXJ5FloNJwm&+#p!(llMAk4=POL%u$Vjq%eGYt&&d6aOOvn
z$Q9rPjY5G7(a?|6AnE0|sVPF=VbhcQ$Ppzb0mQvr#&YoWG~vn@^9*9>l4s%)!bb#Y
zzp8I;!*zYB-t~G=i^S><JRm1l3<{vFG`E2?WH~2KmcfZXTe}&g_O4g*<J&&HWH?d9
zhW4L2p@Bd6C$OZMMSp1wQ2FZCrj2Cb<K!*qyV;T$u#;h`Fv(`0sqvnlUBsTKm;FH`
z8F;tZPMwUJVDOHpl&$DwOomnPG$4w|g@=o_53)-5lHqtcg{TIps*90jfexJEU89&Z
zT>o2w+&U+8FvFdpbe37nPTpqQ{k`K(yM~6*rsvC25$vHP(=-TD{(Mf0fzw-L?jA%7
znI$ZqAAaMcVEgs#5zls##kihvynJ3F8R&lR_x`kCs_W+3&bu_GnnsuAQLINKR|#bk
z(u$hDePbcNEg+h)%0mBHS5r(xlj8^#b;_rLm5+=X1#4ayCJ>T3nukI?zQ0trZgG9U
zkw>f$KDYb97aVby*EHRD?j%8HZRmRB_eIR7G3cbD%SLK@mNE3^^rdDm(zR-^4KOvV
zRhBOheT9!g7ND|q2>C{UTj)kj!&$+5fh|FI6b{P&4dh>Z^1l4}B$L~Tra{<%J_+f8
za3pruzm*`b4nWy^20v#ZR|F2>*8BxZ@rc|_rd~r?Z@!RFFdoAFAO6&-5NgCunmjlx
zj{z^twvEv0iuCM^lCE2D3_18EFgBIn)b(JS`rhXVE3scBKVN>mgTaY!SU`7k{X^1g
z6`_C~x^7%C!DDq}735^!PB&#Q53zzFOBoyujOA|Rq3oW1BAHrr%wBl4p<;sE;2jt=
zW@om#$Dbg*vGP)aV0`ph9o}v86Bo{xR@hG3$=2Ek4C{8U`xff<9zzKLJjVK0kb|~d
zRL=Mq?o#bN+qfA>uH*9@8xoW^#-Qmu2fZRs`7lu1Ps|o8YH3rvOwM~l=pCO2`;yVe
z$5^}Y#iwXD4JUs-)P;nK;w3VCpD9Cb)8#w?abTvjD6m2@)@SA|cmnxTb)u|r(7`6w
zL@%}5G7*Nj@d2mlyrl!0c%Q$abgoW94^af>t1Kdz*uger=538jK!co?S9os{f%U2@
z`G}rJ_OTw;W=tr@5+Th}T`Yof(`{IRwZ=cio2`#H+`9xx0a}j-=g1X^V;9VYs94c_
z=*BNTfGV`U%k8k`HGYdNe^)}=;%0ltBs@1nK7JL=ydQDoy$Zxzw860Dw=OE$OEZp=
zR43RENrU-=5|ry)g0-ZQBb*D}kP6+HQh#ds>L|Ibx*WCPugaiG`*49Fy^h1~o6j?X
zF?YMo<9H8Q)z1Bh%<Ib~>|5;d9mO`_h&Qn-!|t@}Ecf5G`RK3!&smFJXp`%Ho|h0C
zeD#YS12}qM3}E#cC$k{PdP|Qs)fAz!89u;>d7rt|$Pg8LO?i;3X;-eX%uL84ze9z>
zY8G-mpU)kWD=SWPmtL`xK{H{jqLge89Z3flW`8X-LffuS$=A41UGSk6DvEobJp#hZ
z?Ly)uq>=QBIC@Z9Fc|x&iKB>+ALjK5q}{Rc4agi813t#l3$oQ|RuE)?$l9>rF*+?J
zg8A1G1j@U&nuVtv25jrAh7|0DF|f++sAICqWa3|CQ=)V0<dtSd<jCq=C%;I?I5ktM
zbm3?#o8x<5|09^j+yx{)Z7TVdL64F@@j&~LHhl&p$ThA@JtbzhjjehI-6tiAy5Hl`
zxO061ngmWBjluDGhDk-beOaK5c6JFOk{m1jwrW1k`cy<kAa!);%LoX^8ZlaPXx$M#
zrz;dkl^r4@s}lpYhF@V4YXQ7(Io4>hyX2a=hfJTKo8u(XnPWJKqgSWfzqCqw5x+QZ
z<||kb3!uHWMvTEt@;0SNL#4Vi#bOPwC>9{iUTV9IO@yay8rnMVYna>~9e|7ea&x<t
ztIj*u+BKLz_hN7r8&)68kqa9aV;nnne<PW(Z%Q-*M`57>RF6Zhs<!d?5L}R8&(pM}
zUoru*eZp1odQ!U?1mK7(2Nc^FLB+D=PY&T%1lvvH8(UEktM0ujhdjKmQL76OC4gr!
zN=87jtZFTrMw-U%s7O!SvFdAZn6<0hkN*IxSUFipYi~JJBGqXlyS)y^hyX(ouNcmh
z%~$V=ngZFiA+cEU8v^ZyZEkjWiKA!H&4cR;vxa6{qGQdsMK&%MvF<yD4Pqa_u}7Z<
zKP-L3EIj_J>!i9zGWN@VY&PUOFl<@uKXPX$*lkV*R!{1sri1Of!*aZ%t=@=qsL_Qj
zu%?^McU{pUHaM+UCBpe;puH|*aJiSzsS%O9!8z-~p@JV!qktRJt}A(Ys2xgvwptSN
zDDT`q)_nQ<<r|j-hLL(1vUHuve>?CxH&ZWY<}f?kovl!}z3kl=5o3P<J^y!OY;CUf
z*|6A3c-5ttAPfNx@#xsEer+326)O<yo#)syP><i|H-0xHlpjY9O+p~Y)Tieq8DwXv
zH}`uR_sRZDy(#_C$7uOPnK*PN(syKTNz8Eir@*H=`&vDqrfq->r{E-EoZvF4iW2kF
zSt7xg3oQiJJR$5n_(ca`jxPUfO@MQ(Rb!j_y=_TPq7kzpTGTtKf!i1+Gzz&__o+70
z897Z?jqLK<lSGMM<uRIGH%2oujg#>t_S<jX$-8&Z6fo$e(?cw&M;yZaCBSI|*b>6k
zdmz1R%rvv){Ng^A;-pkFPS|4b+cE!<Rf93l)`N$CSW`X$ScSluvUZ3aYs>pkw@o=#
zGar@e)RIamg_t!o4c`hqHf2a&P7Gw7PrqmV!<0tBMjKYt7MTh-WR54;W3}AC?zBz&
z{i0LIR@8X*w=|OEO@KTu%1uSXuSY5JW+rji9p~Fpg^@XWwL)r#TKg0pbwv6h${K}%
zA70QbHQui$4}73c`-1=4SMAM%qBucf|5bDP*BLMDw-t2J^sIrCmWMr0kj=R6@S>|W
zYaIFY=1krmEemU=9(NXrRI$xEhn<F%x@_BJWA5VK3(K3{aYGJvHs%p{TI%CuOQ)EG
z_4h-*q|&*He%48@=eN1`%sVeIL_t=tMBn8+t{S$>-vUN)qBJ<GXJ4;2;z;{Ym+HfB
zR5P*a#;zJZP++g^8#tZ@;C}S{ar%HggSyk^2@XNL6YO5_^CW0a*CI_@^!FS(vnjg`
z3>J~AnaA3szZ&xdL}EH^U6K$^>}|M5ZDo=iSKo=_U`Imb$X&bic*Pdax%V_P?34Q3
zb6mcs*8}|E9K?G&q5=QU!7q$QKvR&$iq@T!5mmSDvkxvy!$M)f%GBz!M5%wp@pe*C
zz-d>aGvBj3Ap~jUz^~T5PxRV*8Rfotm%~XlMC|D@3p0l9cF;3TZ<aAjbPoMm+CCB)
zidb(ba(PzI>5Qz;pr|INKQp782EtV#VnbcK3h?{pTrm43d&n@WK05_PT~)sN6_G+S
zT1U9OOIZHt!rqTF2kC$MD`Z-Mr_|?*1$F){mxbABhLIoqdde5=bt5U-s_SuF+x!U!
z*V7_}&33JM=WI#F@8)c&j?PgaCHF4|;*V|0Li1S2p$*lWzcZ5a^(1`c<cwtZVAUg4
zeSlp1@-~^u=b8$ehbnaUJ)_tBX^B>N=+jrWcThjJxnI}4B4Qy#9h&X`&ba;z)AIAR
z-L5HX-QJKZb6@fP`ZQP&YE>|`@;h7Uo*x1Z8<AWhHWU+4#i~9<+FQ03aLzv!yK7Pw
z69G52tzt-*j_`Mi6P1Fy$E|GEL!n!7?*T<0QfOjt^XtUXdUgrz4R+E{gkCnuW#lIo
zI@!H!xntwE6ydau!*;QkfpJXg5c4kzWijWnl?F|H?^DT8w>D3_P_Whe(*xGUUdtmg
z`qR5N9oiWysh4??bsmcre5pS)#{CIfEK9dM&a@u^U;oxe&Vb(9^l^+u8Ptv(Hv?#E
z8&v)Oc&Mtt!%h#{X4}xhwA7|}XHOc7A^auZ0hM<Xy#?>V!$IfT?{*((E17!6>rn`N
z)@1EhS<~}TM+%I<@pI4O3?gH%ej_Gp?|=4Mro}aeA=LFm)TlCY$UU)AY3J^GX+H0M
z0&i?;u_gnNMBauA!&yok-e+`l0+=hFfgul)G(@jMWhqS{XN-l!mXE7LY??i=(GhKf
z$25z8DCo_%)2eO3t@roKaVc_sj$}Lg=@Kz@iz9-M?UdjYowwwlb2dTn#uz~^6CbLY
ze?~k34LXDOL>=+RMn4^+PcL#haD~6HqxUJc<somY3=Leh3#I*)T5E;wVVB%J#xZyH
zFZ>+OfYVlhNbx7YY$7Xt-R`R^6ytaK-V|7Hg^Fi)Mth1xYPIME9{UmfA+2RD<eyoZ
zJCcYl@s+Yfs-FU>x~lJfGVdDaLV@c&Ll$+;hRAg=%i7$5DLV8BS=vVjh)h1#O!cpe
z7+4a846yf!MmRL}GTXLwtY~VD-RzOo_aKA~9KGsdp-L<DI;uJ#-sXk_2-+hR2pr27
znL7<^`@)SY>KfZfuW34pERTkD@E9|n(N@k6(h5P72&`|@iCyN4s}&oNud8jR7n7@(
zlTVZ(b|++xQ%$&soH(nb69yU`H}kC@I)>4tZ;JemoLE|3y&JaId=}LouBge6y!WC;
z-@XI!Ld5@wlI8x|zq{iN*pYodW(3O5W?hm+e-Xaljki;gE|$MzqKve8gGm;65XeK6
zakrI)HFd4W8MxTxKXK=gn<u>7#hG@sV3N4K&-ZTgp+Eu)kuJeB&V1SIo$_1|8$=IW
zzaa1(_&L8us#)clGS?`nCJZZ!#P4~yoM#4D$D`Bu25Z$`eqk3<wrdj;@<N5Ly&8*i
zX8`BykJpTyjw0iID|h>;anf$yqlkdIzI7COn>bvbB7t@Up)#F>%0azuQelfd&Sw`V
zA5Z?yAD6m)62Ihm)S?LS$&ID<H^2b^RhNjnW{_9;sAp6LV2OkFTrep#<in*9&PlW!
z2TEH=9t|&pO0Geq`sMuy6TMkgb1u`;wdztl3qf-B>8lP<?%<ujTueruI)=q@+2H(A
z@bk{CIaOX5W%yjZY<h2}9=Zi<I=9@811C6sK##sBk=<M=`Y01s!<R^Sm<up<lP)KV
z--KShE{K{c7g5yDc8vS`vU(h(-eC&0lV7$^Ajh_5DF{h8#Sbn$&7}cnAco}JQ><NO
zu@&K6NbVbx=%KRTuq`|avWwodeCWW(J!8ujjXzAQmzKkgMI_0>yS$~G!{@H$<djDc
zUy8f7rg<c^Ztmh%oT|Z!k-+e4)f=Y8`2@-4es<>FIDED>Taiyhxt=uiOt}U$(`;n=
z;Nh`>X9o4gct+LfeHYUHHd!1`!3O>gQlr+YN6{$Xh55p_<D8xBeC(!B#7+Nchz=C}
zHNzSqCWAY*xHiEmAbRC(L)^>065^g$p|s|aY14Foq;OyEFl~qU2B_@6s*XMy$2c}z
z{Lb412jz>&k&K_IP_VE0V_F=GQo1{*7nX~zJif=2Gu+BqkU(pcmq<M-9DS%Y=nzvS
zX$+W_*|GSKi95ZQyhZg-A~gZ$S&MOACDkp%k<{z(r<c0!oS2oNjgQm*1uH81s_6L^
zu;R76HTA?UwbswpVvHRrR%lC2CZB2fsYgcf(gIC2y-ri<xm^53`>?C=R54SQCYy0h
zwpzfhC-MP`cF#E%<LWk=`Jbm{gRI?U30oS{hZ;3?hq_svo$<!pp)DOH@@sbX_h}-i
zRPGRa;3S{SPPqxZx&}mxrbiuh`fnUQTLWSOS;HS{K;%rTJI&bE0XQ{IZA?7=8XbQ@
zg{;&ry5EX}64r^;L<-q{*h%k{r@%#PL4xe=7Qr0U_Nvm~q395NrIWGn265IZ@Q>5h
z%7<?sP@@%1$d@BFKH!7XkCfrLwIDKJy#!1L9cZ}x`ga(^;shtkvcN6>dXZ-A3jwml
zsgc_VUH^K4*WGjnR(|$WD`mb>hZ{yllZqh@Lon=B8+<8eC*O?o8J==T@UBUA{L~_j
zzH&Nk3JQU_nIQMF>u(0`{P3oS9>Id^j@|ib@a-Ppu;{t45Ec<b?V1d}BVrxHbt)<e
zrfQW+@bJQ=J<ym96Sgs_My|VtOQgT1XbO+I%7$@arq%yxP>V*mgG-OOe84eOGC!*W
z9vr{54<*wJx;&y;lnidC9W``T59!b@75)kJ8-vH!ffzx~@cKHCFfun6{HP8@{!$f&
ztU)A>E<4&uUJEC#2Z>|#967J9P*RIMz~IL9Aj(%ttu29F;*{dKV*@7i@W^@)FYw@2
z9lQ0f=j48Eta#j8?GOBH;}F&eA8_Ud^<M|<%>3)gt8y|9>U_H2co_N67<o2lBAMsR
z{d9vk+kysW<#HcUi%~fz_y;|DG(GHDYQq|xA9o(6Eldpa)=UmMZ%$miy)UkrX@cs+
z+d4XK_ol-F<DUY40!GwhfbVl|BwX=k>rqj&LrnjBSWjON-~Gjzc*n!8anRk&+6{c|
zzHzO4#Cc=M`Sdg^;EuE7p5)3ouVa;Jd`_nkxY@Yiep-HJzZse2bx?9Yolrb_i;Jps
zKNR<@;P3lg2RtC12#Y?r+}!t+=dDpa-3IHmw)y!y{Czz9d%An9Q<3Bg&@qk!fq~b4
zxA(_FKx^&8fH&YDInvVuf{Sy_tQp?3Y_7%cU|S$Iu{N(C_kTrBUyNWq!iz&pOjg9(
z#8;`tA|t17Q17Z?pI;qusgVNVOdBS!&bvUdOkc6N#bNFmXxGzHC_GMMBx_>Cr@qtf
zv4}x@?b_%oM#CKB{t3rdyZ3IZ2c>Z55%)>we<=jY4WJ){FVv#bZyM2tp4pawiBiuS
zqP{iM&UZup@S-e$WD!S;sryqXabIKNvSSw$r-I>1qtLG(PCc~Sj1a`5+o4L`5bSv;
z>ovQ54n42@`xlvWQhzNIR)=i^V(IojlT?%mY}=rasKkuK#tEg)p&2W$1iKPAY%45k
ztZv)v7{25A5r7)A5aMma&1F{rd_FU%?|rpp;(+;9CE32>Cr=k;57iqdKe|O))T5+d
zjityEE-#5$CdSNo!Po1arm3O^c&YTo$xhQh_4bVqR>ZY0`+qp1|9hhw7!)eLTR%5p
z$Kff&j#D!DqIem}y@8dQOp8oR_|GRDp8egRF>K6DCqk$?Kalw=cjV>MjA|<Ziv2>n
zW)hPHrSgNqk9~iC)Tlwkx240&;k)C<ui5$UDN}l@dC8_HKQj$k!w(A|k+fQ0J`qgw
zGchm;F*fHHY_PiGvPf-Ci)-E^!#}2gh#KQZK#}OEB@H}odyQn%&#MXOa-yKcYuEWJ
zUU5m?v$)q^`$m1IWcaHihi^(Yrigt(U2w3jYiCD~Q+=PKT66oA)oW@#lvdw=o~|!w
zQU&(wGAHYh=pD;PgN(%ZO}Nr6@$=tQPR#ynfl9JCUq6S=?N_4E5QifxeXilc<!K?N
zm8AYc`tj4pu%xkDW9h;mK+R!}6|+&h<J^=kLsOs7C*D-whI?77hLXK1;iO&0fIX%2
zcs`qQcB>*})fPVlP-DKRXsETe(9kn4n27%y!d=Z>&BN>ZgV-OVP1aTVVCYux_-2OA
zmi}W_*s+xC0e!XDOsX+Yvcf^VyS@T`?a+Nz@+#U=-)4}DaUnq*u+93?q#osF^8?g%
zSYCBO$8tiZ;mhmSJo@KmM-d_~I!g8}JT@JeBR;(n4D*&kH{FK)Jd`MubdFGDg1LIW
zAgkmKy^Su&h4<BCB<eSD!MfQ$nn8-<Eq*RW{P8WwA6O)$G0jak_Z%vx1$FsuN}%CJ
zjjwFHBJU22bdG)E0StDm&~Xc9MpX|F{v~vlIXN7KcQK`4evAde(dXuuzO!`t`*2xJ
z0~#ka?ovgVoTT6LZ>TLKf4n2sr##?KS1|Ry#;Beq_L)8S8aQ$zlqk6Gdf(nwW}RA8
zImo{pv1f2%f`o+h^n`-+UI~QEiA02ihJ^9lLyN?l<g8-_7n=ojV$tsYL-2=#kWx6{
z=yM=7kS$zi4#Yz7e|#P!B>d-l{_n_#XU>7dKydiN97qxr52u(1(fxPr!X@WHB*-d5
saJ_kuC}<uIp9hJ94&ih2Ajbds?cksV5HZLEPPzc1M+=$-JwNmR0YAecTmS$7
delta 49121
zcmaI7V{j&M_pKe<wlT47+nCsPGO_O1wr$(CGqIgaZ2RPS{&h~Bdh66%zy8qGAG)f!
zuf5jV$&HYG4Ulzc&@e|+eRr>cz(7M(jReqB2Y|NfJ+xrnj1m3VoOL8umM4tRpzQed
zDymX+#iUEkt9LNFJEPcr+^*IO7%69le6MjD$G(nKq;1|o>nKFl`*azE>5T+uLa&60
zT8+Bq4ZYo{>a9cT%aG5_zk}`LPW8FC^A8<j$Ae1P^sm3WQkgGlZI~@)CPYqLA#z2H
zQvhF(R}438dy3Q-NU=n2l+Pq=9oha)OQ(r30dfvTc#J&c4wxV|0ibcb-iY8x(|Ps`
z$0{Xdr~Zj0Pv{?*yTeINV7`Pw9bQ9eX6J691sG_2@|~l?t9ysjTvTG=r3L{^`rt%h
zO#N7Sk;te~rbF<kQTZN+@z{Uud(mwT0{~1)U3BBMCo$<uay4(>ea~mEz_z5|#|$Nj
zW}xIHi3QpdH1ho&S-MLTdK_Xfy5%O3Vn>_AYxQFVsOfAJ_`$MlDxC|ZHrqFsEbX0M
zX&SB`Yw9AIQ_i*zvSx3Kp7Ro33i#Som0L;@yzPSIAef*9_WfMfEUzNpi2}5Fuz){=
zx@pN)^|Um-z`{tnX(t1swQ?kF`LL+EX=89R{C|U~m?Wwl4gYCj*5iVokt~XpMLG?_
z)jK11`43cfj1>mCtDcHM9|^(MR~u%*=@Rn<q8ylD(cMbRl%uUPyZ_4}sdxCR3@XEm
zf$rvAk#~Pd80DU_?XtZ|I(YZmb_*!<@@W|}#RO#yW1@W1k6jO1#h7&_#6Dp1iDXWC
z<?Jp7Bpr42HaIfxr9_Fnd^yT6e!@z$v5e71DUg|<7<a*CSs8^ot3;hR8)-f9k@J-R
zR!PkxE<VEU7EzOFMR5_T*TQ=G+N3ghH?3PcY!$0TPK<r-8<4}>P5&?lSpshS>>-0V
zODTcs@sSQE?Q7P7gWKc(`e%$e{T%qut5Zm*3X`&HN$U~+`SBLiKxq06(E#8uGN*~~
zMP_RM3%p02w_cH|us6%9e-d26tSDJo>aG4c&vrQPf*cM{CXK0VrC+L~Te)H9*4wa#
zQmeh({&Dt2x5!VWuj*6|sK76PgXpfD%=BL;3#5M^_DN=4m+r0g^esFv<k7fx^xkTO
z7tDjqW)1(VXQ$m{s4Hv5ipq3|Nugwdd)&@HZ$49%dqqc>;Ku9wwT6Rj{4~)0)bw^g
zHxM-RXXXkO1BI8hEFyt>w(2B6Px5w;j?arjDom+E^-t#u>r@d90Q0<fo|-Q#l**3p
zYWl>2N#u8WYa10yY#o>Tdl(a=9CO|bGG<&GoF|8s>ltUxrbu(kYr9m@vS<ydJ;qrX
zTkZ#wFx$p={WIN^CCyp~pqQim9glP4HZ7Gfy}=$DiJeWO4TzHGphy{Jti?`nA4vWp
zeeL_xwUtbe2%4A-ph35+IvoSQx_oBe?Zt<I+hj>QK&M0TjJ`&?bYU2MHdfQI`DbMn
zJy9e3?c)G8tGR?XzCLv$dE2G(b@(*63=d2p)u56K&!}!mM|QKW8*#sw9A;2YQS+n>
zx`5*K>aI+1HJNr_WDm7~L2vL1Gam{>;QUs>zx#P4RF}pK0CN9JSP`>He!5+Vl0!LE
z|AyW4sKaZpZrBpy<Z9@n?c^$t@j@f?fEDzmp-IWG)(-?j<2-U1qIz0=S^Oo!U``GW
z0~(aAn2HVsL&n92iTYWWy|yf0Rr|6<=aO9jwS2)+4L3pen?_~{xAG(PR^vIuH_eNi
z;Jq&N%!1V&Fu8n<u6{hZQ$XmbA3?VrPS<O`(_5nox6q5`zWsDUEcvnLx!F7H;NSJ?
zPTutu2`YtG8>9wr7Y(YGecxaO1S|1W<*wXqMG^8U;2Xl~kwB_8^R6@FhXa$ujx61n
z@$j!S*oncde|G>EF^eleRAy{XYk(~+&wz;=_K2k%pc8D8xjTC#e1i%KFr`nORjf4q
zgbD>Clo08n(%BAL+^E)8?)_asTW4Z9xP&rcz`b{?BhG7w5|vz|%se4CMDD^lV_uu;
zHnz~I9AO<HsNAgwU3F*ZY@_7z%UKOw3#<uhY{7vJ1mStJS#sAwVeu2P`rWP2)uD}Y
z`SLOgaEfBY9``GpYco;qzN=SlOrqR_!L0{t*+GGsaCI$($IYpE<K<Uv`F5jEeA|fo
zopPxnsyJ~Hfgzs`-bdQ9f38M)3f}BTGr-+3$2#B!V(JgP6C$Hc_-LuW;(d4|QQ`n5
zrVw{j&RvFrvmlB}eC1KLvF#y(MxQ$Eeog=<fD<S+gaB-&jTJ35<cb9~)&(4^$b+AC
z86D_vy2fuHS+cqsAeV2U^RKNk{3@Ehj(}_yUEwJR6mq51kUMD*y<^)I_v7(u+WK(P
zA}>;x@I{+Dm>aD^z0{2<(E4XTBPLFqt3dY6pWD;mg4G0!8jfW>%FkBjzvp5zHQ}5F
zARLrS5L7cwYl#i4c8F%>eY)XA6XZp8-PX*aSyEhYm7upc0H@amwYYU$%Lr@R<8==k
z*F%FaB$-r2%rOAVQJ<M1mz3@^pTi`*eCk;sM}LispFTDlpJ0GsYoz*%y(WyX;+5|&
z28tUJlZ36Ia}6HW>d=i@^DJx~xJ=OmXzqX1^I~Vxm;qf(Dsk_|Y-?MymXw5XW`MtH
z93CFy3@f=`U(?iuU^rVwVDnIy5bI{>QJxlNK(E^xvrT7xQRk}cnUc3QF_TM)1O>r5
z_f6W!_Ik4{r_rzV<<6_}`>ZQkyvS`D*{u5L>K-R8<0d{?WT+UP9|mW%{5IMIG#|xO
zshW-`Mwr|p*X9^V?>>fgA$#7bPYDYcS{_)P_h^!$SFZ2ZY*o}TQYx|?nDKRi%fm=*
zR)r>0{Cos0^I$+-!76%X0XHQm+aGIE1be~Y*P8W;^vTM=wb>#lR<g>AL#bbBTV9kC
zC9@wz=^`{(YtM-iye_#0&vyv`6D-@E9LSD*(=k90u6(?wiLaB1(_!!8-pTbL$NtM~
z0dO<YU)YF6H9O0M9Xmq4hUgpf1~<{m=kZ4N<*+x-VnB2?ax-;ONJT!srO0%i($*rG
zj~ss09DEB`h{VSdgU?%66LB*1QV=pbPMS)=0_5oe3*)vs$w;8g^L{eGFN~@HDzf%>
z!mn7-efYEJpk>){0iO@{L#5Arp93^Al3C*D#A0F-Au!`pjr4(RQ+*Re^SgpvI7TDb
zqlE#g*d$R-f11u~CN<?tGMUW903U6YKDFAM9M#JGKGT@cv!;w`0OQEN+O=Rse|@sg
z`S-;LeU!lInCTjCKi_?TV;+C~$U>sa`NwXK;I3Gz!2l-PaSTFI*fK<;JCqw%8_naq
zhlgDPl3nZ|Fce4cfeXb-FV>FdtkyLoN8jT9o#o{dRG7hJCa?dksC*RTwwz4)M&8|e
z^+cZWgdr+iEn19PfZND9w|m?)YN$azH3grdJB@bCY{q5-k*hJF{#TY*(K`io#aPtH
zzc8&HyLUa;3f9wB4}Qx5rpTInR3TnZUnO3Kdrw|s3q<4=`q#R1Jnd&5NHrS}h5EBt
z&8x{T!ij?&$n`k#aqQessTi+)J${CyNXq)%oE#kHcg5Y^tLoZ8^)`r%UGI7wdJ!YX
z$llWR%2n|8ocVtM{du=OEsU%t|FYF7%y#1_^h_`RRz-pMNpqPS`tCNM+IM_}+D{Zk
zcMf_<!#|Hvl*eq-?f_lr;I9FuzS4n}`5x{1`~cJ<CB26~u=8M>eMR|sb*sA7?eCcI
zg)}*CrEm16^M|Ut@DJlj4^w-R_zIer#C|<=52OkAq!x^T9$azP`aiM-ZL6PIjih-7
zwxA9c@P12&t6Irre(jkTpIAwJzK<VG^s`LO77hAfq6?AzRiaO9&w{?Bwf7_|D^uIJ
zu<RHD6855TgCe%-;9!I=*-K%6@Hh`<tZrS^8-b~78&B1XmJ@hHEnTFVi=B%@gmKv3
zoC>MgvcqWs5zaJj{InO%>FL<l)0}{w)Gr%n$@9WZJL%jt;g(l~mRRLZVzmsX<8OL3
zl0p<D?X^-zP=t=1@mn~(K%y?F+OiuXSC~c>3xY&+MmN`qr}iC?$gW{Y6%R-aT2u4@
zc0sJ}V)O`)eI(#DHu$u1cxzuq_gHZLsl}haZdy42okisxL!#tqno{K*o25Cl^nytR
zHj<C`2exMos1f)0fZUC<)q&j<OTW^xyWkpae5qe`Iec3s%41}=(i)U1XcK5c!VhgE
z`EJ~L{}3%m_X+XknRQlU``A`JUN2HsGcPX^ew1FYbGD=mT*87qE#&6@J2x*y($unR
z)wTunz%rk_1}0(g7JRUev};Xj(io=U&kK&hI<gHqE3FG<K{~5R8SZUHKPXYOR<m!|
zl2L&GcZBIYm(Wd?hh9wvG=C%zKh<o_FmD0fN)d%2HhWuD_~x#Db0dz*aebPZC^O4E
z)UMB-d86qD4z&Y6|2F*CH&dP+1RiS#Fogh!@6R1eV*ecS_2>O#WW@}j$LcURX`n<}
z38v?TBAN~%DYTBNVd@weXdgKNhusISm1uF&`jx6~Cexv>lZE`x6Ow;Z`!W#hPPE6>
z)rBmMGRk4B&6AJdWH5Nz;;{3{B8H6_zjEYoj%;TK>ND(=U}k^3^=QGHO8~K-HsBtR
zEUeaZv#<R;ig-r9#*ao6JMIv?fw8~8WUwdwrBTSwiAr?0K|YT4Pr3^#6xLf9vZv_J
z4kaNX)mNr?{Kmnd>lQ~|NCi;pO5WAzK(XB`tPVvnh63BRA|hE8q<h12LY7}#?ZXra
z=Fl$v@rT%vNpB~NUo3ZTnugpafXNSlf<z)Sd!Q=xYOX6!z_0D0Qs<s}qtL!Zl5~&u
zs&&SG0!vu@uX__&?XVmAYw=_oX?ZiyZZJk6N8&Gw|LK_Sye$)Ys36(%E{&x?+VP~F
ziYVWHaE!p=AsLq@VGXNtNSnz$oaD)*)As#8y2T?Vc8QTU(H;zA!j=C1J3|%#LRMhs
z`p)BHNJ4OV6TXAOA&+foLn8TN72bQH-N|Cf+{()lajuNW_jIm_)AzschLuOdnD1X+
z#k*Z`kvUWm@jmm3R3N3!`(tSR$>5_{HYp+gvbwzO*mYsmlhaK{S3cUev&M}3^0PBA
zkFOoF#8h^_9($=0>YO+xr!<}bBcv*QJG=9gE|R6xY^KV#TiaW1<o|%<#Y1a3Te5OZ
z3tbMs3U*%v8+Yjf@(0}y^QN}_lS>N#xg_5NK%1xOq@eOl#9iefD}_yeR$}q&+x9Md
zwXRZijV+gGVa}&|2+qarZ{}BGzKf&f{;9ZbJom8R<S`Qj@ybVxO!^N3+|FMcp?d0e
zL-C}~1?HAGWEE1Y8>SKrGB#=vc}2*IDJqm6?{Rce*wF>)VA1dK$Giv*NnH)<G6xq^
zZ1ddhf>jSIoE5(L>Zwn%LOvs81>8sZE1i#UIyEg?YN~n7(%9XbnvHn0-qSv#ctJzW
zjbO;>==UQafWFcmrDV+kqS-9Fh_$0x{=s%nNz>S?_+4`yPdmubInLhQ?_;ne_X8WA
z{O%h@A)pl>`1eAr$qX6=fZI7Pw>Nl6Am1w?6;u!i1SqiSFY8YB9?!ncP&$L1!kNg4
zV%OVxQ>a!uuGo0!x&o`w2u#@Ud}Yl27O<dtsj@gu*~W*eh$SaL7;nw!3$Dy?>BnMS
z)<a$rVM<~_Ek#)^#wxuTRnHQ&XUzdd<;f*xoi1$xgb)K_H}Aupk?|;NYvfwC;pqTP
zS0R3huFaZ-M{{Jjhp$lQuT<P!cD607db}j!(J*jQ3?)(wAcuotOQQ(rJvp{6$a$dS
z4iU^Q;H%3z+}$&P)2Yt>G9SL3Fm1#zKbqhR#R5?1ZY0pxGQ2}ic&Gtg>W9!!?uGUW
z5G7RG9?q@-Fxr6iYF@ZBQXUCMAF-=Lo%RBHt1K;EM^^dERY<q0AF9F%F*hfP15j0s
z$YFZ-jjZhLJG*XZnp$8vfosDUXSbZGU{08j#Un}&xYsYhE(Yd%b-3mVRI4ln5s_%1
zfs0U7DNdQ%#L|66>QLv-9*TYM5MZ1Fg|Jt>Nlrhiy0Cab_>#4}`XQch<#&p@TYBW(
zp=>VrB#OqbnyFKkPt$Jm>gPZwPH(Up3S4U-*Z|>j2lcW-Ee8Y;qNeLGI^~D5X~&&R
z8l~gT{@^e`Avcz_)pzuRoCxWe+kI(>aHd|q=|2XFezC=1VV{CWS%{uj;csjH83#)0
z&z@R&HR0@{cy4v@8yXGRAy41w++&7_oKvu89uSkr7PH^HdA}f?)u}Nm@2UAa)5@Hc
zlgMgS{s2CZ<ZZ=Z*F!-s9O>vJMfwTP5?lO-D;FaHufpW$ik<R|7-g4(z)x*(^P`vt
zxS(Egwv@CW1<S56&EI^kobOI@R-p7Id>rby(uME%=bi@g)QgclgK^^aHPgJ&k3ApU
z_?45Uwi`4?K?D)1<~cdypntS(vN!(uSI1p+;UgBl1xyWs+FIbG7ZPFcbMaN`>V!=X
zXgk&c-htu*4IOWaN)ejEwV}O0F0ikSkdg>mtuO0<=Zp#}{lBt91QyJ3-Z}LnME6Lj
zh+&ABh?~p(VB+F3;wma)<pxl0X=zDRY?+tdsz}03nR9Y_qJJu*FK%U`B}V#TUs}-C
zmQ7B3S6Rv=rW!-r)-~SM)~IGa(d<ZX+Z8`R=Nj(<m>6;o*w#(ZB6Qv)G^D~VE@ur<
zi6E%26x#hU*|Y4v<Mr_-cUk7>ql;{;z~OWl{N?8?CK&zvAO^9M4$Fk*d_MHM+24Jh
z>nlg*Xh&(Ui-U|ThP<Ye`ptIq+NZeqsYO;>5YE?d*?jMB!&oP*>x31BbGvlfy}tpg
zcvGRFQ^*WUleHzcK+KOgmSC_`OUS3JcGnrvZ^vtjqj>+R_mtf>gYLW}*QbC%In3C(
zi*Q~5+88QRQ%lZNQWK1L)<K`gIkD`!x?cU?$q|j~zsgYN#70~u%C>CR_UP=u!MNsc
z-Sf#f(xDsBEt8?2#<jPn+mgC><$yUGlwlk&5rIw1#lx27wrpV6m7?n=@+0m=Jakow
zfwXL&DA|5LVm4rrw00cJP@b4YTtSo9mMb6JLHEgVT3rqg?D`do+_Ij659aY0ssjtG
zZDCmZ*sy$p*wCSj5^vF}YVD_18CSE`BD=)yC&OtsySp(B+zf$cXeCHcW`NH!ZELie
zaUGy}C5Jr&<ug8QWH-7;0iN9_&^KRRN+H{V)ge#cO5jw5;p=8+is{?q%gs_!Xe#ye
z#B^fko5w3q2IN!C$5zt!biuqp@<Ra+mQkB@F^8hRP5AP;ncri11TIOUV3Q*gt%(rs
zm7^dgSMOyp&)h}$q}FFn7%<f$xjSQACBU~}zjMDWdqe-4@y4f*d7(;wR2jZ}GJNw;
zNuX&|fOA0lE<0>!jTtzsi6zVH%1!q2mnpTe#*sP*_qMyuorL_Oc^|GHa9D;JIlLfT
z%)BG+e&PXQpui4`cd2nMm1(we66;;8n9PuO5^E|3tZLgCBrvJ43E&_y3Z<PKUZ)yj
z4rGkh-)?GaKPghAUY<=LW;9Bk@3{$M0b-K9jlUt5WAU9g@VaB?Vka0qQ=I%46K3~U
zLnfD-D~!<wD(t{H@O%UBwdQdV^NbHs7Kes^K5u7jP8RwQ+J9RxTlOJ`n<z1d3r2}Q
z+|c->4*4mtMm#L+8Q|*9V;)3I6;h;hDrGE*!__x=3=2L>fYAlziU7HKu>KZ=9-goh
z#?RGq#jLED1OMDreP;uuoRUNSFkXl+bHcpAX-}ysYHC>eSIMqbQz&C#S&BP;qFzXZ
zS<f&@ivb*iF~2Lx<+;+pW113@QG3Unp}#Ypf8sFLsdJc`41oSWkmH(uSStP;f+<p<
zxxpl#(`i0j<RIvxr_J)Vu5rFouApUFRW9(N@G3_xE3$`d!ZG1pVD=BR*7yhlx_enP
zR-2F@K8FjdtPq;Ubs|nSCiAtkgx2MdRJl98@|%dQoQBJj!`ha-dZI`zm+Ck~xpik&
ztVZ(r*&lGErOCZxvJIhLWdA7&H!!}4XNun%hn}iW|2eDXAkSye39)r-ynS?mEir$(
zlP6!QTFKTCopLw)Le`ud;>7peKiosx62qkcXiGHg_5aL3eA2H!-=@;84D7$H@?~-@
z8G3*;7{+Ku=yC9MRj#``&>o~)hXX-!qh=bFL;}1KJD`ozH|Ee>f>jU&Qq8aI?9Cze
zCJ}AXg~yZk=MK3YYT`$;27Y*!e6PW0+KP<mX)B{}0(^`fb?TJ8fhzdfCilQkE6rMU
zRJ%92jTADt9<C>jZ*-HS)f6NDrpIiyF2_os4Rb#i+cMuMd{PgtzHiL>+gbQeGWTV>
zlmUFOwnhC$$-+XE>&aG7D!QVc+Oh%z2-<b1)U2WD!egJt63e&W<6Y(bO9`IWW^gb4
z2b1(xw=?hO_j&bJc+w2F(QPK1QRPWECA9*&+&POpw7c`xVfe3(YnDz~spZmj%a_;Z
zDVbZ}j65{9A5`|iS<?-pkscNhdA;I?&j5t*#UnMOnebv)V9Fi|0`U>yu6o15af>!o
z{h#zCgG*(h#PA5{{o|T*b7`q}tlaW!v25!F3kEA65dcR-*`_Zv!t%=J`CPQ<44qP(
zYVp+zsdp6bJZyyp^PP)fJEZV#MR6h%5L~a>yV9e`?&m;S{;MPM@lKTkBLW_mI6(aI
zb4qtw>PZ-Mz-Cf=u(4-SGCY&{EgQJfKXe-Bi7C35OP|o#DAT9RMa!n8?sy(%$Jf!n
zC5dfVbYC9D8oX__Q<wdii2F;-0GL@3;GuE=6GZ<*y&pHk!YJ(N!KbHLn)0n|`KC16
z_OE<Nb@(in^WI`s0e}IcCr3MxE1)yaxcHGgjAA1e?BV5v#@^SC@e=E8;)JG?4M9U6
z8>p*X`wF@>p9E|Zx8^SS@Xo`Lol%{&t-RNO`Q}BRkE=W@Ng>NPY6!A`*XdU+Ll~+k
z$-^q%^J#9qu=Lj>KTa4@tA=p>ZiwGj;+0iEIQ)T|iB&>R2|^VgamTF+E?}Wd+-FII
zHeZ7qEdIG>bXFsP`rOTgbvfG7z^I>nm?R#&LWbEakhCEh8u>9lx9zkarot0pW5=Oo
zJ^euKDDqdq8HfN0h}ueiMNR0(?_A`PFk(SrY@O@yH=h7kaUfwI(V9L+(T^ddR=iij
z<A|zSfh$vE(0z=^O-*FC89>K2FBeJes5jw6{nHObjEcC(elk8$-*@OR5J+=7hJIwj
zK?A+=U<+?m$%bTD$%YxqXOlSyXklOlW;K`2#vljc0rSzrISt)*{L%p!#F2K5<+ni#
z{es{sM{rUfz`h=X48((TGC<C-i2I|*!$Y%d#xSDv4Pw}71&1W63+S}UwP40e;na)@
z=Y6WDzf-E$;it)tFd=#3e&xcBh;b?8a6h#k{hlA8XzwE1dg{iLEq;k=w+rx0K%=Nt
z^7g^Li>dH3vQS-6kel~f_7nigP9n43aCL&4YROax4R7xDO!r>#Jew3H^;!6P!V&vV
zBhq63V9Hon>mbl$3IK%f0Oew4JQ1wr>XN|)0v-34)G2|7B*c@K0A*5}i)7i(?tNo-
z&FD1j$osPMob3Oqoj%J||2&e<44-PB))&NCU%Z-H0J6D{<4pgjbV?jzekvE%=bwju
zi&Gis-}PF{#)iEz<9Exxz5~qd9UKb)y%n|a)$%Z6`v8)HH7QFI%Wz2Xl=zR1GP-0t
z3G$T6RnTK{p*Q`3VHjju3#${vhq#Xza1SEg2Z6AEAfzSMr>*l}`lMP@1xXBJIYI)S
z`g61;J_P{@!)arrwG51G6}Ka$A7x%w_gj$!XlXf{8K2VolWQ?6Owg^`d|eO#j01n{
zWu+3w^%M;}l0b0>U!o=T>1T-f6$Pq81$kJ55dkrTRvhUc-3BMSrb<<y6#5ULXCx1+
zBom8c8q+!McA+gVH%ib1#?2Qk1qNqTd?FU1kvEn!ZGqWwYKUHmS%|tQ--!ojAHj+|
zBque^n6YMum%sed#v7HU10g?PkzxlvJ8{}nvB}-38g3w)Nyaqps9rD1pFiE~zYfKJ
z3<~zd&pm&w&mL~Ia)Fx>43zC{Hr&TqmdRYw-t&>`qEuU~*k!*@9kj2t5H$y|w`GN)
zzgo^f9!~~)!2*)L-_*d4F`(wx8tX93So_xWw%D?-jo$SLmcKeU0CMu1gRX8f9<RR+
zc66F)#JbBZT<uNn#4D=*eO3=J>+29MTqoBoDstB}Q~Xy6Al||)m*Hp2kn36KB56c)
z!(nTuTUli==)I|t+Od?GXwDL31_|~8NN;9;JkG#Q9r>gX*7fXBg=jAKBRtRHH+^|w
zT@eK(gL;K*sJQ`XTq&H8*ZS(5@oR2QpmV<8n2l=X_?6JbH7ydFVIB*?x%$}-;vz>5
zxaJS72UY$Afm+FJWX!FMXno83%u^f7ef9$;jvZRdBLueR6k}~QU<#I{1VtBp_0JQ{
zVR(j5<F5fFoL>e;H{y%83zKmqGC51o`MpnKuyfS<^jQEk*hbiHF|RXUiLs4%G6o^}
zBjp|~sd)o(r<M4$hmn`D?WtLkC^n&H#x$@g{JLM|lf{3ia>>Y<^jUlB`OQj;Immxi
zx9Qvva)~N?OC4<d?kEkG9&Ws}_`?zx@C#=(aT*Vg=fZvd<M)?_`#+lC3U6RBK-I4Z
zd>HjZ*a=#I!0oPSzf-*{%&q6rs1KY_WM;HHOtH}?EjCEu>+AJXutb(!mtjO~5Ta$q
zZQ{svV(WPQEvI^@7Q36MbD%iw+v2!$kfDemUMa55&Oft1(hwBUZ3LFAmuL_wARaOK
z5`%Z)+bWPI^DGj?A!!CT;Q9Rg@tm-?+X(i@QH*tf!9IG|q=7{Qk5H*s?|j8*2|pFL
zH48$7y3yw1gr%JL|DqQ?w#_xaDz@f)6_JzXK#e9527^m`Fr7wq|B-d+%dax+%Sd7r
zmp!=ojt29(uvaiS<@dKQb@h%SYAWeU8M~rb=GW3qyZQhcv7VF;__qLEm&fwC?!9KU
z6o6^Yr;Wr_f?Z@i5Mc1E)RFrSt&5we=RbqI7>bJ+HWBxQY0QHfDjEFOFjf1*GXY7t
zGwGS!iS7UKE#@m`b4v3UFQ5MyT6M6i=HPOq8Or^}^StZB8UEWY6Z@ytW7Aeil~aI+
zN`uK+r=N3%e6>&?2_Tm5z~|$rislI*KtQ6`2+ACBrde{cYJ(8)TMTlr603}RuvpXL
zSkmi?7&dodork+SPyafOK0&R-iC;1W{wt1@g(TFZm=uMg#_pPbooI&e$A(QT#OLwy
zSdgPhwro*3YA~J=DFS?rZUh(DT+5yuBZ+6<N7t=dzEsI_g0@}JS|)caYjhkSfFm7!
z1d;I{{@6apke?){Qd+>Bw3M7qLGc!%V_uGa+iocrMHMgWCSxl7#KEWZGFNkU!=XJc
zX&O8J;f{ytAMOBAC#@P^l#hdBJ+OSkJGqQlGXHaw)5AbEQ;A+kkMm#h(2u3(r+?JA
zvy)JpGA=EF0S87Sfei<23tB}4Yax?CIjQ`sNVn8U@ix<&(T2(Dqe-EkosygVp^w%p
zmO+uTMbwM!G5od5RiVm2HzpiE>`*L^VG!~e>Y<BR?P}qr{QrQ*Ma&O)H1fc%WfK)J
zVb-Z+Vc;>Bk4)O4jd!p+n$b5W{X0g46R*sHtlv<`YgZLS2Z$zKN`QEJKO}PqaHP4y
z`ItT=Z{a{z)xrU7FGjDAwA&T*`HWR=<SPV3Mk`wT=i@8i)dp`b`u=M32sQdj^m-QQ
zsI;SFkJ^bZmwL<Kt`qM*9Z|5AEyQ_LLu6uwV|yu)h{4ejl-q@SWgV?Na77aH3mPGm
zhAHjXZ^bC<0D!6LBEHu0cTfNSVFRVd)(W9p2R(sfq}o;0q0CH&JE!3SVo&BmIT!D;
zUa9e;ds$35>4;uV2I-1*D9XuoTyG*V@SDwZ+I<)TO=5lt4ysX_3MJpd=!dNE|4j{m
z{}(o-6k_N0{lBn*4Z?81^#>d9yKVmuH7w5@V*G9E#s9x_Mt0ATYFRi66>qYY=~(H6
zfkUysD=&#|ssbn?g%yrg3EPd!@Aa+Q+1Yr3faip~!M^pJCV>t1c@Fq!-+PM~5n41U
z-djk^8Y+q}L669ABpi)2Q#$P{Lk${`e2GJo=@4s;(m2VR#@6YTjSE1Gg{v}7gyz0~
zz6Mz@y_K8d-or(c;?sWdm|H63s0_5SpE?utGsG4SFqDsA5*X3q1Q#IM!Y4wxhiG&I
z5aGR>6ZO9fb~!C0n+*>E?pNNGN6i$%T|^~}f`O+~kI)~jLz_{NhhXhZu!-+j=zF@5
zrf?CHza}k+Me7x>+y7qQsDeMnErF+0<Hk6M;EUIlxufNOt5@VM1hpe}7vtj;rKMvH
zJm?}(Yxi}QrD*_B#v6KL1-k76JpRiHP@_5R_J2vvgnoR19}()<O_W7ruYjFcWJxi%
zDc?+XrLj+}S*PblhwI$;80qf_=)lif{{+s*mQKVzi)M4FKu+eZ5MVvqmT#t1ryf}8
zY@_IbZ=y77czZKpRV-X~FA9^>DMiecBN3?a@z?{?$Py8*rk2YB0E^0W!%P9C=+!k;
z3N&vo;+eD9B+|i0Uk(TOxK#X0t-*|=L-3~~$SQe*V<FyovA$a5xf7ar-Hks3?G|(n
zJQQeq6jJ3O##AV4#D^vITO~rKAzcjxjYH-<Lqzo4t1LwG+@Zq2;LxZgbW~VSMzx`m
z-lfysJ7mp$-h=l891_NlJIi|5kjR};HzuHV3U9=UpMZ%qAOGqRj<{Z*Sve&1S&%TB
zJfa&WDT{j5xzkf#8nbwR;{6LxC-QZ$(H}l`0iB+UhmAx(;}OrnlNYmzszUq1fhbWU
zIA*`UO5%k!Q5vj5Pz7a3PS9@M=q7{wc$?kfN0Gd3pH7McXuvYQUnz(@(<i4Q(mENi
zwd)yDk%>QtGz#MK!Nv^2-GcyrHcG~Eqr}jA2|Ofa1~t>xodQC1vo~L;h%&)L8fcx$
zJ&JS*lQOnn$2+FA#?;l~^-a1Zl!4;qxRRsj)oWGGYZTkQ_V~2aeGA2dI`Jsipq>uK
zcw6dA4K_Fd$QabDP{^V|Eeyvkso#I4U`kK2&D$btR&2**EGA2K(`|+9OZrW%RFuMy
z4Hz$yl2EZJ+p}h(RVYMX!M`TFCP?Kqf@A}6&g6DBPIsOfUSUpU5LVf{e-Bj^4_uk9
zrs>vc$iPM7lCaWZ$V)fBmXSvmNb@n|E_Ls^0(8&-Z3lMscf52bv$LFg*(%GH(Q;-z
z#|@fjesF||WjOYOiiPy&^Oo%=<J(tm2%?$5Ud$oe>l?>+9{E`pyy0W9JsS94o493(
z;e0K;Q;)kS)>=dVp^S_4@{KHg;uq6;3*8?J*3qxkRm(7x4LVa7xt+?(E9E0Wru4}z
zbQxf>KzWRuXfcXE!WFxamG3DDCrxv>&H*aUX6^fCJsd3Bn8Ms+ioWHmGnw=9(d4Zq
z<=k{JLa!mj$aanTmlk8LRNOwy+#k5@^0d+$Bg~jk9io|DIJfI95-<Ep7hc@wPIGch
z@Vy*fJ=M44RFWSPNu~_=<n?KKZsQMgUMN8O_aZ_pAK~f^b;V(vNW6%l(@6$oh;=)=
z!b)E4NfsdwUB9;&yR%C}#@P<Xv)xqlfVK7sCGWYvc&AsL>$5{Ou-OBu_gNduTcr##
za%lEGBd%nZzM8j#CFOT@hYboVL*`*&9+MCh(a-jGKxXt%)iZ2DO}mCS&bc>m32^4M
z!0@H!&3mazn^Z}(1iNARXjF$%$_UnG3T?;M#@6zf7Hx};nzCdFheU`{8)-aeo)Zd+
zbY1;kACIj(2uGaR^vRRi$#G}pF2wEBx}jJjRX8|#856$P)frFi)T)%6QM6PuZhKcg
zXOj>iTh!TUCQ=jcS&6?{4$V$o4qz{qAHAL%h9l+@Me9f84LiIsPB1N%J&HFuO%L?v
z&akk#`!#;i_7^Al&W`SZKf^2)WYHxAzfc|P0)^yp_q3t7y7I<>MW$`pOeHPf{<n-x
zbF}&U$9&`}Kj9e9Q@S9)SO5BJD}!L>@bnKEnI`w|gL$6?{0)tuU`Hsj2LKxICISbh
z)qiyRAYXB-4>*AZgXT%p`}ktqw1$Q(xNX*2d5zHAT$JW9zDo6kyPN~cm8<3Yrvavn
zKzJ~nQpaY%G11YAOTx_&$K8Pwrh#ECaiZ#`wYveF!KwMCb(qbb5iIPxG0ltN$F6h9
ze|2Z;Z<eO7@*SEB_C3I!KXWu+UgG!e|CXH1-&}RATn{&q>-j&7PQKp@rh9z6nLJgt
zSQ~2F0UI{YZ$HkvEJyjnaBX{U9I)6Q+5^+R(C}d!cJnn}H;?fN70j0D!>p7<HkMeh
zKjq^*cjuZcZGuzLkH|{nK~DREjh`11ZvYoYq6PEl2Y{S3{AbRkMFwMQQux2Yz@G*r
z`N!orjm1()?%0nUHHWU@3aQ)1S^AtQvbB<5@+Y-XPcxI6J2X8>0;`cEGZhfwBS5@_
zk1fkXj5K|zzA1Q@jDBe?Ewv|g$&)M9a!oFc6G2E2`NtMOIR+Fu1avN5mOXytJuVEh
znpkU%3_!896^`;6H%=OUU_GUnMU21W?#K5WkIZ#9!i;R2{^t%|A&TMC8Cq4yiNOo#
zSJYzJ20k))r_|%HhNiDQ9wT%OwvP3vAU6A*rzxvulK_*Lh!SY*5}V&G<kgto4P96o
z{Ciy64Bn3qTZAFKxW{s1BZqs_vE05Q_3e(vA5dJ}v!vd*bQRaK*`8RrH6qdenl8Cz
z*QwNgO@E<%Bv<DpkSMpLQ7O;dc&e6<feBOtd6%&Wh3}UPj2&6APYclMDHHVEoRmq>
z*A#^MjEW7rzOp(zeVN_R?pJ0*9XeBctx^wVo#S4`nUG7*3}aIbM`Q38^&h9t7oV^K
z0yxv&J6#KSr$fyPUHDtyY392^<a~mrZ<X{%2jR~F2?uaBWMvB%bLXj`(#{45-aXQw
z$Fh;YUzPii73JHUDN!nDs$P>0fyld0CGp4NY-88fo3-E)qh-!!N2;FW8}v4%G3cUs
zFku@^KrF7<S&QrAT7U99N>xU&Jwkq$1K4IfemaVx?G!uHg3TTELsv--{EkKN4Nqdm
z$0PNtBhW=E=u;F&p!>8vn9uN)e_p#81&sF)*x;p0!e2HXIq1A(1F3gHrDPh<Fd(BH
z%9L$R^~@i0Yu8BPs?x9XS4%dBZ<nK0+*l@;<8Yf<erS0w=L^H2leHmF#H(AI0Mz*S
zD*^;z4_M&P-a8D%NuTNXNtu=`%IE9vub=AjdflthGqtr^FS_T^qr)eF8Zy8ZS@(Rx
zW|K_8^-G%!jko_zi1zfBy~=v(EJsqbk(w3H5+$>;nA}=s2<jhc+c7`GUlN@<7z0ni
zw`EKyvo50z%Q{+gAaH}hW>`*6fT7BEsCHf0;2W|X$L5pdW}QQ)KnzP_KW|N~sWyHo
z!$e}^F!JOqJu6x!Vl75+gGi=ow6%1Ldf}<IN&*oTg34%1e&Bb30kPa<4!TZdfVPN@
z(~VyN-rqHko?RcoE4R47xWJzS;aGO`F#OH>`$0(8P&Cn$|8XgeyFCCS&bYr0Nd@Q!
z)1!-*?gh;Yj52)OqVfc^%8|UN2JBO>)S$kL5hMU^ExT=;#)z!q5qNY1PoF1Bd*e)^
z7YD6R!%(3Cx|hR>?rChYjn_lq<Vh(_E?M{U1(c_rlt&AUz>WochDt6J)m*l|zTRy!
z%fKXZ|4E*7nr>>6xd3d+M55V-*ox&-|Jz>j2ia5YklrdCF*B<CLQ1o=;>A2!U3n-d
zWV`@PmSoA3pD-`hB0#aHi}lKAO+5E7XO6cv%lk${jYq@&2>|A3<>H<e%64!tTTV>4
zJcrC}BneDRBfL{Du~S)G;jEK2NkN(>&J)9(rpcVJ1vuCNq6T8dQDsutaVp63{J>|{
z{u2u{qp?oxg4u|;9)zDnUK;|_PE-DDa?k@&5Z!gCP6tMrOtR|(>?*>fO_HUYy%}x(
zCkWV4G*=&T$umb`O1rVlpaXro`%t)fwYi`mqx-rZF8SMctjGPfcMOWZ|6aJ>8391X
zAkg39y#Y|Pa(WkW2y7CI&!p^&_aKj#7iqVfU}xD>LN})So%c2@tfOpY@<sLR^>Oe4
z=>btl!jp-6n4xH)4_-mquAVXznYS2!?)Z%%0vsS$_=&LHdI>ZBUl-%gY=%5T>X$i3
z!OYds#kOj0tVe6RS=O5FgZ<wV`HCf+EpeGwfRT5?It=>7D<Zv}meIRLTIJk$-A<qH
zG~|2j>U4>0KO)5FVZvgunZ(FHl}K=eV%&HpIL(S^!#^Q)><|c&UF;OG#aQ{ph)C~-
zR1hgf1|-Z+qr=t=Xdy$2{mO9u<W}WqgENjt<E6M<TfdDo{QOfaV7cpnKw%kOO~Vli
z06>L<p&&M%zfnc`O6r&t;?WY)21d|YGP(E=f9xmbEJt6Gf#^hBpuz5=LkM;!hs0BO
z&Q%ixRsy8|S#6-gsM!>#TP)GJI6mqN27k<QS^vCTdgS5aC{dApZ2V*g9yWDKvz;XH
zDH(`K9mM|<2GigebIY7$Fy+>F0;(G9YjRi)1>W&sV#eHw2kA^90%Bx$K*;(=lsOjr
znHpn9V~t(P)|LY9F&o;D&+D)PJQTzxy_iS-F~pbdL06R53c6PW-y3;c3#U36Hqx{G
zZn}Qq0T(`DO7gr7^XA((8+uG2l*<xfkfuBa4`amt5AT~A{D=3oH&_6~sYM$-sbYap
z0d%2x>Wn}mjMfT<Dd<2Vj5ovcQo!Vnv9k3xpM)`Em3-HIu3!OE#bl|okwoUhT?B{^
z#9A4k`_~>k4C}ZCPZ082h?mnfX#r|<&1{9W(am4kZ$)Y&8E=8#s_rrmDG)LbAt@(y
zoI5R|o!J^hSw_Y<1MUDPy@jaf6?DSIVfS4RYP$I@P{%O14DTC%Rc*GNsHEBqn*YkM
zWFDW`FE2d*-_GvH6ld6Qzu4~j24l@k?ReNWqgBxq)Bw!pQ<!pLy7PzBE*#qv+X?K-
zHGUno$7qT#f5rfft%sr+%hmgU^%0~3gx>5pTwOBFA8w%{f=oRr4m;Ty0mxg5<!&!3
z*;YPp(h5H57S_^NeMb9o_JL7)B_?mO_`WZ3y5&35xk{6NKiR?^*!S~Tp=CmOX2u4R
zCQpX>n3xB8K6eI<z$)SXjj?SmN>y?9SSC_m+fQNBi9eDnDL_pS0C7(;xv%O1Zgy3i
zY|%gIdQU}r?|*-EamMhZ{Jo4IqKy>zl^>i}h<ugZ->ukYQ+aGU9Q8p&A~2NTnYIE~
zO%>xebEF=O#DFC1%1`!KBEbD8Y%s=8*7wU+RSV>JH{hA1e(pRE%wA|St4;xz4rcF8
zy36Nhz9(h@)baWTRb4z)9JJ5{AnR3ijMnJ2D%pYzO^%imbmcCQ+}Azlr*W(3rb-7#
zZTzlu?1NLZuDm+1DK102g;CDY@e{O86v&L${u-aXfPMQGPpwrly{;#pZ9A0)OX>S0
z@yZ9~(1lLa7?k^M^o#kqgvKoRVu>V4&uf_H6UI$1UrspF8Hyd+O%He<P#w7Gk$eRo
z$!i+;6Y7^ES*GwnZ$aOgT~gDMldTJ;T2g>LCchEoGz&c{{7*MFp*I(&QQ{2x1sXc9
zrWN&G5=o0)T90?e_t~_VT$>%{d=z@|o;m!E^z0&e1HG>-+l3_sBXtoazqREdTJrD{
z?#_ZSq#>zMT}zjUk@<Plb~<GcSXXq~$ZInTt0@(G%mZAQyTE%t>+;L*FK_WU@27#p
zHCAza`GOzoyNdDUPzN{BzqBY#UY;0U19|T1Bhy1zg`PS@%JfkezK4;ugdFbTGQvMs
zb<|RHC1Yj~ZzkN6FY%hEE!~KV!xWw5?zloX*g%glj8^IHvVc`p;7l_Yqi8~|1~c-{
zc}UPZT~Y?rYK0Pt_f(6whjjkUg#`-(d00aJsbZQ;?WEcsE^^Sv=m|{J=AWd@#zNv_
zyS?nyYAD^S_6wP`k;yWKA`0Q-%0WH%lAJc22LzJlZk&dJ;Yo6?rzu#{kGP$3;-jew
zLW}Oo>r;~P2w1~=c(x#GDIk5zs^=vCk##lyCYWdL$XGWafI+N2Z%rBs_MQL>aApsu
zqa#kY1_-H!uH^Za@Z+T)luL^Uoz3twKdZl&NAHx(CT;OgedsP=b!p09*P2i70v~!y
zAPD4_kuoZi(T0*SPu84G&^9pQIh+`P4!P8;`-67g00sz$bj2ctb8t2TpR|||pz24c
z0Vxd2#Cn4em^x;c^SF_tvCQpk2W5Q~Cicu5%B3`|<GPJY<SPv%t;IDuq}Lw!35!4E
zsw=E-)s*FH@%2;HrnSlf|I+7XEl#l6GhrUta656pMc<6IS9t$3SgUU_4bA4>Xey-m
zWg<HZs0%ZaU5fg4^3;zrDR(BJT<_{M9SCK~Py34(4?;5pNBdho_|REf^ljB@QQ34a
zOqBY0Wy*Ja+U%f4<dl@nYRJm=6qRNW+#xi8-`c1?!s{X0t@vp(I$O;~d*`O$S`)ub
z*RLDk+Fx~j<^JoUM@1FcvY*$Wvr&O$^}vh}AlQQ1JcQ5;2Gn+o^0gZU67hn-z&hoq
z&VaN<^on@J%<+JL5P|$KTGgGOck_6DM@|W4Cb`w;*w2iRBobBsYQ6lXXAl^jUh9sX
zl{u7A9dznPls!-_j;W#6Ilv~PABeD~_=hkDG4>58fL}vpGBhjx>BMg5$jTB3m#AhJ
zFu8>Eyftj9QPU!MSfhGZLW1i#PwSkC!l{mNc3Nr~Lc7Q!|K6vOcT8Z)ZNsGB@k@9;
z9QkLL_;SIpKOc<|UtO@6%xq0x<xTcg{|u>fh_AA#<FHWYvSl5<JUvkHIdleKB&3gX
z{>8}GHTb~zbDhfHal0c%1V7r2JQ%?4lc^HNeho8F?q-618|26xc=qg4A*q7?0tgtA
z*R_(6>?-YR)b26a?$Kpy(}BmL(D_efdLuGw?oIUuJG@=oko{JOOn=9+Ez|JO>B7kJ
zze9P#roh1JX}`7Mdk3I9Bz8KIP0nUQ#?5%b)IA$I+z9(ZkqGsC8@fk#L3e<lyP%f<
zkqr1j4295TqG3WY+m9)1z*oLZ3A%tc_hQKlzbCRZGeg6BsOy;e6V@tn%c@1IxDyk;
zz<PDWF5+9jdya*7DU-Ws(`vWxpo#3>8-cPEA@IW`)@+X}rJRAdm)|pv2>rTGV}m0U
zNjne1tM`gIq1z<RtvdFfKLu!h9@>OM{?~Q7-3R`O0naPdBSI1({wO+MCMK5`<uVeS
zrwil|nx40a3A6af(O;f#G9#$bUok6G?ap`gu#HeJm}_{~ei^1tOTGfB?`{;vY*h=_
zUlVktR&HJy8=NXVmEym2FTNL7JeNmsHrVc}zKXv2tq<dUJ-LMd!(NgR92EJP;mhZO
zpE>aDJaP$>$6|4xziUU6q724EVX4%NY7u`Va(zAudm@ld_m^J5&6|7-boI!dV&TXh
z=o+z|pp^n)Qa>%U<`Z?Thf~W61xQvnlW+Jz*<jQO;`<?gAH6F~%(QX`Cv`e05GIw6
zwIHW<omon??(QQ%Gi2=7sT3kBvnu?f6t0zcm=`!lg&lWyTAn(0?F_XXqG%zI7c-3k
z6!V-=-T)~@S@r&Rw=CXeD^q{6=cx`v;aj<&BC^AODI(ca@zS@xiFA^2S}TLv-N`Xu
zt8B1oE0oB$cEK1y`GyR4Lmwz?Zp@j|O$90;^aBWt1mG}&p!4-=CtvyxX_ljWl)s@u
zl0oVxl0a!4h9sq)8nX3C=X`;_>)Odbvvmb-Fn77F8!{d%w{U4(x{xoQ`wSNw_7EP`
zs}Fbc0Geg+;Eu}UGT9X3?4Z0iZIdE8B+Gm#%WQoUktP@8=#1=HH@?aJ_&)mI<|i{6
z4FeJI`@DF8nvj*6chwP!Q?TxFl=d~cjfOvPIxd>3PUXoBavS+G9@FCtbz9SMNSX)U
z>W_%3%4m5N(G;d6`c94E!kyGtRT@&J-Q$UZvF#8{!`d&wk<fc)n~5}$XzpNHxYY7M
ziLl&vpK-bo;uxkNO1;cGGiGC01!+pL@r7A{MAmaaBa$(UaX!_DyRe}0i>Vw1P@V0c
zbY>;n)l|4s=MDIyjK#wm9n(YtT$tP-EHY%ZcIOlZ&;%%q@#zsZU#4B=E)JMwhv|^{
z<c)|$vXvE^5;OmP2b#pUZskjME$L8?dl9-!(gbBxm%|hE(mieaJ{W$l0%<vnCi!l_
zPZX<&WD@4+S2=YOo)@88>uvmW9-k2lHVt$8wIZGAtaY*eH{hwKo6FYZ^u2OQWV_K{
zE1=!jYqDmaq^ss`AI&vkUAjfN0ZULpuz<{PRAI*V+0BgAFNoYR6ubPJ(>$6i%=2w~
zVWIbmV{j>#?boAqc;B%hoy3Z`yWKqCZ|X)TkM#Sc+;LB3ll<hOZ8^#sqy7$m!3N}_
zTE;UV9^D=pnRH%!2-ZjDbB3<Ah@C^DT-7UpEmVO!#*VM*@V;7FMl?JECx<4^##A!R
zWh2Mm<kMwGxvXj_bF)rt5?z4`oWwjby)_OhH_e)C*QsOm#1QtGzXEi#uI>#G=p&OR
z)JjA;U>DVV?uNK=6@|_a5$t5ipc9fE+`W1|UAj21-RwXOQyB6*u+}$>9JdAB>9^l)
zDp|thM&UgO_gNA{Irr+>uBECajKFkQ?y)*4lMb6>LCNAP2v#lJti+EdIr+sSzAbiU
zz@}eDaM3g+Z|k9dwl!mW{Erg=;rKo28xFT1gu2l3Sc<HtG=ZsJRD++$hP(EHt$hZ?
z#hraV@%(GZeA)%U<qFxuG}C18#PaNCJe5^Wyu?3VD$KLqd6XPF{qVbL`uM=#C8<NW
zBQ>3D?#ySdrny46rQ@BWZLsCyV-X3u_V*--a$|uz13hkzDlVY)@jDUFG7O|!5Amjh
z1m7Tu*yl<<LHO6k%iy#d|E{3BQ2>Dm&`<|Z$N|XG-@N;ppSg*jhb_n_{LwSYMpX3U
z>uNkO9xeRT`57xx3*J=7jkIJ1zErMuA~tCXlDKf#cAJ2-ge_rJz7v<5lp1>bDvh6|
zk*2{=H#yf4So}`PNFD)*sfbdUG5)e~F0LYhV@I;t<+m|lMI2f(DCuhd+w~iRzKT}g
z5%Ld~rz`b!k4TGg`iS30frvN8eMx1vMUS#Y6DWSkx4iIxoj~I&${0x(DQ}-R!!u}a
z&56A&&9JNSTt1nnYsb9M+zJEDu<)MZhB#06*A$<o#-I5NE@nD_C-aMG8P9_tW9xgc
z4mp}{c*1YRifUNm<12I1=Esxl&KI*Cv7m-uf`s3@74ug29ycT!7d90pUPh!_d1qW4
zLM)~wD+@Yq_o8mCGaANZwC}STrW=Bv<r`eo*zGHLx+jYL`FaKun|%wCOTI<wA=xA%
z^v|@9;7ScPiyS}m*Wq(C1v0=m@$UESf5URQ+=`rxT=pqXMCYCNJqvK$lL~n%+w0$Y
zOOoHQslN;xX54kCQzIrxh_4K{gDDVcKRorqv#zEsQRAXK>{4czs3}>+c#z{g5_6>L
zUx=|caF!Qpsr5uoJ3&K~RYlrqYe9N0H8^Bt)FbPH5fdf>dy+(gm|wujD0Gjw`t-Xz
zo_KCd;xXD?shis}B8{nsdh2rLC*r6@EWl0Bk`P5KaWj7TLC*g!Hzu-YB<uR@l~JTH
zIGUD_=OrWwTRs0s5)%7t@=<fM?yG(XlaKptueLB*Qy|c)H;7*|qKVe#Ekdu+5qm8U
zG7S}*tb)h@9=;vjhloKjz@NfK2u4zAJt3a54yXCD3}`j(7S206Z(@)f-0nI$`vR=s
zHg#?C8xQubicDi?8|fB^B3Xy~NqdutH2)W4Zxs}0w5{!;!QI{6-QC^Y-QD$r;2PZB
zA-KB*ch}(V9zuX{$lCj_T1PHU^+jJaUENfTp7qS<9di!Nj~c0P&A(UaJWCWY{BJ{0
z@P*Hd0q-Z5X1n}36gB8@j9=q>xxO^nt^Iah)Gp(Ol@P_OeV2p!^#K%B)cqZplk4W0
z)k<|kQIe0ylPP}RXAFU07Gn-M^TyD_?nt8!?D_)lp2K8>s)#U{!?9N#<}C{5wjxdr
za$Nm_y&S}2xAzz7xSD`j1Pl+F(;11AISY!}>Z)H~`ezOYHMY~a1R6gzbojl*7$p2r
zGFet(9g^6KqBtS^5jbFkXJpjX#$!h;IF6a08<7MFt6Zr`&w;I6ZxuKAjp{d1M4*BU
z1av%=a35v;XG>%70ZcgqlcNhi$=p`hsNC-LuJ>1KQ>0O}Ac2V1a%X-KYM3O8r#<gM
z6@6BGC2mw3Xa0dnJnlS@QM5y3Ywo-R`oP!e6`M5!>IPoW`d2{uukE^8)N|fNY;2PC
zvJ5lTMb`HoLq$B(t5JGriN4B%aiYW*lYgrl7AdP}3%)^e2i+goe@msTbyxIh7ID7_
zcO+wc=F-SHu|E7g@?9td;$W0C2!7}TOI5>Yh;sv9Bm&ePBF=p<0t8-r7eeeYD1Im(
zGuAHg&%7pXU=58sP5hhm{mQT;l3x3>3c+r8*fP<jUJ^lKOzu!W3J=ddz9w22nQJ)-
zQv6sjR(voPB1%$}EXr7(`qgMZYk4ReXka>xrx8(Hg`wJMaZEHZ)uFJjVs{7CGpHOg
zk;>2sGZg*zIMI$q=)YPp2_G)V#EB563}wzg_s+mUfKW*PlZ+zBnCGMmclH`Hh=FYx
zs~d7=a91(#(XG_a9|wqrA2}0(a?5V%fE-c3T$hf_?zq>OtfJKoF&)anwuAO2I2m;1
zMBcXMwYtPt$iQSr>BjcL_v8dzQQP+AY%yEt!i9M;inTAP)5B}=rAnM<0~t@64lOjg
z_NlYkz$Hkr)$m~X%pW4967GQ|oq;I<`CQ0fIiUi~9=}SQdZ(Axg}tx{1%kUmDw~R;
z{_`UKM&96(P9eE;+A@~gIXyaiJY^@dhvtYcjeM>`5YE0_*C-DI6vk<dQTqK0!FlEw
zGr^;&uFzb>&X*aYw^%br(+yq~jgT}MghimAhLkcsik_b{zu|<Ld|~R&Jd}Xnw?|xT
zG;=N8%8H4JsUz>aBK)QLJJ0<P4(Cjs<ngY(b+5nk6w>vqzca8ew!=C~w1nh{>gwr>
z5~|`M^)m$3<kkhAau0S)H^U`|*-CUPSOgz-M1Y-j8YZh^1((f^F)MXKf-Q3AFCQSY
z%SAuO2?t?Gu4rivz6hEq;!ipdRf!kxMBUK$nQ##wgY`?|f_zxCzIELJ32ZflF9PYG
zYpk0*odBnwI}+<>T1=HHL1qu1L(-uvaK!xAfM4GlV}`=8UPY|{gi0+akvB8eaIxvT
zPRRz38ShA1ZSZBt<*4GXEs{tW4&X229UiT4up*PCa-khf(1(jM$UM5SP8p>CWZv1~
zWg)K*8;X9C1>$^PC~ocn^iRbj;!Eqm9%%N^Y2DI!T_Mz4A;fTG_&v{XW)K{aB#ke8
z(4NSb#A^e&JFGLF&MHBmlK5}$W(M2QaUsFc;M^Gpo1YyDgyG(=5aJ;`5P%!>v2sqE
zd=O6x<7HYt5psaso}=!SDi6@m+I1&{cr8gCmps$mKyc3eCwnMjhwbwiaFzFxuz2q#
z>end#CB21K3u*=7YUqF8L4+M0o&=o&br^<PYV5-FKJHw?$Av|OaYP8J(kZ~&YxSPr
ze;QG2cqF?0E9`YbMx2WwK;<ozxG$1!yb~L{aQ&oCb$dFIBuH_7I?-C?Hs?DWtS&{M
z5PlxGBw6h`jcAJ|X<3n<7<bPRK$~L&2O}rS5Hx3#t+k9r1JRXLfz2hHJ6!xl-JL5l
zU0LFF==mJ71h2yRjCYHjKwo=|KF&xWj<_Fxuo&j+wh~f3&)+{90Mpw8Qtm)@5m8@L
zAv+C9DCh_qkrMWHJn6mR3ju~OKB^@i|JyfM2iq{)BT;E(3DbF4FOxb(UhvTRz7vYH
zuU`iV@^%8Y#uC{p%Qutjj4FkWVN&?AHhM1G#$wh@KHk6^lgM>4<D~9kWi#yUX>h_^
zdlzVjMOjfb15E^C0K@TJEb9cKqfwsGVKQ3+?w`iGU=q5_B*o(0V!ruGVx%l^kej>x
z5RbFVu=u_2r+wg>`s@Y<X(pgMu1Ak#ony<A9YytEdFXVr)n?cNtX`KxB6&ktd0cty
zbF^9$VV)h>=Be*q{j?z(+ln!dA3uhYh9}~uLo}njk-fw)fjn4IZ<ug<Bw6&;rY%QQ
zMo={U<gTG;k7==8&@JcehC5ILYNFt2=oXwiz5+aZ_HI}$fjSxrlMNxVE9ex$QfdK4
zH~a6j=Eb>v)`#iPTCy*)7JK;of&Cxeq-IjR+>Ma}4lvl6Gw{`T|40=m;CfB_yO~(E
zP{px1nHU1p$21yJsPoiePQ(ys%-gu?{36==jWl6|IWeE7&gyf7nxQzGg@5N3pT{9>
z#TvQZ!Ve58kFHs^wNy&nENHf4nfgs(@uiA<NPzYsQ-F(l4IOH^4~>l<MriSs;{~wg
zzGO8TIZVu_Y%ct^7#Wp6)~ev(1v~5VJ~Lj(ywC*VwA{@qlSxPiEA6yZQKhP`+#5tl
zk$=8cAUgjTJV*M)aH30)h}5HHu2?T}dPQPYvB~B-ZsevhO^Vu?z`-K_ku=A&Pv|?0
z`1%J<vPz;BHxfWWiE_Ekkm55B&TklS-c@Mqy<`R-4FIwbyN$9p$Pq<C^Jomq1@7lS
zQOAMr>t%y7;N&u3#R>=~F}66p$J(=DExu;KFAJ<fYkyEfSXkWD;VlugheG%Kc~dDp
ztw{)UP=$ZGUl4tCI~1kezqD2sC?UzhZFwmZ1aw?AU-$7BMD!66ruON-K1_RGuJ9od
zT^N)uzTs`P%^&2$+HMI67O=kn?phyuXGuWHW!;w-q?dB?r{nFw2essSET1vb!Lv;$
z_svQ!A}66$BA8?+zE~fxiMR!u!co7+hVTBGyx#~{+ETt9685b8b`M!ViYB#7a20UR
zi3;%X<ebahdyK00^gllBS+%lX|837)&|!8b-*!dQIvaf?tHRo9WEE+<UHm#L+vWp&
zq>FOPwJR)TxUtGT;V$z%haPNYR+Urr1oDP#JR6Y0A$@yNR2wfbU)Ba@zzrG0N#b`>
z&jR=xU3AN(Jf(`+1BjC);6~GP1x^E{1K4_TXl}qWOZj;cS(`rI$y^kME_2%2-n}J4
zqRnNvd?WDDDrd{K#m-ely~j~q)0ly(6=PmX9r9@<Ki@v3d)@m&r%W?-;ee#Fg)aek
zr37*uV|^mmwsq=l;pD>}PV?x?%mgKwX<-~zkI~jq9wog=@|s-y36-Z|pH5KXY?-n1
z#XV->ahdaJEEE4cBbIvBsVb7YVeval%E!0zkq_0bzLA$E@%GoyTmCWk2RPv3DL4o;
zG7(gJ`r5Lgzr~Hg^OeZz@;E(ci^{5vQ`Ld5yb_Uc>RYZL8d1XacY9SVJf!`N;5wb5
zJU5~DCdD8k?_Oqo)tT&4QJQrG)1<3`hG13<+|)mz17esIRlX|K>0)moLA)yMv80r;
zWsv0IH%5LN4QUbYIXS%{CAI+j{C)|OJpQ<=qmHR$6NMgu*<&b&hpid=rp;2g_G|E&
zWbyeekuzH3I@d49M@~U?gu=VK=1~S~6{?I2J#|Cx{Nho%a6jpd&|Q??I^Uyw`VCJ*
zu~-P|JHzD2P(kK}?$GmzHRFWh8oDOj^Ps$Aqe~tpgEs;UR-MrCJ{cfzsFCF*E+4T|
zloN#=O0R<`bc@YwR+QV-NhhVrx>}L4dCD!!Tjca!b@-RpX2aO>qmt`;`74>uqpV=}
zj9{n_QaaX?rwUIUQ3#gf4_c!?o>xV>mb$|9PuN^%J>OHRY}TBs*c2Vg-U0&Hqsuf{
zJILf8*UII&skZ2QO3;ClD7UXLmIB$K=hVDWEY<3|jdLB=5(Zm(L{Nwmn@7S(swne%
z9_lHmDmJMzoA$p-(NiSk45x--QPsxM?M$8&bg0P<x87LL#<;BBOEj8{O<z>fq{~|G
zB}&gm#=f4VZ!0nB__QhXX}8m-`K&&C6s&K*-NO(jZ9@_LaBTxT`LV1;_nDq2et1P^
z{%F;!6K$hS`@O}OdVXDEBXimmaV!__^k_&zitbAhEhrP=B0H8f#pF9Nu2Js!el3!y
z88wL1G^YD22DQnySGk5Gp2o5(!TY^*`|78sBl*$X6NM_u6)BF1;Ep0e+!WSNGXdr|
zSf|veW`sD^VhRLck&~Z;o`U0eiR30eG1|c@lE471CcYj>$}dC2HHBk}VaKrGnv2#4
zE7hjqNJk2nYlR^CNv7JUTbyX4xIaTmnzE>ZBw$j4S9N?kTNQFP&H?q_j~21Hb+>kT
zHe;9c7N|yJw3H6o0)eUFb<KtFaWG8bHW;t@OMYd`cCP_=xxD_~iEfo>6Kl90)?rt8
z!YuG!%f3F{$hb}~Mi|Q<vo)cWEn%bnhc`lSUECluJM$UmTRE9jJ~ANI1u$eYj&?t3
zt%g@ic+0wPZpvTkB{%gWjA&kU58SeE*G^p)d4<eBbK5gtzj_2P^;E7|Q$`7`trQC2
z{mK2MpG^cTY4^XItI(!TFy6=$739JNK|rKsbNt~yn7M|?j0*WhGSvlTuRySO<G%U!
zs?AY$POSh$V=?)3WJz&7N>V<=%%r=*+2mbx>*so$(5*EO$-yrxw@53;Fc{J4xu5IM
zr`aL+l)($kk7K3if3p!ZfoMl#aQRLz8I`G8oyP@ij-of^f=zp|^<=eh+<zl)Evd>@
zKs?yMD)l_7c<tn4nQ1|~OC0*`HZ|zzl3<I&cCLQGp49xbmZ-33clIvDcQA95%!!X*
zK-nvapdDB?{f!e%izVY+;4K%6%yf8?kW%G1BX_9mPX_$TJ@4jz|DSFMD>=Ofr4$hf
z8(nq4BP^9U*)MwxtWuTsC`L-*n<ssOpz&oJAEzh&zKqW0Ic0u_e*8(^Da^Mt+EXj4
zO4lL8PV8vien*lu0<@CzqG~svHmRy`1#lXS8Lqn?Z}Jk`1PmGri(vI0<dmz#DsU6Y
zABEv>-+maTSSJlUJ7Xq+I<p8SV#<$EqwtXe8c?*(?KbI(!EKm?JQ!<k!Zw9NDpqRX
zVWu$DUa86;zO@jd12gE1<3nFEz8X`RQla)PmIaQlfw`lS4zxo{qu<9Exo?NYfUrXg
zhkVZkhsVWO<$;KK#}y7)g04nlr^cb)QPoB`^$(h%;z~Y<o52Dbf~xnp=+NtT=b5qt
zRC~ef4O}yc>=Qv-9l3}Q>ODa8s4zA`TM@f|j$BacM1t3QfUPf~P0S$bq5B8!7PZ6w
z`;#^(j&RfayCrPni=@#9;>FMR$#8&tU;li*Gig(<_h9}lX)L2dyoABXH7J_Qu?Drv
z#$4a^9RVai803mn7>a<KlJZuk6i}f{ogCdAE)1U4hFXAm-~S7FgK)2-I-hykb;3kD
z-O4ZEQV2eNYBvg+1>+DiJ!CELXZJog14m9?La0QQz3*JMW+g@P1;p?|_*T~i1=45J
zhcgt3uB~eNUd2lMN_>meANxmTQmO`s8U<ri(hqe=H)_-_+GsD=fw3H3T!1gg7PKTI
zb_qf_8lj=k#T|Pwry8206wKF$D?43W;E_bwB|ef|%|?^9#`kt%meCiA`!`{PW_v(i
z?M*4ywi+ZkPQ{}W371{7p*rkQ%>AES&iX<|*KL2~Ic&|ibvE1BFsieP_Tb-``|<HO
zT$jzaDvPP)N%@!E0t1`jC;-k*?y9YY&F%nKO)|^t8lw@r?SeEZ2u75?-u|x}hEAo!
zR>sm6U3$*w21rbk%D&i0p;KmNRuz%VAUl$a-@v-q;$k(f{5%#70PZ;&vBd(aKn-dQ
zwlrF@+venJ03(bhtT||50h(2hj1Hm@)OZoC($SsLnt+2HeSzjV01SOY1$Blse#Da8
zQQI`4X6dI1Qf?mwNizrAMNgtf?FA<d90%bJgBXn@DS!z^1-mOmzat5v41&{U_+=MF
zsZ5ptF$W<_6+{VQ{q?8(mpO>f4~ZI-3RIZT0#T4=d;n$NLl(ppMgw6Np=|}X2J`NS
zTUBU8h7v}DXso140dfh>r~GXvKFg{A1|+Ng|CE(5+W(YQ0nGoDRRPSM30Dwh4*6J$
zvMLBWs4!Z8zhgBgNEQfKa|{Tm^&qH^i}M@==g@m-yvs+@^A`@1SYF9&^^nsL7)t|l
zx-Z$IZno<B+|)OUKB_1(rj(ZVG^(V`JV4Q!$D7X_6fw`QS|3PBMB}7O$Yco9Ahl>v
z<0R=HwEg5L>Eh!oGDUk*zcKnd5)yFxJ3w;aP7V2=IJtNA`;VtEUBS7rqFB))20Bj9
zS}jei5ItDDJl%5wCm39j)!fJ-;87KqV&scTjK|MWxKPJekGZi=YE%oc_vJ&$kpX8?
z@tNn2ROtSw$e-haapczkWKo-QNfEOxFSW2Vy*oBBxgy$AuH`a_nsCX|>D(pb(tR}<
z*jfODXPu8TygI)iztL1~elCHIzP24ZBD9!)sK`yZm@_oLI5T53KYuL<GOMz0+Si?4
zoji-61W+BA^i;^uDN$t*(J!9t-GDF)+<1`4?*A@GAn4#_0fqfjo?Y2SdECY>D9R#9
zwCFQ*4g^N1u+RHhry?!4l{cWGdjNi@D0Lw*jBg5DwHED!PvqavwX^=45u}eWhe^O#
zh-I<6VWO3j0KtUe-ZVx()@`aZyT3@R`&~*L$?FM8zicE7oy+MdJYdO4bj(3x<Q1L@
zax6Fb#$iHH^k!11N!=obm>;QV&8<ID*>hYb+fCUmg-*74Wcyx3Lq%rrL5a8c)kO2!
zk#9l_y_m78W#mQW7590pf}O(20VESGsmqTp`i<W_h&9PZBl0IGt)jn*%U7LVM0LfK
z=)C;&S-DXrvFos3H~>U~J>fip(f&lLFg4}6HoDa*^_Y43N?a`}05^YhuRf>ZlAam*
zM0~fW8Ix-p6fX<mj{b->Bo((R0Ga~rm7gP7%vI(`c1q0*!ovD*0i09$tW;sC-2bUm
zUf6=-|5Pefa6GL#?euYDgkx?xUCr>YHRwJP94#OW)7T{y9uLd5?;aOj)9BhzFI;3t
zXp}=8&9BP1Hd%gr0I8OUj3p;~wdg+J&NQ)#OKI!*(r>h(Azrz&FeP!A8|UiQpn36o
zJcjm_gxM_u%e(ggne4R*n+;zwC8aiEg#3oaVaGvgHL!bjMsN?5!7q6hZjej~d~8_p
z4jGVXbtrpwwae?F$Bt%S^d3nQQ^Q{TVBYE}FpNT6+3fQ86%zun;$vr%G|LMG2iU5~
zt9+sC{MiWIm@K}YDK_Cq54U}AQR-i;B8R;>d~jQ}HjSZNTMX74=v_d$Jz0yFJU#)p
zPw-a3U@GfB^xJ$uiyK}0LBXwv{)VxweiCT7-?J(dW8)cTH*S-L08?K7{oWY}lBM;T
zeA2c>T>a%k4$7P0J^RZj;bZ==0WR4Jy<cR{z+?(19gXKZu{#-9vxb|rPyh+>4p#d<
zt6kFp=Y^l>sFF1US1N1q(p}Y#8Kq@eM7{5U1!^QHmq{W5hLaUmlTa%9J%b6h;~tQs
z3SS`}tV4dX?Oj(DFN_ss!U<XDu=j7bp+}P!@Q8ou{+VnmRb&e@*^;o>Cm!ZX7Vy?m
zk533r1`dt6no`H|S4Z2gao@SFJ%&(+rqU?IE;Tx`?&{_>-KwWBt7#H!A#BfyjH8e2
zco#pL(5r|Kc>fQf=eEose{)F&w!362v%(R|d5b-f9v4H3-N~D)OK**<TJ11FB+@dD
z@~*v%0|xE6$N9MN!{Jk_o#@Isq_0Ipw)P!uup*v+x=Hl^^=|yQTB8BoVX<6ckD<1m
zP1~w1&|-BCEZI=R!V|FH+m`YO!o$=A>YXGGtvety(4Zw7-xzj?VK#1n{EC@`yyEdj
zQ040~(+r-Hl(rPp9v!Ut<<M-GSwFeK5xdEp@HU@`2`C*uyzRc(pkas@9w)ca<mU&H
zl*76_MX4hd_cXdkhG8U<<5M#pSG!}Eon$?ngi>=y<n{v`5^^Q{8-b?F6_Wq~BibV@
zvqpjH=N*&klk}h`<)bw~lV8ihpjEEacN~LdAnhk%7E@!bZ~V78gGHhsm0KK0&@h*k
zAZ?gM$_bo85ivL?VX<MqS!OiJdGDqQgFZiH&)6F9N0^zV$Xz9?Zq#1T*^lGkeJpO1
z2JO05*cg@(zTvk9yaXzn9~S7v76p(id@{}UHLZntI6W~xWh#yVxaEz){X&C^j=!om
zo{oH6cYCw{whUp**e2V@CGnESaMukDx^J+fh*3qIThtu7q>c3AAP2Q3|Gi_FEF$4#
z_u76KW2c(e^Lept(oXV&EK-0kR~8O)rvvFIpq2co9XbWGiGP*}uQN7i_IYnKwO%Ma
zYaB37*TtM}7VUO3fGhhfKe|;gLwaf5A-zh&xafA8tYBZuXY2Xpgq_l*SCAw;^DDg$
zXQ_XDeTkh|_z-(8_orezB>SMJ9<N}hOI^vnx}GEd@hP5?V?~SC)Xi-R*=q>#IvpNp
zHgz}L6r#aj{NX%W$8pP+ilFZAXB;Ps)B(q*I|>u5K$tvXuqY#!Ms;nxW<=PFa1Vwv
z5U*ClKIL?>TFaW?A;{md4JKaF7lVP$g2!SR4*zxp0$VHem%BG+%NbS0kNk&iXP&2*
z^fWHAk2@skkAMAZp=L8r<35Gc$>LUS4E9;awWscz-^BcB7Y$(t|FAm^J%1D-V2;XC
zsw>>74bFR1y;K1~ZVdaCDCTex@@3~P_@<5OHY$&z0QEGg*4#pi+b;h07>cIC?Unh5
z`vH;7bdcT*K5YSuCIa62vGmQ6G8U5hcPXg-CRjK-asMpD##2lL4c$nRx~+!-2|lrV
zdLOneoGdk{bv1<_tpVeGm3W``A6ih1Cdzzgd1RB>`8`T4QGpo~XZo5E?F{k-*)?C@
zD#K#(K!-gY$#gw9S60J~&)8(~JaMOV@CQ9us-OZ}hjgYP+ny9O7I7%O=InYE=)&ND
zLJc$yc>SyOC8Y@u+!;IWnbf*KVb^xFQ|Gq1&IiTV=He-yu?11wKWiEXeBntFQ0fh=
zd3%_}XU;C|y*~Ps?=$?*h7~8Oosf;FZaO5cgQlEy1)<|3w&N9?6Brge&I~CJsVk;N
z8!<75e)!Hy63Os}?b_1e<w+?2>rrD?@+r=qqMD6GC8N9}9{<|&E@KpVC_V1A1wTN&
zy2(Kza$POZ+^!4)WaVIfl0M%1ARqon$kF7mVS0u}sL=^oeg(?xc%=^$e?$sCBOBPl
z(68jDDvb4pmDDE-_Hq!f-ko*Q0;B>TVS#F+$wG74?Bf8Ry@a@xCGT`wuJ7;6)6T0B
zHk={RApY`Lj(~>Ejin+ln7vA@BLCx<+=+sZq9CKlB&0%MlG8_Ll9QyNYI5-ZVJy+n
zc9+s|h}bfX%2W2XinW)$CAiXfaY#ucIwji@>8pPkCh`n?>DVmZRQh#E`C`ZTLq~kj
z=pHWo>;v?V#80uhH4`IXz}{(!9r=TlRPJn-<y*#hk&szP3t4tRZ|QKCLw*@JaI;$W
z*dE$d7xI=H6<ph-`u$VH3YKfsE`GE_ytm>ty@fKmsg3)qF#K2@-S4>NV-{MOLTVjz
zBtydhXGBq{ZB(X9*_f>Pm4rNz-KA<{YJ{blPl58IQsA;o)_<`EczbA#(w#WjEc+@b
zK+WL3BjWLR)VkRDaUVqZH_%%h_#16%-lbn-#C~L6a<M~hrprvT6FchI^Kjd+t-(9p
zHy^;K`uZQhVI9mRPZ9iES<tl^e;>5)D0Eu;w{1ZpR|()414#eeD@i{qz5`S3W3%$U
z_M{<>L~{=1+y33}`hdD_!yLaTvMFojPwHGu-)<?QuFp39hf|hf2H=RSL#mELaI4@k
zq)0kD5uU)x)F>|(!QF~GfOLUJ(Cu;ig00owe`GUl)_pV2^tD>M7Tv-v3ddCDX8zr(
zbhG<IjwFYIR8F7t#5{%Yf6%H9(V;eCXu8<<BoErb>-c3c*-$7~<m#0X<s@R%x4e>G
z!2%^@t%gs?zZu985k&1XcW<@2`Enw6TJ{(`bqVt2ota4D;=gpJsu8ILrOD=-|MjUU
zoAvOSu1Sx(HPzc2h=85dbILWNPGMr|FH1HBYYLb7>aDgZ7TWrGDt7u7l?}hvyz>=g
zyf;h+%7seEM|8oJoX@=95ALxdaCYY^)(|*N|5-)SvmwEi{|ODf8)Yt%lo7m?WD^hB
zw4{MokGAl7-lJ+}&+d2`*W`Df%F@qLIs{ZY<<nO>o^Hk=b@fnHT3&6<4ndW?Rvl+(
zc9!$n1{WV&Btq4=zd+lPQc%c%3c?sH`i6Ze2%9VbAOBmh*B(khzE)nyTkcj!4v;JM
zS`-SzwK?!I?!wreFB!UHSc_G^_cvDxd;CVCpP5;3$}w~xZs$$g5Dz-OLol*6b_<A+
zO|k=eFu8vISs=_URT0-|maM?!9xU`QtZiO_1ol3`QWvIi>@}+Z)uD|7j~I;`@k_E1
zF`emp2K|^46_`W9B0bjjY;~b2AoyfC%z2&E`D!0ztSb}cE%kbnWgs}{DtT(F=H;UD
zolD32l%#2ankz&h_rnFo8T*&X`4HM2PHj{Pu5XX<jb52>cRB)0fIvk^=dTb`mCpRz
zpp}-Lx|}WO?)A~hT5K|M0%31<o}Y|Fp*ap)@ml96Hrt{s36{9CYp&L{fYl$OvQ7Ra
zEWRP8YESi`9&D7r$?&dmT1)al{h8~=e+JZD>&liTPkGq$50$VxrILJLEz+zG!x3Jb
zF7oEP7YNU#dLAyR<{qi0j*kd_Fp;&0e3VgMGu13?1PpRCLr(45Amix7C?8JocB&wP
ziIjZtv{oeN0lLG4?ac{Lz<2$t6t6+X_-(6z^&#e_X=nliVrT-)b+^2lR4KbTs>BLK
z>G!bFMSbL64_>vzp=Pz@GVO5;2MSMvdyxeK=^LwN-ElU9$*%z$t0<NXgC8Ri*(0n2
zOP$9VXA=UIt$i2Ln6;2_X5o>hpBF*&VWx@U8k>U*?CXwlPzJ#Qc1q^c(4|58&s>F*
zP3@cIoAn(BjPL+n<qN@Fphu{dSH#mjm%}P!mBk9`CH&ogOWi@vMHx?0Y4jd`<_5eO
zuQXLzlT{v<@bN`q#A>AsB~lqQq7z23ffK#c#}sXGhyq8Kw&s{*m+N_aV}O4hVEA$E
zF$GlHV5%&^SZBvJ8A7@R!nm4q{d&Ctc^$U;uz{t^*4@ORb*;9)-dJfhP`Qs$VK(%5
z*b4res1{Z3MdlvMV36GElIa5u1OIhsgxEYru^S6vi$<~Z)O!NG1`w`CUc`roABcLi
z3t`6J2aAO3#~0Q|ZfXaym2P~YcRcy(I}cD*v6UP!HrgXMZFb?66{Os!saA7ckXDs9
zW9<AxPa?)p38A7QWd(1fW0uxqmI>4i#JukO5dMTdUtR(UB^LU+xi<8U-F&;olHw!R
zh7_kL>84G&I4T~|IeI^z&!Yy+Cf`#3UYB)TxvEso1fL1Jg(dB;5o$W6%RZ^<07DdQ
z5yvsr-}4TfN#^J6i!~ir^`9{9uQpZc462VdHB4q6yp6g7p9{o$JD9jpw@>LwvLOE?
z3n%@;AslgZZRvlD=Ug5{{F1gGs^*XR9LY3e<kM&N_Zt~~SNET%bl*0&&5lI(qdKO<
zXzIK*=m$)94|Y}~=WlW2hg#g$x<E>oJ(p~prWoOTbZSwCT0}RWByDRhiKjb5O!(TI
zzIx23W-fv&{l}ELuyEvh!EI@zL8727vvKckA}@0P!Ad%nC;9Mv>%E4TDT)*ZQOwMd
z(^P|uNfy>uL-AqvS@)m!0H%r5WLPI(LuQ^>zP+jOR^kiW+yqNhC-4N*Fo-U@&`5LS
zhA(4)$J#x9HY8$bE=%-cB|-v}>QpwL4}4|oI)wxt1oEVjSN{W+ypyK70$+$t-G-1E
zt~w|slo*-JXJxmj+^6{xNghB7)fvIk#2>?qzUn3k?`s{}3~m&7)E|T2%?}zNf+)jQ
z>jZK620@Fl&ea0bER$d+y~p*it!44I&v}k^n4Y)K4Yu;jD!zpGn~%(Jcvg)IWKnRV
zGFTCFh0iBsSKdWZPUftAy!qGH`pizV9$urbJM^SNyu<DU^!ze@^?6NhV{%N*R%^H{
z%T;MrvJBchkov9kzD3BfDem8t(lww5b=*R^grqsr;En-!rc|ILlU7J0T?~?agj__T
zm1n_^XK5IG`CP>qBW+bpM^b4~1aqsXP<9cx-w;?@d<{r~N7LJbA8$IGcDJ%tBrsRv
zK-m06@lobjFpz@LDM?uaIyXyhvrANrUieSkU767if?^Ag1+eCtq%nN6n%5VMw1D!Y
zO_;h00>We{<pR@Tk(!{@|FBOVl3MsqLm1>q)JO+$w`dCpS?o5pT*)fuz+R>$ZIRcS
zj!J9$<5)cZOXK&IyHxoSojt;2z2@(N`!}ZoxnDlfG6gh#<fhm0P%;F6e4Wy$PxIp?
z;>3sF%Y{ive#`k7XGc^Zy{d?qjwKy~l!_j(eE65kSEt3=7VE1_$R(*zpM2U0#|o@3
z!_VHLRdS}Vpmi$h7Edm2dr#<kMWP6dDf=poCgAcxUtgLPnFR{|<@5lj&;#uJ2JWBR
z&Hk5=_m0vym}kL-^^BS@DaD&6`ZdztBYV}X>XEw(Ef`H5%~dq}`Y53!<7ppLU^gEE
zn$r7%0}{<&8m)UI7KwSFtD8JyH#y6gcIP5%_n%18>HGF?SEEp0&o^sQtg+Ax9#SU#
zZA(ek989}TJ>|gypvUR?Ppj0a!kH_^OOY2wyLstiH%3WBz8l$==RzjDDUTCjhf@77
zBTv0--A6_e{wGoP1^)&$0O!#N{T5Dzl;lxUHgH%o+Pr8|=l8AciF>hA8Jm&XukQ*x
zS*Bn)e^?L9C?76TlDg&ADKCk?ZA(?%MLW`au!@Uwd&K?n)7HM~A$9wT;MykjDt@`H
zaKEZw-!#^I%7k53GRdmyP#zOU*y?;!%-fH!0X@)rr2m>0$_RN~2xwi_L;4I<GmNq$
z8^MZ`j|`X@!jQqe#UgnA9l3mP#0~5|wBfxCbdW^??_{7D+leN^=1kEY9+8kXo4zcf
zvcDB#oXm8U84Ua1_<TfRHf$&OrNJ_={>g*Ss}kEeDl!@xwnenE$G=q`Rn&*ejaGqw
zMD)CgoSD~y7-*<O-TD8Kr59nZGiqI6J=mzavMf9~C9yDMq&SV3wQc+S*1n<5J>EI(
zC8!L*TGl<_XQEmp{7|3Ua**;7w)sQ!pPapDro0dLPzc50gQ~=@oR^G^@t!R?4{fS|
z`@e$rR84yJZ^)q;^0ZvC<za+3^P{!*I%oX8Q`T}X(%18LaY|g&?*%!{q4kS_FX<+7
zpxD4$&P=A`rkZ^|#$;-dJtK=4{h2`LN40zQGe8sCByJuhcbF5q@Z;g(-skMMN8R*O
zLc5Rp%}=z#s~LX4cj2O(ts-EDV*l+{`-VNWw9wcTxvPf4;NX<zkj#~;xXDw5i~HG>
zFjms$kFcfZenMb|Qon~#NiWXl$-L*_DOEs%?K5ZphWNv5@-rTpeQQ4~dUhY?hIv8U
zo2DA6Pj=5+*rK4zZCu#4H^r)nW1NY4VsMLvwXerQkaI}+g_vB}$uab$Pa-DDT^@;m
zFXUgUeu@d3tFZV$)GW&MDCfAGV0cxOt<*-XVC1mKO_nT1kc@KlqJV+Y=jqE44qQzh
zR5tjr;>vp^c@vs<3P|k?fRImsR9kxe9E~TB+S(TT*_3SYf1}fZ7I34cvD}9tK?bEh
z>5>w1ntrN6JKefDkY;BbcDP|8cKr1E5QtZH^%Hl-l+dB^Hu~oByH*9MX?_!h+>IDP
z7>CFfqKsCV3d9c}cz6x*Pu${n1PHBBHM}-*(P`^iClw!;n(Ecs@J;Ov=Vo>jrkEAz
zG4#@$S*9D>)|tsvP2vp@DT;eo<<igW$DkX9x;jTa=odRUd~!aOkmv$%k0(Un&hdS>
z^muVVwb0DpAB-GZ?uUXmHV5bgF?aCljkT6R)8q-}Y<nxPfvm5;bgqC&T5}}x_DwFx
zA(dFHp2fQS>U0AsTH`cu5L`F41jOUPDxC%XTPOA=?bq}ypW7B<Y@4#XScV8m<FXGE
zUH{Uu8Iq*V2BbL8bQ!UrWO+GN`fn71PCU4iMGd@}bGpyVrm|(xSC6yM^2}Z^<lqAO
zj#r?WJL<;(h1u`f_1%SK`?|FII7x=cPhSaR(&dex*Wn|>u<%5BsQn*2mZ97Z)o8r`
z$_*yu9hHIF$1kOxhk=MjjIQUb+Yn|DpK26Ege?^70L{0L=FltJj8kr{rs+qd4V~Qo
zyGz947^`l4^uMt5W`8wV;uxmyH(*DA$<pl=(h~VI+9%Xhfd+;L-mKC_Fgo&|-emMI
zpfmgDHBl|e(Q+e<;VoC|@Ak?X<e6J)c`Q3!m8uAMXXLE^!D;_YWDwYM7T(!GUD()q
zdlG|x^EH*C0~ieEDpn2QGJWk+8plWA>poa-ao>0Cyy30hc2GT=YhduUEir2c)C?v5
z_WWAi<8Rs3hD6JhFMlox;Q?ijSE?!p0-mCW<eDXK#|ZLeal=`Ev*Q{uWod$8Pv$nL
zC(W_v_#!R*C#}kn;?PEHC6K7SWAY+^XH^Psk=K91s^T?Ctif77)2eJ-NrMr;qLGy1
zY2~4utiqV6L^}flfZWA>xH^CiJC=!q?)CFUjL&w3B{<AeVjQYy?DtI&41em7Z(H@q
z+5(4-p8R_K=utY>t-xx3Phxd9({@bw^;xN^yQWSwrPB8rEpL#MKJYs}?5_;lYNley
z)kYUsMc(!V1v<!B!ulY`!mS)>6pXiFASN*k?)ONq^W)K$0H@%ikV-Czz(c-8lla#5
z9JP!S&9HA1W06V}qubqcT4>yA8V`cdEF~S0`IRK_5jgsWL~j{OUv{-S{`$i2cCV#W
zu9i-tzlrfDZ693C^|1lJgX42nz<&fP!~`k~@s`ZaoJ%tIw%-V8-H$(}3_G=79D{z}
zFt0qw_5T5qBlsHxDt11*B`CbAijCzQk*2ZkwBC6~gRyFr|3j^Ht^HD8-cSjxyWVDO
zBvpGhXd!T$rvr2@EhA;!#mN)CvvzaSJMtU)+g-su1)Odr;+lB*JY6t7Ii9q-k)2H#
z8IfYBSQnriZ;hT?oA>S#X4j0@IlmP%qw3%<0G~cIqN<KaxY@Ln`u{CNh2Q@PQOE=T
z79D!X2*}3jemyx9;liMelP~!l1=>80^rFY1sk^?EA)-yUYyQWm6PP41@@9_{j8jM0
znE_1Px}8?M*PPP-;8o)Kchuw~{vUHR@C(Jid3&>QGD|gpQY1!Mg>#w$O}^OR*+zoM
zr~(z@{nx3NF(J6iN$bzd7FqpeT|#N>2H;O50SZ;@X~DwZ$d)}j#qlHO9{=|YN=N|R
zV^E0A6Lv+=v<gIX^GO!^0kby@pm``%xVRk!pDFdb#1#ekUeJ$pcaVvs2Bd!(@dC$|
z1@r|#yqyX)e$I3DIIaNw{{YyW4fn2uczxn8Nq_MUi?{P<VGj}odJ@Oq&iy;P+=8_Q
zR<{hwfu_hgw&BMUdVZbxG@6z3Iu_h04?9VOGG~tp^?T}^)32BU=rWIOvoj*M-Nt~*
zr499wtl3Lw_80^pIKP`gn?ePWkZQ13%`=!yR8lcHy#+6l2$aAu2{~l`hiPa3n`xD?
zGzpD+aQ_d}euqA6@gubNw35!Fa4O>yb2}&Y`)nC%WqRTA07odRHsoZrp$QJ8#QE}p
zT?)Ny+Mog-ERhxt8v2uJ5hofbz2(DDKZbQON!m%;DODX*&yUV5%hdOjuM;S^sIOpv
zL8ig{eqz5&%);HSu*Jy&y+gl~u@$D<i_GL1qX(b_Z-uGitmS4je=J?5(O>P1zxIi_
zr-wSYDXLHQ$q|gb#ZVTKlXt=x_kP@wGZSe#IDOX?WWt<H_j3H9HT?l_SL0T1eEI%G
zkwH;0?&j~ayC19S>e{)J#{CKGB=JEdTdzc%S0Ju_y9*cH@4q5OTuUfB!?)vKUZfNr
zMzvc4`MJq)Nk783$%YoNO8pSEDsQ+f#g+q|IejQ96u$aWl<0_eOLPVjiM9=fu9R*l
zm7Jnad*x~SEc|kU%#}O?%(P08(_cS3t6ZOvqg+ounVaL?sPDeG<{zCNubxqWO{A1J
zLhTqVG6iV_5vHL<fL$|=VvtIHeIKtL;vLlzVcj<)n6k3h8DczCU*Th2aIiwyRPLp2
zeITj96hYnf)BhU3JGA+QP%_q94ejDady##}fSLR|;iWk)^$AeEB$8m9^d+e*xhdit
zqHF#Ta6b^L_}6L`pt@K1O;w8qYr|rM#m58mf0|D}M4D>(J5iOUEIKz_CbqM3xw}(f
zrKh{>7>f)?m}OE^c^u2Fy0OUxw7xP7^-&4@thbBe9&Dc7yjZK`efVhlH)W4ZXA>wL
zRlE`nV}{fB&u{z=5gF=BFqUp<1WF6U^(xaZX|XT;h{T2AmYhe^ju~fx#g@09oQug~
z7NHoosKifD+jolF<Wftu#nM`x@`!_UzqtZ&?+vQ`E5tQ3P8Y;8_Ha1xll3v5v4HQG
zZFj3>CSa5g^hWB-K#>#G9%y;;K_KZ~3oi+Ww5Z~i+R0Dk0TGjv6s8rX^Wh`F8f`4t
zS>gp7>JJGC;^8()M<MBPh0Bt`oPKSd(RxcPjj0RbxL?s3Zv1>3rZ(@M<eM+@HWnnF
z<S7VbQK2zP{x9KymI`X*+qRjmC*7F^%a)v1MA?utUv6r9y~z$Q_@Fjl{f6hxv?g?>
zuaINX@$PH2X=q7lFzE<&wmdul3WD;akGU}fmbWU}q6Q?o4oK$vX4{{5|G34vxwsw(
z4Y``DMa8O8VL~b?vR5s+RN@*$VI}OjfT~iEV~~fE0sd-T*jq$zrI5|rPF-01iDnYB
zAGS1+tdtS=EE875M-c!1DU!-SZr9P8w2g#gWrkrWpJ#OjNP31W6eM6-HRx-y2LF!`
z$knPM<^CN}qg1)nLVenr&s+HAf`7*7FlvOmR=SgrnWM9!s|GJcYDf=9%>{{8k%5!7
zCq(1s^HG=3_A!<er6;LLx3nrLOGDnTg<yS%R<WA@1U-cT?^Ghe=*KtCW6y}dPzBuv
zy|Hn|GsIF27R%%`7bXETY-y0f)P)_XOJr#~3KML-*iRC=m8Da8(*ngvmdgm53KL|Z
zLjro1PVJjiivx<ms=Iy^nr^BV_n3j;cOINatd>ezz2cNYjqPgj;#*D}zxY(PvvZJ4
zjbealFNa4Q)<#&B(55izyJM)X_(T`a3^s9>KJ2eQm9AaHk&QsorMjNErbC&6a)5D`
zrVE;a3Sd4=FKoWX8w8zA0KR@V*dR`+ZI8cr(i2TVt&!lQGs~7!f>UUudopMi;nJba
zKy{c_;lh&WU(m9Cw7jv|GaOAEX*2|mwb&yR3oGH~hSw<5rpQr@aD&w`TL<UV9oMEO
zn1OOb=<@a<rcarH;semsNo1NLDuswWGMyaPOom4K%fIvFv3kZ3!oUnnE4BI4!D3Mq
zSFXzTZU4HS7EK&AHPTNNUH(aWYvQfXLlyrnyeX{=gIW$-t<KlgZCIH^{l4ErpwCl#
zi&w+~AyHp6&Bk`+F%T^*?qW{CFfX6wjUsF43rWT+fmUrz$@*z@EGe4ii(Ge1kjM1!
z;X(g;mWNTP2?(*KoT(Wuhx7g!UHfw&R>9$`!MIw7bRN<N^ZVFdHr+0e@<U5<4PB$^
z&!UqC<n&+f%ju;%;*fDI^Rj4@{n#+0>I60re<t7R49)BcRQuflEYa@h;MDv}HPB5K
z3Znin_}6dmH=@8u@aMcl8nhg-;q0TZUctp`d&H-wJMGLYb}!)Tb}^dkS^i;Pp*(Rq
zV>F~nMnBN>w?ngl6p@MBbh3dhys9j=Y2A2>zF-M;c}7}$>;YZi`vJ?jiTSQz4b3Px
zyC(fT_=L1d3%+mQAa2&T<?;KGUd6mP8G7crR!LRWDLSAnG}=VhVB%=<B81X;o4h2P
z?aGDS7vE%(Cy=6w7Y0EcRuVm(#5#=OqT8?vYMfAIO-}oVr^!GyejNk}b3~-EG~(lR
z#({8wq~>y)x`A1p#IS)ki+HJeUj+;!nj(sd^q3ThO7Yjgxor4^&yBS|+LT`x2<6(!
zG1C@9gFH~r=Vcc}6yKqde$BTnd_p&sPRC)wOKhJgP^%do7R}7THgv>#8``cbP4YM8
zDU?8w^WpSis<|S9V~!*CmTBL+QtYsZNA-#p&4PD0S|kIZ68AN{SFryPNdX6SmEoK2
zq0@m5$M~>S%j|YE0~2G3>k)r%cfw(#JIuXk`ypV0Ug@JBr{=yF6&+S{Q|+o*8O`|e
zw^06bGVQXZB3TARqI*r2i37_}cx@h3O395CixbLa@XCg05?t@nR>`m7V!IrBp=3@g
ztn^hVhV4x#m?j-l3v$9r&~Ds1OXC|%RD1`43^3PCRFjH6Ub-BOm)fZ~6WP#_lQ8{c
zv_fDsxsn4g9~hBOtJ43&tG<_q#WWM0UWEJ@&k)z{R~D8Ex)9N@)xo^W$k-X0`s}F^
zno2Bct5zgL7jC)g%w2uac}s{9{9A^yH&>_--7E4&{WZs2(Up<-P_gmU@Gmu&oj;C4
z7uubamNuV`gx=pBsE)-{yPS?-JrlO2E;>N&<lZCom4jVx8fiM>^>wIG!KmRDTIT8-
zQvQ)8*V~)MSTIVw^onh{;s}9@Np?mR3m9B{xrN}_ES<9FA>(-~X>r!q$l*&b1!e;e
zEuYh(FQp-G9U-0h<f8FPtbbg0&Wn^(p*ZwpxEKZB%*@||4`+=!jdDwiX0R%0BV2*Q
z#F;Zk*`Kz0X_3Q;vV}CzQM9fazNcdq8s)L+va-QF(3E3{^2kJbA-NVtxOhdC$2S`?
zv5Nl4`#PJvopLj3w~BkK0|Xw@*>@J0sfdr^_J}p2A|1nVzhaVaa7ChC!=J=Ndo$L(
z01!q+3vzTe7Tqp~jQq!K#4Doq1#`gAwTqBH`Wufv+~<JjNg45uG~TIpyOzQ4^TUl*
zqINAMgD;uf4qw3SS}wn!HnTo^7_`*YBNXrL7s;*pIz^zxzlE6oK#TY2aCZ3|y97P3
zXm^Fdy8b*#w}O_pJ8IO)J4Sw4|EoyQ{GupYojKrZZY>vv2FP6)MWA*}fp+SrYECjr
zAg@Y3)Ltelp~z(V60lG4pyxnXIHoe9ZEkASXXmomIB+IM^&HU+Q`Ssw8hY$8z_S?<
zc0geYp;~n3%4SRxeMN~+#?|CBC4Bo=d%(}g73%`B5?=`8q(>mCm^kRCup`OXl~Z@+
zk62}>OVf6fK9bY585SS$>or2)2R^LLT1)#WX9R!Sj||@0(;kgy+J-DXNQ#_6P3;uf
z^yQGwZa?ik6pJ}MO}CSsXgZ#t?t_x=+6xFf?BgzkWs_hX-!4w5%PQlWgJ96NU(n*q
zjJkdkVl)jL;1r36Q*WXJiZBNHm4^IuaEs<6R+18H`?X9HEse~}xqA*k=<wI$&X`lW
z2Bz7GBZQemZzrgz>FEPi-h9%*#CV(ViYPA4<9X=z+N2|h5+DG1w?_a0I>=yiwRYgk
zzsHqlh(Cs}OlJj#7F(rA;j2BO)i*mj_olj7mMrQl)<LvtAE=$*3;Eg=Pjp7AemgQ0
zm9w>zrJ^=UCBOI|Jpn}xqt&?J1Ma@88yqk4W4s9Q4K~trp%LWYzwg(mUQ2A*0xT_W
zC)G)qJ0#xaNg`km`9{vgDZ0F4<hSZsnSAcE^p~Fqd&K+tyE{niOpHp;aA$h|hP_bs
zF@O6T1}qFj-9$oG&sKXKZr=V*yi8(c=?kFy`x|9jC@`OTT?)to6)*uW{(^f-U-aFC
z-&kcM9cd!Qmb5d~?Nuq4KhBcIeM2eUe|Lvnvb)<LTfz5G=ZOS`n;o<I7vL+l-wSq9
zt<Igif-y6!Q>m>73AV)|G<#K<B06&%`L%BGHpD4awWa^S^xANRDEo)iRCY#Wk4WCV
zZeUe5?a|b&;3d#~?N#6ms~eXv)z98sMT5y%S<pOQj_szWTNGOxcakVZ6lCE!#2WQg
z)9`|17<;-RS1Gy41le`Q-dUQ^)IXpgY$1vip&=V}Xz6RhKBLfz6k$Ae#SfAO_wks*
zLZ!K=vQeJ+M0fDzrq44)6?#I9rEWv!O+Bo8*@SNy^1p#HC@cBjn>Ew+!8OK1Dd*+V
z%k=!DLQSsT^9f_VjmM}(j%+W*+PlI-<kydgeL-#D=AOID@k18qY&RejY$_ZH!Vzi0
z=j0Ket&H*siw1Ntf(53(M#Qj>AfbPNnpg*8L4l#!H--%C_C5OB$FdRhve)Hcm6(F2
zlrT-_PVWH;$MQ0~=ZR~FF+b*`EpTj|EKL?=_#2$pyTm6f0;3*kt&ES#Gp8*-zR==l
z!?wD+$E$z=&cj45itOI0nj24OUQ76NWy>QQAb8%l5;%LG1M^<d;V@~$X;h9%#B+>~
ze-an~-iZ0hw%OL6uG~Wrx!pJ94PuT%g^qvQ=|=z>cc~bfm_@-D%xDwb>7D1~+300h
zCV1=~*N;fk;`P)`g)>48y_VZDyBSL8Uc-(p#}<`hneQXvY&A;Xv2tXal5cds`+Q(<
zQ2hBj;=TLxNv3R%l9&TuFJ282N9?O(?f`sz&RC0^O|e-^fh-TmSDUj%iDgDRQm?LG
zLP}-;Pmyb`p&Mx1OIr;ek521%J`z)%s(-ibeSa}{m9yo-oBsA4sq@uh(Xd$Jes2cm
z{mUN?u9-US_bb;5CHgD=&K5jBdqkVt@9oLeEI9|F?~S^5Eyq1KASmA?#P`9zHE=fO
zv9c+=&%#@ge&s~m;?{@x{(F{{>%g<dKqy}iK)QDR<NllTn%RgR%PetEF~2R~rB6`$
z!`6K+f8jmkr$~N@kOeu(#xPz>>9N#Hcwj_;VH5$bE??Y`eB8}Ox}VPZ7?1dC=Az~C
z<%)*g?>^iurHWA7I9~(DMYx$d550=d_L1bkviUuA<I6D-ZM7jnooh1vei0C;U@s_8
zz+Uc}AoS{7aC`rpKo60Jk77cnzb%2w-ZH6+J4qYl+oeZ1?a{5ImU2kj1C*ET?bt}&
zxLd)GyN+SOmCC2n7mrOitG5S)_wdUScE8Mblz9}SF%M%`NF@J*_c3&V`CWHS;Uj+@
zBkin^(P^z78<l-R<811Wl{l$ne>);F09<Jl!UcRO8(by+c^7ILyZ1;2JQUp)eqQ-Z
zj|)Y#iaZyK>4IN1h|10)pTD00a;Qa9RfaL$gtThuo?0kDuOIp~QSYs3bW5k`7oj>R
zQPu7<f%&H<(0bm&pCL?N@Odbvjp9?wF4}!19J?Abbs6h{yFN~6bcwFx@xSB-3|5Pj
z#}69Eo^UvG^;f>%-S>O{d-&<BsPE9Nii{$k?t2R33<8<L##n6!&e_(rW^Z3b^GF)5
zc$B=luyk+5vK?{0RT~keyyHjYFGoqw_A=bMHiMJ(Utph;>fGw$UADuOwsn|6d5rC@
z)p*k<#v2GnU#Ig;k^Z(l$OFLrte9{Q5yuGf2>CQztuQQPW9D2CD%Mr+A(R?0W~O%n
zCK680X3v=);P680Qdv4Za9Ka%6tT?q;Q8kP6J6qME%_<R-X}{kUM;hV-dz!P-Dual
zsx;$fdj=prB&{uSl5d!OIT4eYVu6or0j0nW%O?Nkr)^|$5lSRVA#i|jNeC%9NT7*{
zCHYs^E7SuQm7nD<Qt_H6N`ssd_CS2d9fC}wdk%N*7`s^~AE9GJvMZqmuPcdI@8zB;
z><hOgSsvuk`e8EEp;G73EdLqEH|D_F|EsI742mP@+GTNfhu{z_5Zo4bcMFyP!QEwH
zAcR10Cj?9I;2Io)yM^Gsuxtp6yI<b-yLG>+J3nToy1VAb)H&VHdCuv6_F-f7+@r*f
zx_G&XSy>HQfqvR~a>xBrC()D@XZ)|{K+GfP$-UQRCxG&YmSH!i9i`<}Q@g#WW_Bxw
z<pJ>tG-^7*j>RzitJg9k-Ys%Er^v^1dS``Ks!*O{`%mO0ID={$dPveI4uaiIJB+Uv
zK3205d4AZKN%zO>SVYA+^PI*|DBk3n*;+kmLx~Tj=P7~?pb=*CS_jXcN*I7xSBhh2
zDNz4TkO7^|q2_9EeOq;JpHZidJm2k?OJ0BaLf+d!UyB*lq;WNkGQt+k$f+4BBgls(
zmd+U~y1hqB9bT-TEC~lr%%RDjaV(n|(jBpyx?7jXT<K_J4VtOWXFGV0r&O-ZvK6o(
zFGkbrDMmepr-r458Pt*f{&es!Tub@g4@EunfX_e_Ml%_q2fh`-;iUQCWFjaWU-KI`
zPM`d10pA`Ch8BBzlDw+7)|rvb<U{9oR0~bC*tt2lrL*{FoBLf~)m}<vd+rh+<r<SL
zG^SdS)`Ce6_>w%)kq2@BWJK!*Mk0~bJ7&D6|G@cQj#hL#9Q%&fgLai~q1aB)LFZ8J
zQSqSu3xQ5#FH^NdEkn&cvwyq;1+$|FaqOZvyh_~NvZs%C(n}ZL-2%qAAg0DV>74c#
zLQa<E7C9qAnVnuBDxlcI(nN>ptFW1f$xjjeW0)>2QP1YM1umRJ)(Bzw)!7BYG*4c-
z!R_s!%aQP5qd5(;8bJh-QMP@s;FD1?zJOTO#SlqiN5bwxcMXx0i-r>aZM>qBCEuO=
ztnGlu?rks{a4;h}BcwxvWlY0`r|Tv7@rOp~cHk*)f1T5t8o-G4H!|_dq+a`1^iF^H
z37xqPk|)03%9E@ywWolih6&=+e|&bt6ICmA1LZ<zIN0)n3`hGP31pS=HPbTV<^jF9
zA|<A>DPKhKfs5)A_ib31d2bR<r!p)OB77_uB_2Ozbx3njv*T;gb~OC`ceeXh$geU1
zOdlE@qB>UPQ9wd@RBOkteNHK!jqI#1YreFYcJikFc~QDbjr&TF&*eqa^zU!sYTAmP
zarJk%!dvMV1~2oA{As7pSEv>qxHwMx@>l7L03q?{`&k!X-KIIWvAILM;=1TYC<~5P
zG$IN!k$&-1UR3Ti)z!44ptPRYd~))i$LnKLH$PqFLn*%zSVGu2mS|(D<4>KcnHw+O
z5HcOixX*1)Hm+fK;Ol}okAYvK+Ls*rhYj395-v`nxB6{B5=n$MptcUlO!ooN1DVAM
zIr0{*vx={nGM0tQCEor$LJ9JrK@+(?(ooOm2~r?L4OwtocvP;*ZTS)ndUY)9tHKp5
z5Sy}YKQJAQ5hS;6dXMf&QS_#2#<lV5MpJi@UuxOt9L637xg>`J5!u#GD!%6Hyy==&
z&(O#a*9>z;yJ^vGAK%2GalfU-JfP{Aqw35+s9z;RF(V;}lB1Ka1mKJS@t^Rzu-gds
zl&GHPsIE8X@y!pzf7(7Z#cm;Zr**%G=wg?TA%t~jzP#ZQ#;@TbR|!Utv3`ih#>1i#
z+bY=72sRMHp-B`hw6z=LbfqU{sobLz^&QJWgwx#UE=gG`FR#zgZ7@mR0Y86@G5Rqe
zZ~8g}>=n}f)16vYuynRL;jN|%s2~sffLG}kiS4bTj<0=;`9$eG-D>}?Dw7+nj@;!@
zGIYAn@-Nidx+8DxYZY|!?TQj%kLY3mu`V&StjDG8_sf*n5?U!gT3@0_wZ{xH9hj<H
z-YYXN%&uk1-cP0$M+ilC4#59IqD`sZ;k6q+aIC#SxZxW3wdVcMuH(Fh${eXPK^$Ig
zv8WxQ(^pSgD(<b^Zt{C41%Zb`N<1X&*VKOK$4M>B)1L$)@IMk@<|}}}5D%0jj2(jE
zd}d_uNkbKLN_CT7UC`v$No?8bu<(=6KC4*iUi7{o5`L^A2~2)MTY!VVD1~Qi`Yu{_
z-lexcTj|}(My>Y`kmvgZK1^oS--_;EbQpUh`aWWB&Y1|=yu;m-Kb&wDNzib!hwJ@T
z1IlRjZT`$OI7xn|X#cquq^Bu3{rT^_502#i;1tcrk>>Z$jF7eUpH7p0IllzH$6j<I
z`q3|Wi#GNoxu2;J?TbJQ`ss#sIGu~RU8L#v%k42{BB_k>iSsO~3qRomP_j>2i7NX+
z8{w8viFFza;)TxJWV(d~s~o85xlS!X64Glug1FD+dxP7=bi6A??no>ny2>`i%9Std
zE^OytWg|8;+`fp2WPEqnW#+44i%A#L#rX1<+F8pbdjN0eb`}aeSZ_U@@-88~q;lZS
z>U>>u5i$#w%WpZ2oyY*h=iZ23wuXl?3v4D~8oQr@`gvsOlz*u1l-yUsg|H>+!{L&j
z&D;>P?z$|$-Gf8hXAq+YB|eGeRbS(JB{b^Is3)8%mK=TByc%xFbVnk3Uy@jojt<23
zmD&jJX8sY}CDH(ppD4y@x2)yeWqV$PW^r;qEKQMml_lXd@`)07q5VwfKUR2(Bnrmi
zkdM7DI6I=4Q?T(Zc*CdF5%xac<7x^Wc?(5fgxjg|N2-f~fNCCdmnIew<cET<wXGzP
z^O)(Tt=*fl8XDUzedzk{Lno90J#zOeJ6m9QBMae`0v`wZQQju~5s0q7V)qwKz;7Mz
zt@v6kd(Hix4JB5q^0#ayVu<|9C5yJTat>vqS>fj&c%8Ubw6l8av5w(S0{SD$!>47G
zA?X1gmtoO*oQ8J_U-?brX8ueE+kj<n2|FRR6NJ<*awzEuUwW{YtgI`zPrd)4;=O)o
zR1M2jUMtN4vgA>6-~WqtnJmO0;k6aMA%CFKjhOuD>0p_Nvip&Hx#tT>eTn1P!R)^|
z<0b6&>X+2msx|*qDM)wu-fq6pN#ovhj}I>w)h&svpulJ%a{0<RNLxqVz+ULQEmqex
zKU{T@WF2RB{!NmJO_6eD>~|}~f-Lf{fB(p}3`+Y*JwWf<SHh@?(~6~h;OH_<Aupf5
zxW`1eL&wju`bmFj1P?iOgM6vimhp*nCSi&BRDeWrj`oL=I;kTyt?p4|@sb#lBx)``
zg`vrujab%Job~RDZ?<gZcKLRrjG7}fjTIWr4-5^T^9B&{a3dJ&rqLS)%oIV*Z_8d8
z5uKHzfq*P0xlB^D;@RJ1ysgnc-^HWevf1?%^vc$8%O6ZcM{G)Z+)P$mjoU5Y2a7Mb
z`h9V<nQJWQwmCz|cC;rtlpzs*k@QQIQKukDuRtl<Yh@vq(f?sf45BrfJcDR7+JalZ
zJd#j#Ng~wMAEM7Z_&(1!P1>6DKXFz|_URBnnayw=)Q1B$o4BiK)QITl=A*k7(!1Bd
z(t>&8qOXgek|QLZL#RE@nJTvw_7M`ZT3U?XZ|A*V+Sy#&Mc%$|f;wviZGf2QVx#*}
zXlz_<CEwe9z0IeZMD<&)Pmhro!8h(L??#-~m>KLCY_!u;cGR8k9lLh$b(!!CGW7u=
z2}%^cwm04wzS61|=NA#@HA)x{9Bb0UFV0nO^_7U7q*gZyfBv~=-<Upno835Edj4OA
z_{-(F^7x<Y>X6p|E?@*NvLF>2B+LI$|1Wz-@i6x{RF2ia`Om&rwxvgC)rws1{6qe^
zKs<gavW+r+?^mYFVK4eE02v=y6thDD^R1cJ5kvnt<S;2^C_{cy>!5aCe>H2kdhZ<3
zB#;@ZOHn7@S(o(sOp%B&89hP2x9gDR7H^4z>5~8M`KPss1do6JZ8n=N#aAWkmEiKt
zt1E|>Td`rb&9C6zOoUGJv?3%97yKf1g6`Y;YVe9IbB?>I$FDeK;16ZN_utKKOqixA
zld+k;bXctt+_7p31mb9{g-@eQ4r9_ihm#8LE8b2<+bAjAi2i6K4iSlJ7<yk~BiQ<o
z`X`)ZU~7?Zh*ePLFxNvQD(xbN0NYsMzIRr_I#=l>s3y!f;iT>B3;efhWG0G?q5UIO
z+tXHpvtpq^VfH32W`N*dsUI?Z{^usCC&QD&VzOXeR%RTjnDgV6l-J5ryfK>WA!*!l
zYp!O)S$47rX3tiIifbbq8K1|@l?zo$d49>ONlV6Kw&N0qj#jTT;*l@abb`NfKSV}s
zq+N>@_hGLxcbS}h?4H>xfPb6SF`d(~;yd{C$tUNFQ^EIJ;!?oSMVH5NZqp~+p)THH
z3a%&kfsM(^S_Ae*KCV+=!es<B+n-pesccf398W@$=s;7MD*7rCM2dwmkV97K%J&H;
zDcQ$%I<GVe6m{i?gz>Z)P-p)$a5-g``1(Wh%SrWh2GZV*7-@u_WA~MswDm+ENQD($
z&!6|F!xE-Ue?4FrRPq4Im#8;AmuTbs*n=AqeStTlvl@IR{-wd=HT&R?cGMtm#Ul|V
zOBbSWtxWyJk&HsRLoNZ5;M&bMWEQhq_P}#l{{6%wgycCuhz%F>#G}TMkXm6oVke(+
zX>T50+^@wHatm3d8*(}cj1r$&kI=-ai0rTjMxR_yb%2_2BDlp0HRvKu`~nsA3SrZG
zpgnk|5%Y>a_aSco4Sp`_I8AsU+8*+#*I?kF*jq>XKWF$azaNCBjhqwMLkGMF#fQ^;
zIEajxI)$6<h%9AG%S+B*T~P3aqT>wNi$yp7KNSDKXtg?AtOu*?`jj185i8KT8%1*j
z9N>yfvTAv5Gb&rr?~|e-jBz5>wmNFRDGNsatZw<gu)gX|U7!c+znC?3Z{J9j{$(iX
z-ExGAx@Y59SFu7Z7Y}k+!tiaoby%$>N5}x@U<H&U@#roq6-E@geB)@bKUnuWT#R^<
zmP?9V3?idUm<S)yk8s%?NnvyVf@6iASfc(BX!`66ngq`wUPq22Ye3%-lW2E>`Rks%
z<?d{M|FC)Uq{^4#B%@~-_+PB<;zpaKPU-dis;nbv3O#REr}X4zETr(!cd3D2Bk1f^
z2LUBFQn(umXD_SCKEaE1y{t>Hn-rva*!53&VWOA5o4Q5or?&WiPd<Q7E+l1rtMLun
zlc1S}fos!>SJN0Zca6Xg%==;0vK>N3Vj{XHL%7^KtRB&S-)9{{i_B+}jYIk+yqeW{
z=<5xyLyPf71<lIl|3X?byRg|&d)aV8Gm7no+*qdCI$xsaKc;oO>ugNy9L!xw&hsL7
z(&BM4)o=={f}&fR#UgMOd;4dsf5{pj`sl_sbB2Llf<Pjfdf^~9<cavq-f)mD5N~Jf
zCQ=-1;5uhZ8d{sX_S?b8Y(lGWHa&Z;CR&Z|T$v$tI;4#~EtUdA?U0z=%-3DC`!VC~
z^SRV**y6oxL*qTX`jM=aw^?%dWcA^~=y2KUUuTh%9R`==dY{N|6k&=&GbXuZ<OU3G
ziF7Bw4ikxJFh8PbtbKd+&!J1W9LPtSs#`Dlh}mUOAh_rA*4kbSCudv(=fXFF8=^*Q
z7WB*So5?8jV0~Q9`_!MU9dQj7*d8ZAM^_b%_#2ey6}1gb8sn}l-=jvI)U)HRZRzJE
z>Mz%H33v2|u#xwFmL))Be|a-C_G^I8b*Gb(>{?nN9rBSoF=ob?1H+ym8elKhYtk;_
zNFAWm#Oo9pBq<`8xsP!DH2s={x~#F3)oA#UnCpFVmi|kJu)R`EWCZx1)WrCzH7+=L
zBJlhzG;*A6qd=P3X>>^#f8wv}NX~%T4^H{V@qDpo3N%S<dq;hLXAcLq*t-}kVS5pn
zqS^nuF#QvVOmap?w3~CG8kn?^?+KQW3Fn&M@coxV7B6{8*4JB7?vO>%o{CoSeu6JP
zqGQe9Ja}M7X{%o6$DK`V_kp?zPp|MWvAtI$^x(eUh};fo-vX%A+IuGyA_{(nxv6p}
z1{!f(&roHTPbJ^f_Eo<v3-@PSaKQ-38oAm<#Tv|i6<S{5W4IFvI2m1HN)u=lQTlkK
zs+vRO1LrfW23>THDD<-H!*3C<S(`YMomlb2ZBrMuvNHIYkdr_2DQLc587#rEU*jip
zmHI5S{O`2Vyn67;v9meia3_Y!RUjtN&7r^nmGK<OQ4qbl!dBK$+Uq$%5Rr*WgG$r*
zgjdh13i~I=a~ordwJ7w{S*X<~#3-};BS?tMqRg^VIi$tnaBW|JoqLdoaF2aGAoKhq
zhzEJaER!Yj`PtT$X%Go=M6OQH?2ZJnKbKYjyJQhUudSKFzIuu&e}Rvnz0Z-SJ~?(%
z7wZ>n3tY*>`~-3YLMmdv?mXpcS9O0?`S<PTeK&zuDc1tjr}SJ=^;ni%21}a#L?ps0
zVz7Ip58Haoj^Gk!o|nn%_4Q-onah;GN_KZUs#MM<bp2kFu+=|5Yn7{YBW2o?GjYRc
z5G|*RNfJSg_^Q)xBvxZNr^l8f7R+ykzc1h{9_zV_vdAO`3}%Qou5{HUoz<88sxC5_
z!&&kRh@+glSql8oA#QpkC3X-P_5D{X=||-75KVlnD(Araf;Zu&-s|CNj4E$yT`~g@
zzKH(Lnr5^7hg(aYpjR;(mBwfo_k`PSoq_M;Fe7{d+2*#da8K6r)QX4P-NPt}KE;@`
zTh-*dR-31f0!`^#+tsYctCPbA=Fqg&b{@kW?Hk1fBlPn?Z&KFK?`v52?EOFI5&HA{
zY;p}&bLy<fV)*$RGl(0$d8K)hCmK2(ztC6%%?T7D{PSN+w0h``HoU$jq9N;)u(g$4
z;Sa9}JetLRf}`7A-#XhlxJyMfsUyiaFKz$BmpCRb1b&~Cn^#*2yuk{;Tx7lL;I&;M
zo_yS{6VV@vFbuc*pz;$V#Nc~%tqj!RYQ(rFC^FJN3x~OzovRV;P-6aiEu*cPYeR|@
z_~^4t;??8rQomO!`4E|lT#A>qm#i8zf#4!PNY7!Z-w-?FaVYKn?(`*NZ3pnx@S1tr
zD8gXf11$Oj8w}nfXw!W!v0MZQI8xkmRu0_<S@G)sHk<93`)Lv-rsH|5&`{8>uPBNo
zv0$<bqxF@bEGIBC`<F683EsDFX=rzRhKEpE>R`Lm94YBFWXQy{y%)N@pWZ8oiF?*x
zUUD^Hbw6TeY`x}j_x|_tM?~AJww9@?H-+m-&Va1*wlAX@y$j}ik=KEF+cL`?_E#U^
zToaKtX2C#4i=4NQy6(mA9VMtOw++&DnB+#UL2-~k;)y^mk|{5e5p3nHnJjYpwS!0R
z73_}T<*n%E+tT<dr^}h{mi`^<%77F`pgb_?+-W2IFrQUs!uT`9n@$uxs&`K*_p`q2
zC6I<`PJydYErawXQ`#7ZQd&%q{-2W4D`rFa0h*Ui>MyF9=wqJB$I%tSpc|dgD`DoV
zub_>tA2539mN1!{N*!cO1O{@bf`8ice&q~lb|hks@Q-^SRDJ9~L;Zt_jr-bUd1)E@
z#V7Ky0K(VQ7{W}bR^h8dOd>@_De`U^r@)ob+r)ISoBcI|nd58Ea>~(<lqduAe2o%F
z%woTebE1QCKZ;>L$R5zgJwB$#RDOJ#_rsVEDlDQP91%Qw0M%x^ZnGE{^X5Tf`yG!p
z8;ePDoWxd4#Lu=$FLo(v@W-Db7_E7=K)t(-Gv9cjRQc3dp>s!bb)iXiUGGl%YzHWz
zD&g)O=jgtw7ZW77iQWhfk7E5O_`2ScMXAoRQjKeikd?2aIii1)9!)>8B@gvn-fOMk
z;%`7>6B~yWMQrh$;xvm?b{{>@VOnPT;`Xmkgs-YLC1ZpIrT(lSW5+}rqK+nJg0tsT
zkNbMfj&c^vO$wEjQX1|G8s`IIaj}5jL6f~LJ~Ezaahrj9+$NER+AL3ovufok<J9?I
zK{boMda=%SR#e;m%r274R@ozuBIS+<OA#06+utvxEQ~&WL||-j%rXT>8JxP4I5wyz
zf@ME7{g!JBztxWUe(op{i_81+Cj%G6rzHDd$>>+xuH>T}6cA&+8C+V1mJ-)o_d2JB
z3U|_3hIp9gL%a!k-+lV#8*@(@+%rmalwaIiKjs2mTq)7Y8C+X#X)$25gjytwW!UU0
z31RO9)ZH`JL^BQq$@^5~X}Hs2y{3$B$3L-0C!$@C#V*6%XZFQ{{s1u3R&xt_V`>e^
z(E+HRg2|XxIQl&7&-w+Y_2{useQ9j(-6B!*ufo%K9>Z~_n2DcNdOUe^3);e{N5xl{
zi+|q$s3ccayz0X7LHL(*h4^cq?dx~-MvSbqcaF}uPDNHe?EP(!>z?^tx^rtf<G1i<
zhGg#T<XE1Y>8<U>@{ET2WW~!$+}Ag2W5_t#c)n|1owO5wD_KqrE~hRo_=-Pl2TTVq
z!ot^|9LjLj%0G3N8RdiD-+T!9do|FulrI_b+8#(s952fay?R2+Olks=HT%bd#L%%*
zu{5p97?F@hYcktXp9`x-d||;T4+N<B+145o=x<n=2Y=!kw$G$l-TP$)nrg5MxI^vg
z+1qBG*15b0)&DFjEq!uDv0F=r=ZF#on*>LTc0b%nZ<u%A-UM`Yi?2N%FFy`;g!*=L
zidtB%J)ZaN1CQH5saD->fat0vUA1=e5WOWJx!(PFxAAcGbP25R9vvnU5L`Fh8Mh1^
zJhW%WKKBO&v;<#W9y|<`Zcl7Iij=l*zxtI}dN#2IdJ25pJDAwIFP-vt=g|`JkBX`w
zFkUbwu-Xu}f$qasazX<hZXkDuH{|Ya)jX4>ce6uLAtUXxLy5)|L-407uYsEd@YP<5
zXWKQzyCWn3fIwHyPN;1}*VciywsxR(Wdb;NLz3=X-GKx9tf5`4p*QWk^Y<|e4_@L{
zkFXdkfsT-#wHqJLj_x7kHS=!V<A0Yc5q%r0e|e`K@$w$-V-lNtVe{a}{Y>UeU;}S;
zW0hf@{wbt632xTW`4r@Re>N6>SOZL?fn|Y~TfjpqCm;1NP|A&G{o%26W%lNHi@L9;
zqocQ?Gl+I=e?xZk>B6Ti@B+5Bb^-}6{uE<!c6)%%lSogHUqk;GaDDu^<r5`+a}$=#
za)k@r23_q1+_76dAx{MMM5MmFht6(6S^`A#Ha1@H)T%xB+<8b@bl<PR_JCYE@y4Av
zKoQaf)wS<>*ndcU+=^N0!y9}w7QfiuBgQ_3t-;UcYVvnNdbVd>uVKvHz%<q+6yh6r
zzn#}`MPOBvh!X(-7d1mc!8eKdNA%SmLqj4@+Hsp3)JVHz+K(?QPwBhYUo4y)A5+bq
zTb<VqrJF1m`%7EqPL-Y=x$`n&JS_oHoewat-fPSC?)&Yh94!J~z)8dX_i_;Rx>7ge
zbv{3*a%Dp@bM>|T%^e*Ut8l`w!J(bz1;l0kT#C$gaFS?cz_Wi&SQM%W^?s|LHCm=W
zp(v}gW#iq0`bbMr2|wGFD}z+$e#v2nPH;k$I?dt}mt&hp*ZOhj!!z}+<AB>mXu~Ox
zSY&cddoB1XKjN+PF|DVsWX0N5^kR=OkAnR*rd)kQx3-F9a&FS-@5A}@R`M0{ZW{Dg
zqe5NTl?>jebL2lV_7nSeNLeC#>;<8~jnItWm@?rS?&un%_FmH_Cw@#zL$%J3NPOrh
z;{z%W2FExB@2w;gkGHA@G4L`!f~DB2O1qBpO^T+Ad3A%SZtEbAqI!MmBr8k%>dmd~
z0u8<I<$#yO*TQpqzf;MS9Jp}NufelcbpF<lqz&TvxlAD^hU_C+k5Y33Bgz(Qg649Y
znf!oOypd@nw(5?*^PqN@)CBj|*?c7qWc^KT#*`V3?#wND>Q<p!9l$=d=O^BUEknvR
zzeaigEN3?`b+Jh%vHV^m0oR7?Ctsy5_kc&i_zGcr%ST*K%T}|G1v<8GHg3F2)KjNc
za1E1Zus$0fX0)A_jEN0oavw!bt(+EmikbAVcX>t@IUHkiaWIFu7qz{8$BXlVCP{2>
zv9hjH=OIwxFJ`4@#cv?J!DBVP`hyU$C|q7};ll?Ls(UfvFLD>8nE{i*;kHvlcC$Nv
zL*fo7^w|+{dfLvpV{*ncD*8%DEN4Q#Gm3a4;Wa;FT~F!)ZE(^>_9KWMY3U2<r7XR6
z_P2ZT(uj)6hiTBiRr<YNzD50ubzv$T;Cx*cQPjiA(mzm0!1We@VM7m~XKYfqqqLRL
zTb1Z4M`>b<4uX>L)!nw>E@jB}OjJChHq6ZgRqE##2;5)Y=tNprOyuS(n*)~B?9PkC
zWf$zKGogu}0aETuf$#m^;-Oo$BDn13grkyf%DlhCPO(?Y!h*1c%N%T!5aY@^KXW*v
zw*#@wa`Xi$qpbms81?Wet#FhbZ=G*tqifc?M65|~(sga{uQ55H7E*<C83RW~7J9a&
zZPfvq3X@>JpN7!QeSI5PPOmWPCB*NIjwmr<{<ZfngFqh#Ger8n97-TGqei~-#yhAM
z+-(zO8n1ruG3Xdgth%_>3Adr4L*z5};N(Yu?glsPRRBMTpVGfr_2EX@<RnL-*1RM=
zH)WzZE^o;Qb}45i>Fc;j%I72esg$?8J8|Va(SGD5A|(qXI)<N@n{Ri*sBvptZ)mx*
z!1~t41_k~YdgBM6o$1qSQk5;pAFBS+QK4FEAF>?PT>F=~YY08rj|~p?@>+lNkT}fO
zY0ZN+Z2<3*6IzO-!)w~GE4zOY;Tqrd&}=<F`%?-p=rg1al}@(y+Sp&$_j^ak7;T;H
z+IsQqcI+ijv18gA62e<EkW@b<B4NS}TZ}g|rf#Nb@<n)w%Y>0XN7m$Ij&x?>OrDxg
zXni}w{e}Y#EQ*4ef}^1QL1RN~gf2Xt))<L#IY7Wbwpj9b1D6(|<_T4Lk%tro5rY?1
z{NUMNLV`N*LTg{^=n(mXo}ehYuArz|8HB8${~KeB{c^Ome~Rg<jmt*?>*FEembj9|
zKq=JWoz(%@1feLDA5O}1_l@y4*y<ohAuD%cq!mHR47NfmQsz0433yy4<La<Pj{O_w
z1&B}*!Kkso^w3{5Tvo2b8208e8T|jS4=0p$zW;z!SNM;{3AdB39hRJq(<Y5%<?ag~
zk8?2DZ{^TN+YJPHs0M2NO*V_+NnXGXV;O>fQ?6Jnr_1Fpl&tP|RnaEC&c@%z$e#fP
zj~q;5798gi5l>MgkHloey>4<QbmE*!271oiWylko*JO|eUFV|JB0ZQ5seEa{P=98!
zp?rb(ts72;NaIhd8kRG&?`v_f`2!76^<IGKSC%g$nY+&3gqtxQJLQ}4**%tU;nDG4
zTf`R456~o;mB+6;TyPWooOGUJ@C!F~>9{CZod_z%hRa%j(+L@3ov|E};C_E308C7a
z*?5ndy<D0cv;c}U2vgEtnOj7e-Q+{3NbCYY2<Hf4kN(G;=-5foCy7qDaMN)R*KnHi
zCtwmZ3^nDTcY$Y2VHgMU;F(*nr3cU#c^pFJeLAdBio_G5Md`+nRG&BTspfgVuat!^
zh0y1w^c?GU)GzDa?zmIiKrg9*VFYA7<s!!ICA`)Da!Hu3^%g<;^U(VkoC@=fTkXBg
z;kIY*LNKL$;C4$7smp*DNZv;ltPH?Ml`Z68M8y~s+c%t6TeCN7<=V+Lju)-$m^)!+
zb)2cjdXV@?d%=)Mt4+VstV$I~tZH&@2LDmT$(lAERzk4Pwe*$+Ln!d2z@%$SZrg^Y
z`h-q^fUUoQF0`m}{6a<lM0yr=ayuNCuG7XX#yYv=AVxFi^`iN@9d`Is&vLp?!eW20
zaTt3@QfRoeXRev$yX(;KZchu7k>dcy&8O>*$XwrV%h9YJU_X2|+x4tk_3s@}s!CU|
z4SA&qI0f@9d&eNM?J-b!s1%SdUrQWtL9DWq?Cde9=u{Jqn)e=O(lD<FQ;RdvFWhIH
z0?GEc1E-+1#603S%`6kieoBB<hV$yH4BHf4RBh}>TF*Rt<}h}P;OB;_s7<ct8{xYk
zQWlOBN<E!*Wpijq5~%W%F=`mQrGP@AZMR%wq`wT3X>St?&^97QU&6q7M$WiIt=9s>
z(cPpJC03ld{gQ@BHOvpGPQ?PdV-<M~lF2v)7_t=Y1ZL7CwYJ0^j?>c1`X%@WL~`d1
z^~n5#l5-4Inq=mrZ2NQhab$F~^awOQ+qDTHyeGkvMR2```j=n8q=Z4?pX+zB#EE=z
zisrq_IuS6S&by31fQ8PfKE`b!Eii=UkoSz!(IJVGr}GtM-WtYlQYzPXRBjL1qLZ|8
z{_~z#ds+e79)H(QFvWbpQHkHwH`Ci+%Wd9aZ)z%@*fC%JI#^LrJki*MJWmmwsSqW2
zwIXLUJjA?xEPKH`oV{uKHJ=2;W$cY*1^VxR1?Mo}kyAL7k#LtETOm1sA)=;Nj$!<P
zfN(vE*a&--ZS*j>_>^tXsCq?)dWs%%m?YnMsgCPKw##7^Seg4OzFtI9XAS)~ajpnM
zgf}*yM>gcoJeP#IA61ewR?@%%hPzE*hAVRimuaQA@G%M9-iTXzfgY<C?NK!uhFTBo
z*T5*iY@775?+y{>L-!&^Bm@2s&hkFwo^1>}C2$T~g+fY#u0MP%J24qf5XP;${*pB>
zDv}7V;|Obb7~7hnCPV#LO~?q@g*ZRr+YLQK?$w9U-33U9TxG`_Mr-lJT*b=mP>_nz
z4t2z#`8pCuVaDG8H$s)oPAlT?KKE`jN|qb|dP;ya8l6qx6mn9zq;gtNt-`cfSBZ|8
zB@q%oxLG0_>}jFWbqc8|a!41_<Y51ro-2e6P3L(r!g!K^KY~0|#Ta8iHX8sGkkSN7
zDPXryW+p*tkM7Zk0}fTOO^_BgiPKz&lE&{0idA5yP7Lu4TCzLiVkrjuB=kZof~SC3
z+0WAi_8&YaPLY3dBX?A4M`@9C5n5~ULbTzTTi$DN^{Gl<#~|88jnjb?(TxtIHA^Y*
zCMO2fckGhvH@bV95J*$I!N3ARB>|AX^<t_Rbds}%ta}RKXE`0s)o@QE(x~Lf&*jMQ
zhafwG=hL!q6G3P$xkF!N)OM#pi2&Tt#ngiz;+HWM!_w<~wk(m<`=GtLXBH&FkStK$
zawV)o-$Bz6;Qv;T(QBW$2NL*+!y2h$L;7nYe|outCUmV=2Hk45MMC{YKgy4sA!VEj
zjR)kU;X8{k_KSRn&aDTDV;~t`6aGixK-Oo{10Sd8m>WN7aDPSS7J9Y|0XAQt&}>mn
z{MFlTEp5T#%bZ!`|HgpXe`6l}vGHD0_FW(=@$e=bt^Y0ck^hoK|NWm&(Y5gnNDL7T
zjqGk705vP=`gi1_lde*B;uhY|D{)?U8KTdQsx*sd)f#;qk%$LLGfoY&T~DV?eUUrm
z;ANT|L5kd5E5i@3<x|u}0J6v@Fs{s$ZK6{tp0I|mU;M@6nRNr6t)XAQTPEXFN=y@t
z!;jPXP2*Qx?l<xL8LtEAILP_HiKq%AF7o&iHTJUVg;4c!Z*VDg(aB$CBX=^%D$6BZ
zw4##$iTQ6;a_ME2905$yuh<UNr08V@V&U@E4YDWF6I^RDb!c1WfS7QZWac_s*3gOv
zJi&g~T76k)jeN4j+F$rY>~Hl6L<P~snfH?!f0uY80&eQhkiNjjM9G)P9<Golnb1cX
zd4!uy+AWFL$(j|d^?)=f8~FM6ClkXLJgLW8y>gB<u%R&@xK~K%SL$Q!$)R4OKVH_l
zrCu=)kbS)Rj-(0(0K!b}JeoKz;XHaxJ9>#;qEZBtYWqrVmm|MrOO5v3B14;(V=#mG
z!U*-$73Af9<dMbtfpv!&j>@)SMW<}mk-ye7m)k?QU?`hLy@$7fTuLNHkrz^tS6AvU
z<-j8~uXPXl4x)7~G=+fveA)s5n3yML1>Q&Jkkc>1kF@Rp*Az1-3%s4vq2D$eF4#pt
zHM|KCyx9zV@;&9VgMIxH=*um;+H#@Y-jJs5KFzI%z3wgirJrrmEnLt|IaPFvPYKz0
z*N7Xdfa<XXd0F+9Ud$ag2<=D5Z9Y>*c;c(u0tcIO`8w;7H+ch~$T=Uz)<FR~tb*Y{
zm1R~*DG3gM(-4ss59fP3idoTW2ijV~WDqzZvNgVTdr0eqK?Gsne8P*kBNnjr5;BIq
z{y+%yh+03Zg%5Cgk|*_WIsFJIW_DtZx!~=wt(I2gocdCX{Ar@8POkiVky8ULe}>a^
zrjm2b8?SnXQ@Dt+Y=?LQvW3epv=`&+IQ{A|hmIPUIs@&2kC1KI=?Y6nz1bZp`<Ss9
z5{#c_lX|DJQ0m#PSecdbjM4eY=suWa3h?xQ`+W5e+3~*ev)}-y6b|N*JgC3}CO8X~
z{Ob6u5Mh5nqCjfPe*gBS`naDgEtlHy^ERLBn-LEFHocvj=8mbfZ?~yE1~&=oxbq_+
zs^~)id?xtkw=o~TFx##>Hg_qt=OYr(M~3Va#lrDb;30A{<O`b?=sRyxa*f*(>g89<
z%t@H+wU<6`S!^#2Oco|0G<WX!-V(mtkWKD912dx(ovbQdS_TNB*`BNhU2=HR<wszR
znqw52bkWKnuDw4HbP%vceG&5?0j;Lgth$i`Pyx<)PuFo`{;4P=j*cg475zqCJ(#d9
za|bWOP>2_uGWU5Nvzu39au0ieCRn*@mUeijoyRN2Fu6R9-+rBHH`2cei3)6-+Mv0b
zf?^7O99d$utOFJJJnkaWB{{?}wr-pmvfw0V>nH;UW_6=`x-!@`Q!r?{cW9Vxo)9pu
zh(49GOH6ZQJUJ(#7*S!B!v!zD$U!!R*vnTh=czoS&o(b_HMqOmy7y1~B10aDHM4c*
z2BK$Z_c?~;k}&W*FBs9WEfBGEB1u3su2o0chu`HbO5iSbhR&4pW0&!hcNf@tGud+8
zRK81!S(B~_wg(-1iQc6w2i2t^1YioE!S6C!-L#1mq43EdZx+)$lF{0Z7&qSTv~-k)
z?HL10-6c>vk>%5p^6ZJYIAO7oMcf70-uN<*#i!{?)*n2JS#<USGN7{UZBc33k?KMt
zl^4%GAG}V}MfCp9=4UX4PEOOk!3uOW?wK(-ixEUF8K3(8iiab3D@-E~*x(4IN3CE;
zR;C{ahQ%d7BVZ4s9+dI^?isZN?_esgbE49as!GL;f8_-fVXz#piO672>7nd_rQh@m
zb#1prYl9>3i%!`~5%O}^w<@O{K&b-r=yPg?*+P~nu_O#vQSR43izYXpkmbO-a{So~
zQr3}r+7XVk==aR4Q=lIJm=o!DU<m#&6jDl>O6tn3B*t1q>Ky2T-tXyWcIoC(-ShrP
z#W-oOP=_HGUV%-jl6*~^mMW@7)VJ?ltl&zfrq>~mG|D`x_pbH^VK5HLpP@#t3CN(v
zv)7iPoianl>R<yEogFUgrNRTBe*4W*D?@h8D;twYd1v2pra1#Czb7oj<0?NH<e9Rs
z*&R?z1i-$Zhft?~IyPzWp}%&U{~d8Ga2)1fSrW@m{-d;K>fMKMWAW%P!7hO&+$t@z
zUj}5Vw5ofyJl^!9Q$+TpT>_nO3pW|$so^xL5+(@&syvSW7-PnJuA12ln8x;HkQi$a
z@7T*wa%)pvu@j)Pvd9NA|6qtDUg(=|s1af#r1jY`+#v5q%^fNojCxvuQ{uII_?4^?
z$qTr<Oxmb_)8Mg|H$!iO;#}MKRhNN|R~JduvdQ(~s#7Yw{jt0u{YbQohaLV61q-}i
zM>IK7kviN|)K>Oqrb0b$NMr6OVovPCfDs{mtAF)fwia-&84(MXH`@vcpfJR<ZY{0a
z^o2y3kYNhbhl{dKVGQJrn#XeI&3r6u$|-)Hsr2;<Mb{HiS&n1yk|tZ<uW5cY_e&wk
zOaHK9)CEJaJ#DEgNeZZuAcUDCGo)X=o@-oj79&nq3!P{W|HN#x2UG;+_l_pll5%Az
zE=8GpI|1Nseg9t*?=iUF)G-c_E7HXJgD1<g`%dSS2!|#0Uj-jAE9=2MRU@}_jjDr+
zB?{-i<)oWuH{_(E=xXuQ;E}4fqIZ3C_J}ArUE2G3mLq@g#+S+<<h1%^xv+kt^|C1E
zoq3dDTWiI#h)EPphjIhHX9o$%@a9FNxb?0*>jL0NK6SX@_~K|8MXc=Jb|C)Pq5MNW
zJ2kkeX6{c`N`a|Nz=o2A(w;-jSoMJNGb<aY{-w6k;i3B0YtMpGFg?kR5M%br@h&HT
zb7A7Tzg;Apw5z$}fQ0k!D5Ic2$KZyBLC_hs2JaRB#HZoH4|avW`~Kvp-`_t2<}uk7
z-#37igcf!U3~6uGUdX-RRO*gWT}pzXxuYCY;YXKXxAbNiL$9{J!{!HYO`MKefU-nV
ze=B9uGhgsbM{rSmfScx!Ut)x@Gb!zTP(HR4rTNI%Jn9sCxfa8wD@OKBwLIorsp_nG
zKqrj~Z5w*xhk~O!m;u?8qEvz3^k?)!U4Ty3s(&l|$i|N)OM!HT&2fJtmdi8VACEeX
zF<k$7gPaAvbr|5Dj5ezqJ8nha*jnp7;M=QW7=64ovfcp~rKh#TyZh2hju5Q`T4~Bp
z_dmPoDE_ID{H_H<6<eS7Hx%Lke<kO2j*J4du|f-mcw-xpUNJ>y-5WPW=iqh<MWD?y
zhUJ6jiG@j3)QERdl{!q~Xk{@UI!!qFd$~3Xo=n|AkchvMG+M#r>h`jdAtE0eNkQzS
zQ*$L<D0h;b+);q%c{eiGT9YgK7T?=!V$#sp#i#4r!rMCct8uByL9SG=id31H`sFd<
z#}_mVsm7wzUvi<K%;p$j9;<++8j*t+Jdh^i;rrE&_~Vdv*UzVhIbDft-?%Ua)H<Az
zcQi+at~#^R{>g3J@$Z%0J$A5l4=(-E{*>uf38DbL&~r+x4fJ}iuhC4k=leHcO-1?8
z6HoQ|C1S?CsXLTaxc-Xnt4Vb1t-c<?{vt&tt>*OwA=eT4A>;czu><S$)H`t#vMUXl
zv>;Uq-O2a+Df}BZi-n%|h%CC?t@e{J_Ki-&Ta0ITpb&5r8kKVFofupnIkX}vHYC9k
zjJ+i3@mc$K09ku!S<d*ZfL+v-@6Tm~Ek$5Pn|)1A6tb<C^O862Jp^Eg6^P24eGXkM
zE9+@GD6Z+dO5Pt8X3AE9UI2#E>zB#*wBb%CRIa~U2o7K2uTxB$>2(b(wi|Y3pex+W
z2Ohbxx8cc;I_t}b*BD;WQ6D+&MURWhgZRay|4LHjP3+&b^Fm-sI$CO=f*j-(4AhVN
zy@W8Ibg2TE6+aU6CW|}-vBpc9byxTW`}g%8XUKAo#KfIXUOvfc57n#mt0gPHoqYj$
zs6~McHaz1wgnxP2>i;S8vI@iutgq(Rz+a%)09Wd*gPN8b=#qQjaylj4s}-C4xnl53
zqe+IL^#^TZw=aJJLfjD8Jl!-oRB#m;@rC?%xv(rW2DPJ|uwtl0|EKiHK&Z#^q#HxJ
z{mXFNbyM+Q1%U|xh*J*wrl^hG{p~*JT-}yF3>QE|T>TJ6J^@omUhV=kk2#yzSKcTH
zqpW>Gfu~y<+{)j9Xebp3Yue|^DHHSn6lok=m9OAt{6fS{vx-u`xwoX|&y$Bn8eJ9^
zG1y7~3OL#SB!N^w+K_N~(=fSC<nA<ZY*1VZLkzRVKA!tCppHMR$8D7`>S!WEo!FrC
ziP3m4gLs1m=Xp`}Ndqi??L-?1P!9wi@+lD&GyA7Cimp)3QCa4r6hKS^{v@?5&<c@H
z>L{VSfH5QK!z86;_@`16ciE<(apY+{5un}PLITuyHHzWCN@7&+d@3OAFLn(@gQ?fJ
zN`VAA@+szco`@s=b<CQR#Dp&yQ?l9-L!~xhWv5qE?LRDK(*XQ*zMRvl`nnepUh&3S
z-Bip5UHN{v9N>K^B@J-1)U)(dqfFXEyiLAsU%r3^>?Q&<KO#VBqtwSd3Ud3=v8Rz`
z|8J;XeM@3YOp(F^ZR#u@hEfxN|5~N%j7Vh{99qfJ@f37;)W;cZ+=c>|!pZE~6P)o_
z0;OF0|0$W%G84_7r?Dj2z<4?{UvA`Y$pjThDU(l~Y0_|Qyu-3R&ZYKt|6Eidx%SwI
zscgQTyCj3&tRj=fn?3$O^rlNni@X^!s|e!`V2He%-AmbiaUheHfd336UR^QPX1dot
z+duudZc@)w;}<E`E-O3ql@_ZnjDmDy<AmvJmim6DY0Nwx!IA?wHH2DNsVg(B1bW+T
z44x+{5H%;r%a?KWTdji6bKxdcr^gD8f>;Gw^zS<~XiZMsYa5r243wzvIUgElNx^b?
zBU~Xf0*WvR3xVR%KwiRP3>)UrTiZfYSSC&#hzK2rF8Z`-GLxwe#3ri=iq=Q?;-v~i
znHI8#erM}z*lLhHxXu=)xxt6E;@Cs*Yu!57_=5Gi`ane-bRH(frJ)I%WZDu%oIjlu
zU=qohpX)%vz^ru`H=VPVLL(SMj7a@t@?(?qMHR!()Ap(dR|B-qYF5;%kz0mD3>in#
zXwSs#dso=_JJV7h&R6@)nf>qOH$xnq^9Rcqn7~6#mwbOmF>O!qqu&>Lp<gfOw|yf<
zwmr4*8e<yVEcCmx>A)_?X=$e|u;}u_%xcjz$7cyk%#^AJ5rST3YS)A4kOlZM-JhQt
zj_1yRc21(W!$vl8{dKcM1|?0)$LOn^L>{d4hQCcyzHo1bAICfa3^6Yhf7L*0T#GwL
zQkaE39?@;fhIi7Bn!2M$j2Tyo{(cDhnz>XDVg)f~KGlQ7k=<D`B^p4~xRcSydZf}A
zO5@#UwVCf4K+@Q2?(XYrv~*GrN15dfAX+^A_O_57X<F&L$ssGI%-IGIKkx{%q!?O>
z+P(quwDXVhrg`CgYtPxi#0^X2xRaaFij^p7^R<rq(5IWeyUYqV0(eb#nAC>l<LnfC
zDDYNT(rR52BEC+0b$`F*aW^|wN$0+Ce|A?o($yl>-Mt8$J!>l;{yWi@y0--aLP9QP
zOEK=KpKijQ+Q2za=hyTAeW=&em*D<YlFMgZ;@Ahf>wP!?xtr*`zl`lZ<L$n2?~cse
z=xyHEfr;E)nn7+}9d1a4bV}sa)VMu8=eN7>y`LUWd_qEQy*lor+^;w8&jqor^6~wZ
zEBCj~E>QaS&(Hf{X9P8&q5k)0D0T;@|F$kF?%FHHvitJuly-s7Y2c@ei>>{=vp%bb
zgTYJsjfaaDJOl(7=8rAUQd%d4(SS$>Z&Ge7SM>H>jR?Kgv*`D=bRK$Wi0{+!V}QRX
zJ)IVu0&zsg$92S`P0=02wq=#r{bYp$<x_j*F2}n)@i={*opKM^_QzuSV+lWgIx?O8
zPtKh%>Ny7wfI!X9<pE<PsGDYM(alyP&t;W>u6o~V(#Q?8C;5@!%p#h}Pd`!8Ki7VL
zt9bIN$lt71Ri1d^U}MG4ta*>{G$qHoJ102RwqEyrE&9G~vAzNZD%v+iLzPaA88az7
zcim0!0vazD9>?+r`BcO$C()<RqrI_JxF)<kXn3j;Y#K2}>olM0NVs%X@u;KR`W4@k
z0ap|AJat>pSYD9w$JMuJc@(SOtH|=RBr|lT(R7You1g^7@b9Q-%MuzGr1YJ1m0lD$
z;;{Z7MH|#D4wBKG!P~a(mHR<D1QO1v`Ow50M3b{nT3c>9QW|UI+g{|W^a(n#G%P7G
zhbUt&ID8_9f;gt#6^`}cX4d60=jC;c=6Z&S$=GFXee|@a{XgxLkd1b(Z<t>@To%@3
z1L@bLcl7^|4>7V%jH-?+@iFf-X!;ugsO2G_fX}+5DV{3ZISG^xieRLAo|BvWk<kCC
zzStcDMPZ;l@?KRRG<(fHyAGh~zVaHklVA~e4RN=X7<)-dCLi-X$`y~WS+asZ7c-Sr
zRxiS?J2b5`VDzVcNKj%?ex6FqdY3P!MEKe7znYd>F<_Ke8q62(Ql;mu+%f`pQ5Nuw
zmHODEBc|M!h6sG%acC4Q*SM^W8#_aUh_8ga3Dcg^J&l7J&gMfXdj0@IiED9~1hb{Y
zbhv>c-^*ii8IkGLMcvo*qzk3TmNSFVf2XA9>|=xC=BG^uP->4(w1pfs{<2dQHVTJV
zyz%>$QBlx4KD?Je|Eu99CIvKOzfC<=KHwu&Y|G#iyps?c7Geu?`T6FX0&zipm7(5o
zn5NxEk;h=Ruq{F78Mhjslais+F+tZ~k-9aGt){2etrpKS>J*uP9!LH{TGTBuPZQqY
z;Irey+w$R+a72@N;I|#^uyvHgh;>gY&Ai@QR3^&Iz#9@}3)m9Q6`@(ni1w;#wY*ds
zUeVC#?WW1A2>FBLtJkm8>+btUyO4T(-)RRCMAxisVPW6P$NJ*{Pkd%2cEJP@bNyQX
z5MF~;<z};{)paAb3X4$s$C>$E9-h%7I!lbyJ?SxzRWlyw$Ba2A-J;+mmOZ+{>CLx8
zb#b7e#7qxPy6~reD1w<^0TR!KyiW^bqc=srYOIRYasA&xj1Uw^NJvjlC`dZ$Ammp_
zq)6yUn9nVYNc@?J^PoO#*!Zv0=S3T-pCi+J0i+G$$*f!evD5r-l@1aT(ewEKz3elO
z7C=%UsZ5?lkSs_r(_<0DMEf7tH2@(Yg(3g{!_S8^s~16%ARrU52$BY+XL2urSZV%i
ehR@><P>_%a|M%O^Z|IpWOCV-+j``<ar2hl3A;vra
--
GitLab