diff --git a/src/main_utils.cpp b/src/main_utils.cpp
index 0263fd7f8c950ea32457ecb9ec23e9fa7af73238..14017c7a29743ffdc3b8919f0784307e997c958d 100644
--- a/src/main_utils.cpp
+++ b/src/main_utils.cpp
@@ -4,7 +4,6 @@
* Created on: May 9, 2016
* Author: vlegrand
*/
-#include <stdint.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/sysctl.h>
@@ -52,7 +51,7 @@ void checkDirExists(const string o_file) {
std::size_t o_found = o_file.find_last_of("/");
if (o_found!=string::npos) {
string parent_dir=o_file.substr(0,o_found);
- cout<<parent_dir<<endl;
+ // cout<<parent_dir<<endl;
struct stat info;
if (stat(parent_dir.c_str(),&info)!=0) {
cout<<"parent directory for output files: "<<parent_dir<<" doesn't exist."<<endl;
@@ -62,78 +61,139 @@ void checkDirExists(const string o_file) {
}
-/*
- * Reads the names of input and output fastq files that ROCK must process from text files whose names are passed as
- * argument of the -i/-o options. Fills the appropriate structures.
- * orresponding output files are supposed to be in the same order as input files. No checking here. It is ut to the user to give something correct in input.
- */
-int processInOutFileArgs(const std::string& input_file,const std::string output_file,std::vector<IO_fq_files>& single_files,vector<PE_files>& v_PE_files,int& f_id) {
- ifstream infiles_names(input_file.c_str());
- if (!infiles_names) cout<<"couldn't open file: "<<input_file<<endl;
- ifstream ofiles_names(output_file.c_str());
- if (!ofiles_names) cout<<"couldn't open file: "<<output_file<<endl;
- /*cout<<"processing in/out files: "<<input_file<<" "<<output_file<<endl;
- cout<<f_id<<endl;*/
- while (infiles_names && ofiles_names && f_id<=k_max_input_files) {
- // cout<<"entering while loop"<<endl;
- string iline;
- string oline;
- if (!getline(infiles_names,iline)) break;
- if (!getline(ofiles_names,oline)) break;
- /*cout<< iline<<endl;
- cout<<oline<<endl;*/
- std::size_t i_found = iline.find_first_of(k_sep_pair_end);
- if (i_found!=std::string::npos) {
- // this is PE
- f_id+=2;
- string i_name_PE1=iline.substr(0,i_found);
- string i_name_PE2=iline.substr(i_found+1);
- std::size_t o_found = oline.find_first_of(k_sep_pair_end);
- if (o_found==std::string::npos) {
- cout<< "Inconsistency between input and output files lists!"<<endl;
- return EXIT_FAILURE;
- }
- string o_name_PE1=oline.substr(0,o_found);
- checkDirExists(o_name_PE1);
- string o_name_PE2=oline.substr(o_found+1);
- //cout<<o_name_PE2<<endl;
- checkDirExists(o_name_PE2);
- PE_files pe;
- pe.PE1.in_fq_file=i_name_PE1;
- pe.PE1.out_fq_file=o_name_PE1;
- pe.PE2.in_fq_file=i_name_PE2;
- pe.PE2.out_fq_file=o_name_PE2;
- v_PE_files.push_back(pe);
+void removePathfromFName(string& FName) {
+ std::size_t i_found2 =FName.find_last_of(path_sep); // remove path from filename.
+ if (i_found2!=std::string::npos) {
+ FName=FName.substr(i_found2+1);
+ }
+}
+
+void changeExtension(string& FName) { // changes .fq into .rock.fq or adds .rock.fq
+ std::size_t o_found = FName.find_last_of(k_ext);
+ if (o_found!=std::string::npos) FName.replace(o_found,1,".rock.");
+ else FName.append(".rock.fq");
+}
+
+void genOutFilenames(const std::vector<string>& v_input_lines,std::vector<string>& v_output_lines) {
+ std::vector<string>::const_iterator it_in;
+ string o_line;
+ for (it_in=v_input_lines.begin();it_in!=v_input_lines.end();++it_in) {
+ std::size_t i_found = (*it_in).find_first_of(k_sep_pair_end);
+ if (i_found!=std::string::npos) {// PE files are separared by a ','
+ string i_name_PE1=(*it_in).substr(0,i_found);
+ removePathfromFName(i_name_PE1);
+ string i_name_PE2=(*it_in).substr(i_found+1);
+ removePathfromFName(i_name_PE2);
+ changeExtension(i_name_PE1);
+ changeExtension(i_name_PE2);
+ o_line=i_name_PE1;
+ o_line+=k_sep_pair_end;
+ o_line.append(i_name_PE2);
} else {
- // this is single.
- f_id+=1;
- IO_fq_files p;
- p.in_fq_file=iline;
- checkDirExists(oline);
- p.out_fq_file=oline;
- single_files.push_back(p);
+ o_line=*it_in;
+ removePathfromFName(o_line);
+ changeExtension(o_line);
}
+ v_output_lines.push_back(o_line);
}
- if (!infiles_names.eof())
- {
- std::cout<<"error while reading input or output file"<<std::endl;
- return EXIT_FAILURE;
+}
+
+
+int processInOutFileArgs(const std::vector<string>& v_input_lines,std::vector<string>& v_output_lines,std::vector<IO_fq_files>& single_files,vector<PE_files>& v_PE_files,int& f_id) {
+ if (v_output_lines.empty()) {
+ // in that case, generate output filenames from input filenames
+ genOutFilenames(v_input_lines,v_output_lines);
}
- /*if (!ofiles_names.eof())
- {
- std::cout<<"error while reading input or output file"<<std::endl;
+ if (v_input_lines.size()!=v_output_lines.size()) {
+ cout<< "Inconsistency between input and output files lists!"<<endl;
return EXIT_FAILURE;
- }*/
+ } else {
+ std::vector<string>::const_iterator it_in;
+ std::vector<string>::const_iterator it_out;
+ it_in=v_input_lines.begin();
+ it_out=v_output_lines.begin();
+ while (it_in!=v_input_lines.end()) {
+ std::size_t i_found = (*it_in).find_first_of(k_sep_pair_end);
+ if (i_found!=std::string::npos) {
+ // this is PE
+ f_id+=2;
+ string i_name_PE1=(*it_in).substr(0,i_found);
+ string i_name_PE2=(*it_in).substr(i_found+1);
+ std::size_t o_found = (*it_out).find_first_of(k_sep_pair_end);
+ if (o_found==std::string::npos) {
+ cout<< "Inconsistency between input and output files lists!"<<endl;
+ return EXIT_FAILURE;
+ }
+ string o_name_PE1=(*it_out).substr(0,o_found);
+ checkDirExists(o_name_PE1);
+ string o_name_PE2=(*it_out).substr(o_found+1);
+ //cout<<o_name_PE2<<endl;
+ checkDirExists(o_name_PE2);
+ PE_files pe;
+ pe.PE1.in_fq_file=i_name_PE1;
+ pe.PE1.out_fq_file=o_name_PE1;
+ pe.PE2.in_fq_file=i_name_PE2;
+ pe.PE2.out_fq_file=o_name_PE2;
+ v_PE_files.push_back(pe);
- if (f_id>k_max_input_files) {
- cout<<"ROCK cannot handle more than "<<k_max_input_files<<" input files."<<endl;
- return EXIT_FAILURE;
+ } else {
+ // this is single.
+ f_id+=1;
+ IO_fq_files p;
+ p.in_fq_file=*it_in;
+ checkDirExists(*it_out);
+ p.out_fq_file=*it_out;
+ single_files.push_back(p);
+ }
+ ++it_in;
+ ++it_out;
+ }
+ if (f_id>k_max_input_files) {
+ cout<<"ROCK cannot handle more than "<<k_max_input_files<<" input files."<<endl;
+ return EXIT_FAILURE;
+ }
}
return EXIT_SUCCESS;
}
+/*
+ * Loads the content of a text file (containing input fastq file names to be filtered or names of files that ROCK must generate).
+ */
+int loadFileArgs(const std::string& afile,std::vector<string>& v_lines) {
+ ifstream infiles_names(afile.c_str());
+ if (!infiles_names) cout<<"couldn't open file: "<<afile<<endl;
+ while (infiles_names) {
+ string iline;
+ if (!getline(infiles_names,iline)) break;
+ v_lines.push_back(iline);
+ }
+ if (!infiles_names.eof())
+ {
+ std::cout<<"error while reading input or output file"<<std::endl;
+ return EXIT_FAILURE;
+ }
+ return EXIT_SUCCESS;
+}
+
+/*
+ * Reads the names of input and output fastq files that ROCK must process from text files whose names are passed as
+ * argument of the -i/-o options. Fills the appropriate structures.
+ * orresponding output files are supposed to be in the same order as input files. No checking here. It is ut to the user to give something correct in input.
+ */
+int loadInOutFileArgs(const std::string& input_file,const std::string& output_file,std::vector<string>& v_input_lines,std::vector<string>& v_output_lines) {
+ int reti=EXIT_SUCCESS;
+ int reto=EXIT_SUCCESS;
+ reti=loadFileArgs(input_file,v_input_lines);
+ if (!output_file.empty()) {
+ reto=loadFileArgs(output_file,v_output_lines);
+ }
+ if (reti==EXIT_SUCCESS && reto==EXIT_SUCCESS) return EXIT_SUCCESS;
+ else return EXIT_FAILURE;
+}
+
+
/*
* Minimize lambda for a given number of k-mers.
* We want p<=0.01,
diff --git a/src/main_utils.h b/src/main_utils.h
index b6ce6e5c0168debfe404e961bb51c7175c6b1ec3..7f3a8c296f318bea70a124a56bd6dfc67204c41b 100644
--- a/src/main_utils.h
+++ b/src/main_utils.h
@@ -10,12 +10,15 @@
#define k_max_input_files 15
#define k_sep_pair_end ','
+#define k_ext '.'
+#define path_sep '/'
#include <vector>
#include "rock_commons.h"
unsigned long getNodePhysMemory();
-int processInOutFileArgs(const std::string& input_file,const std::string output_file,std::vector<IO_fq_files>& single_files,std::vector<PE_files>& v_PE_files,int& f_id);
+int loadInOutFileArgs(const std::string& input_file,const std::string& output_file,std::vector<std::string>& v_input_lines,std::vector<std::string>& v_output_lines);
+int processInOutFileArgs(const std::vector<std::string>& v_input_lines,std::vector<std::string>& v_output_lines,std::vector<IO_fq_files>& single_files,std::vector<PE_files>& v_PE_files,int& f_id);
int getBestLambdaForN(const unsigned long& nb_k_mers,int lambda_max);
float getCollisionProba(const unsigned long& nb_k_mers,const int& lambda);
diff --git a/src/unit_test_main_utils.cpp b/src/unit_test_main_utils.cpp
index 57781f05e4c1c627d234c54b51bdf7dd5bb2c957..0e9947dc24af06088496f7007cca6b4a17ce8382 100644
--- a/src/unit_test_main_utils.cpp
+++ b/src/unit_test_main_utils.cpp
@@ -19,18 +19,28 @@ using namespace std;
void test_processIOFileArgs() {
std::vector<IO_fq_files> single_files;
vector<PE_files> v_PE_files;
+ std::vector<string> v_input_lines;
+ std::vector<string> v_output_lines;
int f_id=0;
- int ret=processInOutFileArgs("../test/data/unit/list_input1.txt","../test/data/unit/list_output1.txt",single_files,v_PE_files,f_id);
+ int ret=loadInOutFileArgs("../test/data/unit/list_input1.txt","../test/data/unit/list_output1.txt",v_input_lines,v_output_lines);
+ assert(ret==EXIT_SUCCESS);
+ ret=processInOutFileArgs(v_input_lines,v_output_lines,single_files,v_PE_files,f_id);
assert(ret==EXIT_FAILURE);
f_id=0;
v_PE_files.clear();
single_files.clear();
- ret=processInOutFileArgs("../test/data/unit/list_input2.txt","../test/data/unit/list_output2.txt",single_files,v_PE_files,f_id);
+ v_input_lines.clear();
+ v_output_lines.clear();
+ ret=loadInOutFileArgs("../test/data/unit/list_input2.txt","../test/data/unit/list_output2.txt",v_input_lines,v_output_lines);
+ ret=processInOutFileArgs(v_input_lines,v_output_lines,single_files,v_PE_files,f_id);
assert(ret==EXIT_FAILURE);
f_id=0;
v_PE_files.clear();
single_files.clear();
- ret=processInOutFileArgs("../test/data/unit/list_input3.txt","../test/data/unit/list_output3.txt",single_files,v_PE_files,f_id);
+ v_input_lines.clear();
+ v_output_lines.clear();
+ ret=loadInOutFileArgs("../test/data/unit/list_input3.txt","../test/data/unit/list_output3.txt",v_input_lines,v_output_lines);
+ ret=processInOutFileArgs(v_input_lines,v_output_lines,single_files,v_PE_files,f_id);
assert(ret==EXIT_SUCCESS);
assert(f_id==14);
assert(single_files.size()==2);
@@ -41,6 +51,18 @@ void test_processIOFileArgs() {
PE_files s2=v_PE_files[5];
assert(s2.PE2.in_fq_file.compare("nono")==0);
assert(s2.PE2.out_fq_file.compare("onono")==0);
+
+ v_output_lines.clear();
+ v_input_lines.clear();
+ v_PE_files.clear();
+ single_files.clear();
+ ret=loadInOutFileArgs("../test/data/unit/list_input3.txt","",v_input_lines,v_output_lines);
+ ret=processInOutFileArgs(v_input_lines,v_output_lines,single_files,v_PE_files,f_id);
+ s=single_files[0];
+ assert(s.out_fq_file.compare("fifi.rock.fq")==0);
+ s2=v_PE_files[5];
+ assert(s2.PE2.out_fq_file.compare("nono.rock.fq")==0);
+
}
@@ -50,7 +72,7 @@ void test_getBestLambdaForN() {
int best=getBestLambdaForN(nb_k_mer,lambda_max);
assert(best==2);
nb_k_mer=600000000;
- best=getBestLambdaForN(nb_k_mer,lambda_max);
+ best=getBestLambdaForN(nb_k_mer,lambda_max);
assert(best==4);
nb_k_mer=2000000000;
best=getBestLambdaForN(nb_k_mer,lambda_max);