fix lot of small bugs

use argparse instead of optparse
raise error when system-id is not specify

src/parser/parser.py

@@ -26,9 +26,11 @@ def replicon_parser(replicon_data):
     :rtype: dict
     """
     replicon_db = {}
-    Replicon_info = namedtuple('Replicon_info', ('name', 'taxid', 'strain', 'taxonomy', 'type'))
+    Replicon_info = namedtuple('Replicon_info', ('name', 'ncbi_id', 'taxid', 'strain', 'taxonomy', 'type'))
     with open(replicon_data, 'r') as replicon_file:
+        line_nb = 0
         for line in replicon_file:
+            line_nb += 1
             if not line.startswith('#'):
                 line = line.strip()
                 fields = line.split('\t')
@@ -43,12 +45,13 @@ def replicon_parser(replicon_data):
                         taxonomy = fields[4].split('; ')
                         # remove ending dot or semi-colon from the last term of taxonnomy
                         if taxonomy[-1].endswith('.') or taxonomy[-1].endswith(';'):
-                            taxonomy[-1] = taxonomy[-1][:-1]
+                            taxonomy = taxonomy[-1][:-1]
                         replicon_type = fields[5]
+
                         replicon_db[replicon_id] = Replicon_info(replicon_id, ncbi_id, taxid, strain,
                                                                  taxonomy, replicon_type)
                     except Exception as err:
-                        raise Exception("Error during parsing line : {0} : {1}".format(line, err))
+                        raise Exception("Error during parsing line {0}: {1} : {2}".format(line_nb, line, err))
     return replicon_db
 
 def system_parser(system_data):
@@ -66,14 +69,17 @@ def system_parser(system_data):
                        'score', 'i_evalue', 'coverage', 'match_begin', 'match_end', 'name', 'description')
                       )
   
-    with open(system_data, 'r') as system_file :
+    with open(system_data, 'r') as system_file:
+        line_nb = 0
         for line in system_file:
+            line_nb += 1
             if line[0] != '#':
                 line = line.strip()
                 fields = line.split('\t')
                 gene_code = fields[0]
                 if gene_code in system_db:
                     raise KeyError("duplicate replicon:" + fields[0])
+                try:
                     gene_id = fields[1]
                     protein_length = int(fields[2])
                     strand = fields[3] if fields[3] != '-' else None
@@ -88,10 +94,14 @@ def system_parser(system_data):
                     match_end = int(fields[11]) if fields[11] != '-' else None
                     replicon_id = fields[12]
                     predicted_system = fields[13] if fields[13] != '-' else None
-                system_id = fields[14] if fields[14] != '-' else None
+                    system_id = fields[14]
+                    if system_id == '-':
+                        raise RuntimeError("System-Id is empty")
                     system_status = fields[15] if fields[15] != '-' else None
                     gene_name = fields[16] if fields[16] else None
                     description = fields[17] if fields[17] else None
+                except Exception as err:
+                    raise RuntimeError("Error during parsing line {0}: {1} : {2}".format(line_nb, line, err))
                 gene = Gene(gene_code,
                             gene_id,
                             protein_length,
@@ -182,7 +192,6 @@ def fill_db(server_uri, db_name, user, passwd, replicon_db, system_db, force_upd
         secretion_system.genes = genes
         secreton_db.save_doc(secretion_system, force_update=force_update)
 
-          
 if __name__ == '__main__':
     import argparse
     import sys
@@ -194,7 +203,7 @@ if __name__ == '__main__':
         return user, password
        
     usage = """
-       %prog [options] 
+       %(prog)s [options]
        parse a file containing replicon informations and a file containing system informations
        and fill a couchDB data base with these informations 
 """
@@ -202,23 +211,19 @@ if __name__ == '__main__':
     server_opt = parser.add_argument_group(title="Server Options")
     server_opt.add_argument("-S", "--server",
                             action="store",
-                            type="string",
                             dest="server_url",
                             help="the url of the couchDB server (with the port)")
     server_opt.add_argument("-d", "--database",
                             action="store",
-                            type="string",
                             dest="db_name",
                             help="the name of the data base")
     parsing_opt = parser.add_argument_group(title="Parsing Options")
     parsing_opt.add_argument("-r", "--replicon",
                              action="store",
-                             type="string",
                              dest="replicon_path",
                              help="the path to the replicon file to parse")
     parsing_opt.add_argument("-s", "--system",
                              action="store",
-                             type="string",
                              dest="system_path",
                              help="the path to the system secretion file to parse")
     parsing_opt.add_argument("-f", "--force_update",
@@ -226,43 +231,41 @@ if __name__ == '__main__':
                              dest="force_update",
                              default=False,
                              help="insert document even if there is already a document with the same id (replace it)")
-    options, args = parser.parse_args()
+    args = parser.parse_args()
     
-    if not options.server_url:
+    if not args.server_url:
         print("You must specify a server url", file=sys.stderr)
         parser.print_help(sys.stderr)
         sys.exit(1)
         
-    if not options.db_name:
+    if not args.db_name:
         print("You must specify a data base name", file=sys.stderr)
         parser.print_help(sys.stderr)
         sys.exit(1)
         
-    if not options.replicon_path:
+    if not args.replicon_path:
         print("You must specify the path to the replicon information file", file=sys.stderr)
         parser.print_help(sys.stderr)
         sys.exit(1)
         
-    if not options.system_path:
+    if not args.system_path:
         print("You must specify the path to the secretion system information file", file=sys.stderr)
         parser.print_help(sys.stderr)
         sys.exit(1)
     
-    replicon_db = replicon_parser(options.replicon_path)
-    system_db = system_parser(options.system_path)
+    replicon_db = replicon_parser(args.replicon_path)
+    system_db = system_parser(args.system_path)
    
     try_again = 0
     while True:
         user, password = get_credentials()
         try:
-            fill_db(options.server_url, options.db_name, user, password,
-                    replicon_db, system_db, force_update=options.force_update)
+            fill_db(args.server_url, args.db_name, user, password,
+                    replicon_db, system_db, force_update=args.force_update)
             break
         except restkit.errors.Unauthorized as err:
-            print("Bad authentication, try again", file=sys.stderr)
             try_again += 1
             if try_again > 2:
                 sys.exit("Authentication failure")
-        except Exception, err:
-            print(err, file=sys.stderr)
+        except Exception as err:
             sys.exit(2)