Cython: rearranged input and output format options to have both and

updated commands accordingly

python/obitools3/apps/optiongroups/__init__.py

@@ -7,7 +7,7 @@ def __addInputOption(optionManager):
     
 
     group = optionManager.add_argument_group("Restriction to a sub-part options",
-                    "Allow to limit analysis to a sub-part of the data file")
+                    "Allows to limit analysis to a sub-part of the input")
 
     group.add_argument('--skip',
                      action="store", dest="obi:skip",
@@ -22,61 +22,55 @@ def __addInputOption(optionManager):
                      default=None,
                      type=int,
                      help="treat only N sequences")
-
-    group.add_argument('--na-string',
-                     action="store", dest="obi:nastring",
-                     default="NA",
-                     type=str,
-                     help="String associated to Non Available (NA) values")
     
 
-def __addSequenceInputOption(optionManager):
+def __addImportInputOption(optionManager):
-    group = optionManager.add_argument_group("Input format options for sequence files")
+    group = optionManager.add_argument_group("Input format options for imported files")
 
-    group.add_argument('--fasta',
+    group.add_argument('--fasta-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'fasta',
                      help="Input file is in sanger fasta format")
 
-    group.add_argument('--fastq',
+    group.add_argument('--fastq-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'fastq',
                      help="Input file is in fastq format")
 
-    group.add_argument('--embl',
+    group.add_argument('--embl-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'embl',
                      help="Input file is in embl nucleic format")
 
-    group.add_argument('--genbank',
+    group.add_argument('--genbank-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'genbank',
                      help="Input file is in genbank nucleic format")
 
-    group.add_argument('--ngsfilter',
+    group.add_argument('--ngsfilter-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'ngsfilter',
                      help="Input file is an ngsfilter file")
 
-    group.add_argument('--ecopcr-result',
+    group.add_argument('--ecopcr-result-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'ecopcr',
                      help="Input file is the result of an ecoPCR (version 2)")
 
-    group.add_argument('--ecoprimers-result',
+    group.add_argument('--ecoprimers-result-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'ecoprimers',
                      help="Input file is the result of an ecoprimers")
 
-    group.add_argument('--tabular',
+    group.add_argument('--tabular-input',
-                     action="store_const", dest="obi:format",
+                     action="store_const", dest="obi:inputformat",
                      default=None,
                      const=b'tabular',
                      help="Input file is a tabular file")
@@ -114,6 +108,13 @@ def __addSequenceInputOption(optionManager):
                      default=None,
                      const=b'pep',
                      help="Input file contains protein sequences")
+
+    group.add_argument('--input-na-string',
+                     action="store", dest="obi:inputnastring",
+                     default="NA",
+                     type=str,
+                     help="String associated with Non Available (NA) values in the input")
+
   
 def __addTabularInputOption(optionManager):
     group = optionManager.add_argument_group("Input format options for tabular files")
@@ -151,7 +152,17 @@ def __addTabularInputOption(optionManager):
                      type=str,
                      help="Lines starting by this char are considered as comment")
 
-def __addTaxonomyInputOption(optionManager):
+
+def __addTaxdumpInputOption(optionManager):  # TODO maybe not the best way to do it
+    group = optionManager.add_argument_group("Input format options for taxdump")
+
+    group.add_argument('--taxdump',
+                     action="store_true", dest="obi:taxdump",
+                     default=False,
+                     help="Whether the input is a taxdump")
+
+
+def __addTaxonomyOption(optionManager):
     group = optionManager.add_argument_group("Input format options for taxonomy")
 
     group.add_argument('--taxonomy',
@@ -161,46 +172,46 @@ def __addTaxonomyInputOption(optionManager):
     
     #TODO option bool to download taxo if URI doesn't exist
 
-def __addTaxdumpInputOption(optionManager):
-    group = optionManager.add_argument_group("Input format options for taxdump")
-
-    group.add_argument('--taxdump',
-                     action="store_true", dest="obi:taxdump",
-                     default=False,
-                     help="Whether the input is a taxdump")
 
 def addMinimalInputOption(optionManager):
     __addInputOption(optionManager)
+
     
-def addSequenceInputOption(optionManager):
+def addImportInputOption(optionManager):
     __addInputOption(optionManager)
-    __addSequenceInputOption(optionManager)
+    __addImportInputOption(optionManager)
-    
+
+
 def addTabularInputOption(optionManager):
-    #__addInputOption(optionManager)  # TODO discuss conflict
     __addTabularInputOption(optionManager)
 
-def addTaxonomyInputOption(optionManager):
+
-    __addTaxonomyInputOption(optionManager)
+def addTaxonomyOption(optionManager):
+    __addTaxonomyOption(optionManager)
+
 
 def addTaxdumpInputOption(optionManager):
     __addTaxdumpInputOption(optionManager)
 
+
 def addAllInputOption(optionManager):
     __addInputOption(optionManager)
-    __addSequenceInputOption(optionManager)
+    __addImportInputOption(optionManager)
     __addTabularInputOption(optionManager)
-    __addTaxonomyInputOption(optionManager)
+    __addTaxonomyOption(optionManager)
     __addTaxdumpInputOption(optionManager)    
-    
+
+
 def __addOutputOption(optionManager):
     
     optionManager.add_argument(
                     dest='obi:outputURI',  
                     metavar='OUTPUT', 
                     help='Data destination URI')
-    
+
-    group = optionManager.add_argument_group("Management of DMS for output")
+
+def __addDMSOutputOption(optionManager):
+    group = optionManager.add_argument_group("Output options for DMS data")
 
     group.add_argument('--no-create-dms',
                  action="store_true", dest="obi:nocreatedms",
@@ -218,5 +229,44 @@ def __addOutputOption(optionManager):
                       "is greater than N, the values are stored as character strings")
 
 
+def __addExportOutputOption(optionManager):
+    group = optionManager.add_argument_group("Output format options for exported files")
+
+    group.add_argument('--fasta-output',
+                     action="store_const", dest="obi:outputformat",
+                     default=None,
+                     const=b'fasta',
+                     help="Output file is in sanger fasta format")
+
+    group.add_argument('--fastq-output',
+                     action="store_const", dest="obi:outputformat",
+                     default=None,
+                     const=b'fastq',
+                     help="Output file is in fastq format")
+
+    group.add_argument('--print-na',
+                     action="store_true", dest="obi:printna",
+                     default=False,
+                     help="Print Non Available (NA) values in the output")
+
+    group.add_argument('--output-na-string',
+                     action="store", dest="obi:outputnastring",
+                     default="NA",
+                     type=str,
+                     help="String associated with Non Available (NA) values in the output")
+
+
 def addMinimalOutputOption(optionManager):
     __addOutputOption(optionManager)
+    __addDMSOutputOption(optionManager)
+
+def addExportOutputOption(optionManager):
+    __addOutputOption(optionManager)
+    __addExportOutputOption(optionManager)
+
+
+def addAllOutputOption(optionManager):
+    __addOutputOption(optionManager)
+    __addDMSOutputOption(optionManager)
+    __addExportOutputOption(optionManager)
+

python/obitools3/commands/alignpairedend.pyx

@@ -8,7 +8,7 @@ from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 from obitools3.dms.column.column cimport Column, Column_line
 from obitools3.dms.capi.obiview cimport QUALITY_COLUMN, COUNT_COLUMN, NUC_SEQUENCE_COLUMN, ID_COLUMN
 from obitools3.dms.capi.obitypes cimport OBI_INT, OBI_STR, index_t, OBI_QUAL
-from obitools3.apps.optiongroups import addSequenceInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.align._qsassemble import QSolexaReverseAssemble
@@ -29,7 +29,7 @@ __title__="Aligns paired-ended reads"
 
 def addOptions(parser):
 
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
  
     group = parser.add_argument_group('obi alignpairedend specific options')

python/obitools3/commands/annotate.pyx

@@ -4,7 +4,7 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms import DMS
 from obitools3.dms.view.view cimport View, Line_selection
 from obitools3.uri.decode import open_uri
-from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption
 from obitools3.dms.view import RollbackException
 from functools import reduce
 from obitools3.apps.config import logger
@@ -29,7 +29,7 @@ SPECIAL_COLUMNS = [NUC_SEQUENCE_COLUMN, ID_COLUMN, DEFINITION_COLUMN, QUALITY_CO
 def addOptions(parser):
     
     addMinimalInputOption(parser)
-    addTaxonomyInputOption(parser)
+    addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
 
     group=parser.add_argument_group('obi annotate specific options')

python/obitools3/commands/clean.pyx

@@ -5,7 +5,7 @@ from obitools3.dms.dms cimport DMS
 from obitools3.dms.capi.obidms cimport OBIDMS_p
 from obitools3.dms.view import RollbackException
 from obitools3.dms.capi.obiclean cimport obi_clean
-from obitools3.apps.optiongroups import addSequenceInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes
@@ -20,7 +20,7 @@ __title__="Tag a set of sequences for PCR and sequencing errors identification"
 
 def addOptions(parser):
 
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
  
     group = parser.add_argument_group('obi clean specific options')

python/obitools3/commands/count.pyx

@@ -4,7 +4,7 @@
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.dms import DMS
-from obitools3.apps.optiongroups import addSequenceInputOption
+from obitools3.apps.optiongroups import addMinimalInputOption
 from obitools3.dms.capi.obiview cimport COUNT_COLUMN
 
 
@@ -13,7 +13,7 @@ __title__="Counts sequence records"
  
 def addOptions(parser):
  
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
  
     group = parser.add_argument_group('obi count specific options')
  

python/obitools3/commands/ecopcr.pyx

@@ -5,7 +5,7 @@ from obitools3.dms.dms cimport DMS
 from obitools3.dms.capi.obidms cimport OBIDMS_p
 from obitools3.dms.view import RollbackException
 from obitools3.dms.capi.obiecopcr cimport obi_ecopcr
-from obitools3.apps.optiongroups import addSequenceInputOption, addMinimalOutputOption, addTaxonomyInputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addTaxonomyOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes
@@ -24,9 +24,9 @@ __title__="in silico PCR"
 # TODO: add option to output unique ids
 def addOptions(parser):
 
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
+    addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
-    addTaxonomyInputOption(parser)
 
  
     group = parser.add_argument_group('obi ecopcr specific options')

python/obitools3/commands/grep.pyx

@@ -4,7 +4,7 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms import DMS
 from obitools3.dms.view.view cimport View, Line_selection
 from obitools3.uri.decode import open_uri
-from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption
 from obitools3.dms.view import RollbackException
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes
@@ -24,7 +24,7 @@ __title__="Grep view lines that match the given predicates"
 def addOptions(parser):
     
     addMinimalInputOption(parser)
-    addTaxonomyInputOption(parser)
+    addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
 
     group=parser.add_argument_group("obi grep specific options")

python/obitools3/commands/import.pyx

@@ -22,10 +22,10 @@ from obitools3.dms.capi.obitypes cimport obitype_t, \
 
 from obitools3.dms.capi.obierrno cimport obi_errno
 
-from obitools3.apps.optiongroups import addSequenceInputOption, \
+from obitools3.apps.optiongroups import addImportInputOption, \
                                         addTabularInputOption, \
-                                        addMinimalOutputOption, \
+                                        addTaxdumpInputOption, \
-                                        addTaxdumpInputOption
+                                        addMinimalOutputOption
 
 from obitools3.uri.decode import open_uri
 
@@ -45,10 +45,10 @@ default_config = {   'destview'     : None,
 
 def addOptions(parser):
     
-    addSequenceInputOption(parser)
+    addImportInputOption(parser)
     addTabularInputOption(parser)
-    addMinimalOutputOption(parser)
     addTaxdumpInputOption(parser)
+    addMinimalOutputOption(parser)
 
 
 def run(config):
@@ -123,7 +123,7 @@ def run(config):
     
     entries = input[1]
     
-    NA_value = tobytes(config['obi']['nastring'])   # TODO
+    NA_value = tobytes(config['obi']['inputnastring'])
     
     NUC_SEQS_view = False
     if isinstance(output[1], View) :

python/obitools3/commands/less.pyx

@@ -1,6 +1,6 @@
 #cython: language_level=3
 
-from obitools3.apps.optiongroups import addSequenceInputOption
+from obitools3.apps.optiongroups import addMinimalInputOption
 from obitools3.uri.decode import open_uri
 from obitools3.dms import DMS
 
@@ -10,7 +10,7 @@ __title__="Less equivalent"
 
 def addOptions(parser):
 
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
  
     group=parser.add_argument_group('obi less specific options')
  

python/obitools3/commands/ngsfilter.pyx

@@ -5,7 +5,7 @@ from obitools3.dms import DMS
 from obitools3.dms.view import RollbackException
 from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 from obitools3.dms.column.column cimport Column, Column_line
-from obitools3.apps.optiongroups import addSequenceInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.align._freeendgapfm import FreeEndGapFullMatch
@@ -26,7 +26,7 @@ __title__="Assigns sequence records to the corresponding experiment/sample based
 
 def addOptions(parser):
 
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
  
     group = parser.add_argument_group('obi ngsfilter specific options')
@@ -38,13 +38,6 @@ def addOptions(parser):
                      default=None,
                      help="URI to the view containing the samples definition (with tags, primers, sample names,...)")
 
-#     group.add_argument('-F', '--forward-reads',
-#                      action="store", dest="ngsfilter:forward",
-#                      metavar="<URI>",
-#                      default=None,
-#                      type=str,
-#                      help="URI to the forward reads if the paired-end reads haven't been aligned yet")
-
     group.add_argument('-R', '--reverse-reads',
                      action="store", dest="ngsfilter:reverse",
                      metavar="<URI>",

python/obitools3/commands/uniq.pyx

@@ -9,7 +9,7 @@ from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 from obitools3.dms.column.column cimport Column, Column_line
 from obitools3.dms.capi.obiview cimport QUALITY_COLUMN, COUNT_COLUMN, NUC_SEQUENCE_COLUMN, ID_COLUMN
 from obitools3.dms.capi.obitypes cimport OBI_INT, OBI_STR, index_t
-from obitools3.apps.optiongroups import addSequenceInputOption, addMinimalOutputOption, addTaxonomyInputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addTaxonomyOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes
@@ -23,9 +23,9 @@ __title__="Group sequence records together"
  
 def addOptions(parser):
  
-    addSequenceInputOption(parser)
+    addMinimalInputOption(parser)
+    addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
-    addTaxonomyInputOption(parser)
  
     group = parser.add_argument_group('obi uniq specific options')