Cython: added possibility to output in tabular format

python/obitools3/apps/optiongroups/__init__.py

@@ -115,9 +115,9 @@ def __addImportInputOption(optionManager):
                      type=str,
                      help="String associated with Non Available (NA) values in the input")
 
-  
+
-def __addTabularInputOption(optionManager):
+def __addTabularOption(optionManager):
-    group = optionManager.add_argument_group("Input format options for tabular files")
+    group = optionManager.add_argument_group("Input and output format options for tabular files")
 
     group.add_argument('--header',
                      action="store_true", dest="obi:header",
@@ -126,9 +126,15 @@ def __addTabularInputOption(optionManager):
 
     group.add_argument('--sep',
                      action="store", dest="obi:sep",
-                     default=None,
+                     default="\t",
                      type=str,
                      help="Column separator")
+
+
+def __addTabularInputOption(optionManager):
+    group = optionManager.add_argument_group("Input format options for tabular files")
+    
+    __addTabularOption(optionManager)
     
     group.add_argument('--dec',
                      action="store", dest="obi:dec",
@@ -244,6 +250,12 @@ def __addExportOutputOption(optionManager):
                      const=b'fastq',
                      help="Output file is in fastq format")
 
+    group.add_argument('--tab-output',
+                     action="store_const", dest="obi:outputformat",
+                     default=None,
+                     const=b'tabular',
+                     help="Output file is in tabular format")
+
     group.add_argument('--print-na',
                      action="store_true", dest="obi:printna",
                      default=False,
@@ -260,13 +272,20 @@ def addMinimalOutputOption(optionManager):
     __addOutputOption(optionManager)
     __addDMSOutputOption(optionManager)
 
+
+def addTabularOutputOption(optionManager):
+    __addTabularOption(optionManager)
+
+
 def addExportOutputOption(optionManager):
     __addOutputOption(optionManager)
     __addExportOutputOption(optionManager)
+    __addTabularOption(optionManager)
 
 
 def addAllOutputOption(optionManager):
     __addOutputOption(optionManager)
     __addDMSOutputOption(optionManager)
     __addExportOutputOption(optionManager)
+    __addTabularOption(optionManager)
 

python/obitools3/uri/decode.pyx

@@ -15,14 +15,18 @@ from obitools3.parsers.universal import entryIteratorFactory
 
 from obitools3.writers.fasta import FastaNucWriter
 from obitools3.writers.fastq import FastqWriter
+from obitools3.writers.tab import TabWriter
 from obitools3.format.fasta import FastaFormat
 from obitools3.format.fastq import FastqFormat
+from obitools3.format.tab import TabFormat
 
 from obitools3.dms.obiseq import Nuc_Seq
 from obitools3.apps.config import getConfiguration,logger
 from obitools3.apps.temp import get_temp_dms
 from obitools3.utils cimport tobytes, count_entries  # TODO tobytes because can't read options as bytes
 
+from obitools3.files.universalopener cimport uopen
+
 from obitools3.dms.capi.obierrno cimport obi_errno, \
                                          OBIVIEW_ALREADY_EXISTS_ERROR
 
@@ -283,17 +287,18 @@ def open_uri(uri,
                     format=config["obi"][formatkey]
                 except KeyError:
                     format=None
-        
+                
         if b'seqtype' in qualifiers:
             seqtype=qualifiers[b'seqtype'][0]
         else:
-            if format == b"ngsfilter":  # TODO discuss
+            if format == b"ngsfilter" or format == b"tabular":  # TODO discuss
                 seqtype=None
             else:
                 try:
                     seqtype=config["obi"]["seqtype"]
                 except KeyError:
                     seqtype=b"nuc"
+        config["obi"]["seqtype"] = seqtype
         
         if b'skip' in qualifiers:
             skip=int(qualifiers[b"skip"][0])
@@ -504,7 +509,11 @@ def open_uri(uri,
                                        skip = skip,
                                        only = only)
                 else:
-                    raise NotImplementedError('Output sequence file format not implemented')
+                    iseq = TabWriter(TabFormat(header=header, NAString=nastring, sep=sep), 
+                                               file,
+                                               skip=skip,
+                                               only=only,
+                                               header=header)
             elif format==b"ngsfilter":
                 objclass = dict
                 if input: