Add automanagement of input and output format

src/obitools/format/options.py

@@ -14,32 +14,33 @@ from obitools.fnaqual.quality import qualityIterator
 from obitools.fasta import formatFasta
 from obitools.fastq import formatFastq
 
-
+import sys
 from array import array
+from itertools import chain
 
 
 
 def addInputFormatOption(optionManager):
     optionManager.add_option('--genbank',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='genbank',
                              help="input file is in genbank format")
     optionManager.add_option('--embl',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='embl',
                              help="input file is in embl format")
 
     optionManager.add_option('--fasta',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='fasta',
                              help="input file is in fasta nucleic format (including obitools fasta extentions)")
 
     optionManager.add_option('--fna',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='fna',
                              help="input file is in fasta nucleic format produced by 454 sequencer pipeline")
 
@@ -51,19 +52,19 @@ def addInputFormatOption(optionManager):
 
     optionManager.add_option('--sanger',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='sanger',
                              help="input file is in sanger fastq nucleic format (standard fastq)")
 
     optionManager.add_option('--solexa',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='solexa',
                              help="input file is in fastq nucleic format produced by solexa sequencer")
 
     optionManager.add_option('--illumina',
                              action="store_const", dest="seqinformat",
-                             default='fasta',
+                             default=None,
                              const='illumina',
                              help="input file is in fastq nucleic format produced by old solexa sequencer")
 
@@ -79,20 +80,94 @@ def addInputFormatOption(optionManager):
                              help="input file is protein sequences")
         
 
-def addOutputOption(optionManager):
+def addOutputFormatOption(optionManager):
     optionManager.add_option('--fastq-output',
                              action="store_const", dest="output",
-                             default=formatFasta,
+                             default=None,
                              const=formatFastq,
                              help="output sequences in sanger fastq format")
     optionManager.add_option('--fasta-output',
                              action="store_const", dest="output",
-                             default=formatFasta,
+                             default=None,
                              const=formatFasta,
                              help="output sequences in obitools fasta format")
 
 
-def entriesIterator(options):
+
+def addInOutputOption(optionManager):
+    addInputFormatOption(optionManager)
+    addOutputFormatOption(optionManager)
+
+
+def printOutput(options,seq,output=sys.stdout):
+    if options.output is not None:
+        r=options.output(seq)
+    elif options.outputFormater is not None:
+        r=options.outputFormater(seq)
+    else:
+        r=formatFasta(seq)
+
+    try:
+        print >>output,r
+    except IOError:
+        sys.exit(0)
+
+
+
+def autoEntriesIterator(options):
+    options.outputFormater=formatFasta
+    
+    def annotatedIterator(formatIterator):
+        options.outputFormater=formatFasta
+        def iterator(lineiterator):
+            for s in formatIterator(lineiterator):
+                s.extractTaxon()
+                yield s
+
+        return iterator
+
+    def withQualIterator(qualityfile):
+        options.outputFormater=formatFastq
+        def iterator(lineiterator):
+            for s in fnaFastaIterator(lineiterator):
+                q = qualityfile.next()
+                quality = array('d',(10.**(-x/10.) for x in q))
+                s.quality=quality
+                yield s
+                
+        return iterator
+
+    def autoSequenceIterator(lineiterator):
+        options.outputFormater=formatFasta
+        first = lineiterator.next()
+        if first[0]==">":
+            if options.withqualfile is not None:
+                qualfile=qualityIterator(options.withqualfile)
+                reader=withQualIterator(qualfile)
+                options.outputFormater=formatFastq
+            elif options.moltype=='nuc':
+                reader=fastaNucIterator
+            elif options.moltype=='pep':
+                reader=fastaAAIterator
+            else:
+                reader=fastaIterator
+        elif first[0]=='@':
+            reader=fastqSangerIterator
+            options.outputFormater=formatFastq
+        elif first[0:3]=='ID ':
+            reader=emblIterator
+        elif first[0:6]=='LOCUS ':
+            reader=genbankIterator
+        else:
+            raise AssertionError,'file is not in fasta, fasta, embl, or genbank format'
+        
+        input = reader(chain([first],lineiterator))
+        
+        return input
+                
+    if options.seqinformat is None:
+        reader = autoSequenceIterator
+    else:
         if options.seqinformat=='fasta':
             if options.moltype=='nuc':
                 reader=fastaNucIterator
@@ -101,33 +176,27 @@ def entriesIterator(options):
             else:
                 reader=fastaIterator
         elif options.seqinformat=='genbank':
-        reader=genbankIterator
+            reader=annotatedIterator(genbankIterator)
         elif options.seqinformat=='embl':
-        reader=emblIterator
+            reader=annotatedIterator(emblIterator)
         elif options.seqinformat=='fna':
             reader=fnaFastaIterator
         elif options.seqinformat=='sanger':
+            options.outputFormater=formatFastq
             reader=fastqSangerIterator
         elif options.seqinformat=='solexa':
+            options.outputFormater=formatFastq
             reader=fastqSolexaIterator
         elif options.seqinformat=='illumina':
+            options.outputFormater=formatFastq
             reader=fastqIlluminaIterator
             
         if options.seqinformat=='fna' and options.withqualfile is not None:
             qualfile=qualityIterator(options.withqualfile)
-    else:
-        qualfile=None
+            reader=withQualIterator(qualfile)
+            options.outputFormater=formatFastq
         
-    def iterator(lineiterator):
-        for s in reader(lineiterator):
-            if qualfile is not None:
-                q = qualfile.next()
-                quality = array('d',(10.**(-x/10.) for x in q))
-                s.quality=quality
-            s.extractTaxon()
-            yield s
-            
-    return iterator
+    return reader