Add --length option to obiannotate to replace the old fastalength script

src/obitools/options/bioseqedittag.py

@@ -49,6 +49,25 @@ def addSequenceEditTagOptions(optionManager):
                              default=False,
                              help="only valid python expressions are allowed")
     
+    optionManager.add_option('-C','--clear',
+                             action="store_true", 
+                             dest='clear',
+                             default=False,
+                             help="clear all tags associated to the sequences")
+    
+    optionManager.add_option('-k','--keep',
+                             action='append',
+                             dest='keep',
+                             default=[],
+                             type="string",
+                             help="only keep this tag")
+    
+    optionManager.add_option('--length',
+                             action="store_true", 
+                             dest='length',
+                             default=False,
+                             help="add seqLength tag with sequence length")
+    
     optionManager.add_option('--with-taxon-at-rank',
                              action='append',
                              dest='taxonrank',
@@ -65,19 +84,28 @@ def sequenceTaggerGenerator(options):
     toSet    = [x.split(':',1) for x in options.setTags if len(x.split(':',1))==2]
     newId    = options.setIdentifier
     newDef   = options.setDefinition
+    clear    = options.clear
+    keep     = set(options.keep)
+    length   = options.length
     if options.taxonomy is not None:
         annoteRank=options.taxonrank
     else:
         annoteRank=[]
     
     def sequenceTagger(seq):
-        for i in toDelete:
-            if i in seq:
-                del seq[i]
-        for o,n in toRename:
-            if o in seq:
-                seq[n]=seq[o]
-                del seq[o]
+        if clear or keep:
+            ks = seq.keys()
+            for k in ks:
+                if k not in keep:
+                    del seq[k]
+        else:
+            for i in toDelete:
+                if i in seq:
+                    del seq[i]                
+            for o,n in toRename:
+                if o in seq:
+                    seq[n]=seq[o]
+                    del seq[o]
         for i,v in toSet:
             try:
                 val = eval(v,{'sequence':seq},seq)
@@ -86,6 +114,10 @@ def sequenceTaggerGenerator(options):
                     raise e
                 val = v
             seq[i]=val
+            
+        if length:
+            seq['seqLength']=len(seq)
+            
         if newId is not None:
             try:
                 val = eval(newId,{'sequence':seq},seq)