annotate: fixed a bug where a column type could be wrongly guessed and

switch to version 3.0.1b16

python/obitools3/commands/annotate.pyx

@@ -16,6 +16,8 @@ from obitools3.dms.capi.obiview cimport NUC_SEQUENCE_COLUMN, \
                                         QUALITY_COLUMN, \
                                         COUNT_COLUMN, \
                                         TAXID_COLUMN
+from obitools3.dms.capi.obitypes cimport OBI_STR
+from obitools3.dms.column.column cimport Column
 
 import time
 import math 
@@ -187,6 +189,8 @@ def sequenceTaggerGenerator(config, taxo=None):
                     else:
                         scn=None
                     seq[rank]=rtaxid
+                    if "%s_name"%rank not in seq.view:
+                        Column.new_column(seq.view, "%s_name"%rank, OBI_STR)
                     seq["%s_name"%rank]=scn
              
         if add_rank:

python/obitools3/commands/split.pyx

@@ -0,0 +1,105 @@
+#cython: language_level=3
+
+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, addTaxonomyOption, addMinimalOutputOption, addNoProgressBarOption
+from obitools3.dms.view import RollbackException
+from obitools3.apps.config import logger
+from obitools3.utils cimport tobytes
+
+import sys
+from cpython.exc cimport PyErr_CheckSignals
+
+ 
+__title__="Split"
+
+ 
+def addOptions(parser):
+    
+    addMinimalInputOption(parser)
+    addNoProgressBarOption(parser)
+
+    group=parser.add_argument_group("obi split specific options")
+ 
+    group.add_argument('-p','--prefix',
+                       action="store", dest="split:prefix",
+                       metavar="<PREFIX>",
+                       help="Prefix added to each subview name (included undefined)")
+
+    group.add_argument('-t','--tag-name',
+                       action="store", dest="split:tagname",
+                       metavar="<TAG_NAME>",
+                       help="Attribute/tag used to split the input")
+      
+    group.add_argument('-u','--undefined',
+                       action="store", dest="split:undefined",
+                       default=b'UNDEFINED',
+                       metavar="<VIEW_NAME>",
+                       help="Name of the view where undefined sequenced are stored (will be PREFIX_VIEW_NAME)")
+
+
+def run(config):
+     
+    DMS.obi_atexit()
+    
+    logger("info", "obi split")
+
+    # Open the input
+    input = open_uri(config["obi"]["inputURI"])
+    if input is None:
+        raise Exception("Could not read input view")
+    i_dms = input[0]
+    i_view = input[1]
+
+    # Initialize the progress bar
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(i_view), config)
+    else:
+        pb = None
+    
+    tag_to_split = config["split"]["tagname"]
+    undefined = tobytes(config["split"]["undefined"])
+    selections = {}
+    
+    # Go through input view and split
+    for i in range(len(i_view)):
+        PyErr_CheckSignals()
+        if pb is not None:
+            pb(i)
+        line = i_view[i]
+        if tag_to_split not in line or line[tag_to_split] is None or len(line[tag_to_split])==0:
+            value = undefined
+        else:
+            value = line[tag_to_split]
+        if value not in selections:
+            selections[value] = Line_selection(i_view)
+        selections[value].append(i)
+        
+    if pb is not None:
+        pb(len(i_view), force=True)
+        print("", file=sys.stderr)
+
+    # Create output views with the line selection
+    try:
+        for cat in selections:
+            o_view_name = config["split"]["prefix"].encode()+cat
+            o_view = selections[cat].materialize(o_view_name)
+            # Save command config in View and DMS comments
+            command_line = " ".join(sys.argv[1:])
+            input_dms_name=[input[0].name]
+            input_view_name=[input[1].name]
+            if 'taxoURI' in config['obi'] and config['obi']['taxoURI'] is not None:
+                input_dms_name.append(config['obi']['taxoURI'].split("/")[-3])
+                input_view_name.append("taxonomy/"+config['obi']['taxoURI'].split("/")[-1])
+            o_view.write_config(config, "split", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
+            o_view.close()
+    except Exception, e:
+        raise RollbackException("obi split error, rollbacking view: "+str(e), o_view)
+
+    i_dms.record_command_line(command_line)
+    i_dms.close(force=True)
+
+    logger("info", "Done.")
+

python/obitools3/version.py

@@ -1,5 +1,5 @@
 major = 3
 minor = 0
-serial= '1b14'
+serial= '1b16'
 
 version ="%d.%d.%s" % (major,minor,serial)