switch to version 3.0.0b28

and lengths. Removed minimum score option

python/obitools3/apps/progress.pyx

@@ -30,12 +30,12 @@ cdef class ProgressBar:
                  off_t maxi,
                  dict  config={},
                  str head="",
-                 double seconde=0.1,
+                 double seconds=5,
                  cut=False):
         
         self.starttime = self.clock()
         self.lasttime  = self.starttime
-        self.tickcount = <clock_t> (seconde * CLOCKS_PER_SEC)
+        self.tickcount = <clock_t> (seconds * CLOCKS_PER_SEC)
         self.freq      = 1
         self.cycle     = 0
         self.arrow     = 0

python/obitools3/commands/align.pyx

@@ -4,7 +4,7 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms import DMS
 from obitools3.dms.view.view cimport View
 from obitools3.uri.decode import open_uri
-from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.dms.view import RollbackException
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes, str2bytes
@@ -12,6 +12,9 @@ from obitools3.utils cimport tobytes, str2bytes
 from obitools3.dms.capi.obilcsalign cimport obi_lcs_align_one_column, \
                                             obi_lcs_align_two_columns
 
+from io import BufferedWriter
+from cpython.exc cimport PyErr_CheckSignals
+
 import time
 import sys
 
@@ -23,6 +26,7 @@ def addOptions(parser):
    
    addMinimalInputOption(parser)
    addMinimalOutputOption(parser)
+   addNoProgressBarOption(parser)
 
    group=parser.add_argument_group('obi align specific options')
 
@@ -201,20 +205,20 @@ def run(config):
     if output is None:
         raise Exception("Could not create output")
     o_dms = output[0]
+    output_0 = output[0]
     o_dms_name = o_dms.name 
     final_o_view_name = output[1]
+    o_view_name = final_o_view_name
 
-    # If the input and output DMS are not the same, align creating a temporary view in the input dms that will be exported to 
+    # If stdout output or the input and output DMS are not the same, align creating a temporary view in the input dms that will be exported to 
     # the right DMS and deleted in the other afterwards.
-    if i_dms != o_dms:
-        temporary_view_name = final_o_view_name
-        i=0
-        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
-            temporary_view_name = final_o_view_name+b"_"+str2bytes(str(i))
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
+        o_view_name = b"temp"
+        while o_view_name in i_dms:   # Making sure view name is unique in input DMS
+            o_view_name = o_view_name+b"_"+str2bytes(str(i))
             i+=1
-        o_view_name = temporary_view_name
-    else:
-        o_view_name = final_o_view_name
         
     # Save command config in View comments
     command_line = " ".join(sys.argv[1:])
@@ -263,8 +267,15 @@ def run(config):
         View.delete_view(i_dms, i_view_name_2)
         i_dms_2.close()
 
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view = o_dms[o_view_name]
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     # If the input and the output DMS are different, delete the temporary result view in the input DMS
-    if i_dms != o_dms:
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     

python/obitools3/commands/alignpairedend.pyx

@@ -6,7 +6,7 @@ from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 from obitools3.dms.column.column cimport Column
 from obitools3.dms.capi.obiview cimport QUALITY_COLUMN
 from obitools3.dms.capi.obitypes cimport OBI_QUAL
-from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.libalign._qsassemble import QSolexaReverseAssemble
@@ -15,7 +15,9 @@ from obitools3.libalign._solexapairend import buildConsensus, buildJoinedSequenc
 from obitools3.dms.obiseq cimport Nuc_Seq
 from obitools3.libalign.shifted_ali cimport Kmer_similarity, Ali_shifted
 from obitools3.dms.capi.obiview cimport REVERSE_SEQUENCE_COLUMN, REVERSE_QUALITY_COLUMN
+from obitools3.utils cimport str2bytes
 
+from io import BufferedWriter
 import sys
 import os
 
@@ -29,6 +31,7 @@ def addOptions(parser):
 
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
  
     group = parser.add_argument_group('obi alignpairedend specific options')
 
@@ -39,12 +42,13 @@ def addOptions(parser):
                      type=str,
                      help="URI to the reverse reads if they are in a different view than the forward reads")
 
-    group.add_argument('--score-min',
-                     action="store", dest="alignpairedend:smin",
-                     metavar="#.###",
-                     default=None,
-                     type=float,
-                     help="Minimum score for keeping alignments")
+#     group.add_argument('--score-min',
+#                      action="store", dest="alignpairedend:smin",
+#                      metavar="#.###",
+#                      default=None,
+#                      type=float,
+#                      help="Minimum score for keeping alignments. " 
+#                           "(for kmer alignment) The score is an approximation of the number of nucleotides matching in the overlap of the alignment.")
 
 #     group.add_argument('-A', '--true-ali',
 #                        action="store_true", dest="alignpairedend:trueali",
@@ -170,18 +174,29 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
 
-    view = output[1]
+    output_0 = output[0]
+    o_dms = output[0]
     
-    Column.new_column(view, QUALITY_COLUMN, OBI_QUAL)   #TODO output URI quality option?
-
-    if 'smin' in config['alignpairedend']:
-        smin = config['alignpairedend']['smin']
+    # stdout output: create temporary view
+    if type(output_0)==BufferedWriter:
+        i_dms = forward.dms   # using any dms
+        o_dms = i_dms
+        i=0
+        o_view_name = b"temp"
+        while o_view_name in i_dms:   # Making sure view name is unique in input DMS
+            o_view_name = o_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view = View_NUC_SEQS.new(o_dms, o_view_name, quality=True)
     else:
-        smin = 0
+        o_view = output[1]
+        Column.new_column(o_view, QUALITY_COLUMN, OBI_QUAL)
 
     # Initialize the progress bar
-    pb = ProgressBar(entries_len, config, seconde=5)
-    
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(entries_len, config)
+    else:
+        pb = None
+         
     #if config['alignpairedend']['trueali']:
     #    kmer_ali = False
     #    aligner = buildAlignment
@@ -206,18 +221,19 @@ def run(config):
     i = 0
     for ali in ba:
         
-        pb(i)
+        if pb is not None:
+            pb(i)
         
         PyErr_CheckSignals()
 
-        consensus = view[i]
+        consensus = o_view[i]
                
         if not two_views:
             seqF = entries[i]
         else:
             seqF = forward[i]
                         
-        if ali.score > smin and ali.consensus_len > 0 :
+        if ali.overlap_len > 0 :
             buildConsensus(ali, consensus, seqF)
         else:
             if not two_views:
@@ -225,32 +241,43 @@ def run(config):
             else:
                 seqR = reverse[i]
             buildJoinedSequence(ali, seqR, consensus, forward=seqF)
-        
-        consensus[b"smin"] = smin
-           
+                   
         if kmer_ali :
             ali.free()
     
         i+=1
 
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
 
     if kmer_ali :
         aligner.free()
 
     # Save command config in View and DMS comments
     command_line = " ".join(sys.argv[1:])
-    view.write_config(config, "alignpairedend", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
-    output[0].record_command_line(command_line)
+    o_view.write_config(config, "alignpairedend", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
+    o_dms.record_command_line(command_line)
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(view), file=sys.stderr)
-    
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
+    # If stdout output, delete the temporary imported view used to create the final file
+    if type(output_0)==BufferedWriter:
+        View_NUC_SEQS.delete_view(o_dms, o_view_name)
+        output_0.close()
+
+    # Close all DMS
     input[0].close(force=True)
     if two_views:
         rinput[0].close(force=True)
-    output[0].close(force=True)
+    o_dms.close(force=True)
 
     logger("info", "Done.")
     

python/obitools3/commands/annotate.pyx

@@ -4,11 +4,12 @@ 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
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.dms.view import RollbackException
 from functools import reduce
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes, str2bytes
+from io import BufferedWriter
 from obitools3.dms.capi.obiview cimport NUC_SEQUENCE_COLUMN, \
                                         ID_COLUMN, \
                                         DEFINITION_COLUMN, \
@@ -34,6 +35,7 @@ def addOptions(parser):
     addMinimalInputOption(parser)
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
     group=parser.add_argument_group('obi annotate specific options')
  
@@ -278,8 +280,19 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
+    output_0 = output[0]
     o_view_name = output[1]
-    
+
+    # stdout output: create temporary view
+    if type(output_0)==BufferedWriter:
+        o_dms = i_dms
+        i=0
+        o_view_name = b"temp"
+        while o_view_name in i_dms: # Making sure view name is unique in output DMS
+            o_view_name = o_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        imported_view_name = o_view_name
+
     # If the input and output DMS are not the same, import the input view in the output DMS before cloning it to modify it
     # (could be the other way around: clone and modify in the input DMS then import the new view in the output DMS)
     if i_dms != o_dms:
@@ -290,7 +303,7 @@ def run(config):
             i+=1
         View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], i_view_name, imported_view_name)
         i_view = o_dms[imported_view_name]
-        
+
     # Clone output view from input view
     o_view = i_view.clone(o_view_name)
     if o_view is None:
@@ -307,7 +320,10 @@ def run(config):
         taxo = None
 
     # Initialize the progress bar
-    pb = ProgressBar(len(o_view), config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(o_view), config)
+    else:
+        pb = None
 
     try:
 
@@ -346,14 +362,16 @@ def run(config):
         sequenceTagger = sequenceTaggerGenerator(config, taxo=taxo)
         for i in range(len(o_view)):
             PyErr_CheckSignals()
-            pb(i)
+            if pb is not None:
+                pb(i)
             sequenceTagger(o_view[i])
 
     except Exception, e:
         raise RollbackException("obi annotate error, rollbacking view: "+str(e), o_view)
 
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
     
     # Save command config in View and DMS comments
     command_line = " ".join(sys.argv[1:])
@@ -363,13 +381,19 @@ def run(config):
         input_dms_name.append(config['obi']['taxoURI'].split("/")[-3])
         input_view_name.append("taxonomy/"+config['obi']['taxoURI'].split("/")[-1])
     o_view.write_config(config, "annotate", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
-    output[0].record_command_line(command_line)
+    o_dms.record_command_line(command_line)
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary imported view used to create the final view
-    if i_dms != o_dms:
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(o_dms, imported_view_name)
         o_dms.close(force=True)
     i_dms.close(force=True)

python/obitools3/commands/build_ref_db.pyx

@@ -4,14 +4,16 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms.dms cimport DMS
 from obitools3.dms.view import RollbackException
 from obitools3.dms.capi.build_reference_db cimport build_reference_db
-from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes, str2bytes
 from obitools3.dms.view.view cimport View
 from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 
+from io import BufferedWriter
 import sys
+from cpython.exc cimport PyErr_CheckSignals
 
 
 __title__="Tag a set of sequences for PCR and sequencing errors identification"
@@ -22,6 +24,7 @@ def addOptions(parser):
     addMinimalInputOption(parser)
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
  
     group = parser.add_argument_group('obi build_ref_db specific options')
 
@@ -56,17 +59,20 @@ def run(config):
     if output is None:
         raise Exception("Could not create output")
     o_dms = output[0]
+    output_0 = output[0]
     final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, build the database creating a temporary view that will be exported to 
+    # If stdout output or the input and output DMS are not the same, build the database creating a temporary view that will be exported to 
     # the right DMS and deleted in the other afterwards.
-    if i_dms != o_dms:
-        temporary_view_name = final_o_view_name
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
         while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
-            temporary_view_name = final_o_view_name+b"_"+str2bytes(str(i))
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
             i+=1
         o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
     else:
         o_view_name = final_o_view_name
 
@@ -80,22 +86,29 @@ def run(config):
     input_dms_name.append(config['obi']['taxoURI'].split("/")[-3])
     input_view_name.append("taxonomy/"+config['obi']['taxoURI'].split("/")[-1])
     comments = View.print_config(config, "build_ref_db", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
-
+    
     if build_reference_db(i_dms.name_with_full_path, tobytes(i_view_name), tobytes(taxonomy_name), tobytes(o_view_name), comments, config['build_ref_db']['threshold']) < 0:
         raise Exception("Error building a reference database")
     
     # If the input and output DMS are not the same, export result view to output DMS
     if i_dms != o_dms:
         View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
-    
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view = o_dms[o_view_name]
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     # Save command config in DMS comments
     o_dms.record_command_line(command_line)
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_dms[final_o_view_name]), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary result view in the input DMS
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     

python/obitools3/commands/cat.pyx

@@ -15,6 +15,7 @@ from obitools3.dms.capi.obiview cimport NUC_SEQUENCE_COLUMN, REVERSE_SEQUENCE_CO
 from obitools3.dms.capi.obitypes cimport OBI_SEQ, OBI_QUAL
 from obitools3.dms.column.column cimport Column
 
+from io import BufferedWriter
 import time
 import sys
  
@@ -76,53 +77,70 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
+    output_0 = output[0]
     o_view = output[1]
     
+    # stdout output
+    if type(output_0)==BufferedWriter: 
+        o_dms = i_dms
+    
     # Initialize quality columns and their associated sequence columns if needed
-    if not remove_qual:
-        if NUC_SEQUENCE_COLUMN not in o_view:
-            Column.new_column(o_view, NUC_SEQUENCE_COLUMN, OBI_SEQ)
-        Column.new_column(o_view, QUALITY_COLUMN, OBI_QUAL, associated_column_name=NUC_SEQUENCE_COLUMN, associated_column_version=o_view[NUC_SEQUENCE_COLUMN].version)    
-    if not remove_rev_qual:
-        Column.new_column(o_view, REVERSE_SEQUENCE_COLUMN, OBI_SEQ)
-        Column.new_column(o_view, REVERSE_QUALITY_COLUMN, OBI_QUAL, associated_column_name=REVERSE_SEQUENCE_COLUMN, associated_column_version=o_view[REVERSE_SEQUENCE_COLUMN].version)
+    if type(output_0) != BufferedWriter: 
+        if not remove_qual:
+            if NUC_SEQUENCE_COLUMN not in o_view:
+                Column.new_column(o_view, NUC_SEQUENCE_COLUMN, OBI_SEQ)
+            Column.new_column(o_view, QUALITY_COLUMN, OBI_QUAL, associated_column_name=NUC_SEQUENCE_COLUMN, associated_column_version=o_view[NUC_SEQUENCE_COLUMN].version)    
+        if not remove_rev_qual:
+            Column.new_column(o_view, REVERSE_SEQUENCE_COLUMN, OBI_SEQ)
+            Column.new_column(o_view, REVERSE_QUALITY_COLUMN, OBI_QUAL, associated_column_name=REVERSE_SEQUENCE_COLUMN, associated_column_version=o_view[REVERSE_SEQUENCE_COLUMN].version)
         
     # Initialize multiple elements columns
-    dict_cols = {}
-    for v in iview_list:
-        for coln in v.keys():
-            if v[coln].nb_elements_per_line > 1:
-                if coln not in dict_cols:
-                    dict_cols[coln] = {}
-                    dict_cols[coln]['eltnames'] = set(v[coln].elements_names)
-                    dict_cols[coln]['nbelts'] = v[coln].nb_elements_per_line
-                    dict_cols[coln]['obitype'] = v[coln].data_type_int
-                else:
-                    dict_cols[coln]['eltnames'] = set(v[coln].elements_names + list(dict_cols[coln]['eltnames']))
-                    dict_cols[coln]['nbelts'] = len(dict_cols[coln]['eltnames'])
-    for coln in dict_cols:
-        Column.new_column(o_view, coln, dict_cols[coln]['obitype'], 
-                          nb_elements_per_line=dict_cols[coln]['nbelts'], elements_names=list(dict_cols[coln]['eltnames']))
+    if type(output_0)==BufferedWriter: 
+        dict_cols = {}
+        for v in iview_list:
+            for coln in v.keys():
+                if v[coln].nb_elements_per_line > 1:
+                    if coln not in dict_cols:
+                        dict_cols[coln] = {}
+                        dict_cols[coln]['eltnames'] = set(v[coln].elements_names)
+                        dict_cols[coln]['nbelts'] = v[coln].nb_elements_per_line
+                        dict_cols[coln]['obitype'] = v[coln].data_type_int
+                    else:
+                        dict_cols[coln]['eltnames'] = set(v[coln].elements_names + list(dict_cols[coln]['eltnames']))
+                        dict_cols[coln]['nbelts'] = len(dict_cols[coln]['eltnames'])
+        for coln in dict_cols:
+            Column.new_column(o_view, coln, dict_cols[coln]['obitype'], 
+                              nb_elements_per_line=dict_cols[coln]['nbelts'], elements_names=list(dict_cols[coln]['eltnames']))
     
     # Initialize the progress bar
-    pb = ProgressBar(total_len, config, seconde=5)
+    if not config['obi']['noprogressbar']:
+        pb = ProgressBar(total_len, config)
+    else:
+        pb = None
     
     i = 0
     for v in iview_list:
-        for l in v:
+        for entry in v:
             PyErr_CheckSignals()
-            pb(i)
-            o_view[i] = l
+            if pb is not None:
+                pb(i)
+            if type(output_0)==BufferedWriter: 
+                rep = repr(entry)
+                output_0.write(str2bytes(rep)+b"\n")
+            else:
+                o_view[i] = entry
             i+=1
 
     # Deletes quality columns if needed
-    if QUALITY_COLUMN in o_view and remove_qual :
-        o_view.delete_column(QUALITY_COLUMN)
-    if REVERSE_QUALITY_COLUMN in o_view and remove_rev_qual :
-        o_view.delete_column(REVERSE_QUALITY_COLUMN)
-
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if type(output_0)!=BufferedWriter: 
+        if QUALITY_COLUMN in o_view and remove_qual :
+            o_view.delete_column(QUALITY_COLUMN)
+        if REVERSE_QUALITY_COLUMN in o_view and remove_rev_qual :
+            o_view.delete_column(REVERSE_QUALITY_COLUMN)
+    
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
     
     # Save command config in DMS comments
     command_line = " ".join(sys.argv[1:])

python/obitools3/commands/clean.pyx

@@ -4,13 +4,14 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms.dms cimport DMS
 from obitools3.dms.view import RollbackException
 from obitools3.dms.capi.obiclean cimport obi_clean
-from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes, str2bytes
 from obitools3.dms.view.view cimport View
 from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 
+from io import BufferedWriter
 import sys
 
 
@@ -21,7 +22,8 @@ def addOptions(parser):
 
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
- 
+    addNoProgressBarOption(parser)
+    
     group = parser.add_argument_group('obi clean specific options')
 
     group.add_argument('--distance', '-d',
@@ -36,8 +38,7 @@ def addOptions(parser):
                        dest="clean:sample-tag-name",
                        metavar="<SAMPLE TAG NAME>",
                        type=str,
-                       default="merged_sample",
-                       help="Name of the tag where sample counts are kept.")
+                       help="Name of the tag where merged sample count informations are kept (typically generated by obi uniq, usually MERGED_sample, default: None).")
     
     group.add_argument('--ratio', '-r',
                        action="store", dest="clean:ratio",
@@ -89,17 +90,20 @@ def run(config):
     if output is None:
         raise Exception("Could not create output")
     o_dms = output[0]
+    output_0 = output[0]
     final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, run obiclean creating a temporary view that will be exported to 
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported to 
     # the right DMS and deleted in the other afterwards.
-    if i_dms != o_dms:
-        temporary_view_name = final_o_view_name
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
         while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
-            temporary_view_name = final_o_view_name+b"_"+str2bytes(str(i))
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
             i+=1
         o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
     else:
         o_view_name = final_o_view_name
         
@@ -107,6 +111,9 @@ def run(config):
     command_line = " ".join(sys.argv[1:])
     comments = View.print_config(config, "clean", command_line, input_dms_name=[i_dms_name], input_view_name=[i_view_name])
 
+    if 'sample-tag-name' not in config['clean']:
+        config['clean']['sample-tag-name'] = ""
+        
     if obi_clean(i_dms.name_with_full_path, tobytes(i_view_name), tobytes(config['clean']['sample-tag-name']), tobytes(o_view_name), comments, \
               config['clean']['distance'], config['clean']['ratio'], config['clean']['heads-only'], config['clean']['thread-count']) < 0:
         raise Exception("Error running obiclean")
@@ -114,15 +121,22 @@ def run(config):
     # If the input and output DMS are not the same, export result view to output DMS
     if i_dms != o_dms:
         View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
-    
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view = o_dms[o_view_name]
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     # Save command config in DMS comments
     o_dms.record_command_line(command_line)
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_dms[final_o_view_name]), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary result view in the input DMS
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     

python/obitools3/commands/count.pyx

@@ -22,7 +22,7 @@ def addOptions(parser):
     group.add_argument('-s','--sequence',
                         action="store_true", dest="count:sequence",
                         default=False,
-                        help="Prints only the number of sequence records.")
+                        help="Prints only the number of sequence records (much faster, default: False).")
  
     group.add_argument('-a','--all',
                         action="store_true", dest="count:all",

python/obitools3/commands/ecopcr.pyx

@@ -5,10 +5,10 @@ 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 addMinimalInputOption, addMinimalOutputOption, addTaxonomyOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addTaxonomyOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
-from obitools3.utils cimport tobytes
+from obitools3.utils cimport tobytes, str2bytes
 from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 from obitools3.dms.view import View
 
@@ -16,6 +16,7 @@ from libc.stdlib  cimport malloc, free
 from libc.stdint  cimport int32_t
 
 import sys
+from io import BufferedWriter
 
 
 __title__="in silico PCR"
@@ -27,6 +28,7 @@ def addOptions(parser):
     addMinimalInputOption(parser)
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
  
     group = parser.add_argument_group('obi ecopcr specific options')
@@ -35,12 +37,14 @@ def addOptions(parser):
                        action="store", dest="ecopcr:primer1",
                        metavar='<PRIMER>',
                        type=str,
+                       required=True,
                        help="Forward primer, length must be less than or equal to 32")
 
     group.add_argument('--primer2', '-R',
                        action="store", dest="ecopcr:primer2",
                        metavar='<PRIMER>',
                        type=str,
+                       required=True,
                        help="Reverse primer, length must be less than or equal to 32")
 
     group.add_argument('--error', '-e',
@@ -167,11 +171,20 @@ def run(config):
     if output is None:
         raise Exception("Could not create output")
     o_dms = output[0]
+    output_0 = output[0]
     o_dms_name = output[0].name
     o_view_name = output[1]
 
     # Read taxonomy name    
     taxonomy_name = config['obi']['taxoURI'].split("/")[-1]   # Robust in theory
+
+    # If stdout output create a temporary view in the input dms that will be deleted afterwards.
+    if type(output_0)==BufferedWriter:
+        o_dms = i_dms
+        o_view_name = b"temp"
+        while o_view_name in i_dms:   # Making sure view name is unique in input DMS
+            o_view_name = o_view_name+b"_"+str2bytes(str(i))
+            i+=1
     
     # Save command config in View comments
     command_line = " ".join(sys.argv[1:])
@@ -199,10 +212,21 @@ def run(config):
 
     free(restrict_to_taxids_p)
     free(ignore_taxids_p)
-    
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view = o_dms[o_view_name]
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_dms[o_view_name]), file=sys.stderr)
 
+    # If stdout output, delete the temporary result view in the input DMS
+    if type(output_0)==BufferedWriter:
+        View.delete_view(i_dms, o_view_name)
+
     i_dms.close(force=True)
     o_dms.close(force=True)
 

python/obitools3/commands/ecotag.pyx

@@ -4,7 +4,7 @@ from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms.dms cimport DMS
 from obitools3.dms.view import RollbackException
 from obitools3.dms.capi.obiecotag cimport obi_ecotag
-from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addTaxonomyOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.utils cimport tobytes, str2bytes
@@ -12,6 +12,7 @@ from obitools3.dms.view.view cimport View
 from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
 
 import sys
+from io import BufferedWriter
 
 
 __title__="Taxonomic assignment of sequences"
@@ -22,6 +23,7 @@ def addOptions(parser):
     addMinimalInputOption(parser)
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
  
     group = parser.add_argument_group('obi ecotag specific options')
 
@@ -75,17 +77,19 @@ def run(config):
     if output is None:
         raise Exception("Could not create output")
     o_dms = output[0]
+    output_0 = output[0]
     final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, run ecotag creating a temporary view that will be exported to 
-    # the right DMS and deleted in the other afterwards.
-    if i_dms != o_dms:
-        temporary_view_name = final_o_view_name
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported and deleted.
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
         while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
-            temporary_view_name = final_o_view_name+b"_"+str2bytes(str(i))
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
             i+=1
         o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
     else:
         o_view_name = final_o_view_name
 
@@ -120,11 +124,18 @@ def run(config):
     # Save command config in DMS comments
     o_dms.record_command_line(command_line)
 
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view = o_dms[o_view_name]
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_dms[final_o_view_name]), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary result view in the input DMS
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     

python/obitools3/commands/export.pyx

@@ -74,7 +74,7 @@ def run(config):
     if config['obi']['noprogressbar']:
         pb = None
     else:
-        pb = ProgressBar(withoutskip - skip, config, seconde=5)
+        pb = ProgressBar(withoutskip - skip, config)
 
     i=0
     for seq in iview :

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, addTaxonomyOption, addMinimalOutputOption
+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, str2bytes
@@ -14,6 +14,7 @@ import time
 import re
 import sys
 import ast
+from io import BufferedWriter
 from cpython.exc cimport PyErr_CheckSignals
 
  
@@ -28,6 +29,7 @@ def addOptions(parser):
     addMinimalInputOption(parser)
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
     group=parser.add_argument_group("obi grep specific options")
  
@@ -161,8 +163,7 @@ def obi_eval(compiled_expr, loc_env, line):
     return obi_eval_result
     
 
-def Filter_generator(options, tax_filter):
-    #taxfilter = taxonomyFilterGenerator(options)
+def Filter_generator(options, tax_filter, i_view):
 
     # Initialize conditions
     predicates = None
@@ -171,6 +172,9 @@ def Filter_generator(options, tax_filter):
     attributes = None
     if "attributes" in options and len(options["attributes"]) > 0:
         attributes = options["attributes"]
+        for attribute in attributes:
+            if attribute not in i_view:
+                return None
     lmax = None
     if "lmax" in options:
         lmax = options["lmax"]
@@ -196,6 +200,8 @@ def Filter_generator(options, tax_filter):
     if "attribute_patterns" in options and len(options["attribute_patterns"]) > 0:
         for p in options["attribute_patterns"]:
             attribute, pattern = p.split(":", 1)
+            if attribute not in i_view:
+                return None
             attribute_patterns[tobytes(attribute)] = re.compile(tobytes(pattern))
     
     def filter(line, loc_env):
@@ -300,16 +306,21 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
-    o_view_name_final = output[1]
-    o_view_name = o_view_name_final
-    
-    # If the input and output DMS are not the same, create output view in input DMS first, then export it
-    # to output DMS, making sure the temporary view name is unique in the input DMS 
-    if i_dms != o_dms:
+    output_0 = output[0]
+    final_o_view_name = output[1]
+
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported and deleted afterwards.
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
-        while o_view_name in i_dms:
-            o_view_name = o_view_name_final+b"_"+str2bytes(str(i))
-            i+=1        
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
+    else:
+        o_view_name = final_o_view_name
 
     if 'taxoURI' in config['obi'] and config['obi']['taxoURI'] is not None:
         taxo_uri = open_uri(config["obi"]["taxoURI"])
@@ -320,33 +331,49 @@ def run(config):
         taxo = None
 
     # Initialize the progress bar
-    pb = ProgressBar(len(i_view), config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(i_view), config)
+    else:
+        pb = None
  
     # Apply filter
     tax_filter = Taxonomy_filter_generator(taxo, config["grep"])
-    filter = Filter_generator(config["grep"], tax_filter)
+    filter = Filter_generator(config["grep"], tax_filter, i_view)        
     selection = Line_selection(i_view)
-    for i in range(len(i_view)):
-        PyErr_CheckSignals()
-        pb(i)
-        line = i_view[i]
-                 
-        loc_env = {"sequence": line, "line": line, "taxonomy": taxo, "obi_eval_result": False}
-        
-        good = filter(line, loc_env)
- 
-        if good :
+    
+    if filter is None and config["grep"]["invert_selection"]: # all sequences are selected: filter is None if no line will be selected because some columns don't exist
+        for i in range(len(i_view)):
+            PyErr_CheckSignals()
+            if pb is not None:
+                pb(i)
             selection.append(i)
-
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    
+    elif filter is not None :   # filter is None if no line will be selected because some columns don't exist
+        for i in range(len(i_view)):
+            PyErr_CheckSignals()
+            if pb is not None:
+                pb(i)
+            line = i_view[i]
+            
+            loc_env = {"sequence": line, "line": line, "taxonomy": taxo, "obi_eval_result": False}
+            
+            good = filter(line, loc_env)
+     
+            if good :
+                selection.append(i)
+   
+    if pb is not None:
+        pb(len(i_view), force=True)
+        print("", file=sys.stderr)
 
     # Create output view with the line selection
     try:
         o_view = selection.materialize(o_view_name)
     except Exception, e:
         raise RollbackException("obi grep error, rollbacking view: "+str(e), o_view)
-    
+
+    logger("info", "Grepped %d entries" % len(o_view))
+
     # Save command config in View and DMS comments
     command_line = " ".join(sys.argv[1:])
     input_dms_name=[input[0].name]
@@ -361,14 +388,20 @@ def run(config):
     # and delete the temporary view in the input DMS
     if i_dms != o_dms:
         o_view.close()
-        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, o_view_name_final)
-        o_view = o_dms[o_view_name_final]
+        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
+        o_view = o_dms[final_o_view_name]
+    
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary imported view used to create the final view
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     i_dms.close(force=True)

python/obitools3/commands/head.pyx

@@ -4,14 +4,15 @@ 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, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.dms.view import RollbackException
 from obitools3.apps.config import logger
 from obitools3.utils cimport str2bytes
 
 import time
 import sys
- 
+from io import BufferedWriter
+
 from cpython.exc cimport PyErr_CheckSignals
 
 
@@ -22,6 +23,7 @@ def addOptions(parser):
     
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
     group=parser.add_argument_group('obi head specific options')
 
@@ -53,31 +55,41 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
-    o_view_name_final = output[1]
-    o_view_name = o_view_name_final
+    output_0 = output[0]
+    final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, create output view in input DMS first, then export it
-    # to output DMS, making sure the temporary view name is unique in the input DMS 
-    if i_dms != o_dms:
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported and deleted.
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
-        while o_view_name in i_dms:
-            o_view_name = o_view_name_final+b"_"+str2bytes(str(i))
-            i+=1        
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
+    else:
+        o_view_name = final_o_view_name
     
     n = min(config['head']['count'], len(i_view))
 
     # Initialize the progress bar
-    pb = ProgressBar(n, config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(n, config)
+    else:
+        pb = None
         
     selection = Line_selection(i_view)
     
     for i in range(n):
         PyErr_CheckSignals()
-        pb(i)
+        if pb is not None:
+            pb(i)
         selection.append(i)
 
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
 
     # Create output view with the line selection
     try:
@@ -94,14 +106,20 @@ def run(config):
     # and delete the temporary view in the input DMS
     if i_dms != o_dms:
         o_view.close()
-        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, o_view_name_final)
-        o_view = o_dms[o_view_name_final]
+        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
+        o_view = o_dms[final_o_view_name]
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(view), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary imported view used to create the final view
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     i_dms.close(force=True)

python/obitools3/commands/import.pyx

@@ -47,6 +47,8 @@ from obitools3.apps.config import logger
 
 from cpython.exc cimport PyErr_CheckSignals
 
+from io import BufferedWriter
+
 
 __title__="Imports sequences from different formats into a DMS"
  
@@ -130,7 +132,7 @@ def run(config):
         if entry_count > 0:
             logger("info", "Importing %d entries", entry_count)
         else:
-            logger("info", "Importing an unknow number of entries")
+            logger("info", "Importing an unknown number of entries")
         
         # TODO a bit dirty?
         if input[2]==Nuc_Seq or input[2]==View_NUC_SEQS:
@@ -178,7 +180,7 @@ def run(config):
         return
 
     if entry_count >= 0:
-        pb = ProgressBar(entry_count, config, seconde=5)
+        pb = ProgressBar(entry_count, config)
         
     NUC_SEQS_view = False
     if isinstance(output[1], View) :
@@ -243,7 +245,7 @@ def run(config):
         if isinstance(input[0], CompressedFile):
             input_is_file = True
         if entry_count >= 0:
-            pb = ProgressBar(entry_count, config, seconde=5)
+            pb = ProgressBar(entry_count, config)
         try:
             input[0].close()
         except AttributeError:
@@ -260,7 +262,6 @@ def run(config):
 
         if entry is None:  # error or exception handled at lower level, not raised because Python generators can't resume after any exception is raised
             if config['obi']['skiperror']:
-                i-=1
                 continue
             else:
                 raise RollbackException("obi import error, rollbacking view", view)

python/obitools3/commands/ls.pyx

@@ -34,9 +34,10 @@ def run(config):
     if input[2] == DMS and not config['ls']['longformat']:
         dms = input[0]
         l = []
-        for view in input[0]:
-            l.append(tostr(view) + "\t(Date created: " + str(bytes2str_object(dms[view].comments["Date created"]))+")")
-            dms[view].close()
+        for viewname in input[0]:
+            view = dms[viewname]
+            l.append(tostr(viewname) + "\t(Date created: " + str(bytes2str_object(view.comments["Date created"]))+")")
+            view.close()
         l.sort()
         for v in l:
             print(v)

python/obitools3/commands/ngsfilter.pyx

@@ -2,10 +2,10 @@
 
 from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
 from obitools3.dms import DMS
-from obitools3.dms.view import RollbackException
+from obitools3.dms.view import RollbackException, View
 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 addMinimalInputOption, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
 from obitools3.libalign._freeendgapfm import FreeEndGapFullMatch
@@ -14,13 +14,14 @@ from obitools3.dms.obiseq cimport Nuc_Seq
 from obitools3.dms.capi.obitypes cimport OBI_SEQ, OBI_QUAL
 from obitools3.dms.capi.apat cimport MAX_PATTERN
 from obitools3.dms.capi.obiview cimport REVERSE_SEQUENCE_COLUMN, REVERSE_QUALITY_COLUMN
-from obitools3.utils cimport tobytes
+from obitools3.utils cimport tobytes, str2bytes
 
 from libc.stdint cimport INT32_MAX
 from functools import reduce
 import math
 import sys
 from cpython.exc cimport PyErr_CheckSignals
+from io import BufferedWriter
 
 
 #REVERSE_SEQ_COLUMN_NAME = b"REVERSE_SEQUENCE"      # used by alignpairedend tool
@@ -34,7 +35,8 @@ def addOptions(parser):
 
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
- 
+    addNoProgressBarOption(parser)
+    
     group = parser.add_argument_group('obi ngsfilter specific options')
 
     group.add_argument('-t','--info-view',
@@ -42,6 +44,7 @@ def addOptions(parser):
                      metavar="<URI>",
                      type=str,
                      default=None,
+                     required=True,
                      help="URI to the view containing the samples definition (with tags, primers, sample names,...).\n"
                           "\nWarning: primer lengths must be less than or equal to 32")
 
@@ -57,7 +60,7 @@ def addOptions(parser):
                      metavar="<URI>",
                      type=str,
                      default=None,
-                     help="URI to the view used to store the sequences unassigned to any sample")
+                     help="URI to the view used to store the sequences unassigned to any sample. Those sequences are untrimmed.")
 
     group.add_argument('--no-tags',
                      action="store_true", dest="ngsfilter:notags",
@@ -478,6 +481,8 @@ cdef tuple annotate(sequences, infos, no_tags, verbose=False):
     if not directmatch[0].forward:
         sequences[0] = sequences[0].reverse_complement
         sequences[0][b'reversed'] = True   # used by the alignpairedend tool (in kmer_similarity.c)
+    else:
+        sequences[0][b'reversed'] = False   # used by the alignpairedend tool (in kmer_similarity.c)
 
     sample=None
     if not no_tags:
@@ -505,7 +510,7 @@ cdef tuple annotate(sequences, infos, no_tags, verbose=False):
                     sample=None
         
         if sample is None:
-            sequences[0][b'error']=b"No tags found"
+            sequences[0][b'error']=b"No sample with that tag combination"
             return False, sequences[0]
     
         sequences[0].update(sample)
@@ -536,7 +541,8 @@ def run(config):
         raise Exception("Could not open input reads")
     if input[2] != View_NUC_SEQS:
         raise NotImplementedError('obi ngsfilter only works on NUC_SEQS views')
-
+    i_dms = input[0]
+    
     if "reverse" in config["ngsfilter"]:
         
         forward = input[1]        
@@ -577,8 +583,19 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     
+    o_dms = output[0]
+    output_0 = output[0]
     o_view = output[1]
     
+    # If stdout output, create a temporary view in the input dms that will be deleted afterwards.
+    if type(output_0)==BufferedWriter:
+        o_dms = i_dms
+        o_view_name = b"temp"
+        while o_view_name in i_dms:   # Making sure view name is unique in input DMS
+            o_view_name = o_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view = View_NUC_SEQS.new(i_dms, o_view_name)
+
     # Open the view containing the informations about the tags and the primers
     info_input = open_uri(config['ngsfilter']['info_view'])
     if info_input is None:
@@ -599,7 +616,10 @@ def run(config):
         unidentified = None
         
     # Initialize the progress bar
-    pb = ProgressBar(entries_len, config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(entries_len, config)
+    else:
+        pb = None
 
     # Check and store primers and tags
     try:
@@ -633,7 +653,8 @@ def run(config):
     try:
         for i in range(entries_len):
             PyErr_CheckSignals()
-            pb(i)
+            if pb is not None:
+                pb(i)
             if not_aligned:
                 modseq = [Nuc_Seq.new_from_stored(forward[i]), Nuc_Seq.new_from_stored(reverse[i])]
             else:
@@ -643,7 +664,13 @@ def run(config):
                 o_view[g].set(oseq.id, oseq.seq, definition=oseq.definition, quality=oseq.quality, tags=oseq)
                 g+=1
             elif unidentified is not None:
-                unidentified[u].set(oseq.id, oseq.seq, definition=oseq.definition, quality=oseq.quality, tags=oseq)
+                # Untrim sequences (put original back)
+                if len(modseq) > 1:
+                    oseq[REVERSE_SEQUENCE_COLUMN] = reverse[i].seq
+                    oseq[REVERSE_QUALITY_COLUMN] = reverse[i].quality
+                    unidentified[u].set(oseq.id, forward[i].seq, definition=oseq.definition, quality=forward[i].quality, tags=oseq)
+                else:
+                    unidentified[u].set(oseq.id, entries[i].seq, definition=oseq.definition, quality=entries[i].quality, tags=oseq)
                 u+=1
     except Exception, e:
         if unidentified is not None:
@@ -651,8 +678,9 @@ def run(config):
         else:
             raise RollbackException("obi ngsfilter error, rollbacking view: "+str(e), o_view)
     
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
 
     # Save command config in View and DMS comments
     command_line = " ".join(sys.argv[1:])
@@ -661,13 +689,23 @@ def run(config):
         unidentified.write_config(config, "ngsfilter", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
         # Add comment about unidentified seqs
         unidentified.comments["info"] = "View containing sequences categorized as unidentified by the ngsfilter command"
-    output[0].record_command_line(command_line)
+    o_dms.record_command_line(command_line)
     
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
+
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
 
-    input[0].close(force=True)
-    output[0].close(force=True)
+    # If stdout output, delete the temporary result view in the input DMS
+    if type(output_0)==BufferedWriter:
+        View.delete_view(i_dms, o_view_name)
+
+    i_dms.close(force=True)
+    o_dms.close(force=True)
     info_input[0].close(force=True)
     if unidentified is not None:
         unidentified_input[0].close(force=True)

python/obitools3/commands/sort.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, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.dms.view import RollbackException
 from obitools3.apps.config import logger
 from obitools3.utils cimport str2bytes
@@ -24,6 +24,7 @@ from obitools3.dms.capi.obitypes cimport OBI_BOOL, \
 import time
 import sys
 from cpython.exc cimport PyErr_CheckSignals
+from io import BufferedWriter
 
  
 NULL_VALUE = {OBI_BOOL: OBIBool_NA, 
@@ -42,6 +43,7 @@ def addOptions(parser):
     
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
     group=parser.add_argument_group('obi sort specific options')
 
@@ -59,7 +61,7 @@ def addOptions(parser):
 
 
 def line_cmp(line, key, pb): 
-    pb   
+    pb
     if line[key] is None:
         return NULL_VALUE[line.view[key].data_type_int]
     else:
@@ -86,20 +88,28 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
-    o_view_name_final = output[1]
-    o_view_name = o_view_name_final
+    output_0 = output[0]
+    final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, create output view in input DMS first, then export it
-    # to output DMS, making sure the temporary view name is unique in the input DMS 
-    if i_dms != o_dms:
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported and deleted.
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
-        while o_view_name in i_dms:
-            o_view_name = o_view_name_final+b"_"+str2bytes(str(i))
-            i+=1        
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
+    else:
+        o_view_name = final_o_view_name
 
     # Initialize the progress bar
-    pb = ProgressBar(len(i_view), config, seconde=5)
-    
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(i_view), config)
+    else:
+        pb = None
+
     keys = config['sort']['keys']
         
     selection = Line_selection(i_view)
@@ -110,10 +120,14 @@ def run(config):
     
     for k in keys:  # TODO order?
         PyErr_CheckSignals()
-        selection.sort(key=lambda line_idx: line_cmp(i_view[line_idx], k, pb(line_idx)), reverse=config['sort']['reverse'])
-
-    pb(len(i_view), force=True)
-    print("", file=sys.stderr)
+        if pb is not None:
+            selection.sort(key=lambda line_idx: line_cmp(i_view[line_idx], k, pb(line_idx)), reverse=config['sort']['reverse'])
+        else:
+            selection.sort(key=lambda line_idx: line_cmp(i_view[line_idx], k, None), reverse=config['sort']['reverse'])
+    
+    if pb is not None:
+        pb(len(i_view), force=True)
+        print("", file=sys.stderr)
 
     # Create output view with the sorted line selection
     try:
@@ -132,16 +146,23 @@ def run(config):
     # and delete the temporary view in the input DMS
     if i_dms != o_dms:
         o_view.close()
-        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, o_view_name_final)
-        o_view = o_dms[o_view_name_final]
+        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
+        o_view = o_dms[final_o_view_name]
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary imported view used to create the final view
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
+
     i_dms.close(force=True)
 
     logger("info", "Done.")

python/obitools3/commands/stats.pyx

@@ -162,7 +162,7 @@ def run(config):
     lcat=0
     
     # Initialize the progress bar
-    pb = ProgressBar(len(i_view), config, seconde=5)
+    pb = ProgressBar(len(i_view), config)
     
     for i in range(len(i_view)):
         PyErr_CheckSignals()

python/obitools3/commands/tail.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, addMinimalOutputOption
+from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption, addNoProgressBarOption
 from obitools3.dms.view import RollbackException
 from obitools3.apps.config import logger
 from obitools3.utils cimport str2bytes
@@ -12,6 +12,7 @@ from obitools3.utils cimport str2bytes
 import time
 import sys
 from cpython.exc cimport PyErr_CheckSignals
+from io import BufferedWriter
 
 
 __title__="Keep the N last lines of a view."
@@ -21,6 +22,7 @@ def addOptions(parser):
     
     addMinimalInputOption(parser)
     addMinimalOutputOption(parser)
+    addNoProgressBarOption(parser)
 
     group=parser.add_argument_group('obi tail specific options')
 
@@ -52,31 +54,41 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     o_dms = output[0]
-    o_view_name_final = output[1]
-    o_view_name = o_view_name_final
+    output_0 = output[0]
+    final_o_view_name = output[1]
     
-    # If the input and output DMS are not the same, create output view in input DMS first, then export it
-    # to output DMS, making sure the temporary view name is unique in the input DMS 
-    if i_dms != o_dms:
+    # If stdout output or the input and output DMS are not the same, create a temporary view that will be exported and deleted.
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
         i=0
-        while o_view_name in i_dms:
-            o_view_name = o_view_name_final+b"_"+str2bytes(str(i))
-            i+=1        
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+        if type(output_0)==BufferedWriter:
+            o_dms = i_dms
+    else:
+        o_view_name = final_o_view_name
     
     start = max(len(i_view) - config['tail']['count'], 0)
 
     # Initialize the progress bar
-    pb = ProgressBar(len(i_view) - start, config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(i_view) - start, config)
+    else:
+        pb = None
         
     selection = Line_selection(i_view)
     
     for i in range(start, len(i_view)):
         PyErr_CheckSignals()
-        pb(i)
+        if pb is not None:
+            pb(i)
         selection.append(i)
 
-    pb(i, force=True)
-    print("", file=sys.stderr)
+    if pb is not None:
+        pb(i, force=True)
+        print("", file=sys.stderr)
 
     # Save command config in View comments
     command_line = " ".join(sys.argv[1:])
@@ -97,14 +109,20 @@ def run(config):
     # and delete the temporary view in the input DMS
     if i_dms != o_dms:
         o_view.close()
-        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, o_view_name_final)
-        o_view = o_dms[o_view_name_final]
+        View.import_view(i_dms.full_path[:-7], o_dms.full_path[:-7], o_view_name, final_o_view_name)
+        o_view = o_dms[final_o_view_name]
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
 
-    # If the input and the output DMS are different, delete the temporary imported view used to create the final view
-    if i_dms != o_dms:
+    # If the input and the output DMS are different or if stdout output, delete the temporary imported view used to create the final view
+    if i_dms != o_dms or type(output_0)==BufferedWriter:
         View.delete_view(i_dms, o_view_name)
         o_dms.close(force=True)
     i_dms.close(force=True)

python/obitools3/commands/uniq.pyx

@@ -14,13 +14,15 @@ from obitools3.dms.capi.obitypes cimport OBI_INT, OBI_STR, index_t
 from obitools3.apps.optiongroups import addMinimalInputOption, \
                                         addMinimalOutputOption, \
                                         addTaxonomyOption, \
-                                        addEltLimitOption
+                                        addEltLimitOption, \
+                                        addNoProgressBarOption
 from obitools3.uri.decode import open_uri
 from obitools3.apps.config import logger
-from obitools3.utils cimport tobytes, tostr
+from obitools3.utils cimport tobytes, tostr, str2bytes
 
 import sys
 from cpython.exc cimport PyErr_CheckSignals
+from io import BufferedWriter
 
 
 __title__="Group sequence records together"
@@ -32,6 +34,7 @@ def addOptions(parser):
     addTaxonomyOption(parser)
     addMinimalOutputOption(parser)
     addEltLimitOption(parser)
+    addNoProgressBarOption(parser)
  
     group = parser.add_argument_group('obi uniq specific options')
  
@@ -143,12 +146,16 @@ cdef merge_taxonomy_classification(View_NUC_SEQS o_view, Taxonomy taxonomy, dict
     scientific_name_column = o_view[b"scientific_name"]  
     
     # Initialize the progress bar
-    pb = ProgressBar(len(o_view), config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(o_view), config)
+    else:
+        pb = None
     
     i=0
     for seq in o_view:
         PyErr_CheckSignals()
-        pb(i)
+        if pb is not None:
+            pb(i)
         if MERGED_TAXID_COLUMN in seq :
             m_taxids = []            
             m_taxids_dict = seq[MERGED_TAXID_COLUMN]
@@ -191,7 +198,8 @@ cdef merge_taxonomy_classification(View_NUC_SEQS o_view, Taxonomy taxonomy, dict
             scientific_name_column[i] = taxonomy.get_scientific_name(taxid)
         i+=1
     
-    pb(len(o_view), force=True)
+    if pb is not None:
+        pb(len(o_view), force=True)
     
 
 cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, dict config, list mergedKeys_list=None, Taxonomy taxonomy=None, bint mergeIds=False, list categories=None, int max_elts=1000000) :
@@ -297,7 +305,8 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
     iter_view = iter(view)
     for i_seq in iter_view :
         PyErr_CheckSignals()
-        pb(i)
+        if pb is not None:
+            pb(i)
          
         # This can't be done in the same line as the unique_id tuple creation because it generates a bug
         # where Cython (version 0.25.2) does not detect the reference to the categs_list variable and deallocates 
@@ -307,6 +316,7 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
         for x in categories :
             catl.append(i_seq[x])    
           
+        #unique_id = tuple(catl) + (i_seq_col[i],)
         unique_id = tuple(catl) + (i_seq_col.get_line_idx(i),)
         #unique_id = tuple(i_seq[x] for x in categories) + (seq_col.get_line_idx(i),)  # The line that cython can't read properly
          
@@ -414,17 +424,22 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
     o_count_col = o_view[COUNT_COLUMN]
     if COUNT_COLUMN in view:
         i_count_col = view[COUNT_COLUMN]
+    
+    if pb is not None:
+        pb(len(view), force=True)
+        print("")
         
-    pb(len(view), force=True)
-    print("")
     logger("info", "Second browsing through the input")
+    
     # Initialize the progress bar
-    pb = ProgressBar(len(uniques), seconde=5)
+    if pb is not None:
+        pb = ProgressBar(len(view))
+        
     o_idx = 0
+    total_treated = 0
     
     for unique_id in uniques :
         PyErr_CheckSignals()
-        pb(o_idx)
         
         merged_sequences = uniques[unique_id]
         
@@ -453,7 +468,11 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
             merged_dict[mkey] = {}
 
         for i_idx in merged_sequences:
-                        
+            PyErr_CheckSignals()
+            
+            if pb is not None:
+                pb(total_treated)
+ 
             i_id = i_id_col[i_idx]
             i_seq = view[i_idx]
 
@@ -504,7 +523,9 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
                 if key != COUNT_COLUMN and key != ID_COLUMN and key != NUC_SEQUENCE_COLUMN and key in o_seq and o_seq[key] != i_seq[key] \
                     and key not in merged_dict :
                     o_seq[key] = None
-
+            
+            total_treated += 1
+            
         # Write merged dicts
         for mkey in merged_dict: 
             if mkey in str_merged_cols:
@@ -526,7 +547,8 @@ cdef uniq_sequences(View_NUC_SEQS view, View_NUC_SEQS o_view, ProgressBar pb, di
         o_count_col[o_idx] = o_count
         o_idx += 1
     
-    pb(len(uniques), force=True)
+    if pb is not None:
+        pb(len(view), force=True)
     
     # Deletes quality columns if there is one because the matching between sequence and quality will be broken (quality set to NA when sequence not)
     if QUALITY_COLUMN in view:
@@ -574,8 +596,23 @@ def run(config):
     if output is None:
         raise Exception("Could not create output view")
     
+    i_dms = input[0]
     entries = input[1]
-    o_view = output[1]
+    o_dms = output[0]
+    output_0 = output[0]
+    
+    # If stdout output create a temporary view that will be exported and deleted.
+    if type(output_0)==BufferedWriter:
+        temporary_view_name = b"temp"
+        i=0
+        while temporary_view_name in i_dms:  # Making sure view name is unique in input DMS
+            temporary_view_name = temporary_view_name+b"_"+str2bytes(str(i))
+            i+=1
+        o_view_name = temporary_view_name
+        o_dms = i_dms
+        o_view = View_NUC_SEQS.new(i_dms, o_view_name)
+    else:
+        o_view = output[1]
     
     if 'taxoURI' in config['obi'] and config['obi']['taxoURI'] is not None:
         taxo_uri = open_uri(config['obi']['taxoURI'])
@@ -586,14 +623,19 @@ def run(config):
         taxo = None
 
     # Initialize the progress bar
-    pb = ProgressBar(len(entries), config, seconde=5)
+    if config['obi']['noprogressbar'] == False:
+        pb = ProgressBar(len(entries), config)
+    else:
+        pb = None
     
-    try:
-        uniq_sequences(entries, o_view, pb, config, mergedKeys_list=config['uniq']['merge'], taxonomy=taxo, mergeIds=config['uniq']['mergeids'], categories=config['uniq']['categories'], max_elts=config['obi']['maxelts'])       
-    except Exception, e:
-        raise RollbackException("obi uniq error, rollbacking view: "+str(e), o_view)
+    if len(entries) > 0:
+        try:
+            uniq_sequences(entries, o_view, pb, config, mergedKeys_list=config['uniq']['merge'], taxonomy=taxo, mergeIds=config['uniq']['mergeids'], categories=config['uniq']['categories'], max_elts=config['obi']['maxelts'])       
+        except Exception, e:
+            raise RollbackException("obi uniq error, rollbacking view: "+str(e), o_view)
     
-    print("", file=sys.stderr)
+    if pb is not None:
+        print("", file=sys.stderr)
 
     # Save command config in View and DMS comments
     command_line = " ".join(sys.argv[1:])
@@ -603,13 +645,23 @@ def run(config):
         input_dms_name.append(config['obi']['taxoURI'].split("/")[-3])
         input_view_name.append("taxonomy/"+config['obi']['taxoURI'].split("/")[-1])
     o_view.write_config(config, "uniq", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)
-    output[0].record_command_line(command_line)
+    o_dms.record_command_line(command_line)
+
+    # stdout output: write to buffer
+    if type(output_0)==BufferedWriter:
+        logger("info", "Printing to output...")
+        o_view.print_to_output(output_0, noprogressbar=config['obi']['noprogressbar'])
+        o_view.close()
 
     #print("\n\nOutput view:\n````````````", file=sys.stderr)
     #print(repr(o_view), file=sys.stderr)
-    
-    input[0].close(force=True)
-    output[0].close(force=True)
+
+    # If stdout output, delete the temporary result view in the input DMS
+    if type(output_0)==BufferedWriter:
+        View.delete_view(i_dms, o_view_name)
+
+    i_dms.close(force=True)
+    o_dms.close(force=True)
 
     logger("info", "Done.")
 

python/obitools3/dms/column/column.pxd

@@ -22,6 +22,7 @@ cdef class Column(OBIWrapper) :
     
     cdef inline OBIDMS_column_p pointer(self)
     cdef read_elements_names(self)
+    cpdef list keys(self)
     
     @staticmethod
     cdef type get_column_class(obitype_t obitype, bint multi_elts, bint tuples)

python/obitools3/dms/column/column.pyx

@@ -323,7 +323,10 @@ cdef class Column(OBIWrapper) :
         free(elts_names_b)
         return elts_names_list
     
-    
+    cpdef list keys(self):
+        return self._elements_names
+
+
     # Column alias property getter and setter
     @property
     def name(self):
@@ -340,7 +343,7 @@ cdef class Column(OBIWrapper) :
     @property
     def elements_names(self):
         return self._elements_names
- 
+       
     # nb_elements_per_line property getter
     @property
     def nb_elements_per_line(self):

python/obitools3/dms/dms.pyx

@@ -227,7 +227,9 @@ cdef class DMS(OBIWrapper):
         cdef str s 
         s=""
         for view_name in self.keys():
-            s = s + repr(self.get_view(view_name)) + "\n"
+            view = self.get_view(view_name)
+            s = s + repr(view) + "\n"
+            view.close()
         return s
     
 

python/obitools3/dms/obiseq.pxd

@@ -39,4 +39,6 @@ cdef class Nuc_Seq_Stored(Seq_Stored) :
     cpdef set_quality_char(self, object new_qual, int offset=*)
     cpdef object build_quality_array(self, list quality)
     cpdef bytes build_reverse_complement(self)
-    cpdef str get_str(self)
+    cpdef str get_str(self)
+    cpdef repr_bytes(self)
+    

python/obitools3/dms/obiseq.pyx

@@ -431,9 +431,12 @@ cdef class Nuc_Seq_Stored(Seq_Stored) :
         return len(self._view.get_column(NUC_SEQUENCE_COLUMN).get_line(self.index))
 
     def __repr__(self):
+        return bytes2str(self.repr_bytes())
+    
+    cpdef repr_bytes(self):
         if self.quality is None:
             formatter = FastaFormat()
         else:
             formatter = FastqFormat()
-        return bytes2str(formatter(self))
+        return formatter(self)
 

python/obitools3/dms/view/view.pxd

@@ -20,6 +20,10 @@ cdef class View(OBIWrapper):
     cdef DMS _dms
     
     cdef inline Obiview_p pointer(self)   
+
+    cpdef print_to_output(self, 
+                          object output, 
+                          bint noprogressbar=*)
         
     cpdef delete_column(self, 
                         object column_name,
@@ -61,6 +65,8 @@ cdef class Line :
     cdef index_t _index
     cdef View    _view
     
+    cpdef repr_bytes(self)
+
 
 cdef register_view_class(bytes view_type_name,
                          type view_class)

python/obitools3/dms/view/view.pyx

@@ -6,6 +6,8 @@ cdef dict __VIEW_CLASS__= {}
 
 from libc.stdlib  cimport malloc
 
+from obitools3.apps.progress cimport ProgressBar  # @UnresolvedImport
+
 from ..capi.obiview cimport Alias_column_pair_p, \
                             obi_new_view, \
                             obi_open_view, \
@@ -48,10 +50,13 @@ from ..capi.obidms cimport obi_import_view
 
 from obitools3.format.tab import TabFormat
 
+from cpython.exc cimport PyErr_CheckSignals
+
 import importlib
 import inspect
 import pkgutil
 import json
+import sys
 
 
 cdef class View(OBIWrapper) :
@@ -184,7 +189,31 @@ cdef class View(OBIWrapper) :
         for column_name in self.keys() :
             s = s + repr(self[column_name]) + '\n'
         return s
-    
+
+
+    cpdef print_to_output(self, object output, bint noprogressbar=False):      
+        
+        cdef int i
+        cdef Line entry
+        
+        self.checkIsActive(self)
+        
+        # Initialize the progress bar
+        if noprogressbar == False:
+            pb = ProgressBar(len(self))
+        else:
+            pb = None
+        i=0
+        for entry in self:
+            PyErr_CheckSignals()
+            if pb is not None:
+                pb(i)
+            output.write(entry.repr_bytes()+b"\n")
+            i+=1
+        if pb is not None:
+            pb(len(self), force=True)
+            print("", file=sys.stderr)
+
      
     def keys(self):
         
@@ -533,6 +562,7 @@ cdef class View(OBIWrapper) :
             for command in command_list:
                 s+=b"obi "
                 s+=command
+                s+=b"\n"
         return s
 
 
@@ -756,8 +786,12 @@ cdef class Line :
 
 
     def __repr__(self):
+        return bytes2str(self).repr_bytes()
+
+
+    cpdef repr_bytes(self):
         formatter = TabFormat(header=False)
-        return bytes2str(formatter(self))
+        return formatter(self)
 
 
     # View property getter

python/obitools3/format/tab.pyx

@@ -3,7 +3,7 @@
 cimport cython
 from obitools3.dms.view.view cimport Line
 from obitools3.utils cimport bytes2str_object, str2bytes, tobytes
-from obitools3.dms.column.column cimport Column_line
+from obitools3.dms.column.column cimport Column_line, Column_multi_elts
 
 
 cdef class TabFormat:
@@ -25,19 +25,29 @@ cdef class TabFormat:
         for k in self.tags:
             
             if self.header and self.first_line:
-                value = tobytes(k)
+                if isinstance(data.view[k], Column_multi_elts):
+                    for k2 in data.view[k].keys():
+                        line.append(tobytes(k)+b':'+tobytes(k2))
+                else:
+                    line.append(tobytes(k))
             else:
                 value = data[k]
-                if value is not None:
-                    if type(value) == Column_line:
-                        value = value.bytes()
+                if isinstance(data.view[k], Column_multi_elts):
+                    if value is None:  # all keys at None
+                        for k2 in data.view[k].keys(): # TODO could be much more efficient
+                            line.append(self.NAString)
                     else:
-                        value = str2bytes(str(bytes2str_object(value))) # genius programming
-                if value is None:
-                    value = self.NAString
-            
-            line.append(value)
-      	
+                        for k2 in data.view[k].keys(): # TODO could be much more efficient
+                            if value[k2] is not None:
+                                line.append(str2bytes(str(bytes2str_object(value[k2]))))  # genius programming
+                            else:
+                                line.append(self.NAString)
+                else:
+                    if value is not None:
+                        line.append(str2bytes(str(bytes2str_object(value))))
+                    else:
+                        line.append(self.NAString)
+                  	
         if self.first_line:
             self.first_line = False
       		

python/obitools3/libalign/_solexapairend.pyx

@@ -183,12 +183,13 @@ def buildConsensus(ali, seq, ref_tags=None):
         # doesn't work because uint8_t* are forced into bytes by cython (nothing read/kept beyond 0 values)
         #obi_set_qual_int_with_elt_idx_and_col_p_in_view(view_p, col_p, seq.index, 0, ali.consensus_qual, ali.consensus_len)
         seq.set(ref_tags.id+b"_CONS", ali.consensus_seq, quality=ali.consensus_qual)
-        seq[b'ali_length'] = ali.consensus_len
-        seq[b'score_norm']=float(ali.score)/ali.consensus_len
+        seq[b"seq_length"] = ali.consensus_len
+        seq[b"overlap_length"] = ali.overlap_len
+        seq[b'score_norm']=round(float(ali.score)/ali.overlap_len, 3)
         seq[b'shift']=ali.shift
     else:
         if len(ali[0])>999:   # TODO why?
-            raise AssertionError,"Too long alignemnt"
+            raise AssertionError,"Too long alignment"
     
         ic=IterOnConsensus(ali)
     
@@ -250,11 +251,22 @@ def buildJoinedSequence(ali, reverse, seq, forward=None):
     quality.extend(reverse.quality)
     seq.set(forward.id +b"_PairedEnd", s, definition=forward.definition, quality=quality)
     seq[b"score"]=ali.score
-    seq[b"ali_direction"]=ali.direction
+    if len(ali.direction) > 0:
+        seq[b"ali_direction"]=ali.direction
+    else:
+        seq[b"ali_direction"]=None
     seq[b"mode"]=b"joined"
-    seq[b"pairedend_limit"]=len(forward)    
+    seq[b"pairedend_limit"]=len(forward)
+    seq[b"seq_length"] = ali.consensus_len
+    seq[b"overlap_length"] = ali.overlap_len
+    if ali.consensus_len > 0:
+        seq[b'score_norm']=round(float(ali.score)/ali.overlap_len, 3)
+    else:
+        seq[b"score_norm"]=0.0
+  
     for tag in forward:
-        if tag != REVERSE_SEQUENCE_COLUMN and tag != REVERSE_QUALITY_COLUMN:
+        if tag != REVERSE_SEQUENCE_COLUMN and tag != REVERSE_QUALITY_COLUMN and \
+            tag != NUC_SEQUENCE_COLUMN and tag != QUALITY_COLUMN:
             seq[tag] = forward[tag]
     return seq
 

python/obitools3/uri/decode.pyx

@@ -210,10 +210,11 @@ def open_uri(uri,
     
     error = None
     
-    if scheme==b"dms" or \
-        (scheme==b"" and \
-         (((not input) and "outputformat" not in config["obi"]) or \
-         (input and "inputformat" not in config["obi"]))):   # TODO maybe not best way
+    if urib != b"-" and \
+        (scheme==b"dms" or \
+         (scheme==b"" and \
+          (((not input) and "outputformat" not in config["obi"]) or \
+          (input and "inputformat" not in config["obi"])))):   # TODO maybe not best way
         
         if default_dms is not None and b"/" not in urip.path:   # assuming view to open in default DMS (TODO discuss)
             dms=(default_dms, urip.path)
@@ -275,10 +276,10 @@ def open_uri(uri,
             iseq = urib
             objclass = bytes
     else:  # TODO update uopen to be able to write? 
-        if urip.path:
-            file = open(urip.path, 'wb')
-        else:
+        if urip.path == b'-':
             file = sys.stdout.buffer
+        elif urip.path :
+            file = open(urip.path, 'wb')
         
     if file is not None:
         qualifiers=parse_qs(urip.query)

python/obitools3/utils.pyx

@@ -166,7 +166,9 @@ cdef object bytes2str_object(object value):  # Only works if complex types are d
                     value[k] = bytes2str(v)
             if type(k) == bytes:
                 value[bytes2str(k)] = value.pop(k)
-    elif isinstance(value, list):
+    elif isinstance(value, list) or isinstance(value, tuple):
+        if isinstance(value, tuple):
+            value = list(value)
         for i in range(len(value)):
             if isinstance(value[i], list) or isinstance(value[i], dict):
                 value[i] = bytes2str_object(value[i])

python/obitools3/version.py

@@ -1,5 +1,5 @@
 major = 3
 minor = 0
-serial= '0b21'
+serial= '0b28'
 
 version ="%d.%d.%s" % (major,minor,serial)

setup.py

@@ -27,10 +27,11 @@ class Distribution(ori_Distribution):
         
         ori_Distribution.__init__(self, attrs)
         
-        self.global_options.insert(0,('cobitools3', None, "intall location of the C library"
+        self.global_options.insert(0,('cobitools3', None, "install location of the C library"
                                      ))
 
 from distutils.command.build import build as build_ori
+from setuptools.command.bdist_egg import bdist_egg as bdist_egg_ori
 from distutils.core import Command
 
 
@@ -71,6 +72,12 @@ class build(build_ori):
         build_ori.run(self)
 
 
+class bdist_egg(bdist_egg_ori):
+    def run(self):
+        self.run_command('build_clib')
+        bdist_egg_ori.run(self)
+
+
 sys.path.append(os.path.abspath("python"))
 
 
@@ -166,6 +173,7 @@ setup(name=PACKAGE,
       ext_modules=xx,
       distclass=Distribution,
       cmdclass={'build': build,
+                'bdist_egg': bdist_egg,
                 'build_clib': build_clib},
       cobitools3=get_python_lib(),
       packages = findPackage('python'),

src/kmer_similarity.c

@@ -413,7 +413,13 @@ Obi_ali_p kmer_similarity(Obiview_p view1, OBIDMS_column_p column1, index_t idx1
 	 	return NULL;
 	}
 
-	score = max_common_kmers + kmer_size - 1;  // aka the number of nucleotides in the longest stretch of kmers perfectly matching
+	if (max_common_kmers > 0)
+		score = max_common_kmers + kmer_size - 1; // aka an approximation of the number of nucleotides matching in the overlap of the alignment.
+												  // It's an approximation because one mismatch produces kmer_size kmer mismatches if in the middle of the overlap,
+												  // and less for mismatches located towards the ends of the overlap. The case where there are the most mismatches is assumed,
+												  // meaning that the score will be often underestimated and never overestimated.
+	else
+		score = 0;
 	abs_shift = abs(best_shift);
 
 	// Save result in Obi_ali structure
@@ -423,10 +429,15 @@ Obi_ali_p kmer_similarity(Obiview_p view1, OBIDMS_column_p column1, index_t idx1
 	ali->shift = abs_shift;
 	ali->consensus_seq = NULL;
 	ali->consensus_qual = NULL;
-	if (((best_shift <= 0) && (!switched_seqs)) || ((best_shift > 0) && switched_seqs))
-		strcpy(ali->direction, "left");
+	if (score == 0)
+		ali->direction[0] = '\0';
 	else
-		strcpy(ali->direction, "right");
+	{
+		if (((best_shift <= 0) && (!switched_seqs)) || ((best_shift > 0) && switched_seqs))
+			strcpy(ali->direction, "left");
+		else
+			strcpy(ali->direction, "right");
+	}
 
 	// Build the consensus sequence if asked
 	if (build_consensus)

src/kmer_similarity.h

@@ -27,7 +27,11 @@
  * @brief Alignment structure, with informations about the similarity and to rebuild the alignment.
  */
 typedef struct Obi_ali {
-	int      score;	    		/**< Alignment score, corresponding to the number of nucleotides in the longest stretch of kmers perfectly matching.
+	int      score;	    		/**< Alignment score, corresponding to an approximation of the number of
+								 *   nucleotides matching in the overlap of the alignment.
+								 *   It's an approximation because one mismatch produces kmer_size kmer mismatches if in the middle of the overlap,
+							     *   and less for mismatches located towards the ends of the overlap. The case where there are the most mismatches is assumed,
+							     *   meaning that the score will be often underestimated and never overestimated.
 	 	 	 	 	 	 	 	 */
 	int      consensus_length; 	/**< Length of the final consensus sequence.
 	 	 	 	 	 	 	 	 */

src/obi_clean.c

@@ -246,7 +246,16 @@ int obi_clean(const char* dms_name,
 
 	// Open the sample column if there is one
 	if ((strcmp(sample_column_name, "") == 0) || (sample_column_name == NULL))
-		sample_column = NULL;
+	{
+		fprintf(stderr, "Info: No sample information provided, assuming one sample.\n");
+		sample_column = obi_view_get_column(i_view, COUNT_COLUMN);
+		if (sample_column == NULL)
+		{
+			obidebug(1, "\nError getting the COUNT column");
+			return -1;
+		}
+		sample_count = 1;
+	}
 	else
 	{
 		sample_column = obi_view_get_column(i_view, sample_column_name);
@@ -255,6 +264,13 @@ int obi_clean(const char* dms_name,
 			obidebug(1, "\nError getting the sample column");
 			return -1;
 		}
+		sample_count = (sample_column->header)->nb_elements_per_line;
+		// Check that the sample column is a merged column with all sample informations
+		if (sample_count == 1)
+		{
+			obidebug(1, "\n\nError: If a sample column is provided, it must contain 'merged' sample counts as built by obi uniq with the -m option\n");
+			return -1;
+		}
 	}
 
 	// Create the output view, or a temporary one if heads only
@@ -279,8 +295,6 @@ int obi_clean(const char* dms_name,
 		return -1;
 	}
 
-	sample_count = (sample_column->header)->nb_elements_per_line;
-
 	// Create the output columns
 	if (create_output_columns(o_view, sample_column, sample_count) < 0)
 	{
@@ -549,7 +563,7 @@ int obi_clean(const char* dms_name,
 
 	if (heads_only)
 	{
-		line_selection = malloc((o_view->infos)->line_count * sizeof(index_t));
+		line_selection = malloc((((o_view->infos)->line_count) + 1) * sizeof(index_t));
 		if (line_selection == NULL)
 		{
 			obi_set_errno(OBI_MALLOC_ERROR);

src/obi_clean.h

@@ -52,7 +52,8 @@
  *
  * @param dms A pointer on an OBIDMS.
  * @param i_view_name The name of the input view.
- * @param sample_column_name The name of the OBI_STR column in the input view where the sample information is kept.
+ * @param sample_column_name The name of the column in the input view where the sample information is kept.
+ *                           Must be merged informations as built by the obi uniq tool (checked by the function).
  *                           NULL or "" (empty string) if there is no sample information.
  * @param o_view_name The name of the output view where the results should be written (should not already exist).
  * @param o_view_comments The comments that should be associated with the output view.

src/obi_ecotag.c

@@ -71,9 +71,12 @@ static int create_output_columns(Obiview_p o_view);
  * @param name The assigned scientific name.
  * @param assigned_status_column A pointer on the column where the assigned status should be written.
  * @param assigned The assigned status (whether the sequence was assigned to a taxon or not).
- * @param best_match_column A pointer on the column where the list of ids of the best matches should be written.
+ * @param best_match_ids_column A pointer on the column where the list of ids of the best matches should be written.
  * @param best_match_ids The list of ids of the best matches as an array of the concatenated ids separated by '\0'.
  * @param best_match_ids_length The total length of the array of ids of best matches.
+ * @param best_match_taxids_column A pointer on the column where the list of taxids of the best matches should be written.
+ * @param best_match_taxids The list of taxids of the best matches as an array of the taxids.
+ * @param best_match_taxids_length The length of the array of taxids of best matches.
  * @param score_column A pointer on the column where the score should be written.
  * @param score The similarity score of the sequence with its best match(es).
  *
@@ -87,7 +90,8 @@ int print_assignment_result(Obiview_p output_view, index_t line,
 							OBIDMS_column_p assigned_taxid_column, int32_t taxid,
 							OBIDMS_column_p assigned_name_column, const char* name,
 							OBIDMS_column_p assigned_status_column, bool assigned,
-							OBIDMS_column_p best_match_column, const char* best_match_ids, int best_match_ids_length,
+							OBIDMS_column_p best_match_ids_column, const char* best_match_ids, int best_match_ids_length,
+							OBIDMS_column_p best_match_taxids_column, const int32_t* best_match_taxids, int best_match_taxids_length,
 							OBIDMS_column_p score_column, double score);
 
 
@@ -130,7 +134,14 @@ static int create_output_columns(Obiview_p o_view)
 	// Column for array of best match ids
 	if (obi_view_add_column(o_view, ECOTAG_BEST_MATCH_IDS_COLUMN_NAME, -1, NULL, OBI_STR, 0, 1, NULL, false, true, false, NULL, NULL, -1, "{}", true) < 0)
 	{
-		obidebug(1, "\nError creating the column for the array of ids of the best match in ecotag");
+		obidebug(1, "\nError creating the column for the array of ids of best matches in ecotag");
+		return -1;
+	}
+
+	// Column for array of best match taxids
+	if (obi_view_add_column(o_view, ECOTAG_BEST_MATCH_TAXIDS_COLUMN_NAME, -1, NULL, OBI_INT, 0, 1, NULL, false, true, false, NULL, NULL, -1, "{}", true) < 0)
+	{
+		obidebug(1, "\nError creating the column for the array of taxids of best matches in ecotag");
 		return -1;
 	}
 
@@ -142,7 +153,8 @@ int print_assignment_result(Obiview_p output_view, index_t line,
 							OBIDMS_column_p assigned_taxid_column, int32_t taxid,
 							OBIDMS_column_p assigned_name_column, const char* name,
 							OBIDMS_column_p assigned_status_column, bool assigned,
-							OBIDMS_column_p best_match_column, const char* best_match_ids, int best_match_ids_length,
+							OBIDMS_column_p best_match_ids_column, const char* best_match_ids, int best_match_ids_length,
+							OBIDMS_column_p best_match_taxids_column, const int32_t* best_match_taxids, int best_match_taxids_length,
 							OBIDMS_column_p score_column, double score)
 {
 	// Write the assigned taxid
@@ -167,9 +179,16 @@ int print_assignment_result(Obiview_p output_view, index_t line,
 	}
 
 	// Write the best match ids
-	if (obi_set_array_with_col_p_in_view(output_view, best_match_column, line, best_match_ids, (uint8_t)(sizeof(char)*8), best_match_ids_length) < 0)
+	if (obi_set_array_with_col_p_in_view(output_view, best_match_ids_column, line, best_match_ids, (uint8_t)(sizeof(char)*8), best_match_ids_length) < 0)
 	{
-		obidebug(1, "\nError writing a assignment status in a column when writing ecotag results");
+		obidebug(1, "\nError writing the array of best match ids in a column when writing ecotag results");
+		return -1;
+	}
+
+	// Write the best match taxids
+	if (obi_set_array_with_col_p_in_view(output_view, best_match_taxids_column, line, best_match_taxids, (uint8_t)(sizeof(OBI_INT)*8), best_match_taxids_length) < 0)
+	{
+		obidebug(1, "\nError writing the array of best match taxids in a column when writing ecotag results");
 		return -1;
 	}
 
@@ -235,6 +254,8 @@ int obi_ecotag(const char* dms_name,
 	char*     		best_match_ids;
 	char*			best_match_ids_to_store;
 	int32_t         best_match_ids_length;
+	int32_t*   		best_match_taxids;
+	int32_t*		best_match_taxids_to_store;
 	int				best_match_count;
 	int             buffer_size;
 	int 			best_match_ids_buffer_size;
@@ -263,7 +284,8 @@ int obi_ecotag(const char* dms_name,
 	OBIDMS_column_p assigned_taxid_column = NULL;
 	OBIDMS_column_p assigned_name_column = NULL;
 	OBIDMS_column_p assigned_status_column = NULL;
-	OBIDMS_column_p best_match_column = NULL;
+	OBIDMS_column_p best_match_ids_column = NULL;
+	OBIDMS_column_p best_match_taxids_column = NULL;
 	OBIDMS_column_p lca_taxid_a_column = NULL;
 	OBIDMS_column_p score_a_column = NULL;
 	OBIDMS_column_p ref_taxid_column = NULL;
@@ -396,7 +418,8 @@ int obi_ecotag(const char* dms_name,
 	assigned_taxid_column = obi_view_get_column(output_view, ECOTAG_TAXID_COLUMN_NAME);
 	assigned_name_column = obi_view_get_column(output_view, ECOTAG_NAME_COLUMN_NAME);
 	assigned_status_column = obi_view_get_column(output_view, ECOTAG_STATUS_COLUMN_NAME);
-	best_match_column = obi_view_get_column(output_view, ECOTAG_BEST_MATCH_IDS_COLUMN_NAME);
+	best_match_ids_column = obi_view_get_column(output_view, ECOTAG_BEST_MATCH_IDS_COLUMN_NAME);
+	best_match_taxids_column = obi_view_get_column(output_view, ECOTAG_BEST_MATCH_TAXIDS_COLUMN_NAME);
 	score_column = obi_view_get_column(output_view, ECOTAG_SCORE_COLUMN_NAME);
 
 	// Open the used reference columns
@@ -453,6 +476,14 @@ int obi_ecotag(const char* dms_name,
 		return -1;
 	}
 
+	best_match_taxids = (int32_t*) malloc(buffer_size* sizeof(int32_t));
+	if (best_match_taxids == NULL)
+	{
+		obi_set_errno(OBI_MALLOC_ERROR);
+		obidebug(1, "\nError allocating memory for the best match taxid array in ecotag");
+		return -1;
+	}
+
 	for (i=0; i < query_count; i++)
 	{
 		if (i%1000 == 0)
@@ -514,7 +545,7 @@ int obi_ecotag(const char* dms_name,
 
 				// Store in best match array
 
-				// Grow match array if needed
+				// Grow match and taxid array if needed
 				if (best_match_count == buffer_size)
 				{
 					buffer_size = buffer_size*2;
@@ -525,6 +556,13 @@ int obi_ecotag(const char* dms_name,
 						obidebug(1, "\nError reallocating match array when assigning");
 						return -1;
 					}
+					best_match_taxids = (int32_t*) realloc(best_match_taxids, buffer_size*sizeof(int32_t));
+					if (best_match_taxids == NULL)
+					{
+						obi_set_errno(OBI_MALLOC_ERROR);
+						obidebug(1, "\nError reallocating match taxids array when assigning");
+						return -1;
+					}
 				}
 
 				id = obi_get_str_with_elt_idx_and_col_p_in_view(ref_view, ref_id_column, j, 0);
@@ -545,6 +583,7 @@ int obi_ecotag(const char* dms_name,
 
 				// Save match
 				best_match_array[best_match_count] = j;
+				best_match_taxids[best_match_count] = obi_get_int_with_elt_idx_and_col_p_in_view(ref_view, ref_taxid_column, j, 0);
 				best_match_count++;
 				strcpy(best_match_ids+best_match_ids_length, id);
 				best_match_ids_length = best_match_ids_length + id_len + 1;
@@ -629,6 +668,7 @@ int obi_ecotag(const char* dms_name,
 			else
 				lca_name = lca->name;
 			best_match_ids_to_store = best_match_ids;
+			best_match_taxids_to_store = best_match_taxids;
 		}
 		else
 		{
@@ -636,6 +676,7 @@ int obi_ecotag(const char* dms_name,
 			lca_name = OBIStr_NA;
 			lca_taxid = OBIInt_NA;
 			best_match_ids_to_store = OBITuple_NA;
+			best_match_taxids_to_store = OBITuple_NA;
 			score = OBIFloat_NA;
 		}
 
@@ -644,7 +685,8 @@ int obi_ecotag(const char* dms_name,
 								    assigned_taxid_column, lca_taxid,
 									assigned_name_column, lca_name,
 									assigned_status_column, assigned,
-									best_match_column, best_match_ids_to_store, best_match_ids_length,
+									best_match_ids_column, best_match_ids_to_store, best_match_ids_length,
+									best_match_taxids_column, best_match_taxids_to_store, best_match_count,
 									score_column, best_score
 									) < 0)
 							return -1;
@@ -652,6 +694,7 @@ int obi_ecotag(const char* dms_name,
 
 	free(best_match_array);
 	free(best_match_ids);
+	free(best_match_taxids);
 
 	obi_close_taxonomy(taxonomy);
 	obi_save_and_close_view(query_view);

src/obi_ecotag.h

@@ -23,7 +23,8 @@
 #define ECOTAG_TAXID_COLUMN_NAME "TAXID"
 #define ECOTAG_NAME_COLUMN_NAME "SCIENTIFIC_NAME"
 #define ECOTAG_STATUS_COLUMN_NAME "ID_STATUS"
-#define ECOTAG_BEST_MATCH_IDS_COLUMN_NAME "BEST_MATCH"
+#define ECOTAG_BEST_MATCH_IDS_COLUMN_NAME "BEST_MATCH_IDS"
+#define ECOTAG_BEST_MATCH_TAXIDS_COLUMN_NAME "BEST_MATCH_TAXIDS"
 #define ECOTAG_SCORE_COLUMN_NAME "BEST_IDENTITY"
 
 

src/obiavl.c

@@ -2259,7 +2259,13 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_blob_p value)
 		parent = next;
 
 		// Compare the crc of the value with the crc of the current node
-		comp = (current_node->crc64) - crc;
+		//comp = (current_node->crc64) - crc;
+		if ((current_node->crc64) == crc)
+			comp = 0;
+		else if ((current_node->crc64) > crc)
+			comp = 1;
+		else
+			comp = -1;
 
 		if (comp == 0)
 		{ // check if really same value
@@ -2354,7 +2360,13 @@ index_t obi_avl_find(OBIDMS_avl_p avl, Obi_blob_p value)
 		current_node = (avl->tree)+next;
 
 		// Compare the crc of the value with the crc of the current node
-		comp = (current_node->crc64) - crc;
+		//comp = (current_node->crc64) - crc;
+		if ((current_node->crc64) == crc)
+			comp = 0;
+		else if ((current_node->crc64) > crc)
+			comp = 1;
+		else
+			comp = -1;
 
 		if (comp == 0)
 		{ // Check if really same value

src/obidms.c

@@ -1496,7 +1496,7 @@ obiversion_t obi_import_column(const char* dms_path_1, const char* dms_path_2, c
 	memcpy(column_2->data, column_1->data, header_1->data_size);
 
 	// Copy the AVL files if there are some (overwriting the automatically created files)
-	if ((header_1->returned_data_type == OBI_STR) || (header_1->returned_data_type == OBI_SEQ) || (header_1->returned_data_type == OBI_QUAL))
+	if ((header_1->tuples) || ((header_1->returned_data_type == OBI_STR) || (header_1->returned_data_type == OBI_SEQ) || (header_1->returned_data_type == OBI_QUAL)))
 	{
 		avl_name_1 = (char*) malloc((strlen(header_1->indexer_name) + 1)  * sizeof(char));
 		if (avl_name_1 == NULL)

src/obidmscolumn.c

@@ -1350,6 +1350,8 @@ OBIDMS_column_p obi_create_column(OBIDMS_p     dms,
 		}
 		strncpy(header->indexer_name, final_indexer_name, INDEXER_MAX_NAME);
 	}
+	else
+		new_column->indexer = NULL;
 
 	// Fill the data with NA values
 	obi_ini_to_NA_values(new_column, 0, nb_lines);
@@ -1558,6 +1560,8 @@ OBIDMS_column_p obi_open_column(OBIDMS_p     dms,
 			return NULL;
 		}
 	}
+	else
+		column->indexer = NULL;
 
 	if (close(column_file_descriptor) < 0)
 	{
@@ -1693,8 +1697,8 @@ int obi_close_column(OBIDMS_column_p column)
 		if (obi_dms_unlist_column(column->dms, column) < 0)
 			ret_val = -1;
 
-		// If the data type is OBI_STR, OBI_SEQ or OBI_QUAL, the associated indexer is closed
-		if (((column->header)->returned_data_type == OBI_STR) || ((column->header)->returned_data_type == OBI_SEQ) || ((column->header)->returned_data_type == OBI_QUAL))
+		// If it's a tuple column or the data type is OBI_STR, OBI_SEQ or OBI_QUAL, the associated indexer is closed
+		if ((column->indexer) != NULL)
 			if (obi_close_indexer(column->indexer) < 0)
 				ret_val = -1;