Ability to map identifiers with cadbiom names

command_line/cadbiom_cmd/cadbiom_cmd.py

@@ -147,6 +147,20 @@ def solutions_2_occcurrences_matrix(args):
         transposed=params['transpose_csv']
     )
 
+
+def identifiers_mapping(args):
+    """Mapping of identifiers from external databases.
+
+    This function exports a CSV formated file presenting the list of known
+    Cadbiom identifiers for each given external identifier.
+    """
+
+    # Module import
+    import solution_repr
+    params = args_to_param(args)
+    solution_repr.identifiers_mapping(**params)
+
+
 def model_comparison(args):
     """Isomorphism test.
 
@@ -437,6 +451,25 @@ def main():
 
     ## Model-related commands ##################################################
 
+    # subparser: Mapping of identifiers
+    # output: CSV file
+    parser_identifiers_mapping = subparsers.add_parser(
+        'identifiers_mapping',
+        help=identifiers_mapping.__doc__,
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter
+    )
+    parser_identifiers_mapping.add_argument('model_file',
+        help="bcx model file.")
+    group = parser_identifiers_mapping.add_mutually_exclusive_group(required=True)
+    group.add_argument('--external_file',
+        help="File with 1 external identifiers to be mapped per line."
+    )
+    group.add_argument('--external_identifiers', nargs='+',
+        help="Multiple external identifiers to be mapped."
+    )
+    parser_identifiers_mapping.set_defaults(func=identifiers_mapping)
+
+
     # subparser: Model comparison
     # 2 models
     parser_model_comparison = subparsers.add_parser(

command_line/cadbiom_cmd/solution_repr.py

@@ -54,7 +54,8 @@ from tools.solutions import load_solutions, convert_solutions_to_json
 from tools.models import Reporter
 from tools.models import get_transitions, \
                          get_transitions_from_model_file, \
-                         get_frontier_places
+                         get_frontier_places, \
+                         get_identifiers_mapping
 from tools.models import get_places_data
 from tools.graphs import build_graph, get_json_graph, export_graph
 
@@ -553,6 +554,34 @@ def low_model_info(model_file,
     return info
 
 
+def identifiers_mapping(model_file, *args, **kwargs):
+    """Entry point for the mapping of identifiers from external databases
+
+    :param model_file: File for the model.
+    :key external_file: File with 1 external identifier per line.
+    :key external_identifiers: List of external identifiers to be mapped.
+    :type model_file: <str>
+    :type external_file: <str>
+    :type external_identifiers: <list>
+    """
+    if kwargs['external_csv_file']:
+        with open(kwargs['external_file'], 'r') as f_d:
+            external_identifiers = set(line.strip('\n') for line in f_d)
+    else:
+        external_identifiers = set(kwargs['external_identifiers'])
+
+    mapping = get_identifiers_mapping(model_file, external_identifiers)
+
+    # Make CSV file
+    with open("mapping.csv", 'w') as csvfile:
+        writer = csv.writer(csvfile, delimiter=str(';'))
+
+        # Join multiple Cadbiom names with a |
+        g = ((external_id, "|".join(cadbiom_names))
+             for external_id, cadbiom_names in mapping.iteritems())
+        writer.writerows(g)
+
+
 def model_graph(model_file, output_dir='./graphs/',
                 centralities=False,
                 **kwargs):

command_line/cadbiom_cmd/tools/models.py

@@ -547,3 +547,45 @@ def get_places_data(places, model):
 
     # v1 model: return only the name of the place
     return [{'cadbiomName': place_name} for place_name in places]
+
+
+def get_identifiers_mapping(model_file, external_identifiers):
+    """Get Cadbiom names corresponding to the given external identifiers (xrefs)
+
+    .. note:: This function works only on v2 formated models with JSON additional data
+
+    :param model_file: Model file.
+    :param external_identifiers: Set of external identifiers to be mapped.
+    :type model_file: <str>
+    :type external_identifiers: <set>
+    :return: Mapping dictionary with external identifiers as keys
+        and cadbiom names as values.
+    :rtype: <dict <str>:<list>>
+    """
+    # Get the model
+    parser = MakeModelFromXmlFile(model_file)
+    model = parser.handler.model
+
+    assert model.xml_namespace == 'http://cadbiom.genouest.org/v2/', \
+        "Operation not supported: Only v2 models are supported."
+
+    # Get all nodes
+    places_data = get_places_data(parser.handler.node_dict.iterkeys(), model)
+
+    # {'xrefs': {'bdd': [values],}, 'cadbiomName': '',}
+    g = {place['cadbiomName']: frozenset(it.chain(*place['xrefs'].itervalues()))
+         for place in places_data}
+
+    # Mapping: external_identifiers as keys and Cadbiom names as values
+    mapping = defaultdict(set)
+    for place, identifiers in g.iteritems():
+
+        common_identifiers = identifiers & external_identifiers
+        if common_identifiers:
+            [mapping[common_id].add(place) for common_id in common_identifiers]
+
+    not_found_identifiers = external_identifiers - set(mapping.keys())
+    if not_found_identifiers:
+        LOGGER.info("Some identifiers were not found: %s", not_found_identifiers)
+
+    return mapping