doc/tutorials/dataimport/diseasome_import.py
author Julien Cristau <julien.cristau@logilab.fr>
Thu, 03 Jul 2014 12:18:34 +0200
changeset 9867 7f36f31d8828
parent 8927 885dea8f16a0
child 10662 10942ed172de
permissions -rw-r--r--
Added tag cubicweb-version-3.19.2, cubicweb-debian-version-3.19.2-1, cubicweb-centos-version-3.19.2-1 for changeset 8ac2202866e7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
8836
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     2
# copyright 2012 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     3
# contact http://www.logilab.fr -- mailto:contact@logilab.fr
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     4
#
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     5
# This program is free software: you can redistribute it and/or modify it under
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     6
# the terms of the GNU Lesser General Public License as published by the Free
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     7
# Software Foundation, either version 2.1 of the License, or (at your option)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     8
# any later version.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
     9
#
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    10
# This program is distributed in the hope that it will be useful, but WITHOUT
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    11
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    12
# FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    13
# details.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    14
#
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    15
# You should have received a copy of the GNU Lesser General Public License along
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    16
# with this program. If not, see <http://www.gnu.org/licenses/>.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    17
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    18
"""This module imports the Diseasome data into a CubicWeb instance.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    19
"""
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    20
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    21
# Python imports
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    22
import sys
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    23
import argparse
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    24
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    25
# Logilab import, for timing
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    26
from logilab.common.decorators import timed
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    27
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    28
# CubicWeb imports
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    29
import cubicweb.dataimport as cwdi
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    30
from cubes.dataio import dataimport as mcwdi
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    31
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    32
# Diseasome parser import
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    33
import diseasome_parser as parser
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    34
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    35
def _is_of_class(instance, class_name):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    36
    """Helper function to determine whether an instance is
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    37
    of a specified class or not.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    38
    Returns a True if this is the case and False otherwise.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    39
    """
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    40
    if instance.__class__.__name__ == class_name:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    41
        return True
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    42
    else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    43
        return False
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    44
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    45
@timed
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    46
def diseasome_import(session, file_name, store):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    47
    """Main function for importing Diseasome data.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    48
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    49
    It uses the Diseasome data parser to get the contents of the
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    50
    data from a file, then uses a store for importing the data
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    51
    into a CubicWeb instance.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    52
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    53
    >>> diseasome_import(session, 'file_name', Store)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    54
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    55
    """
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    56
    exturis = dict(session.execute('Any U, X WHERE X is ExternalUri, X uri U'))
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    57
    uri_to_eid = {}
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    58
    uri_to_etype = {}
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    59
    all_relations = {}
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    60
    etypes = {('http://www4.wiwiss.fu-berlin.de/'
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    61
               'diseasome/resource/diseasome/genes'): 'Gene',
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    62
              ('http://www4.wiwiss.fu-berlin.de/'
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    63
               'diseasome/resource/diseasome/diseases'): 'Disease'}
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    64
    # Read the parsed data
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    65
    for entity, relations in parser.entities_from_rdf(file_name, 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    66
                                                      ('gene', 'disease')):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    67
        uri = entity.get('cwuri', None)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    68
        types = list(relations.get('types', []))
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    69
        if not types:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    70
            continue
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    71
        etype = etypes.get(types[0])
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    72
        if not etype:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    73
            sys.stderr.write('Entity type %s not recognized.', types[0])
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    74
            sys.stderr.flush()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    75
        if _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    76
            for relation in (set(relations).intersection(('classes', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    77
                            'possible_drugs', 'omim', 'omim_page', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    78
                            'chromosomal_location', 'same_as', 'gene_id',
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    79
                            'hgnc_id', 'hgnc_page'))):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    80
                store.init_rtype_table(etype, relation, 'ExternalUri')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    81
            for relation in set(relations).intersection(('subtype_of',)):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    82
                store.init_rtype_table(etype, relation, 'Disease')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    83
            for relation in set(relations).intersection(('associated_genes',)):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    84
                store.init_rtype_table(etype, relation, 'Gene')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    85
        # Create the entities
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    86
        ent = store.create_entity(etype, **entity)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    87
        if not _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    88
            uri_to_eid[uri] = ent.eid
8927
885dea8f16a0 [cubicweb/doc] Replace dc_type() by cw_etype
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8836
diff changeset
    89
            uri_to_etype[uri] = ent.cw_etype
8836
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    90
        else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    91
            uri_to_eid[uri] = uri
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    92
            uri_to_etype[uri] = etype
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    93
        # Store relations for after
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    94
        all_relations[uri] = relations
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    95
    # Perform a first commit, of the entities
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    96
    store.flush()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    97
    kwargs = {}
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    98
    for uri, relations in all_relations.iteritems():
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
    99
        from_eid = uri_to_eid.get(uri)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   100
        # ``subjtype`` should be initialized if ``SQLGenObjectStore`` is used
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   101
        # and there are inlined relations in the schema.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   102
        # If ``subjtype`` is not given, while ``SQLGenObjectStore`` is used
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   103
        # and there are inlined relations in the schema, the store
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   104
        # tries to infer the type of the subject, but this does not always 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   105
        # work, e.g. when there are several object types for the relation.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   106
        # ``subjtype`` is ignored for other stores, or if there are no
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   107
        # inlined relations in the schema.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   108
        kwargs['subjtype'] = uri_to_etype.get(uri)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   109
        if not from_eid:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   110
            continue
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   111
        for rtype, rels in relations.iteritems():
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   112
            if rtype in ('classes', 'possible_drugs', 'omim', 'omim_page',
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   113
                         'chromosomal_location', 'same_as', 'gene_id',
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   114
                         'hgnc_id', 'hgnc_page'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   115
                for rel in list(rels):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   116
                    if rel not in exturis:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   117
                        # Create the "ExternalUri" entities, which are the
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   118
                        # objects of the relations
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   119
                        extu = store.create_entity('ExternalUri', uri=rel)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   120
                        if not _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   121
                            rel_eid = extu.eid
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   122
                        else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   123
                            # For the "MassiveObjectStore", the EIDs are 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   124
                            # in fact the URIs.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   125
                            rel_eid = rel
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   126
                        exturis[rel] = rel_eid
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   127
                    else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   128
                        rel_eid = exturis[rel]
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   129
                    # Create the relations that have "ExternalUri"s as objects
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   130
                    if not _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   131
                        store.relate(from_eid, rtype, rel_eid, **kwargs)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   132
                    else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   133
                        store.relate_by_iid(from_eid, rtype, rel_eid)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   134
            elif rtype in ('subtype_of', 'associated_genes'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   135
                for rel in list(rels):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   136
                    to_eid = uri_to_eid.get(rel)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   137
                    if to_eid:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   138
                        # Create relations that have objects of other type 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   139
                        # than "ExternalUri"
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   140
                        if not _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   141
                            store.relate(from_eid, rtype, to_eid, **kwargs)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   142
                        else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   143
                            store.relate_by_iid(from_eid, rtype, to_eid)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   144
                    else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   145
                        sys.stderr.write('Missing entity with URI %s '
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   146
                                         'for relation %s' % (rel, rtype))
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   147
                        sys.stderr.flush()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   148
    # Perform a second commit, of the "ExternalUri" entities.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   149
    # when the stores in the CubicWeb ``dataimport`` module are used,
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   150
    # relations are also committed.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   151
    store.flush()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   152
    # If the ``MassiveObjectStore`` is used, then entity and relation metadata
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   153
    # are pushed as well. By metadata we mean information on the creation
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   154
    # time and author.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   155
    if _is_of_class(store, 'MassiveObjectStore'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   156
        store.flush_meta_data()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   157
        for relation in ('classes', 'possible_drugs', 'omim', 'omim_page', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   158
                         'chromosomal_location', 'same_as'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   159
            # Afterwards, relations are actually created in the database.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   160
            store.convert_relations('Disease', relation, 'ExternalUri',
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   161
                                    'cwuri', 'uri')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   162
        store.convert_relations('Disease', 'subtype_of', 'Disease', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   163
                                'cwuri', 'cwuri')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   164
        store.convert_relations('Disease', 'associated_genes', 'Gene', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   165
                                'cwuri', 'cwuri')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   166
        for relation in ('gene_id', 'hgnc_id', 'hgnc_page', 'same_as'):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   167
            store.convert_relations('Gene', relation, 'ExternalUri', 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   168
                                    'cwuri', 'uri')
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   169
        # Clean up temporary tables in the database
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   170
        store.cleanup()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   171
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   172
if __name__ == '__main__':
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   173
    # Change sys.argv so that ``cubicweb-ctl shell`` can work out the options
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   174
    # we give to our ``diseasome_import.py`` script.
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   175
    sys.argv = [arg for 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   176
                arg in sys.argv[sys.argv.index("--") - 1:] if arg != "--"]
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   177
    PARSER = argparse.ArgumentParser(description="Import Diseasome data")
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   178
    PARSER.add_argument("-df", "--datafile", type=str,
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   179
                        help="RDF data file name")
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   180
    PARSER.add_argument("-st", "--store", type=str,
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   181
                        default="RQLObjectStore",
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   182
                        help="data import store")
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   183
    ARGS = PARSER.parse_args()
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   184
    if ARGS.datafile:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   185
        FILENAME = ARGS.datafile
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   186
        if ARGS.store in (st + "ObjectStore" for 
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   187
                          st in ("RQL", "NoHookRQL", "SQLGen")):
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   188
            IMPORT_STORE = getattr(cwdi, ARGS.store)(session)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   189
        elif ARGS.store == "MassiveObjectStore":
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   190
            IMPORT_STORE = mcwdi.MassiveObjectStore(session)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   191
        else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   192
            sys.exit("Import store unknown")
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   193
        diseasome_import(session, FILENAME, IMPORT_STORE)
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   194
    else:
8a57802d40d3 [cubicweb/doc] Add tutorial on data import in CubicWeb.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents:
diff changeset
   195
        sys.exit("Data file not found or not specified")