dataimport.py
author Pierre-Yves David <pierre-yves.david@logilab.fr>
Thu, 27 Jun 2013 18:14:35 +0200
changeset 9119 cc85def2ac57
parent 8970 0a1bd0c590e2
child 9181 2eac0aa1d3f6
permissions -rw-r--r--
[doc] add documentation for the new API in test
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
     2
# copyright 2003-2012 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
5421
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     3
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     4
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     5
# This file is part of CubicWeb.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     6
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     7
# CubicWeb is free software: you can redistribute it and/or modify it under the
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     8
# terms of the GNU Lesser General Public License as published by the Free
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
     9
# Software Foundation, either version 2.1 of the License, or (at your option)
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    10
# any later version.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    11
#
5424
8ecbcbff9777 replace logilab-common by CubicWeb in disclaimer
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5421
diff changeset
    12
# CubicWeb is distributed in the hope that it will be useful, but WITHOUT
5421
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    13
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    14
# FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    15
# details.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    16
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    17
# You should have received a copy of the GNU Lesser General Public License along
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5097
diff changeset
    18
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    19
"""This module provides tools to import tabular data.
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    20
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    21
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    22
Example of use (run this with `cubicweb-ctl shell instance import-script.py`):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    23
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    24
.. sourcecode:: python
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    25
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    26
  from cubicweb.dataimport import *
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    27
  # define data generators
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    28
  GENERATORS = []
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    29
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    30
  USERS = [('Prenom', 'firstname', ()),
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    31
           ('Nom', 'surname', ()),
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    32
           ('Identifiant', 'login', ()),
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    33
           ]
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    34
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    35
  def gen_users(ctl):
6133
6f3eabbbdf2e use iter_and_commit in example
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 6122
diff changeset
    36
      for row in ctl.iter_and_commit('utilisateurs'):
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    37
          entity = mk_entity(row, USERS)
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    38
          entity['upassword'] = 'motdepasse'
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    39
          ctl.check('login', entity['login'], None)
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    40
          entity = ctl.store.create_entity('CWUser', **entity)
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    41
          email = ctl.store.create_entity('EmailAddress', address=row['email'])
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    42
          ctl.store.relate(entity.eid, 'use_email', email.eid)
3003
2944ee420dca R [dataimport] rename uid to eid
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 2974
diff changeset
    43
          ctl.store.rql('SET U in_group G WHERE G name "users", U eid %(x)s', {'x':entity['eid']})
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    44
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    45
  CHK = [('login', check_doubles, 'Utilisateurs Login',
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    46
          'Deux utilisateurs ne devraient pas avoir le même login.'),
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    47
         ]
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    48
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    49
  GENERATORS.append( (gen_users, CHK) )
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    50
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    51
  # create controller
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    52
  if 'cnx' in globals():
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    53
      ctl = CWImportController(RQLObjectStore(cnx))
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    54
  else:
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    55
      print 'debug mode (not connected)'
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    56
      print 'run through cubicweb-ctl shell to access an instance'
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    57
      ctl = CWImportController(ObjectStore())
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
    58
  ctl.askerror = 1
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    59
  ctl.generators = GENERATORS
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    60
  ctl.data['utilisateurs'] = lazytable(ucsvreader(open('users.csv')))
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    61
  # run
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    62
  ctl.run()
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    63
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
    64
.. BUG file with one column are not parsable
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
    65
.. TODO rollback() invocation is not possible yet
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    66
"""
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    67
__docformat__ = "restructuredtext en"
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    68
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    69
import csv
4186
ca7e526b07b6 import cleanup, check data file exists
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4173
diff changeset
    70
import sys
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    71
import threading
4186
ca7e526b07b6 import cleanup, check data file exists
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4173
diff changeset
    72
import traceback
8926
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
    73
import warnings
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    74
import cPickle
4186
ca7e526b07b6 import cleanup, check data file exists
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4173
diff changeset
    75
import os.path as osp
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
    76
import inspect
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    77
from collections import defaultdict
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
    78
from copy import copy
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    79
from datetime import date, datetime
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    80
from time import asctime
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    81
from StringIO import StringIO
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    82
7159
3bcccd3ab6b6 [dataimport] ObjectStore.create_entity should return something that looks like an entity (eg no more using dict protocol to access to attributes)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7158
diff changeset
    83
from logilab.common import shellutils, attrdict
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
    84
from logilab.common.date import strptime
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
    85
from logilab.common.decorators import cached
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
    86
from logilab.common.deprecation import deprecated
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
    87
7171
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
    88
from cubicweb import QueryError
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    89
from cubicweb.utils import make_uid
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    90
from cubicweb.schema import META_RTYPES, VIRTUAL_RTYPES
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    91
from cubicweb.server.edition import EditedEntity
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
    92
from cubicweb.server.sqlutils import SQL_PREFIX
5066
bf5cbc351e99 [repo] move eschema_eid function from hooks.metadata to server.utils
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5063
diff changeset
    93
from cubicweb.server.utils import eschema_eid
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
    94
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
    95
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    96
def count_lines(stream_or_filename):
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    97
    if isinstance(stream_or_filename, basestring):
6492
47a284c0d012 fix some pylint detected errors
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6427
diff changeset
    98
        f = open(stream_or_filename)
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
    99
    else:
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   100
        f = stream_or_filename
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   101
        f.seek(0)
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   102
    for i, line in enumerate(f):
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   103
        pass
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   104
    f.seek(0)
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   105
    return i+1
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   106
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   107
def ucsvreader_pb(stream_or_path, encoding='utf-8', separator=',', quote='"',
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   108
                  skipfirst=False, withpb=True):
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   109
    """same as ucsvreader but a progress bar is displayed as we iter on rows"""
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   110
    if isinstance(stream_or_path, basestring):
6492
47a284c0d012 fix some pylint detected errors
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6427
diff changeset
   111
        if not osp.exists(stream_or_path):
47a284c0d012 fix some pylint detected errors
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6427
diff changeset
   112
            raise Exception("file doesn't exists: %s" % stream_or_path)
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   113
        stream = open(stream_or_path)
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   114
    else:
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   115
        stream = stream_or_path
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   116
    rowcount = count_lines(stream)
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   117
    if skipfirst:
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   118
        rowcount -= 1
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   119
    if withpb:
4140
46ddd27a4ca4 tweaks output
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4136
diff changeset
   120
        pb = shellutils.ProgressBar(rowcount, 50)
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   121
    for urow in ucsvreader(stream, encoding, separator, quote, skipfirst):
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   122
        yield urow
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   123
        if withpb:
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   124
            pb.update()
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   125
    print ' %s rows imported' % rowcount
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   126
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   127
def ucsvreader(stream, encoding='utf-8', separator=',', quote='"',
8637
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   128
               skipfirst=False, ignore_errors=False):
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   129
    """A csv reader that accepts files with any encoding and outputs unicode
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   130
    strings
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   131
    """
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   132
    it = iter(csv.reader(stream, delimiter=separator, quotechar=quote))
8637
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   133
    if not ignore_errors:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   134
        if skipfirst:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   135
            it.next()
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   136
        for row in it:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   137
            yield [item.decode(encoding) for item in row]
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   138
    else:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   139
        # Skip first line
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   140
        try:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   141
            row = it.next()
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   142
        except csv.Error:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   143
            pass
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   144
        # Safe version, that can cope with error in CSV file
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   145
        while True:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   146
            try:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   147
                row = it.next()
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   148
            # End of CSV, break
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   149
            except StopIteration:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   150
                break
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   151
            # Error in CSV, ignore line and continue
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   152
            except csv.Error:
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   153
                continue
e16561083d84 [dataimport] new ignore_errors argument to ucsvreader, default to False. Closes #2547200
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8631
diff changeset
   154
            yield [item.decode(encoding) for item in row]
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   155
6136
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   156
def callfunc_every(func, number, iterable):
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   157
    """yield items of `iterable` one by one and call function `func`
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   158
    every `number` iterations. Always call function `func` at the end.
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   159
    """
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   160
    for idx, item in enumerate(iterable):
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   161
        yield item
7227
23d9c1f89c96 [dataimport] actually commit every desired number...
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7214
diff changeset
   162
        if not idx % number:
6136
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   163
            func()
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   164
    func()
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   165
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   166
def lazytable(reader):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   167
    """The first row is taken to be the header of the table and
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   168
    used to output a dict for each row of data.
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   169
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   170
    >>> data = lazytable(ucsvreader(open(filename)))
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   171
    """
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   172
    header = reader.next()
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   173
    for row in reader:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   174
        yield dict(zip(header, row))
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   175
7201
52f5831400b2 [dataimport] allow to specify columns on which result should be sorted in lazydbtable
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7171
diff changeset
   176
def lazydbtable(cu, table, headers, orderby=None):
7160
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   177
    """return an iterator on rows of a sql table. On each row, fetch columns
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   178
    defined in headers and return values as a dictionary.
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   179
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   180
    >>> data = lazydbtable(cu, 'experimentation', ('id', 'nickname', 'gps'))
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   181
    """
7201
52f5831400b2 [dataimport] allow to specify columns on which result should be sorted in lazydbtable
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7171
diff changeset
   182
    sql = 'SELECT %s FROM %s' % (','.join(headers), table,)
52f5831400b2 [dataimport] allow to specify columns on which result should be sorted in lazydbtable
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7171
diff changeset
   183
    if orderby:
52f5831400b2 [dataimport] allow to specify columns on which result should be sorted in lazydbtable
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7171
diff changeset
   184
        sql += ' ORDER BY %s' % ','.join(orderby)
52f5831400b2 [dataimport] allow to specify columns on which result should be sorted in lazydbtable
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7171
diff changeset
   185
    cu.execute(sql)
7160
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   186
    while True:
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   187
        row = cu.fetchone()
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   188
        if row is None:
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   189
            break
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   190
        yield dict(zip(headers, row))
923013173031 [dataimport] new 'lazydbtable' generator function to feed data from a database table
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7159
diff changeset
   191
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   192
def mk_entity(row, map):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   193
    """Return a dict made from sanitized mapped values.
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   194
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   195
    ValueError can be raised on unexpected values found in checkers
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   196
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   197
    >>> row = {'myname': u'dupont'}
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   198
    >>> map = [('myname', u'name', (call_transform_method('title'),))]
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   199
    >>> mk_entity(row, map)
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   200
    {'name': u'Dupont'}
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   201
    >>> row = {'myname': u'dupont', 'optname': u''}
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   202
    >>> map = [('myname', u'name', (call_transform_method('title'),)),
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   203
    ...        ('optname', u'MARKER', (optional,))]
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   204
    >>> mk_entity(row, map)
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   205
    {'name': u'Dupont', 'optname': None}
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   206
    """
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   207
    res = {}
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   208
    assert isinstance(row, dict)
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   209
    assert isinstance(map, list)
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   210
    for src, dest, funcs in map:
8406
f3bc8ca0b715 [data import] don't crash if value isn't in the file, simply no key/value in the output dict. Closes #2356328
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8403
diff changeset
   211
        try:
f3bc8ca0b715 [data import] don't crash if value isn't in the file, simply no key/value in the output dict. Closes #2356328
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8403
diff changeset
   212
            res[dest] = row[src]
f3bc8ca0b715 [data import] don't crash if value isn't in the file, simply no key/value in the output dict. Closes #2356328
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8403
diff changeset
   213
        except KeyError:
f3bc8ca0b715 [data import] don't crash if value isn't in the file, simply no key/value in the output dict. Closes #2356328
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8403
diff changeset
   214
            continue
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   215
        try:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   216
            for func in funcs:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   217
                res[dest] = func(res[dest])
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   218
                if res[dest] is None:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   219
                    break
8695
358d8bed9626 [toward-py3k] rewrite to "except AnException as exc:" (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8637
diff changeset
   220
        except ValueError as err:
7170
32b5d9d43a7e [dataimport] propagate stack
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7160
diff changeset
   221
            raise ValueError('error with %r field: %s' % (src, err)), None, sys.exc_info()[-1]
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   222
    return res
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   223
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   224
# user interactions ############################################################
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   225
3029
bc573d5fb5b7 F [devtools] by default dataimport prints message on stdout
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 3003
diff changeset
   226
def tell(msg):
bc573d5fb5b7 F [devtools] by default dataimport prints message on stdout
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 3003
diff changeset
   227
    print msg
bc573d5fb5b7 F [devtools] by default dataimport prints message on stdout
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 3003
diff changeset
   228
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   229
def confirm(question):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   230
    """A confirm function that asks for yes/no/abort and exits on abort."""
4721
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   231
    answer = shellutils.ASK.ask(question, ('Y', 'n', 'abort'), 'Y')
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   232
    if answer == 'abort':
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   233
        sys.exit(1)
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   234
    return answer == 'Y'
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   235
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   236
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   237
class catch_error(object):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   238
    """Helper for @contextmanager decorator."""
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   239
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   240
    def __init__(self, ctl, key='unexpected error', msg=None):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   241
        self.ctl = ctl
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   242
        self.key = key
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   243
        self.msg = msg
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   244
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   245
    def __enter__(self):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   246
        return self
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   247
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   248
    def __exit__(self, type, value, traceback):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   249
        if type is not None:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   250
            if issubclass(type, (KeyboardInterrupt, SystemExit)):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   251
                return # re-raise
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   252
            if self.ctl.catcherrors:
4173
cfd5d3270f99 msg isn't defined there, but we've to give traceback information to record error
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4152
diff changeset
   253
                self.ctl.record_error(self.key, None, type, value, traceback)
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   254
                return True # silent
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   255
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   256
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   257
# base sanitizing/coercing functions ###########################################
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   258
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   259
def optional(value):
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   260
    """checker to filter optional field
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   261
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   262
    If value is undefined (ex: empty string), return None that will
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   263
    break the checkers validation chain
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   264
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   265
    General use is to add 'optional' check in first condition to avoid
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   266
    ValueError by further checkers
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   267
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   268
    >>> MAPPER = [(u'value', 'value', (optional, int))]
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   269
    >>> row = {'value': u'XXX'}
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   270
    >>> mk_entity(row, MAPPER)
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   271
    {'value': None}
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   272
    >>> row = {'value': u'100'}
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   273
    >>> mk_entity(row, MAPPER)
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   274
    {'value': 100}
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   275
    """
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   276
    if value:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   277
        return value
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   278
    return None
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   279
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   280
def required(value):
6136
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   281
    """raise ValueError if value is empty
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   282
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   283
    This check should be often found in last position in the chain.
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   284
    """
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   285
    if value:
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   286
        return value
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   287
    raise ValueError("required")
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   288
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   289
def todatetime(format='%d/%m/%Y'):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   290
    """return a transformation function to turn string input value into a
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   291
    `datetime.datetime` instance, using given format.
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   292
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   293
    Follow it by `todate` or `totime` functions from `logilab.common.date` if
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   294
    you want a `date`/`time` instance instead of `datetime`.
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   295
    """
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   296
    def coerce(value):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   297
        return strptime(value, format)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   298
    return coerce
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   299
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   300
def call_transform_method(methodname, *args, **kwargs):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   301
    """return value returned by calling the given method on input"""
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   302
    def coerce(value):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   303
        return getattr(value, methodname)(*args, **kwargs)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   304
    return coerce
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   305
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   306
def call_check_method(methodname, *args, **kwargs):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   307
    """check value returned by calling the given method on input is true,
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   308
    else raise ValueError
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   309
    """
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   310
    def check(value):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   311
        if getattr(value, methodname)(*args, **kwargs):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   312
            return value
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   313
        raise ValueError('%s not verified on %r' % (methodname, value))
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   314
    return check
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   315
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   316
# base integrity checking functions ############################################
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   317
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   318
def check_doubles(buckets):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   319
    """Extract the keys that have more than one item in their bucket."""
4721
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   320
    return [(k, len(v)) for k, v in buckets.items() if len(v) > 1]
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   321
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   322
def check_doubles_not_none(buckets):
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   323
    """Extract the keys that have more than one item in their bucket."""
4721
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   324
    return [(k, len(v)) for k, v in buckets.items()
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   325
            if k is not None and len(v) > 1]
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   326
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   327
# sql generator utility functions #############################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   328
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   329
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   330
def _import_statements(sql_connect, statements, nb_threads=3,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   331
                       dump_output_dir=None,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   332
                       support_copy_from=True, encoding='utf-8'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   333
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   334
    Import a bunch of sql statements, using different threads.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   335
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   336
    try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   337
        chunksize = (len(statements) / nb_threads) + 1
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   338
        threads = []
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   339
        for i in xrange(nb_threads):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   340
            chunks = statements[i*chunksize:(i+1)*chunksize]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   341
            thread = threading.Thread(target=_execmany_thread,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   342
                                      args=(sql_connect, chunks,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   343
                                            dump_output_dir,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   344
                                            support_copy_from,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   345
                                            encoding))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   346
            thread.start()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   347
            threads.append(thread)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   348
        for t in threads:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   349
            t.join()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   350
    except Exception:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   351
        print 'Error in import statements'
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   352
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   353
def _execmany_thread_not_copy_from(cu, statement, data, table=None,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   354
                                   columns=None, encoding='utf-8'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   355
    """ Execute thread without copy from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   356
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   357
    cu.executemany(statement, data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   358
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   359
def _execmany_thread_copy_from(cu, statement, data, table,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   360
                               columns, encoding='utf-8'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   361
    """ Execute thread with copy from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   362
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   363
    buf = _create_copyfrom_buffer(data, columns, encoding)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   364
    if buf is None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   365
        _execmany_thread_not_copy_from(cu, statement, data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   366
    else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   367
        if columns is None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   368
            cu.copy_from(buf, table, null='NULL')
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   369
        else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   370
            cu.copy_from(buf, table, null='NULL', columns=columns)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   371
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   372
def _execmany_thread(sql_connect, statements, dump_output_dir=None,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   373
                     support_copy_from=True, encoding='utf-8'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   374
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   375
    Execute sql statement. If 'INSERT INTO', try to use 'COPY FROM' command,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   376
    or fallback to execute_many.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   377
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   378
    if support_copy_from:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   379
        execmany_func = _execmany_thread_copy_from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   380
    else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   381
        execmany_func = _execmany_thread_not_copy_from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   382
    cnx = sql_connect()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   383
    cu = cnx.cursor()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   384
    try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   385
        for statement, data in statements:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   386
            table = None
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   387
            columns = None
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   388
            try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   389
                if not statement.startswith('INSERT INTO'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   390
                    cu.executemany(statement, data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   391
                    continue
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   392
                table = statement.split()[2]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   393
                if isinstance(data[0], (tuple, list)):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   394
                    columns = None
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   395
                else:
8696
0bb18407c053 [toward py3k] rewrite dict.keys() and dict.values() (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8695
diff changeset
   396
                    columns = list(data[0])
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   397
                execmany_func(cu, statement, data, table, columns, encoding)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   398
            except Exception:
8970
0a1bd0c590e2 [dataimport] minor typo in error handling
Dimitri Papadopoulos <dimitri.papadopoulos@cea.fr>
parents: 8930
diff changeset
   399
                print 'unable to copy data into table %s' % table
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   400
                # Error in import statement, save data in dump_output_dir
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   401
                if dump_output_dir is not None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   402
                    pdata = {'data': data, 'statement': statement,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   403
                             'time': asctime(), 'columns': columns}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   404
                    filename = make_uid()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   405
                    try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   406
                        with open(osp.join(dump_output_dir,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   407
                                           '%s.pickle' % filename), 'w') as fobj:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   408
                            fobj.write(cPickle.dumps(pdata))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   409
                    except IOError:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   410
                        print 'ERROR while pickling in', dump_output_dir, filename+'.pickle'
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   411
                        pass
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   412
                cnx.rollback()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   413
                raise
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   414
    finally:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   415
        cnx.commit()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   416
        cu.close()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   417
8631
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   418
def _create_copyfrom_buffer(data, columns, encoding='utf-8', replace_sep=None):
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   419
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   420
    Create a StringIO buffer for 'COPY FROM' command.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   421
    Deals with Unicode, Int, Float, Date...
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   422
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   423
    # Create a list rather than directly create a StringIO
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   424
    # to correctly write lines separated by '\n' in a single step
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   425
    rows = []
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   426
    if isinstance(data[0], (tuple, list)):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   427
        columns = range(len(data[0]))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   428
    for row in data:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   429
        # Iterate over the different columns and the different values
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   430
        # and try to convert them to a correct datatype.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   431
        # If an error is raised, do not continue.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   432
        formatted_row = []
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   433
        for col in columns:
8926
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   434
            try:
8834
6947201033be [dataimport] Handle various data formats when creating buffers from data.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8833
diff changeset
   435
                value = row[col]
8926
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   436
            except KeyError:
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   437
                warnings.warn(u"Column %s is not accessible in row %s" 
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   438
                              % (col, row), RuntimeWarning)
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   439
                # XXX 'value' set to None so that the import does not end in 
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   440
                # error. 
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   441
                # Instead, the extra keys are set to NULL from the 
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   442
                # database point of view.
336e4971dc50 [dataimport] backout 6947201033be (related to #2788402)
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8900
diff changeset
   443
                value = None
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   444
            if value is None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   445
                value = 'NULL'
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   446
            elif isinstance(value, (long, int, float)):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   447
                value = str(value)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   448
            elif isinstance(value, (str, unicode)):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   449
                # Remove separators used in string formatting
8631
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   450
                for _char in (u'\t', u'\r', u'\n'):
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   451
                    if _char in value:
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   452
                        # If a replace_sep is given, replace
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   453
                        # the separator instead of returning None
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   454
                        # (and thus avoid empty buffer)
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   455
                        if replace_sep:
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   456
                            value = value.replace(_char, replace_sep)
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   457
                        else:
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   458
                            return
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   459
                value = value.replace('\\', r'\\')
8631
1053b9d0fdf7 [dataimport] Allow to replace escape char in the create_copyfrom_buffer
Vincent Michel <vincent.michel@logilab.fr>
parents: 8625
diff changeset
   460
                if value is None:
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   461
                    return
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   462
                if isinstance(value, unicode):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   463
                    value = value.encode(encoding)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   464
            elif isinstance(value, (date, datetime)):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   465
                # Do not use strftime, as it yields issue
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   466
                # with date < 1900
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   467
                value = '%04d-%02d-%02d' % (value.year,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   468
                                            value.month,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   469
                                            value.day)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   470
            else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   471
                return None
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   472
            # We push the value to the new formatted row
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   473
            # if the value is not None and could be converted to a string.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   474
            formatted_row.append(value)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   475
        rows.append('\t'.join(formatted_row))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   476
    return StringIO('\n'.join(rows))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   477
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   478
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   479
# object stores #################################################################
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   480
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   481
class ObjectStore(object):
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   482
    """Store objects in memory for *faster* validation (development mode)
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   483
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   484
    But it will not enforce the constraints of the schema and hence will miss some problems
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   485
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   486
    >>> store = ObjectStore()
7158
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
   487
    >>> user = store.create_entity('CWUser', login=u'johndoe')
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
   488
    >>> group = store.create_entity('CWUser', name=u'unknown')
0f31a50b144e [dataimport] cleanups, update docstring to up-to-date usage
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7118
diff changeset
   489
    >>> store.relate(user.eid, 'in_group', group.eid)
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   490
    """
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   491
    def __init__(self):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   492
        self.items = []
3003
2944ee420dca R [dataimport] rename uid to eid
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 2974
diff changeset
   493
        self.eids = {}
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   494
        self.types = {}
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   495
        self.relations = set()
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   496
        self.indexes = {}
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   497
        self._rql = None
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   498
        self._commit = None
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   499
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   500
    def _put(self, type, item):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   501
        self.items.append(item)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   502
        return len(self.items) - 1
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   503
6990
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   504
    def create_entity(self, etype, **data):
7159
3bcccd3ab6b6 [dataimport] ObjectStore.create_entity should return something that looks like an entity (eg no more using dict protocol to access to attributes)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7158
diff changeset
   505
        data = attrdict(data)
3bcccd3ab6b6 [dataimport] ObjectStore.create_entity should return something that looks like an entity (eg no more using dict protocol to access to attributes)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7158
diff changeset
   506
        data['eid'] = eid = self._put(etype, data)
6990
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   507
        self.eids[eid] = data
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   508
        self.types.setdefault(etype, []).append(eid)
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   509
        return data
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   510
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   511
    @deprecated("[3.11] add is deprecated, use create_entity instead")
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   512
    def add(self, etype, item):
3486
ea6bf6f9ba0c [cwctl] improve dialog messages
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 3318
diff changeset
   513
        assert isinstance(item, dict), 'item is not a dict but a %s' % type(item)
6990
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   514
        data = self.create_entity(etype, **item)
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   515
        item['eid'] = data['eid']
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   516
        return item
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   517
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   518
    def relate(self, eid_from, rtype, eid_to, **kwargs):
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   519
        """Add new relation"""
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   520
        relation = eid_from, rtype, eid_to
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   521
        self.relations.add(relation)
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   522
        return relation
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   523
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   524
    def commit(self):
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   525
        """this commit method do nothing by default
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   526
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   527
        This is voluntary to use the frequent autocommit feature in CubicWeb
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   528
        when you are using hooks or another
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   529
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   530
        If you want override commit method, please set it by the
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   531
        constructor
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   532
        """
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   533
        pass
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   534
8833
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   535
    def flush(self):
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   536
        """The method is provided so that all stores share a common API.
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   537
        It just tries to call the commit method.
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   538
        """
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   539
        print 'starting flush'
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   540
        try:
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   541
            self.commit()
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   542
        except:
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   543
            print 'failed to flush'
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   544
        else:
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   545
            print 'flush done'
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
   546
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   547
    def rql(self, *args):
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   548
        if self._rql is not None:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   549
            return self._rql(*args)
7033
ddc1b4d80dbd [dataimport] remove eid_key deprecation warning and prints
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6990
diff changeset
   550
        return []
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   551
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   552
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   553
    def nb_inserted_entities(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   554
        return len(self.eids)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   555
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   556
    def nb_inserted_types(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   557
        return len(self.types)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   558
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   559
    def nb_inserted_relations(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   560
        return len(self.relations)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   561
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   562
class RQLObjectStore(ObjectStore):
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   563
    """ObjectStore that works with an actual RQL repository (production mode)"""
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   564
    _rql = None # bw compat
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   565
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   566
    def __init__(self, session=None, commit=None):
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   567
        ObjectStore.__init__(self)
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   568
        if session is None:
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   569
            sys.exit('please provide a session of run this script with cubicweb-ctl shell and pass cnx as session')
7398
26695dd703d8 [repository api] definitly kill usage of word 'pool' to refer to connections set used by a session
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7227
diff changeset
   570
        if not hasattr(session, 'set_cnxset'):
8403
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   571
            if hasattr(session, 'request'):
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   572
                # connection object
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   573
                cnx = session
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   574
                session = session.request()
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   575
            else: # object is already a request
a6ee3cd783e1 [data import] allow a request to be given as argument, ease use from web ui
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7815
diff changeset
   576
                cnx = session.cnx
7398
26695dd703d8 [repository api] definitly kill usage of word 'pool' to refer to connections set used by a session
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7227
diff changeset
   577
            session.set_cnxset = lambda : None
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   578
            commit = commit or cnx.commit
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   579
        else:
7398
26695dd703d8 [repository api] definitly kill usage of word 'pool' to refer to connections set used by a session
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7227
diff changeset
   580
            session.set_cnxset()
6122
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   581
        self.session = session
4d2b04b32cdc improvements in dataimport.py
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 5557
diff changeset
   582
        self._commit = commit or session.commit
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   583
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   584
    def commit(self):
5063
2a94b61837e1 [dataimport] stop disabling undo ; commit return transaction id
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5054
diff changeset
   585
        txuuid = self._commit()
7398
26695dd703d8 [repository api] definitly kill usage of word 'pool' to refer to connections set used by a session
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7227
diff changeset
   586
        self.session.set_cnxset()
5063
2a94b61837e1 [dataimport] stop disabling undo ; commit return transaction id
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5054
diff changeset
   587
        return txuuid
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   588
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   589
    def rql(self, *args):
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   590
        if self._rql is not None:
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   591
            return self._rql(*args)
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   592
        return self.session.execute(*args)
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   593
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   594
    def create_entity(self, *args, **kwargs):
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   595
        entity = self.session.create_entity(*args, **kwargs)
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   596
        self.eids[entity.eid] = entity
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   597
        self.types.setdefault(args[0], []).append(entity.eid)
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   598
        return entity
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   599
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   600
    def _put(self, type, item):
6989
4a999a647f52 [dataimport] make it possible to insert 'empty' (i.e. no-attrs) entities
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6492
diff changeset
   601
        query = 'INSERT %s X' % type
4a999a647f52 [dataimport] make it possible to insert 'empty' (i.e. no-attrs) entities
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6492
diff changeset
   602
        if item:
4a999a647f52 [dataimport] make it possible to insert 'empty' (i.e. no-attrs) entities
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6492
diff changeset
   603
            query += ': ' + ', '.join('X %s %%(%s)s' % (k, k)
4a999a647f52 [dataimport] make it possible to insert 'empty' (i.e. no-attrs) entities
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6492
diff changeset
   604
                                      for k in item)
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   605
        return self.rql(query, item)[0][0]
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   606
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   607
    def relate(self, eid_from, rtype, eid_to, **kwargs):
4721
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   608
        eid_from, rtype, eid_to = super(RQLObjectStore, self).relate(
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   609
            eid_from, rtype, eid_to, **kwargs)
4136
47060a66c97f dataimport refactoring / improvments, keeping bw compat (for now)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3486
diff changeset
   610
        self.rql('SET X %s Y WHERE X eid %%(x)s, Y eid %%(y)s' % rtype,
7033
ddc1b4d80dbd [dataimport] remove eid_key deprecation warning and prints
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6990
diff changeset
   611
                 {'x': int(eid_from), 'y': int(eid_to)})
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   612
7116
dfd4680a23f0 [session] add find_entities and find_one_entity to session/request API (closes #1550045)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7033
diff changeset
   613
    def find_entities(self, *args, **kwargs):
dfd4680a23f0 [session] add find_entities and find_one_entity to session/request API (closes #1550045)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7033
diff changeset
   614
        return self.session.find_entities(*args, **kwargs)
dfd4680a23f0 [session] add find_entities and find_one_entity to session/request API (closes #1550045)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7033
diff changeset
   615
dfd4680a23f0 [session] add find_entities and find_one_entity to session/request API (closes #1550045)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7033
diff changeset
   616
    def find_one_entity(self, *args, **kwargs):
dfd4680a23f0 [session] add find_entities and find_one_entity to session/request API (closes #1550045)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7033
diff changeset
   617
        return self.session.find_one_entity(*args, **kwargs)
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   618
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   619
# the import controller ########################################################
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   620
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   621
class CWImportController(object):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   622
    """Controller of the data import process.
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   623
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   624
    >>> ctl = CWImportController(store)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   625
    >>> ctl.generators = list_of_data_generators
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   626
    >>> ctl.data = dict_of_data_tables
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   627
    >>> ctl.run()
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   628
    """
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   629
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   630
    def __init__(self, store, askerror=0, catcherrors=None, tell=tell,
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   631
                 commitevery=50):
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   632
        self.store = store
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   633
        self.generators = None
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   634
        self.data = {}
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   635
        self.errors = None
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   636
        self.askerror = askerror
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   637
        if  catcherrors is None:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   638
            catcherrors = askerror
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   639
        self.catcherrors = catcherrors
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   640
        self.commitevery = commitevery # set to None to do a single commit
3029
bc573d5fb5b7 F [devtools] by default dataimport prints message on stdout
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 3003
diff changeset
   641
        self._tell = tell
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   642
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   643
    def check(self, type, key, value):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   644
        self._checks.setdefault(type, {}).setdefault(key, []).append(value)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   645
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   646
    def check_map(self, entity, key, map, default):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   647
        try:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   648
            entity[key] = map[entity[key]]
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   649
        except KeyError:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   650
            self.check(key, entity[key], None)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   651
            entity[key] = default
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   652
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   653
    def record_error(self, key, msg=None, type=None, value=None, tb=None):
4186
ca7e526b07b6 import cleanup, check data file exists
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4173
diff changeset
   654
        tmp = StringIO()
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   655
        if type is None:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   656
            traceback.print_exc(file=tmp)
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   657
        else:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   658
            traceback.print_exception(type, value, tb, file=tmp)
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   659
        # use a list to avoid counting a <nb lines> errors instead of one
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   660
        errorlog = self.errors.setdefault(key, [])
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   661
        if msg is None:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   662
            errorlog.append(tmp.getvalue().splitlines())
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   663
        else:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   664
            errorlog.append( (msg, tmp.getvalue().splitlines()) )
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   665
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   666
    def run(self):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   667
        self.errors = {}
7171
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   668
        if self.commitevery is None:
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   669
            self.tell('Will commit all or nothing.')
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   670
        else:
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   671
            self.tell('Will commit every %s iterations' % self.commitevery)
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   672
        for func, checks in self.generators:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   673
            self._checks = {}
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   674
            func_name = func.__name__
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   675
            self.tell("Run import function '%s'..." % func_name)
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   676
            try:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   677
                func(self)
7815
2a164a9cf81c [exceptions] stop catching any exception in various places (closes #1942716)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7471
diff changeset
   678
            except Exception:
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   679
                if self.catcherrors:
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   680
                    self.record_error(func_name, 'While calling %s' % func.__name__)
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   681
                else:
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   682
                    self._print_stats()
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   683
                    raise
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   684
            for key, func, title, help in checks:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   685
                buckets = self._checks.get(key)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   686
                if buckets:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   687
                    err = func(buckets)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   688
                    if err:
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   689
                        self.errors[title] = (help, err)
7171
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   690
        try:
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   691
            txuuid = self.store.commit()
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   692
            if txuuid is not None:
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   693
                self.tell('Transaction commited (txuuid: %s)' % txuuid)
8695
358d8bed9626 [toward-py3k] rewrite to "except AnException as exc:" (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8637
diff changeset
   694
        except QueryError as ex:
7171
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   695
            self.tell('Transaction aborted: %s' % ex)
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   696
        self._print_stats()
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   697
        if self.errors:
4721
8f63691ccb7f pylint style fixes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4613
diff changeset
   698
            if self.askerror == 2 or (self.askerror and confirm('Display errors ?')):
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   699
                from pprint import pformat
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   700
                for errkey, error in self.errors.items():
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   701
                    self.tell("\n%s (%s): %d\n" % (error[0], errkey, len(error[1])))
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   702
                    self.tell(pformat(sorted(error[1])))
7171
4297be67bbe4 [dataimport] tell more and nicely about transaction status
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7170
diff changeset
   703
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   704
    def _print_stats(self):
8696
0bb18407c053 [toward py3k] rewrite dict.keys() and dict.values() (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8695
diff changeset
   705
        nberrors = sum(len(err) for err in self.errors.itervalues())
4912
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   706
        self.tell('\nImport statistics: %i entities, %i types, %i relations and %i errors'
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   707
                  % (self.store.nb_inserted_entities,
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   708
                     self.store.nb_inserted_types,
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   709
                     self.store.nb_inserted_relations,
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   710
                     nberrors))
9767cc516b4f [R] dataimport: changes
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4847
diff changeset
   711
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   712
    def get_data(self, key):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   713
        return self.data.get(key)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   714
4527
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   715
    def index(self, name, key, value, unique=False):
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   716
        """create a new index
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   717
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   718
        If unique is set to True, only first occurence will be kept not the following ones
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   719
        """
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   720
        if unique:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   721
            try:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   722
                if value in self.store.indexes[name][key]:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   723
                    return
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   724
            except KeyError:
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   725
                # we're sure that one is the first occurence; so continue...
67ab70e98488 [R] devtools: improve default data import mechanism
Julien Jehannet <julien.jehannet@logilab.fr>
parents: 4252
diff changeset
   726
                pass
2974
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   727
        self.store.indexes.setdefault(name, {}).setdefault(key, []).append(value)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   728
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   729
    def tell(self, msg):
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   730
        self._tell(msg)
3dfe497e5afa F tools to import data
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents:
diff changeset
   731
4152
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   732
    def iter_and_commit(self, datakey):
30fd1229137d new catch_error context manager, nicer controller __init__ and new iter_and_commit(datakey) method
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4140
diff changeset
   733
        """iter rows, triggering commit every self.commitevery iterations"""
6136
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   734
        if self.commitevery is None:
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   735
            return self.get_data(datakey)
79da6f969b15 [dataimport] refactor commitevery to gain readability
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 6133
diff changeset
   736
        else:
6169
55378e1bab1b fix order of parameters in call to callfunc_every
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 6136
diff changeset
   737
            return callfunc_every(self.store.commit,
55378e1bab1b fix order of parameters in call to callfunc_every
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 6136
diff changeset
   738
                                  self.commitevery,
55378e1bab1b fix order of parameters in call to callfunc_every
Alexandre Fayolle <alexandre.fayolle@logilab.fr>
parents: 6136
diff changeset
   739
                                  self.get_data(datakey))
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   740
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   741
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   742
class NoHookRQLObjectStore(RQLObjectStore):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   743
    """ObjectStore that works with an actual RQL repository (production mode)"""
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   744
    _rql = None # bw compat
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   745
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   746
    def __init__(self, session, metagen=None, baseurl=None):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   747
        super(NoHookRQLObjectStore, self).__init__(session)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   748
        self.source = session.repo.system_source
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   749
        self.rschema = session.repo.schema.rschema
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   750
        self.add_relation = self.source.add_relation
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   751
        if metagen is None:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   752
            metagen = MetaGenerator(session, baseurl)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   753
        self.metagen = metagen
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   754
        self._nb_inserted_entities = 0
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   755
        self._nb_inserted_types = 0
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   756
        self._nb_inserted_relations = 0
5054
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   757
        self.rql = session.execute
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   758
        # deactivate security
8807
d9aaad2c52e9 [session] drop useless getter and setter for security
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8797
diff changeset
   759
        session.read_security = False
d9aaad2c52e9 [session] drop useless getter and setter for security
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8797
diff changeset
   760
        session.write_security = False
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   761
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   762
    def create_entity(self, etype, **kwargs):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   763
        for k, v in kwargs.iteritems():
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   764
            kwargs[k] = getattr(v, 'eid', v)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   765
        entity, rels = self.metagen.base_etype_dicts(etype)
7471
bf9443f8725f [dataimport] fix #1732685: cached entity and shared cw_edited data with NoHookRQLObjectStore
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 7398
diff changeset
   766
        # make a copy to keep cached entity pristine
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   767
        entity = copy(entity)
7471
bf9443f8725f [dataimport] fix #1732685: cached entity and shared cw_edited data with NoHookRQLObjectStore
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 7398
diff changeset
   768
        entity.cw_edited = copy(entity.cw_edited)
5557
1a534c596bff [entity] continue cleanup of Entity/AnyEntity namespace
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   769
        entity.cw_clear_relation_cache()
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   770
        self.metagen.init_entity(entity)
6142
8bc6eac1fac1 [session] cleanup hook / operation / entity edition api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6122
diff changeset
   771
        entity.cw_edited.update(kwargs, skipsec=False)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   772
        session = self.session
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   773
        self.source.add_entity(session, entity)
5054
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   774
        self.source.add_info(session, entity, self.source, None, complete=False)
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   775
        kwargs = dict()
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   776
        if inspect.getargspec(self.add_relation).keywords:
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8835
diff changeset
   777
            kwargs['subjtype'] = entity.cw_etype
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   778
        for rtype, targeteids in rels.iteritems():
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   779
            # targeteids may be a single eid or a list of eids
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   780
            inlined = self.rschema(rtype).inlined
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   781
            try:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   782
                for targeteid in targeteids:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   783
                    self.add_relation(session, entity.eid, rtype, targeteid,
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   784
                                      inlined, **kwargs)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   785
            except TypeError:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   786
                self.add_relation(session, entity.eid, rtype, targeteids,
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   787
                                  inlined, **kwargs)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   788
        self._nb_inserted_entities += 1
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   789
        return entity
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   790
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   791
    def relate(self, eid_from, rtype, eid_to, **kwargs):
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   792
        assert not rtype.startswith('reverse_')
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   793
        self.add_relation(self.session, eid_from, rtype, eid_to,
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   794
                          self.rschema(rtype).inlined)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   795
        self._nb_inserted_relations += 1
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   796
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   797
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   798
    def nb_inserted_entities(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   799
        return self._nb_inserted_entities
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   800
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   801
    def nb_inserted_types(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   802
        return self._nb_inserted_types
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   803
    @property
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   804
    def nb_inserted_relations(self):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   805
        return self._nb_inserted_relations
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   806
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   807
    def _put(self, type, item):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   808
        raise RuntimeError('use create entity')
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   809
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   810
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   811
class MetaGenerator(object):
6427
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   812
    META_RELATIONS = (META_RTYPES
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   813
                      - VIRTUAL_RTYPES
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   814
                      - set(('eid', 'cwuri',
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   815
                             'is', 'is_instance_of', 'cw_source')))
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   816
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   817
    def __init__(self, session, baseurl=None):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   818
        self.session = session
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   819
        self.source = session.repo.system_source
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   820
        self.time = datetime.now()
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   821
        if baseurl is None:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   822
            config = session.vreg.config
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   823
            baseurl = config['base-url'] or config.default_base_url()
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   824
        if not baseurl[-1] == '/':
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   825
            baseurl += '/'
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   826
        self.baseurl =  baseurl
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   827
        # attributes/relations shared by all entities of the same type
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   828
        self.etype_attrs = []
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   829
        self.etype_rels = []
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   830
        # attributes/relations specific to each entity
5054
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   831
        self.entity_attrs = ['cwuri']
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   832
        #self.entity_rels = [] XXX not handled (YAGNI?)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   833
        schema = session.vreg.schema
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   834
        rschema = schema.rschema
6427
c8a5ac2d1eaa [schema / sources] store data sources as cubicweb entities
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6173
diff changeset
   835
        for rtype in self.META_RELATIONS:
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   836
            if rschema(rtype).final:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   837
                self.etype_attrs.append(rtype)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   838
            else:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   839
                self.etype_rels.append(rtype)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   840
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   841
    @cached
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   842
    def base_etype_dicts(self, etype):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   843
        entity = self.session.vreg['etypes'].etype_class(etype)(self.session)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   844
        # entity are "surface" copied, avoid shared dict between copies
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   845
        del entity.cw_extra_kwargs
6142
8bc6eac1fac1 [session] cleanup hook / operation / entity edition api
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6122
diff changeset
   846
        entity.cw_edited = EditedEntity(entity)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   847
        for attr in self.etype_attrs:
6990
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   848
            entity.cw_edited.edited_attribute(attr, self.generate(entity, attr))
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   849
        rels = {}
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   850
        for rel in self.etype_rels:
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   851
            rels[rel] = self.generate(entity, rel)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   852
        return entity, rels
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   853
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   854
    def init_entity(self, entity):
5054
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   855
        entity.eid = self.source.create_eid(self.session)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   856
        for attr in self.entity_attrs:
6990
353ad06867a8 [dataimport] implement create_entity() on ObjectStore to provide a consistent interface
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 6989
diff changeset
   857
            entity.cw_edited.edited_attribute(attr, self.generate(entity, attr))
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   858
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   859
    def generate(self, entity, rtype):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   860
        return getattr(self, 'gen_%s' % rtype)(entity)
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   861
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   862
    def gen_cwuri(self, entity):
5054
cb066d29166a fix dataimport for 3.7.2
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4913
diff changeset
   863
        return u'%seid/%s' % (self.baseurl, entity.eid)
4818
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   864
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   865
    def gen_creation_date(self, entity):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   866
        return self.time
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   867
    def gen_modification_date(self, entity):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   868
        return self.time
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   869
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   870
    def gen_created_by(self, entity):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   871
        return self.session.user.eid
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   872
    def gen_owned_by(self, entity):
9f9bfbcdecfd le patch massiveimport a été importé
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4734
diff changeset
   873
        return self.session.user.eid
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   874
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   875
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   876
###########################################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   877
## SQL object store #######################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   878
###########################################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   879
class SQLGenObjectStore(NoHookRQLObjectStore):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   880
    """Controller of the data import process. This version is based
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   881
    on direct insertions throught SQL command (COPY FROM or execute many).
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   882
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   883
    >>> store = SQLGenObjectStore(session)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   884
    >>> store.create_entity('Person', ...)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   885
    >>> store.flush()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   886
    """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   887
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   888
    def __init__(self, session, dump_output_dir=None, nb_threads_statement=3):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   889
        """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   890
        Initialize a SQLGenObjectStore.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   891
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   892
        Parameters:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   893
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   894
          - session: session on the cubicweb instance
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   895
          - dump_output_dir: a directory to dump failed statements
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   896
            for easier recovery. Default is None (no dump).
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   897
          - nb_threads_statement: number of threads used
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   898
            for SQL insertion (default is 3).
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   899
        """
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   900
        super(SQLGenObjectStore, self).__init__(session)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   901
        ### hijack default source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   902
        self.source = SQLGenSourceWrapper(
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   903
            self.source, session.vreg.schema,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   904
            dump_output_dir=dump_output_dir,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   905
            nb_threads_statement=nb_threads_statement)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   906
        ### XXX This is done in super().__init__(), but should be
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   907
        ### redone here to link to the correct source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   908
        self.add_relation = self.source.add_relation
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   909
        self.indexes_etypes = {}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   910
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   911
    def flush(self):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   912
        """Flush data to the database"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   913
        self.source.flush()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   914
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   915
    def relate(self, subj_eid, rtype, obj_eid, **kwargs):
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   916
        if subj_eid is None or obj_eid is None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   917
            return
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   918
        # XXX Could subjtype be inferred ?
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   919
        self.source.add_relation(self.session, subj_eid, rtype, obj_eid,
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
   920
                                 self.rschema(rtype).inlined, **kwargs)
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   921
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   922
    def drop_indexes(self, etype):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   923
        """Drop indexes for a given entity type"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   924
        if etype not in self.indexes_etypes:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   925
            cu = self.session.cnxset['system']
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   926
            def index_to_attr(index):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   927
                """turn an index name to (database) attribute name"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   928
                return index.replace(etype.lower(), '').replace('idx', '').strip('_')
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   929
            indices = [(index, index_to_attr(index))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   930
                       for index in self.source.dbhelper.list_indices(cu, etype)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   931
                       # Do not consider 'cw_etype_pkey' index
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   932
                       if not index.endswith('key')]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   933
            self.indexes_etypes[etype] = indices
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   934
        for index, attr in self.indexes_etypes[etype]:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   935
            self.session.system_sql('DROP INDEX %s' % index)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   936
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   937
    def create_indexes(self, etype):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   938
        """Recreate indexes for a given entity type"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   939
        for index, attr in self.indexes_etypes.get(etype, []):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   940
            sql = 'CREATE INDEX %s ON cw_%s(%s)' % (index, etype, attr)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   941
            self.session.system_sql(sql)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   942
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   943
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   944
###########################################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   945
## SQL Source #############################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   946
###########################################################################
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   947
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   948
class SQLGenSourceWrapper(object):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   949
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   950
    def __init__(self, system_source, schema,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   951
                 dump_output_dir=None, nb_threads_statement=3):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   952
        self.system_source = system_source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   953
        self._sql = threading.local()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   954
        # Explicitely backport attributes from system source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   955
        self._storage_handler = self.system_source._storage_handler
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   956
        self.preprocess_entity = self.system_source.preprocess_entity
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   957
        self.sqlgen = self.system_source.sqlgen
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   958
        self.copy_based_source = self.system_source.copy_based_source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   959
        self.uri = self.system_source.uri
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   960
        self.eid = self.system_source.eid
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   961
        # Directory to write temporary files
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   962
        self.dump_output_dir = dump_output_dir
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   963
        # Allow to execute code with SQLite backend that does
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   964
        # not support (yet...) copy_from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   965
        # XXX Should be dealt with in logilab.database
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   966
        spcfrom = system_source.dbhelper.dbapi_module.support_copy_from
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   967
        self.support_copy_from = spcfrom
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   968
        self.dbencoding = system_source.dbhelper.dbencoding
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   969
        self.nb_threads_statement = nb_threads_statement
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   970
        # initialize thread-local data for main thread
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   971
        self.init_thread_locals()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   972
        self._inlined_rtypes_cache = {}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   973
        self._fill_inlined_rtypes_cache(schema)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   974
        self.schema = schema
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   975
        self.do_fti = False
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   976
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   977
    def _fill_inlined_rtypes_cache(self, schema):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   978
        cache = self._inlined_rtypes_cache
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   979
        for eschema in schema.entities():
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   980
            for rschema in eschema.ordered_relations():
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   981
                if rschema.inlined:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   982
                    cache[eschema.type] = SQL_PREFIX + rschema.type
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   983
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   984
    def init_thread_locals(self):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   985
        """initializes thread-local data"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   986
        self._sql.entities = defaultdict(list)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   987
        self._sql.relations = {}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   988
        self._sql.inlined_relations = {}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   989
        # keep track, for each eid of the corresponding data dict
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   990
        self._sql.eid_insertdicts = {}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   991
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   992
    def flush(self):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   993
        print 'starting flush'
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   994
        _entities_sql = self._sql.entities
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   995
        _relations_sql = self._sql.relations
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   996
        _inlined_relations_sql = self._sql.inlined_relations
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   997
        _insertdicts = self._sql.eid_insertdicts
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   998
        try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
   999
            # try, for each inlined_relation, to find if we're also creating
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1000
            # the host entity (i.e. the subject of the relation).
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1001
            # In that case, simply update the insert dict and remove
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1002
            # the need to make the
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1003
            # UPDATE statement
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1004
            for statement, datalist in _inlined_relations_sql.iteritems():
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1005
                new_datalist = []
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1006
                # for a given inlined relation,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1007
                # browse each couple to be inserted
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1008
                for data in datalist:
8696
0bb18407c053 [toward py3k] rewrite dict.keys() and dict.values() (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8695
diff changeset
  1009
                    keys = list(data)
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1010
                    # For inlined relations, it exists only two case:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1011
                    # (rtype, cw_eid) or (cw_eid, rtype)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1012
                    if keys[0] == 'cw_eid':
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1013
                        rtype = keys[1]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1014
                    else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1015
                        rtype = keys[0]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1016
                    updated_eid = data['cw_eid']
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1017
                    if updated_eid in _insertdicts:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1018
                        _insertdicts[updated_eid][rtype] = data[rtype]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1019
                    else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1020
                        # could not find corresponding insert dict, keep the
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1021
                        # UPDATE query
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1022
                        new_datalist.append(data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1023
                _inlined_relations_sql[statement] = new_datalist
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1024
            _import_statements(self.system_source.get_connection,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1025
                               _entities_sql.items()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1026
                               + _relations_sql.items()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1027
                               + _inlined_relations_sql.items(),
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1028
                               dump_output_dir=self.dump_output_dir,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1029
                               nb_threads=self.nb_threads_statement,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1030
                               support_copy_from=self.support_copy_from,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1031
                               encoding=self.dbencoding)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1032
        except:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1033
            print 'failed to flush'
8833
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
  1034
        else:
39f81e2db2fc [dataimport] Add a ``flush`` method for all stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8832
diff changeset
  1035
            print 'flush done'
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1036
        finally:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1037
            _entities_sql.clear()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1038
            _relations_sql.clear()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1039
            _insertdicts.clear()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1040
            _inlined_relations_sql.clear()
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1041
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1042
    def add_relation(self, session, subject, rtype, object,
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
  1043
                     inlined=False, **kwargs):
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1044
        if inlined:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1045
            _sql = self._sql.inlined_relations
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1046
            data = {'cw_eid': subject, SQL_PREFIX + rtype: object}
8832
26cdfc6dd6f8 [dataimport] Uniformize the API across the different stores.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8807
diff changeset
  1047
            subjtype = kwargs.get('subjtype')
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1048
            if subjtype is None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1049
                # Try to infer it
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1050
                targets = [t.type for t in
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1051
                           self.schema.rschema(rtype).targets()]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1052
                if len(targets) == 1:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1053
                    subjtype = targets[0]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1054
                else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1055
                    raise ValueError('You should give the subject etype for '
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1056
                                     'inlined relation %s'
8835
3612b760488b [dataimport] Slight message modification in exception handling code.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8834
diff changeset
  1057
                                     ', as it cannot be inferred: '
3612b760488b [dataimport] Slight message modification in exception handling code.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8834
diff changeset
  1058
                                     'this type is given as keyword argument '
3612b760488b [dataimport] Slight message modification in exception handling code.
Vladimir Popescu <vladimir.popescu@logilab.fr>
parents: 8834
diff changeset
  1059
                                     '``subjtype``'% rtype)
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1060
            statement = self.sqlgen.update(SQL_PREFIX + subjtype,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1061
                                           data, ['cw_eid'])
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1062
        else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1063
            _sql = self._sql.relations
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1064
            data = {'eid_from': subject, 'eid_to': object}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1065
            statement = self.sqlgen.insert('%s_relation' % rtype, data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1066
        if statement in _sql:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1067
            _sql[statement].append(data)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1068
        else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1069
            _sql[statement] = [data]
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1070
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1071
    def add_entity(self, session, entity):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1072
        with self._storage_handler(entity, 'added'):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1073
            attrs = self.preprocess_entity(entity)
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8835
diff changeset
  1074
            rtypes = self._inlined_rtypes_cache.get(entity.cw_etype, ())
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1075
            if isinstance(rtypes, str):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1076
                rtypes = (rtypes,)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1077
            for rtype in rtypes:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1078
                if rtype not in attrs:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1079
                    attrs[rtype] = None
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8835
diff changeset
  1080
            sql = self.sqlgen.insert(SQL_PREFIX + entity.cw_etype, attrs)
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1081
            self._sql.eid_insertdicts[entity.eid] = attrs
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1082
            self._append_to_entities(sql, attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1083
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1084
    def _append_to_entities(self, sql, attrs):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1085
        self._sql.entities[sql].append(attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1086
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1087
    def _handle_insert_entity_sql(self, session, sql, attrs):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1088
        # We have to overwrite the source given in parameters
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1089
        # as here, we directly use the system source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1090
        attrs['source'] = 'system'
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1091
        attrs['asource'] = self.system_source.uri
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1092
        self._append_to_entities(sql, attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1093
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1094
    def _handle_is_relation_sql(self, session, sql, attrs):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1095
        self._append_to_entities(sql, attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1096
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1097
    def _handle_is_instance_of_sql(self, session, sql, attrs):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1098
        self._append_to_entities(sql, attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1099
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1100
    def _handle_source_relation_sql(self, session, sql, attrs):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1101
        self._append_to_entities(sql, attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1102
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1103
    # XXX add_info is similar to the one in NativeSQLSource. It is rewritten
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1104
    # here to correctly used the _handle_xxx of the SQLGenSourceWrapper. This
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1105
    # part should be rewritten in a more clearly way.
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1106
    def add_info(self, session, entity, source, extid, complete):
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1107
        """add type and source info for an eid into the system table"""
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1108
        # begin by inserting eid/type/source/extid into the entities table
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1109
        if extid is not None:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1110
            assert isinstance(extid, str)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1111
            extid = b64encode(extid)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1112
        uri = 'system' if source.copy_based_source else source.uri
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8835
diff changeset
  1113
        attrs = {'type': entity.cw_etype, 'eid': entity.eid, 'extid': extid,
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1114
                 'source': uri, 'asource': source.uri, 'mtime': datetime.utcnow()}
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1115
        self._handle_insert_entity_sql(session, self.sqlgen.insert('entities', attrs), attrs)
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1116
        # insert core relations: is, is_instance_of and cw_source
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1117
        try:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1118
            self._handle_is_relation_sql(session, 'INSERT INTO is_relation(eid_from,eid_to) VALUES (%s,%s)',
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1119
                                         (entity.eid, eschema_eid(session, entity.e_schema)))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1120
        except IndexError:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1121
            # during schema serialization, skip
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1122
            pass
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1123
        else:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1124
            for eschema in entity.e_schema.ancestors() + [entity.e_schema]:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1125
                self._handle_is_relation_sql(session,
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1126
                                             'INSERT INTO is_instance_of_relation(eid_from,eid_to) VALUES (%s,%s)',
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1127
                                             (entity.eid, eschema_eid(session, eschema)))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1128
        if 'CWSource' in self.schema and source.eid is not None: # else, cw < 3.10
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1129
            self._handle_is_relation_sql(session, 'INSERT INTO cw_source_relation(eid_from,eid_to) VALUES (%s,%s)',
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1130
                                         (entity.eid, source.eid))
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1131
        # now we can update the full text index
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8835
diff changeset
  1132
        if self.do_fti and self.need_fti_indexation(entity.cw_etype):
8625
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1133
            if complete:
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1134
                entity.complete(entity.e_schema.indexable_attributes())
7ee0752178e5 [dataimport] Add SQL Store for faster import - works ONLY with Postgres for now, as it requires "copy from" command - closes #2410822
Vincent Michel <vincent.michel@logilab.fr>
parents: 8406
diff changeset
  1135
            self.index_entity(session, entity=entity)