dataimport/deprecated.py
author Sylvain Thénault <sylvain.thenault@logilab.fr>
Tue, 24 Nov 2015 10:16:55 +0100
changeset 10915 cf800aa43f48
parent 10907 9ae707db5265
child 10939 b30c2f49da57
permissions -rw-r--r--
[server/sources] fix name error in eid_type_source_pre_131 i.e. the eid_type_source implementation used when some pre 3.13.1 database is detected. Regression from 5fbdbbe51867.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     1
# copyright 2003-2015 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     2
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     3
#
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     4
# This file is part of CubicWeb.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     5
#
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     6
# CubicWeb is free software: you can redistribute it and/or modify it under the
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     7
# terms of the GNU Lesser General Public License as published by the Free
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     8
# Software Foundation, either version 2.1 of the License, or (at your option)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
     9
# any later version.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    10
#
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    11
# CubicWeb is distributed in the hope that it will be useful, but WITHOUT
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    12
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    13
# FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License for more
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    14
# details.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    15
#
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    16
# You should have received a copy of the GNU Lesser General Public License along
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    17
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    18
"""Old and deprecated dataimport API that provides tools to import tabular data.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    19
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    20
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    21
Example of use (run this with `cubicweb-ctl shell instance import-script.py`):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    22
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    23
.. sourcecode:: python
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    24
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    25
  from cubicweb.dataimport import *
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    26
  # define data generators
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    27
  GENERATORS = []
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    28
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    29
  USERS = [('Prenom', 'firstname', ()),
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    30
           ('Nom', 'surname', ()),
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    31
           ('Identifiant', 'login', ()),
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    32
           ]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    33
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    34
  def gen_users(ctl):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    35
      for row in ctl.iter_and_commit('utilisateurs'):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    36
          entity = mk_entity(row, USERS)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    37
          entity['upassword'] = 'motdepasse'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    38
          ctl.check('login', entity['login'], None)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    39
          entity = ctl.store.prepare_insert_entity('CWUser', **entity)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    40
          email = ctl.store.prepare_insert_entity('EmailAddress', address=row['email'])
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    41
          ctl.store.prepare_insert_relation(entity, 'use_email', email)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    42
          ctl.store.rql('SET U in_group G WHERE G name "users", U eid %(x)s', {'x': entity})
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    43
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    44
  CHK = [('login', check_doubles, 'Utilisateurs Login',
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    45
          'Deux utilisateurs ne devraient pas avoir le meme login.'),
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    46
         ]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    47
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    48
  GENERATORS.append( (gen_users, CHK) )
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    49
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    50
  # create controller
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    51
  ctl = CWImportController(RQLObjectStore(cnx))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    52
  ctl.askerror = 1
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    53
  ctl.generators = GENERATORS
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    54
  ctl.data['utilisateurs'] = lazytable(ucsvreader(open('users.csv')))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    55
  # run
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    56
  ctl.run()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    57
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    58
.. BUG file with one column are not parsable
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    59
.. TODO rollback() invocation is not possible yet
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    60
"""
10589
7c23b7de2b8d [py3k] print function
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10513
diff changeset
    61
from __future__ import print_function
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    62
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    63
import sys
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    64
import traceback
10808
b6b2e1bcd1b2 [dataimport] import StringIO from io
Julien Cristau <julien.cristau@logilab.fr>
parents: 10688
diff changeset
    65
from io import StringIO
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    66
10688
fa29f3628a1b [py3k] use six.add_metaclass
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10669
diff changeset
    67
from six import add_metaclass
fa29f3628a1b [py3k] use six.add_metaclass
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10669
diff changeset
    68
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    69
from logilab.common import attrdict, shellutils
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    70
from logilab.common.date import strptime
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    71
from logilab.common.deprecation import deprecated, class_deprecated
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    72
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    73
from cubicweb import QueryError
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    74
from cubicweb.dataimport import callfunc_every
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    75
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    76
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    77
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    78
def lazytable(reader):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    79
    """The first row is taken to be the header of the table and
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    80
    used to output a dict for each row of data.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    81
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    82
    >>> data = lazytable(ucsvreader(open(filename)))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    83
    """
10669
155c29e0ed1c [py3k] use next builtin instead of next method
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10663
diff changeset
    84
    header = next(reader)
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    85
    for row in reader:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    86
        yield dict(zip(header, row))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    87
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    88
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    89
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    90
def lazydbtable(cu, table, headers, orderby=None):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    91
    """return an iterator on rows of a sql table. On each row, fetch columns
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    92
    defined in headers and return values as a dictionary.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    93
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    94
    >>> data = lazydbtable(cu, 'experimentation', ('id', 'nickname', 'gps'))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    95
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    96
    sql = 'SELECT %s FROM %s' % (','.join(headers), table,)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    97
    if orderby:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    98
        sql += ' ORDER BY %s' % ','.join(orderby)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
    99
    cu.execute(sql)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   100
    while True:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   101
        row = cu.fetchone()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   102
        if row is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   103
            break
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   104
        yield dict(zip(headers, row))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   105
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   106
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   107
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   108
def tell(msg):
10589
7c23b7de2b8d [py3k] print function
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10513
diff changeset
   109
    print(msg)
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   110
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   111
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   112
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   113
def confirm(question):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   114
    """A confirm function that asks for yes/no/abort and exits on abort."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   115
    answer = shellutils.ASK.ask(question, ('Y', 'n', 'abort'), 'Y')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   116
    if answer == 'abort':
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   117
        sys.exit(1)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   118
    return answer == 'Y'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   119
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   120
10688
fa29f3628a1b [py3k] use six.add_metaclass
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10669
diff changeset
   121
@add_metaclass(class_deprecated)
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   122
class catch_error(object):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   123
    """Helper for @contextmanager decorator."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   124
    __deprecation_warning__ = '[3.21] deprecated'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   125
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   126
    def __init__(self, ctl, key='unexpected error', msg=None):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   127
        self.ctl = ctl
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   128
        self.key = key
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   129
        self.msg = msg
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   130
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   131
    def __enter__(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   132
        return self
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   133
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   134
    def __exit__(self, type, value, traceback):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   135
        if type is not None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   136
            if issubclass(type, (KeyboardInterrupt, SystemExit)):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   137
                return # re-raise
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   138
            if self.ctl.catcherrors:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   139
                self.ctl.record_error(self.key, None, type, value, traceback)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   140
                return True # silent
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   141
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   142
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   143
def mk_entity(row, map):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   144
    """Return a dict made from sanitized mapped values.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   145
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   146
    ValueError can be raised on unexpected values found in checkers
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   147
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   148
    >>> row = {'myname': u'dupont'}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   149
    >>> map = [('myname', u'name', (call_transform_method('title'),))]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   150
    >>> mk_entity(row, map)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   151
    {'name': u'Dupont'}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   152
    >>> row = {'myname': u'dupont', 'optname': u''}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   153
    >>> map = [('myname', u'name', (call_transform_method('title'),)),
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   154
    ...        ('optname', u'MARKER', (optional,))]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   155
    >>> mk_entity(row, map)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   156
    {'name': u'Dupont', 'optname': None}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   157
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   158
    res = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   159
    assert isinstance(row, dict)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   160
    assert isinstance(map, list)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   161
    for src, dest, funcs in map:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   162
        try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   163
            res[dest] = row[src]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   164
        except KeyError:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   165
            continue
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   166
        try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   167
            for func in funcs:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   168
                res[dest] = func(res[dest])
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   169
                if res[dest] is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   170
                    break
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   171
        except ValueError as err:
10590
7629902e7554 [py3k] Fix raise with embedded traceback
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10589
diff changeset
   172
            exc = ValueError('error with %r field: %s' % (src, err))
7629902e7554 [py3k] Fix raise with embedded traceback
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10589
diff changeset
   173
            exc.__traceback__ = sys.exc_info()[-1]
7629902e7554 [py3k] Fix raise with embedded traceback
Samuel Trégouët <samuel.tregouet@logilab.fr>
parents: 10589
diff changeset
   174
            raise exc
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   175
    return res
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   176
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   177
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   178
# base sanitizing/coercing functions ###########################################
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   179
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   180
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   181
def optional(value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   182
    """checker to filter optional field
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   183
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   184
    If value is undefined (ex: empty string), return None that will
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   185
    break the checkers validation chain
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   186
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   187
    General use is to add 'optional' check in first condition to avoid
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   188
    ValueError by further checkers
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   189
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   190
    >>> MAPPER = [(u'value', 'value', (optional, int))]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   191
    >>> row = {'value': u'XXX'}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   192
    >>> mk_entity(row, MAPPER)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   193
    {'value': None}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   194
    >>> row = {'value': u'100'}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   195
    >>> mk_entity(row, MAPPER)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   196
    {'value': 100}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   197
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   198
    if value:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   199
        return value
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   200
    return None
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   201
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   202
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   203
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   204
def required(value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   205
    """raise ValueError if value is empty
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   206
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   207
    This check should be often found in last position in the chain.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   208
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   209
    if value:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   210
        return value
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   211
    raise ValueError("required")
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   212
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   213
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   214
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   215
def todatetime(format='%d/%m/%Y'):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   216
    """return a transformation function to turn string input value into a
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   217
    `datetime.datetime` instance, using given format.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   218
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   219
    Follow it by `todate` or `totime` functions from `logilab.common.date` if
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   220
    you want a `date`/`time` instance instead of `datetime`.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   221
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   222
    def coerce(value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   223
        return strptime(value, format)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   224
    return coerce
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   225
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   226
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   227
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   228
def call_transform_method(methodname, *args, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   229
    """return value returned by calling the given method on input"""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   230
    def coerce(value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   231
        return getattr(value, methodname)(*args, **kwargs)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   232
    return coerce
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   233
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   234
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   235
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   236
def call_check_method(methodname, *args, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   237
    """check value returned by calling the given method on input is true,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   238
    else raise ValueError
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   239
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   240
    def check(value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   241
        if getattr(value, methodname)(*args, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   242
            return value
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   243
        raise ValueError('%s not verified on %r' % (methodname, value))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   244
    return check
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   245
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   246
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   247
# base integrity checking functions ############################################
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   248
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   249
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   250
def check_doubles(buckets):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   251
    """Extract the keys that have more than one item in their bucket."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   252
    return [(k, len(v)) for k, v in buckets.items() if len(v) > 1]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   253
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   254
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   255
@deprecated('[3.21] deprecated')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   256
def check_doubles_not_none(buckets):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   257
    """Extract the keys that have more than one item in their bucket."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   258
    return [(k, len(v)) for k, v in buckets.items()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   259
            if k is not None and len(v) > 1]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   260
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   261
10688
fa29f3628a1b [py3k] use six.add_metaclass
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10669
diff changeset
   262
@add_metaclass(class_deprecated)
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   263
class ObjectStore(object):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   264
    """Store objects in memory for *faster* validation (development mode)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   265
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   266
    But it will not enforce the constraints of the schema and hence will miss some problems
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   267
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   268
    >>> store = ObjectStore()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   269
    >>> user = store.prepare_insert_entity('CWUser', login=u'johndoe')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   270
    >>> group = store.prepare_insert_entity('CWUser', name=u'unknown')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   271
    >>> store.prepare_insert_relation(user, 'in_group', group)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   272
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   273
    __deprecation_warning__ = '[3.21] use the new importer API'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   274
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   275
    def __init__(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   276
        self.items = []
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   277
        self.eids = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   278
        self.types = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   279
        self.relations = set()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   280
        self.indexes = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   281
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   282
    def prepare_insert_entity(self, etype, **data):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   283
        """Given an entity type, attributes and inlined relations, return an eid for the entity that
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   284
        would be inserted with a real store.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   285
        """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   286
        data = attrdict(data)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   287
        data['eid'] = eid = len(self.items)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   288
        self.items.append(data)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   289
        self.eids[eid] = data
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   290
        self.types.setdefault(etype, []).append(eid)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   291
        return eid
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   292
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   293
    def prepare_update_entity(self, etype, eid, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   294
        """Given an entity type and eid, updates the corresponding fake entity with specified
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   295
        attributes and inlined relations.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   296
        """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   297
        assert eid in self.types[etype], 'Trying to update with wrong type {}'.format(etype)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   298
        data = self.eids[eid]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   299
        data.update(kwargs)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   300
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   301
    def prepare_insert_relation(self, eid_from, rtype, eid_to, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   302
        """Store into the `relations` attribute that a relation ``rtype`` exists between entities
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   303
        with eids ``eid_from`` and ``eid_to``.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   304
        """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   305
        relation = eid_from, rtype, eid_to
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   306
        self.relations.add(relation)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   307
        return relation
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   308
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   309
    def flush(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   310
        """Nothing to flush for this store."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   311
        pass
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   312
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   313
    def commit(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   314
        """Nothing to commit for this store."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   315
        return
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   316
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   317
    def finish(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   318
        """Nothing to do once import is terminated for this store."""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   319
        pass
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   320
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   321
    @property
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   322
    def nb_inserted_entities(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   323
        return len(self.eids)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   324
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   325
    @property
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   326
    def nb_inserted_types(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   327
        return len(self.types)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   328
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   329
    @property
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   330
    def nb_inserted_relations(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   331
        return len(self.relations)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   332
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   333
    @deprecated('[3.21] use prepare_insert_entity instead')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   334
    def create_entity(self, etype, **data):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   335
        self.prepare_insert_entity(etype, **data)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   336
        return attrdict(data)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   337
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   338
    @deprecated('[3.21] use prepare_insert_relation instead')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   339
    def relate(self, eid_from, rtype, eid_to, **kwargs):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   340
        self.prepare_insert_relation(eid_from, rtype, eid_to, **kwargs)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   341
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   342
10688
fa29f3628a1b [py3k] use six.add_metaclass
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10669
diff changeset
   343
@add_metaclass(class_deprecated)
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   344
class CWImportController(object):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   345
    """Controller of the data import process.
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   346
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   347
    >>> ctl = CWImportController(store)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   348
    >>> ctl.generators = list_of_data_generators
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   349
    >>> ctl.data = dict_of_data_tables
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   350
    >>> ctl.run()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   351
    """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   352
    __deprecation_warning__ = '[3.21] use the new importer API'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   353
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   354
    def __init__(self, store, askerror=0, catcherrors=None, tell=tell,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   355
                 commitevery=50):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   356
        self.store = store
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   357
        self.generators = None
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   358
        self.data = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   359
        self.errors = None
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   360
        self.askerror = askerror
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   361
        if  catcherrors is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   362
            catcherrors = askerror
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   363
        self.catcherrors = catcherrors
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   364
        self.commitevery = commitevery # set to None to do a single commit
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   365
        self._tell = tell
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   366
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   367
    def check(self, type, key, value):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   368
        self._checks.setdefault(type, {}).setdefault(key, []).append(value)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   369
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   370
    def check_map(self, entity, key, map, default):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   371
        try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   372
            entity[key] = map[entity[key]]
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   373
        except KeyError:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   374
            self.check(key, entity[key], None)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   375
            entity[key] = default
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   376
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   377
    def record_error(self, key, msg=None, type=None, value=None, tb=None):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   378
        tmp = StringIO()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   379
        if type is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   380
            traceback.print_exc(file=tmp)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   381
        else:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   382
            traceback.print_exception(type, value, tb, file=tmp)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   383
        # use a list to avoid counting a <nb lines> errors instead of one
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   384
        errorlog = self.errors.setdefault(key, [])
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   385
        if msg is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   386
            errorlog.append(tmp.getvalue().splitlines())
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   387
        else:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   388
            errorlog.append( (msg, tmp.getvalue().splitlines()) )
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   389
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   390
    def run(self):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   391
        self.errors = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   392
        if self.commitevery is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   393
            self.tell('Will commit all or nothing.')
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   394
        else:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   395
            self.tell('Will commit every %s iterations' % self.commitevery)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   396
        for func, checks in self.generators:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   397
            self._checks = {}
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   398
            func_name = func.__name__
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   399
            self.tell("Run import function '%s'..." % func_name)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   400
            try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   401
                func(self)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   402
            except Exception:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   403
                if self.catcherrors:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   404
                    self.record_error(func_name, 'While calling %s' % func.__name__)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   405
                else:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   406
                    self._print_stats()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   407
                    raise
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   408
            for key, func, title, help in checks:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   409
                buckets = self._checks.get(key)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   410
                if buckets:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   411
                    err = func(buckets)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   412
                    if err:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   413
                        self.errors[title] = (help, err)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   414
        try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   415
            txuuid = self.store.commit()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   416
            if txuuid is not None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   417
                self.tell('Transaction commited (txuuid: %s)' % txuuid)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   418
        except QueryError as ex:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   419
            self.tell('Transaction aborted: %s' % ex)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   420
        self._print_stats()
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   421
        if self.errors:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   422
            if self.askerror == 2 or (self.askerror and confirm('Display errors ?')):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   423
                from pprint import pformat
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   424
                for errkey, error in self.errors.items():
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   425
                    self.tell("\n%s (%s): %d\n" % (error[0], errkey, len(error[1])))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   426
                    self.tell(pformat(sorted(error[1])))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   427
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   428
    def _print_stats(self):
10663
54b8a1f249fb [py3k] dict.itervalues → dict.values
Rémi Cardona <remi.cardona@logilab.fr>
parents: 10590
diff changeset
   429
        nberrors = sum(len(err) for err in self.errors.values())
10513
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   430
        self.tell('\nImport statistics: %i entities, %i types, %i relations and %i errors'
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   431
                  % (self.store.nb_inserted_entities,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   432
                     self.store.nb_inserted_types,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   433
                     self.store.nb_inserted_relations,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   434
                     nberrors))
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   435
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   436
    def get_data(self, key):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   437
        return self.data.get(key)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   438
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   439
    def index(self, name, key, value, unique=False):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   440
        """create a new index
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   441
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   442
        If unique is set to True, only first occurence will be kept not the following ones
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   443
        """
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   444
        if unique:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   445
            try:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   446
                if value in self.store.indexes[name][key]:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   447
                    return
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   448
            except KeyError:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   449
                # we're sure that one is the first occurence; so continue...
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   450
                pass
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   451
        self.store.indexes.setdefault(name, {}).setdefault(key, []).append(value)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   452
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   453
    def tell(self, msg):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   454
        self._tell(msg)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   455
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   456
    def iter_and_commit(self, datakey):
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   457
        """iter rows, triggering commit every self.commitevery iterations"""
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   458
        if self.commitevery is None:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   459
            return self.get_data(datakey)
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   460
        else:
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   461
            return callfunc_every(self.store.commit,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   462
                                  self.commitevery,
7bec01a59f92 [dataimport] dispatch and deprecate old code
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents:
diff changeset
   463
                                  self.get_data(datakey))