server/checkintegrity.py
author Christophe de Vienne <cdevienne@gmail.com>
Thu, 12 Dec 2013 16:17:25 +0100
changeset 9382 ac0661486568
parent 9171 be9596750678
child 9463 d62e13eba033
child 9492 c7fc56eecd1a
permissions -rw-r--r--
[release notes] Documents one() and find() addition
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
8562
0d2fb4604265 [session] fix arguments default value and promote usage of security_enabled as session method. Closes #2481820
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8527
diff changeset
     1
# 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: 5341
diff changeset
     2
# 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: 5341
diff changeset
     3
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5341
diff changeset
     4
# 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: 5341
diff changeset
     5
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5341
diff changeset
     6
# 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: 5341
diff changeset
     7
# 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: 5341
diff changeset
     8
# 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: 5341
diff changeset
     9
# any later version.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5341
diff changeset
    10
#
5424
8ecbcbff9777 replace logilab-common by CubicWeb in disclaimer
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5421
diff changeset
    11
# 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: 5341
diff changeset
    12
# 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: 5341
diff changeset
    13
# 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: 5341
diff changeset
    14
# details.
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5341
diff changeset
    15
#
8167de96c523 proper licensing information (LGPL-2.1). Hope I get it right this time.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5341
diff changeset
    16
# 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: 5341
diff changeset
    17
# with CubicWeb.  If not, see <http://www.gnu.org/licenses/>.
6127
747e423093fc [ms, c-c] new command checking for consistency / potentian flaws and enhancements of mapping file of a multi-sources instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6112
diff changeset
    18
"""Integrity checking tool for instances:
747e423093fc [ms, c-c] new command checking for consistency / potentian flaws and enhancements of mapping file of a multi-sources instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6112
diff changeset
    19
747e423093fc [ms, c-c] new command checking for consistency / potentian flaws and enhancements of mapping file of a multi-sources instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6112
diff changeset
    20
* integrity of a CubicWeb repository. Hum actually only the system database is
747e423093fc [ms, c-c] new command checking for consistency / potentian flaws and enhancements of mapping file of a multi-sources instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6112
diff changeset
    21
  checked.
5999
eaf8219f8b7d [migration] fix rename_entity_type to avoid to loose some relations on the way
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5954
diff changeset
    22
"""
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    23
__docformat__ = "restructuredtext en"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    24
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    25
import sys
1016
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
    26
from datetime import datetime
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    27
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    28
from logilab.common.shellutils import ProgressBar
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    29
7468
c202aaf71489 [server] fix NameError (missing VIRTUAL_RTYPES import) in checkintegrity.py
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 7040
diff changeset
    30
from cubicweb.schema import PURE_VIRTUAL_RTYPES, VIRTUAL_RTYPES
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
    31
from cubicweb.server.sqlutils import SQL_PREFIX
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
    32
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
    33
def notify_fixed(fix):
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
    34
    if fix:
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
    35
        sys.stderr.write(' [FIXED]')
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
    36
    sys.stderr.write('\n')
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
    37
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    38
def has_eid(session, sqlcursor, eid, eids):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    39
    """return true if the eid is a valid eid"""
5341
0de53140bd29 [db-check] cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5340
diff changeset
    40
    if eid in eids:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    41
        return eids[eid]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    42
    sqlcursor.execute('SELECT type, source FROM entities WHERE eid=%s' % eid)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    43
    try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    44
        etype, source = sqlcursor.fetchone()
7815
2a164a9cf81c [exceptions] stop catching any exception in various places (closes #1942716)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7687
diff changeset
    45
    except Exception:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    46
        eids[eid] = False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    47
        return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    48
    if source and source != 'system':
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    49
        try:
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    50
            # insert eid *and* etype to attempt checking entity has not been
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    51
            # replaced by another subsquently to a restore of an old dump
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    52
            if session.execute('Any X WHERE X is %s, X eid %%(x)s' % etype,
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    53
                               {'x': eid}):
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    54
                eids[eid] = True
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    55
                return True
7815
2a164a9cf81c [exceptions] stop catching any exception in various places (closes #1942716)
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7687
diff changeset
    56
        except Exception: # TypeResolverError, Unauthorized...
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    57
            pass
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    58
        eids[eid] = False
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    59
        return False
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
    60
    if etype not in session.vreg.schema:
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
    61
        eids[eid] = False
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
    62
        return False
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
    63
    sqlcursor.execute('SELECT * FROM %s%s WHERE %seid=%s' % (SQL_PREFIX, etype,
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
    64
                                                             SQL_PREFIX, eid))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    65
    result = sqlcursor.fetchall()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    66
    if len(result) == 0:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    67
        eids[eid] = False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    68
        return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    69
    elif len(result) > 1:
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
    70
        msg = ('  More than one entity with eid %s exists in source !\n'
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
    71
               '  WARNING : Unable to fix this, do it yourself !\n')
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
    72
        sys.stderr.write(msg % eid)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    73
    eids[eid] = True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    74
    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    75
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    76
# XXX move to yams?
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    77
def etype_fti_containers(eschema, _done=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    78
    if _done is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    79
        _done = set()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    80
    _done.add(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    81
    containers = tuple(eschema.fulltext_containers())
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    82
    if containers:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    83
        for rschema, target in containers:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    84
            if target == 'object':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    85
                targets = rschema.objects(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    86
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    87
                targets = rschema.subjects(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    88
            for targeteschema in targets:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    89
                if targeteschema in _done:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    90
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    91
                _done.add(targeteschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    92
                for container in etype_fti_containers(targeteschema, _done):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    93
                    yield container
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    94
    else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    95
        yield eschema
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
    96
5850
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
    97
def reindex_entities(schema, session, withpb=True, etypes=None):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    98
    """reindex all entities in the repository"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    99
    # deactivate modification_date hook since we don't want them
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   100
    # to be updated due to the reindexation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   101
    repo = session.repo
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: 7040
diff changeset
   102
    cursor = session.cnxset['system']
5954
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   103
    dbhelper = session.repo.system_source.dbhelper
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   104
    if not dbhelper.has_fti_table(cursor):
2248
cbf043a2134a try to create fti table if not existant on rebuild-fti
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1977
diff changeset
   105
        print 'no text index table'
5954
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   106
        dbhelper.init_fti(cursor)
4806
4f12f59b1a13 [fti] refactor and fix full text indexation handling
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4691
diff changeset
   107
    repo.system_source.do_fti = True  # ensure full-text indexation is activated
5850
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   108
    if etypes is None:
5954
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   109
        print 'Reindexing entities'
5850
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   110
        etypes = set()
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   111
        for eschema in schema.entities():
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   112
            if eschema.final:
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   113
                continue
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   114
            indexable_attrs = tuple(eschema.indexable_attributes()) # generator
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   115
            if not indexable_attrs:
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   116
                continue
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   117
            for container in etype_fti_containers(eschema):
fabff2813ee4 [migration] schema should be accessed through .repo
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5693
diff changeset
   118
                etypes.add(container)
5954
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   119
        # clear fti table first
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   120
        session.system_sql('DELETE FROM %s' % dbhelper.fti_table)
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   121
    else:
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   122
        print 'Reindexing entities of type %s' % \
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   123
              ', '.join(sorted(str(e) for e in etypes))
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   124
        # clear fti table first. Use subquery for sql compatibility
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   125
        session.system_sql("DELETE FROM %s WHERE EXISTS(SELECT 1 FROM ENTITIES "
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   126
                           "WHERE eid=%s AND type IN (%s))" % (
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   127
                               dbhelper.fti_table, dbhelper.fti_uid_attr,
987086484876 [fti migration] test and fix reindexation of some specific entity types
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5850
diff changeset
   128
                               ','.join("'%s'" % etype for etype in etypes)))
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   129
    if withpb:
6112
913979c79244 [db-fti-index] simple fix fpr progressbar-related crash when etypes is None in reindex_entities()
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 5999
diff changeset
   130
        pb = ProgressBar(len(etypes) + 1)
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   131
        pb.update()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   132
    # reindex entities by generating rql queries which set all indexable
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   133
    # attribute to their current value
4816
c02583cb80a9 repair stuff broken by fti handling changes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4806
diff changeset
   134
    source = repo.system_source
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   135
    for eschema in etypes:
8450
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   136
        etype_class = session.vreg['etypes'].etype_class(str(eschema))
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   137
        for fti_rql in etype_class.cw_fti_index_rql_queries(session):
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   138
            rset = session.execute(fti_rql)
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   139
            source.fti_index_entities(session, rset.entities())
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   140
            # clear entity cache to avoid high memory consumption on big tables
11063635c4e4 [fti] allow usage of custom RQL to fetch entities to index (closes #2410509)
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 8447
diff changeset
   141
            session.drop_entity_cache()
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   142
        if withpb:
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   143
            pb.update()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   144
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   145
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   146
def check_schema(schema, session, eids, fix=1):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   147
    """check serialized schema"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   148
    print 'Checking serialized schema'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   149
    unique_constraints = ('SizeConstraint', 'FormatConstraint',
5523
4bf975c049a6 [db-check] RQLConstraint is not a 'unique' constraint
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   150
                          'VocabularyConstraint',
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   151
                          'RQLVocabularyConstraint')
5338
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   152
    rql = ('Any COUNT(X),RN,SN,ON,CTN GROUPBY RN,SN,ON,CTN ORDERBY 1 '
1398
5fe84a5f7035 rename internal entity types to have CW prefix instead of E
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
   153
           'WHERE X is CWConstraint, R constrained_by X, '
5338
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   154
           'R relation_type RT, RT name RN, R from_entity ST, ST name SN, '
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   155
           'R to_entity OT, OT name ON, X cstrtype CT, CT name CTN')
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   156
    for count, rn, sn, on, cstrname in session.execute(rql):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   157
        if count == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   158
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   159
        if cstrname in unique_constraints:
5338
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   160
            print "ERROR: got %s %r constraints on relation %s.%s.%s" % (
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   161
                count, cstrname, sn, rn, on)
5523
4bf975c049a6 [db-check] RQLConstraint is not a 'unique' constraint
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   162
            if fix:
4bf975c049a6 [db-check] RQLConstraint is not a 'unique' constraint
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   163
                print 'dunno how to fix, do it yourself'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   164
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   165
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   166
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   167
def check_text_index(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   168
    """check all entities registered in the text index"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   169
    print 'Checking text index'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   170
    msg = '  Entity with eid %s exists in the text index but in no source (autofix will remove from text index)'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   171
    cursor = session.system_sql('SELECT uid FROM appears;')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   172
    for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   173
        eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   174
        if not has_eid(session, cursor, eid, eids):
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
   175
            sys.stderr.write(msg % eid)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   176
            if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   177
                session.system_sql('DELETE FROM appears WHERE uid=%s;' % eid)
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   178
            notify_fixed(fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   179
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   180
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   181
def check_entities(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   182
    """check all entities registered in the repo system table"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   183
    print 'Checking entities system table'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   184
    # system table but no source
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   185
    msg = '  Entity %s with eid %s exists in the system table but in no source (autofix will delete the entity)'
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   186
    cursor = session.system_sql('SELECT eid,type FROM entities;')
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   187
    for row in cursor.fetchall():
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   188
        eid, etype = row
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   189
        if not has_eid(session, cursor, eid, eids):
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   190
            sys.stderr.write(msg % (etype, eid))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   191
            if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   192
                session.system_sql('DELETE FROM entities WHERE eid=%s;' % eid)
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   193
            notify_fixed(fix)
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   194
    # source in entities, but no relation cw_source
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   195
    applcwversion = session.repo.get_versions().get('cubicweb')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   196
    if applcwversion >= (3,13,1): # entities.asource appeared in 3.13.1
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   197
        cursor = session.system_sql('SELECT e.eid FROM entities as e, cw_CWSource as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   198
                                    'WHERE s.cw_name=e.asource AND '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   199
                                    'NOT EXISTS(SELECT 1 FROM cw_source_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   200
                                    '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   201
                                    'ORDER BY e.eid')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   202
        msg = ('  Entity with eid %s refers to source in entities table, '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   203
               'but is missing relation cw_source (autofix will create the relation)\n')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   204
        for row in cursor.fetchall():
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   205
            sys.stderr.write(msg % row[0])
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   206
        if fix:
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   207
            session.system_sql('INSERT INTO cw_source_relation (eid_from, eid_to) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   208
                               'SELECT e.eid, s.cw_eid FROM entities as e, cw_CWSource as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   209
                               'WHERE s.cw_name=e.asource AND NOT EXISTS(SELECT 1 FROM cw_source_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   210
                               '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid)')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   211
            notify_fixed(True)
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   212
    # inconsistencies for 'is'
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   213
    msg = '  %s #%s is missing relation "is" (autofix will create the relation)\n'
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   214
    cursor = session.system_sql('SELECT e.type, e.eid FROM entities as e, cw_CWEType as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   215
                                'WHERE s.cw_name=e.type AND NOT EXISTS(SELECT 1 FROM is_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   216
                                '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   217
                                'ORDER BY e.eid')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   218
    for row in cursor.fetchall():
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   219
        sys.stderr.write(msg % row)
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   220
    if fix:
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   221
        session.system_sql('INSERT INTO is_relation (eid_from, eid_to) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   222
                           'SELECT e.eid, s.cw_eid FROM entities as e, cw_CWEType as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   223
                           'WHERE s.cw_name=e.type AND NOT EXISTS(SELECT 1 FROM is_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   224
                           '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid)')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   225
        notify_fixed(True)
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   226
    # inconsistencies for 'is_instance_of'
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   227
    msg = '  %s #%s is missing relation "is_instance_of" (autofix will create the relation)\n'
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   228
    cursor = session.system_sql('SELECT e.type, e.eid FROM entities as e, cw_CWEType as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   229
                                'WHERE s.cw_name=e.type AND NOT EXISTS(SELECT 1 FROM is_instance_of_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   230
                                '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   231
                                'ORDER BY e.eid')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   232
    for row in cursor.fetchall():
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   233
        sys.stderr.write(msg % row)
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   234
    if fix:
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   235
        session.system_sql('INSERT INTO is_instance_of_relation (eid_from, eid_to) '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   236
                           'SELECT e.eid, s.cw_eid FROM entities as e, cw_CWEType as s '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   237
                           'WHERE s.cw_name=e.type AND NOT EXISTS(SELECT 1 FROM is_instance_of_relation as cs '
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   238
                           '  WHERE cs.eid_from=e.eid AND cs.eid_to=s.cw_eid)')
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   239
        notify_fixed(True)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   240
    print 'Checking entities tables'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   241
    msg = '  Entity with eid %s exists in the %s table but not in the system table (autofix will delete the entity)'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   242
    for eschema in schema.entities():
3689
deb13e88e037 follow yams 0.25 api changes to improve performance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3374
diff changeset
   243
        if eschema.final:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   244
            continue
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   245
        table = SQL_PREFIX + eschema.type
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   246
        column = SQL_PREFIX +  'eid'
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   247
        cursor = session.system_sql('SELECT %s FROM %s;' % (column, table))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   248
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   249
            eid = row[0]
5341
0de53140bd29 [db-check] cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5340
diff changeset
   250
            # eids is full since we have fetched everything from the entities table,
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   251
            # no need to call has_eid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   252
            if not eid in eids or not eids[eid]:
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
   253
                sys.stderr.write(msg % (eid, eschema.type))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   254
                if fix:
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   255
                    session.system_sql('DELETE FROM %s WHERE %s=%s;' % (table, column, eid))
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   256
                notify_fixed(fix)
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   257
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   258
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   259
def bad_related_msg(rtype, target, eid, fix):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   260
    msg = '  A relation %s with %s eid %s exists but no such entity in sources'
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
   261
    sys.stderr.write(msg % (rtype, target, eid))
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   262
    notify_fixed(fix)
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   263
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   264
def bad_inlined_msg(rtype, parent_eid, eid, fix):
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   265
    msg = ('  An inlined relation %s from %s to %s exists but the latter '
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   266
           'entity does not exist')
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   267
    sys.stderr.write(msg % (rtype, parent_eid, eid))
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   268
    notify_fixed(fix)
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   269
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   270
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   271
def check_relations(schema, session, eids, fix=1):
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   272
    """check that eids referenced by relations are registered in the repo system
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   273
    table
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   274
    """
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   275
    print 'Checking relations'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   276
    for rschema in schema.relations():
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   277
        if rschema.final or rschema.type in PURE_VIRTUAL_RTYPES:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   278
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   279
        if rschema.inlined:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   280
            for subjtype in rschema.subjects():
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   281
                table = SQL_PREFIX + str(subjtype)
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   282
                column = SQL_PREFIX +  str(rschema)
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   283
                sql = 'SELECT cw_eid,%s FROM %s WHERE %s IS NOT NULL;' % (
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   284
                    column, table, column)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   285
                cursor = session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   286
                for row in cursor.fetchall():
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   287
                    parent_eid, eid = row
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   288
                    if not has_eid(session, cursor, eid, eids):
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   289
                        bad_inlined_msg(rschema, parent_eid, eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   290
                        if fix:
3374
d5bd1b659ce8 [db-check] fix sql to fix bad eid referenced by inlined relation
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2596
diff changeset
   291
                            sql = 'UPDATE %s SET %s=NULL WHERE %s=%s;' % (
d5bd1b659ce8 [db-check] fix sql to fix bad eid referenced by inlined relation
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2596
diff changeset
   292
                                table, column, column, eid)
381
sylvain.thenault@logilab.fr
parents: 380
diff changeset
   293
                            session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   294
            continue
6185
229006accd26 [c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6132
diff changeset
   295
        try:
229006accd26 [c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6132
diff changeset
   296
            cursor = session.system_sql('SELECT eid_from FROM %s_relation;' % rschema)
8695
358d8bed9626 [toward-py3k] rewrite to "except AnException as exc:" (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8694
diff changeset
   297
        except Exception as ex:
6185
229006accd26 [c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6132
diff changeset
   298
            # usually because table doesn't exist
229006accd26 [c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6132
diff changeset
   299
            print 'ERROR', ex
229006accd26 [c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6132
diff changeset
   300
            continue
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   301
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   302
            eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   303
            if not has_eid(session, cursor, eid, eids):
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   304
                bad_related_msg(rschema, 'subject', eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   305
                if fix:
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   306
                    sql = 'DELETE FROM %s_relation WHERE eid_from=%s;' % (
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   307
                        rschema, eid)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   308
                    session.system_sql(sql)
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   309
        cursor = session.system_sql('SELECT eid_to FROM %s_relation;' % rschema)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   310
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   311
            eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   312
            if not has_eid(session, cursor, eid, eids):
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   313
                bad_related_msg(rschema, 'object', eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   314
                if fix:
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   315
                    sql = 'DELETE FROM %s_relation WHERE eid_to=%s;' % (
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   316
                        rschema, eid)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   317
                    session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   318
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   319
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   320
def check_mandatory_relations(schema, session, eids, fix=1):
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   321
    """check entities missing some mandatory relation"""
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   322
    print 'Checking mandatory relations'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   323
    msg = '%s #%s is missing mandatory %s relation %s (autofix will delete the entity)'
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   324
    for rschema in schema.relations():
8527
6151b0199bc1 [c-c db-check] don't skip is/is_instance_of mandatory relation to avoid telling two opposite messages when the relation is missing. Closes #2465659
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8450
diff changeset
   325
        if rschema.final or rschema in PURE_VIRTUAL_RTYPES or rschema in ('is', 'is_instance_of'):
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   326
            continue
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   327
        smandatory = set()
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   328
        omandatory = set()
8696
0bb18407c053 [toward py3k] rewrite dict.keys() and dict.values() (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8695
diff changeset
   329
        for rdef in rschema.rdefs.itervalues():
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   330
            if rdef.cardinality[0] in '1+':
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   331
                smandatory.add(rdef.subject)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   332
            if rdef.cardinality[1] in '1+':
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   333
                omandatory.add(rdef.object)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   334
        for role, etypes in (('subject', smandatory), ('object', omandatory)):
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   335
            for etype in etypes:
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   336
                if role == 'subject':
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   337
                    rql = 'Any X WHERE NOT X %s Y, X is %s' % (rschema, etype)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   338
                else:
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   339
                    rql = 'Any X WHERE NOT Y %s X, X is %s' % (rschema, etype)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   340
                for entity in session.execute(rql).entities():
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8696
diff changeset
   341
                    sys.stderr.write(msg % (entity.cw_etype, entity.eid, role, rschema))
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   342
                    if fix:
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   343
                        #if entity.cw_describe()['source']['uri'] == 'system': XXX
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   344
                        entity.cw_delete() # XXX this is BRUTAL!
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   345
                    notify_fixed(fix)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   346
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   347
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   348
def check_mandatory_attributes(schema, session, eids, fix=1):
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   349
    """check for entities stored in the system source missing some mandatory
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   350
    attribute
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   351
    """
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   352
    print 'Checking mandatory attributes'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   353
    msg = '%s #%s is missing mandatory attribute %s (autofix will delete the entity)'
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   354
    for rschema in schema.relations():
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   355
        if not rschema.final or rschema in VIRTUAL_RTYPES:
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   356
            continue
8696
0bb18407c053 [toward py3k] rewrite dict.keys() and dict.values() (part of #2711624)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 8695
diff changeset
   357
        for rdef in rschema.rdefs.itervalues():
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   358
            if rdef.cardinality[0] in '1+':
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   359
                rql = 'Any X WHERE X %s NULL, X is %s, X cw_source S, S name "system"' % (
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   360
                    rschema, rdef.subject)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   361
                for entity in session.execute(rql).entities():
8900
010a59e12d89 use cw_etype instead of __regid__
Pierre-Yves David <pierre-yves.david@logilab.fr>
parents: 8696
diff changeset
   362
                    sys.stderr.write(msg % (entity.cw_etype, entity.eid, rschema))
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   363
                    if fix:
7682
07b592a62be3 [c-c db-check] drop some deprecation warning
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7473
diff changeset
   364
                        entity.cw_delete()
7036
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   365
                    notify_fixed(fix)
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   366
63386b35ec69 [c-c db-check] new checks for entities missing a mandatory relation/attribute
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 7035
diff changeset
   367
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   368
def check_metadata(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   369
    """check entities has required metadata
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   370
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   371
    FIXME: rewrite using RQL queries ?
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   372
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   373
    print 'Checking metadata'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   374
    cursor = session.system_sql("SELECT DISTINCT type FROM entities;")
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   375
    eidcolumn = SQL_PREFIX + 'eid'
7965
d45c0eb39e72 [server] fix integrity checks that destroy data (closes #1972842)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7896
diff changeset
   376
    msg = '  %s with eid %s has no %s (autofix will set it to now)'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   377
    for etype, in cursor.fetchall():
9171
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   378
        if etype not in session.vreg.schema:
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   379
            sys.stderr.write('entities table references unknown type %s\n' %
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   380
                             etype)
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   381
            if fix:
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   382
                session.system_sql("DELETE FROM entities WHERE type = %(type)s",
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   383
                                   {'type': etype})
be9596750678 Fix two crashes in db-check (closes #3024964)
Denis Laxalde <denis.laxalde@logilab.fr>
parents: 8900
diff changeset
   384
            continue
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   385
        table = SQL_PREFIX + etype
1016
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
   386
        for rel, default in ( ('creation_date', datetime.now()),
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
   387
                              ('modification_date', datetime.now()), ):
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   388
            column = SQL_PREFIX + rel
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   389
            cursor = session.system_sql("SELECT %s FROM %s WHERE %s is NULL"
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   390
                                        % (eidcolumn, table, column))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   391
            for eid, in cursor.fetchall():
7896
4c954e1e73ef [lint] remove uses of "print >> sys.stderr" (closes #1908571)
Nicolas Chauvat <nicolas.chauvat@logilab.fr>
parents: 7815
diff changeset
   392
                sys.stderr.write(msg % (etype, eid, rel))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   393
                if fix:
1251
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   394
                    session.system_sql("UPDATE %s SET %s=%%(v)s WHERE %s=%s ;"
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   395
                                       % (table, column, eidcolumn, eid),
af40e615dc89 introduce a 'cw_' prefix on entity table and column names so we don't conflict with sql or DBMS specific keywords
sylvain.thenault@logilab.fr
parents: 1161
diff changeset
   396
                                       {'v': default})
7035
8d2cf36bd79d [c-c db-check] factorize code by introducing notify_fixed dumb function
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 6889
diff changeset
   397
                notify_fixed(fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   398
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   399
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   400
def check(repo, cnx, checks, reindex, fix, withpb=True):
2476
1294a6bdf3bf application -> instance where it makes sense
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2248
diff changeset
   401
    """check integrity of instance's repository,
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   402
    using given user and password to locally connect to the repository
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   403
    (no running cubicweb server needed)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   404
    """
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: 7040
diff changeset
   405
    session = repo._get_session(cnx.sessionid, setcnxset=True)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   406
    # yo, launch checks
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   407
    if checks:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   408
        eids_cache = {}
8562
0d2fb4604265 [session] fix arguments default value and promote usage of security_enabled as session method. Closes #2481820
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8527
diff changeset
   409
        with session.security_enabled(read=False, write=False): # ensure no read security
4835
13b0b96d7982 [repo] enhanced security handling: deprecates unsafe_execute, in favor of explicit read/write security control using the `enabled_security` context manager. Also code executed on the repository side is now unsafe by default.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4834
diff changeset
   410
            for check in checks:
13b0b96d7982 [repo] enhanced security handling: deprecates unsafe_execute, in favor of explicit read/write security control using the `enabled_security` context manager. Also code executed on the repository side is now unsafe by default.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4834
diff changeset
   411
                check_func = globals()['check_%s' % check]
13b0b96d7982 [repo] enhanced security handling: deprecates unsafe_execute, in favor of explicit read/write security control using the `enabled_security` context manager. Also code executed on the repository side is now unsafe by default.
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4834
diff changeset
   412
                check_func(repo.schema, session, eids_cache, fix=fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   413
        if fix:
8578
f6ea4ea63d57 [check integrity] use session consistently
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8562
diff changeset
   414
            session.commit()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   415
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   416
            print
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   417
        if not fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   418
            print 'WARNING: Diagnostic run, nothing has been corrected'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   419
    if reindex:
8578
f6ea4ea63d57 [check integrity] use session consistently
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8562
diff changeset
   420
        session.rollback()
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: 7040
diff changeset
   421
        session.set_cnxset()
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   422
        reindex_entities(repo.schema, session, withpb=withpb)
8578
f6ea4ea63d57 [check integrity] use session consistently
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 8562
diff changeset
   423
        session.commit()