server/checkintegrity.py
author Sylvain Thénault <sylvain.thenault@logilab.fr>
Thu, 17 Jun 2010 15:15:52 +0200
changeset 5774 0d792bceb25d
parent 5693 8af6623f3d4e
child 5850 fabff2813ee4
permissions -rw-r--r--
backport stable
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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
     1
# copyright 2003-2010 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
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/>.
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    18
"""Check integrity of a CubicWeb repository. Hum actually only the system database
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    19
is checked.
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    20
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    21
"""
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
    22
from __future__ import with_statement
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
    23
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    24
__docformat__ = "restructuredtext en"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    25
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    26
import sys
1016
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
    27
from datetime import datetime
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    28
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    29
from logilab.common.shellutils import ProgressBar
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    30
2596
d02eed70937f [R repo, schema] use VIRTUAL_RTYPES const
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 2476
diff changeset
    31
from cubicweb.schema import PURE_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
    32
from cubicweb.server.sqlutils import SQL_PREFIX
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
    33
from cubicweb.server.session import security_enabled
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
    34
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    35
def has_eid(session, sqlcursor, eid, eids):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    36
    """return true if the eid is a valid eid"""
5341
0de53140bd29 [db-check] cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5340
diff changeset
    37
    if eid in eids:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    38
        return eids[eid]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    39
    sqlcursor.execute('SELECT type, source FROM entities WHERE eid=%s' % eid)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    40
    try:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    41
        etype, source = sqlcursor.fetchone()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    42
    except:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    43
        eids[eid] = False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    44
        return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    45
    if source and source != 'system':
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    46
        try:
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    47
            # 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
    48
            # 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
    49
            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
    50
                               {'x': eid}):
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    51
                eids[eid] = True
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    52
                return True
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    53
        except: # TypeResolverError, Unauthorized...
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    54
            pass
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    55
        eids[eid] = False
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
    56
        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
    57
    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
    58
                                                             SQL_PREFIX, eid))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    59
    result = sqlcursor.fetchall()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    60
    if len(result) == 0:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    61
        eids[eid] = False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    62
        return False
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    63
    elif len(result) > 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    64
        msg = '  More than one entity with eid %s exists in source !'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    65
        print >> sys.stderr, msg % eid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    66
        print >> sys.stderr, '  WARNING : Unable to fix this, do it yourself !'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    67
    eids[eid] = True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    68
    return True
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    69
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    70
# XXX move to yams?
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    71
def etype_fti_containers(eschema, _done=None):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    72
    if _done is None:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    73
        _done = set()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    74
    _done.add(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    75
    containers = tuple(eschema.fulltext_containers())
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    76
    if containers:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    77
        for rschema, target in containers:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    78
            if target == 'object':
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    79
                targets = rschema.objects(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    80
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    81
                targets = rschema.subjects(eschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    82
            for targeteschema in targets:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    83
                if targeteschema in _done:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    84
                    continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    85
                _done.add(targeteschema)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    86
                for container in etype_fti_containers(targeteschema, _done):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    87
                    yield container
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    88
    else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    89
        yield eschema
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
    90
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
    91
def reindex_entities(schema, session, withpb=True):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    92
    """reindex all entities in the repository"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    93
    # deactivate modification_date hook since we don't want them
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    94
    # to be updated due to the reindexation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
    95
    repo = session.repo
2248
cbf043a2134a try to create fti table if not existant on rebuild-fti
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 1977
diff changeset
    96
    cursor = session.pool['system']
4831
c5aec27c1bf7 [repo] use logilab.db instead of lgc.adbh/lgc.db/lgc.sqlgen/indexer, test new date extranction functions
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4816
diff changeset
    97
    if not repo.system_source.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
    98
        print 'no text index table'
5693
8af6623f3d4e [pylint] fix detected name errors
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5523
diff changeset
    99
        repo.system_source.dbhelper.init_fti(cursor)
4806
4f12f59b1a13 [fti] refactor and fix full text indexation handling
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4691
diff changeset
   100
    repo.system_source.do_fti = True  # ensure full-text indexation is activated
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   101
    etypes = set()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   102
    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
   103
        if eschema.final:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   104
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   105
        indexable_attrs = tuple(eschema.indexable_attributes()) # generator
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   106
        if not indexable_attrs:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   107
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   108
        for container in etype_fti_containers(eschema):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   109
            etypes.add(container)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   110
    print 'Reindexing entities of type %s' % \
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   111
          ', '.join(sorted(str(e) for e in etypes))
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   112
    if withpb:
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   113
        pb = ProgressBar(len(etypes) + 1)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   114
    # first monkey patch Entity.check to disable validation
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   115
    # clear fti table first
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   116
    session.system_sql('DELETE FROM %s' % session.repo.system_source.dbhelper.fti_table)
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   117
    if withpb:
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   118
        pb.update()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   119
    # reindex entities by generating rql queries which set all indexable
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   120
    # attribute to their current value
4816
c02583cb80a9 repair stuff broken by fti handling changes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4806
diff changeset
   121
    source = repo.system_source
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   122
    for eschema in etypes:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   123
        for entity in session.execute('Any X WHERE X is %s' % eschema).entities():
4816
c02583cb80a9 repair stuff broken by fti handling changes
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4806
diff changeset
   124
            source.fti_index_entity(session, entity)
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   125
        if withpb:
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   126
            pb.update()
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   127
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   128
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   129
def check_schema(schema, session, eids, fix=1):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   130
    """check serialized schema"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   131
    print 'Checking serialized schema'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   132
    unique_constraints = ('SizeConstraint', 'FormatConstraint',
5523
4bf975c049a6 [db-check] RQLConstraint is not a 'unique' constraint
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   133
                          'VocabularyConstraint',
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   134
                          'RQLVocabularyConstraint')
5338
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   135
    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
   136
           '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
   137
           '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
   138
           '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
   139
    for count, rn, sn, on, cstrname in session.execute(rql):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   140
        if count == 1:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   141
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   142
        if cstrname in unique_constraints:
5338
3e5a256d17ba [db-check] fix duplicated schema constraint detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4835
diff changeset
   143
            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
   144
                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
   145
            if fix:
4bf975c049a6 [db-check] RQLConstraint is not a 'unique' constraint
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5424
diff changeset
   146
                print 'dunno how to fix, do it yourself'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   147
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   148
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   149
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   150
def check_text_index(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   151
    """check all entities registered in the text index"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   152
    print 'Checking text index'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   153
    cursor = session.system_sql('SELECT uid FROM appears;')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   154
    for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   155
        eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   156
        if not has_eid(session, cursor, eid, eids):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   157
            msg = '  Entity with eid %s exists in the text index but in no source'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   158
            print >> sys.stderr, msg % eid,
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   159
            if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   160
                session.system_sql('DELETE FROM appears WHERE uid=%s;' % eid)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   161
                print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   162
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   163
                print >> sys.stderr
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
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   166
def check_entities(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   167
    """check all entities registered in the repo system table"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   168
    print 'Checking entities system table'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   169
    cursor = session.system_sql('SELECT eid FROM entities;')
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   170
    for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   171
        eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   172
        if not has_eid(session, cursor, eid, eids):
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   173
            msg = '  Entity with eid %s exists in the system table but in no source'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   174
            print >> sys.stderr, msg % eid,
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   175
            if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   176
                session.system_sql('DELETE FROM entities WHERE eid=%s;' % eid)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   177
                print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   178
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   179
                print >> sys.stderr
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   180
    print 'Checking entities tables'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   181
    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
   182
        if eschema.final:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   183
            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
   184
        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
   185
        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
   186
        cursor = session.system_sql('SELECT %s FROM %s;' % (column, table))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   187
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   188
            eid = row[0]
5341
0de53140bd29 [db-check] cleanup
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5340
diff changeset
   189
            # 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
   190
            # no need to call has_eid
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   191
            if not eid in eids or not eids[eid]:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   192
                msg = '  Entity with eid %s exists in the %s table but not in the system table'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   193
                print >> sys.stderr, msg % (eid, eschema.type),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   194
                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
   195
                    session.system_sql('DELETE FROM %s WHERE %s=%s;' % (table, column, eid))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   196
                    print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   197
                else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   198
                    print >> sys.stderr
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   199
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   200
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   201
def bad_related_msg(rtype, target, eid, fix):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   202
    msg = '  A relation %s with %s eid %s exists but no such entity in sources'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   203
    print >> sys.stderr, msg % (rtype, target, eid),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   204
    if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   205
        print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   206
    else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   207
        print >> sys.stderr
1802
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   208
d628defebc17 delete-trailing-whitespace + some copyright update
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents: 1398
diff changeset
   209
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   210
def check_relations(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   211
    """check all relations registered in the repo system table"""
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   212
    print 'Checking relations'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   213
    for rschema in schema.relations():
3689
deb13e88e037 follow yams 0.25 api changes to improve performance
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 3374
diff changeset
   214
        if rschema.final or rschema in PURE_VIRTUAL_RTYPES:
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   215
            continue
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   216
        if rschema.inlined:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   217
            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
   218
                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
   219
                column = SQL_PREFIX +  str(rschema)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   220
                sql = 'SELECT %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
   221
                    column, table, column)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   222
                cursor = session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   223
                for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   224
                    eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   225
                    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
   226
                        bad_related_msg(rschema, 'object', eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   227
                        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
   228
                            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
   229
                                table, column, column, eid)
381
sylvain.thenault@logilab.fr
parents: 380
diff changeset
   230
                            session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   231
            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
   232
        cursor = session.system_sql('SELECT eid_from FROM %s_relation;' % rschema)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   233
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   234
            eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   235
            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
   236
                bad_related_msg(rschema, 'subject', eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   237
                if fix:
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   238
                    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
   239
                        rschema, eid)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   240
                    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
   241
        cursor = session.system_sql('SELECT eid_to FROM %s_relation;' % rschema)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   242
        for row in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   243
            eid = row[0]
5339
b83327846450 [db-check] fix unexistent multisource entity detection
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5338
diff changeset
   244
            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
   245
                bad_related_msg(rschema, 'object', eid, fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   246
                if fix:
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   247
                    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
   248
                        rschema, eid)
380
06e7f2932afe db-check fixes
sylvain.thenault@logilab.fr
parents: 0
diff changeset
   249
                    session.system_sql(sql)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   250
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   251
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   252
def check_metadata(schema, session, eids, fix=1):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   253
    """check entities has required metadata
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   254
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   255
    FIXME: rewrite using RQL queries ?
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   256
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   257
    print 'Checking metadata'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   258
    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
   259
    eidcolumn = SQL_PREFIX + 'eid'
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   260
    for etype, in cursor.fetchall():
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
   261
        table = SQL_PREFIX + etype
1016
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
   262
        for rel, default in ( ('creation_date', datetime.now()),
26387b836099 use datetime instead of mx.DateTime
sylvain.thenault@logilab.fr
parents: 713
diff changeset
   263
                              ('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
   264
            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
   265
            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
   266
                                        % (eidcolumn, table, column))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   267
            for eid, in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   268
                msg = '  %s with eid %s has no %s'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   269
                print >> sys.stderr, msg % (etype, eid, rel),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   270
                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
   271
                    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
   272
                                       % (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
   273
                                       {'v': default})
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   274
                    print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   275
                else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   276
                    print >> sys.stderr
1398
5fe84a5f7035 rename internal entity types to have CW prefix instead of E
sylvain.thenault@logilab.fr
parents: 1263
diff changeset
   277
    cursor = session.system_sql('SELECT MIN(%s) FROM %sCWUser;' % (eidcolumn,
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
   278
                                                                  SQL_PREFIX))
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   279
    default_user_eid = cursor.fetchone()[0]
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   280
    assert default_user_eid is not None, 'no user defined !'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   281
    for rel, default in ( ('owned_by', default_user_eid), ):
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   282
        cursor = session.system_sql("SELECT eid, type FROM entities "
5340
4de474016568 [db-check] don't check entities from external sources have owned_by
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 5339
diff changeset
   283
                                    "WHERE source='system' AND NOT EXISTS "
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   284
                                    "(SELECT 1 FROM %s_relation WHERE eid_from=eid);"
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   285
                                    % rel)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   286
        for eid, etype in cursor.fetchall():
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   287
            msg = '  %s with eid %s has no %s relation'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   288
            print >> sys.stderr, msg % (etype, eid, rel),
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   289
            if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   290
                session.system_sql('INSERT INTO %s_relation VALUES (%s, %s) ;'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   291
                                   % (rel, eid, default))
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   292
                print >> sys.stderr, ' [FIXED]'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   293
            else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   294
                print >> sys.stderr
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   295
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   296
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   297
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
   298
    """check integrity of instance's repository,
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   299
    using given user and password to locally connect to the repository
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   300
    (no running cubicweb server needed)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   301
    """
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   302
    session = repo._get_session(cnx.sessionid, setpool=True)
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   303
    # yo, launch checks
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   304
    if checks:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   305
        eids_cache = {}
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
   306
        with security_enabled(session, read=False): # ensure no read security
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
   307
            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
   308
                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
   309
                check_func(repo.schema, session, eids_cache, fix=fix)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   310
        if fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   311
            cnx.commit()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   312
        else:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   313
            print
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   314
        if not fix:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   315
            print 'WARNING: Diagnostic run, nothing has been corrected'
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   316
    if reindex:
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   317
        cnx.rollback()
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   318
        session.set_pool()
4675
9233a8350420 [test] don't display progress bar when testing checkintegrity
Sylvain Thénault <sylvain.thenault@logilab.fr>
parents: 4252
diff changeset
   319
        reindex_entities(repo.schema, session, withpb=withpb)
0
b97547f5f1fa Showtime !
Adrien Di Mascio <Adrien.DiMascio@logilab.fr>
parents:
diff changeset
   320
        cnx.commit()