SQLCatalog.py 105 KB
Newer Older
1
# -*- coding: utf-8 -*-
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2 3
##############################################################################
#
4
# Copyright (c) 2002-2009 Nexedi SARL. All Rights Reserved.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
5 6 7 8 9 10 11 12 13 14 15
# Copyright (c) 2001 Zope Corporation and Contributors. All Rights Reserved.
#
# This software is subject to the provisions of the Zope Public License,
# Version 2.0 (ZPL).  A copy of the ZPL should accompany this distribution.
# THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED
# WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
# WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS
# FOR A PARTICULAR PURPOSE
#
##############################################################################

16
from Persistence import Persistent, PersistentMapping
Jean-Paul Smets's avatar
Jean-Paul Smets committed
17 18
import Acquisition
import ExtensionClass
19
import OFS.History
20 21
from App.class_init import default__class_init__ as InitializeClass
from App.special_dtml import DTMLFile
22
from thread import allocate_lock, get_ident
23
from OFS.Folder import Folder
24
from AccessControl import ClassSecurityInfo
25
from AccessControl.SimpleObjectPolicies import ContainerAssertions
26
from BTrees.OIBTree import OIBTree
27 28
from App.config import getConfiguration
from BTrees.Length import Length
29
from Shared.DC.ZRDB.TM import TM
Jean-Paul Smets's avatar
Jean-Paul Smets committed
30

31
from Acquisition import aq_parent, aq_inner, aq_base
32
from zLOG import LOG, WARNING, INFO, TRACE, ERROR
33
from ZODB.POSException import ConflictError
34
from Products.PythonScripts.Utility import allow_class
Jean-Paul Smets's avatar
Jean-Paul Smets committed
35 36

import time
Jean-Paul Smets's avatar
Jean-Paul Smets committed
37
import sys
38 39
import urllib
import string
40
import pprint
41 42
import re
import warnings
43
from contextlib import contextmanager
44
from cStringIO import StringIO
45
from xml.dom.minidom import parse
46
from xml.sax.saxutils import escape, quoteattr
47
import os
48
from hashlib import md5
49

50
from interfaces.query_catalog import ISearchKeyCatalog
51 52
from zope.interface.verify import verifyClass
from zope.interface import implements
53

54
from SearchText import isAdvancedSearchText, dequote
55

56 57 58 59
# Try to import ActiveObject in order to make SQLCatalog active
try:
  from Products.CMFActivity.ActiveObject import ActiveObject
except ImportError:
60
  ActiveObject = ExtensionClass.Base
61

62 63 64 65
try:
  from Products.CMFCore.Expression import Expression
  from Products.PageTemplates.Expressions import getEngine
  from Products.CMFCore.utils import getToolByName
66
  new_context_search = re.compile(r'\bcontext\b').search
67 68 69 70
  withCMF = 1
except ImportError:
  withCMF = 0

71 72 73 74
try:
  import psyco
except ImportError:
  psyco = None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
75

76 77 78
@contextmanager
def noReadOnlyTransactionCache():
  yield
79
try:
80 81
  from Products.ERP5Type.Cache import \
    readOnlyTransactionCache, caching_instance_method
82
except ImportError:
83
  LOG('SQLCatalog', WARNING, 'Count not import caching_instance_method, expect slowness.')
84
  def caching_instance_method(*args, **kw):
85
    return lambda method: method
86
  readOnlyTransactionCache = noReadOnlyTransactionCache
87 88 89 90

try:
  from Products.ERP5Type.TransactionalVariable import getTransactionalVariable
except ImportError:
91
  LOG('SQLCatalog', WARNING, 'Count not import getTransactionalVariable, expect slowness.')
92
  def getTransactionalVariable():
93 94
    return {}

95 96
def getInstanceID(instance):
  # XXX: getPhysicalPath is overkill for a unique cache identifier.
97 98 99
  # What I would like to use instead of it is:
  #   (self._p_jar.db().database_name, self._p_oid)
  # but database_name is not unique in at least ZODB 3.4 (Zope 2.8.8).
100 101 102 103 104
  try:
    instance_id = instance._v_physical_path
  except AttributeError:
    instance._v_physical_path = instance_id = instance.getPhysicalPath()
  return instance_id
105 106 107

def generateCatalogCacheId(method_id, self, *args, **kwd):
  return str((method_id, self.getCacheSequenceNumber(), getInstanceID(self),
Julien Muchembled's avatar
Julien Muchembled committed
108
    args, kwd))
109

110 111 112 113 114 115 116 117 118 119
class transactional_cache_decorator:
  """
    Implements singleton-style caching.
    Wrapped method must have no parameters (besides "self").
  """
  def __init__(self, cache_id):
    self.cache_id = cache_id

  def __call__(self, method):
    def wrapper(wrapped_self):
120
      transactional_cache = getTransactionalVariable()
121
      cache_id = str((self.cache_id,
122
        wrapped_self.getCacheSequenceNumber(),
123 124
        getInstanceID(wrapped_self),
      ))
125
      try:
126
        result = transactional_cache[cache_id]
127
      except KeyError:
128 129
        result = transactional_cache[cache_id] = method(wrapped_self)
      return result
130 131
    return wrapper

132
list_type_list = list, tuple, set, frozenset
133 134 135 136
try:
  from ZPublisher.HTTPRequest import record
except ImportError:
  dict_type_list = (dict, )
137 138
else:
  dict_type_list = (dict, record)
139 140


141
UID_BUFFER_SIZE = 300
142
OBJECT_LIST_SIZE = 300 # XXX 300 is arbitrary value of catalog object list
143
MAX_PATH_LEN = 255
144
RESERVED_KEY_LIST = ('where_expression', 'sort-on', 'sort_on', 'sort-order', 'sort_order', 'limit',
145 146
                     'format', 'search_mode', 'operator', 'selection_domain', 'selection_report',
                     'extra_key_list', )
147

148
valid_method_meta_type_list = ('Z SQL Method', 'LDIF Method', 'Script (Python)') # Nicer
149

150
manage_addSQLCatalogForm = DTMLFile('dtml/addSQLCatalog',globals())
151

152 153 154 155 156
# Here go uid buffers
# Structure:
#  global_uid_buffer_dict[catalog_path][thread_id] = UidBuffer
global_uid_buffer_dict = {}

157
def manage_addSQLCatalog(self, id, title,
158
             vocab_id='create_default_catalog_', # vocab_id is a strange name - not abbreviation
159 160 161
             REQUEST=None):
  """Add a Catalog object
  """
162 163 164
  id = str(id)
  title = str(title)
  vocab_id = str(vocab_id)
165 166 167
  if vocab_id == 'create_default_catalog_':
    vocab_id = None

168
  c = Catalog(id, title, self)
169 170 171 172
  self._setObject(id, c)
  if REQUEST is not None:
    return self.manage_main(self, REQUEST,update_menu=1)

173 174
class UidBuffer(TM):
  """Uid Buffer class caches a list of reserved uids in a transaction-safe way."""
175

Yoshinori Okuji's avatar
Yoshinori Okuji committed
176
  def __init__(self):
177
    """Initialize some variables.
178

179
      temporary_buffer is used to hold reserved uids created by non-committed transactions.
180

181
      finished_buffer is used to hold reserved uids created by committed-transactions.
182

183
      This distinction is important, because uids by non-committed transactions might become
Yoshinori Okuji's avatar
Yoshinori Okuji committed
184
      invalid afterwards, so they may not be used by other transactions."""
185 186
    self.temporary_buffer = {}
    self.finished_buffer = []
187

188 189 190 191 192 193 194 195
  def _finish(self):
    """Move the uids in the temporary buffer to the finished buffer."""
    tid = get_ident()
    try:
      self.finished_buffer.extend(self.temporary_buffer[tid])
      del self.temporary_buffer[tid]
    except KeyError:
      pass
196

197 198 199 200 201 202 203
  def _abort(self):
    """Erase the uids in the temporary buffer."""
    tid = get_ident()
    try:
      del self.temporary_buffer[tid]
    except KeyError:
      pass
204

205 206 207 208 209 210 211 212
  def __len__(self):
    tid = get_ident()
    l = len(self.finished_buffer)
    try:
      l += len(self.temporary_buffer[tid])
    except KeyError:
      pass
    return l
213

214 215 216 217 218 219 220 221 222 223 224
  def remove(self, value):
    self._register()
    for uid_list in self.temporary_buffer.values():
      try:
        uid_list.remove(value)
      except ValueError:
        pass
    try:
      self.finished_buffer.remove(value)
    except ValueError:
      pass
225

226 227 228 229
  def pop(self):
    self._register()
    tid = get_ident()
    try:
230
      uid = self.temporary_buffer[tid].pop()
231
    except (KeyError, IndexError):
232 233
      uid = self.finished_buffer.pop()
    return uid
234

235 236 237
  def extend(self, iterable):
    self._register()
    tid = get_ident()
Yoshinori Okuji's avatar
Yoshinori Okuji committed
238
    self.temporary_buffer.setdefault(tid, []).extend(iterable)
239

240 241 242 243
class DummyDict(dict):
  def __setitem__(self, key, value):
    pass

244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
class LazyIndexationParameterList(tuple):
  def __new__(cls, document_list, attribute, global_cache):
    self = super(LazyIndexationParameterList, cls).__new__(cls)
    self._document_list = document_list
    self._attribute = attribute
    self._global_cache = global_cache
    return self

  def __getitem__(self, index):
    document = self._document_list[index]
    attribute = self._attribute
    global_cache_key = (document.uid, attribute)
    global_cache = self._global_cache
    if global_cache_key in global_cache:
      value = global_cache[global_cache_key]
    else:
      value = getattr(document, attribute, None)
      if callable(value):
        try:
          value = value()
        except ConflictError:
          raise
        except Exception:
          LOG('SQLCatalog', WARNING,
            'Failed to call method %s on %r' % (attribute, document),
            error=True,
          )
          value = None
      global_cache[global_cache_key] = value
    return value

  def __iter__(self):
    for index in xrange(len(self)):
      yield self[index]

  def __len__(self):
    return len(self._document_list)

  def __repr__(self):
    return '<%s(%i documents, %r) at %x>' % (self.__class__.__name__,
      len(self), self._attribute, id(self))

ContainerAssertions[LazyIndexationParameterList] = 1

288
related_key_warned_column_set = set()
289

290 291 292
class Catalog(Folder,
              Persistent,
              Acquisition.Implicit,
293
              ActiveObject,
294
              OFS.History.Historical):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
295 296 297 298
  """ An Object Catalog

  An Object Catalog maintains a table of object metadata, and a
  series of manageable indexes to quickly search for objects
299
  (references in the metadata) that satisfy a search where_expression.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
300 301 302 303 304 305 306 307

  This class is not Zope specific, and can be used in any python
  program to build catalogs of objects.  Note that it does require
  the objects to be Persistent, and thus must be used with ZODB3.

  uid -> the (local) universal ID of objects
  path -> the (local) path of objects

308 309
  If you pass it a keyword argument which is present in sql_catalog_full_text_search_keys
  (in catalog properties), it does a MySQL full-text search.
310 311 312
  Additionally you can pass it a search_mode argument ('natural', 'in_boolean_mode'
  or 'with_query_expansion') to use an advanced search mode ('natural'
  is the default).
313 314 315
  search_mode arg can be given for all full_text keys, or for a specific key by naming
  the argument search_mode_KeyName, or even more specifically, search_mode_Table.Key
  or search_mode_Table_Key
316

317
 """
318

319
  implements(ISearchKeyCatalog)
320

321

322
  meta_type = "SQLCatalog"
323
  icon = 'misc_/ZCatalog/ZCatalog.gif' # FIXME: use a different icon
324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352
  security = ClassSecurityInfo()

  manage_options = (
    {'label': 'Contents',       # TAB: Contents
     'action': 'manage_main',
     'help': ('OFSP','ObjectManager_Contents.stx')},
    {'label': 'Catalog',      # TAB: Catalogged Objects
     'action': 'manage_catalogView',
     'help':('ZCatalog','ZCatalog_Cataloged-Objects.stx')},
    {'label': 'Properties',     # TAB: Properties
     'action': 'manage_propertiesForm',
     'help': ('OFSP','Properties.stx')},
    {'label': 'Filter',     # TAB: Filter
     'action': 'manage_catalogFilter',},
    {'label': 'Find Objects',     # TAB: Find Objects
     'action': 'manage_catalogFind',
     'help':('ZCatalog','ZCatalog_Find-Items-to-ZCatalog.stx')},
    {'label': 'Advanced',       # TAB: Advanced
     'action': 'manage_catalogAdvanced',
     'help':('ZCatalog','ZCatalog_Advanced.stx')},
    {'label': 'Undo',         # TAB: Undo
     'action': 'manage_UndoForm',
     'help': ('OFSP','Undo.stx')},
    {'label': 'Security',       # TAB: Security
     'action': 'manage_access',
     'help': ('OFSP','Security.stx')},
    {'label': 'Ownership',      # TAB: Ownership
     'action': 'manage_owner',
     'help': ('OFSP','Ownership.stx'),}
353
    ) + OFS.History.Historical.manage_options
354

355
  __ac_permissions__= (
356 357 358 359 360

    ('Manage ZCatalog Entries',
     ['manage_catalogObject', 'manage_uncatalogObject',

      'manage_catalogView', 'manage_catalogFind',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
361 362
      'manage_catalogFilter',
      'manage_catalogAdvanced',
363 364

      'manage_catalogReindex', 'manage_catalogFoundItems',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
365 366
      'manage_catalogClear',
      'manage_main',
367 368 369 370 371 372
      'manage_editFilter',
      ],
     ['Manager']),

    ('Search ZCatalog',
     ['searchResults', '__call__', 'uniqueValuesFor',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
373 374
      'all_meta_types', 'valid_roles',
      'getCatalogSearchTableIds',
375
      'getFilterableMethodList',],
376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
     ['Anonymous', 'Manager']),

    ('Import/Export objects',
     ['manage_exportProperties', 'manage_importProperties', ],
     ['Manager']),

    )

  _properties = (
    { 'id'      : 'title',
      'description' : 'The title of this catalog',
      'type'    : 'string',
      'mode'    : 'w' },

    # Z SQL Methods
    { 'id'      : 'sql_catalog_produce_reserved',
      'description' : 'A method to produce new uid values in advance',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_clear_reserved',
      'description' : 'A method to clear reserved uid values',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
401 402 403 404 405
    { 'id'      : 'sql_catalog_reserve_uid',
      'description' : 'A method to reserve a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
406 407 408 409 410
    { 'id'      : 'sql_catalog_delete_uid',
      'description' : 'A method to delete a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
411 412
    { 'id'      : 'sql_catalog_object_list',
      'description' : 'Methods to be called to catalog the list of objects',
413 414 415 416 417 418 419 420
      'type'    : 'multiple selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_uncatalog_object',
      'description' : 'Methods to be called to uncatalog an object',
      'type'    : 'multiple selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
421 422 423
    { 'id'      : 'sql_catalog_translation_list',
      'description' : 'Methods to be called to catalog the list of translation objects',
      'type'    : 'selection',
424 425
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
426 427 428
    { 'id'      : 'sql_delete_translation_list',
      'description' : 'Methods to be called to delete translations',
      'type'    : 'selection',
429 430
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
431 432
    { 'id'      : 'sql_clear_catalog',
      'description' : 'The methods which should be called to clear the catalog',
433 434 435
      'type'    : 'multiple selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
436
    { 'id'      : 'sql_record_object_list',
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455
      'description' : 'Method to record catalog information',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_read_recorded_object_list',
      'description' : 'Method to get recorded information',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_delete_recorded_object_list',
      'description' : 'Method to delete recorded information',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_search_results',
      'description' : 'Main method to search the catalog',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
Aurel's avatar
Aurel committed
456 457 458 459 460
    { 'id'      : 'sql_search_security',
      'description' : 'Main method to search security',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495
    { 'id'      : 'sql_search_tables',
      'description' : 'Tables to join in the result',
      'type'    : 'multiple selection',
      'select_variable' : 'getTableIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_search_result_keys',
      'description' : 'Keys to display in the result',
      'type'    : 'multiple selection',
      'select_variable' : 'getResultColumnIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_count_results',
      'description' : 'Main method to search the catalog',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_getitem_by_path',
      'description' : 'Get a catalog brain by path',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_getitem_by_uid',
      'description' : 'Get a catalog brain by uid',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_tables',
      'description' : 'Method to get the main catalog tables',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_schema',
      'description' : 'Method to get the main catalog schema',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
496 497 498 499 500
    { 'id'      : 'sql_catalog_index',
      'description' : 'Method to get the main catalog index',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
501 502 503 504 505 506 507 508 509 510
    { 'id'      : 'sql_unique_values',
      'description' : 'Find unique disctinct values in a column',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_paths',
      'description' : 'List all object paths in catalog',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
511 512 513 514 515
    { 'id': 'sql_catalog_search_keys',
      'title': 'Search Key Mappings',
      'description': 'A list of Search Key mappings',
      'type': 'lines',
      'mode': 'w' },
516
    { 'id'      : 'sql_catalog_keyword_search_keys',
517
      'description' : 'Columns which should be considered as keyword search',
518 519 520
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
521
    { 'id'      : 'sql_catalog_datetime_search_keys',
522
      'description' : 'Columns which should be considered as datetime search',
523 524 525
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
526 527 528 529 530 531 532 533 534 535 536 537 538 539 540
    { 'id'      : 'sql_catalog_full_text_search_keys',
      'description' : 'Columns which should be considered as full text search',
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_request_keys',
      'description' : 'Columns which should be ignore in the REQUEST in order to accelerate caching',
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_multivalue_keys',
      'description' : 'Keys which hold multiple values',
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
541 542 543 544 545
    { 'id'      : 'sql_catalog_index_on_order_keys',
      'description' : 'Columns which should be used by specifying an index when sorting on them',
      'type'    : 'multiple selection',
      'select_variable' : 'getSortColumnIds',
      'mode'    : 'w' },
546 547 548 549 550 551 552 553 554
    { 'id'      : 'sql_catalog_topic_search_keys',
      'description' : 'Columns which should be considered as topic index',
      'type'    : 'lines',
      'mode'    : 'w' },
    { 'id'      : 'sql_catalog_related_keys',
      'title'   : 'Related keys',
      'description' : 'Additional columns obtained through joins',
      'type'    : 'lines',
      'mode'    : 'w' },
555 556 557 558 559
    { 'id'      : 'sql_catalog_scriptable_keys',
      'title'   : 'Related keys',
      'description' : 'Virtual columns to generate scriptable scriptable queries',
      'type'    : 'lines',
      'mode'    : 'w' },
560 561 562 563 564 565 566 567 568 569 570
    { 'id': 'sql_catalog_role_keys',
      'title': 'Role keys',
      'description': 'Columns which should be used to map a monovalued role',
      'type': 'lines',
      'mode': 'w' },
    { 'id': 'sql_catalog_local_role_keys',
      'title': 'Local Role keys',
      'description': 'Columns which should be used to map' \
                      'a monovalued local role',
      'type': 'lines',
      'mode': 'w' },
571 572 573 574 575 576 577 578
    { 'id': 'sql_catalog_security_uid_columns',
      'title': 'Security Uid Columns',
      'description': 'A list of mappings "local_roles_group_id | security_uid_column"'
                     ' local_roles_group_id will be the name of a local roles'
                     ' group, and security_uid_column is the corresponding'
                     ' column in catalog table',
      'type': 'lines',
      'mode': 'w' },
579 580 581 582 583 584
    { 'id': 'sql_catalog_table_vote_scripts',
      'title': 'Table vote scripts',
      'description': 'Scripts helping column mapping resolution',
      'type': 'multiple selection',
      'select_variable' : 'getPythonMethodIds',
      'mode': 'w' },
585 586 587 588 589 590
    { 'id': 'sql_catalog_raise_error_on_uid_check',
      'title': 'Raise error on UID check',
      'description': 'Boolean used to tell if we raise error when uid check fail',
      'type': 'boolean',
      'default' : True,
      'mode': 'w' },
591

592 593
  )

594
  sql_catalog_produce_reserved = ''
595
  sql_catalog_delete_uid = ''
596 597 598 599 600 601 602 603 604 605 606
  sql_catalog_clear_reserved = ''
  sql_catalog_reserve_uid = ''
  sql_catalog_object_list = ()
  sql_uncatalog_object = ()
  sql_clear_catalog = ()
  sql_catalog_translation_list = ''
  sql_delete_translation_list = ''
  sql_record_object_list = ''
  sql_read_recorded_object_list = ''
  sql_delete_recorded_object_list = ''
  sql_search_results = ''
Aurel's avatar
Aurel committed
607
  sql_search_security = ''
608 609 610 611 612 613
  sql_count_results = ''
  sql_getitem_by_path = ''
  sql_getitem_by_uid = ''
  sql_catalog_tables = ''
  sql_search_tables = ()
  sql_catalog_schema = ''
614
  sql_catalog_index = ''
615 616
  sql_unique_values = ''
  sql_catalog_paths = ''
617
  sql_catalog_search_keys = ()
618
  sql_catalog_keyword_search_keys =  ()
619
  sql_catalog_datetime_search_keys = ()
620 621 622 623 624
  sql_catalog_full_text_search_keys = ()
  sql_catalog_request_keys = ()
  sql_search_result_keys = ()
  sql_catalog_topic_search_keys = ()
  sql_catalog_multivalue_keys = ()
625
  sql_catalog_index_on_order_keys = ()
626
  sql_catalog_related_keys = ()
627
  sql_catalog_scriptable_keys = ()
628 629
  sql_catalog_role_keys = ()
  sql_catalog_local_role_keys = ()
630
  sql_catalog_security_uid_columns = (' | security_uid',)
631
  sql_catalog_table_vote_scripts = ()
632
  sql_catalog_raise_error_on_uid_check = True
633

634 635 636 637 638 639
  # These are ZODB variables, so shared by multiple Zope instances.
  # This is set to the last logical time when clearReserved is called.
  _last_clear_reserved_time = 0
  # This is to record the maximum value of uids. Because this uses the class Length
  # in BTrees.Length, this does not generate conflict errors.
  _max_uid = None
640

641 642 643 644 645 646 647
  # These are class variable on memory, so shared only by threads in the same Zope instance.
  # This is set to the time when reserved uids are cleared in this Zope instance.
  _local_clear_reserved_time = None
  # This is used for exclusive access to the list of reserved uids.
  _reserved_uid_lock = allocate_lock()
  # This is an instance id which specifies who owns which reserved uids.
  _instance_id = getattr(getConfiguration(), 'instance_id', None)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
648

649 650 651 652 653
  manage_catalogView = DTMLFile('dtml/catalogView',globals())
  manage_catalogFilter = DTMLFile('dtml/catalogFilter',globals())
  manage_catalogFind = DTMLFile('dtml/catalogFind',globals())
  manage_catalogAdvanced = DTMLFile('dtml/catalogAdvanced', globals())

654 655
  _cache_sequence_number = 0

656 657 658 659 660
  def __init__(self, id, title='', container=None):
    if container is not None:
      self=self.__of__(container)
    self.id=id
    self.title=title
Jean-Paul Smets's avatar
Jean-Paul Smets committed
661 662 663
    self.schema = {}  # mapping from attribute name to column
    self.names = {}   # mapping from column to attribute name
    self.indexes = {}   # empty mapping
664
    self.filter_dict = PersistentMapping()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
665

666 667 668 669 670 671 672
  def manage_afterClone(self, item):
    try:
      del self._v_physical_path
    except AttributeError:
      pass
    super(Catalog, self).manage_afterClone(item)

673 674 675 676 677 678
  def getCacheSequenceNumber(self):
    return self._cache_sequence_number

  def _clearCaches(self):
    self._cache_sequence_number += 1

679 680 681 682
  def getSQLCatalogRoleKeysList(self):
    """
    Return the list of role keys.
    """
683 684 685 686 687
    role_key_dict = {}
    for role_key in self.sql_catalog_role_keys:
      role, column = role_key.split('|')
      role_key_dict[role.strip()] = column.strip()
    return role_key_dict.items()
688

689 690 691 692 693 694 695 696 697 698 699 700
  def getSQLCatalogSecurityUidGroupsColumnsDict(self):
    """
    Return a mapping of local_roles_group_id name to the name of the column
    storing corresponding security_uid.
    The default mappiny is {'': 'security_uid'}
    """
    local_roles_group_id_dict = {}
    for local_roles_group_id_key in self.sql_catalog_security_uid_columns:
      local_roles_group_id, column = local_roles_group_id_key.split('|')
      local_roles_group_id_dict[local_roles_group_id.strip()] = column.strip()
    return local_roles_group_id_dict

701 702 703 704
  def getSQLCatalogLocalRoleKeysList(self):
    """
    Return the list of local role keys.
    """
705 706 707 708
    local_role_key_dict = {}
    for role_key in self.sql_catalog_local_role_keys:
      role, column = role_key.split('|')
      local_role_key_dict[role.strip()] = column.strip()
709
    return local_role_key_dict.items()
710

711 712 713 714 715 716
  def manage_exportProperties(self, REQUEST=None, RESPONSE=None):
    """
      Export properties to an XML file.
    """
    f = StringIO()
    f.write('<?xml version="1.0"?>\n<SQLCatalogData>\n')
717 718 719 720 721 722 723 724
    property_id_list = self.propertyIds()
    # Get properties and values
    property_list = []
    for property_id in property_id_list:
      value = self.getProperty(property_id)
      if value is not None:
        property_list.append((property_id, value))
    # Sort for easy diff
725
    property_list.sort(key=lambda x: x[0])
726 727 728
    for property in property_list:
      property_id = property[0]
      value       = property[1]
729
      if isinstance(value, basestring):
730
        f.write('  <property id=%s type="str">%s</property>\n' % (quoteattr(property_id), escape(value)))
731
      elif isinstance(value, (tuple, list)):
732 733 734
        f.write('  <property id=%s type="tuple">\n' % quoteattr(property_id))
        # Sort for easy diff
        item_list = []
735
        for item in value:
736
          if isinstance(item, basestring):
737 738 739 740
            item_list.append(item)
        item_list.sort()
        for item in item_list:
          f.write('    <item type="str">%s</item>\n' % escape(str(item)))
741
        f.write('  </property>\n')
742 743 744
    # XXX Although filters are not properties, output filters here.
    # XXX Ideally, filters should be properties in Z SQL Methods, shouldn't they?
    if hasattr(self, 'filter_dict'):
745 746
      filter_list = []
      for filter_id in self.filter_dict.keys():
747
        filter_definition = self.filter_dict[filter_id]
748 749
        filter_list.append((filter_id, filter_definition))
      # Sort for easy diff
750
      filter_list.sort(key=lambda x: x[0])
751 752 753 754
      for filter_item in filter_list:
        filter_id  = filter_item[0]
        filter_def = filter_item[1]
        if not filter_def['filtered']:
755 756
          # If a filter is not activated, no need to output it.
          continue
757
        if not filter_def['expression']:
758 759
          # If the expression is not specified, meaningless to specify it.
          continue
760
        f.write('  <filter id=%s expression=%s />\n' % (quoteattr(filter_id), quoteattr(filter_def['expression'])))
761
        # For now, portal types are not exported, because portal types are too specific to each site.
762 763 764 765 766 767 768 769 770 771 772 773
    f.write('</SQLCatalogData>\n')

    if RESPONSE is not None:
      RESPONSE.setHeader('Content-type','application/data')
      RESPONSE.setHeader('Content-Disposition',
                          'inline;filename=properties.xml')
    return f.getvalue()

  def manage_importProperties(self, file):
    """
      Import properties from an XML file.
    """
774
    with open(file) as f:
775 776 777 778 779 780 781
      doc = parse(f)
      root = doc.documentElement
      try:
        for prop in root.getElementsByTagName("property"):
          id = prop.getAttribute("id")
          type = prop.getAttribute("type")
          if not id or not hasattr(self, id):
782
            raise CatalogError, 'unknown property id %r' % (id,)
783
          if type not in ('str', 'tuple'):
784
            raise CatalogError, 'unknown property type %r' % (type,)
785 786 787 788
          if type == 'str':
            value = ''
            for text in prop.childNodes:
              if text.nodeType == text.TEXT_NODE:
789
                value = str(text.data)
790 791 792 793 794 795
                break
          else:
            value = []
            for item in prop.getElementsByTagName("item"):
              item_type = item.getAttribute("type")
              if item_type != 'str':
796
                raise CatalogError, 'unknown item type %r' % (item_type,)
797 798
              for text in item.childNodes:
                if text.nodeType == text.TEXT_NODE:
799
                  value.append(str(text.data))
800 801 802 803
                  break
            value = tuple(value)

          setattr(self, id, value)
804

805 806 807
        if not hasattr(self, 'filter_dict'):
          self.filter_dict = PersistentMapping()
        for filt in root.getElementsByTagName("filter"):
808
          id = str(filt.getAttribute("id"))
809 810 811 812 813 814 815 816 817 818 819 820
          expression = filt.getAttribute("expression")
          if not self.filter_dict.has_key(id):
            self.filter_dict[id] = PersistentMapping()
          self.filter_dict[id]['filtered'] = 1
          self.filter_dict[id]['type'] = []
          if expression:
            expr_instance = Expression(expression)
            self.filter_dict[id]['expression'] = expression
            self.filter_dict[id]['expression_instance'] = expr_instance
          else:
            self.filter_dict[id]['expression'] = ""
            self.filter_dict[id]['expression_instance'] = None
821 822
      finally:
        doc.unlink()
Aurel's avatar
Aurel committed
823

824 825 826 827 828 829 830
  def manage_historyCompare(self, rev1, rev2, REQUEST,
                            historyComparisonResults=''):
    return Catalog.inheritedAttribute('manage_historyCompare')(
          self, rev1, rev2, REQUEST,
          historyComparisonResults=OFS.History.html_diff(
              pprint.pformat(rev1.__dict__),
              pprint.pformat(rev2.__dict__)))
831 832 833

  def _clearSecurityCache(self):
    self.security_uid_dict = OIBTree()
834
    self.security_uid_index = None
835

836 837 838 839
  def _clearSubjectCache(self):
    self.subject_set_uid_dict = OIBTree()
    self.subject_set_uid_index = None

840 841
  security.declarePrivate('getSecurityUidDict')
  def getSecurityUidDict(self, wrapped_object):
842
    """
843 844
    returns a tuple with a dict of security uid by local group id, and a tuple
    containing optimised_roles_and_users that might have been created.
845
    """
846
    if getattr(aq_base(self), 'security_uid_dict', None) is None:
847
      self._clearSecurityCache()
848 849

    optimised_roles_and_users = []
850
    local_roles_group_id_to_security_uid_mapping = {}
851 852

    # Get security information
853 854 855
    security_uid = None
    for key in wrapped_object.getLocalRolesGroupIdDict().iteritems():
      local_roles_group_id, allowed_roles_and_users = key
856 857 858 859 860 861 862 863 864 865 866 867
      allowed_roles_and_users = tuple(sorted(allowed_roles_and_users))
      if self.security_uid_dict.has_key(key):
        local_roles_group_id_to_security_uid_mapping[local_roles_group_id] \
                = self.security_uid_dict[key]
      elif self.security_uid_dict.has_key(allowed_roles_and_users)\
           and not local_roles_group_id:
        # This key is present in security_uid_dict without
        # local_roles_group_id, it has been inserted before
        # local_roles_group_id were introduced.
        local_roles_group_id_to_security_uid_mapping[local_roles_group_id] = \
          self.security_uid_dict[allowed_roles_and_users]
      else:
868 869 870 871 872 873 874 875 876 877 878
        if not security_uid:
          getTransactionalVariable().pop('getSecurityUidDictAndRoleColumnDict',
                                         None)
          id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
          # We must keep compatibility with existing sites
          security_uid = getattr(self, 'security_uid_index', None)
          if security_uid is None:
            security_uid = 0
          # At some point, it was a Length
          elif isinstance(security_uid, Length):
            security_uid = security_uid()
879 880 881 882
        # If the id_tool is there, it is better to use it, it allows
        # to create many new security uids by the same time
        # because with this tool we are sure that we will have 2 different
        # uids if two instances are doing this code in the same time
883
        security_uid += 1
884 885
        if id_tool is not None:
          security_uid = int(id_tool.generateNewId(id_generator='uid',
886
              id_group='security_uid_index', default=security_uid))
887
        else:
888 889 890 891 892 893 894 895 896 897 898 899
          self.security_uid_index = security_uid

        self.security_uid_dict[key] = security_uid
        local_roles_group_id_to_security_uid_mapping[local_roles_group_id]\
            = security_uid

        # If some optimised_roles_and_users are returned by this method it
        # means that new entries will have to be added to roles_and_users table.
        for user in allowed_roles_and_users:
          optimised_roles_and_users.append((security_uid, local_roles_group_id, user))

    return (local_roles_group_id_to_security_uid_mapping, optimised_roles_and_users)
900

901 902
  def getRoleAndSecurityUidList(self):
    """
903
      Return a list of 3-tuples, suitable for direct use in a zsqlmethod.
904 905 906
      Goal: make it possible to regenerate a table containing this data.
    """
    result = []
907 908
    for role_list, security_uid in getattr(
            aq_base(self), 'security_uid_dict', {}).iteritems():
909 910 911 912 913 914 915
      if role_list:
        if isinstance(role_list[-1], tuple):
          local_role_group_id, role_list = role_list
        else:
          local_role_group_id = ''
        result += [(local_role_group_id, role, security_uid)
                  for role in role_list]
916 917
    return result

918 919 920 921 922 923 924 925 926 927 928 929 930
  security.declarePrivate('getSubjectSetUid')
  def getSubjectSetUid(self, wrapped_object):
    """
    Cache a uid for each unique subject tuple.
    Return a tuple with a subject uid (string) and a new subject tuple
    if not exist already.
    """
    getSubjectList = getattr(wrapped_object, 'getSubjectList', None)
    if getSubjectList is None:
      return (None, None)
    # Get subject information
    # XXX if more collation is available, we can have smaller number of
    # unique subject sets.
931
    subject_list = tuple(sorted({(x or '').lower() for x in getSubjectList()}))
932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960
    if not subject_list:
      return (None, None)
    # Make sure no duplicates
    if getattr(aq_base(self), 'subject_set_uid_dict', None) is None:
      self._clearSubjectCache()
    elif self.subject_set_uid_dict.has_key(subject_list):
      return (self.subject_set_uid_dict[subject_list], None)
    # If the id_tool is there, it is better to use it, it allows
    # to create many new subject uids by the same time
    # because with this tool we are sure that we will have 2 different
    # uids if two instances are doing this code in the same time
    id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
    if id_tool is not None:
      default = 1
      # We must keep compatibility with existing sites
      previous_subject_set_uid = getattr(self, 'subject_set_uid_index', None)
      if previous_subject_set_uid is not None:
        default = previous_subject_set_uid
      subject_set_uid = int(id_tool.generateNewId(id_generator='uid',
          id_group='subject_set_uid_index', default=default))
    else:
      previous_subject_set_uid = getattr(self, 'subject_set_uid_index', None)
      if previous_subject_set_uid is None:
        previous_subject_set_uid = 0
      subject_set_uid = previous_subject_set_uid + 1
      self.subject_set_uid_index = subject_set_uid
    self.subject_set_uid_dict[subject_list] = subject_set_uid
    return (subject_set_uid, subject_list)

Jean-Paul Smets's avatar
Jean-Paul Smets committed
961 962 963 964 965 966
  def clear(self):
    """
    Clears the catalog by calling a list of methods
    """
    methods = self.sql_clear_catalog
    for method_name in methods:
967
      method = getattr(self, method_name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
968 969
      try:
        method()
970 971
      except ConflictError:
        raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
972
      except:
973
        LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
974
            'could not clear catalog with %s' % method_name, error=sys.exc_info())
975
        raise
976

977
    # Reserved uids have been removed.
978
    self.clearReserved()
979

980 981 982 983
    id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
    if id_tool is None:
      # Add a dummy item so that SQLCatalog will not use existing uids again.
      self.insertMaxUid()
984

985
    self._clearSecurityCache()
986
    self._clearSubjectCache()
987
    self._clearCaches()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
988

989 990 991 992 993 994 995 996 997 998
  def insertMaxUid(self):
    """
      Add a dummy item so that SQLCatalog will not use existing uids again.
    """
    if self._max_uid is not None and self._max_uid() != 0:
      method_id = self.sql_catalog_reserve_uid
      method = getattr(self, method_id)
      self._max_uid.change(1)
      method(uid = [self._max_uid()])

999 1000 1001 1002 1003 1004
  def clearReserved(self):
    """
    Clears reserved uids
    """
    method_id = self.sql_catalog_clear_reserved
    method = getattr(self, method_id)
Romain Courteaud's avatar
Romain Courteaud committed
1005 1006 1007 1008 1009
    try:
      method()
    except ConflictError:
      raise
    except:
1010
      LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
1011 1012 1013
          'could not clear reserved catalog with %s' % \
              method_id, error=sys.exc_info())
      raise
1014
    self._last_clear_reserved_time += 1
1015

1016
  def getRecordForUid(self, uid):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1017 1018 1019 1020
    """
    Get an object by UID
    Note: brain is defined in Z SQL Method object
    """
1021
    # this method used to be __getitem__(self, uid) but was found to hurt more
1022
    # than it helped: It would be inadvertently called by
1023 1024 1025 1026 1027 1028
    # (un)restrictedTraverse and if there was any error in rendering the SQL
    # expression or contacting the database, an error different from KeyError
    # would be raised, causing confusion.
    # It could also have a performance impact for traversals to objects in
    # the acquisition context on Zope 2.12 even when it didn't raise a weird
    # error.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1029 1030 1031 1032
    method = getattr(self,  self.sql_getitem_by_uid)
    search_result = method(uid = uid)
    if len(search_result) > 0:
      return search_result[0]
1033
    raise KeyError, uid
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053

  def editSchema(self, names_list):
    """
    Builds a schema from a list of strings
    Splits each string to build a list of attribute names
    Columns on the database should not change during this operations
    """
    i = 0
    schema = {}
    names = {}
    for cid in self.getColumnIds():
      name_list = []
      for name in names_list[i].split():
        schema[name] = cid
        name_list += [name,]
      names[cid] = tuple(name_list)
      i += 1
    self.schema = schema
    self.names = names

1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
  def getCatalogSearchTableIds(self):
    """Return selected tables of catalog which are used in JOIN.
       catalaog is always first
    """
    search_tables = self.sql_search_tables
    if len(search_tables) > 0:
      if search_tables[0] != 'catalog':
        result = ['catalog']
        for t in search_tables:
          if t != 'catalog':
            result.append(t)
        self.sql_search_tables = result
    else:
      self.sql_search_tables = ['catalog']

    return self.sql_search_tables

1071
  security.declarePublic('getCatalogSearchResultKeys')
1072 1073 1074 1075
  def getCatalogSearchResultKeys(self):
    """Return search result keys.
    """
    return self.sql_search_result_keys
1076

1077
  def _getCatalogSchema(self, table=None):
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090
    result_list = []
    try:
      method_name = self.sql_catalog_schema
      method = getattr(self, method_name)
      search_result = method(table=table)
      for c in search_result:
        result_list.append(c.Field)
    except ConflictError:
      raise
    except:
      LOG('SQLCatalog', WARNING, '_getCatalogSchema failed with the method %s' % method_name, error=sys.exc_info())
      pass
    return tuple(result_list)
1091

1092
  @caching_instance_method(id='SQLCatalog.getColumnIds',
1093
    cache_factory='erp5_content_long',
1094
    cache_id_generator=generateCatalogCacheId,
1095
  )
1096
  def _getColumnIds(self):
1097 1098
    keys = set()
    add_key = keys.add
1099
    for table in self.getCatalogSearchTableIds():
1100
      for field in self._getCatalogSchema(table=table):
1101 1102
        add_key(field)
        add_key('%s.%s' % (table, field))  # Is this inconsistent ?
1103 1104
    for related in self.getSQLCatalogRelatedKeyList():
      related_tuple = related.split('|')
1105
      add_key(related_tuple[0].strip())
1106 1107
    for scriptable in self.getSQLCatalogScriptableKeyList():
      scriptable_tuple = scriptable.split('|')
1108 1109
      add_key(scriptable_tuple[0].strip())
    return sorted(keys)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1110

1111 1112 1113 1114 1115 1116 1117
  def getColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
    return self._getColumnIds()[:]

1118
  @transactional_cache_decorator('SQLCatalog.getColumnMap')
1119 1120
  @caching_instance_method(id='SQLCatalog.getColumnMap',
    cache_factory='erp5_content_long',
1121
    cache_id_generator=generateCatalogCacheId,
1122
  )
1123 1124 1125 1126 1127
  def getColumnMap(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
1128 1129 1130 1131 1132 1133
    result = {}
    for table in self.getCatalogSearchTableIds():
      for field in self._getCatalogSchema(table=table):
        result.setdefault(field, []).append(table)
        result.setdefault('%s.%s' % (table, field), []).append(table) # Is this inconsistent ?
    return result
1134

1135 1136
  @transactional_cache_decorator('SQLCatalog.getResultColumnIds')
  @caching_instance_method(id='SQLCatalog.getResultColumnIds',
1137
    cache_factory='erp5_content_long',
1138
    cache_id_generator=generateCatalogCacheId,
1139
  )
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1140 1141 1142 1143 1144
  def getResultColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
1145
    keys = set()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1146
    for table in self.getCatalogSearchTableIds():
1147 1148 1149
      for field in self._getCatalogSchema(table=table):
        keys.add('%s.%s' % (table, field))
    return sorted(keys)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1150

1151 1152 1153
  @transactional_cache_decorator('SQLCatalog.getSortColumnIds')
  @caching_instance_method(id='SQLCatalog.getSortColumnIds',
      cache_factory='erp5_content_long',
1154
      cache_id_generator=generateCatalogCacheId,
1155
  )
1156 1157 1158 1159 1160
  def getSortColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids that can be used for a sort
    """
1161
    keys = set()
1162
    for table in self.getTableIds():
1163 1164 1165
      for field in self._getCatalogSchema(table=table):
        keys.add('%s.%s' % (table, field))
    return sorted(keys)
1166

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1167 1168 1169
  def getTableIds(self):
    """
    Calls the show table method and returns dictionnary of
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1170 1171 1172
    Field Ids
    """
    keys = []
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1173 1174
    method_name = self.sql_catalog_tables
    try:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1175 1176 1177
      method = getattr(self,  method_name)
      search_result = method()
      for c in search_result:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1178
        keys.append(c[0])
1179 1180
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1181 1182
    except:
      pass
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1183 1184
    return keys

1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
  def getUIDBuffer(self, force_new_buffer=False):
    global global_uid_buffer_dict
    klass = self.__class__
    assert klass._reserved_uid_lock.locked()
    assert getattr(self, 'aq_base', None) is not None
    instance_key = self.getPhysicalPath()
    if instance_key not in global_uid_buffer_dict:
      global_uid_buffer_dict[instance_key] = {}
    uid_buffer_dict = global_uid_buffer_dict[instance_key]
    thread_key = get_ident()
    if force_new_buffer or thread_key not in uid_buffer_dict:
      uid_buffer_dict[thread_key] = UidBuffer()
    return uid_buffer_dict[thread_key]
1198

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1199
  # the cataloging API
1200 1201 1202
  def produceUid(self):
    """
      Produces reserved uids in advance
1203
    """
1204 1205 1206 1207
    klass = self.__class__
    assert klass._reserved_uid_lock.locked()
    # This checks if the list of local reserved uids was cleared after clearReserved
    # had been called.
1208 1209 1210 1211
    force_new_buffer = (klass._local_clear_reserved_time != self._last_clear_reserved_time)
    uid_buffer = self.getUIDBuffer(force_new_buffer=force_new_buffer)
    klass._local_clear_reserved_time = self._last_clear_reserved_time
    if len(uid_buffer) == 0:
1212 1213 1214
      id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
      if id_tool is not None:
        if self._max_uid is None:
1215
          self._max_uid = Length(1)
1216
        uid_list = id_tool.generateNewIdList(id_generator='uid', id_group='catalog_uid',
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
                     id_count=UID_BUFFER_SIZE, default=self._max_uid())
        # TODO: if this method is kept and former uid allocation code is
        # discarded, self._max_uid duplicates work done by portal_ids: it
        # already keeps track of the highest allocated number for all id
        # generator groups.
      else:
        method_id = self.sql_catalog_produce_reserved
        method = getattr(self, method_id)
        # Generate an instance id randomly. Note that there is a small possibility that this
        # would conflict with others.
        random_factor_list = [time.time(), os.getpid(), os.times()]
        try:
          random_factor_list.append(os.getloadavg())
        except (OSError, AttributeError): # AttributeError is required under cygwin
          pass
1232
        instance_id = md5(str(random_factor_list)).hexdigest()
1233
        uid_list = [x.uid for x in method(count = UID_BUFFER_SIZE, instance_id = instance_id) if x.uid != 0]
1234
      uid_buffer.extend(uid_list)
1235

1236 1237 1238 1239 1240 1241
  def isIndexable(self):
    """
    This is required to check in many methods that
    the site root and zope root are indexable
    """
    zope_root = self.getZopeRoot()
1242
    site_root = self.getSiteRoot() # XXX-JPS - Why don't we use getPortalObject here ?
1243 1244 1245 1246 1247 1248

    root_indexable = int(getattr(zope_root, 'isIndexable', 1))
    site_indexable = int(getattr(site_root, 'isIndexable', 1))
    if not (root_indexable and site_indexable):
      return False
    return True
Aurel's avatar
Aurel committed
1249

1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
  def getSiteRoot(self):
    """
    Returns the root of the site
    """
    if withCMF:
      site_root = getToolByName(self, 'portal_url').getPortalObject()
    else:
      site_root = self.aq_parent
    return site_root

  def getZopeRoot(self):
    """
    Returns the root of the zope
    """
    if withCMF:
      zope_root = getToolByName(self, 'portal_url').getPortalObject().aq_parent
    else:
      zope_root = self.getPhysicalRoot()
    return zope_root

1270 1271 1272
  def newUid(self):
    """
      This is where uid generation takes place. We should consider a multi-threaded environment
1273 1274
      with multiple ZEO clients on a single ZEO server.

1275
      The main risk is the following:
1276

1277
      - objects a/b/c/d/e/f are created (a is parent of b which is parent of ... of f)
1278

1279
      - one reindexing node N1 starts reindexing f
1280

1281
      - another reindexing node N2 starts reindexing e
1282

1283 1284 1285
      - there is a strong risk that N1 and N2 start reindexing at the same time
        and provide different uid values for a/b/c/d/e

1286
      Similar problems may happen with relations and acquisition of uid values (ex. order_uid)
1287
      with the risk of graph loops
1288
    """
1289
    if not self.isIndexable():
1290 1291
      return None

1292 1293 1294 1295
    klass = self.__class__
    try:
      klass._reserved_uid_lock.acquire()
      self.produceUid()
1296 1297 1298
      uid_buffer = self.getUIDBuffer()
      if len(uid_buffer) > 0:
        uid = uid_buffer.pop()
1299
        if self._max_uid is None:
1300
          self._max_uid = Length(1)
1301 1302
        if uid > self._max_uid():
          self._max_uid.set(uid)
1303
        return long(uid)
1304 1305 1306 1307
      else:
        raise CatalogError("Could not retrieve new uid")
    finally:
      klass._reserved_uid_lock.release()
1308

1309 1310 1311
  def manage_catalogObject(self, REQUEST, RESPONSE, URL1, urls=None):
    """ index Zope object(s) that 'urls' point to """
    if urls:
1312
      if isinstance(urls, str):
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
        urls=(urls,)

      for url in urls:
        obj = self.resolve_path(url)
        if not obj:
          obj = self.resolve_url(url, REQUEST)
        if obj is not None:
          self.aq_parent.catalog_object(obj, url, sql_catalog_id=self.id)

    RESPONSE.redirect(URL1 + '/manage_catalogView?manage_tabs_message=Object%20Cataloged')

  def manage_uncatalogObject(self, REQUEST, RESPONSE, URL1, urls=None):
    """ removes Zope object(s) 'urls' from catalog """

    if urls:
1328
      if isinstance(urls, str):
1329 1330 1331 1332 1333 1334 1335
        urls=(urls,)

      for url in urls:
        self.aq_parent.uncatalog_object(url, sql_catalog_id=self.id)

    RESPONSE.redirect(URL1 + '/manage_catalogView?manage_tabs_message=Object%20Uncataloged')

1336
  def manage_catalogReindex(self, REQUEST, RESPONSE, URL1, urls=None):
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
    """ clear the catalog, then re-index everything """
    elapse = time.time()
    c_elapse = time.clock()

    self.aq_parent.refreshCatalog(clear=1, sql_catalog_id=self.id)

    elapse = time.time() - elapse
    c_elapse = time.clock() - c_elapse

    RESPONSE.redirect(URL1 +
              '/manage_catalogAdvanced?manage_tabs_message=' +
              urllib.quote('Catalog Updated<br>'
                     'Total time: %s<br>'
                     'Total CPU time: %s' % (`elapse`, `c_elapse`)))

1352
  def manage_catalogClear(self, REQUEST=None, RESPONSE=None,
Romain Courteaud's avatar
Romain Courteaud committed
1353
                          URL1=None, sql_catalog_id=None):
1354
    """ clears the whole enchilada """
1355 1356
    self.beforeCatalogClear()

1357 1358 1359
    self.clear()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1360
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1361
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1362 1363

  def manage_catalogClearReserved(self, REQUEST=None, RESPONSE=None, URL1=None):
1364
    """ clears reserved uids """
1365 1366 1367
    self.clearReserved()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1368
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1369
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406

  def manage_catalogFoundItems(self, REQUEST, RESPONSE, URL2, URL1,
                 obj_metatypes=None,
                 obj_ids=None, obj_searchterm=None,
                 obj_expr=None, obj_mtime=None,
                 obj_mspec=None, obj_roles=None,
                 obj_permission=None):
    """ Find object according to search criteria and Catalog them
    """
    elapse = time.time()
    c_elapse = time.clock()

    words = 0
    obj = REQUEST.PARENTS[1]
    path = string.join(obj.getPhysicalPath(), '/')

    results = self.aq_parent.ZopeFindAndApply(obj,
                    obj_metatypes=obj_metatypes,
                    obj_ids=obj_ids,
                    obj_searchterm=obj_searchterm,
                    obj_expr=obj_expr,
                    obj_mtime=obj_mtime,
                    obj_mspec=obj_mspec,
                    obj_permission=obj_permission,
                    obj_roles=obj_roles,
                    search_sub=1,
                    REQUEST=REQUEST,
                    apply_func=self.aq_parent.catalog_object,
                    apply_path=path,
                    sql_catalog_id=self.id)

    elapse = time.time() - elapse
    c_elapse = time.clock() - c_elapse

    RESPONSE.redirect(URL1 + '/manage_catalogView?manage_tabs_message=' +
              urllib.quote('Catalog Updated<br>Total time: %s<br>Total CPU time: %s' % (`elapse`, `c_elapse`)))

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1407
  def catalogObject(self, object, path, is_object_moved=0):
1408 1409
    """Add an object to the Catalog by calling all SQL methods and
    providing needed arguments.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1410

1411 1412
    'object' is the object to be catalogged."""
    self._catalogObjectList([object])
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1413

1414
  def catalogObjectList(self, object_list, method_id_list=None,
1415 1416 1417
                        disable_cache=0, check_uid=1, idxs=None):
    """Add objects to the Catalog by calling all SQL methods and
    providing needed arguments.
1418

1419 1420
      method_id_list : specify which methods should be used. If not
                       set, it will take the default value of portal_catalog.
1421 1422

      disable_cache : do not use cache, so values will be computed each time,
1423 1424 1425 1426 1427 1428
                      only useful in some particular cases, most of the time
                      you don't need to use it.

    Each element of 'object_list' is an object to be catalogged.

    'uid' is the unique Catalog identifier for this object.
1429

1430 1431 1432 1433
    Note that this method calls _catalogObjectList with fragments of
    the object list, because calling _catalogObjectList with too many
    objects at a time bloats the process's memory consumption, due to
    caching."""
1434 1435
    for i in xrange(0, len(object_list), OBJECT_LIST_SIZE):
      self._catalogObjectList(object_list[i:i + OBJECT_LIST_SIZE],
1436 1437 1438 1439
                              method_id_list=method_id_list,
                              disable_cache=disable_cache,
                              check_uid=check_uid,
                              idxs=idxs)
1440 1441

  def _catalogObjectList(self, object_list, method_id_list=None,
1442 1443 1444 1445 1446
                         disable_cache=0, check_uid=1, idxs=None):
    """This is the real method to catalog objects.

    XXX: For now newUid is used to allocated UIDs. Is this good?
    Is it better to INSERT then SELECT?"""
1447
    LOG('SQLCatalog', TRACE, 'catalogging %d objects' % len(object_list))
1448
    #LOG('catalogObjectList', 0, 'called with %r' % (object_list,))
1449

1450
    if idxs not in (None, []):
1451
      LOG('ZSLQCatalog.SQLCatalog:catalogObjectList', WARNING,
1452
          'idxs is ignored in this function and is only provided to be compatible with CMFCatalogAware.reindexObject.')
1453

1454
    if not self.isIndexable():
1455
      return
1456

1457 1458
    # Reminder about optimization: It might be possible to issue just one
    # query to get enought results to check uid & path consistency.
1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476
    path_uid_dict = {}
    uid_path_dict = {}

    if check_uid:
      path_list = []
      path_list_append = path_list.append
      uid_list = []
      uid_list_append = uid_list.append
      for object in object_list:
        path = object.getPath()
        if path is not None:
          path_list_append(path)
        uid = object.uid
        if uid is not None:
          uid_list_append(uid)
      path_uid_dict = self.getUidDictForPathList(path_list=path_list)
      uid_path_dict = self.getPathDictForUidList(uid_list=uid_list)

1477 1478 1479 1480 1481
    # This dict will store uids and objects which are verified below.
    # The purpose is to prevent multiple objects from having the same
    # uid inside object_list.
    assigned_uid_dict = {}

1482
    for object in object_list:
1483 1484 1485 1486
      uid = getattr(aq_base(object), 'uid', None)
      # Several Tool objects have uid=0 (not 0L) from the beginning, but
      # we need an unique uid for each object.
      if uid is None or isinstance(uid, int) and uid == 0:
1487
        try:
1488
          object.uid = self.newUid()
1489 1490
        except ConflictError:
          raise
1491
        except:
1492
          raise RuntimeError, 'could not set missing uid for %r' % (object,)
1493
      elif check_uid:
1494
        if uid in assigned_uid_dict:
1495 1496 1497 1498 1499 1500 1501
            error_message = 'uid of %r is %r and ' \
                  'is already assigned to %s in catalog !!! This can be fatal.' % \
                  (object, uid, assigned_uid_dict[uid])
            if not self.sql_catalog_raise_error_on_uid_check:
                LOG("SQLCatalog.catalogObjectList", ERROR, error_message)
            else:
                raise ValueError(error_message)
1502

1503
        path = object.getPath()
1504
        index = path_uid_dict.get(path)
1505 1506 1507
        if index is not None:
          if index < 0:
            raise CatalogError, 'A negative uid %d is used for %s. Your catalog is broken. Recreate your catalog.' % (index, path)
1508 1509
          if uid != index or isinstance(uid, int):
            # We want to make sure that uid becomes long if it is an int
1510 1511 1512 1513 1514 1515
            error_message = 'uid of %r changed from %r (property) to %r '\
	                    '(catalog, by path) !!! This can be fatal' % (object, uid, index)
            if not self.sql_catalog_raise_error_on_uid_check:
              LOG("SQLCatalog.catalogObjectList", ERROR, error_message)
            else:
              raise ValueError(error_message)
1516 1517 1518 1519
        else:
          # Make sure no duplicates - ie. if an object with different path has same uid, we need a new uid
          # This can be very dangerous with relations stored in a category table (CMFCategory)
          # This is why we recommend completely reindexing subobjects after any change of id
1520
          if uid in uid_path_dict:
1521 1522 1523
            catalog_path = uid_path_dict.get(uid)
          else:
            catalog_path = self.getPathForUid(uid)
1524 1525 1526
          #LOG('catalogObject', 0, 'uid = %r, catalog_path = %r' % (uid, catalog_path))
          if catalog_path == "reserved":
            # Reserved line in catalog table
1527
            lock = self.__class__._reserved_uid_lock
1528
            try:
1529
              lock.acquire()
1530 1531
              uid_buffer = self.getUIDBuffer()
              if uid_buffer is not None:
1532 1533 1534 1535 1536 1537 1538 1539
                # This is the case where:
                #   1. The object got an uid.
                #   2. The catalog was cleared.
                #   3. The catalog produced the same reserved uid.
                #   4. The object was reindexed.
                # In this case, the uid is not reserved any longer, but
                # SQLCatalog believes that it is still reserved. So it is
                # necessary to remove the uid from the list explicitly.
1540
                try:
1541
                  uid_buffer.remove(uid)
1542 1543
                except ValueError:
                  pass
1544
            finally:
1545
              lock.release()
1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562
          elif catalog_path == 'deleted':
            # Two possible cases:
            # - Reindexed object's path changed (ie, it or at least one of its
            #   parents was renamed) but unindexObject was not called yet.
            #   Reindexing is harmelss: unindexObject and then an
            #   immediateReindexObject will be called.
            # - Reindexed object was deleted by a concurrent transaction, which
            #   committed after we got our ZODB snapshot of this object.
            #   Reindexing is harmless: unindexObject will be called, and
            #   cannot be executed in parallel thanks to activity's
            #   serialisation_tag (so we cannot end up with a fantom object in
            #   catalog).
            # So we index object.
            # We could also not index it to save the time needed to index, but
            # this would slow down all regular case to slightly improve an
            # exceptional case.
            pass
1563 1564
          elif catalog_path is not None:
            # An uid conflict happened... Why?
1565
            # can be due to path length
1566
            if len(path) > MAX_PATH_LEN:
1567
              LOG('SQLCatalog', ERROR, 'path of object %r is too long for catalog. You should use a shorter path.' %(object,))
1568

1569
            LOG('SQLCatalog', ERROR,
1570 1571
                'uid of %r changed from %r to %r as old one is assigned'
                ' to %s in catalog !!! This can be fatal.' % (
1572 1573 1574 1575 1576 1577 1578 1579 1580
                object, uid, object.uid, catalog_path))

            error_message = 'uid of %r is %r and ' \
                            'is already assigned to %s in catalog !!! This can be fatal.' \
                            % (object, uid, catalog_path)
            if not self.sql_catalog_raise_error_on_uid_check:
                LOG('SQLCatalog', ERROR, error_message)
            else:
                raise ValueError(error_message)
1581 1582 1583
            uid = object.uid

        assigned_uid_dict[uid] = object
1584

1585 1586
    if method_id_list is None:
      method_id_list = self.sql_catalog_object_list
1587
    econtext = getEngine().getContext()
1588 1589 1590 1591
    if disable_cache:
      argument_cache = DummyDict()
    else:
      argument_cache = {}
1592

1593 1594
    with (noReadOnlyTransactionCache if disable_cache else
          readOnlyTransactionCache)():
1595
      filter_dict = self.filter_dict
1596
      catalogged_object_list_cache = {}
1597
      for method_name in method_id_list:
1598 1599 1600 1601 1602
        # We will check if there is an filter on this
        # method, if so we may not call this zsqlMethod
        # for this object
        expression = None
        try:
1603
          filter = filter_dict[method_name]
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
          if filter['filtered']:
            if filter.get('type'):
              expression = Expression('python: context.getPortalType() in '
                                      + repr(tuple(filter['type'])))
              LOG('SQLCatalog', WARNING,
                  "Convert deprecated type filter for %r into %r expression"
                  % (method_name, expression.text))
              filter['type'] = ()
              filter['expression'] = expression.text
              filter['expression_instance'] = expression
            else:
              expression = filter['expression_instance']
        except KeyError:
          pass
        if expression is None:
          catalogged_object_list = object_list
        else:
          text = expression.text
          catalogged_object_list = catalogged_object_list_cache.get(text)
          if catalogged_object_list is None:
            catalogged_object_list_cache[text] = catalogged_object_list = []
            append = catalogged_object_list.append
            old_context = new_context_search(text) is None
            if old_context:
              warnings.warn("Filter expression for %r (%r): using variables"
                            " other than 'context' is deprecated and slower."
                            % (method_name, text), DeprecationWarning)
            expression_cache_key_list = filter.get('expression_cache_key', ())
            expression_result_cache = {}
            for object in object_list:
1634
              if expression_cache_key_list:
1635 1636 1637 1638
                # Expressions are slow to evaluate because they are executed
                # in restricted environment. So we try to save results of
                # expressions by portal_type or any other key.
                # This cache is built each time we reindex
1639 1640
                # objects but we could also use over multiple transactions
                # if this can improve performance significantly
1641 1642 1643
                # ZZZ - we could find a way to compute this once only
                cache_key = tuple(object.getProperty(key) for key
                                  in expression_cache_key_list)
1644
                try:
1645 1646 1647
                  if expression_result_cache[cache_key]:
                    append(object)
                  continue
1648
                except KeyError:
1649 1650 1651
                  pass
              if old_context:
                result = expression(self.getExpressionContext(object))
1652
              else:
1653
                econtext.setLocal('context', object)
1654
                result = expression(econtext)
1655
              if expression_cache_key_list:
1656
                expression_result_cache[cache_key] = result
1657 1658
              if result:
                append(object)
1659

1660
        if not catalogged_object_list:
1661
          continue
1662

1663 1664
        #LOG('catalogObjectList', 0, 'method_name = %s' % (method_name,))
        method = getattr(self, method_name)
1665
        if method.meta_type in ("Z SQL Method", "LDIF Method"):
1666
          # Build the dictionnary of values
1667
          arguments = method.arguments_src.split()
1668 1669 1670
        elif method.meta_type == "Script (Python)":
          arguments = \
            method.func_code.co_varnames[:method.func_code.co_argcount]
1671 1672
        else:
          arguments = []
1673 1674 1675
        kw = {x: LazyIndexationParameterList(catalogged_object_list,
                                             x, argument_cache)
          for x in arguments}
1676

1677 1678 1679
        # Alter/Create row
        try:
          #start_time = DateTime()
1680
          #LOG('catalogObjectList', DEBUG, 'kw = %r, method_name = %r' % (kw, method_name))
1681 1682 1683 1684 1685 1686 1687
          method(**kw)
          #end_time = DateTime()
          #if method_name not in profile_dict:
          #  profile_dict[method_name] = end_time.timeTime() - start_time.timeTime()
          #else:
          #  profile_dict[method_name] += end_time.timeTime() - start_time.timeTime()
          #LOG('catalogObjectList', 0, '%s: %f seconds' % (method_name, profile_dict[method_name]))
1688

1689 1690 1691
        except ConflictError:
          raise
        except:
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1692
          LOG('SQLCatalog', WARNING, 'could not catalog objects %s with method %s' % (object_list, method_name),
1693 1694
              error=sys.exc_info())
          raise
1695

1696 1697
  if psyco is not None:
    psyco.bind(_catalogObjectList)
1698

1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710
  def beforeUncatalogObject(self, path=None,uid=None):
    """
    Set the path as deleted
    """
    if not self.isIndexable():
      return None

    if uid is None and path is not None:
      uid = self.getUidForPath(path)
    method_name = self.sql_catalog_delete_uid
    if uid is None:
      return None
1711 1712 1713 1714
    if method_name in (None,''):
      # This should exist only if the site is not up to date.
      LOG('ZSQLCatalog.beforeUncatalogObject',0,'The sql_catalog_delete_uid'\
                                                + ' method is not defined')
Sebastien Robin's avatar
Sebastien Robin committed
1715
      return self.uncatalogObject(path=path,uid=uid)
1716 1717 1718
    method = getattr(self, method_name)
    method(uid = uid)

1719
  def uncatalogObject(self, path=None, uid=None):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
    """
    Uncatalog and object from the Catalog.

    Note, the uid must be the same as when the object was
    catalogued, otherwise it will not get removed from the catalog

    This method should not raise an exception if the uid cannot
    be found in the catalog.

    XXX Add filter of methods

    """
1732
    if not self.isIndexable():
1733 1734
      return None

1735 1736
    if uid is None and path is not None:
      uid = self.getUidForPath(path)
1737 1738
    methods = self.sql_uncatalog_object
    if uid is None:
1739
      return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1740
    for method_name in methods:
1741 1742
      # Do not put try/except here, it is required to raise error
      # if uncatalog does not work.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1743
      method = getattr(self, method_name)
1744
      method(uid = uid)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1745

1746 1747 1748 1749
  def catalogTranslationList(self, object_list):
    """Catalog translations.
    """
    method_name = self.sql_catalog_translation_list
1750 1751
    return self.catalogObjectList(object_list, method_id_list = (method_name,),
                                  check_uid=0)
1752

1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763
  def deleteTranslationList(self):
    """Delete translations.
    """
    method_name = self.sql_delete_translation_list
    method = getattr(self, method_name)
    try:
      method()
    except ConflictError:
      raise
    except:
      LOG('SQLCatalog', WARNING, 'could not delete translations', error=sys.exc_info())
1764

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1765 1766 1767
  def uniqueValuesFor(self, name):
    """ return unique values for FieldIndex name """
    method = getattr(self, self.sql_unique_values)
1768
    return method(column=name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1769 1770 1771 1772 1773 1774 1775 1776

  def getPaths(self):
    """ Returns all object paths stored inside catalog """
    method = getattr(self, self.sql_catalog_paths)
    return method()

  def getUidForPath(self, path):
    """ Looks up into catalog table to convert path into uid """
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
    #try:
    if path is None:
      return None
    # Get the appropriate SQL Method
    method = getattr(self, self.sql_getitem_by_path)
    search_result = method(path = path, uid_only=1)
    # If not empty, return first record
    if len(search_result) > 0:
      return long(search_result[0].uid)
    else:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1787 1788
      return None

1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827
  def getUidDictForPathList(self, path_list):
    """ Looks up into catalog table to convert path into uid """
    # Get the appropriate SQL Method
    method = getattr(self, self.sql_getitem_by_path)
    path_uid_dict = {}
    try:
      search_result = method(path_list = path_list)
      # If not empty, return first record
      for result in search_result:
        path_uid_dict[result.path] = result.uid
    except ValueError, message:
      # This code is only there for backward compatibility
      # XXX this must be removed one day
      # This means we have the previous zsql method
      # and we must call the method for every path
      for path in path_list:
        search_result = method(path = path)
        if len(search_result) > 0:
          path_uid_dict[path] = search_result[0].uid
    return path_uid_dict

  def getPathDictForUidList(self, uid_list):
    """ Looks up into catalog table to convert uid into path """
    # Get the appropriate SQL Method
    method = getattr(self, self.sql_getitem_by_uid)
    uid_path_dict = {}
    try:
      search_result = method(uid_list = uid_list)
      # If not empty, return first record
      for result in search_result:
        uid_path_dict[result.uid] = result.path
    except ValueError, message:
      # This code is only there for backward compatibility
      # XXX this must be removed one day
      # This means we have the previous zsql method
      # and we must call the method for every path
      for uid in uid_list:
        search_result = method(uid = uid)
        if len(search_result) > 0:
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1828
          uid_path_dict[uid] = search_result[0].path
1829 1830
    return uid_path_dict

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1831 1832 1833 1834 1835 1836 1837 1838 1839
  def hasPath(self, path):
    """ Checks if path is catalogued """
    return self.getUidForPath(path) is not None

  def getPathForUid(self, uid):
    """ Looks up into catalog table to convert uid into path """
    try:
      if uid is None:
        return None
1840 1841 1842 1843
      try:
        int(uid)
      except ValueError:
        return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1844 1845 1846 1847 1848 1849 1850 1851
      # Get the appropriate SQL Method
      method = getattr(self, self.sql_getitem_by_uid)
      search_result = method(uid = uid)
      # If not empty return first record
      if len(search_result) > 0:
        return search_result[0].path
      else:
        return None
1852 1853
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1854 1855 1856
    except:
      # This is a real LOG message
      # which is required in order to be able to import .zexp files
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1857
      LOG('SQLCatalog', WARNING, "could not find path from uid %s" % (uid,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
      return None

  def getMetadataForUid(self, uid):
    """ Accesses a single record for a given uid """
    if uid is None:
      return None
    # Get the appropriate SQL Method
    method = getattr(self, self.sql_getitem_by_uid)
    brain = method(uid = uid)[0]
    result = {}
    for k in brain.__record_schema__.keys():
      result[k] = getattr(brain,k)
    return result

  def getIndexDataForUid(self, uid):
    """ Accesses a single record for a given uid """
    return self.getMetadataForUid(uid)

  def getMetadataForPath(self, path):
    """ Accesses a single record for a given path """
    try:
      # Get the appropriate SQL Method
      method = getattr(self, self.sql_getitem_by_path)
      brain = method(path = path)[0]
      result = {}
      for k in brain.__record_schema__.keys():
        result[k] = getattr(brain,k)
      return result
1886 1887
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1888 1889 1890
    except:
      # This is a real LOG message
      # which is required in order to be able to import .zexp files
1891 1892
      LOG('SQLCatalog', WARNING,
          "could not find metadata from path %s" % (path,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1893 1894 1895 1896 1897 1898
      return None

  def getIndexDataForPath(self, path):
    """ Accesses a single record for a given path """
    return self.getMetadataForPath(path)

1899 1900
  def getCatalogMethodIds(self,
      valid_method_meta_type_list=valid_method_meta_type_list):
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911
    """Find Z SQL methods in the current folder and above
    This function return a list of ids.
    """
    ids={}
    have_id=ids.has_key

    while self is not None:
      if hasattr(self, 'objectValues'):
        for o in self.objectValues(valid_method_meta_type_list):
          if hasattr(o,'id'):
            id=o.id
1912 1913
            if not isinstance(id, str):
              id=id()
1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924
            if not have_id(id):
              if hasattr(o,'title_and_id'): o=o.title_and_id()
              else: o=id
              ids[id]=id
      if hasattr(self, 'aq_parent'): self=self.aq_parent
      else: self=None

    ids=map(lambda item: (item[1], item[0]), ids.items())
    ids.sort()
    return ids

1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936
  def getPythonMethodIds(self):
    """
      Returns a list of all python scripts available in
      current sql catalog.
    """
    return self.getCatalogMethodIds(valid_method_meta_type_list=('Script (Python)', ))

  @transactional_cache_decorator('SQLCatalog._getSQLCatalogRelatedKeyList')
  def _getSQLCatalogRelatedKeySet(self):
    column_map = self.getColumnMap()
    column_set = set(column_map)
    for related_key in self.sql_catalog_related_keys:
1937 1938
      split_entire_definition = related_key.split('|')
      if len(split_entire_definition) != 2:
1939
        LOG('SQLCatalog', WARNING, 'Malformed related key definition: %r. Ignored.' % (related_key, ))
1940 1941
        continue
      related_key_id = split_entire_definition[0].strip()
1942 1943 1944
      if related_key_id in column_set and \
         related_key_id not in related_key_warned_column_set:
        related_key_warned_column_set.add(related_key_id)
1945
        if related_key_id in column_map:
1946
          LOG('SQLCatalog', WARNING, 'Related key %r has the same name as an existing column on tables %r' % (related_key_id, column_map[related_key_id]))
1947
        else:
1948
          LOG('SQLCatalog', WARNING, 'Related key %r is declared more than once.' % (related_key_id, ))
1949 1950 1951
      column_set.add(related_key_id)
    return column_set

1952
  def getSQLCatalogRelatedKeyList(self, key_list=None):
1953 1954
    """
    Return the list of related keys.
1955
    This method can be overidden in order to implement
1956 1957
    dynamic generation of some related keys.
    """
1958 1959
    if key_list is None:
      key_list = []
1960
    column_map = self._getSQLCatalogRelatedKeySet()
1961 1962 1963 1964
    return self.getDynamicRelatedKeyList(
      [k for k in key_list if k not in column_map],
      sql_catalog_id=self.id,
    ) + list(self.sql_catalog_related_keys)
1965

1966 1967 1968
  # Compatibililty SQL Sql
  getSqlCatalogRelatedKeyList = getSQLCatalogRelatedKeyList

1969 1970 1971 1972 1973
  def getSQLCatalogScriptableKeyList(self):
    """
    Return the list of scriptable keys.
    """
    return self.sql_catalog_scriptable_keys
1974

1975
  @caching_instance_method(id='SQLCatalog.getTableIndex',
1976
    cache_factory='erp5_content_long',
1977
    cache_id_generator=generateCatalogCacheId,
1978
  )
1979
  def _getTableIndex(self, table):
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
    table_index = {}
    method = getattr(self, self.sql_catalog_index, '')
    if method in ('', None):
      return {}
    index = list(method(table=table))
    for line in index:
      if table_index.has_key(line.KEY_NAME):
        table_index[line.KEY_NAME].append(line.COLUMN_NAME)
      else:
        table_index[line.KEY_NAME] = [line.COLUMN_NAME,]
1990 1991 1992 1993 1994 1995 1996
    return table_index

  def getTableIndex(self, table):
    """
    Return a map between index and column for a given table
    """
    return self._getTableIndex(table).copy()
1997

1998 1999 2000 2001
  def isValidColumn(self, column_id):
    """
      Tells wether given name is or not an existing column.

2002 2003
      Warning: This includes "virtual" columns, such as related keys and
      scriptable keys.
2004
    """
2005
    result = self.getScriptableKeyScript(column_id) is not None
2006
    if not result:
2007 2008 2009
      result = column_id in self.getColumnMap()
      if not result:
        result = self.getRelatedKeyDefinition(column_id) is not None
2010 2011
    return result

2012 2013 2014 2015 2016
  def getRelatedKeyDefinition(self, key):
    """
      Returns the definition of given related key name if found, None
      otherwise.
    """
2017 2018
    related_key_definition_cache = getTransactionalVariable().setdefault(
      'SQLCatalog.getRelatedKeyDefinition', {})
2019 2020 2021 2022
    try:
      result = related_key_definition_cache[key]
    except KeyError:
      for entire_definition in self.getSQLCatalogRelatedKeyList([key]):
2023 2024
        split_entire_definition = entire_definition.split('|')
        if len(split_entire_definition) != 2:
2025
          LOG('SQLCatalog', WARNING, 'Malformed related key definition: %r. Ignored.' % (entire_definition, ))
2026
          continue
2027 2028
        if split_entire_definition[0].strip() == key:
          result = split_entire_definition[1].strip()
2029
          break
2030 2031 2032
      else:
        result = None
      related_key_definition_cache[key] = result
2033
    return result
2034

2035 2036 2037 2038 2039 2040
  @transactional_cache_decorator('SQLCatalog._getgetScriptableKeyDict')
  def _getgetScriptableKeyDict(self):
    result = {}
    for scriptable_key_definition in self.sql_catalog_scriptable_keys:
      split_scriptable_key_definition = scriptable_key_definition.split('|')
      if len(split_scriptable_key_definition) != 2:
2041
        LOG('SQLCatalog', WARNING, 'Malformed scriptable key definition: %r. Ignored.' % (scriptable_key_definition, ))
2042 2043 2044 2045
        continue
      key, script_id = [x.strip() for x in split_scriptable_key_definition]
      script = getattr(self, script_id, None)
      if script is None:
2046
        LOG('SQLCatalog', WARNING, 'Scriptable key %r script %r is missing.' \
2047 2048 2049 2050 2051 2052 2053 2054
            ' Skipped.' % (key, script_id))
      else:
        result[key] = script
    return result

  def getScriptableKeyScript(self, key):
    return self._getgetScriptableKeyDict().get(key)

2055 2056 2057 2058 2059 2060 2061 2062 2063 2064
  def getColumnSearchKey(self, key, search_key_name=None):
    """
      Return a SearchKey instance for given key, using search_key_name
      as a SearchKey name if given, otherwise guessing from catalog
      configuration. If there is no search_key_name given and no
      SearchKey can be found, return None.

      Also return a related key definition string with following rules:
       - If returned SearchKey is a RelatedKey, value is its definition
       - Otherwise, value is None
2065 2066 2067

      If both a related key and a real column are found, the related key
      is used.
2068
    """
2069 2070 2071 2072 2073
    # Is key a related key or a "real" column ?
    related_key_definition = self.getRelatedKeyDefinition(key)
    if related_key_definition is None:
      if key in self.getColumnMap():
        search_key = self.getSearchKey(key, search_key_name)
2074
      else:
2075 2076 2077
        search_key = None
    else:
      search_key = self.getSearchKey(key, 'RelatedKey')
2078 2079
    return search_key, related_key_definition

2080 2081 2082
  def hasColumn(self, column):
    return self.getColumnSearchKey(column)[0] is not None

2083
  def getColumnDefaultSearchKey(self, key, search_key_name=None):
2084 2085 2086 2087
    """
      Return a SearchKey instance which would ultimately receive the value
      associated with given key.
    """
2088 2089
    search_key, related_key_definition = self.getColumnSearchKey(key,
      search_key_name=search_key_name)
2090 2091 2092 2093
    if search_key is None:
      result = None
    else:
      if related_key_definition is not None:
2094 2095
        search_key = search_key.getSearchKey(sql_catalog=self,
          related_key_definition=related_key_definition)
2096 2097 2098 2099 2100 2101 2102
    return search_key

  def buildSingleQuery(self, key, value, search_key_name=None, logical_operator=None, comparison_operator=None):
    """
      From key and value, determine the SearchKey to use and generate a Query
      from it.
    """
2103 2104 2105 2106 2107
    script = self.getScriptableKeyScript(key)
    if script is None:
      search_key, related_key_definition = self.getColumnSearchKey(key, search_key_name)
      if search_key is None:
        result = None
Ivan Tyagov's avatar
Ivan Tyagov committed
2108
      else:
2109
        if related_key_definition is None:
2110
          build_key = search_key
2111
        else:
2112 2113 2114
          build_key = search_key.getSearchKey(sql_catalog=self,
            related_key_definition=related_key_definition,
            search_key_name=search_key_name)
2115 2116 2117
        result = build_key.buildQuery(value, logical_operator=logical_operator,
                                      comparison_operator=comparison_operator)
        if related_key_definition is not None:
2118 2119 2120
          result = search_key.buildQuery(sql_catalog=self,
            related_key_definition=related_key_definition,
            search_value=result)
2121 2122
    else:
      result = script(value)
2123 2124
    return result

2125
  def _buildQueryFromAbstractSyntaxTreeNode(self, node, search_key, wrap):
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135
    """
    node
      Abstract syntax tree node (see SearchText/AdvancedSearchTextParser.py,
      classes inheriting from Node).
    search_key
      Search key to generate queries from values found during syntax tree walk.
    wrap
      Callback transforming a value just before it is passed to
      search_key.buildQuery .
    """
2136 2137 2138 2139
    if search_key.dequoteParsedText():
      _dequote = dequote
    else:
      _dequote = lambda x: x
2140
    if node.isLeaf():
2141
      result = search_key.buildQuery(wrap(_dequote(node.getValue())),
2142
        comparison_operator=node.getComparisonOperator())
2143 2144 2145 2146 2147 2148 2149 2150
    elif node.isColumn():
      result = self.buildQueryFromAbstractSyntaxTreeNode(node.getSubNode(), node.getColumnName())
    else:
      query_list = []
      value_dict = {}
      append = query_list.append
      for subnode in node.getNodeList():
        if subnode.isLeaf():
2151
          value_dict.setdefault(subnode.getComparisonOperator(),
2152
            []).append(wrap(_dequote(subnode.getValue())))
2153
        else:
2154
          subquery = self._buildQueryFromAbstractSyntaxTreeNode(subnode, search_key, wrap)
2155 2156
          if subquery is not None:
            append(subquery)
2157
      logical_operator = node.getLogicalOperator()
2158 2159 2160 2161
      if logical_operator == 'not':
        query_logical_operator = None
      else:
        query_logical_operator = logical_operator
2162
      for comparison_operator, value_list in value_dict.iteritems():
2163
        append(search_key.buildQuery(value_list, comparison_operator=comparison_operator, logical_operator=query_logical_operator))
2164 2165
      if logical_operator == 'not' or len(query_list) > 1:
        result = ComplexQuery(query_list, logical_operator=logical_operator)
2166 2167
      elif len(query_list) == 1:
        result = query_list[0]
2168
      else:
2169 2170 2171
        result = None
    return result

2172
  def buildQueryFromAbstractSyntaxTreeNode(self, node, key, wrap=lambda x: x):
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182
    """
      Build a query from given Abstract Syntax Tree (AST) node by recursing in
      its childs.
      This method calls itself recursively when walking the tree.

      node
        AST node being treated.
      key
        Default column (used when there is no explicit column in an AST leaf).

2183
      Expected node API is described in interfaces/abstract_syntax_node.py .
2184
    """
2185 2186 2187 2188 2189 2190
    script = self.getScriptableKeyScript(key)
    if script is None:
      search_key, related_key_definition = self.getColumnSearchKey(key)
    else:
      search_key = SearchKeyWrapperForScriptableKey(key, script)
      related_key_definition = None
2191 2192
    if search_key is None:
      # Unknown, skip loudly
2193
      LOG('SQLCatalog', WARNING, 'Unknown column %r, skipped.' % (key, ))
2194
      result = None
2195 2196 2197 2198
    else:
      if related_key_definition is None:
        build_key = search_key
      else:
2199 2200
        build_key = search_key.getSearchKey(sql_catalog=self,
          related_key_definition=related_key_definition)
2201 2202
      result = self._buildQueryFromAbstractSyntaxTreeNode(node, build_key,
        wrap)
2203
      if related_key_definition is not None:
2204 2205 2206
        result = search_key.buildQuery(sql_catalog=self,
          related_key_definition=related_key_definition,
          search_value=result)
2207 2208
    return result

2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221
  def _parseSearchText(self, search_key, search_text, is_valid=None):
    if is_valid is None:
      is_valid = self.isValidColumn
    return search_key.parseSearchText(search_text, is_valid)

  def parseSearchText(self, search_text, column=None, search_key=None,
                      is_valid=None):
    if column is None and search_key is None:
      raise ValueError, 'One of column and search_key must be different '\
                        'from None'
    return self._parseSearchText(self.getSearchKey(
      column, search_key=search_key), search_text, is_valid=is_valid)

2222 2223 2224 2225 2226 2227 2228 2229 2230 2231
  def buildQuery(self, kw, ignore_empty_string=True, operator='and'):
    query_list = []
    append = query_list.append
    # unknown_column_dict: contains all (key, value) pairs which could not be
    # changed into queries. This is here for backward compatibility, because
    # scripts can invoke this method and expect extra parameters (such as
    # from_expression) to be handled. As they are normaly handled at
    # buildSQLQuery level, we must store them into final ComplexQuery, which
    # will handle them.
    unknown_column_dict = {}
2232 2233 2234
    # empty_value_dict: contains all keys whose value causes them to be
    # discarded.
    empty_value_dict = {}
2235 2236 2237 2238 2239 2240 2241
    for key, value in kw.iteritems():
      result = None
      if isinstance(value, dict_type_list):
        # Cast dict-ish types into plain dicts.
        value = dict(value)
      if ignore_empty_string and (
          value == ''
2242
          or (isinstance(value, list_type_list) and not value)
2243 2244 2245
          or (isinstance(value, dict) and (
            'query' not in value
            or value['query'] == ''
2246 2247
            or (isinstance(value['query'], list_type_list)
              and not value['query'])))):
2248
        # We have an empty value, do not create a query from it
2249
        empty_value_dict[key] = value
2250
      else:
2251
        script = self.getScriptableKeyScript(key)
2252
        if isinstance(value, dict):
2253 2254 2255 2256 2257 2258
          # Dictionnary: might contain the search key to use.
          search_key_name = value.get('key')
          # Backward compatibility: former "Keyword" key is now named
          # "KeywordKey".
          if search_key_name == 'Keyword':
            search_key_name = value['key'] = 'KeywordKey'
2259 2260 2261 2262
          # Backward compatibility: former "ExactMatch" is now only available
          # as "RawKey"
          elif search_key_name == 'ExactMatch':
            search_key_name = value['key'] = 'RawKey'
2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
        if isinstance(value, _Query):
          # Query instance: use as such, ignore key.
          result = value
        elif script is not None:
          result = script(value)
        elif isinstance(value, (basestring, dict)):
          # String: parse using key's default search key.
          raw_value = value
          if isinstance(value, dict):
            # De-wrap value for parsing, and re-wrap when building queries.
            def wrap(x):
              result = raw_value.copy()
              result['query'] = x
              return result
            value = value['query']
          else:
            wrap = lambda x: x
            search_key_name = None
          search_key = self.getColumnDefaultSearchKey(key,
            search_key_name=search_key_name)
          if search_key is not None:
            if isinstance(value, basestring):
              abstract_syntax_tree = self._parseSearchText(search_key, value)
            else:
              abstract_syntax_tree = None
            if abstract_syntax_tree is None:
              # Parsing failed, create a query from the bare string.
              result = self.buildSingleQuery(key, raw_value, search_key_name)
            else:
              result = self.buildQueryFromAbstractSyntaxTreeNode(
                abstract_syntax_tree, key, wrap)
2294 2295 2296 2297 2298 2299 2300 2301
        else:
          # Any other type, just create a query. (can be a DateTime, ...)
          result = self.buildSingleQuery(key, value)
        if result is None:
          # No query could be created, emit a log, add to unknown column dict.
          unknown_column_dict[key] = value
        else:
          append(result)
2302 2303
    if len(empty_value_dict):
      LOG('SQLCatalog', WARNING, 'Discarding columns with empty values: %r' % (empty_value_dict, ))
2304
    if len(unknown_column_dict):
2305
      LOG('SQLCatalog', WARNING, 'Unknown columns %r, skipped.' % (unknown_column_dict.keys(), ))
2306 2307
    return ComplexQuery(query_list, logical_operator=operator,
        unknown_column_dict=unknown_column_dict)
2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326

  def buildOrderByList(self, sort_on=None, sort_order=None, order_by_expression=None):
    """
      Internal method. Should not be used by code outside buildSQLQuery.

      It is in a separate method because this code is here to keep backward
      compatibility with an ambiguous API, and as such is ugly. So it's better
      to conceal it to its own method.

      It does not preserve backward compatibility for:
        'sort-on' parameter
        'sort-on' property
        'sort-order' parameter
        'sort-order' property
    """
    order_by_list = []
    append = order_by_list.append
    if sort_on is not None:
      if order_by_expression is not None:
2327
        LOG('SQLCatalog', WARNING, 'order_by_expression (%r) and sort_on (%r) were given. Ignoring order_by_expression.' % (order_by_expression, sort_on))
2328 2329 2330 2331 2332
      if not isinstance(sort_on, (tuple, list)):
        sort_on = [[sort_on]]
      for item in sort_on:
        if isinstance(item, (tuple, list)):
          item = list(item)
2333
        else:
2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351
          item = [item]
        if sort_order is not None and len(item) == 1:
          item.append(sort_order)
        if len(item) > 1:
          if item[1] in ('descending', 'reverse', 'DESC'):
            item[1] = 'DESC'
          else:
            item[1] = 'ASC'
          if len(item) > 2:
            if item[2] == 'int':
              item[2] = 'SIGNED'
        append(item)
    elif order_by_expression is not None:
      if not isinstance(order_by_expression, basestring):
        raise TypeError, 'order_by_expression must be a basestring instance. Got %r.' % (order_by_expression, )
      order_by_list = [[x.strip()] for x in order_by_expression.split(',')]
    return order_by_list

2352 2353
  def buildEntireQuery(self, kw, query_table='catalog', ignore_empty_string=1,
                       limit=None, extra_column_list=()):
2354
    group_by_list = kw.pop('group_by_list', kw.pop('group_by', kw.pop('group_by_expression', ())))
2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370
    if isinstance(group_by_list, basestring):
      group_by_list = [x.strip() for x in group_by_list.split(',')]
    select_dict = kw.pop('select_dict', kw.pop('select_list', kw.pop('select_expression', None)))
    if isinstance(select_dict, basestring):
      if len(select_dict):
        real_select_dict = {}
        for column in select_dict.split(','):
          index = column.lower().find(' as ')
          if index != -1:
            real_select_dict[column[index + 4:].strip()] = column[:index].strip()
          else:
            real_select_dict[column.strip()] = None
        select_dict = real_select_dict
      else:
        select_dict = None
    elif isinstance(select_dict, (list, tuple)):
2371
      select_dict = dict.fromkeys(select_dict)
2372 2373
    # Handle left_join_list
    left_join_list = kw.pop('left_join_list', ())
2374 2375 2376 2377 2378
    # Handle implicit_join. It's True by default, as there's a lot of code
    # in BT5s and elsewhere that calls buildSQLQuery() expecting implicit
    # join. self._queryResults() defaults it to False for those using
    # catalog.searchResults(...) or catalog(...) directly.
    implicit_join = kw.pop('implicit_join', True)
2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389
    # Handle order_by_list
    order_by_list = kw.pop('order_by_list', None)
    sort_on = kw.pop('sort_on', None)
    sort_order = kw.pop('sort_order', None)
    order_by_expression = kw.pop('order_by_expression', None)
    if order_by_list is None:
      order_by_list = self.buildOrderByList(
        sort_on=sort_on,
        sort_order=sort_order,
        order_by_expression=order_by_expression
      )
2390
    else:
2391
      if sort_on is not None:
2392
        LOG('SQLCatalog', WARNING, 'order_by_list and sort_on were given, ignoring sort_on.')
2393
      if sort_order is not None:
2394
        LOG('SQLCatalog', WARNING, 'order_by_list and sort_order were given, ignoring sort_order.')
2395
      if order_by_expression is not None:
2396
        LOG('SQLCatalog', WARNING, 'order_by_list and order_by_expression were given, ignoring order_by_expression.')
2397 2398 2399 2400 2401
    # Handle from_expression
    from_expression = kw.pop('from_expression', None)
    # Handle where_expression
    where_expression = kw.get('where_expression', None)
    if isinstance(where_expression, basestring) and len(where_expression):
2402
      LOG('SQLCatalog', INFO, 'Giving where_expression a string value is deprecated.')
2403 2404 2405 2406 2407 2408
      # Transform given where_expression into a query, and update kw.
      kw['where_expression'] = SQLQuery(where_expression)
    # Handle select_expression_key
    # It is required to support select_expression_key parameter for backward
    # compatiblity, but I'm not sure if there can be a serious use for it in
    # new API.
2409
    order_by_override_list = kw.pop('select_expression_key', ())
2410
    return EntireQuery(
2411 2412 2413 2414 2415
      query=self.buildQuery(kw, ignore_empty_string=ignore_empty_string),
      order_by_list=order_by_list,
      order_by_override_list=order_by_override_list,
      group_by_list=group_by_list,
      select_dict=select_dict,
2416
      left_join_list=left_join_list,
2417
      implicit_join=implicit_join,
2418 2419 2420 2421
      limit=limit,
      catalog_table_name=query_table,
      extra_column_list=extra_column_list,
      from_expression=from_expression)
2422 2423 2424 2425 2426 2427 2428 2429

  def buildSQLQuery(self, query_table='catalog', REQUEST=None,
                          ignore_empty_string=1, only_group_columns=False,
                          limit=None, extra_column_list=(),
                          **kw):
    query = self.buildEntireQuery(kw, query_table=query_table,
      ignore_empty_string=ignore_empty_string, limit=limit,
      extra_column_list=extra_column_list)
2430 2431
    result = query.asSQLExpression(self, only_group_columns).asSQLExpressionDict()
    return result
2432

2433 2434 2435
  # Compatibililty SQL Sql
  buildSqlQuery = buildSQLQuery

2436
  @transactional_cache_decorator('SQLCatalog._getSearchKeyDict')
2437 2438
  @caching_instance_method(id='SQLCatalog._getSearchKeyDict',
    cache_factory='erp5_content_long',
2439
    cache_id_generator=generateCatalogCacheId,
2440
  )
2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
  def _getSearchKeyDict(self):
    result = {}
    search_key_column_dict = {
      'KeywordKey': self.sql_catalog_keyword_search_keys,
      'FullTextKey': self.sql_catalog_full_text_search_keys,
      'DateTimeKey': self.sql_catalog_datetime_search_keys,
    }
    for key, column_list in search_key_column_dict.iteritems():
      for column in column_list:
        if column in result:
2451
          LOG('SQLCatalog', WARNING, 'Ambiguous configuration: column %r is set to use %r key, but also to use %r key. Former takes precedence.' % (column, result[column], key))
2452 2453
        else:
          result[column] = key
2454 2455 2456 2457 2458 2459
    for line in self.sql_catalog_search_keys:
      try:
        column, key = [x.strip() for x in line.split('|', 2)]
        result[column] = key
      except ValueError:
        LOG('SQLCatalog', WARNING, 'Wrong configuration for sql_catalog_search_keys: %r' % line)
2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486
    return result

  def getSearchKey(self, column, search_key=None):
    """
      Return an instance of a SearchKey class.

      column (string)
        The column for which the search key will be returned.
      search_key (string)
        If given, must be the name of a SearchKey class to be returned.
        Returned value will be an instance of that class, even if column has
        been configured to use a different one.
    """
    if search_key is None:
      search_key = self._getSearchKeyDict().get(column, 'DefaultKey')
    return getSearchKeyInstance(search_key, column)

  def getComparisonOperator(self, operator):
    """
      Return an instance of an Operator class.

      operator (string)
        String defining the expected operator class.
        See Operator module to have a list of available operators.
    """
    return getComparisonOperatorInstance(operator)

2487 2488 2489 2490 2491
  PROPAGATE_PARAMETER_SET = ('selection_domain',
                             'selection_report',
                             # XXX should get the next parameters from
                             # the ZSQLMethod class itself
                             'zsql_brain')
2492
  def _queryResults(self, REQUEST=None, build_sql_query_method=None, **kw):
2493
    """ Returns a list of brains from a set of constraints on variables """
2494 2495
    if build_sql_query_method is None:
      build_sql_query_method = self.buildSQLQuery
2496
    kw.setdefault('implicit_join', False)
2497
    query = build_sql_query_method(REQUEST=REQUEST, **kw)
2498 2499 2500 2501
    # XXX: decide if this should be made normal
    ENFORCE_SEPARATION = True
    if ENFORCE_SEPARATION:
      # Some parameters must be propagated:
2502 2503
      kw = {name: kw[name] for name in self.PROPAGATE_PARAMETER_SET
                           if name in kw}
2504 2505 2506
    kw['where_expression'] = query['where_expression']
    kw['sort_on'] = query['order_by_expression']
    kw['from_table_list'] = query['from_table_list']
2507
    kw['from_expression'] = query['from_expression']
Yoshinori Okuji's avatar
Yoshinori Okuji committed
2508
    kw['limit_expression'] = query['limit_expression']
2509
    kw['select_expression'] = query['select_expression']
2510
    kw['group_by_expression'] = query['group_by_expression']
2511
    # XXX: why not kw.update(query)??
2512
    return kw
2513

2514 2515 2516
  def queryResults(self, sql_method, REQUEST=None, src__=0, build_sql_query_method=None, **kw):
    sql_kw = self._queryResults(REQUEST=REQUEST, build_sql_query_method=build_sql_query_method, **kw)
    return sql_method(src__=src__, **sql_kw)
2517

2518 2519 2520
  def getSearchResultsMethod(self):
    return getattr(self, self.sql_search_results)

2521
  def searchResults(self, REQUEST=None, **kw):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2522
    """ Returns a list of brains from a set of constraints on variables """
2523 2524 2525 2526
    if 'only_group_columns' in kw:
      # searchResults must be consistent in API with countResults
      raise ValueError('only_group_columns does not belong to this API '
        'level, use queryResults directly')
2527 2528 2529 2530 2531 2532
    return self.queryResults(
      self.getSearchResultsMethod(),
      REQUEST=REQUEST,
      extra_column_list=self.getCatalogSearchResultKeys(),
      **kw
    )
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2533 2534 2535

  __call__ = searchResults

2536 2537 2538
  def getCountResultsMethod(self):
    return getattr(self, self.sql_count_results)

2539
  def countResults(self, REQUEST=None, **kw):
2540
    """ Returns the number of items which satisfy the where_expression """
2541 2542 2543 2544 2545 2546 2547
    return self.queryResults(
      self.getCountResultsMethod(),
      REQUEST=REQUEST,
      extra_column_list=self.getCatalogSearchResultKeys(),
      only_group_columns=True,
      **kw
    )
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2548

2549 2550 2551
  def isAdvancedSearchText(self, search_text):
    return isAdvancedSearchText(search_text, self.isValidColumn)

2552
  def recordObjectList(self, path_list, catalog=1):
2553
    """
2554
      Record the path of an object being catalogged or uncatalogged.
2555
    """
2556 2557
    method = getattr(self, self.sql_record_object_list)
    method(path_list=path_list, catalog=catalog)
2558

2559
  def deleteRecordedObjectList(self, uid_list=()):
2560 2561 2562 2563
    """
      Delete all objects which contain any path.
    """
    method = getattr(self, self.sql_delete_recorded_object_list)
2564
    method(uid_list=uid_list)
2565

2566
  def readRecordedObjectList(self, catalog=1):
2567 2568 2569 2570
    """
      Read objects. Note that this might not return all objects since ZMySQLDA limits the max rows.
    """
    method = getattr(self, self.sql_read_recorded_object_list)
2571
    return method(catalog=catalog)
2572

2573 2574 2575 2576 2577 2578 2579 2580
  # Filtering
  def manage_editFilter(self, REQUEST=None, RESPONSE=None, URL1=None):
    """
    This methods allows to set a filter on each zsql method called,
    so we can test if we should or not call a zsql method, so we can
    increase a lot the speed.
    """
    if withCMF:
2581
      method_id_list = [zsql_method.id for zsql_method in self.getFilterableMethodList()]
2582

2583 2584 2585 2586
      # Remove unused filters.
      for id in self.filter_dict.keys():
        if id not in method_id_list:
          del self.filter_dict[id]
2587

2588
      for id in method_id_list:
2589 2590 2591
        # We will first look if the filter is activated
        if not self.filter_dict.has_key(id):
          self.filter_dict[id] = PersistentMapping()
2592

2593 2594 2595 2596 2597
        if REQUEST.has_key('%s_box' % id):
          self.filter_dict[id]['filtered'] = 1
        else:
          self.filter_dict[id]['filtered'] = 0

2598 2599 2600 2601
        expression = REQUEST.get('%s_expression' % id, '').strip()
        self.filter_dict[id]['expression'] = expression
        if expression:
          self.filter_dict[id]['expression_instance'] = Expression(expression)
2602 2603 2604 2605 2606
        else:
          self.filter_dict[id]['expression_instance'] = None

        if REQUEST.has_key('%s_type' % id):
          list_type = REQUEST['%s_type' % id]
2607
          if isinstance(list_type, str):
2608 2609 2610 2611 2612
            list_type = [list_type]
          self.filter_dict[id]['type'] = list_type
        else:
          self.filter_dict[id]['type'] = []

2613 2614
        self.filter_dict[id]['expression_cache_key'] = \
          tuple(sorted(REQUEST.get('%s_expression_cache_key' % id, '').split()))
2615

2616 2617 2618 2619 2620 2621 2622 2623 2624 2625
    if RESPONSE and URL1:
      RESPONSE.redirect(URL1 + '/manage_catalogFilter?manage_tabs_message=Filter%20Changed')

  def isMethodFiltered(self, method_name):
    """
    Returns 1 if the method is already filtered,
    else it returns 0
    """
    if withCMF:
      # Reset Filtet dict
2626
      if getattr(aq_base(self), 'filter_dict', None) is None:
2627 2628
        self.filter_dict = PersistentMapping()
        return 0
2629
      try:
2630
        return self.filter_dict[method_name]['filtered']
2631 2632
      except KeyError:
        return 0
2633 2634 2635
    return 0

  def getExpression(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2636
    """ Get the filter expression text for this method.
2637 2638
    """
    if withCMF:
2639
      if getattr(aq_base(self), 'filter_dict', None) is None:
2640 2641
        self.filter_dict = PersistentMapping()
        return ""
2642
      try:
2643
        return self.filter_dict[method_name]['expression']
2644 2645
      except KeyError:
        return ""
2646 2647
    return ""

2648 2649 2650 2651 2652 2653 2654 2655 2656
  def getExpressionCacheKey(self, method_name):
    """ Get the key string which is used to cache results
        for the given expression.
    """
    if withCMF:
      if getattr(aq_base(self), 'filter_dict', None) is None:
        self.filter_dict = PersistentMapping()
        return ""
      try:
2657
        return ' '.join(self.filter_dict[method_name]['expression_cache_key'])
2658 2659 2660 2661
      except KeyError:
        return ""
    return ""

2662
  def getExpressionInstance(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2663
    """ Get the filter expression instance for this method.
2664 2665
    """
    if withCMF:
2666
      if getattr(aq_base(self), 'filter_dict', None) is None:
2667 2668
        self.filter_dict = PersistentMapping()
        return None
2669
      try:
2670
        return self.filter_dict[method_name]['expression_instance']
2671 2672
      except KeyError:
        return None
2673 2674
    return None

Rafael Monnerat's avatar
Rafael Monnerat committed
2675
  def setFilterExpression(self, method_name, expression):
2676
    """ Set the Expression for a certain method name. This allow set
Rafael Monnerat's avatar
Rafael Monnerat committed
2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688
        expressions by scripts.
    """
    if withCMF:
      if getattr(aq_base(self), 'filter_dict', None) is None:
        self.filter_dict = PersistentMapping()
        return None
      self.filter_dict[method_name]['expression'] = expression
      if expression:
        self.filter_dict[method_name]['expression_instance'] = Expression(expression)
      else:
        self.filter_dict[method_name]['expression_instance'] = None

Jérome Perrin's avatar
Jérome Perrin committed
2689 2690
  def isPortalTypeSelected(self, method_name, portal_type):
    """ Returns true if the portal type is selected for this method.
2691
      XXX deprecated
2692 2693
    """
    if withCMF:
2694
      if getattr(aq_base(self), 'filter_dict', None) is None:
2695 2696
        self.filter_dict = PersistentMapping()
        return 0
2697 2698 2699 2700
      try:
        return portal_type in (self.filter_dict[method_name]['type'])
      except KeyError:
        return 0
2701 2702
    return 0

2703 2704 2705
  def getFilteredPortalTypeList(self, method_name):
    """ Returns the list of portal types which define
        the filter.
2706
      XXX deprecated
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716
    """
    if withCMF:
      if getattr(aq_base(self), 'filter_dict', None) is None:
        self.filter_dict = PersistentMapping()
        return []
      try:
        return self.filter_dict[method_name]['type']
      except KeyError:
        return []
    return []
2717

Rafael Monnerat's avatar
Rafael Monnerat committed
2718 2719 2720
  def getFilterDict(self):
    """
      Utility Method.
2721
      Filter Dict is a dictionary and used at Python Scripts,
Rafael Monnerat's avatar
Rafael Monnerat committed
2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736
      This method returns a filter dict as a dictionary.
    """
    if withCMF:
      if getattr(aq_base(self), 'filter_dict', None) is None:
        self.filter_dict = PersistentMapping()
        return None
      filter_dict = {}
      for key in self.filter_dict:
        # Filter is also a Persistence dict.
        filter_dict[key] = {}
        for sub_key in self.filter_dict[key]:
           filter_dict[key][sub_key] = self.filter_dict[key][sub_key]
      return filter_dict
    return None

2737 2738 2739 2740 2741 2742
  def getFilterableMethodList(self):
    """
    Returns only zsql methods wich catalog or uncatalog objets
    """
    method_dict = {}
    if withCMF:
2743 2744 2745 2746
      methods = getattr(self,'sql_catalog_object',()) + \
                getattr(self,'sql_uncatalog_object',()) + \
                getattr(self,'sql_update_object',()) + \
                getattr(self,'sql_catalog_object_list',())
2747 2748 2749 2750 2751 2752 2753 2754
      for method_id in methods:
        method_dict[method_id] = 1
    method_list = map(lambda method_id: getattr(self, method_id, None), method_dict.keys())
    return filter(lambda method: method is not None, method_list)

  def getExpressionContext(self, ob):
      '''
      An expression context provides names for TALES expressions.
2755
      XXX deprecated
2756 2757 2758 2759 2760
      '''
      if withCMF:
        data = {
            'here':         ob,
            'container':    aq_parent(aq_inner(ob)),
2761 2762 2763 2764
            #'root':         ob.getPhysicalRoot(),
            #'request':      getattr( ob, 'REQUEST', None ),
            #'modules':      SecureModuleImporter,
            #'user':         getSecurityManager().getUser(),
2765 2766 2767 2768 2769 2770 2771 2772 2773
            # XXX these below are defined, because there is no
            # accessor for some attributes, and restricted environment
            # may not access them directly.
            'isDelivery':   ob.isDelivery,
            'isMovement':   ob.isMovement,
            'isPredicate':  ob.isPredicate,
            'isDocument':   ob.isDocument,
            'isInventory':  ob.isInventory,
            'isInventoryMovement': ob.isInventoryMovement,
2774 2775 2776 2777
            }
        return getEngine().getContext(data)


2778
InitializeClass(Catalog)
2779

Jean-Paul Smets's avatar
Jean-Paul Smets committed
2780
class CatalogError(Exception): pass
Ivan Tyagov's avatar
Ivan Tyagov committed
2781

2782 2783
from Query.Query import Query as _Query
from Query.SimpleQuery import SimpleQuery
Ivan Tyagov's avatar
Ivan Tyagov committed
2784
from Query.ComplexQuery import ComplexQuery
2785 2786
from Query.AutoQuery import AutoQuery
Query = AutoQuery
2787 2788

def NegatedQuery(query):
2789
  return ComplexQuery(query, logical_operator='not')
2790

2791 2792 2793 2794 2795 2796
def AndQuery(*args):
  return ComplexQuery(logical_operator='and', *args)

def OrQuery(*args):
  return ComplexQuery(logical_operator='or', *args)

2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821
allow_class(SimpleQuery)
allow_class(ComplexQuery)

import SearchKey
SEARCH_KEY_INSTANCE_POOL = {}
SEARCH_KEY_CLASS_CACHE = {}

def getSearchKeyInstance(search_key_class_name, column):
  assert isinstance(search_key_class_name, basestring)
  try:
    search_key_class = SEARCH_KEY_CLASS_CACHE[search_key_class_name]
  except KeyError:
    search_key_class = getattr(getattr(SearchKey, search_key_class_name),
                               search_key_class_name)
    SEARCH_KEY_CLASS_CACHE[search_key_class_name] = search_key_class
  try:
    instance_dict = SEARCH_KEY_INSTANCE_POOL[search_key_class]
  except KeyError:
    instance_dict = SEARCH_KEY_INSTANCE_POOL[search_key_class] = {}
  try:
    result = instance_dict[column]
  except KeyError:
    result = instance_dict[column] = search_key_class(column)
  return result

2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848
class SearchKeyWrapperForScriptableKey(SearchKey.SearchKey.SearchKey):
  """
    This SearchKey is a simple wrapper around a ScriptableKey, so such script
    can be used in place of a regular SearchKey.
  """
  default_comparison_operator = None
  get_operator_from_value = False

  def __init__(self, column, script):
    self.script = script
    super(SearchKeyWrapperForScriptableKey, self).__init__(column)

  def buildQuery(self, search_value, group=None, logical_operator=None,
                 comparison_operator=None):
    # XXX: It would be better to extend ScriptableKey API to support other
    # parameters.
    if group is not None:
      raise ValueError, 'ScriptableKey cannot be used inside a group ' \
        '(%r given).' % (group, )
    if logical_operator is not None:
      raise ValueError, 'ScriptableKey ignores logical operators ' \
        '(%r given).' % (logical_operator, )
    if comparison_operator != '':
      raise ValueError, 'ScriptableKey ignores comparison operators ' \
        '(%r given).' % (comparison_operator, )
    return self.script(search_value)

2849 2850 2851 2852 2853 2854 2855 2856
from Operator import operator_dict
def getComparisonOperatorInstance(operator):
  return operator_dict[operator]

from Query.EntireQuery import EntireQuery
from Query.SQLQuery import SQLQuery

verifyClass(ISearchKeyCatalog, Catalog)