SQLCatalog.py 104 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 BTrees.OIBTree import OIBTree
26 27
from App.config import getConfiguration
from BTrees.Length import Length
28
from Shared.DC.ZRDB.TM import TM
Jean-Paul Smets's avatar
Jean-Paul Smets committed
29

30
from Acquisition import aq_parent, aq_inner, aq_base
31
from zLOG import LOG, WARNING, INFO, TRACE, ERROR
32
from ZODB.POSException import ConflictError
33
from Products.PythonScripts.Utility import allow_class
Jean-Paul Smets's avatar
Jean-Paul Smets committed
34 35

import time
Jean-Paul Smets's avatar
Jean-Paul Smets committed
36
import sys
37 38
import urllib
import string
39
import pprint
40 41
import re
import warnings
42
from contextlib import contextmanager
43
from cStringIO import StringIO
44
from xml.dom.minidom import parse
45
from xml.sax.saxutils import escape, quoteattr
46
import os
47
from hashlib import md5
48

49
from interfaces.query_catalog import ISearchKeyCatalog
50 51
from zope.interface.verify import verifyClass
from zope.interface import implements
52

53
from SearchText import isAdvancedSearchText, dequote
54

55 56 57 58
# Try to import ActiveObject in order to make SQLCatalog active
try:
  from Products.CMFActivity.ActiveObject import ActiveObject
except ImportError:
59
  ActiveObject = ExtensionClass.Base
60 61


62 63 64 65 66 67
PROFILING_ENABLED = False
if PROFILING_ENABLED:
  from tiny_profiler import profiler_decorator, profiler_report, profiler_reset
else:
  def profiler_decorator(func):
    return func
68 69 70 71 72

try:
  from Products.CMFCore.Expression import Expression
  from Products.PageTemplates.Expressions import getEngine
  from Products.CMFCore.utils import getToolByName
73
  new_context_search = re.compile(r'\bcontext\b').search
74 75 76 77
  withCMF = 1
except ImportError:
  withCMF = 0

78 79 80 81
try:
  import psyco
except ImportError:
  psyco = None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
82

83 84 85
@contextmanager
def noReadOnlyTransactionCache():
  yield
86
try:
87 88
  from Products.ERP5Type.Cache import \
    readOnlyTransactionCache, caching_instance_method
89
except ImportError:
90
  LOG('SQLCatalog', WARNING, 'Count not import caching_instance_method, expect slowness.')
91
  def caching_instance_method(*args, **kw):
92
    return lambda method: method
93
  readOnlyTransactionCache = noReadOnlyTransactionCache
94 95 96 97

try:
  from Products.ERP5Type.TransactionalVariable import getTransactionalVariable
except ImportError:
98
  LOG('SQLCatalog', WARNING, 'Count not import getTransactionalVariable, expect slowness.')
99
  def getTransactionalVariable():
100 101
    return {}

102 103
def getInstanceID(instance):
  # XXX: getPhysicalPath is overkill for a unique cache identifier.
104 105 106
  # 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).
107 108 109 110
  return instance.getPhysicalPath()

def generateCatalogCacheId(method_id, self, *args, **kwd):
  return str((method_id, self.getCacheSequenceNumber(), getInstanceID(self),
Julien Muchembled's avatar
Julien Muchembled committed
111
    args, kwd))
112

113 114 115 116 117 118 119 120 121 122
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):
123
      transactional_cache = getTransactionalVariable()
124
      cache_id = str((self.cache_id,
125
        wrapped_self.getCacheSequenceNumber(),
126 127
        getInstanceID(wrapped_self),
      ))
128
      try:
129
        result = transactional_cache[cache_id]
130
      except KeyError:
131 132
        result = transactional_cache[cache_id] = method(wrapped_self)
      return result
133 134
    return wrapper

135
list_type_list = list, tuple, set, frozenset
136 137 138 139
try:
  from ZPublisher.HTTPRequest import record
except ImportError:
  dict_type_list = (dict, )
140 141
else:
  dict_type_list = (dict, record)
142 143


144
UID_BUFFER_SIZE = 300
145
OBJECT_LIST_SIZE = 300 # XXX 300 is arbitrary value of catalog object list
146
MAX_PATH_LEN = 255
147
RESERVED_KEY_LIST = ('where_expression', 'sort-on', 'sort_on', 'sort-order', 'sort_order', 'limit',
148 149
                     'format', 'search_mode', 'operator', 'selection_domain', 'selection_report',
                     'extra_key_list', )
150

151
valid_method_meta_type_list = ('Z SQL Method', 'LDIF Method', 'Script (Python)') # Nicer
152

153
manage_addSQLCatalogForm = DTMLFile('dtml/addSQLCatalog',globals())
154

155 156 157 158 159
# Here go uid buffers
# Structure:
#  global_uid_buffer_dict[catalog_path][thread_id] = UidBuffer
global_uid_buffer_dict = {}

160
def manage_addSQLCatalog(self, id, title,
161
             vocab_id='create_default_catalog_', # vocab_id is a strange name - not abbreviation
162 163 164
             REQUEST=None):
  """Add a Catalog object
  """
165 166 167
  id = str(id)
  title = str(title)
  vocab_id = str(vocab_id)
168 169 170
  if vocab_id == 'create_default_catalog_':
    vocab_id = None

171
  c = Catalog(id, title, self)
172 173 174 175
  self._setObject(id, c)
  if REQUEST is not None:
    return self.manage_main(self, REQUEST,update_menu=1)

176 177
class UidBuffer(TM):
  """Uid Buffer class caches a list of reserved uids in a transaction-safe way."""
178

Yoshinori Okuji's avatar
Yoshinori Okuji committed
179
  def __init__(self):
180
    """Initialize some variables.
181

182
      temporary_buffer is used to hold reserved uids created by non-committed transactions.
183

184
      finished_buffer is used to hold reserved uids created by committed-transactions.
185

186
      This distinction is important, because uids by non-committed transactions might become
Yoshinori Okuji's avatar
Yoshinori Okuji committed
187
      invalid afterwards, so they may not be used by other transactions."""
188 189
    self.temporary_buffer = {}
    self.finished_buffer = []
190

191 192 193 194 195 196 197 198
  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
199

200 201 202 203 204 205 206
  def _abort(self):
    """Erase the uids in the temporary buffer."""
    tid = get_ident()
    try:
      del self.temporary_buffer[tid]
    except KeyError:
      pass
207

208 209 210 211 212 213 214 215
  def __len__(self):
    tid = get_ident()
    l = len(self.finished_buffer)
    try:
      l += len(self.temporary_buffer[tid])
    except KeyError:
      pass
    return l
216

217 218 219 220 221 222 223 224 225 226 227
  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
228

229 230 231 232
  def pop(self):
    self._register()
    tid = get_ident()
    try:
233
      uid = self.temporary_buffer[tid].pop()
234
    except (KeyError, IndexError):
235 236
      uid = self.finished_buffer.pop()
    return uid
237

238 239 240
  def extend(self, iterable):
    self._register()
    tid = get_ident()
Yoshinori Okuji's avatar
Yoshinori Okuji committed
241
    self.temporary_buffer.setdefault(tid, []).extend(iterable)
242

243
related_key_definition_cache = {}
244
related_key_warned_column_set = set()
245

246 247 248
class Catalog(Folder,
              Persistent,
              Acquisition.Implicit,
249
              ActiveObject,
250
              OFS.History.Historical):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
251 252 253 254
  """ An Object Catalog

  An Object Catalog maintains a table of object metadata, and a
  series of manageable indexes to quickly search for objects
255
  (references in the metadata) that satisfy a search where_expression.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
256 257 258 259 260 261 262 263

  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

264 265
  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.
266 267 268
  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).
269 270 271
  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
272

273
 """
274

275
  implements(ISearchKeyCatalog)
276

277

278
  meta_type = "SQLCatalog"
279
  icon = 'misc_/ZCatalog/ZCatalog.gif' # FIXME: use a different icon
280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
  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'),}
309
    ) + OFS.History.Historical.manage_options
310

311
  __ac_permissions__= (
312 313 314 315 316

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

      'manage_catalogView', 'manage_catalogFind',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
317 318
      'manage_catalogFilter',
      'manage_catalogAdvanced',
319 320

      'manage_catalogReindex', 'manage_catalogFoundItems',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
321 322
      'manage_catalogClear',
      'manage_main',
323 324 325 326 327 328
      'manage_editFilter',
      ],
     ['Manager']),

    ('Search ZCatalog',
     ['searchResults', '__call__', 'uniqueValuesFor',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
329 330
      'all_meta_types', 'valid_roles',
      'getCatalogSearchTableIds',
331
      'getFilterableMethodList',],
332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
     ['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' },
357 358 359 360 361
    { 'id'      : 'sql_catalog_reserve_uid',
      'description' : 'A method to reserve a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
362 363 364 365 366
    { 'id'      : 'sql_catalog_delete_uid',
      'description' : 'A method to delete a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
367 368
    { 'id'      : 'sql_catalog_object_list',
      'description' : 'Methods to be called to catalog the list of objects',
369 370 371 372 373 374 375 376
      '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' },
377 378 379
    { 'id'      : 'sql_catalog_translation_list',
      'description' : 'Methods to be called to catalog the list of translation objects',
      'type'    : 'selection',
380 381
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
382 383 384
    { 'id'      : 'sql_delete_translation_list',
      'description' : 'Methods to be called to delete translations',
      'type'    : 'selection',
385 386
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
387 388
    { 'id'      : 'sql_clear_catalog',
      'description' : 'The methods which should be called to clear the catalog',
389 390 391
      'type'    : 'multiple selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
392
    { 'id'      : 'sql_record_object_list',
393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
      '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
412 413 414 415 416
    { 'id'      : 'sql_search_security',
      'description' : 'Main method to search security',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
    { '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' },
452 453 454 455 456
    { 'id'      : 'sql_catalog_index',
      'description' : 'Method to get the main catalog index',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
457 458 459 460 461 462 463 464 465 466
    { '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' },
467 468 469 470 471
    { 'id': 'sql_catalog_search_keys',
      'title': 'Search Key Mappings',
      'description': 'A list of Search Key mappings',
      'type': 'lines',
      'mode': 'w' },
472
    { 'id'      : 'sql_catalog_keyword_search_keys',
473
      'description' : 'Columns which should be considered as keyword search',
474 475 476
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
477
    { 'id'      : 'sql_catalog_datetime_search_keys',
478
      'description' : 'Columns which should be considered as datetime search',
479 480 481
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496
    { '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' },
497 498 499 500 501
    { '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' },
502 503 504 505 506 507 508 509 510
    { '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' },
511 512 513 514 515
    { 'id'      : 'sql_catalog_scriptable_keys',
      'title'   : 'Related keys',
      'description' : 'Virtual columns to generate scriptable scriptable queries',
      'type'    : 'lines',
      'mode'    : 'w' },
516 517 518 519 520 521 522 523 524 525 526
    { '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' },
527 528 529 530 531 532 533 534
    { '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' },
535 536 537 538 539 540
    { 'id': 'sql_catalog_table_vote_scripts',
      'title': 'Table vote scripts',
      'description': 'Scripts helping column mapping resolution',
      'type': 'multiple selection',
      'select_variable' : 'getPythonMethodIds',
      'mode': 'w' },
541 542 543 544 545 546
    { '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' },
547

548 549
  )

550
  sql_catalog_produce_reserved = ''
551
  sql_catalog_delete_uid = ''
552 553 554 555 556 557 558 559 560 561 562
  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
563
  sql_search_security = ''
564 565 566 567 568 569
  sql_count_results = ''
  sql_getitem_by_path = ''
  sql_getitem_by_uid = ''
  sql_catalog_tables = ''
  sql_search_tables = ()
  sql_catalog_schema = ''
570
  sql_catalog_index = ''
571 572
  sql_unique_values = ''
  sql_catalog_paths = ''
573
  sql_catalog_search_keys = ()
574
  sql_catalog_keyword_search_keys =  ()
575
  sql_catalog_datetime_search_keys = ()
576 577 578 579 580
  sql_catalog_full_text_search_keys = ()
  sql_catalog_request_keys = ()
  sql_search_result_keys = ()
  sql_catalog_topic_search_keys = ()
  sql_catalog_multivalue_keys = ()
581
  sql_catalog_index_on_order_keys = ()
582
  sql_catalog_related_keys = ()
583
  sql_catalog_scriptable_keys = ()
584 585
  sql_catalog_role_keys = ()
  sql_catalog_local_role_keys = ()
586
  sql_catalog_security_uid_columns = (' | security_uid',)
587
  sql_catalog_table_vote_scripts = ()
588
  sql_catalog_raise_error_on_uid_check = True
589

590 591 592 593 594 595
  # 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
596

597 598 599 600 601 602 603
  # 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
604

605 606 607 608 609
  manage_catalogView = DTMLFile('dtml/catalogView',globals())
  manage_catalogFilter = DTMLFile('dtml/catalogFilter',globals())
  manage_catalogFind = DTMLFile('dtml/catalogFind',globals())
  manage_catalogAdvanced = DTMLFile('dtml/catalogAdvanced', globals())

610 611
  _cache_sequence_number = 0

612 613 614 615 616
  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
617 618 619
    self.schema = {}  # mapping from attribute name to column
    self.names = {}   # mapping from column to attribute name
    self.indexes = {}   # empty mapping
620
    self.filter_dict = PersistentMapping()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
621

622 623 624 625 626 627
  def getCacheSequenceNumber(self):
    return self._cache_sequence_number

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

628 629 630 631
  def getSQLCatalogRoleKeysList(self):
    """
    Return the list of role keys.
    """
632 633 634 635 636
    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()
637

638 639 640 641 642 643 644 645 646 647 648 649
  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

650 651 652 653
  def getSQLCatalogLocalRoleKeysList(self):
    """
    Return the list of local role keys.
    """
654 655 656 657
    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()
658
    return local_role_key_dict.items()
659

660 661 662 663 664 665
  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')
666 667 668 669 670 671 672 673
    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
674
    property_list.sort(key=lambda x: x[0])
675 676 677
    for property in property_list:
      property_id = property[0]
      value       = property[1]
678
      if isinstance(value, basestring):
679
        f.write('  <property id=%s type="str">%s</property>\n' % (quoteattr(property_id), escape(value)))
680
      elif isinstance(value, (tuple, list)):
681 682 683
        f.write('  <property id=%s type="tuple">\n' % quoteattr(property_id))
        # Sort for easy diff
        item_list = []
684
        for item in value:
685
          if isinstance(item, basestring):
686 687 688 689
            item_list.append(item)
        item_list.sort()
        for item in item_list:
          f.write('    <item type="str">%s</item>\n' % escape(str(item)))
690
        f.write('  </property>\n')
691 692 693
    # 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'):
694 695
      filter_list = []
      for filter_id in self.filter_dict.keys():
696
        filter_definition = self.filter_dict[filter_id]
697 698
        filter_list.append((filter_id, filter_definition))
      # Sort for easy diff
699
      filter_list.sort(key=lambda x: x[0])
700 701 702 703
      for filter_item in filter_list:
        filter_id  = filter_item[0]
        filter_def = filter_item[1]
        if not filter_def['filtered']:
704 705
          # If a filter is not activated, no need to output it.
          continue
706
        if not filter_def['expression']:
707 708
          # If the expression is not specified, meaningless to specify it.
          continue
709
        f.write('  <filter id=%s expression=%s />\n' % (quoteattr(filter_id), quoteattr(filter_def['expression'])))
710
        # For now, portal types are not exported, because portal types are too specific to each site.
711 712 713 714 715 716 717 718 719 720 721 722
    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.
    """
723
    with open(file) as f:
724 725 726 727 728 729 730
      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):
731
            raise CatalogError, 'unknown property id %r' % (id,)
732
          if type not in ('str', 'tuple'):
733
            raise CatalogError, 'unknown property type %r' % (type,)
734 735 736 737
          if type == 'str':
            value = ''
            for text in prop.childNodes:
              if text.nodeType == text.TEXT_NODE:
738
                value = str(text.data)
739 740 741 742 743 744
                break
          else:
            value = []
            for item in prop.getElementsByTagName("item"):
              item_type = item.getAttribute("type")
              if item_type != 'str':
745
                raise CatalogError, 'unknown item type %r' % (item_type,)
746 747
              for text in item.childNodes:
                if text.nodeType == text.TEXT_NODE:
748
                  value.append(str(text.data))
749 750 751 752
                  break
            value = tuple(value)

          setattr(self, id, value)
753

754 755 756
        if not hasattr(self, 'filter_dict'):
          self.filter_dict = PersistentMapping()
        for filt in root.getElementsByTagName("filter"):
757
          id = str(filt.getAttribute("id"))
758 759 760 761 762 763 764 765 766 767 768 769
          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
770 771
      finally:
        doc.unlink()
Aurel's avatar
Aurel committed
772

773 774 775 776 777 778 779
  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__)))
780 781 782

  def _clearSecurityCache(self):
    self.security_uid_dict = OIBTree()
783
    self.security_uid_index = None
784

785 786 787 788
  def _clearSubjectCache(self):
    self.subject_set_uid_dict = OIBTree()
    self.subject_set_uid_index = None

789 790
  security.declarePrivate('getSecurityUidDict')
  def getSecurityUidDict(self, wrapped_object):
791
    """
792 793
    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.
794
    """
795
    if getattr(aq_base(self), 'security_uid_dict', None) is None:
796
      self._clearSecurityCache()
797

798
    id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835

    optimised_roles_and_users = []
    local_roles_group_id_to_security_uid_mapping= dict()

    # Get security information
    for local_roles_group_id, allowed_roles_and_users in\
          wrapped_object.getLocalRolesGroupIdDict().iteritems():
      allowed_roles_and_users = tuple(sorted(allowed_roles_and_users))

      key = (local_roles_group_id, 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:
        # 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
        if id_tool is not None:
          default = 1
          # We must keep compatibility with existing sites
          previous_security_uid = getattr(self, 'security_uid_index', None)
          if previous_security_uid is not None:
            # At some point, it was a Length
            if isinstance(previous_security_uid, Length):
              default = previous_security_uid() + 1
            else:
              default = previous_security_uid
          security_uid = int(id_tool.generateNewId(id_generator='uid',
              id_group='security_uid_index', default=default))
836
        else:
837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855
          previous_security_uid = getattr(self, 'security_uid_index', None)
          if previous_security_uid is None:
            previous_security_uid = 0
          # At some point, it was a Length
          if isinstance(previous_security_uid, Length):
            previous_security_uid = previous_security_uid()
          security_uid = previous_security_uid + 1
          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)
856

857 858 859 860 861 862 863
  def getRoleAndSecurityUidList(self):
    """
      Return a list of 2-tuples, suitable for direct use in a zsqlmethod.
      Goal: make it possible to regenerate a table containing this data.
    """
    result = []
    extend = result.extend
864 865
    for role_list, security_uid in getattr(
            aq_base(self), 'security_uid_dict', {}).iteritems():
866 867 868
      extend([(role, security_uid) for role in role_list])
    return result

869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911
  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.
    subject_list = tuple(sorted(set([(x or '').lower() for x in getSubjectList()])))
    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
912 913 914 915 916 917
  def clear(self):
    """
    Clears the catalog by calling a list of methods
    """
    methods = self.sql_clear_catalog
    for method_name in methods:
918
      method = getattr(self, method_name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
919 920
      try:
        method()
921 922
      except ConflictError:
        raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
923
      except:
924
        LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
925
            'could not clear catalog with %s' % method_name, error=sys.exc_info())
926
        raise
927

928
    # Reserved uids have been removed.
929
    self.clearReserved()
930

931 932 933 934
    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()
935

936
    self._clearSecurityCache()
937
    self._clearSubjectCache()
938
    self._clearCaches()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
939

940 941 942 943 944 945 946 947 948 949
  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()])

950 951 952 953 954 955
  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
956 957 958 959 960
    try:
      method()
    except ConflictError:
      raise
    except:
961
      LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
962 963 964
          'could not clear reserved catalog with %s' % \
              method_id, error=sys.exc_info())
      raise
965
    self._last_clear_reserved_time += 1
966

967
  def getRecordForUid(self, uid):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
968 969 970 971
    """
    Get an object by UID
    Note: brain is defined in Z SQL Method object
    """
972
    # this method used to be __getitem__(self, uid) but was found to hurt more
973
    # than it helped: It would be inadvertently called by
974 975 976 977 978 979
    # (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
980 981 982 983
    method = getattr(self,  self.sql_getitem_by_uid)
    search_result = method(uid = uid)
    if len(search_result) > 0:
      return search_result[0]
984
    raise KeyError, uid
Jean-Paul Smets's avatar
Jean-Paul Smets committed
985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004

  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

1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
  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

1022
  security.declarePublic('getCatalogSearchResultKeys')
1023 1024 1025 1026
  def getCatalogSearchResultKeys(self):
    """Return search result keys.
    """
    return self.sql_search_result_keys
1027

1028
  def _getCatalogSchema(self, table=None):
1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
    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)
1042

1043
  @caching_instance_method(id='SQLCatalog.getColumnIds',
1044
    cache_factory='erp5_content_long',
1045
    cache_id_generator=generateCatalogCacheId,
1046
  )
1047
  def _getColumnIds(self):
1048 1049
    keys = set()
    add_key = keys.add
1050 1051 1052
    for table in self.getCatalogSearchTableIds():
      field_list = self._getCatalogSchema(table=table)
      for field in field_list:
1053 1054
        add_key(field)
        add_key('%s.%s' % (table, field))  # Is this inconsistent ?
1055 1056
    for related in self.getSQLCatalogRelatedKeyList():
      related_tuple = related.split('|')
1057
      add_key(related_tuple[0].strip())
1058 1059
    for scriptable in self.getSQLCatalogScriptableKeyList():
      scriptable_tuple = scriptable.split('|')
1060 1061
      add_key(scriptable_tuple[0].strip())
    return sorted(keys)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1062

1063 1064 1065 1066 1067 1068 1069
  def getColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
    return self._getColumnIds()[:]

1070 1071 1072
  @profiler_decorator
  @transactional_cache_decorator('SQLCatalog.getColumnMap')
  @profiler_decorator
1073 1074
  @caching_instance_method(id='SQLCatalog.getColumnMap',
    cache_factory='erp5_content_long',
1075
    cache_id_generator=generateCatalogCacheId,
1076
  )
1077
  @profiler_decorator
1078 1079 1080 1081 1082
  def getColumnMap(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
1083 1084 1085 1086 1087 1088
    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
1089

1090
  @profiler_decorator
1091
  @transactional_cache_decorator('SQLCatalog.getResultColumnIds')
1092
  @profiler_decorator
1093
  @caching_instance_method(id='SQLCatalog.getResultColumnIds',
1094
    cache_factory='erp5_content_long',
1095
    cache_id_generator=generateCatalogCacheId,
1096 1097
  )
  @profiler_decorator
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1098 1099 1100 1101 1102 1103 1104
  def getResultColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
    keys = {}
    for table in self.getCatalogSearchTableIds():
1105 1106 1107
      field_list = self._getCatalogSchema(table=table)
      for field in field_list:
        keys['%s.%s' % (table, field)] = 1
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1108 1109 1110 1111
    keys = keys.keys()
    keys.sort()
    return keys

1112 1113 1114 1115 1116
  @profiler_decorator
  @transactional_cache_decorator('SQLCatalog.getSortColumnIds')
  @profiler_decorator
  @caching_instance_method(id='SQLCatalog.getSortColumnIds',
      cache_factory='erp5_content_long',
1117
      cache_id_generator=generateCatalogCacheId,
1118 1119
  )
  @profiler_decorator
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133
  def getSortColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids that can be used for a sort
    """
    keys = {}
    for table in self.getTableIds():
      field_list = self._getCatalogSchema(table=table)
      for field in field_list:
        keys['%s.%s' % (table, field)] = 1
    keys = keys.keys()
    keys.sort()
    return keys

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1134 1135 1136
  def getTableIds(self):
    """
    Calls the show table method and returns dictionnary of
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1137 1138 1139
    Field Ids
    """
    keys = []
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1140 1141
    method_name = self.sql_catalog_tables
    try:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1142 1143 1144
      method = getattr(self,  method_name)
      search_result = method()
      for c in search_result:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1145
        keys.append(c[0])
1146 1147
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1148 1149
    except:
      pass
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1150 1151
    return keys

1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
  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]
1165

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1166
  # the cataloging API
1167 1168 1169
  def produceUid(self):
    """
      Produces reserved uids in advance
1170
    """
1171 1172 1173 1174
    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.
1175 1176 1177 1178
    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:
1179 1180 1181
      id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
      if id_tool is not None:
        if self._max_uid is None:
1182
          self._max_uid = Length(1)
1183
        uid_list = id_tool.generateNewIdList(id_generator='uid', id_group='catalog_uid',
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
                     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
1199
        instance_id = md5(str(random_factor_list)).hexdigest()
1200
        uid_list = [x.uid for x in method(count = UID_BUFFER_SIZE, instance_id = instance_id) if x.uid != 0]
1201
      uid_buffer.extend(uid_list)
1202

1203 1204 1205 1206 1207 1208
  def isIndexable(self):
    """
    This is required to check in many methods that
    the site root and zope root are indexable
    """
    zope_root = self.getZopeRoot()
1209
    site_root = self.getSiteRoot() # XXX-JPS - Why don't we use getPortalObject here ?
1210 1211 1212 1213 1214 1215

    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
1216

1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
  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

1237 1238 1239
  def newUid(self):
    """
      This is where uid generation takes place. We should consider a multi-threaded environment
1240 1241
      with multiple ZEO clients on a single ZEO server.

1242
      The main risk is the following:
1243

1244
      - objects a/b/c/d/e/f are created (a is parent of b which is parent of ... of f)
1245

1246
      - one reindexing node N1 starts reindexing f
1247

1248
      - another reindexing node N2 starts reindexing e
1249

1250 1251 1252
      - 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

1253
      Similar problems may happen with relations and acquisition of uid values (ex. order_uid)
1254
      with the risk of graph loops
1255
    """
1256
    if not self.isIndexable():
1257 1258
      return None

1259 1260 1261 1262
    klass = self.__class__
    try:
      klass._reserved_uid_lock.acquire()
      self.produceUid()
1263 1264 1265
      uid_buffer = self.getUIDBuffer()
      if len(uid_buffer) > 0:
        uid = uid_buffer.pop()
1266
        if self._max_uid is None:
1267
          self._max_uid = Length(1)
1268 1269
        if uid > self._max_uid():
          self._max_uid.set(uid)
1270
        return long(uid)
1271 1272 1273 1274
      else:
        raise CatalogError("Could not retrieve new uid")
    finally:
      klass._reserved_uid_lock.release()
1275

1276 1277 1278
  def manage_catalogObject(self, REQUEST, RESPONSE, URL1, urls=None):
    """ index Zope object(s) that 'urls' point to """
    if urls:
1279
      if isinstance(urls, str):
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
        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:
1295
      if isinstance(urls, str):
1296 1297 1298 1299 1300 1301 1302
        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')

1303
  def manage_catalogReindex(self, REQUEST, RESPONSE, URL1, urls=None):
1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318
    """ 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`)))

1319
  def manage_catalogClear(self, REQUEST=None, RESPONSE=None,
Romain Courteaud's avatar
Romain Courteaud committed
1320
                          URL1=None, sql_catalog_id=None):
1321
    """ clears the whole enchilada """
1322 1323
    self.beforeCatalogClear()

1324 1325 1326
    self.clear()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1327
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1328
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1329 1330

  def manage_catalogClearReserved(self, REQUEST=None, RESPONSE=None, URL1=None):
1331
    """ clears reserved uids """
1332 1333 1334
    self.clearReserved()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1335
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1336
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373

  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
1374
  def catalogObject(self, object, path, is_object_moved=0):
1375 1376
    """Add an object to the Catalog by calling all SQL methods and
    providing needed arguments.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1377

1378 1379
    'object' is the object to be catalogged."""
    self._catalogObjectList([object])
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1380

1381
  def catalogObjectList(self, object_list, method_id_list=None,
1382 1383 1384
                        disable_cache=0, check_uid=1, idxs=None):
    """Add objects to the Catalog by calling all SQL methods and
    providing needed arguments.
1385

1386 1387
      method_id_list : specify which methods should be used. If not
                       set, it will take the default value of portal_catalog.
1388 1389

      disable_cache : do not use cache, so values will be computed each time,
1390 1391 1392 1393 1394 1395
                      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.
1396

1397 1398 1399 1400
    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."""
1401 1402
    for i in xrange(0, len(object_list), OBJECT_LIST_SIZE):
      self._catalogObjectList(object_list[i:i + OBJECT_LIST_SIZE],
1403 1404 1405 1406
                              method_id_list=method_id_list,
                              disable_cache=disable_cache,
                              check_uid=check_uid,
                              idxs=idxs)
1407 1408

  def _catalogObjectList(self, object_list, method_id_list=None,
1409 1410 1411 1412 1413
                         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?"""
1414
    LOG('SQLCatalog', TRACE, 'catalogging %d objects' % len(object_list))
1415
    #LOG('catalogObjectList', 0, 'called with %r' % (object_list,))
1416

1417
    if idxs not in (None, []):
1418
      LOG('ZSLQCatalog.SQLCatalog:catalogObjectList', WARNING,
1419
          'idxs is ignored in this function and is only provided to be compatible with CMFCatalogAware.reindexObject.')
1420

1421
    if not self.isIndexable():
1422
      return
1423

1424 1425
    # Reminder about optimization: It might be possible to issue just one
    # query to get enought results to check uid & path consistency.
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
    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)

1444 1445 1446 1447 1448
    # 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 = {}

1449
    for object in object_list:
1450 1451 1452 1453
      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:
1454
        try:
1455
          object.uid = self.newUid()
1456 1457
        except ConflictError:
          raise
1458
        except:
1459
          raise RuntimeError, 'could not set missing uid for %r' % (object,)
1460
      elif check_uid:
1461
        if uid in assigned_uid_dict:
1462 1463 1464 1465 1466 1467 1468
            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)
1469

1470
        path = object.getPath()
1471
        index = path_uid_dict.get(path)
1472 1473 1474
        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)
1475 1476
          if uid != index or isinstance(uid, int):
            # We want to make sure that uid becomes long if it is an int
1477 1478 1479 1480 1481 1482
            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)
1483 1484 1485 1486
        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
1487
          if uid in uid_path_dict:
1488 1489 1490
            catalog_path = uid_path_dict.get(uid)
          else:
            catalog_path = self.getPathForUid(uid)
1491 1492 1493
          #LOG('catalogObject', 0, 'uid = %r, catalog_path = %r' % (uid, catalog_path))
          if catalog_path == "reserved":
            # Reserved line in catalog table
1494
            lock = self.__class__._reserved_uid_lock
1495
            try:
1496
              lock.acquire()
1497 1498
              uid_buffer = self.getUIDBuffer()
              if uid_buffer is not None:
1499 1500 1501 1502 1503 1504 1505 1506
                # 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.
1507
                try:
1508
                  uid_buffer.remove(uid)
1509 1510
                except ValueError:
                  pass
1511
            finally:
1512
              lock.release()
1513 1514
          elif catalog_path is not None:
            # An uid conflict happened... Why?
1515
            # can be due to path length
1516
            if len(path) > MAX_PATH_LEN:
1517
              LOG('SQLCatalog', ERROR, 'path of object %r is too long for catalog. You should use a shorter path.' %(object,))
1518

1519
            LOG('SQLCatalog', ERROR,
1520 1521
                'uid of %r changed from %r to %r as old one is assigned'
                ' to %s in catalog !!! This can be fatal.' % (
1522 1523 1524 1525 1526 1527 1528 1529 1530
                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)
1531 1532 1533
            uid = object.uid

        assigned_uid_dict[uid] = object
1534

1535 1536
    if method_id_list is None:
      method_id_list = self.sql_catalog_object_list
1537
    econtext = getEngine().getContext()
1538 1539
    argument_cache = {}

1540 1541
    with (noReadOnlyTransactionCache if disable_cache else
          readOnlyTransactionCache)():
1542
      filter_dict = self.filter_dict
1543
      catalogged_object_list_cache = {}
1544
      for method_name in method_id_list:
1545 1546 1547 1548 1549
        # 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:
1550
          filter = filter_dict[method_name]
1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
          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:
1581
              if expression_cache_key_list:
1582 1583 1584 1585
                # 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
1586 1587
                # objects but we could also use over multiple transactions
                # if this can improve performance significantly
1588 1589 1590
                # ZZZ - we could find a way to compute this once only
                cache_key = tuple(object.getProperty(key) for key
                                  in expression_cache_key_list)
1591
                try:
1592 1593 1594
                  if expression_result_cache[cache_key]:
                    append(object)
                  continue
1595
                except KeyError:
1596 1597 1598
                  pass
              if old_context:
                result = expression(self.getExpressionContext(object))
1599
              else:
1600
                econtext.setLocal('context', object)
1601
                result = expression(econtext)
1602
              if expression_cache_key_list:
1603
                expression_result_cache[cache_key] = result
1604 1605
              if result:
                append(object)
1606

1607
        if not catalogged_object_list:
1608
          continue
1609

1610 1611
        #LOG('catalogObjectList', 0, 'method_name = %s' % (method_name,))
        method = getattr(self, method_name)
1612
        if method.meta_type in ("Z SQL Method", "LDIF Method"):
1613
          # Build the dictionnary of values
1614
          arguments = method.arguments_src.split()
1615 1616 1617
        elif method.meta_type == "Script (Python)":
          arguments = \
            method.func_code.co_varnames[:method.func_code.co_argcount]
1618 1619
        else:
          arguments = []
1620
        kw = {}
1621 1622 1623 1624
        for arg in arguments:
          value_list = []
          append = value_list.append
          for object in catalogged_object_list:
1625
            argument_cache_key = (object.uid, arg)
1626
            try:
1627
              value = argument_cache[argument_cache_key]
1628
            except KeyError:
1629
              try:
1630 1631 1632 1633 1634 1635
                value = getattr(object, arg, None)
                if callable(value):
                  value = value()
              except ConflictError:
                raise
              except:
1636
                LOG('SQLCatalog', WARNING, 'Failed to call method %s on %r' %
1637
                    (arg, object), error=sys.exc_info())
1638 1639
                value = None
              if not disable_cache:
1640
                argument_cache[argument_cache_key] = value
1641 1642
            append(value)
          kw[arg] = value_list
1643

1644 1645 1646
        # Alter/Create row
        try:
          #start_time = DateTime()
1647
          #LOG('catalogObjectList', DEBUG, 'kw = %r, method_name = %r' % (kw, method_name))
1648 1649 1650 1651 1652 1653 1654
          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]))
1655

1656 1657 1658
        except ConflictError:
          raise
        except:
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1659
          LOG('SQLCatalog', WARNING, 'could not catalog objects %s with method %s' % (object_list, method_name),
1660 1661
              error=sys.exc_info())
          raise
1662

1663 1664
  if psyco is not None:
    psyco.bind(_catalogObjectList)
1665

1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
  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
1678 1679 1680 1681
    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
1682
      return self.uncatalogObject(path=path,uid=uid)
1683 1684 1685
    method = getattr(self, method_name)
    method(uid = uid)

1686
  def uncatalogObject(self, path=None, uid=None):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
    """
    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

    """
1699
    if not self.isIndexable():
1700 1701
      return None

1702 1703
    if uid is None and path is not None:
      uid = self.getUidForPath(path)
1704 1705
    methods = self.sql_uncatalog_object
    if uid is None:
1706
      return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1707
    for method_name in methods:
1708 1709
      # 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
1710
      method = getattr(self, method_name)
1711
      method(uid = uid)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1712

1713 1714 1715 1716
  def catalogTranslationList(self, object_list):
    """Catalog translations.
    """
    method_name = self.sql_catalog_translation_list
1717 1718
    return self.catalogObjectList(object_list, method_id_list = (method_name,),
                                  check_uid=0)
1719

1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730
  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())
1731

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1732 1733 1734
  def uniqueValuesFor(self, name):
    """ return unique values for FieldIndex name """
    method = getattr(self, self.sql_unique_values)
1735
    return method(column=name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1736 1737 1738 1739 1740 1741 1742 1743

  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 """
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
    #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
1754 1755
      return None

1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794
  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
1795
          uid_path_dict[uid] = search_result[0].path
1796 1797
    return uid_path_dict

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1798 1799 1800 1801 1802 1803 1804 1805 1806
  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
1807 1808 1809 1810
      try:
        int(uid)
      except ValueError:
        return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1811 1812 1813 1814 1815 1816 1817 1818
      # 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
1819 1820
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1821 1822 1823
    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
1824
      LOG('SQLCatalog', WARNING, "could not find path from uid %s" % (uid,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
      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
1853 1854
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1855 1856 1857
    except:
      # This is a real LOG message
      # which is required in order to be able to import .zexp files
1858 1859
      LOG('SQLCatalog', WARNING,
          "could not find metadata from path %s" % (path,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1860 1861 1862 1863 1864 1865
      return None

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

1866 1867
  def getCatalogMethodIds(self,
      valid_method_meta_type_list=valid_method_meta_type_list):
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878
    """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
1879 1880
            if not isinstance(id, str):
              id=id()
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891
            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

1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
  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)', ))

  @profiler_decorator
  @transactional_cache_decorator('SQLCatalog._getSQLCatalogRelatedKeyList')
  @profiler_decorator
  def _getSQLCatalogRelatedKeySet(self):
    column_map = self.getColumnMap()
    column_set = set(column_map)
    for related_key in self.sql_catalog_related_keys:
1906 1907
      split_entire_definition = related_key.split('|')
      if len(split_entire_definition) != 2:
1908
        LOG('SQLCatalog', WARNING, 'Malformed related key definition: %r. Ignored.' % (related_key, ))
1909 1910
        continue
      related_key_id = split_entire_definition[0].strip()
1911 1912 1913
      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)
1914
        if related_key_id in column_map:
1915
          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]))
1916
        else:
1917
          LOG('SQLCatalog', WARNING, 'Related key %r is declared more than once.' % (related_key_id, ))
1918 1919 1920
      column_set.add(related_key_id)
    return column_set

1921
  def getSQLCatalogRelatedKeyList(self, key_list=None):
1922 1923
    """
    Return the list of related keys.
1924
    This method can be overidden in order to implement
1925 1926
    dynamic generation of some related keys.
    """
1927 1928
    if key_list is None:
      key_list = []
1929
    column_map = self._getSQLCatalogRelatedKeySet()
1930
    # Do not generate dynamic related key for acceptable_keys
1931
    dynamic_key_list = [k for k in key_list if k not in column_map]
1932
    dynamic_list = self.getDynamicRelatedKeyList(dynamic_key_list)
1933 1934 1935
    full_list = list(dynamic_list) + list(self.sql_catalog_related_keys)
    return full_list

1936 1937 1938
  # Compatibililty SQL Sql
  getSqlCatalogRelatedKeyList = getSQLCatalogRelatedKeyList

1939 1940 1941 1942 1943
  def getSQLCatalogScriptableKeyList(self):
    """
    Return the list of scriptable keys.
    """
    return self.sql_catalog_scriptable_keys
1944

1945
  @caching_instance_method(id='SQLCatalog.getTableIndex',
1946
    cache_factory='erp5_content_long',
1947
    cache_id_generator=generateCatalogCacheId,
1948
  )
1949
  def _getTableIndex(self, table):
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959
    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,]
1960 1961 1962 1963 1964 1965 1966
    return table_index

  def getTableIndex(self, table):
    """
    Return a map between index and column for a given table
    """
    return self._getTableIndex(table).copy()
1967

1968 1969 1970 1971 1972
  @profiler_decorator
  def isValidColumn(self, column_id):
    """
      Tells wether given name is or not an existing column.

1973 1974
      Warning: This includes "virtual" columns, such as related keys and
      scriptable keys.
1975
    """
1976
    result = self.getScriptableKeyScript(column_id) is not None
1977
    if not result:
1978 1979 1980
      result = column_id in self.getColumnMap()
      if not result:
        result = self.getRelatedKeyDefinition(column_id) is not None
1981 1982
    return result

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
  @profiler_decorator
  def getRelatedKeyDefinition(self, key):
    """
      Returns the definition of given related key name if found, None
      otherwise.
    """
    try:
      result = related_key_definition_cache[key]
    except KeyError:
      result = None
      for entire_definition in self.getSQLCatalogRelatedKeyList([key]):
1994 1995
        split_entire_definition = entire_definition.split('|')
        if len(split_entire_definition) != 2:
1996
          LOG('SQLCatalog', WARNING, 'Malformed related key definition: %r. Ignored.' % (entire_definition, ))
1997 1998
          continue
        name, definition = [x.strip() for x in split_entire_definition]
1999 2000 2001 2002 2003 2004
        if name == key:
          result = definition
          break
      if result is not None:
        related_key_definition_cache[key] = result
    return result
2005

2006 2007 2008 2009 2010 2011
  @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:
2012
        LOG('SQLCatalog', WARNING, 'Malformed scriptable key definition: %r. Ignored.' % (scriptable_key_definition, ))
2013 2014 2015 2016
        continue
      key, script_id = [x.strip() for x in split_scriptable_key_definition]
      script = getattr(self, script_id, None)
      if script is None:
2017
        LOG('SQLCatalog', WARNING, 'Scriptable key %r script %r is missing.' \
2018 2019 2020 2021 2022 2023 2024 2025
            ' Skipped.' % (key, script_id))
      else:
        result[key] = script
    return result

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

2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036
  @profiler_decorator
  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
2037 2038 2039

      If both a related key and a real column are found, the related key
      is used.
2040
    """
2041 2042 2043 2044 2045
    # 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)
2046
      else:
2047 2048 2049
        search_key = None
    else:
      search_key = self.getSearchKey(key, 'RelatedKey')
2050 2051
    return search_key, related_key_definition

2052 2053 2054
  def hasColumn(self, column):
    return self.getColumnSearchKey(column)[0] is not None

2055
  @profiler_decorator
2056
  def getColumnDefaultSearchKey(self, key, search_key_name=None):
2057 2058 2059 2060
    """
      Return a SearchKey instance which would ultimately receive the value
      associated with given key.
    """
2061 2062
    search_key, related_key_definition = self.getColumnSearchKey(key,
      search_key_name=search_key_name)
2063 2064 2065 2066
    if search_key is None:
      result = None
    else:
      if related_key_definition is not None:
2067 2068
        search_key = search_key.getSearchKey(sql_catalog=self,
          related_key_definition=related_key_definition)
2069 2070 2071 2072 2073 2074 2075 2076
    return search_key

  @profiler_decorator
  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.
    """
2077 2078 2079 2080 2081
    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
2082
      else:
2083
        if related_key_definition is None:
2084
          build_key = search_key
2085
        else:
2086 2087 2088
          build_key = search_key.getSearchKey(sql_catalog=self,
            related_key_definition=related_key_definition,
            search_key_name=search_key_name)
2089 2090 2091
        result = build_key.buildQuery(value, logical_operator=logical_operator,
                                      comparison_operator=comparison_operator)
        if related_key_definition is not None:
2092 2093 2094
          result = search_key.buildQuery(sql_catalog=self,
            related_key_definition=related_key_definition,
            search_value=result)
2095 2096
    else:
      result = script(value)
2097 2098 2099
    return result

  @profiler_decorator
2100
  def _buildQueryFromAbstractSyntaxTreeNode(self, node, search_key, wrap):
2101 2102 2103 2104
    if search_key.dequoteParsedText():
      _dequote = dequote
    else:
      _dequote = lambda x: x
2105
    if node.isLeaf():
2106
      result = search_key.buildQuery(wrap(_dequote(node.getValue())),
2107
        comparison_operator=node.getComparisonOperator())
2108 2109 2110 2111 2112 2113 2114 2115
    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():
2116
          value_dict.setdefault(subnode.getComparisonOperator(),
2117
            []).append(wrap(_dequote(subnode.getValue())))
2118
        else:
2119
          subquery = self._buildQueryFromAbstractSyntaxTreeNode(subnode, search_key, wrap)
2120 2121
          if subquery is not None:
            append(subquery)
2122
      logical_operator = node.getLogicalOperator()
2123 2124 2125 2126
      if logical_operator == 'not':
        query_logical_operator = None
      else:
        query_logical_operator = logical_operator
2127
      for comparison_operator, value_list in value_dict.iteritems():
2128
        append(search_key.buildQuery(value_list, comparison_operator=comparison_operator, logical_operator=query_logical_operator))
2129 2130
      if logical_operator == 'not' or len(query_list) > 1:
        result = ComplexQuery(query_list, logical_operator=logical_operator)
2131 2132
      elif len(query_list) == 1:
        result = query_list[0]
2133
      else:
2134 2135 2136
        result = None
    return result

2137
  @profiler_decorator
2138
  def buildQueryFromAbstractSyntaxTreeNode(self, node, key, wrap=lambda x: x):
2139 2140 2141 2142 2143 2144 2145 2146 2147 2148
    """
      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).

2149
      Expected node API is described in interfaces/abstract_syntax_node.py .
2150
    """
2151 2152 2153 2154 2155 2156
    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
2157 2158
    if search_key is None:
      # Unknown, skip loudly
2159
      LOG('SQLCatalog', WARNING, 'Unknown column %r, skipped.' % (key, ))
2160
      result = None
2161 2162 2163 2164
    else:
      if related_key_definition is None:
        build_key = search_key
      else:
2165 2166
        build_key = search_key.getSearchKey(sql_catalog=self,
          related_key_definition=related_key_definition)
2167 2168
      result = self._buildQueryFromAbstractSyntaxTreeNode(node, build_key,
        wrap)
2169
      if related_key_definition is not None:
2170 2171 2172
        result = search_key.buildQuery(sql_catalog=self,
          related_key_definition=related_key_definition,
          search_value=result)
2173 2174
    return result

2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
  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)

2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198
  @profiler_decorator
  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 = {}
2199 2200 2201
    # empty_value_dict: contains all keys whose value causes them to be
    # discarded.
    empty_value_dict = {}
2202 2203 2204 2205 2206 2207 2208
    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 == ''
2209
          or (isinstance(value, list_type_list) and not value)
2210 2211 2212
          or (isinstance(value, dict) and (
            'query' not in value
            or value['query'] == ''
2213 2214
            or (isinstance(value['query'], list_type_list)
              and not value['query'])))):
2215
        # We have an empty value, do not create a query from it
2216
        empty_value_dict[key] = value
2217
      else:
2218
        script = self.getScriptableKeyScript(key)
2219
        if isinstance(value, dict):
2220 2221 2222 2223 2224 2225
          # 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'
2226 2227 2228 2229
          # Backward compatibility: former "ExactMatch" is now only available
          # as "RawKey"
          elif search_key_name == 'ExactMatch':
            search_key_name = value['key'] = 'RawKey'
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260
        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)
2261 2262 2263 2264 2265 2266 2267 2268
        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)
2269 2270
    if len(empty_value_dict):
      LOG('SQLCatalog', WARNING, 'Discarding columns with empty values: %r' % (empty_value_dict, ))
2271
    if len(unknown_column_dict):
2272
      LOG('SQLCatalog', WARNING, 'Unknown columns %r, skipped.' % (unknown_column_dict.keys(), ))
2273 2274
    return ComplexQuery(query_list, logical_operator=operator,
        unknown_column_dict=unknown_column_dict)
2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294

  @profiler_decorator
  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:
2295
        LOG('SQLCatalog', WARNING, 'order_by_expression (%r) and sort_on (%r) were given. Ignoring order_by_expression.' % (order_by_expression, sort_on))
2296 2297 2298 2299 2300
      if not isinstance(sort_on, (tuple, list)):
        sort_on = [[sort_on]]
      for item in sort_on:
        if isinstance(item, (tuple, list)):
          item = list(item)
2301
        else:
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320
          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

  @profiler_decorator
2321 2322
  def buildEntireQuery(self, kw, query_table='catalog', ignore_empty_string=1,
                       limit=None, extra_column_list=()):
2323
    group_by_list = kw.pop('group_by_list', kw.pop('group_by', kw.pop('group_by_expression', ())))
2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340
    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)):
      select_dict = dict([(x, None) for x in select_dict])
2341 2342
    # Handle left_join_list
    left_join_list = kw.pop('left_join_list', ())
2343 2344 2345 2346 2347
    # 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)
2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358
    # 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
      )
2359
    else:
2360
      if sort_on is not None:
2361
        LOG('SQLCatalog', WARNING, 'order_by_list and sort_on were given, ignoring sort_on.')
2362
      if sort_order is not None:
2363
        LOG('SQLCatalog', WARNING, 'order_by_list and sort_order were given, ignoring sort_order.')
2364
      if order_by_expression is not None:
2365
        LOG('SQLCatalog', WARNING, 'order_by_list and order_by_expression were given, ignoring order_by_expression.')
2366 2367 2368 2369 2370
    # 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):
2371
      LOG('SQLCatalog', INFO, 'Giving where_expression a string value is deprecated.')
2372 2373 2374 2375 2376 2377
      # 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.
2378
    order_by_override_list = kw.pop('select_expression_key', ())
2379
    return EntireQuery(
2380 2381 2382 2383 2384
      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,
2385
      left_join_list=left_join_list,
2386
      implicit_join=implicit_join,
2387 2388 2389 2390
      limit=limit,
      catalog_table_name=query_table,
      extra_column_list=extra_column_list,
      from_expression=from_expression)
2391 2392 2393 2394 2395 2396 2397 2398 2399

  @profiler_decorator
  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)
2400 2401
    result = query.asSQLExpression(self, only_group_columns).asSQLExpressionDict()
    return result
2402

2403 2404 2405
  # Compatibililty SQL Sql
  buildSqlQuery = buildSQLQuery

2406 2407 2408
  @profiler_decorator
  @transactional_cache_decorator('SQLCatalog._getSearchKeyDict')
  @profiler_decorator
2409 2410
  @caching_instance_method(id='SQLCatalog._getSearchKeyDict',
    cache_factory='erp5_content_long',
2411
    cache_id_generator=generateCatalogCacheId,
2412
  )
2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423
  @profiler_decorator
  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:
2424
          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))
2425 2426
        else:
          result[column] = key
2427 2428 2429 2430 2431 2432
    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)
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460
    return result

  @profiler_decorator
  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)

2461 2462 2463 2464 2465 2466
  PROPAGATE_PARAMETER_SET = set(['selection_domain',
                                 'selection_report',
                                 # XXX should get the next parameters from
                                 # the ZSQLMethod class itself
                                 'zsql_brain',
                               ])
2467 2468
  @profiler_decorator
  def _queryResults(self, REQUEST=None, build_sql_query_method=None, **kw):
2469
    """ Returns a list of brains from a set of constraints on variables """
2470 2471
    if build_sql_query_method is None:
      build_sql_query_method = self.buildSQLQuery
2472
    kw.setdefault('implicit_join', False)
2473
    query = build_sql_query_method(REQUEST=REQUEST, **kw)
2474 2475 2476 2477
    # XXX: decide if this should be made normal
    ENFORCE_SEPARATION = True
    if ENFORCE_SEPARATION:
      # Some parameters must be propagated:
2478 2479 2480
      new_kw = dict((name, kw[name])
                    for name in self.PROPAGATE_PARAMETER_SET & set(kw))
      # discard all others:
2481
      kw = new_kw
2482 2483 2484
    kw['where_expression'] = query['where_expression']
    kw['sort_on'] = query['order_by_expression']
    kw['from_table_list'] = query['from_table_list']
2485
    kw['from_expression'] = query['from_expression']
Yoshinori Okuji's avatar
Yoshinori Okuji committed
2486
    kw['limit_expression'] = query['limit_expression']
2487
    kw['select_expression'] = query['select_expression']
2488
    kw['group_by_expression'] = query['group_by_expression']
2489
    # XXX: why not kw.update(query)??
2490
    return kw
2491

2492 2493 2494
  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)
2495

2496 2497 2498
  def getSearchResultsMethod(self):
    return getattr(self, self.sql_search_results)

2499
  def searchResults(self, REQUEST=None, **kw):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2500
    """ Returns a list of brains from a set of constraints on variables """
2501 2502 2503 2504
    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')
2505 2506 2507 2508 2509 2510
    return self.queryResults(
      self.getSearchResultsMethod(),
      REQUEST=REQUEST,
      extra_column_list=self.getCatalogSearchResultKeys(),
      **kw
    )
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2511 2512 2513

  __call__ = searchResults

2514 2515 2516
  def getCountResultsMethod(self):
    return getattr(self, self.sql_count_results)

2517
  def countResults(self, REQUEST=None, **kw):
2518
    """ Returns the number of items which satisfy the where_expression """
2519 2520 2521 2522 2523 2524 2525
    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
2526

2527 2528 2529
  def isAdvancedSearchText(self, search_text):
    return isAdvancedSearchText(search_text, self.isValidColumn)

2530
  def recordObjectList(self, path_list, catalog=1):
2531
    """
2532
      Record the path of an object being catalogged or uncatalogged.
2533
    """
2534 2535
    method = getattr(self, self.sql_record_object_list)
    method(path_list=path_list, catalog=catalog)
2536

2537
  def deleteRecordedObjectList(self, uid_list=()):
2538 2539 2540 2541
    """
      Delete all objects which contain any path.
    """
    method = getattr(self, self.sql_delete_recorded_object_list)
2542
    method(uid_list=uid_list)
2543

2544
  def readRecordedObjectList(self, catalog=1):
2545 2546 2547 2548
    """
      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)
2549
    return method(catalog=catalog)
2550

2551 2552 2553 2554 2555 2556 2557 2558
  # 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:
2559
      method_id_list = [zsql_method.id for zsql_method in self.getFilterableMethodList()]
2560

2561 2562 2563 2564
      # Remove unused filters.
      for id in self.filter_dict.keys():
        if id not in method_id_list:
          del self.filter_dict[id]
2565

2566
      for id in method_id_list:
2567 2568 2569
        # We will first look if the filter is activated
        if not self.filter_dict.has_key(id):
          self.filter_dict[id] = PersistentMapping()
2570

2571 2572 2573 2574 2575
        if REQUEST.has_key('%s_box' % id):
          self.filter_dict[id]['filtered'] = 1
        else:
          self.filter_dict[id]['filtered'] = 0

2576 2577 2578 2579
        expression = REQUEST.get('%s_expression' % id, '').strip()
        self.filter_dict[id]['expression'] = expression
        if expression:
          self.filter_dict[id]['expression_instance'] = Expression(expression)
2580 2581 2582 2583 2584
        else:
          self.filter_dict[id]['expression_instance'] = None

        if REQUEST.has_key('%s_type' % id):
          list_type = REQUEST['%s_type' % id]
2585
          if isinstance(list_type, str):
2586 2587 2588 2589 2590
            list_type = [list_type]
          self.filter_dict[id]['type'] = list_type
        else:
          self.filter_dict[id]['type'] = []

2591 2592
        self.filter_dict[id]['expression_cache_key'] = \
          tuple(sorted(REQUEST.get('%s_expression_cache_key' % id, '').split()))
2593

2594 2595 2596 2597 2598 2599 2600 2601 2602 2603
    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
2604
      if getattr(aq_base(self), 'filter_dict', None) is None:
2605 2606
        self.filter_dict = PersistentMapping()
        return 0
2607
      try:
2608
        return self.filter_dict[method_name]['filtered']
2609 2610
      except KeyError:
        return 0
2611 2612 2613
    return 0

  def getExpression(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2614
    """ Get the filter expression text for this method.
2615 2616
    """
    if withCMF:
2617
      if getattr(aq_base(self), 'filter_dict', None) is None:
2618 2619
        self.filter_dict = PersistentMapping()
        return ""
2620
      try:
2621
        return self.filter_dict[method_name]['expression']
2622 2623
      except KeyError:
        return ""
2624 2625
    return ""

2626 2627 2628 2629 2630 2631 2632 2633 2634
  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:
2635
        return ' '.join(self.filter_dict[method_name]['expression_cache_key'])
2636 2637 2638 2639
      except KeyError:
        return ""
    return ""

2640
  def getExpressionInstance(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2641
    """ Get the filter expression instance for this method.
2642 2643
    """
    if withCMF:
2644
      if getattr(aq_base(self), 'filter_dict', None) is None:
2645 2646
        self.filter_dict = PersistentMapping()
        return None
2647
      try:
2648
        return self.filter_dict[method_name]['expression_instance']
2649 2650
      except KeyError:
        return None
2651 2652
    return None

Rafael Monnerat's avatar
Rafael Monnerat committed
2653
  def setFilterExpression(self, method_name, expression):
2654
    """ Set the Expression for a certain method name. This allow set
Rafael Monnerat's avatar
Rafael Monnerat committed
2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666
        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
2667 2668
  def isPortalTypeSelected(self, method_name, portal_type):
    """ Returns true if the portal type is selected for this method.
2669
      XXX deprecated
2670 2671
    """
    if withCMF:
2672
      if getattr(aq_base(self), 'filter_dict', None) is None:
2673 2674
        self.filter_dict = PersistentMapping()
        return 0
2675 2676 2677 2678
      try:
        return portal_type in (self.filter_dict[method_name]['type'])
      except KeyError:
        return 0
2679 2680
    return 0

2681 2682 2683
  def getFilteredPortalTypeList(self, method_name):
    """ Returns the list of portal types which define
        the filter.
2684
      XXX deprecated
2685 2686 2687 2688 2689 2690 2691 2692 2693 2694
    """
    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 []
2695

Rafael Monnerat's avatar
Rafael Monnerat committed
2696 2697 2698
  def getFilterDict(self):
    """
      Utility Method.
2699
      Filter Dict is a dictionary and used at Python Scripts,
Rafael Monnerat's avatar
Rafael Monnerat committed
2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714
      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

2715 2716 2717 2718 2719 2720
  def getFilterableMethodList(self):
    """
    Returns only zsql methods wich catalog or uncatalog objets
    """
    method_dict = {}
    if withCMF:
2721 2722 2723 2724
      methods = getattr(self,'sql_catalog_object',()) + \
                getattr(self,'sql_uncatalog_object',()) + \
                getattr(self,'sql_update_object',()) + \
                getattr(self,'sql_catalog_object_list',())
2725 2726 2727 2728 2729 2730 2731 2732
      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.
2733
      XXX deprecated
2734 2735 2736 2737 2738
      '''
      if withCMF:
        data = {
            'here':         ob,
            'container':    aq_parent(aq_inner(ob)),
2739 2740 2741 2742
            #'root':         ob.getPhysicalRoot(),
            #'request':      getattr( ob, 'REQUEST', None ),
            #'modules':      SecureModuleImporter,
            #'user':         getSecurityManager().getUser(),
2743 2744 2745 2746 2747 2748 2749 2750 2751
            # 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,
2752 2753 2754 2755
            }
        return getEngine().getContext(data)


2756
InitializeClass(Catalog)
2757

Jean-Paul Smets's avatar
Jean-Paul Smets committed
2758
class CatalogError(Exception): pass
Ivan Tyagov's avatar
Ivan Tyagov committed
2759

2760 2761
from Query.Query import Query as _Query
from Query.SimpleQuery import SimpleQuery
Ivan Tyagov's avatar
Ivan Tyagov committed
2762
from Query.ComplexQuery import ComplexQuery
2763 2764 2765
from Query.AutoQuery import AutoQuery as Query

def NegatedQuery(query):
2766
  return ComplexQuery(query, logical_operator='not')
2767

2768 2769 2770 2771 2772 2773
def AndQuery(*args):
  return ComplexQuery(logical_operator='and', *args)

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

2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799
allow_class(SimpleQuery)
allow_class(ComplexQuery)

import SearchKey
SEARCH_KEY_INSTANCE_POOL = {}
SEARCH_KEY_CLASS_CACHE = {}

@profiler_decorator
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

2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826
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)

2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844
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)

if PROFILING_ENABLED:
  def Catalog_dumpProfilerData(self):
    return profiler_report()

  def Catalog_resetProfilerData(self):
    profiler_reset()

  Catalog.dumpProfilerData = Catalog_dumpProfilerData
  Catalog.resetProfilerData = Catalog_resetProfilerData
2845