SQLCatalog.py 99.5 KB
Newer Older
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
##############################################################################
#
# Copyright (c) 2002 Nexedi SARL. All Rights Reserved.
# 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
#
##############################################################################

from Persistence import Persistent
import Acquisition
import ExtensionClass
18
import Globals
19
import OFS.History
20
from Globals import DTMLFile, PersistentMapping
21
from string import split, join
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 DateTime import DateTime
31 32
from Acquisition import aq_parent, aq_inner, aq_base
from zLOG import LOG, WARNING, INFO, TRACE
33
from ZODB.POSException import ConflictError
34
from DocumentTemplate.DT_Var import sql_quote
35
from Products.PythonScripts.Utility import allow_class
Jean-Paul Smets's avatar
Jean-Paul Smets committed
36 37

import time
Jean-Paul Smets's avatar
Jean-Paul Smets committed
38
import sys
39 40
import urllib
import string
41
import pprint
42
from cStringIO import StringIO
43
from xml.dom.minidom import parse
44
from xml.sax.saxutils import escape, quoteattr
45 46
import os
import md5
47 48 49 50 51 52 53 54 55

try:
  from Products.CMFCore.Expression import Expression
  from Products.PageTemplates.Expressions import getEngine
  from Products.CMFCore.utils import getToolByName
  withCMF = 1
except ImportError:
  withCMF = 0

56 57 58 59
try:
  import psyco
except ImportError:
  psyco = None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
60

61
try:
62 63
  from Products.ERP5Type.Cache import enableReadOnlyTransactionCache
  from Products.ERP5Type.Cache import disableReadOnlyTransactionCache, CachingMethod
64 65 66
except ImportError:
  def doNothing(context):
    pass
67 68 69 70 71 72
  class CachingMethod:
    """
      Dummy CachingMethod class.
    """
    def __init__(self, callable, **kw):
      self.function = callable
Yoshinori Okuji's avatar
Yoshinori Okuji committed
73
    def __call__(self, *opts, **kw):
74
      return self.function(*opts, **kw)
75 76
  enableReadOnlyTransactionCache = doNothing
  disableReadOnlyTransactionCache = doNothing
77

78
UID_BUFFER_SIZE = 300
79 80
OBJECT_LIST_SIZE = 300
MAX_PATH_LEN = 255
81
RESERVED_KEY_LIST = ('where_expression', 'sort-on', 'sort_on', 'sort-order', 'sort_order', 'limit',
82
                     'format', 'search_mode', 'operator', 'selection_domain', 'selection_report')
83

84
valid_method_meta_type_list = ('Z SQL Method', 'LDIF Method', 'Script (Python)') # Nicer
85

86 87 88 89 90
full_text_search_modes = { 'natural': '',                                   # XXX-JPS probably not right place
                           'in_boolean_mode': 'IN BOOLEAN MODE',            # full_text_search_modes wrong naming
                           'with_query_expansion': 'WITH QUERY EXPANSION' } # according to ERP5 conventions
                                                                            # we really need a good grammar
                                                                            # and some cleanup
91

92
manage_addSQLCatalogForm = DTMLFile('dtml/addSQLCatalog',globals())
93

94 95 96 97 98
# Here go uid buffers
# Structure:
#  global_uid_buffer_dict[catalog_path][thread_id] = UidBuffer
global_uid_buffer_dict = {}

99
def manage_addSQLCatalog(self, id, title,
100
             vocab_id='create_default_catalog_', # vocab_id is a strange name - not abbreviation
101 102 103
             REQUEST=None):
  """Add a Catalog object
  """
104 105 106
  id = str(id)
  title = str(title)
  vocab_id = str(vocab_id)
107 108 109
  if vocab_id == 'create_default_catalog_':
    vocab_id = None

110
  c = Catalog(id, title, self)
111 112 113 114
  self._setObject(id, c)
  if REQUEST is not None:
    return self.manage_main(self, REQUEST,update_menu=1)

115 116 117 118 119 120
def isSimpleType(value):
  return isinstance(value, basestring) or \
         isinstance(value, int) or \
         isinstance(value, long) or \
         isinstance(value, float)

121

122 123
class UidBuffer(TM):
  """Uid Buffer class caches a list of reserved uids in a transaction-safe way."""
124

Yoshinori Okuji's avatar
Yoshinori Okuji committed
125
  def __init__(self):
126
    """Initialize some variables.
127

128
      temporary_buffer is used to hold reserved uids created by non-committed transactions.
129

130
      finished_buffer is used to hold reserved uids created by committed-transactions.
131

132
      This distinction is important, because uids by non-committed transactions might become
Yoshinori Okuji's avatar
Yoshinori Okuji committed
133
      invalid afterwards, so they may not be used by other transactions."""
134 135
    self.temporary_buffer = {}
    self.finished_buffer = []
136

137 138 139 140 141 142 143 144
  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
145

146 147 148 149 150 151 152
  def _abort(self):
    """Erase the uids in the temporary buffer."""
    tid = get_ident()
    try:
      del self.temporary_buffer[tid]
    except KeyError:
      pass
153

154 155 156 157 158 159 160 161
  def __len__(self):
    tid = get_ident()
    l = len(self.finished_buffer)
    try:
      l += len(self.temporary_buffer[tid])
    except KeyError:
      pass
    return l
162

163 164 165 166 167 168 169 170 171 172 173
  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
174

175 176 177 178
  def pop(self):
    self._register()
    tid = get_ident()
    try:
179
      uid = self.temporary_buffer[tid].pop()
180
    except (KeyError, IndexError):
181 182
      uid = self.finished_buffer.pop()
    return uid
183

184 185 186
  def extend(self, iterable):
    self._register()
    tid = get_ident()
Yoshinori Okuji's avatar
Yoshinori Okuji committed
187
    self.temporary_buffer.setdefault(tid, []).extend(iterable)
188

189 190 191 192 193

# valid search modes for queries
FULL_TEXT_SEARCH_MODE = 'FullText'
EXACT_MATCH_SEARCH_MODE = 'ExactMatch'
KEYWORD_SEARCH_MODE = 'Keyword'
194
DATETIME_SEARCH_MODE = 'DateTime'
195 196


197
class QueryMixin:
198 199 200 201 202
  """
    Mixing class which implements methods which are
    common to all kinds of Queries
  """

203 204 205
  operator = None
  format = None
  type = None
206 207 208 209

  def getOperator(self):
    return self.operator

210 211 212 213 214 215
  def getFormat(self):
    return self.format

  def getType(self):
    return self.type

Sebastien Robin's avatar
Sebastien Robin committed
216 217 218
  def getLogicalOperator(self):
    return self.logical_operator

219 220 221
  def _quoteSQLString(self, value):
    """Return a quoted string of the value.
    """
Sebastien Robin's avatar
Sebastien Robin committed
222 223 224 225 226 227 228 229
    format = self.getFormat()
    type = self.getType()
    if format is not None and type is not None:
      if type == 'date':
        if hasattr(value, 'strftime'):
          value = value.strftime(format)
        if isinstance(value, basestring):
          value = "STR_TO_DATE('%s','%s')" % (value, format)
230 231 232
      if type == 'float':
        # Make sure there is no space in float values
        value = value.replace(' ','')
233
        value = "'%s'" % value
234
    else:
235 236
      if getattr(value, 'ISO', None) is not None:
        value = "'%s'" % value.toZone('UTC').ISO()
Sebastien Robin's avatar
Sebastien Robin committed
237 238
      else:
        value = "'%s'" % sql_quote(str(value))
239 240
    return value

Sebastien Robin's avatar
Sebastien Robin committed
241 242 243 244 245 246 247 248
  def _quoteSQLKey(self, key):
    """Return a quoted string of the value.
    """
    format = self.getFormat()
    type = self.getType()
    if format is not None and type is not None:
      if type == 'date':
        key = "STR_TO_DATE(DATE_FORMAT(%s,'%s'),'%s')" % (key, format, format)
249 250 251 252 253
      if type == 'float':
        float_format = format.replace(' ','')
        if float_format.find('.') >= 0:
          precision = len(float_format.split('.')[1])
          key = "TRUNCATE(%s,%s)" % (key, precision)
Sebastien Robin's avatar
Sebastien Robin committed
254 255
    return key

256 257
  def asSQLExpression(self, key_alias_dict=None,
                      keyword_search_keys=None,
258
                      datetime_search_keys=None,
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
                      full_text_search_keys=None,
                      ignore_empty_string=1, stat__=0):
    """
      Return a dictionnary containing the keys and value types:
        'where_expression': string
        'select_expression_list': string
    """
    raise NotImplementedError

  def getSQLKeyList(self):
    """
      Return a list of keys used by this query and its subqueries.
    """
    raise NotImplementedError
  
  def getRelatedTableMapDict(self):
    """
      Return for each key used by this query (plus ones used by its
      subqueries) the table alias mapping.
    """
    raise NotImplementedError

281
class NegatedQuery(QueryMixin): # XXX Bad name JPS - NotQuery or NegativeQuery is better NegationQuery
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
  """
    Do a boolean negation of given query.
  """

  def __init__(self, query):
    self._query = query

  def asSQLExpression(self, *args, **kw):
    sql_expression_dict = self._query.asSQLExpression(*args, **kw)
    sql_expression_dict['where_expression'] = '(NOT (%s))' % \
      (sql_expression_dict['where_expression'], )
    return sql_expression_dict

  def getSQLKeyList(self, *args, **kw):
    return self._query.getSQLKeyList(*args, **kw)

  def getRelatedTableMapDict(self, *args, **kw):
    return self._query.getRelatedTableMapDict(*args, **kw)

301 302
  # asSearchTextExpression is still not implemented

303 304
allow_class(NegatedQuery)

305 306 307
class Query(QueryMixin):
  """
  This allow to define constraints on a sql column
308

309 310
  format - type date : %d/%m/%Y
           type float : 1 234.12
311
  """
312
  def __init__(self, format=None, operator=None, range=None, key=None,
Sebastien Robin's avatar
Sebastien Robin committed
313
                     search_mode=None, table_alias_list=None, type=None, **kw):
314 315 316 317 318 319
    self.format = format
    if operator is None:
      operator = 'OR'
    self.operator = operator
    self.range = range
    self.search_mode = search_mode
320
    self.table_alias_list = table_alias_list
321
    key_list = kw.keys()
322
    if len(key_list) != 1:
323 324 325
      raise KeyError, 'Query must have only one key'
    self.key = key_list[0]
    self.value = kw[self.key]
Sebastien Robin's avatar
Sebastien Robin committed
326
    self.type = type
327
    self.search_key = key
328

Aurel's avatar
Aurel committed
329 330
  def __call__(self, **kw):
    return self.asSQLExpression(**kw)
331 332 333 334

  def getRange(self):
    return self.range

335 336 337 338 339 340 341 342 343 344
  def getTableAliasList(self):
    return self.table_alias_list

  def getRelatedTableMapDict(self):
    result = {}
    table_alias_list = self.getTableAliasList()
    if table_alias_list is not None:
      result[self.getKey()] = table_alias_list
    return result

345
  def getSearchMode(self):
346 347
    """Search mode used for Full Text search
    """
348 349
    return self.search_mode

350 351 352 353 354 355 356 357 358 359 360
  def asSearchTextExpression(self):
    # This will be the standard way to represent
    # complex values in listbox. Some fixed
    # point must be garanteed
    value = self.value
    if isSimpleType(value) or isinstance(value, DateTime):
      return str(value)
    elif isinstance(value, (list, tuple)):
      value = map(lambda x:str(x), value)
      return (' %s ' % self.operator).join(value)

361 362
  def asSQLExpression(self, key_alias_dict=None,
                            keyword_search_keys=None,
363
                            datetime_search_keys=None,
364
                            full_text_search_keys=None,
365
                            ignore_empty_string=1, stat__=0):
366 367 368 369 370 371
    """
    Build the sql string
    """
    sql_expression = ''
    value = self.getValue()
    key = self.getKey()
372 373
    search_key = self.search_key
    ignore_key = 0
374 375 376 377 378 379 380 381 382 383 384 385
    if key_alias_dict is not None:
      # Try to find the alias
      if key not in key_alias_dict:
        ignore_key=1
      else:
        key = key_alias_dict.get(key)
        if key is None:
          ignore_key=1
    where_expression = []
    select_expression = []
    # Default case: variable equality
    range_value = self.getRange()
Sebastien Robin's avatar
Sebastien Robin committed
386
    format = self.getFormat()
387
    if ignore_key:
Aurel's avatar
Aurel committed
388
      pass    
389 390
    elif range_value is not None:
      if isinstance(value, (list, tuple)):
Sebastien Robin's avatar
Sebastien Robin committed
391 392 393 394 395 396
        if format is None:
          query_min = min(value)
          query_max = max(value)
        else:
          query_min = value[0]
          query_max = value[1]
397 398
      else:
        query_min=query_max=value
Sebastien Robin's avatar
Sebastien Robin committed
399 400
      query_min = self._quoteSQLString(query_min)
      query_max = self._quoteSQLString(query_max)
401
      if range_value == 'min' :
Sebastien Robin's avatar
Sebastien Robin committed
402
        where_expression.append("%s >= %s" % (key, query_min))
403
      elif range_value == 'max' :
Sebastien Robin's avatar
Sebastien Robin committed
404
        where_expression.append("%s < %s" % (key, query_max))
405
      elif range_value == 'minmax' :
Sebastien Robin's avatar
Sebastien Robin committed
406
        where_expression.append("%s >= %s and %s < %s" % (key, query_min, key, query_max))
407
      elif range_value == 'minngt' :
Sebastien Robin's avatar
Sebastien Robin committed
408
        where_expression.append("%s >= %s and %s <= %s" % (key, query_min, key, query_max))
409
      elif range_value == 'ngt' :
Sebastien Robin's avatar
Sebastien Robin committed
410
        where_expression.append("%s <= %s" % (key, query_max))
411 412
      elif range_value == 'nlt' :
        where_expression.append("%s > %s" % (key, query_max))
413
    elif isSimpleType(value) or isinstance(value, DateTime) \
414
        or (isinstance(value, (list, tuple)) and self.operator.upper() != 'IN'):
415 416 417 418
      # Convert into lists any value which contain 'OR'
      # Refer to _listGlobalActions DCWorkflow patch for example of use
      if isinstance(value, basestring) \
                and search_key != EXACT_MATCH_SEARCH_MODE:
419
        value = value.split(' OR ')
420
        value = map(lambda x:x.strip(), value)
421
      value_list = value
422
      if isSimpleType(value) or isinstance(value, DateTime):
423 424 425
        value_list = [value]
      # For security.
      for value in value_list:
Sebastien Robin's avatar
Sebastien Robin committed
426 427 428
        comparison_operator = None
        if (value != '' or not ignore_empty_string) \
                        and isinstance(value, basestring):
429
          if '%' in value and search_key != EXACT_MATCH_SEARCH_MODE:
Sebastien Robin's avatar
Sebastien Robin committed
430
            comparison_operator = 'LIKE'
431 432 433 434 435 436 437 438 439 440 441 442 443 444 445
          elif search_key == DATETIME_SEARCH_MODE  or (
               datetime_search_keys is not None and key in datetime_search_keys):
            if len(value) >= 1 and value[0:2] in ('<=','!=','>='):
              comparison_operator = value[0:2]
              value = value[2:]
            elif len(value) >= 1 and value[0] in ('=','>','<'):
              comparison_operator = value[0]
              value = value[1:]
            if comparison_operator is None:
              comparison_operator = '='
            # this seems like a DateTime bug!
            # 2002/02/01 ==>(UTC) 2002-01-31 22:00:00
            # 2002-02-01 ==>(UTC) 2002-02-01 00:00:00 (!)
            value = value.replace('-', '/') 
            value = DateTime(value).toZone('UTC').ISO()
Sebastien Robin's avatar
Sebastien Robin committed
446 447 448 449 450 451
          elif len(value) >= 1 and value[0:2] in ('<=','!=','>='):
            comparison_operator = value[0:2]
            value = value[2:]
          elif len(value) >= 1 and value[0] in ('=','>','<'):
            comparison_operator = value[0]
            value = value[1:]
452 453 454
          elif search_key == KEYWORD_SEARCH_MODE or (
                   key in keyword_search_keys and
                    search_key != EXACT_MATCH_SEARCH_MODE):
455
            # We must add % in the request to simulate the catalog
Sebastien Robin's avatar
Sebastien Robin committed
456 457
            comparison_operator = 'LIKE'
            value = '%%%s%%' % value
458 459 460
          elif search_key == FULL_TEXT_SEARCH_MODE or (
                  key in full_text_search_keys
                  and search_key != EXACT_MATCH_SEARCH_MODE):
461 462 463 464 465 466 467 468 469 470
            # We must add % in the request to simulate the catalog
            # we first check if there is a special search_mode for this key
            # incl. table name, or for all keys of that name,
            # or there is a search_mode supplied for all fulltext keys
            # or we fall back to natural mode
            search_mode=self.getSearchMode()
            if search_mode is None:
              search_mode = 'natural'
            search_mode=search_mode.lower()
            mode = full_text_search_modes.get(search_mode,'')
Sebastien Robin's avatar
Sebastien Robin committed
471 472
            where_expression.append(
                        "MATCH %s AGAINST ('%s' %s)" % (key, value, mode))
473 474
            if not stat__:
              # we return relevance as Table_Key_relevance
Sebastien Robin's avatar
Sebastien Robin committed
475 476 477
              select_expression.append(
                     "MATCH %s AGAINST ('%s' %s) AS %s_relevance" 
                     % (key, value, mode,key.replace('.','_')))
478 479
              # and for simplicity as Key_relevance
              if '.' in key:
Sebastien Robin's avatar
Sebastien Robin committed
480 481 482
                select_expression.append(
                     "MATCH %s AGAINST ('%s' %s) AS %s_relevance" % 
                     (key, value, mode,key.split('.')[1]))
483
          else:
Sebastien Robin's avatar
Sebastien Robin committed
484
            comparison_operator = '='
485
        elif not isinstance(value, basestring):
Sebastien Robin's avatar
Sebastien Robin committed
486 487 488 489 490 491
          comparison_operator = '='
        if comparison_operator is not None:
          key = self._quoteSQLKey(key)
          value = self._quoteSQLString(value)
          where_expression.append("%s %s %s" % 
                                  (key, comparison_operator, value))
Aurel's avatar
Aurel committed
492

Nicolas Delaby's avatar
Nicolas Delaby committed
493 494
    elif value is None:
      where_expression.append("%s is NULL" % (key))
495
    elif isinstance(value, (tuple, list)) and self.operator.upper() == 'IN':
496
      if len(value) > 1:
497 498 499
        escaped_value_list = [self._quoteSQLString(x) for x in value]
        escaped_value_string = ', '.join(escaped_value_list)
        where_expression.append("%s IN (%s)" % (key, escaped_value_string))
500
      elif len(value) == 1:
501
        where_expression.append("%s = %s" % (key, self._quoteSQLString(value[0])))
502
      else:
Vincent Pelletier's avatar
Vincent Pelletier committed
503
        where_expression.append('0') # "foo IN ()" is invalid SQL syntax, so use a "false" value.
504
    else:
Sebastien Robin's avatar
Sebastien Robin committed
505 506
      where_expression.append("%s = %s" % 
           (self._quoteSQLKey(key), self._quoteSQLString(value)))
507 508 509 510 511 512

    if len(where_expression)>0:
      if len(where_expression)==1:
        where_expression = where_expression[0]
      else:
        where_expression = '(%s)' % (' %s ' % self.getOperator()).join(where_expression)
513
    else:
514
      where_expression = '1' # It is better to have a valid default
515 516
    return {'where_expression':where_expression,
            'select_expression_list':select_expression}
Aurel's avatar
Aurel committed
517

518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536
  def getKey(self):
    return self.key

  def getValue(self):
    return self.value

  def getSQLKeyList(self):
    """
    Returns the list of keys used by this
    instance
    """
    return [self.getKey()]

allow_class(Query)

class ComplexQuery(QueryMixin):
  """
  Used in order to concatenate many queries
  """
537
  def __init__(self, *args, **kw):
538
    self.query_list = args
539 540
    self.operator = kw.pop('operator', 'AND')
    # XXX: What is that used for ?! It's utterly dangerous.
541 542
    self.__dict__.update(kw)

Aurel's avatar
Aurel committed
543 544
  def __call__(self, **kw):
    return self.asSQLExpression(**kw)
545 546 547 548

  def getQueryList(self):
    return self.query_list

549 550 551
  def getRelatedTableMapDict(self):
    result = {}
    for query in self.getQueryList():
552 553
      if not(isinstance(query, basestring)):
        result.update(query.getRelatedTableMapDict())
554 555
    return result

556 557 558
  def asSQLExpression(self, key_alias_dict=None,
                            ignore_empty_string=1,
                            keyword_search_keys=None,
559
                            datetime_search_keys=None,
560 561
                            full_text_search_keys=None,
                            stat__=0):
562 563 564 565 566 567
    """
    Build the sql string
    """
    sql_expression_list = []
    select_expression_list = []
    for query in self.getQueryList():
568 569 570 571 572 573 574 575 576 577
      if isinstance(query, basestring):
        sql_expression_list.append(query)
      else:
        query_result = query.asSQLExpression( key_alias_dict=key_alias_dict,
                               ignore_empty_string=ignore_empty_string,
                               keyword_search_keys=keyword_search_keys,
                               full_text_search_keys=full_text_search_keys,
                               stat__=stat__)
        sql_expression_list.append(query_result['where_expression'])
        select_expression_list.extend(query_result['select_expression_list'])
578 579
    operator = self.getOperator()
    result = {'where_expression':('(%s)' %  \
580
                         (' %s ' % operator).join(['(%s)' % x for x in sql_expression_list])),
581 582 583 584 585 586 587 588 589 590
              'select_expression_list':select_expression_list}
    return result

  def getSQLKeyList(self):
    """
    Returns the list of keys used by this
    instance
    """
    key_list=[]
    for query in self.getQueryList():
591 592
      if not(isinstance(query, basestring)):
        key_list.extend(query.getSQLKeyList())
593 594 595 596
    return key_list

allow_class(ComplexQuery)

597 598 599 600 601
class Catalog(Folder,
              Persistent,
              Acquisition.Implicit,
              ExtensionClass.Base,
              OFS.History.Historical):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
602 603 604 605
  """ An Object Catalog

  An Object Catalog maintains a table of object metadata, and a
  series of manageable indexes to quickly search for objects
606
  (references in the metadata) that satisfy a search where_expression.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
607 608 609 610 611 612 613 614

  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

615 616
  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.
617 618 619
  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).
620 621 622
  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
623

Jean-Paul Smets's avatar
Jean-Paul Smets committed
624

625
  brain defined in methods...
Jean-Paul Smets's avatar
Jean-Paul Smets committed
626 627 628 629 630 631

  TODO:

    - optmization: indexing objects should be deferred
      until timeout value or end of transaction
  """
632
  meta_type = "SQLCatalog"
633
  icon = 'misc_/ZCatalog/ZCatalog.gif' # FIXME: use a different icon
634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
  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'),}
663
    ) + OFS.History.Historical.manage_options
664

665
  __ac_permissions__= (
666 667 668 669 670

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

      'manage_catalogView', 'manage_catalogFind',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
671 672
      'manage_catalogFilter',
      'manage_catalogAdvanced',
673 674

      'manage_catalogReindex', 'manage_catalogFoundItems',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
675 676
      'manage_catalogClear',
      'manage_main',
677 678 679 680 681 682
      'manage_editFilter',
      ],
     ['Manager']),

    ('Search ZCatalog',
     ['searchResults', '__call__', 'uniqueValuesFor',
Yoshinori Okuji's avatar
Yoshinori Okuji committed
683 684
      'all_meta_types', 'valid_roles',
      'getCatalogSearchTableIds',
685
      'getFilterableMethodList',],
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
     ['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' },
711 712 713 714 715
    { 'id'      : 'sql_catalog_reserve_uid',
      'description' : 'A method to reserve a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
716 717 718 719 720
    { 'id'      : 'sql_catalog_delete_uid',
      'description' : 'A method to delete a uid value',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
721 722
    { 'id'      : 'sql_catalog_object_list',
      'description' : 'Methods to be called to catalog the list of objects',
723 724 725 726 727 728 729 730
      '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' },
731 732 733
    { 'id'      : 'sql_catalog_translation_list',
      'description' : 'Methods to be called to catalog the list of translation objects',
      'type'    : 'selection',
734 735
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
736 737 738
    { 'id'      : 'sql_delete_translation_list',
      'description' : 'Methods to be called to delete translations',
      'type'    : 'selection',
739 740
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
741 742
    { 'id'      : 'sql_clear_catalog',
      'description' : 'The methods which should be called to clear the catalog',
743 744 745
      'type'    : 'multiple selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
746
    { 'id'      : 'sql_record_object_list',
747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765
      '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
766 767 768 769 770
    { 'id'      : 'sql_search_security',
      'description' : 'Main method to search security',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
    { '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' },
806 807 808 809 810
    { 'id'      : 'sql_catalog_index',
      'description' : 'Method to get the main catalog index',
      'type'    : 'selection',
      'select_variable' : 'getCatalogMethodIds',
      'mode'    : 'w' },
811 812 813 814 815 816 817 818 819 820 821 822 823 824 825
    { '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' },
    { 'id'      : 'sql_catalog_keyword_search_keys',
      'description' : 'Columns which should be considered as full text search',
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
826 827 828 829 830
    { 'id'      : 'sql_catalog_datetime_search_keys',
      'description' : 'Columns which should be considered as full text search',
      'type'    : 'multiple selection',
      'select_variable' : 'getColumnIds',
      'mode'    : 'w' },
831 832 833 834 835 836 837 838 839 840 841 842 843 844 845
    { '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' },
846 847 848 849 850
    { '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' },
851 852 853 854 855 856 857 858 859
    { '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' },
860 861 862 863 864
    { 'id'      : 'sql_catalog_scriptable_keys',
      'title'   : 'Related keys',
      'description' : 'Virtual columns to generate scriptable scriptable queries',
      'type'    : 'lines',
      'mode'    : 'w' },
865 866
  )

867
  sql_catalog_produce_reserved = ''
868
  sql_catalog_delete_uid = ''
869 870 871 872 873 874 875 876 877 878 879
  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
880
  sql_search_security = ''
881 882 883 884 885 886
  sql_count_results = ''
  sql_getitem_by_path = ''
  sql_getitem_by_uid = ''
  sql_catalog_tables = ''
  sql_search_tables = ()
  sql_catalog_schema = ''
887
  sql_catalog_index = ''
888 889 890
  sql_unique_values = ''
  sql_catalog_paths = ''
  sql_catalog_keyword_search_keys =  ()
891
  sql_catalog_datetime_search_keys = ()
892 893 894 895 896
  sql_catalog_full_text_search_keys = ()
  sql_catalog_request_keys = ()
  sql_search_result_keys = ()
  sql_catalog_topic_search_keys = ()
  sql_catalog_multivalue_keys = ()
897
  sql_catalog_index_on_order_keys = ()
898
  sql_catalog_related_keys = ()
899
  sql_catalog_scriptable_keys = ()
900

901 902 903 904 905 906
  # 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
907

908 909 910 911 912 913 914
  # 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
915

916 917 918 919 920 921 922 923 924 925
  manage_catalogView = DTMLFile('dtml/catalogView',globals())
  manage_catalogFilter = DTMLFile('dtml/catalogFilter',globals())
  manage_catalogFind = DTMLFile('dtml/catalogFind',globals())
  manage_catalogAdvanced = DTMLFile('dtml/catalogAdvanced', globals())

  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
926 927 928
    self.schema = {}  # mapping from attribute name to column
    self.names = {}   # mapping from column to attribute name
    self.indexes = {}   # empty mapping
929
    self.filter_dict = PersistentMapping()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
930

931 932 933 934 935 936
  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')
937 938 939 940 941 942 943 944 945 946 947 948
    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
    property_list.sort(lambda x, y: cmp(x[0], y[0]))
    for property in property_list:
      property_id = property[0]
      value       = property[1]
949
      if isinstance(value, basestring):
950
        f.write('  <property id=%s type="str">%s</property>\n' % (quoteattr(property_id), escape(value)))
951
      elif isinstance(value, (tuple, list)):
952 953 954
        f.write('  <property id=%s type="tuple">\n' % quoteattr(property_id))
        # Sort for easy diff
        item_list = []
955
        for item in value:
956
          if isinstance(item, basestring):
957 958 959 960
            item_list.append(item)
        item_list.sort()
        for item in item_list:
          f.write('    <item type="str">%s</item>\n' % escape(str(item)))
961
        f.write('  </property>\n')
962 963 964
    # 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'):
965 966
      filter_list = []
      for filter_id in self.filter_dict.keys():
967
        filter_definition = self.filter_dict[filter_id]
968 969 970 971 972 973 974
        filter_list.append((filter_id, filter_definition))
      # Sort for easy diff
      filter_list.sort(lambda x, y: cmp(x[0], y[0]))
      for filter_item in filter_list:
        filter_id  = filter_item[0]
        filter_def = filter_item[1]
        if not filter_def['filtered']:
975 976
          # If a filter is not activated, no need to output it.
          continue
977
        if not filter_def['expression']:
978 979
          # If the expression is not specified, meaningless to specify it.
          continue
980
        f.write('  <filter id=%s expression=%s />\n' % (quoteattr(filter_id), quoteattr(filter_def['expression'])))
981
        # For now, portal types are not exported, because portal types are too specific to each site.
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002
    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.
    """
    f = open(file)
    try:
      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):
1003
            raise CatalogError, 'unknown property id %r' % (id,)
1004
          if type not in ('str', 'tuple'):
1005
            raise CatalogError, 'unknown property type %r' % (type,)
1006 1007 1008 1009
          if type == 'str':
            value = ''
            for text in prop.childNodes:
              if text.nodeType == text.TEXT_NODE:
1010
                value = str(text.data)
1011 1012 1013 1014 1015 1016
                break
          else:
            value = []
            for item in prop.getElementsByTagName("item"):
              item_type = item.getAttribute("type")
              if item_type != 'str':
1017
                raise CatalogError, 'unknown item type %r' % (item_type,)
1018 1019
              for text in item.childNodes:
                if text.nodeType == text.TEXT_NODE:
1020
                  value.append(str(text.data))
1021 1022 1023 1024
                  break
            value = tuple(value)

          setattr(self, id, value)
1025

1026 1027 1028
        if not hasattr(self, 'filter_dict'):
          self.filter_dict = PersistentMapping()
        for filt in root.getElementsByTagName("filter"):
1029
          id = str(filt.getAttribute("id"))
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
          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
1042 1043 1044 1045
      finally:
        doc.unlink()
    finally:
      f.close()
Aurel's avatar
Aurel committed
1046

1047 1048 1049 1050 1051 1052 1053
  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__)))
1054 1055 1056

  def _clearSecurityCache(self):
    self.security_uid_dict = OIBTree()
1057
    self.security_uid_index = None
1058 1059

  security.declarePrivate('getSecurityUid')
1060
  def getSecurityUid(self, wrapped_object):
1061 1062 1063 1064 1065 1066 1067
    """
      Cache a uid for each security permission

      We try to create a unique security (to reduce number of lines)
      and to assign security only to root document
    """
    # Get security information
1068
    allowed_roles_and_users = wrapped_object.allowedRolesAndUsers()
1069 1070 1071 1072
    # Sort it
    allowed_roles_and_users = list(allowed_roles_and_users)
    allowed_roles_and_users.sort()
    allowed_roles_and_users = tuple(allowed_roles_and_users)
1073 1074
    # Make sure no duplicates
    if getattr(aq_base(self), 'security_uid_dict', None) is None:
1075 1076 1077
      self._clearSecurityCache()
    if self.security_uid_dict.has_key(allowed_roles_and_users):
      return (self.security_uid_dict[allowed_roles_and_users], None)
1078 1079 1080 1081 1082 1083 1084 1085
    # 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
    id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
    if id_tool is not None:
      default = 1
      # We must keep compatibility with existing sites
1086 1087
      previous_security_uid = getattr(self, 'security_uid_index', None)
      if previous_security_uid is not None:
1088 1089 1090 1091
        # At some point, it was a Length
        if isinstance(previous_security_uid, Length):
          default = previous_security_uid() + 1
        else:
1092
          default = previous_security_uid
1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
      security_uid = id_tool.generateNewLengthId(id_group='security_uid_index',
                                        default=default)
    else:
      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
1104 1105
    self.security_uid_dict[allowed_roles_and_users] = security_uid
    return (security_uid, allowed_roles_and_users)
1106

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1107 1108 1109 1110 1111 1112
  def clear(self):
    """
    Clears the catalog by calling a list of methods
    """
    methods = self.sql_clear_catalog
    for method_name in methods:
1113
      method = getattr(self, method_name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1114 1115
      try:
        method()
1116 1117
      except ConflictError:
        raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1118
      except:
1119
        LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
1120
            'could not clear catalog with %s' % method_name, error=sys.exc_info())
1121

1122
    # Reserved uids have been removed.
1123
    self.clearReserved()
1124

1125
    # Add a dummy item so that SQLCatalog will not use existing uids again.
1126
    self.insertMaxUid()
1127

1128
    # Remove the cache of catalog schema.
1129 1130
    if hasattr(self, '_v_catalog_schema_dict') :
      del self._v_catalog_schema_dict
1131

1132
    self._clearSecurityCache()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1133

1134 1135 1136 1137 1138 1139 1140 1141 1142 1143
  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()])

1144 1145 1146 1147 1148 1149
  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
1150 1151 1152 1153 1154
    try:
      method()
    except ConflictError:
      raise
    except:
1155
      LOG('SQLCatalog', WARNING,
Romain Courteaud's avatar
Romain Courteaud committed
1156 1157 1158
          'could not clear reserved catalog with %s' % \
              method_id, error=sys.exc_info())
      raise
1159
    self._last_clear_reserved_time += 1
1160

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1161 1162 1163 1164 1165 1166 1167 1168 1169
  def __getitem__(self, uid):
    """
    Get an object by UID
    Note: brain is defined in Z SQL Method object
    """
    method = getattr(self,  self.sql_getitem_by_uid)
    search_result = method(uid = uid)
    if len(search_result) > 0:
      return search_result[0]
1170
    raise KeyError, uid
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190

  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

1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
  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

1208
  security.declarePublic('getCatalogSearchResultKeys')
1209 1210 1211 1212
  def getCatalogSearchResultKeys(self):
    """Return search result keys.
    """
    return self.sql_search_result_keys
1213

1214
  def _getCatalogSchema(self, table=None):
1215 1216 1217
    # XXX: Using a volatile as a cache makes it impossible to flush
    # consistently on all connections containing the volatile. Another
    # caching scheme must be used here.
1218
    catalog_schema_dict = getattr(aq_base(self), '_v_catalog_schema_dict', {})
1219

1220 1221 1222 1223 1224 1225 1226 1227 1228
    if table not in catalog_schema_dict:
      result_list = []
      try:
        method_name = self.sql_catalog_schema
        method = getattr(self, method_name)
        #LOG('_getCatalogSchema', 0, 'method_name = %r, method = %r, table = %r' % (method_name, method, table))
        search_result = method(table=table)
        for c in search_result:
          result_list.append(c.Field)
1229 1230
      except ConflictError:
        raise
1231
      except:
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1232
        LOG('SQLCatalog', WARNING, '_getCatalogSchema failed with the method %s' % method_name, error=sys.exc_info())
1233 1234 1235
        pass
      catalog_schema_dict[table] = tuple(result_list)
      self._v_catalog_schema_dict= catalog_schema_dict
1236

1237
    return catalog_schema_dict[table]
1238

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1239 1240
  def getColumnIds(self):
    """
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1241 1242 1243
    Calls the show column method and returns dictionnary of
    Field Ids
    """
1244
    def _getColumnIds():
1245 1246 1247 1248 1249 1250
      keys = {}
      for table in self.getCatalogSearchTableIds():
        field_list = self._getCatalogSchema(table=table)
        for field in field_list:
          keys[field] = 1
          keys['%s.%s' % (table, field)] = 1  # Is this inconsistent ?
1251
      for related in self.getSQLCatalogRelatedKeyList():
1252 1253 1254
        related_tuple = related.split('|')
        related_key = related_tuple[0].strip()
        keys[related_key] = 1
1255 1256 1257 1258
      for scriptable in self.getSQLCatalogScriptableKeyList():
        scriptable_tuple = scriptable.split('|')
        scriptable = scriptable_tuple[0].strip()
        keys[scriptable] = 1
1259 1260 1261
      keys = keys.keys()
      keys.sort()
      return keys
Aurel's avatar
Aurel committed
1262
    return CachingMethod(_getColumnIds, id='SQLCatalog.getColumnIds', cache_factory='erp5_content_long')()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1263

1264 1265 1266 1267 1268
  def getColumnMap(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
1269
    def _getColumnMap():
1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
      keys = {}
      for table in self.getCatalogSearchTableIds():
        field_list = self._getCatalogSchema(table=table)
        for field in field_list:
          key = field
          if not keys.has_key(key): keys[key] = []
          keys[key].append(table)
          key = '%s.%s' % (table, key)
          if not keys.has_key(key): keys[key] = []
          keys[key].append(table) # Is this inconsistent ?
      return keys
Aurel's avatar
Aurel committed
1281
    return CachingMethod(_getColumnMap, id='SQLCatalog.getColumnMap', cache_factory='erp5_content_long')()
1282

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1283 1284 1285 1286 1287 1288 1289
  def getResultColumnIds(self):
    """
    Calls the show column method and returns dictionnary of
    Field Ids
    """
    keys = {}
    for table in self.getCatalogSearchTableIds():
1290 1291 1292
      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
1293 1294 1295 1296
    keys = keys.keys()
    keys.sort()
    return keys

1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310
  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
1311 1312 1313
  def getTableIds(self):
    """
    Calls the show table method and returns dictionnary of
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1314 1315 1316
    Field Ids
    """
    keys = []
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1317 1318
    method_name = self.sql_catalog_tables
    try:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1319 1320 1321
      method = getattr(self,  method_name)
      search_result = method()
      for c in search_result:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1322
        keys.append(c[0])
1323 1324
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1325 1326
    except:
      pass
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1327 1328
    return keys

1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342
  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]
  
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1343
  # the cataloging API
1344 1345 1346
  def produceUid(self):
    """
      Produces reserved uids in advance
1347
    """
1348 1349 1350 1351
    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.
1352 1353 1354 1355
    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:
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377
      id_tool = getattr(self.getPortalObject(), 'portal_ids', None)
      if id_tool is not None:
        if self._max_uid is None:
          self._max_uid = Length()
        uid_list = id_tool.generateNewLengthIdList(id_group='catalog_uid',
                     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
        instance_id = md5.new(str(random_factor_list)).hexdigest()
        uid_list = [x.uid for x in method(count = UID_BUFFER_SIZE, instance_id = instance_id) if x.uid != 0]
1378
      uid_buffer.extend(uid_list)
1379

1380 1381 1382 1383 1384 1385
  def isIndexable(self):
    """
    This is required to check in many methods that
    the site root and zope root are indexable
    """
    zope_root = self.getZopeRoot()
1386
    site_root = self.getSiteRoot() # XXX-JPS - Why don't we use getPortalObject here ?
1387 1388 1389 1390 1391 1392

    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
1393

1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
  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

1414 1415 1416
  def newUid(self):
    """
      This is where uid generation takes place. We should consider a multi-threaded environment
1417 1418
      with multiple ZEO clients on a single ZEO server.

1419
      The main risk is the following:
1420

1421
      - objects a/b/c/d/e/f are created (a is parent of b which is parent of ... of f)
1422

1423
      - one reindexing node N1 starts reindexing f
1424

1425
      - another reindexing node N2 starts reindexing e
1426

1427 1428 1429
      - 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

1430
      Similar problems may happen with relations and acquisition of uid values (ex. order_uid)
1431
      with the risk of graph loops
1432
    """
1433
    if not self.isIndexable():
1434 1435
      return None

1436 1437 1438 1439
    klass = self.__class__
    try:
      klass._reserved_uid_lock.acquire()
      self.produceUid()
1440 1441 1442
      uid_buffer = self.getUIDBuffer()
      if len(uid_buffer) > 0:
        uid = uid_buffer.pop()
1443 1444 1445 1446 1447 1448
        # Vincent added this 2006/01/25
        #if uid > 4294967296: # 2**32
        #if uid > 10000000: # arbitrary level : below it's normal, above it's suspicious
        #   LOG('SQLCatalog', WARNING, 'Newly generated UID (%s) seems too big ! - vincent' % (uid,))
        #   raise RuntimeError, 'Newly generated UID (%s) seems too big ! - vincent' % (uid,)
        # end
1449 1450 1451 1452
        if self._max_uid is None:
          self._max_uid = Length()
        if uid > self._max_uid():
          self._max_uid.set(uid)
1453
        return long(uid)
1454 1455 1456 1457
      else:
        raise CatalogError("Could not retrieve new uid")
    finally:
      klass._reserved_uid_lock.release()
1458

1459 1460 1461
  def manage_catalogObject(self, REQUEST, RESPONSE, URL1, urls=None):
    """ index Zope object(s) that 'urls' point to """
    if urls:
1462
      if isinstance(urls, str):
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477
        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:
1478
      if isinstance(urls, str):
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
        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')

  def manage_catalogReindex(self, REQUEST, RESPONSE, URL1):
    """ 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`)))

1502
  def manage_catalogClear(self, REQUEST=None, RESPONSE=None,
Romain Courteaud's avatar
Romain Courteaud committed
1503
                          URL1=None, sql_catalog_id=None):
1504
    """ clears the whole enchilada """
1505 1506
    self.beforeCatalogClear()

1507 1508 1509
    self.clear()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1510
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1511
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1512 1513

  def manage_catalogClearReserved(self, REQUEST=None, RESPONSE=None, URL1=None):
1514
    """ clears reserved uids """
1515 1516 1517
    self.clearReserved()

    if RESPONSE and URL1:
Romain Courteaud's avatar
Romain Courteaud committed
1518
      RESPONSE.redirect('%s/manage_catalogAdvanced?' \
1519
                        'manage_tabs_message=Catalog%%20Cleared' % URL1)
1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556

  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
1557
  def catalogObject(self, object, path, is_object_moved=0):
1558 1559
    """Add an object to the Catalog by calling all SQL methods and
    providing needed arguments.
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1560

1561 1562
    'object' is the object to be catalogged."""
    self._catalogObjectList([object])
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1563

1564 1565 1566 1567
  def catalogObjectList(self, object_list, method_id_list=None, 
                        disable_cache=0, check_uid=1, idxs=None):
    """Add objects to the Catalog by calling all SQL methods and
    providing needed arguments.
1568

1569 1570
      method_id_list : specify which methods should be used. If not
                       set, it will take the default value of portal_catalog.
1571 1572

      disable_cache : do not use cache, so values will be computed each time,
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584
                      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.
    
    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."""
    # XXX 300 is arbitrary.
1585 1586
    for i in xrange(0, len(object_list), OBJECT_LIST_SIZE):
      self._catalogObjectList(object_list[i:i + OBJECT_LIST_SIZE],
1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597
                              method_id_list=method_id_list,
                              disable_cache=disable_cache,
                              check_uid=check_uid,
                              idxs=idxs)
    
  def _catalogObjectList(self, object_list, method_id_list=None, 
                         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?"""
1598
    LOG('SQLCatalog', TRACE, 'catalogging %d objects' % len(object_list))
1599
    #LOG('catalogObjectList', 0, 'called with %r' % (object_list,))
1600

1601
    if idxs not in (None, []):
1602 1603
      LOG('ZSLQCatalog.SQLCatalog:catalogObjectList', WARNING, 
          'idxs is ignored in this function and is only provided to be compatible with CMFCatalogAware.reindexObject.')
1604

1605 1606
    if not self.isIndexable():
      return None
1607

1608 1609 1610
    portal_catalog = self.getSiteRoot().portal_catalog # XXX-JPS - This is a hardcoded name. Weird
                                                       # Isn't self == self.getSiteRoot().portal_catalog
                                                       # in this case ?
1611

1612 1613
    # Reminder about optimization: It might be possible to issue just one
    # query to get enought results to check uid & path consistency.
1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
    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)

1632
    for object in object_list:
1633
      if not getattr(aq_base(object), 'uid', None):
1634
        try:
1635
          object.uid = self.newUid()
1636 1637
        except ConflictError:
          raise
1638
        except:
1639
          raise RuntimeError, 'could not set missing uid for %r' % (object,)
1640
      elif check_uid:
1641 1642
        uid = object.uid
        path = object.getPath()
1643
        index = path_uid_dict.get(path, None)
1644
        try:
1645
          index = long(index)
1646
        except TypeError:
1647
          index = None
1648 1649 1650
        if index is not None and index < 0:
          raise CatalogError, 'A negative uid %d is used for %s. Your catalog is broken. Recreate your catalog.' % (index, path)
        if index:
1651 1652
          if uid != index or isinstance(uid, int):
            # We want to make sure that uid becomes long if it is an int
1653
            LOG('SQLCatalog', WARNING, 'uid of %r changed from %r (property) to %r (catalog, by path) !!! This can be fatal. You should reindex the whole site immediately.' % (object, uid, index))
1654 1655 1656 1657 1658 1659
            uid = index
            object.uid = uid
        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
1660
          if uid in uid_path_dict:
1661 1662 1663
            catalog_path = uid_path_dict.get(uid)
          else:
            catalog_path = self.getPathForUid(uid)
1664 1665 1666 1667 1668 1669
          #LOG('catalogObject', 0, 'uid = %r, catalog_path = %r' % (uid, catalog_path))
          if catalog_path == "reserved":
            # Reserved line in catalog table
            klass = self.__class__
            try:
              klass._reserved_uid_lock.acquire()
1670 1671
              uid_buffer = self.getUIDBuffer()
              if uid_buffer is not None:
1672 1673 1674 1675 1676 1677 1678 1679
                # 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.
1680
                try:
1681
                  uid_buffer.remove(uid)
1682 1683
                except ValueError:
                  pass
1684 1685 1686 1687
            finally:
              klass._reserved_uid_lock.release()
          elif catalog_path is not None:
            # An uid conflict happened... Why?
1688
            # can be due to path length
1689
            if len(path) > MAX_PATH_LEN:
1690 1691
              LOG('SQLCatalog', WARNING, 'path of object %r is too long for catalog. You should use a shorter path.' %(object,))

1692
            object.uid = self.newUid()
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1693
            LOG('SQLCatalog', WARNING,
1694
                'uid of %r changed from %r to %r as old one is assigned to %s in catalog !!! This can be fatal. You should reindex the whole site immediately.' % (object, uid, object.uid, catalog_path))
1695

1696 1697
    if method_id_list is None:
      method_id_list = self.sql_catalog_object_list
1698
    econtext_cache = {}
1699
    expression_result_cache = {}
1700 1701
    argument_cache = {}

1702
    try:
1703
      if not disable_cache:
1704
        enableReadOnlyTransactionCache(self)
1705

1706
      method_kw_dict = {}
1707
      for method_name in method_id_list:
1708
        kw = {}
1709
        if self.isMethodFiltered(method_name):
1710 1711
          catalogged_object_list = []
          type_list = self.filter_dict[method_name]['type']
1712
          type_dict = dict(zip(type_list, type_list)) or None
1713
          expression = self.filter_dict[method_name]['expression_instance']
1714
          expression_cache_key_list = self.filter_dict[method_name].get('expression_cache_key', '').split()
1715 1716 1717 1718
          for object in object_list:
            # We will check if there is an filter on this
            # method, if so we may not call this zsqlMethod
            # for this object
1719
            if type_dict is not None and object.getPortalType() not in type_dict:
1720 1721
              continue
            elif expression is not None:
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748
              if expression_cache_key_list:
                # We try to save results of expressions by portal_type
                # or by anyother key which can prevent us from evaluating
                # expressions. This cache is built each time we reindex
                # objects but we could also use over multiple transactions
                # if this can improve performance significantly
                try:
                  cache_key = map(lambda key: object.getProperty(key, None), expression_cache_key_list)
                    # ZZZ - we could find a way to compute this once only
                  cache_key = (method_name, tuple(cache_key))
                  result = expression_result_cache[cache_key]
                  compute_result = 0
                except KeyError:
                  cache_result = 1
                  compute_result = 1
              else:
                cache_result = 0
                compute_result = 1
              if compute_result:
                try:
                  econtext = econtext_cache[object.uid]
                except KeyError:
                  econtext = self.getExpressionContext(object)
                  econtext_cache[object.uid] = econtext
                result = expression(econtext)
              if cache_result:
                expression_result_cache[cache_key] = result
1749 1750 1751 1752 1753
              if not result:
                continue
            catalogged_object_list.append(object)
        else:
          catalogged_object_list = object_list
1754

1755 1756
        if len(catalogged_object_list) == 0:
          continue
1757

1758
        method_kw_dict[method_name] = kw
1759

1760 1761
        #LOG('catalogObjectList', 0, 'method_name = %s' % (method_name,))
        method = getattr(self, method_name)
1762
        if method.meta_type in ("Z SQL Method", "LDIF Method"):
1763
          # Build the dictionnary of values
1764
          arguments = split(method.arguments_src)
1765 1766 1767
        elif method.meta_type == "Script (Python)":
          arguments = \
            method.func_code.co_varnames[:method.func_code.co_argcount]
1768 1769 1770 1771 1772 1773 1774 1775 1776
        else:
          arguments = []
        for arg in arguments:
          value_list = []
          append = value_list.append
          for object in catalogged_object_list:
            try:
              value = argument_cache[(object.uid, arg)]
            except KeyError:
1777
              try:
1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788
                value = getattr(object, arg, None)
                if callable(value):
                  value = value()
              except ConflictError:
                raise
              except:
                value = None
              if not disable_cache:
                argument_cache[(object.uid, arg)] = value
            append(value)
          kw[arg] = value_list
1789

1790
      for method_name in method_kw_dict.keys():
1791 1792
        kw = method_kw_dict[method_name]
        method = getattr(self, method_name)
Jérome Perrin's avatar
Jérome Perrin committed
1793
        method = aq_base(method).__of__(portal_catalog) # Use method in
1794 1795 1796 1797
                # the context of portal_catalog
        # Alter/Create row
        try:
          #start_time = DateTime()
1798
          #LOG('catalogObjectList', DEBUG, 'kw = %r, method_name = %r' % (kw, method_name))
1799 1800 1801 1802 1803 1804 1805
          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]))
1806

1807 1808 1809
        except ConflictError:
          raise
        except:
Yoshinori Okuji's avatar
Yoshinori Okuji committed
1810
          LOG('SQLCatalog', WARNING, 'could not catalog objects %s with method %s' % (object_list, method_name),
1811 1812 1813
              error=sys.exc_info())
          raise
    finally:
1814
      if not disable_cache:
1815
        disableReadOnlyTransactionCache(self)
1816

1817 1818
  if psyco is not None:
    psyco.bind(_catalogObjectList)
1819

1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831
  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
1832 1833 1834 1835
    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
1836
      return self.uncatalogObject(path=path,uid=uid)
1837 1838 1839
    method = getattr(self, method_name)
    method(uid = uid)

1840
  def uncatalogObject(self, path=None, uid=None):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
    """
    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

    """
1853
    if not self.isIndexable():
1854 1855
      return None

1856 1857
    if uid is None and path is not None:
      uid = self.getUidForPath(path)
1858 1859
    methods = self.sql_uncatalog_object
    if uid is None:
1860
      return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1861
    for method_name in methods:
1862 1863
      # 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
1864
      method = getattr(self, method_name)
1865
      method(uid = uid)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1866

1867 1868 1869 1870
  def catalogTranslationList(self, object_list):
    """Catalog translations.
    """
    method_name = self.sql_catalog_translation_list
1871 1872
    return self.catalogObjectList(object_list, method_id_list = (method_name,),
                                  check_uid=0)
1873

1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
  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())
1885

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1886 1887 1888
  def uniqueValuesFor(self, name):
    """ return unique values for FieldIndex name """
    method = getattr(self, self.sql_unique_values)
1889
    return method(column=name)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1890 1891 1892 1893 1894 1895 1896 1897

  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 """
1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
    #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
1908 1909
      return None

1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948
  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
1949
          uid_path_dict[uid] = search_result[0].path
1950 1951
    return uid_path_dict

Jean-Paul Smets's avatar
Jean-Paul Smets committed
1952 1953 1954 1955 1956 1957 1958 1959 1960
  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
1961 1962 1963 1964
      try:
        int(uid)
      except ValueError:
        return None
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1965 1966 1967 1968 1969 1970 1971 1972
      # 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
1973 1974
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1975 1976 1977
    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
1978
      LOG('SQLCatalog', WARNING, "could not find path from uid %s" % (uid,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
      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
2007 2008
    except ConflictError:
      raise
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2009 2010 2011
    except:
      # This is a real LOG message
      # which is required in order to be able to import .zexp files
2012 2013
      LOG('SQLCatalog', WARNING,
          "could not find metadata from path %s" % (path,))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2014 2015 2016 2017 2018 2019
      return None

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

2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
  def getCatalogMethodIds(self):
    """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
2032 2033
            if not isinstance(id, str):
              id=id()
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
            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

2045
  def getSQLCatalogRelatedKeyList(self, key_list=None):
2046 2047
    """
    Return the list of related keys.
2048
    This method can be overidden in order to implement
2049 2050
    dynamic generation of some related keys.
    """
2051 2052
    if key_list is None:
      key_list = []
2053
    # Do not generate dynamic related key for acceptable_keys
2054
    dynamic_key_list = [k for k in key_list \
2055 2056
        if k not in self.getColumnMap().keys()]

2057
    dynamic_list = self.getDynamicRelatedKeyList(dynamic_key_list)
2058 2059 2060
    full_list = list(dynamic_list) + list(self.sql_catalog_related_keys)
    return full_list

2061 2062 2063
  # Compatibililty SQL Sql
  getSqlCatalogRelatedKeyList = getSQLCatalogRelatedKeyList

2064 2065 2066 2067 2068
  def getSQLCatalogScriptableKeyList(self):
    """
    Return the list of scriptable keys.
    """
    return self.sql_catalog_scriptable_keys
2069

2070 2071
  def getTableIndex(self, table):
    """
2072
    Return a map between index and column for a given table
2073 2074 2075 2076 2077 2078 2079 2080
    """
    def _getTableIndex(table):
      table_index = {}
      method = getattr(self, self.sql_catalog_index, '')
      if method in ('', None):
        return {}
      index = list(method(table=table))
      for line in index:
2081 2082
        if table_index.has_key(line.KEY_NAME):
          table_index[line.KEY_NAME].append(line.COLUMN_NAME)
2083
        else:
2084 2085
          table_index[line.KEY_NAME] = [line.COLUMN_NAME,]
      LOG("SQLCatalog.getTableIndex", INFO, "index = %s for table = %s" \
2086 2087 2088
          %(table_index, table))
      return table_index
    return CachingMethod(_getTableIndex, id='SQLCatalog.getTableIndex', \
Aurel's avatar
Aurel committed
2089
                         cache_factory='erp5_content_long')(table=table)
2090 2091


2092
  def getIndex(self, table, column_list, all_column_list):
2093 2094 2095
    """
    Return possible index for a column list in a given table
    """
2096
    def _getIndex(table, column_list, all_column_list):
2097 2098 2099
      index_dict = self.getTableIndex(table)
      if isinstance(column_list, str):
        column_list = [column_list,]
2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
      # Get possible that can be used
      possible_index = []
      for index in index_dict.keys():
        index_columns = index_dict[index]
        for column in index_columns:
          if column in column_list:
            if index not in possible_index:
              possible_index.append(index)
      if len(possible_index) == 0:
        return []
      # Get the most suitable index
      for index in possible_index:
        # Make sure all column in index are used by the query
        index_column = index_dict[index]
        for column in index_column:
          if column in column_list or column in all_column_list:
            continue
          else:
            possible_index.remove(index)
      LOG("SQLCatalog.getIndex", INFO, "index = %s for table %s and columns %s" \
          %(possible_index, table, column_list))
      return possible_index
Aurel's avatar
Aurel committed
2122
    return CachingMethod(_getIndex, id='SQLCatalog.getIndex', cache_factory='erp5_content_long')\
2123 2124
          (table=table, column_list=column_list, all_column_list=all_column_list)

2125

2126
  def buildSQLQuery(self, query_table='catalog', REQUEST=None,
2127
                          ignore_empty_string=1, query=None, stat__=0, **kw):
2128
    """ Builds a complex SQL query to simulate ZCatalog behaviour """
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2129 2130 2131 2132 2133 2134 2135
    # Get search arguments:
    if REQUEST is None and (kw is None or kw == {}):
      # We try to get the REQUEST parameter
      # since we have nothing handy
      try: REQUEST=self.REQUEST
      except AttributeError: pass

2136
    #LOG('SQLCatalog.buildSQLQuery, kw',0,kw)
2137 2138
    # If kw and query are not set, then use REQUEST instead
    if query is None and (kw is None or kw == {}):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2139 2140
      kw = REQUEST

2141
    acceptable_key_map = self.getColumnMap()
2142 2143
    full_text_search_keys = list(self.sql_catalog_full_text_search_keys)
    keyword_search_keys = list(self.sql_catalog_keyword_search_keys)
2144
    datetime_search_keys = list(self.sql_catalog_datetime_search_keys)
2145
    topic_search_keys = self.sql_catalog_topic_search_keys
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2146
    multivalue_keys = self.sql_catalog_multivalue_keys
Aurel's avatar
Aurel committed
2147 2148


2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
    # Compute "sort_index", which is a sort index, or none:
    if kw.has_key('sort-on'):
      sort_index=kw['sort-on']
    elif hasattr(self, 'sort-on'):
      sort_index=getattr(self, 'sort-on')
    elif kw.has_key('sort_on'):
      sort_index=kw['sort_on']
    else: sort_index=None

    # Compute the sort order
    if kw.has_key('sort-order'):
      so=kw['sort-order']
    elif hasattr(self, 'sort-order'):
      so=getattr(self, 'sort-order')
    elif kw.has_key('sort_order'):
      so=kw['sort_order']
    else: so=None

    # We must now turn sort_index into
    # a dict with keys as sort keys and values as sort order
    if isinstance(sort_index, basestring):
      sort_index = [(sort_index, so)]
    elif not isinstance(sort_index, (list, tuple)):
      sort_index = None
2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192

    # Rebuild keywords to behave as new style query (_usage='toto:titi' becomes {'toto':'titi'})
    new_kw = {}
    usage_len = len('_usage')
    for k, v in kw.items():
      if k.endswith('_usage'):
        new_k = k[0:-usage_len]
        if not new_kw.has_key(new_k):
          new_kw[new_k] = {}
        if not isinstance(new_kw[new_k], dict):
          new_kw[new_k] = {'query': new_kw[new_k]}
        split_v = v.split(':')
        new_kw[new_k] = {split_v[0]: split_v[1]}
      else:
        if not new_kw.has_key(k):
          new_kw[k] = v
        else:
          new_kw[k]['query'] = v
    kw = new_kw

2193 2194 2195 2196 2197 2198 2199 2200
    # Initialise Scriptable Dict
    scriptable_key_dict = {}
    for t in self.sql_catalog_scriptable_keys:
      t = t.split('|')
      key = t[0].strip()
      method_id = t[1].strip()
      scriptable_key_dict[key] = method_id

2201 2202 2203 2204
    # Build the list of Queries and ComplexQueries
    query_dict = {}
    key_list = [] # the list of column keys
    key_alias_dict = {}
2205
    query_group_by_list = None # Useful to keep a default group_by passed by scriptable keys
2206 2207 2208
    query_related_table_map_dict = {}
    if query is not None:
      kw ['query'] = query
2209
    for key in kw.keys():
2210
      if key not in RESERVED_KEY_LIST:
2211 2212 2213
        value = kw[key]
        current_query = None
        new_query_dict = {}
2214 2215 2216 2217 2218
        if isinstance(value, (Query, ComplexQuery)):
          current_query = value
        elif scriptable_key_dict.has_key(key):
          # Turn this key into a query by invoking a script
          method = getattr(self, scriptable_key_dict[key])
2219 2220
          current_query = method(value) # May return None
          if hasattr(current_query, 'order_by'): query_group_by_list = current_query.order_by
2221
        else:
2222
          if isinstance(value, dict):
2223
            for value_key in value.keys():
2224 2225
              if value_key == 'query':
                new_query_dict[key] = value['query']
2226
              else:
2227
                new_query_dict[value_key] = value[value_key]
2228
          else:
2229
            new_query_dict[key] = value
2230
          current_query = Query(**new_query_dict)
2231 2232 2233
        if current_query is not None:
          query_dict[key] = current_query
          key_list.extend(current_query.getSQLKeyList())
2234
          query_related_table_map_dict.update(current_query.getRelatedTableMapDict())
2235

2236 2237
    # if we have a sort index, we must take it into account to get related
    # keys.
2238
    sort_key_dict = dict()
2239 2240
    if sort_index:
      for sort_info in sort_index:
2241 2242 2243
        sort_key = sort_info[0]
        if sort_key not in key_list:
          key_list.append(sort_key)
2244
        sort_key_dict[sort_key] = 1
2245

2246
    related_tuples = self.getSQLCatalogRelatedKeyList(key_list=key_list)
Aurel's avatar
Aurel committed
2247

2248
    # Define related maps
2249 2250
    # each tuple from `related_tuples` has the form (key,
    # 'table1,table2,table3/column/where_expression')
2251
    related_keys = {}
2252 2253 2254
    related_method = {}
    related_table_map = {}
    related_column = {}
2255
    related_table_list = {}
2256
    table_rename_index = 0
2257
    related_methods = {} # related methods which need to be used
2258 2259 2260
    for t in related_tuples:
      t_tuple = t.split('|')
      key = t_tuple[0].strip()
2261
      if key in key_list:
2262
        if ignore_empty_string \
2263 2264 2265 2266
            and kw.get(key, None) in ('', [], ())\
            and key not in sort_key_dict:
              # We don't ignore 0 and None, but if the key is used for sorting,
              # we should not discard this key
2267
          continue
2268
        join_tuple = t_tuple[1].strip().split('/')
2269
        related_keys[key] = None
2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285
        method_id = join_tuple[2]
        table_list = tuple(join_tuple[0].split(','))
        related_method[key] = method_id
        related_table_list[key] = table_list
        related_column[key] = join_tuple[1]
        # Check if some aliases where specified in queries
        map_list = query_related_table_map_dict.get(key,None)
        # Rename tables to prevent conflicts
        if not related_table_map.has_key((table_list,method_id)):
          if map_list is None:
            map_list = []
            for table_id in table_list:
              map_list.append((table_id,
                 "related_%s_%s" % (table_id, table_rename_index))) # We add an index in order to alias tables in the join
              table_rename_index += 1 # and prevent name conflicts
          related_table_map[(table_list,method_id)] = map_list
2286

Jean-Paul Smets's avatar
Jean-Paul Smets committed
2287 2288
    # We take additional parameters from the REQUEST
    # and give priority to the REQUEST
2289
    if REQUEST is not None:
2290
      for key in acceptable_key_map.iterkeys():
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2291 2292
        if REQUEST.has_key(key):
          # Only copy a few keys from the REQUEST
2293
          if key in self.sql_catalog_request_keys:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2294
            kw[key] = REQUEST[key]
2295 2296

    def getNewKeyAndUpdateVariables(key):
2297
      key_is_acceptable = key in acceptable_key_map # Only calculate once
2298 2299 2300 2301 2302 2303 2304 2305 2306 2307
      key_is_related = key in related_keys
      new_key = None
      if key_is_acceptable or key_is_related:
        if key_is_related: # relation system has priority (ex. security_uid)
          # We must rename the key
          method_id = related_method[key]
          table_list = related_table_list[key]
          if not related_methods.has_key((table_list,method_id)):
            related_methods[(table_list,method_id)] = 1
          # Prepend renamed table name
Aurel's avatar
Aurel committed
2308
          new_key = "%s.%s" % (related_table_map[(table_list,method_id)][-1][-1],
2309 2310 2311 2312 2313 2314 2315 2316
                               related_column[key])
        elif key_is_acceptable:
          if key.find('.') < 0:
            # if the key is only used by one table, just append its name
            if len(acceptable_key_map[key]) == 1 :
              new_key = '%s.%s' % (acceptable_key_map[key][0], key)
            # query_table specifies what table name should be used by default
            elif query_table and \
2317
                '%s.%s' % (query_table, key) in acceptable_key_map:
2318 2319 2320
              new_key = '%s.%s' % (query_table, key)
            elif key == 'uid':
              # uid is always ambiguous so we can only change it here
2321
              new_key = 'catalog.uid'
2322 2323 2324 2325
          else:
            new_key = key
          if new_key is not None:
            # Add table to table dict, we use catalog by default
Aurel's avatar
Aurel committed
2326
            from_table_dict[acceptable_key_map[new_key][0]] = acceptable_key_map[new_key][0]
2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
      key_alias_dict[key] = new_key
      return new_key

    where_expression_list = []
    select_expression_list = []
    group_by_expression_list = []
    where_expression = ''
    select_expression = ''
    group_by_expression = ''

2337
    from_table_dict = {'catalog' : 'catalog'} # Always include catalog table
2338
    if len(kw):
2339
      if kw.has_key('select_expression'):
Jérome Perrin's avatar
Jérome Perrin committed
2340
        select_expression_list.append(kw['select_expression'])
2341
      if kw.has_key('group_by_expression'):
2342
        group_by_expression_list.append(kw['group_by_expression'])
2343
      # Grouping
2344
      group_by_list = kw.get('group_by', query_group_by_list)
2345 2346 2347 2348
      if type(group_by_list) is type('a'): group_by_list = [group_by_list]
      if group_by_list is not None:
        try:
          for key in group_by_list:
2349 2350
            new_key = getNewKeyAndUpdateVariables(key)
            group_by_expression_list.append(new_key)
2351 2352 2353 2354
        except ConflictError:
          raise
        except:
          LOG('SQLCatalog', WARNING, 'buildSQLQuery could not build the new group by expression', error=sys.exc_info())
2355 2356 2357 2358
          group_by_expression = ''
      if len(group_by_expression_list)>0:
        group_by_expression = ','.join(group_by_expression_list)
        group_by_expression = str(group_by_expression)
2359

2360
    sort_on = None
2361
    sort_key_list = []
2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373
    if sort_index is not None:
      new_sort_index = []
      for sort in sort_index:
        if len(sort) == 2:
          # Try to analyse expressions of the form "title AS unsigned"
          sort_key_list = sort[0].split()
          if len(sort_key_list) == 3:
            sort_key = sort_key_list[0]
            sort_type = sort_key_list[2]
          elif len(sort_key_list):
            sort_key = sort_key_list[0]
            sort_type = None
2374
          else:
2375 2376 2377 2378 2379 2380 2381 2382
            sort_key = sort[0]
            sort_type = None
          new_sort_index.append((sort_key, sort[1], sort_type))
        elif len(sort) == 3:
          new_sort_index.append(sort)
      sort_index = new_sort_index
      try:
        new_sort_index = []
Jérome Perrin's avatar
Jérome Perrin committed
2383
        for (original_key, so, as_type) in sort_index:
2384
          key = getNewKeyAndUpdateVariables(original_key)
2385
          if key is not None:
2386
            sort_key_list.append(key)
2387 2388 2389 2390 2391 2392
            if as_type == 'int':
              key = 'CAST(%s AS SIGNED)' % key
            elif as_type:
              key = 'CAST(%s AS %s)' % (key, as_type) # Different casts are possible
            if so in ('descending', 'reverse', 'DESC'):
              new_sort_index.append('%s DESC' % key)
2393
            else:
2394 2395
              new_sort_index.append('%s' % key)
          else:
2396 2397
            LOG('SQLCatalog', WARNING, 'buildSQLQuery could not build sort '
                'index (%s -> %s)' % (original_key, key))
2398 2399 2400 2401 2402 2403 2404
        sort_index = join(new_sort_index,',')
        sort_on = str(sort_index)
      except ConflictError:
        raise
      except:
        LOG('SQLCatalog', WARNING, 'buildSQLQuery could not build the new sort index', error=sys.exc_info())
        sort_on = ''
2405
        sort_key_list = []
2406 2407 2408 2409 2410

    for key in key_list:
      if not key_alias_dict.has_key(key):
        getNewKeyAndUpdateVariables(key)
    if len(query_dict):
2411
      for key, query in query_dict.items():
2412 2413 2414
        query_result = query.asSQLExpression(key_alias_dict=key_alias_dict,
                                    full_text_search_keys=full_text_search_keys,
                                    keyword_search_keys=keyword_search_keys,
2415
                                    datetime_search_keys=datetime_search_keys,
2416 2417 2418 2419 2420
                                    ignore_empty_string=ignore_empty_string,
                                    stat__=stat__)
        if query_result['where_expression'] not in ('',None):
          where_expression_list.append(query_result['where_expression'])
        select_expression_list.extend(query_result['select_expression_list'])
Aurel's avatar
Aurel committed
2421

2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438
    # Calculate extra where_expression based on required joins
    for k, tid in from_table_dict.items():
      if k != query_table:
        where_expression_list.append('%s.uid = %s.uid' % (query_table, tid))
    # Calculate extra where_expressions based on related definition
    for (table_list, method_id) in related_methods.keys():
      related_method = getattr(self, method_id, None)
      if related_method is not None:
        table_id = {'src__' : 1} # Return query source, do not evaluate
        table_id['query_table'] = query_table
        table_index = 0
        for t_tuple in related_table_map[(table_list,method_id)]:
          table_id['table_%s' % table_index] = t_tuple[1] # table_X is set to mapped id
          from_table_dict[t_tuple[1]] = t_tuple[0]
          table_index += 1
        where_expression_list.append(related_method(**table_id))
    # Concatenate expressions
2439 2440
    if kw.get('where_expression',None) not in (None,''):
      where_expression_list.append(kw['where_expression'])
2441 2442
    if len(where_expression_list)>1:
      where_expression_list = ['(%s)' % x for x in where_expression_list]
2443 2444 2445 2446 2447 2448 2449 2450
    where_expression = join(where_expression_list, ' AND ')
    select_expression= join(select_expression_list,',')

    limit_expression = kw.get('limit', None)
    if isinstance(limit_expression, (list, tuple)):
      limit_expression = '%s,%s' % (limit_expression[0], limit_expression[1])
    elif limit_expression is not None:
      limit_expression = str(limit_expression)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2451

2452
    # force index if exists when doing sort as mysql doesn't manage them efficiently
2453
    if len(sort_key_list) > 0:
2454 2455
      index_from_table = {}
      # first group columns from a same table
2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466
      for key in sort_key_list:
        try:
          related_table, column = key.split('.')
        except ValueError:
          # key is not of the form table.column
          # so get table from dict
          if len(from_table_dict) != 1:
            continue
          column = key
          related_table = from_table_dict.keys()[0]

2467
        table = from_table_dict[related_table]
2468 2469 2470 2471 2472 2473
        # Check if it's a column for which we want to specify index
        index_columns = getattr(self, 'sql_catalog_index_on_order_keys', [])
        sort_column = '%s.%s' %(table, column)
        if not sort_column in index_columns:
          continue
        # Group columns
2474 2475 2476 2477 2478 2479
        if not index_from_table.has_key(table):
          index_from_table[table] = [column,]
        else:
          index_from_table[table].append(column)
      # second ask index
      for table in index_from_table.keys():
2480
        available_index_list = self.getIndex(table, index_from_table[table], key_list)
2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
        if len(available_index_list) > 0:
          # tell mysql to use these index
          table = from_table_dict.pop(related_table)
          index_list_string = ""
          for index in available_index_list:
            if len(index_list_string) == 0:
              index_list_string += "%s" %index
            else:
              index_list_string += ", %s" %index
          table_with_index =  "%s use index(%s)"  %(related_table, index_list_string)
          from_table_dict[table_with_index] = table

2493
    # Use a dictionary at the moment.
2494
    return { 'from_table_list' : from_table_dict.items(),
2495
             'order_by_expression' : sort_on,
Yoshinori Okuji's avatar
Yoshinori Okuji committed
2496
             'where_expression' : where_expression,
2497
             'limit_expression' : limit_expression,
2498 2499
             'select_expression': select_expression,
             'group_by_expression' : group_by_expression}
2500

2501 2502 2503
  # Compatibililty SQL Sql
  buildSqlQuery = buildSQLQuery

2504
  def queryResults(self, sql_method, REQUEST=None, used=None, src__=0, build_sql_query_method=None, **kw):
2505
    """ Returns a list of brains from a set of constraints on variables """
2506 2507 2508
    if build_sql_query_method is None:
      build_sql_query_method = self.buildSQLQuery
    query = build_sql_query_method(REQUEST=REQUEST, **kw)
2509 2510 2511
    kw['where_expression'] = query['where_expression']
    kw['sort_on'] = query['order_by_expression']
    kw['from_table_list'] = query['from_table_list']
Yoshinori Okuji's avatar
Yoshinori Okuji committed
2512
    kw['limit_expression'] = query['limit_expression']
2513
    kw['select_expression'] = query['select_expression']
2514
    kw['group_by_expression'] = query['group_by_expression']
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2515
    # Return the result
2516

2517 2518 2519
    #LOG('acceptable_keys',0,'acceptable_keys: %s' % str(acceptable_keys))
    #LOG('acceptable_key_map',0,'acceptable_key_map: %s' % str(acceptable_key_map))
    #LOG('queryResults',0,'kw: %s' % str(kw))
2520
    #LOG('queryResults',0,'from_table_list: %s' % str(query['from_table_list']))
2521
    return sql_method(src__=src__, **kw)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2522

2523
  def searchResults(self, REQUEST=None, used=None, **kw):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2524
    """ Returns a list of brains from a set of constraints on variables """
2525
    # The used argument is deprecated and is ignored
2526
    method = getattr(self, self.sql_search_results)
2527
    return self.queryResults(method, REQUEST=REQUEST, used=used, **kw)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2528 2529 2530

  __call__ = searchResults

2531
  def countResults(self, REQUEST=None, used=None, stat__=1, **kw):
2532
    """ Returns the number of items which satisfy the where_expression """
2533 2534
    # Get the search method
    method = getattr(self, self.sql_count_results)
2535
    return self.queryResults(method, REQUEST=REQUEST, used=used, stat__=stat__, **kw)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
2536

2537
  def recordObjectList(self, path_list, catalog=1):
2538
    """
2539
      Record the path of an object being catalogged or uncatalogged.
2540
    """
2541 2542
    method = getattr(self, self.sql_record_object_list)
    method(path_list=path_list, catalog=catalog)
2543

2544
  def deleteRecordedObjectList(self, uid_list=()):
2545 2546 2547 2548
    """
      Delete all objects which contain any path.
    """
    method = getattr(self, self.sql_delete_recorded_object_list)
2549
    method(uid_list=uid_list)
2550

2551
  def readRecordedObjectList(self, catalog=1):
2552 2553 2554 2555
    """
      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)
2556
    return method(catalog=catalog)
2557 2558 2559 2560 2561 2562 2563 2564 2565

  # 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:
2566
      method_id_list = [zsql_method.id for zsql_method in self.getFilterableMethodList()]
2567

2568 2569 2570 2571
      # Remove unused filters.
      for id in self.filter_dict.keys():
        if id not in method_id_list:
          del self.filter_dict[id]
2572

2573
      for id in method_id_list:
2574 2575 2576
        # We will first look if the filter is activated
        if not self.filter_dict.has_key(id):
          self.filter_dict[id] = PersistentMapping()
2577 2578 2579 2580 2581
          self.filter_dict[id]['filtered'] = 0
          self.filter_dict[id]['type'] = []
          self.filter_dict[id]['expression'] = ""
          self.filter_dict[id]['expression_cache_key'] = "portal_type"

2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
        if REQUEST.has_key('%s_box' % id):
          self.filter_dict[id]['filtered'] = 1
        else:
          self.filter_dict[id]['filtered'] = 0

        if REQUEST.has_key('%s_expression' % id):
          expression = REQUEST['%s_expression' % id]
          if expression == "":
            self.filter_dict[id]['expression'] = ""
            self.filter_dict[id]['expression_instance'] = None
          else:
            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

        if REQUEST.has_key('%s_type' % id):
          list_type = REQUEST['%s_type' % id]
2602
          if isinstance(list_type, str):
2603 2604 2605 2606 2607
            list_type = [list_type]
          self.filter_dict[id]['type'] = list_type
        else:
          self.filter_dict[id]['type'] = []

2608 2609 2610 2611 2612 2613 2614 2615 2616
        if REQUEST.has_key('%s_expression_cache_key' % id):
          expression_cache_key = REQUEST['%s_expression_cache_key' % id]
          if expression_cache_key == "":
            self.filter_dict[id]['expression_cache_key'] = expression_cache_key
          else:
            self.filter_dict[id]['expression_cache_key'] = ""
        else:
          self.filter_dict[id]['expression_cache_key'] = ""

2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
    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
2627
      if getattr(aq_base(self), 'filter_dict', None) is None:
2628 2629
        self.filter_dict = PersistentMapping()
        return 0
2630
      try:
2631
        return self.filter_dict[method_name]['filtered']
2632 2633
      except KeyError:
        return 0
2634 2635 2636
    return 0

  def getExpression(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2637
    """ Get the filter expression text for this method.
2638 2639
    """
    if withCMF:
2640
      if getattr(aq_base(self), 'filter_dict', None) is None:
2641 2642
        self.filter_dict = PersistentMapping()
        return ""
2643
      try:
2644
        return self.filter_dict[method_name]['expression']
2645 2646
      except KeyError:
        return ""
2647 2648
    return ""

2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662
  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:
        return self.filter_dict[method_name]['expression_cache_key']
      except KeyError:
        return ""
    return ""

2663
  def getExpressionInstance(self, method_name):
Jérome Perrin's avatar
Jérome Perrin committed
2664
    """ Get the filter expression instance for this method.
2665 2666
    """
    if withCMF:
2667
      if getattr(aq_base(self), 'filter_dict', None) is None:
2668 2669
        self.filter_dict = PersistentMapping()
        return None
2670
      try:
2671
        return self.filter_dict[method_name]['expression_instance']
2672 2673
      except KeyError:
        return None
2674 2675
    return None

Jérome Perrin's avatar
Jérome Perrin committed
2676 2677
  def isPortalTypeSelected(self, method_name, portal_type):
    """ Returns true if the portal type is selected for this method.
2678 2679
    """
    if withCMF:
2680
      if getattr(aq_base(self), 'filter_dict', None) is None:
2681 2682
        self.filter_dict = PersistentMapping()
        return 0
2683 2684 2685 2686
      try:
        return portal_type in (self.filter_dict[method_name]['type'])
      except KeyError:
        return 0
2687 2688
    return 0

2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
  def getFilteredPortalTypeList(self, method_name):
    """ Returns the list of portal types which define
        the filter.
    """
    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 []
2702 2703 2704 2705 2706 2707 2708

  def getFilterableMethodList(self):
    """
    Returns only zsql methods wich catalog or uncatalog objets
    """
    method_dict = {}
    if withCMF:
2709 2710 2711 2712
      methods = getattr(self,'sql_catalog_object',()) + \
                getattr(self,'sql_uncatalog_object',()) + \
                getattr(self,'sql_update_object',()) + \
                getattr(self,'sql_catalog_object_list',())
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726
      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.
      '''
      if withCMF:
        data = {
            'here':         ob,
            'container':    aq_parent(aq_inner(ob)),
            'nothing':      None,
2727 2728 2729 2730 2731 2732 2733
            #'root':         ob.getPhysicalRoot(),
            #'request':      getattr( ob, 'REQUEST', None ),
            #'modules':      SecureModuleImporter,
            #'user':         getSecurityManager().getUser(),
            'isDelivery':   ob.isDelivery, # XXX
            'isMovement':   ob.isMovement, # XXX
            'isPredicate':  ob.isPredicate, # XXX
2734
            'isDocument':   ob.isDocument, # XXX
2735 2736
            'isInventory':  ob.isInventory, # XXX
            'isInventoryMovement': ob.isInventoryMovement, # XXX
2737 2738 2739 2740 2741 2742
            }
        return getEngine().getContext(data)


Globals.default__class_init__(Catalog)

Jean-Paul Smets's avatar
Jean-Paul Smets committed
2743
class CatalogError(Exception): pass