ContributionTool.py 27.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
##############################################################################
#
# Copyright (c) 2007 Nexedi SARL and Contributors. All Rights Reserved.
#                    Jean-Paul Smets <jp@nexedi.com>
#
# WARNING: This program as such is intended to be used by professional
# programmers who take the whole responsability of assessing all potential
# consequences resulting from its eventual inadequacies and bugs
# End users who are looking for a ready-to-use solution with commercial
# garantees and support are strongly adviced to contract a Free Software
# Service Company
#
# This program is Free Software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License
# as published by the Free Software Foundation; either version 2
# of the License, or (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
#
##############################################################################

29
import cStringIO
30 31
import re
import string
32 33
import socket
import md5
Jean-Paul Smets's avatar
Jean-Paul Smets committed
34
import urllib2, urllib
35
import inspect
36

Bartek Górny's avatar
Bartek Górny committed
37
from AccessControl import ClassSecurityInfo, getSecurityManager
38
from Globals import InitializeClass, DTMLFile
39
from Products.CMFCore.utils import getToolByName, _checkPermission
40 41 42
from Products.ERP5Type.Tool.BaseTool import BaseTool
from Products.ERP5Type import Permissions
from Products.ERP5 import _dtmldir
Jean-Paul Smets's avatar
Jean-Paul Smets committed
43 44
from Products.ERP5.Document.Url import no_crawl_protocol_list, no_host_protocol_list

45 46 47
from zLOG import LOG
from DateTime import DateTime
from Acquisition import aq_base
48
from zExceptions import BadRequest
49

50 51 52 53 54 55 56
# Install openers
import ContributionOpener
opener = urllib2.build_opener(ContributionOpener.DirectoryFileHandler)
urllib2.install_opener(opener)

# A temporary hack until urllib2 supports timeout setting - XXX
import socket
57
socket.setdefaulttimeout(600) # 1 minute timeout
58 59

# Global parameters
60
TEMP_NEW_OBJECT_KEY = '_v_new_object'
61
MAX_REPEAT = 10
62 63

_marker = []  # Create a new marker object.
64 65 66 67

class ContributionTool(BaseTool):
  """
    ContributionTool provides an abstraction layer to unify the contribution
68
    of documents into an ERP5 Site.
69

70 71
    ContributionTool needs to be configured in portal_types (allowed contents) so
    that it can store Text, Spreadsheet, PDF, etc. 
72

73 74 75
    The main method of ContributionTool is newContent. This method can
    be provided various parameters from which the portal type and document
    metadata can be derived. 
76 77

    Configuration Scripts:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
78

79 80 81
      - ContributionTool_getPropertyDictFromFileName: receives file name and a 
        dict derived from filename by regular expression, and does any necesary
        operations (e.g. mapping document type id onto a real portal_type).
Jean-Paul Smets's avatar
Jean-Paul Smets committed
82 83 84 85 86 87

    Problems which are not solved

      - handling of relative links in HTML contents (or others...)
        some text rewriting is necessary.

88 89 90 91 92 93
  """
  title = 'Contribution Tool'
  id = 'portal_contributions'
  meta_type = 'ERP5 Contribution Tool'
  portal_type = 'Contribution Tool'

Jean-Paul Smets's avatar
Jean-Paul Smets committed
94 95 96
  # Regular expressions
  simple_normaliser = re.compile('#.*')

97 98 99 100 101 102 103
  # Declarative Security
  security = ClassSecurityInfo()

  security.declareProtected(Permissions.ManagePortal, 'manage_overview' )
  manage_overview = DTMLFile( 'explainContributionTool', _dtmldir )

  security.declareProtected(Permissions.AddPortalContent, 'newContent')
104
  def newContent(self, id=None, portal_type=None, url=None, container=None,
Jean-Paul Smets's avatar
Jean-Paul Smets committed
105
                       container_path=None,
106
                       discover_metadata=1, temp_object=0,
107
                       user_login=None, data=None, file_name=None, **kw):
108 109 110 111 112 113 114
    """
      The newContent method is overriden to implement smart content
      creation by detecting the portal type based on whatever information
      was provided and finding out the most appropriate module to store
      the content.

      user_login is the name under which the content will be created
115 116
      XXX - this is a security hole which needs to be fixed by
      making sure only Manager can use this parameter
117

118 119 120 121
      container -- if specified, it is possible to define
      where to contribute the content. Else, ContributionTool
      tries to guess.

Jean-Paul Smets's avatar
Jean-Paul Smets committed
122 123 124
      container_path -- if specified, defines the container path
      and has precedence over container

125 126
      url -- if specified, content is download from the URL.

127 128 129 130
      NOTE:
        We always generate ID. So, we must prevent using the one
        which we were provided.
    """
131 132 133 134
    if file_name is not None: kw['file_name'] = file_name
    if data is not None: kw['data'] = data # This is only used to make sure
                                           # we can pass file as parameter to ZPublisher
                                           # whenever we ingest email
135 136 137 138 139
    # Temp objects use the standard newContent from Folder
    if temp_object:
      # For temp_object creation, use the standard method
      return BaseTool.newContent(self, id=id, portal_type=portal_type, temp_object=1, **kw)

140 141
    document = None

142
    # Try to find the file_name
143
    file_name = None
144
    mime_type = None
145
    if not url:
146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
      # check if file was provided
      file = kw.get('file', None)
      if file is not None:
        file_name = file.filename
      else:
        # some channels supply data and file-name separately
        # this is the case for example for email ingestion
        # in this case, we build a file wrapper for it
        data = kw.get('data', None)
        if data is not None:
          file_name = kw.get('file_name', None)
          if file_name is not None:
            file = cStringIO.StringIO()
            file.write(data)
            file.seek(0)
161 162 163
            kw['file'] = file
            del kw['data']
            del kw['file_name']
Jean-Paul Smets's avatar
Jean-Paul Smets committed
164
    else:
165
      # build a new file from the url
166 167
      url_file = urllib2.urlopen(url)
      data = url_file.read() # time out must be set or ... too long XXX
168 169 170
      file = cStringIO.StringIO()
      file.write(data)
      file.seek(0)
171
      # Create a file name based on the URL and quote it
Jean-Paul Smets's avatar
Jean-Paul Smets committed
172
      file_name = url.split('/')[-1] or url.split('/')[-2]
173 174 175
      file_name = urllib.quote(file_name, safe='')
      file_name = file_name.replace('%', '')
      # For URLs, we want an id by default equal to the encoded URL 
176
      if id is None: id = self.encodeURL(url)
177 178
      if hasattr(url_file, 'headers'):
        headers = url_file.headers
179 180 181
        if hasattr(headers, 'type'):
          mime_type = headers.type
      kw['file'] = file
182 183

    # If the portal_type was provided, we can go faster
184 185 186
    if portal_type and container is None:
      # We know the portal_type, let us find the default module
      # and use it as container
187 188 189 190
      try:
        container = self.getDefaultModule(portal_type)
      except ValueError:
        container = None
191

192 193
    if portal_type and container is not None:
      # We could simplify things here and return a document immediately
194
      # NOTE: we use the module ID generator rather than the provided ID
195 196 197 198 199
      #document = module.newContent(portal_type=portal_type, **kw)
      #if discover_metadata:
      #  document.activate().discoverMetadata(file_name=file_name, user_login=user_login)
      #return document
      pass # XXX - This needs to be implemented once the rest is stable
200

201
    # From here, there is no hope unless a file was provided
202 203 204 205
    if file is None:
      raise ValueError, "could not determine portal type"

    # So we will simulate WebDAV to get an empty object
Jean-Paul Smets's avatar
Jean-Paul Smets committed
206
    # with PUT_factory - we provide the mime_type as
207
    # parameter
208
    # LOG('new content', 0, "%s -- %s" % (file_name, mime_type))
209

210 211 212

    #
    # Check if same file is already exists. if it exists, then update it.
213
    #
214
    if portal_type is None:
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233
      registry = getToolByName(self, 'portal_contribution_registry', None)
      if registry is not None:
        portal_type = registry.findPortalTypeName(file_name, None, None)
        property_dict = self.getMatchedFileNamePatternDict(file_name)
        reference = property_dict.get('reference', None)
        version  = property_dict.get('version', None)
        language  = property_dict.get('language', None)
        if portal_type and reference and version and language:
          portal_catalog = getToolByName(self, 'portal_catalog')
          document = portal_catalog.getResultValue(portal_type=portal_type,
                                                   reference=reference,
                                                   version=version,
                                                   language=language)
          if document is not None:
            # document is already uploaded. So overrides file.
            if not _checkPermission(Permissions.ModifyPortalContent, document):
              raise Unauthorized, "[DMS] You are not allowed to update the existing document which has the same coordinates (id %s)" % document.getId()
            document.edit(file=kw['file'])
            return document
234 235 236 237

    #
    # Strong possibility of a new file.
    #
238 239 240 241 242 243 244 245
    try:
      self._checkId(file_name)
    except BadRequest:
      extension = ''
      if '.' in file_name:
        extension = '.%s' % file_name.split('.')[-1]
      file_name = '%s%s' % (self.generateNewId(), extension)

246
    ob = self.PUT_factory(file_name, mime_type, data)
247

Jean-Paul Smets's avatar
Jean-Paul Smets committed
248 249 250 251
    # Raise an error if we could not guess the portal type
    if ob is None:
      raise ValueError, "Could not determine the document type"

252 253
    object_id = ob.getId()

254 255 256
    # Prevent any reindexing operations
    ob.isIndexable = 0

257
    # Then put the file inside ourselves for a short while
258 259
    BaseTool._setObject(self, object_id, ob)
    document = BaseTool._getOb(self, object_id)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
260

261 262 263 264 265 266
    rewrite_method = document._getTypeBasedMethod('rewriteIngestionData')
    if rewrite_method is not None:
      modified_kw = rewrite_method(**kw.copy())
      if modified_kw is not None:
        kw.update(modified_kw)

267 268
    try:
      # Then edit the document contents (so that upload can happen)
269 270 271 272 273
      if 'set_filename__' in inspect.getargspec(document._edit)[0]:
        # Only a few classes supports set_filename__.
        document._edit(set_filename__=0, **kw)
      else:
        document._edit(**kw)
274 275
      if url:
        document.fromURL(url)
276 277
    finally:
      # Remove the object from ourselves
278
      BaseTool._delObject(self, object_id)
279

Jean-Paul Smets's avatar
Jean-Paul Smets committed
280
    # Move the document to where it belongs
Jean-Paul Smets's avatar
Jean-Paul Smets committed
281 282
    if container_path is not None:
      container = self.getPortalObject().restrictedTraverse(container_path)
283
    document = self._setObject(object_id, ob, user_login=user_login, id=id,
Jean-Paul Smets's avatar
Jean-Paul Smets committed
284 285
                               container=container, discover_metadata=discover_metadata,
                               )
286
    document = self._getOb(object_id) # Call _getOb to purge cache
Jean-Paul Smets's avatar
Jean-Paul Smets committed
287

Jean-Paul Smets's avatar
Jean-Paul Smets committed
288
    # Notify workflows
289
    #document.notifyWorkflowCreated()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
290

291 292
    # Allow reindexing, reindex it and return the document
    delattr(document, 'isIndexable')
293
    document.reindexObject()
294 295
    return document

296
  security.declareProtected( Permissions.AddPortalContent, 'newXML' )
297 298 299 300 301 302 303
  def newXML(self, xml):
    """
      Create a new content based on XML data. This is intended for contributing
      to ERP5 from another application.
    """
    pass

304 305
  security.declareProtected(Permissions.ModifyPortalContent,'getMatchedFileNamePatternDict')
  def getMatchedFileNamePatternDict(self, file_name):
306
    """
307
      Get matched group dict of file name parsing regular expression.
308
    """
309
    property_dict = {}
310 311 312 313 314 315

    if file_name is None:
      return property_dict

    regex_text = self.portal_preferences.getPreferredDocumentFileNameRegularExpression()
    if regex_text in ('', None):
316 317
      return property_dict

318 319 320
    if regex_text:
      pattern = re.compile(regex_text)
      if pattern is not None:
321
        try:
322
          property_dict = pattern.match(file_name).groupdict()
323 324
        except AttributeError: # no match
          pass
325 326 327 328 329 330 331 332 333 334 335 336
    return property_dict

  security.declareProtected(Permissions.ModifyPortalContent,'getPropertyDictFromFileName')
  def getPropertyDictFromFileName(self, file_name):
    """
      Gets properties from filename. File name is parsed with a regular expression
      set in preferences. The regexp should contain named groups.
    """
    if file_name is None:
      return {}
    property_dict = self.getMatchedFileNamePatternDict(file_name)
    method = self._getTypeBasedMethod('getPropertyDictFromFileName',
337
        fallback_script_id = 'ContributionTool_getPropertyDictFromFileName')
338
    property_dict = method(file_name, property_dict)
Ivan Tyagov's avatar
Ivan Tyagov committed
339
    if property_dict.get('portal_type', None) is not None:
340
      # we have to return portal_type as a tuple
341
      # because we should allow for having multiple candidate types
342 343
      property_dict['portal_type'] = (property_dict['portal_type'],)
    else:
344
      # we have to find candidates by file extenstion
Ivan Tyagov's avatar
Ivan Tyagov committed
345 346 347
      if file_name.rfind('.')!= -1:
        ext = file_name.split('.')[-1]
        property_dict['portal_type'] = self.ContributionTool_getCandidateTypeListByExtension(ext)
348 349
    return property_dict

350
  # WebDAV virtual folder support
Jean-Paul Smets's avatar
Jean-Paul Smets committed
351 352
  def _setObject(self, name, ob, user_login=None, container=None,
                       id=None, discover_metadata=1):
353 354 355 356 357 358 359 360 361 362 363
    """
      The strategy is to let NullResource.PUT do everything as
      usual and at the last minute put the object in a different
      location with a different portal type. This means that
      NullResource.PUT creates an empty document with PUT_factory
      then upload document data by invoking PUT on the empty
      document and finally sets the object. By overriding _setObject
      we get a chance to fix the portal_type of the document
      (as long as the one we find is compatible) and move the
      document to the appropriate module.

364 365
      portal_contribution_registry will find appropriate portal type
      name by file_name and content itself.
366 367 368 369 370 371

      The ContributionTool instance must be configured in such
      way that _verifyObjectPaste will return TRUE.

      Refer to: NullResource.PUT
    """
372 373 374 375 376 377 378 379 380 381 382 383 384
    # _setObject is called by constructInstance at a time
    # when the object has no portal_type defined yet. It
    # will be removed later on. We can safely store the
    # document inside us at this stage. Else we
    # must find out where to store it.
    if not ob.__dict__.has_key('portal_type'):
      BaseTool._setObject(self, name, ob)
      document = self[name]
    else:
      # We give the system a last chance to analyse the
      # portal_type based on the document content
      # (ex. a Memo is a kind of Text which can be identified
      # by the fact it includes some specific content)
385

386 387
      # Now we know the portal_type, let us find the module
      # to which we should move the document to
388 389 390 391
      if container is None:
        module = self.getDefaultModule(ob.portal_type)
      else:
        module = container
Jean-Paul Smets's avatar
Jean-Paul Smets committed
392 393 394 395
      if id is None:
        new_id = module.generateNewId()
      else:
        new_id = id
396
      ob.id = new_id
397 398 399 400 401 402 403 404 405 406 407 408
      existing_document = module.get(new_id, None)
      if existing_document is None:
        # There is no preexisting document - we can therefore
        # set the new object
        module._setObject(new_id, ob)
        # We can now discover metadata
        document = module[new_id]
        if discover_metadata:
          # Metadata disovery is done as an activity by default
          # If we need to discoverMetadata synchronously, it must
          # be for user interface and should thus be handled by
          # ZODB scripts
409 410 411 412 413 414 415 416

          # XXX converting state is for only document which is necessary to
          # convert base format.
          portal_workflow = self.getPortalObject().portal_workflow
          if 'processing_status_workflow' in portal_workflow.getChainFor(document):
            document.processFile() # move to converting state.
          document.activate().Document_convertToBaseFormatAndDiscoverMetadata(
            file_name=name, user_login=user_login)
417 418
      else:
        if document.isExternalDocument():
419
          document = existing_document
420
          # If this is an external document, update its content
421
          # document.activate().updateContentFromURL() # XXX I think this is no longer useful with alarms
Jean-Paul Smets's avatar
Jean-Paul Smets committed
422 423 424 425
          # XXX - Make sure this does not increase ZODB
          # XXX - what to do also with parameters (put again edit_kw) ?
          # Providing some information to the use about the fact
          # this was an existing document would also be great
426
        else:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
427 428
          # We may have to implement additional revision support
          # to support in place contribution (ie. for a given ID)
429
          # but is this really useful ?
430
          raise NotImplementedError
431

432 433
      # Keep the document close to us - this is only useful for
      # file upload from webdav
434 435 436
      if not hasattr(self, '_v_document_cache'):
        self._v_document_cache = {}
      self._v_document_cache[name] = document.getRelativeUrl()
437 438 439

    # Return document to newContent method
    return document
440

441 442 443 444 445
  def _getOb(self, id, default=_marker):
    """
    Check for volatile temp object info first
    and try to find it
    """
446 447
    # Use the document cache if possible and return result immediately
    # this is only useful for webdav
Jean-Paul Smets's avatar
Jean-Paul Smets committed
448 449 450
    if hasattr(self, '_v_document_cache'):
      document_url = self._v_document_cache.get(id, None)
      if document_url is not None:
451
        del self._v_document_cache[id]
Jean-Paul Smets's avatar
Jean-Paul Smets committed
452 453
        return self.getPortalObject().unrestrictedTraverse(document_url)

454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470
    # Try first to return the real object inside
    # This is much safer than trying to access objects displayed by listDAVObjects
    # because the behaviour of catalog is unpredicatble if a string is passed
    # for a UID. For example 
    #   select path from catalog where uid = "001193.html";
    # will return the same as
    #   select path from catalog where uid = 1193;
    # This was the source of an error in which the contribution tool
    # was creating a web page and was returning a Base Category
    # when
    #   o = folder._getOb(id)
    # was called in DocumentConstructor
    result = BaseTool._getOb(self, id, default=default)
    if result is not _marker:
      return result

    # Return an object listed by listDAVObjects
471 472 473
    uid = str(id).split('-')[-1]
    object = self.getPortalObject().portal_catalog.unrestrictedGetResultValue(uid=uid)
    if object is not None:
474
      return object.getObject() # Make sure this does not break security. XXX
475

476 477 478
    # Raise an AttributeError the same way as in OFS.ObjectManager._getOb
    raise AttributeError, id

479

Bartek Górny's avatar
Bartek Górny committed
480
  def listDAVObjects(self):
481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504
    """
      Get all contents contributed by the current user. This is
      delegated to a script in order to help customisation.
    """
    method = getattr(self, 'ContributionTool_getMyContentList', None)
    if method is not None:
      object_list = method()
    else:
      sm = getSecurityManager()
      user = sm.getUser()
      object_list = self.portal_catalog(portal_type=self.getPortalMyDocumentTypeList(),
                                        owner=str(user))

    def wrapper(o_list):
      for o in o_list:
        o = o.getObject()
        reference = o.getReference()
        if reference:
          id = '%s-%s' % (reference, o.getUid())
        else:
          id = '%s' % o.getUid()
        yield o.getObject().asContext(id=id)

    return wrapper(object_list)
Bartek Górny's avatar
Bartek Górny committed
505

Jean-Paul Smets's avatar
Jean-Paul Smets committed
506
  # Crawling methods
507 508
  security.declareProtected(Permissions.View, 'normaliseURL')
  def normaliseURL(self, url, base_url=None):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
    """
      Returns a normalised version of the url so
      that we do not download twice the same content.
      URL normalisation is an important part in crawlers.
      The current implementation is obviously simplistic.
      Refer to http://en.wikipedia.org/wiki/Web_crawler
      and study Harvestman for more ideas.
    """
    url = self.simple_normaliser.sub('', url)
    url_split = url.split(':')
    url_protocol = url_split[0]
    if url_protocol in no_host_protocol_list:
      return url
    if base_url and len(url_split) == 1:
      # Make relative URL absolute
      url = '%s/%s' % (base_url, url)
    return url

527 528
  security.declareProtected(Permissions.View, 'encodeURL')
  def encodeURL(self, url):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
529 530 531 532 533 534 535 536 537 538
    """
    Returns the URL as an ID. ID should be chosen in such
    way that it is optimal with HBTreeFolder (ie. so that
    distribution of access time on a cluster is possible)

    NOTE: alternate approach is based on a url table
    and catalog lookup. It is faster ? Not sure. Since
    we must anyway insert objects in btrees and this
    is simimar in cost to accessing them.
    """
539 540 541
    # Produce an MD5 from the URL
    hex_md5 = md5.md5(url).hexdigest()
    # Take the first part in the URL which is not empty
542
    # LOG("encodeURL", 0, url)
543 544 545 546 547 548 549 550 551 552 553 554 555
    url_segment = url.split(':')[1]
    url_segment_list = url_segment.split('/')
    url_domain = None
    for url_part in url_segment_list:
      if url_part:
        url_domain = url_part
        break
    # Return encoded url
    if url_domain:
      url_domain = urllib.quote(url_domain, safe='')
      url_domain = url_domain.replace('%', '')
      return "%s-%s" % (url_domain, hex_md5)
    return hex_md5
Jean-Paul Smets's avatar
Jean-Paul Smets committed
556 557 558 559 560 561
    url = urllib.quote(url, safe='')
    url = url.replace('_', '__')
    url = url.replace('%', '_')
    return url

  security.declareProtected(Permissions.AddPortalContent, 'crawlContent')
562
  def crawlContent(self, content, container=None):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
563 564 565 566 567 568 569
    """
      Analyses content and download linked pages

      XXX: missing is the conversion of content local href to something
      valid.
    """
    depth = content.getCrawlingDepth()
570 571 572 573 574 575 576 577
    if depth < 0:
      # Do nothing if crawling depth is reached
      # (this is not a duplicate code but a way to prevent
      # calling isIndexContent unnecessarily)
      return
    if not content.isIndexContent(): # Decrement depth only if it is a content document
      depth = depth - 1
    if depth < 0:
Jean-Paul Smets's avatar
Jean-Paul Smets committed
578 579 580
      # Do nothing if crawling depth is reached
      return
    base_url = content.getContentBaseURL()
581
    url_list = map(lambda url: self.normaliseURL(url, base_url), set(content.getContentURLList()))
Jean-Paul Smets's avatar
Jean-Paul Smets committed
582
    for url in set(url_list):
583
      # LOG('trying to crawl', 0, url)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
584 585 586
      # Some url protocols should not be crawled
      if url.split(':')[0] in no_crawl_protocol_list:
        continue
587 588 589 590
      if container is None:
        #if content.getParentValue()
        # in place of not ?
        container = content.getParentValue()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
591
      # Calculate the id under which content will be stored
592
      id = self.encodeURL(url)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
593 594 595 596 597 598
      # Try to access the document if it already exists
      document = container.get(id, None)
      if document is None:
        # XXX - This call is not working due to missing group_method_id
        # therefore, multiple call happen in parallel and eventually fail
        # (the same URL is created multiple times)
599
        # LOG('activate newContentFromURL', 0, url)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
600
        self.activate(activity="SQLQueue").newContentFromURL(container_path=container.getRelativeUrl(),
601 602 603 604 605
                                                      id=id, url=url, crawling_depth=depth)
      elif depth and document.getCrawlingDepth() < depth:
        # Update the crawling depth if necessary
        document._setCrawlingDepth(depth)
        document.activate().crawlContent()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
606 607

  security.declareProtected(Permissions.AddPortalContent, 'updateContentFromURL')
608
  def updateContentFromURL(self, content, repeat=MAX_REPEAT, crawling_depth=0):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
609 610 611
    """
      Updates an existing content.
    """
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
    # First, test if the document is updatable according to
    # its workflow states (if it has a workflow associated with)
    if content.isUpdatable():
      # Step 0: update crawling_depth if required
      if crawling_depth > content.getCrawlingDepth():
        content._setCrawlingDepth(crawling_depth)
      # Step 1: download new content
      try:
        url = content.asURL()
        data = urllib2.urlopen(url).read()
        file = cStringIO.StringIO()
        file.write(data)
        file.seek(0)
      except urllib2.HTTPError, error:
        if repeat == 0:
          # XXX - Call the extendBadURLList method,--NOT Implemented--
628
          # IDEA : ajouter l'url en question dans une list "bad_url_list" puis lors du crawling au lieu que de boucler sur 
629
          #        la liste des url extraites de la page web on fait un test supplementaire qui verifie que l'url n'est pas 
630
          #        dans la liste bad_url_lis
Jérome Perrin's avatar
Jérome Perrin committed
631
          raise
632 633 634 635 636
        content.activate(at_date=DateTime() + 1).updateContentFromURL(repeat=repeat - 1)
        return
      except urllib2.URLError, error:
        if repeat == 0:
          # XXX - Call the extendBadURLList method,--NOT Implemented--
Jérome Perrin's avatar
Jérome Perrin committed
637
          raise
638 639 640 641 642 643 644
        content.activate(at_date=DateTime() + 1).updateContentFromURL(repeat=repeat - 1)
        return

      # Step 2: compare and update if necessary (md5)
      # md5 stuff to compare contents
      new_content_md5 = md5.md5(data).hexdigest()
      content_md5 = content.getContentMd5()
645
      if content_md5 == new_content_md5:
646 647 648 649 650
        return
      content._edit(file=file)# Please make sure that if content is the same
                              # we do not update it
                              # This feature must be implemented by Base or File
                              # not here (look at _edit in Base)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
651
      # Step 3: convert to base format
652
      content.convertToBaseFormat()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
653
      # Step 4: activate populate (unless interaction workflow does it)
654
      content.activate().populateContent()
Jean-Paul Smets's avatar
Jean-Paul Smets committed
655
      # Step 5: activate crawlContent
656 657 658 659
      depth = content.getCrawlingDepth()
      if depth > 0:
        content.activate().crawlContent()
      content.setContentMd5(new_content_md5)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
660 661

  security.declareProtected(Permissions.AddPortalContent, 'newContentFromURL')
662
  def newContentFromURL(self, container_path=None, id=None, repeat=MAX_REPEAT, **kw):
Jean-Paul Smets's avatar
Jean-Paul Smets committed
663 664 665 666 667 668 669 670 671 672
    """
      A wrapper method for newContent which provides extra safety
      in case or errors (ie. download, access, conflict, etc.).
      The method is able to handle a certain number of exceptions
      and can postpone itself through an activity based on
      the type of exception (ex. for a 404, postpone 1 day), using
      the at_date parameter and some standard values.

      NOTE: implementation needs to be done.
    """
673 674 675 676 677 678 679 680 681
    # First of all, make sure do not try to create an existing document
    if container_path is not None and id is not None:
      container = self.restrictedTraverse(container_path)
      document = container.get(id, None)
      if document is not None:
        # Document aleardy exists: no need to keep on crawling
        return
    try:
      document = self.newContent(container_path=container_path, id=id, **kw)
682 683 684 685 686 687
      if document.isIndexContent() and document.getCrawlingDepth() >= 0:
        # If this is an index document, keep on crawling even if crawling_depth is 0
        document.activate().crawlContent()
      elif document.getCrawlingDepth() > 0:
        # If this is an index document, stop crawling if crawling_depth is 0
        document.activate().crawlContent()
688
    except urllib2.HTTPError, error:
689 690 691 692
      if repeat == 0:
        # here we must call the extendBadURLList method,--NOT Implemented--
        # which had to add this url to bad URL list, so next time we avoid
        # crawling bad URL
Jérome Perrin's avatar
Jérome Perrin committed
693
        raise
694 695 696 697 698
      # Catch any HTTP error
      self.activate(at_date=DateTime() + 1).newContentFromURL(
                        container_path=container_path, id=id,
                        repeat=repeat - 1, **kw)
    except urllib2.URLError, error:
699 700
      if repeat == 0:
        # XXX - Call the extendBadURLList method, --NOT Implemented--
Jérome Perrin's avatar
Jérome Perrin committed
701
        raise
702 703 704 705 706 707 708 709
      print error.reason
      #if getattr(error.reason,'args',None):
        #if error.reason.args[0] == socket.EAI_AGAIN:
          ## Temporary failure in name resolution - try again in 1 day
      self.activate(at_date=DateTime() + 1,
                    activity="SQLQueue").newContentFromURL(
                      container_path=container_path, id=id,
                      repeat=repeat - 1, **kw)
Jean-Paul Smets's avatar
Jean-Paul Smets committed
710

Ivan Tyagov's avatar
Ivan Tyagov committed
711
InitializeClass(ContributionTool)