def _use_rule_wrappers(self, source, page, raw_text):
     """
     Look if there is any wrapper in the database for the given source.
     """
     log.info('Attempting to extract reference with ruled wrappers') #@UndefinedVariable
     fields = {}
     reference = Reference()
     wrapper_manager = WrapperGateway(max_wrappers=self.max_wrappers)
     wrapper_field_collections = wrapper_manager.find_wrapper_collections(source)
     
     for collection in wrapper_field_collections:
         # Get the wrappers for the current collection
         url, field = collection.url, collection.field
         wrappers = wrapper_manager.get_wrappers(url, field)
         log.debug('Collection %s:%s has %d wrappers' % (url, field, #@UndefinedVariable
                                                         len(wrappers)))
         
         # Get field validator
         try:
             validator = self.field_validation[collection.field][1]
         except KeyError:
             validator = None
         
         # Extract information using the wrappers we have
         for wrapper in wrappers:
             info = wrapper.extract_info(page)
             # we expect 'info' to be a string
             if type(info) == list and not (collection.field == 'author' 
                  or collection.field == 'editor'):
                 continue 
             log.debug('Info extracted by wrapper: %s' % info) #@UndefinedVariable
             
             valid = validator.validate(info, raw_text) if validator else True
             # Save the extracted info even if it's not correct. It will
             # be overwritten afterwards if necessary
             reference.set_field(field, info, valid)
             
             if not valid: 
                 log.debug('The extracted information is not valid. ' #@UndefinedVariable
                           'Downvoting wrapper.') 
                 wrapper.downvotes += 1
                 wrapper_manager.update_wrapper(wrapper)
             else:
                 log.debug('The extracted information is valid. ' #@UndefinedVariable
                           'Upvoting wrapper') 
                 wrapper.upvotes += 1
                 wrapper_manager.update_wrapper(wrapper)
                 fields[field] = info
                 break
             
     if len(reference.fields) > 0:
         log.info('Extracted reference')  #@UndefinedVariable
         return [reference]
     else:
         log.info('Could not extract reference using ruled wrappers')  #@UndefinedVariable
         return []
Example #2
0
    def _use_rule_wrappers(self, source, page, raw_text):
        """
        Look if there is any wrapper in the database for the given source.
        """
        log.info('Attempting to extract reference with ruled wrappers'
                 )  #@UndefinedVariable
        fields = {}
        reference = Reference()
        wrapper_manager = WrapperGateway(max_wrappers=self.max_wrappers)
        wrapper_field_collections = wrapper_manager.find_wrapper_collections(
            source)

        for collection in wrapper_field_collections:
            # Get the wrappers for the current collection
            url, field = collection.url, collection.field
            wrappers = wrapper_manager.get_wrappers(url, field)
            log.debug('Collection %s:%s has %d wrappers' % (
                url,
                field,  #@UndefinedVariable
                len(wrappers)))

            # Get field validator
            try:
                validator = self.field_validation[collection.field][1]
            except KeyError:
                validator = None

            # Extract information using the wrappers we have
            for wrapper in wrappers:
                info = wrapper.extract_info(page)
                # we expect 'info' to be a string
                if type(info) == list and not (collection.field == 'author' or
                                               collection.field == 'editor'):
                    continue
                log.debug('Info extracted by wrapper: %s' %
                          info)  #@UndefinedVariable

                valid = validator.validate(info,
                                           raw_text) if validator else True
                # Save the extracted info even if it's not correct. It will
                # be overwritten afterwards if necessary
                reference.set_field(field, info, valid)

                if not valid:
                    log.debug(
                        'The extracted information is not valid. '  #@UndefinedVariable
                        'Downvoting wrapper.')
                    wrapper.downvotes += 1
                    wrapper_manager.update_wrapper(wrapper)
                else:
                    log.debug(
                        'The extracted information is valid. '  #@UndefinedVariable
                        'Upvoting wrapper')
                    wrapper.upvotes += 1
                    wrapper_manager.update_wrapper(wrapper)
                    fields[field] = info
                    break

        if len(reference.fields) > 0:
            log.info('Extracted reference')  #@UndefinedVariable
            return [reference]
        else:
            log.info('Could not extract reference using ruled wrappers'
                     )  #@UndefinedVariable
            return []
class TestWrapperManager(unittest.TestCase):
    
    def setUp(self):
        self.wm = WrapperGateway(create_session(
            sql_uri='sqlite:///:memory:', debug=True))
    
    def test_find_collection(self):
        # Do not create
        collection1 = self.wm.find_wrapper_collection(u'some_url',
                                                     u'some_field')
        self.failIf(collection1)
        
        # New collection
        collection1 = self.wm.find_wrapper_collection(u'some_url',
                                                     u'some_field', True)
        self.failUnless(collection1)
        self.failUnless(type(collection1) == mappers.WrapperCollection)
        
        # Existent collection
        collection2 = self.wm.find_wrapper_collection(u'some_url',
                                                     u'some_field')
        self.failUnless(collection2)
        self.failUnless(collection1 is collection2)
    
    def test_find_collections(self):
        collection11 = self.wm.find_wrapper_collection(u'c01', u'f01', True) #@UnusedVariable
        collection12 = self.wm.find_wrapper_collection(u'c01', u'f02', True) #@UnusedVariable
        collection21 = self.wm.find_wrapper_collection(u'c02', u'f01', True) #@UnusedVariable
        collection22 = self.wm.find_wrapper_collection(u'c02', u'f02', True) #@UnusedVariable
        
        collections = self.wm.find_wrapper_collections()
        self.failUnless(collections.count() >= 4)
        
        collections = self.wm.find_wrapper_collections(url=u'c02')
        self.failUnless(collections.count() == 2)
    
        collections = self.wm.find_wrapper_collections(field=u'f02')
        self.failUnless(collections.count() == 2)
            
    def test_get_unavailable_wrappers(self):
        wrappers = self.wm.get_wrappers(u'non_existent_url', u'no_field')
        self.failUnless(wrappers == [])

    def xtest_persist_wrapper_with_incorrect_rules(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(MockRule02(33)))
        self.failUnlessRaises(TypeError, self.wm.persist_wrapper,
                              u'some_url', u'some_field', wrapper)
    
    def xtest_persist_and_get_wrapper(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(33))
        wrapper.add_rule(MockRule01(55))
        wrapper.add_rule(MockRule02([1, [2, 3, 4, 5], 6]))
        self.wm.persist_wrapper(u'some_url', u'some_field', wrapper)

        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(66))
        wrapper.add_rule(MockRule01(77))
        wrapper.add_rule(MockRule02([[2, 3, 4, 5], 4]))
        self.wm.persist_wrapper(u'some_url', u'some_field', wrapper)

        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(11))
        wrapper.add_rule(MockRule01(22))
        wrapper.add_rule(MockRule01(33))
        self.wm.persist_wrapper(u'some_url', u'some_other_field', wrapper)
    
        # Get non-existent wrapper
        wrappers = self.wm.get_wrappers(u'some_url', u'non_existent_field')
        self.failIf(wrappers, 'Get non-existent wrapper')
        
        # Get wrappers
        wrappers = self.wm.get_wrappers(u'some_url', u'some_field')
        self.failUnless(len(wrappers) == 2)
        wrappers = self.wm.get_wrappers(u'some_url', u'some_other_field')
        self.failUnless(len(wrappers) == 1)
        
    def test_persist_and_update_wrapper(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(33))
        wrapper.add_rule(MockRule01(55))
        wrapper.add_rule(MockRule02([1, [2, 3, 4, 5], 6]))
        self.wm.persist_wrapper(u'concrete_url', u'concrete_field', wrapper)
        
        # Get wrappers
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        
        # Update wrapper
        wrapper = wrappers[0]
        wrapper.upvotes += 1
        wrapper.downvotes -= 1
        self.wm.update_wrapper(wrapper)
        
        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(wrappers[0].upvotes == 1)
        self.failUnless(wrappers[0].downvotes == -1)
        
        # Update wrapper rules
        wrapper = wrappers[0]
        wrapper.rules[0].pattern = 223
        wrapper.rules[2].pattern = [1, 6]
        self.wm.update_wrapper(wrapper)
        
        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(wrappers[0].rules[0].pattern == 223)
        self.failUnless(wrappers[0].rules[2].pattern == [1, 6])
        
        # Add another wrapper rule
        wrapper = wrappers[0]
        wrapper.rules.append(MockRule02([2, 3, 4, 5]))
        self.wm.update_wrapper(wrapper)
        
        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(len(wrappers[0].rules) == 4)
        self.failUnless(wrappers[0].rules[3].pattern == [2, 3, 4, 5])
class TestWrapperManager(unittest.TestCase):
    def setUp(self):
        self.wm = WrapperGateway(
            create_session(sql_uri='sqlite:///:memory:', debug=True))

    def test_find_collection(self):
        # Do not create
        collection1 = self.wm.find_wrapper_collection(u'some_url',
                                                      u'some_field')
        self.failIf(collection1)

        # New collection
        collection1 = self.wm.find_wrapper_collection(u'some_url',
                                                      u'some_field', True)
        self.failUnless(collection1)
        self.failUnless(type(collection1) == mappers.WrapperCollection)

        # Existent collection
        collection2 = self.wm.find_wrapper_collection(u'some_url',
                                                      u'some_field')
        self.failUnless(collection2)
        self.failUnless(collection1 is collection2)

    def test_find_collections(self):
        collection11 = self.wm.find_wrapper_collection(u'c01', u'f01',
                                                       True)  #@UnusedVariable
        collection12 = self.wm.find_wrapper_collection(u'c01', u'f02',
                                                       True)  #@UnusedVariable
        collection21 = self.wm.find_wrapper_collection(u'c02', u'f01',
                                                       True)  #@UnusedVariable
        collection22 = self.wm.find_wrapper_collection(u'c02', u'f02',
                                                       True)  #@UnusedVariable

        collections = self.wm.find_wrapper_collections()
        self.failUnless(collections.count() >= 4)

        collections = self.wm.find_wrapper_collections(url=u'c02')
        self.failUnless(collections.count() == 2)

        collections = self.wm.find_wrapper_collections(field=u'f02')
        self.failUnless(collections.count() == 2)

    def test_get_unavailable_wrappers(self):
        wrappers = self.wm.get_wrappers(u'non_existent_url', u'no_field')
        self.failUnless(wrappers == [])

    def xtest_persist_wrapper_with_incorrect_rules(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(MockRule02(33)))
        self.failUnlessRaises(TypeError, self.wm.persist_wrapper, u'some_url',
                              u'some_field', wrapper)

    def xtest_persist_and_get_wrapper(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(33))
        wrapper.add_rule(MockRule01(55))
        wrapper.add_rule(MockRule02([1, [2, 3, 4, 5], 6]))
        self.wm.persist_wrapper(u'some_url', u'some_field', wrapper)

        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(66))
        wrapper.add_rule(MockRule01(77))
        wrapper.add_rule(MockRule02([[2, 3, 4, 5], 4]))
        self.wm.persist_wrapper(u'some_url', u'some_field', wrapper)

        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(11))
        wrapper.add_rule(MockRule01(22))
        wrapper.add_rule(MockRule01(33))
        self.wm.persist_wrapper(u'some_url', u'some_other_field', wrapper)

        # Get non-existent wrapper
        wrappers = self.wm.get_wrappers(u'some_url', u'non_existent_field')
        self.failIf(wrappers, 'Get non-existent wrapper')

        # Get wrappers
        wrappers = self.wm.get_wrappers(u'some_url', u'some_field')
        self.failUnless(len(wrappers) == 2)
        wrappers = self.wm.get_wrappers(u'some_url', u'some_other_field')
        self.failUnless(len(wrappers) == 1)

    def test_persist_and_update_wrapper(self):
        wrapper = Wrapper()
        wrapper.add_rule(MockRule01(33))
        wrapper.add_rule(MockRule01(55))
        wrapper.add_rule(MockRule02([1, [2, 3, 4, 5], 6]))
        self.wm.persist_wrapper(u'concrete_url', u'concrete_field', wrapper)

        # Get wrappers
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)

        # Update wrapper
        wrapper = wrappers[0]
        wrapper.upvotes += 1
        wrapper.downvotes -= 1
        self.wm.update_wrapper(wrapper)

        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(wrappers[0].upvotes == 1)
        self.failUnless(wrappers[0].downvotes == -1)

        # Update wrapper rules
        wrapper = wrappers[0]
        wrapper.rules[0].pattern = 223
        wrapper.rules[2].pattern = [1, 6]
        self.wm.update_wrapper(wrapper)

        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(wrappers[0].rules[0].pattern == 223)
        self.failUnless(wrappers[0].rules[2].pattern == [1, 6])

        # Add another wrapper rule
        wrapper = wrappers[0]
        wrapper.rules.append(MockRule02([2, 3, 4, 5]))
        self.wm.update_wrapper(wrapper)

        # Get the wrapper again
        wrappers = self.wm.get_wrappers(u'concrete_url', u'concrete_field')
        self.failUnless(len(wrappers) == 1)
        self.failUnless(len(wrappers[0].rules) == 4)
        self.failUnless(wrappers[0].rules[3].pattern == [2, 3, 4, 5])