def __extract_title__(self):
"Helper function extracts title information from MODS"
title_entities = []
titleInfos = self.mods_xml.findall('{{{0}}}titleInfo'.format(MODS_NS))
for titleInfo in titleInfos:
output = {}
if titleInfo.attrib.get('type')is None:
# equalvant to MARC 245 $a
titleValue = titleInfo.find('{{{0}}}title'.format(MODS_NS))
if titleValue is not None and len(titleValue.text) > 0:
output['titleValue'] = titleValue.text
output['label'] = output['titleValue']
# equalvant to MARC 245 $b
subtitle = titleInfo.find('{{{0}}}subTitle'.format(MODS_NS))
if subtitle is not None and len(subtitle.text) > 0:
output['subtitle'] = subtitle.text
output['label'] = '{0}: {1}'.format(output.get('label'),
output['subtitle'])
# equalivant to MARC 245 $p
partTitle = titleInfo.find('{{{0}}}partName'.format(MODS_NS))
if partTitle is not None and len(partTitle.text) > 0:
output['partTitle'] = partTitle.text
if len(output) > 0:
title_entity = TitleEntity(redis_datastore=self.redis_datastore,
**output)
title_entity.save()
index_title(title_entity, self.redis_datastore)
title_entities.append(title_entity.redis_key)
return title_entities
def __add_title_entity__(self):
"Helper method adds a new TitleEntity"
self.title_entity = TitleEntity(redis_datastore=self.redis_datastore)
for key, value in self.entity_info.iteritems():
if key is not None and value is not None:
setattr(self.title_entity,
key,
value)
self.title_entity.save()
class TestWorkClassifier(unittest.TestCase):
def setUp(self):
self.title_entity = TitleEntity(redis_datastore = TEST_REDIS,
label='Pride and Prejudice',
titleValue='Pride and Prejudice')
self.title_entity.save()
index_title(self.title_entity, TEST_REDIS)
self.work = Book(redis_datastore=TEST_REDIS,
associatedAgent=set(["bf:Person:1"]),
title=self.title_entity.redis_key)
setattr(self.work, 'rda:isCreatedBy', 'bf:Person:1')
self.work.save()
TEST_REDIS.sadd(
"{0}:relatedResources".format(self.title_entity.redis_key),
self.work.redis_key)
def test_init(self):
entity_info = {'title': 'Pride and Prejudice',
'rda:isCreatedBy': set(['bf:Person:1'])}
classifier = simple_fuzzy.WorkClassifier(redis_datastore = TEST_REDIS,
entity_info = entity_info,
work_class=Book)
self.assert_(classifier is not None)
self.assert_(classifier.strict is True)
self.assert_(classifier.creative_work is None)
self.assertEquals(classifier.entity_info,
entity_info)
def test_exact_match(self):
entity_info = {'rda:isCreatedBy': set(['bf:Person:1']),
'title': self.title_entity.redis_key,}
classifier = simple_fuzzy.WorkClassifier(redis_datastore = TEST_REDIS,
entity_info = entity_info,
work_class=Book)
classifier.classify()
self.assertEquals(classifier.creative_work.redis_key,
self.work.redis_key)
def tearDown(self):
TEST_REDIS.flushdb()
def setUp(self):
self.title_entity = TitleEntity(redis_datastore = TEST_REDIS,
label='Pride and Prejudice',
titleValue='Pride and Prejudice')
self.title_entity.save()
index_title(self.title_entity, TEST_REDIS)
self.work = Book(redis_datastore=TEST_REDIS,
associatedAgent=set(["bf:Person:1"]),
title=self.title_entity.redis_key)
setattr(self.work, 'rda:isCreatedBy', 'bf:Person:1')
self.work.save()
TEST_REDIS.sadd(
"{0}:relatedResources".format(self.title_entity.redis_key),
self.work.redis_key)
class MARC21toTitleEntity(MARCParser):
"Extracts BIBFRAME TitleEntity info from MARC21 record"
def __init__(self, **kwargs):
"""Initializes MARC21toTitleEntity object
Parameters:
"""
kwargs['rules_filename'] = 'bibframe-title-entity-map.json'
super(MARC21toTitleEntity, self).__init__(**kwargs)
self.title_entity = None
def __add_title_entity__(self):
"Helper method adds a new TitleEntity"
self.title_entity = TitleEntity(redis_datastore=self.redis_datastore)
for key, value in self.entity_info.iteritems():
if key is not None and value is not None:
setattr(self.title_entity,
key,
value)
self.title_entity.save()
def __get_or_add_title_entity__(self):
"Helper method returns new or existing TitleEntity"
existing_titles = []
if self.entity_info.get('titleValue') is not None:
title_string = title
if self.entity_info.get('subtitle') is not None:
title_string += " {0}".format(
self.entity_info.get('subtitle'))
self.entity_info['label'] = title_string
def ingest(self):
"Method finds or creates a TitleEntity in RLSP"
self.parse()
self.__add_title_entity__()
def ingest(self):
"Method ingests MARC Record into RLSP"
self.__classify_work_class__()
self.creative_work = self.work_class(
redis_datastore=self.redis_datastore)
work_titles = []
for attribute, rules in self.creative_work.marc_map.iteritems():
values = []
#! NEED TitleEntity to check for duplicates
if attribute == 'uniformTitle':
pass
if attribute == 'title':
rule = rules[0]
titleValue = ' '.join(self.__rule_one__(rule))
title_entity = TitleEntity(redis_datastore=self.redis_datastore,
titleValue=titleValue,
label=self.record.title())
title_entity.save()
index_title(title_entity, self.redis_datastore)
self.entity_info[attribute] = title_entity.redis_key
work_titles.append(title_entity.redis_key)
continue
for rule in rules:
result = list(set(self.__rule_one__(rule)))
values.extend(result)
if len(values) > 0:
self.entity_info[attribute] = values
# List of specific methods that haven't had Rule regex developed
self.extract_creators()
self.extract_note()
self.extract_performerNote()
self.get_or_add_work()
if self.creative_work is not None:
for title_key in work_titles:
self.redis_datastore.sadd(
"{0}:relatedResources".format(title_key),
self.creative_work.redis_key)
