class TestVIAFClient(DatabaseTest):
def setup(self):
super(TestVIAFClient, self).setup()
self.client = VIAFClient(self._db)
self.log = logging.getLogger("VIAF Client Test")
def sample_data(self, filename):
return sample_data(filename, "viaf")
def queue_file_in_mock_http(self, filename):
h = DummyHTTPClient()
xml = self.sample_data(filename)
h.queue_response(200, media_type='text/xml', content=xml)
return h
def test_process_contributor(self):
client = MockVIAFClientLookup(self._db, self.log)
contributor = self._contributor()[0]
# If lookup returns an empty array (as in the case of
# VIAFParser#parse_multiple), the contributor is not updated.
client.queue_lookup([])
client.process_contributor(contributor)
eq_(contributor.sort_name, '2001')
eq_(contributor.display_name, None)
def queue_lookup_result():
http = self.queue_file_in_mock_http("mindy_kaling.xml")
lookup = self.client.lookup_by_viaf(viaf="9581122", do_get=http.do_get)
client.results = [lookup]
# When lookup is successful, the contributor is updated.
queue_lookup_result()
client.process_contributor(contributor)
eq_(contributor.sort_name, "Kaling, Mindy")
eq_(contributor.display_name, "Mindy Kaling")
# If a contributor with the same VIAF number already exists,
# the original contributor will be updated with VIAF data
# and the processed contributor will be merged into the original.
earliest_contributor = contributor
# Reset the contributors sort name to confirm the data update.
earliest_contributor.sort_name = None
# Create a new contributor and contribution to confirm the merge.
contributor = self._contributor()[0]
edition = self._edition(authors=contributor.sort_name)
eq_(edition.contributors, set([contributor]))
queue_lookup_result()
client.process_contributor(contributor)
eq_(earliest_contributor.sort_name, "Kaling, Mindy")
eq_(edition.contributors, set([earliest_contributor]))
# The new contributor has been deleted.
assert contributor not in self._db
# If the display name of the original contributor is suspiciously
# different from the VIAF display name, the new contributor will be
# updated without being merged.
earliest_contributor.display_name = "Mindy L. Kaling"
earliest_contributor.sort_name = None
contributor = self._contributor()[0]
edition = self._edition(authors=contributor.sort_name)
queue_lookup_result()
client.process_contributor(contributor)
eq_(contributor.viaf, "9581122")
eq_(contributor.sort_name, "Kaling, Mindy")
# Earlier contributor has not been updated or merged.
eq_(earliest_contributor.sort_name, None)
assert earliest_contributor not in edition.contributors
def test_lookup_by_viaf(self):
# there can be one and only one Mindy
h = self.queue_file_in_mock_http("mindy_kaling.xml")
contributor_candidate = self.client.lookup_by_viaf(viaf="9581122", do_get=h.do_get)
(selected_candidate, match_confidences, contributor_titles) = contributor_candidate
eq_(selected_candidate.viaf, "9581122")
eq_(selected_candidate.sort_name, "Kaling, Mindy")
def test_lookup_by_name(self):
# there can be one and only one Mindy
h = self.queue_file_in_mock_http("mindy_kaling.xml")
(selected_candidate,
match_confidences,
contributor_titles) = self.client.lookup_by_name(sort_name="Mindy Kaling", do_get=h.do_get)
eq_(selected_candidate.viaf, "9581122")
eq_(selected_candidate.sort_name, "Kaling, Mindy")
Contributor,
)
from viaf import VIAFClient
_db = production_session()
viaf_client = VIAFClient(_db)
from sqlalchemy.sql import text
contributors = _db.query(Contributor).filter(
text("contributors.display_name ~ '^Q[0-9]'")).order_by(Contributor.id)
print contributors.count()
for contributor in contributors:
if contributor.viaf:
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_viaf(
contributor.viaf)
else:
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_name(
contributor.name)
print "%s: %s => %s, %s => %s" % (contributor.id, contributor.display_name,
display_name, contributor.wikipedia_name,
wikipedia_name)
contributor.display_name = display_name
contributor.wikipedia_name = wikipedia_name
contributor.family_name = family_name
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_viaf(
contributor.viaf)
for contribution in contributor.contributions:
edition = contribution.edition
if edition.work:
edition.work.calculate_presentation()
else:
edition.calculate_presentation()
_db.commit()
class TestVIAFClient(DatabaseTest):
def setup(self):
super(TestVIAFClient, self).setup()
self.client = VIAFClient(self._db)
self.log = logging.getLogger("VIAF Client Test")
def sample_data(self, filename):
return sample_data(filename, "viaf")
def queue_file_in_mock_http(self, filename):
h = DummyHTTPClient()
xml = self.sample_data(filename)
h.queue_response(200, media_type='text/xml', content=xml)
return h
def test_process_contributor(self):
client = MockVIAFClientLookup(self._db, self.log)
contributor = self._contributor()[0]
# If lookup returns an empty array (as in the case of
# VIAFParser#parse_multiple), the contributor is not updated.
client.queue_lookup([])
client.process_contributor(contributor)
eq_(contributor.sort_name, '2001')
eq_(contributor.display_name, None)
def queue_lookup_result():
http = self.queue_file_in_mock_http("mindy_kaling.xml")
lookup = self.client.lookup_by_viaf(viaf="9581122",
do_get=http.do_get)
client.results = [lookup]
# When lookup is successful, the contributor is updated.
queue_lookup_result()
client.process_contributor(contributor)
eq_(contributor.sort_name, "Kaling, Mindy")
eq_(contributor.display_name, "Mindy Kaling")
# If a contributor with the same VIAF number already exists,
# the original contributor will be updated with VIAF data
# and the processed contributor will be merged into the original.
earliest_contributor = contributor
# Reset the contributors sort name to confirm the data update.
earliest_contributor.sort_name = None
# Create a new contributor and contribution to confirm the merge.
contributor = self._contributor()[0]
edition = self._edition(authors=contributor.sort_name)
eq_(edition.contributors, set([contributor]))
queue_lookup_result()
client.process_contributor(contributor)
eq_(earliest_contributor.sort_name, "Kaling, Mindy")
eq_(edition.contributors, set([earliest_contributor]))
# The new contributor has been deleted.
assert contributor not in self._db
# If the display name of the original contributor is suspiciously
# different from the VIAF display name, the new contributor will be
# updated without being merged.
earliest_contributor.display_name = "Mindy L. Kaling"
earliest_contributor.sort_name = None
contributor = self._contributor()[0]
edition = self._edition(authors=contributor.sort_name)
queue_lookup_result()
client.process_contributor(contributor)
eq_(contributor.viaf, "9581122")
eq_(contributor.sort_name, "Kaling, Mindy")
# Earlier contributor has not been updated or merged.
eq_(earliest_contributor.sort_name, None)
assert earliest_contributor not in edition.contributors
def test_lookup_by_viaf(self):
# there can be one and only one Mindy
h = self.queue_file_in_mock_http("mindy_kaling.xml")
contributor_candidate = self.client.lookup_by_viaf(viaf="9581122",
do_get=h.do_get)
(selected_candidate, match_confidences,
contributor_titles) = contributor_candidate
eq_(selected_candidate.viaf, "9581122")
eq_(selected_candidate.sort_name, "Kaling, Mindy")
def test_lookup_by_name(self):
# there can be one and only one Mindy
h = self.queue_file_in_mock_http("mindy_kaling.xml")
(selected_candidate, match_confidences,
contributor_titles) = self.client.lookup_by_name(
sort_name="Mindy Kaling", do_get=h.do_get)
eq_(selected_candidate.viaf, "9581122")
eq_(selected_candidate.sort_name, "Kaling, Mindy")
Contributor,
)
from viaf import VIAFClient
_db = production_session()
viaf_client = VIAFClient(_db)
from sqlalchemy.sql import text
contributors = _db.query(Contributor).filter(
text("contributors.display_name ~ '^Q[0-9]'")
).order_by(Contributor.id)
print contributors.count()
for contributor in contributors:
if contributor.viaf:
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_viaf(contributor.viaf)
else:
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_name(contributor.name)
print "%s: %s => %s, %s => %s" % (
contributor.id,
contributor.display_name, display_name,
contributor.wikipedia_name, wikipedia_name
)
contributor.display_name = display_name
contributor.wikipedia_name = wikipedia_name
contributor.family_name = family_name
viaf, display_name, family_name, sort_name, wikipedia_name = viaf_client.lookup_by_viaf(contributor.viaf)
for contribution in contributor.contributions:
edition = contribution.edition
if edition.work:
edition.work.calculate_presentation()
else:
edition.calculate_presentation()
