def _do_test(self, input, output, urn_fallback=False):
if isinstance(output, Exception):
with pytest.raises(type(output)) as e:
IRILink().execute(input)
assert e.value.args == output.args
else:
assert {
k: v
for k, v in IRILink(
urn_fallback=urn_fallback).execute(input).items()
if k in output
} == output
def normalize(self, node, graph):
try:
ret = IRILink().execute(node.attrs['uri'])
except ValueError as e:
logger.warning('Discarding invalid identifier %s with error %s',
node.attrs['uri'], e)
graph.remove(node)
return
if ret['authority'] in {'issn', 'orcid.org'
} or ret['scheme'] in {'mailto'}:
logger.warning(
'Discarding %s %s as an invalid identifier for works',
ret['authority'], ret['IRI'])
graph.remove(node)
return
if node.attrs['uri'] != ret['IRI']:
logger.debug('Normalized %s to %s', node.attrs['uri'], ret['IRI'])
node.attrs = {
'uri': ret['IRI'],
'host': ret['authority'],
'scheme': ret['scheme'],
}
def add_source_identifier(self, source_id, jsonld, root_ref):
uri = IRILink(urn_fallback=True).execute(str(source_id))['IRI']
if any(n['@type'].lower() == 'workidentifier' and n['uri'] == uri for n in jsonld['@graph']):
return
identifier_ref = {
'@id': '_:' + uuid.uuid4().hex,
'@type': 'workidentifier'
}
identifier = {
'uri': uri,
'creative_work': root_ref,
**identifier_ref
}
ctx.pool[root_ref].setdefault('identifiers', []).append(identifier_ref)
jsonld['@graph'].append(identifier)
def _parse(self, *args, **kwargs):
if self.attrs.pop('parse'):
parsed = IRILink().execute(self.attrs['uri'])
self.attrs['scheme'] = parsed['scheme']
self.attrs['host'] = parsed['authority']
def test_benchmark(self, input, benchmark):
benchmark(IRILink().execute, input)
def handle(self, id_file, *args, **options):
id_map = yaml.load(id_file)
temp_id_query = 'select nextval(%s);'
fk_query = '''
SELECT kcu.table_name, kcu.column_name
FROM information_schema.table_constraints AS tc
JOIN information_schema.constraint_column_usage AS ccu
USING (constraint_schema, constraint_name)
JOIN information_schema.key_column_usage AS kcu
USING (constraint_schema, constraint_name)
WHERE tc.constraint_type = 'FOREIGN KEY'
AND ccu.table_name = %s
AND ccu.column_name = %s;
'''
work_id_query = '''
SELECT DISTINCT wi.creative_work_id
FROM share_workidentifier wi
WHERE wi.uri IN %s;
'''
with transaction.atomic():
with connection.cursor() as c:
c.execute(temp_id_query, ['share_creativework_id_seq'])
temp_id = c.fetchone()[0]
c.execute(fk_query, ['share_creativework', 'id'])
foreign_keys = c.fetchall()
tables = ['share_creativework', *[t for t, _ in foreign_keys]]
self.disable_triggers(c, tables)
for new_id, identifiers in id_map.items():
uris = []
for app_label, identifier in identifiers:
Context().config = apps.get_app_config(app_label)
uris.append(
IRILink(
urn_fallback=True).execute(identifier)['IRI'])
c.execute(work_id_query, [tuple(uris)])
old_id = c.fetchone()
if old_id:
old_id = old_id[0]
if old_id == new_id:
continue
print('Moving {} to {}...'.format(old_id, new_id))
if AbstractCreativeWork.objects.filter(
id=new_id).exists():
self.update_id(c, new_id, temp_id, foreign_keys)
self.update_id(c, old_id, new_id, foreign_keys)
self.update_id(c, temp_id, old_id, foreign_keys)
else:
self.update_id(c, old_id, new_id, foreign_keys)
else:
print('Skipping {}! No work found for identifiers: {}'.
format(new_id, uris))
# Cannot update a table and then alter it in the same transaction
with transaction.atomic():
with connection.cursor() as c:
self.enable_triggers(c, tables)