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 regulate_node(self, node):
old_iri = node[self.iri_field]
try:
ret = IRILink(urn_fallback=self.urn_fallback).execute(old_iri)
node[self.authority_field] = ret['authority']
node[self.scheme_field] = ret['scheme']
new_iri = ret['IRI']
if old_iri != new_iri:
node[self.iri_field] = new_iri
self.info(
'Normalized IRI "{}" into "{}"'.format(old_iri, new_iri),
node.id)
if self.blocked_schemes and ret['scheme'] in self.blocked_schemes:
self.info(
'Discarding identifier based on invalid scheme "{}"'.
format(ret['scheme']), node.id)
node.delete()
elif self.blocked_authorities and ret[
'authority'] in self.blocked_authorities:
self.info(
'Discarding identifier based on invalid authority "{}"'.
format(ret['authority']), node.id)
node.delete()
except InvalidIRI as e:
self.info(
'Discarding identifier based on unrecognized IRI "{}"'.format(
old_iri), node.id)
node.delete()
def normalize(self, node, graph):
try:
ret = IRILink().execute(node.attrs['uri'])
except InvalidIRI 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 normalize(self, node, graph):
try:
ret = IRILink().execute(node.attrs['uri'])
except InvalidIRI as e:
logger.warning('Discarding invalid identifier %s with error %s',
node.attrs['uri'], e)
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):
from share.transform.chain.links import IRILink
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}
root_node = next(
n for n in jsonld['@graph']
if n['@id'] == root_ref['@id'] and n['@type'] == root_ref['@type'])
root_node.setdefault('identifiers', []).append(identifier_ref)
jsonld['@graph'].append(identifier)
def test_benchmark(self, input, benchmark):
benchmark(IRILink().execute, input)
def is_valid_iri(iri):
try:
IRILink().execute(iri)
except InvalidIRI:
return False
return True
def is_valid_iri(iri):
if not isinstance(iri, str):
return False
IRILink().execute(iri)
return True
def parse(self, _, parse, **kwargs):
if parse:
parsed = IRILink().execute(self['uri'])
self['uri'] = parsed['IRI']
self['scheme'] = parsed['scheme']
self['host'] = parsed['authority']
def _parse(self, *args, **kwargs):
if self.attrs.pop('parse'):
parsed = IRILink().execute(self.attrs['uri'])
self.attrs['uri'] = parsed['IRI']
self.attrs['scheme'] = parsed['scheme']
self.attrs['host'] = parsed['authority']