def _get_putcodes_and_recids_iter(self, putcodes):
for putcode, url in self._get_urls_for_putcodes_iter(putcodes):
# Filter out putcodes that do not belong to Inspire.
if INSPIRE_WORK_URL_REGEX.match(url):
recid = get_pid_from_record_uri(url)[1]
if not recid:
logger.error('OrcidPutcodeGetter: cannot parse recid from url={} for orcid={}'.format(
url, self.orcid))
continue
yield putcode, recid
def resolve_conference_record_as_root(self, pub_info_item):
conference_record = pub_info_item.get('conference_record')
if conference_record is None:
return {}
_, recid = get_pid_from_record_uri(conference_record.get('$ref'))
conference = get_db_record('con', recid).dumps()
titles = conference.get('titles')
if titles is None:
return {}
return conference
def _get_putcodes_and_recids_iter(self, putcodes):
for putcode, url in self._get_urls_for_putcodes_iter(putcodes):
# Filter out putcodes that do not belong to Inspire.
if INSPIRE_WORK_URL_REGEX.match(url):
recid = get_pid_from_record_uri(url)[1]
if not recid:
logger.error(
'OrcidPutcodeGetter: cannot parse recid from url={} for orcid={}'
.format(url, self.orcid))
continue
yield putcode, recid
def resolve_conference_record_as_root(self, pub_info_item):
conference_record = pub_info_item.get('conference_record')
if conference_record is None:
return {}
_, recid = get_pid_from_record_uri(conference_record.get('$ref'))
try:
conference = get_db_record('con', recid)
except RecordGetterError:
return {}
titles = conference.get('titles')
if titles is None:
return {}
return conference.to_dict()
def _cache_all_author_putcodes(self):
logger.info(
'New OrcidPusher cache all author putcodes for orcid={}'.format(
self.orcid))
putcode_getter = OrcidPutcodeGetter(self.orcid, self.oauth_token)
putcodes_urls = list(putcode_getter.get_all_inspire_putcodes(
)) # Can raise exceptions.InputDataInvalidException.
putcode = None
for fetched_putcode, fetched_url in putcodes_urls:
fetched_recid = get_pid_from_record_uri(fetched_url)[1]
if not fetched_recid:
logger.error(
'OrcidPusher cache all author putcodes: cannot parse recid from url={} for orcid={}'
.format(fetched_url, self.orcid))
continue
if fetched_recid == str(self.recid):
putcode = fetched_putcode
cache = OrcidCache(self.orcid, fetched_recid)
cache.write_work_putcode(fetched_putcode)
if not putcode:
raise exceptions.PutcodeNotFoundInOrcidException(
'No putcode was found in ORCID API for orcid={} and recid={}.'
' And the POST has previously failed for the same recid because'
' the work had already existed'.format(self.orcid, self.recid))
# Ensure the putcode is actually in cache.
# Note: this step is not really necessary and it can be skipped, but
# at this moment it helps isolate a potential issue.
if not self.cache.read_work_putcode():
raise exceptions.PutcodeNotFoundInCacheAfterCachingAllPutcodes(
'No putcode={} found in cache for recid={} after having'
' cached all author putcodes for orcid={}'.format(
self.putcode, self.recid, self.orcid))
return putcode
def _get_ids_from_refs(references):
return set([
get_pid_from_record_uri(ref['record']['$ref'])
for ref in references
if 'record' in ref
])
def test_get_pid_from_record_uri_non_url():
record_uri = 'non-url-string'
assert not get_pid_from_record_uri(record_uri)
def test_get_pid_from_record_uri_ending_slash():
record_uri = 'http://labs.inspirehep.net/api/literature/1273685/'
assert get_pid_from_record_uri(record_uri) == ('lit', '1273685')
def test_get_pid_from_record_uri_happy_flow():
record_uri = 'http://labs.inspirehep.net/api/literature/1273685'
assert get_pid_from_record_uri(record_uri) == ('lit', '1273685')
def _get_ids_from_refs(references):
return set([
get_pid_from_record_uri(ref['record']['$ref'])
for ref in references if 'record' in ref
])
def test_get_pid_from_record_uri_non_url():
record_uri = 'non-url-string'
assert not get_pid_from_record_uri(record_uri)
def test_get_pid_from_record_uri_ending_slash():
record_uri = 'http://labs.inspirehep.net/api/literature/1273685/'
assert get_pid_from_record_uri(record_uri) == ('lit', '1273685')
def test_get_pid_from_record_uri_happy_flow():
record_uri = 'http://labs.inspirehep.net/api/literature/1273685'
assert get_pid_from_record_uri(record_uri) == ('lit', '1273685')
