def delete_work_for_orcid(orcid):
oauth_token = TestOrcidPusherBase._oauth_token(orcid)
client = OrcidClient(oauth_token, orcid)
all_work = client.get_all_works_summary()
for work in all_work.get("group", []):
putcode = work["work-summary"][0]["put-code"]
client.delete_work(putcode)
def test_get_putcodes_for_source_source_client_id_none(self):
orcid = '0000-0002-4490-1930'
client = OrcidClient('mytoken', orcid)
response = client.get_all_works_summary()
response.raise_for_result()
putcodes = list(
response.get_putcodes_for_source_iter('0000-0001-8607-8906'))
assert len(putcodes) == 90
class OrcidPutcodeGetter(object):
def __init__(self, orcid, oauth_token):
self.orcid = orcid
self.oauth_token = oauth_token
self.client = OrcidClient(self.oauth_token, self.orcid)
self.source_client_id_path = current_app.config[
'ORCID_APP_CREDENTIALS']['consumer_key']
def get_all_inspire_putcodes(self):
"""
Get all the Inspire putcodes for the given ORCID.
"""
putcodes = self._get_all_putcodes()
if not putcodes:
return
# Filter out putcodes that do not belong to Inspire.
for putcode, url in self._get_urls_for_putcodes(putcodes):
if INSPIRE_WORK_URL_REGEX.match(url):
yield putcode, url
def _get_all_putcodes(self):
response = self.client.get_all_works_summary()
utils.log_service_response(logger, response,
'in OrcidPutcodeGetter works summary')
try:
response.raise_for_result()
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
return list(
response.get_putcodes_for_source(self.source_client_id_path))
def _get_urls_for_putcodes(self, putcodes):
# The call get_bulk_works_details_iter() can be very expensive for an
# author with many works (if each work also has many *contributors*).
# Fi. for an ATLAS author with ~750 works, 8 calls would be performed
# with a total data transfer > 0.5 Gb.
chained = []
for response in self.client.get_bulk_works_details_iter(putcodes):
utils.log_service_response(logger, response,
'in OrcidPutcodeGetter works details')
try:
response.raise_for_result()
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
chained = itertools.chain(chained,
response.get_putcodes_and_urls())
return chained
def get_putcode_for_work(orcid, token, recid):
client = OrcidClient(token, orcid)
response = client.get_all_works_summary()
response.raise_for_result()
source_client_id_path = config.get('orcid-api', 'consumer_key')
putcodes = list(
response.get_putcodes_for_source_iter(source_client_id_path))
if not putcodes:
return None
# TODO: this has to be simplified when we push recids as external
# identifier (thus just the get_all_works_summary() call is required to
# match recids with putcodes).
for response in client.get_bulk_works_details_iter(putcodes):
response.raise_for_result()
for putcode, url in response.get_putcodes_and_urls_iter():
if url.endswith('/{}'.format(recid)):
return putcode
class OrcidPutcodeGetter(object):
def __init__(self, orcid, oauth_token):
self.orcid = orcid
self.oauth_token = oauth_token
self.client = OrcidClient(self.oauth_token, self.orcid)
self.source_client_id_path = current_app.config[
"ORCID_APP_CREDENTIALS"]["consumer_key"]
def get_all_inspire_putcodes_and_recids_iter(self):
"""
Query ORCID api and get all the Inspire putcodes for the given ORCID.
"""
summary_response = self._get_all_works_summary()
# `putcodes_recids` is a list like: [('43326850', 20), ('43255490', None)]
putcodes_recids = list(
summary_response.get_putcodes_and_recids_for_source_iter(
self.source_client_id_path))
putcodes_with_recids = [x for x in putcodes_recids if x[1]]
putcodes_without_recids = [x[0] for x in putcodes_recids if not x[1]]
for putcode, recid in putcodes_with_recids:
yield putcode, recid
if not putcodes_without_recids:
return
for putcode, recid in self._get_putcodes_and_recids_iter(
putcodes_without_recids):
yield putcode, recid
def _get_all_works_summary(self):
"""
Query ORCID api and get all the putcodes with their embedded recids
for the given ORCID.
An embedded recid is a recid written as external-identifier.
"""
response = self.client.get_all_works_summary()
LOGGER.info("Get ORCID work summary",
response=response,
orcid=self.orcid)
try:
response.raise_for_result()
except (
orcid_client_exceptions.TokenInvalidException,
orcid_client_exceptions.TokenMismatchException,
orcid_client_exceptions.TokenWithWrongPermissionException,
):
LOGGER.info(
"OrcidPutcodeGetter: deleting Orcid push access",
token=self.oauth_token,
orcid=self.orcid,
)
push_access_tokens.delete_access_token(self.oauth_token,
self.orcid)
raise exceptions.TokenInvalidDeletedException
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
return response
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 = PidStoreBase.get_pid_from_record_uri(url)[1]
if not recid:
LOGGER.error(
"OrcidPutcodeGetter: cannot parse recid from url",
url=url,
orcid=self.orcid,
)
continue
yield putcode, recid
def _get_urls_for_putcodes_iter(self, putcodes):
# The call `get_bulk_works_details_iter()` can be expensive for an
# author with many works (if each work also has many *contributors*).
# Fi. for an ATLAS author with ~750 works (each of them with many
# authors), 8 calls would be performed with a total data transfer > 0.5 Gb.
chained = []
for response in self.client.get_bulk_works_details_iter(putcodes):
# Note: this log can be large. Consider removing it when this part
# is considered mature.
LOGGER.info("ORCID work details",
response=response,
orcid=self.orcid)
try:
response.raise_for_result()
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
chained = itertools.chain(chained,
response.get_putcodes_and_urls_iter())
return chained
def get_putcodes_and_recids_by_identifiers_iter(self, identifiers):
"""
Yield putcode and recid for each work matched by the external
identifiers.
Note: external identifiers of type 'other-id' are skipped.
Args:
identifiers (List[inspirehep.orcid.converter.ExternalIdentifier]):
list af all external identifiers added after the xml conversion.
"""
summary_response = self._get_all_works_summary()
for (
putcode,
ids,
) in summary_response.get_putcodes_and_external_identifiers_iter():
# ids is a list like:
# [
# {'external-id-relationship': 'SELF',
# 'external-id-type': 'other-id',
# 'external-id-url': {'value': 'http://inspireheptest.cern.ch/record/20'},
# 'external-id-value': '20'
# },...
# ]
# Get the recid.
recid = self._get_recid_for_work(ids, str(putcode))
for identifier in ids:
id_type = identifier.get("external-id-type")
# We are interested only in doi, arxiv, isbns.
if not id_type or id_type.lower() == "other-id":
continue
id_value = identifier.get("external-id-value")
if not id_value:
continue
if ExternalIdentifier(id_type, id_value) in identifiers:
yield putcode, recid
def _get_recid_for_work(self, external_identifiers, putcode):
"""
Get the recid for a work given its external identifiers and putcode.
The recid might be in the external identifiers or a get_work_details()
might be called to find it.
Args:
external_identifier (List[Dict]): a list like:
[
{'external-id-relationship': 'SELF',
'external-id-type': 'other-id',
'external-id-url': {'value': 'http://inspireheptest.cern.ch/record/20'},
'external-id-value': '20'
},...
]
putcode: putcode of the given work.
Returns: the Inspire recid mathcing the work.
"""
for identifier in external_identifiers:
id_type = identifier.get("external-id-type")
if not id_type or id_type.lower() != "other-id":
continue
id_url = inspire_service_orcid_utils.smartget(
identifier, "external-id-url.value", "")
if not re.match(r".*inspire.*", id_url, re.I):
continue
id_value = identifier.get("external-id-value")
if not id_value:
continue
# recid found.
return id_value
# The recid was not found in the external_identifiers.
# Thus we call get_bulk_works_details_iter().
putcodes_recid = list(self._get_putcodes_and_recids_iter([putcode]))
if putcodes_recid:
return putcodes_recid[0][1]
class OrcidPutcodeGetter(object):
def __init__(self, orcid, oauth_token):
self.orcid = orcid
self.oauth_token = oauth_token
self.client = OrcidClient(self.oauth_token, self.orcid)
self.source_client_id_path = current_app.config['ORCID_APP_CREDENTIALS'][
'consumer_key']
def get_all_inspire_putcodes_and_recids_iter(self):
"""
Query ORCID api and get all the Inspire putcodes for the given ORCID.
"""
summary_response = self._get_all_works_summary()
# `putcodes_recids` is a list like: [('43326850', 20), ('43255490', None)]
putcodes_recids = list(summary_response.get_putcodes_and_recids_for_source_iter(
self.source_client_id_path))
putcodes_with_recids = [x for x in putcodes_recids if x[1]]
putcodes_without_recids = [x[0] for x in putcodes_recids if not x[1]]
for putcode, recid in putcodes_with_recids:
yield putcode, recid
if not putcodes_without_recids:
return
for putcode, recid in self._get_putcodes_and_recids_iter(putcodes_without_recids):
yield putcode, recid
def _get_all_works_summary(self):
"""
Query ORCID api and get all the putcodes with their embedded recids
for the given ORCID.
An embedded recid is a recid written as external-identifier.
"""
response = self.client.get_all_works_summary()
utils.log_service_response(logger, response, 'in OrcidPutcodeGetter works summary')
try:
response.raise_for_result()
except (orcid_client_exceptions.TokenInvalidException,
orcid_client_exceptions.TokenMismatchException,
orcid_client_exceptions.TokenWithWrongPermissionException):
logger.info('OrcidPutcodeGetter: deleting Orcid push access token={} for orcid={}'.format(
self.oauth_token, self.orcid))
push_access_tokens.delete_access_token(self.oauth_token, self.orcid)
raise exceptions.TokenInvalidDeletedException
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
return response
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 _get_urls_for_putcodes_iter(self, putcodes):
# The call `get_bulk_works_details_iter()` can be expensive for an
# author with many works (if each work also has many *contributors*).
# Fi. for an ATLAS author with ~750 works (each of them with many
# authors), 8 calls would be performed with a total data transfer > 0.5 Gb.
chained = []
for response in self.client.get_bulk_works_details_iter(putcodes):
# Note: this log can be large. Consider removing it when this part
# is considered mature.
utils.log_service_response(logger, response, 'in OrcidPutcodeGetter works details')
try:
response.raise_for_result()
except orcid_client_exceptions.BaseOrcidClientJsonException as exc:
raise exceptions.InputDataInvalidException(from_exc=exc)
chained = itertools.chain(chained, response.get_putcodes_and_urls_iter())
return chained
def get_putcodes_and_recids_by_identifiers_iter(self, identifiers):
"""
Yield putcode and recid for each work matched by the external
identifiers.
Note: external identifiers of type 'other-id' are skipped.
Args:
identifiers (List[inspirehep.modules.orcid.converter.ExternalIdentifier]):
list af all external identifiers added after the xml conversion.
"""
summary_response = self._get_all_works_summary()
for putcode, ids in summary_response.get_putcodes_and_external_identifiers_iter():
# ids is a list like:
# [
# {'external-id-relationship': 'SELF',
# 'external-id-type': 'other-id',
# 'external-id-url': {'value': 'http://inspireheptest.cern.ch/record/20'},
# 'external-id-value': '20'
# },...
# ]
# Get the recid.
recid = self._get_recid_for_work(ids, str(putcode))
for identifier in ids:
id_type = identifier.get('external-id-type')
# We are interested only in doi, arxiv, isbns.
if not id_type or id_type.lower() == 'other-id':
continue
id_value = identifier.get('external-id-value')
if not id_value:
continue
if ExternalIdentifier(id_type, id_value) in identifiers:
yield putcode, recid
def _get_recid_for_work(self, external_identifiers, putcode):
"""
Get the recid for a work given its external identifiers and putcode.
The recid might be in the external identifiers or a get_work_details()
might be called to find it.
Args:
external_identifier (List[Dict]): a list like:
[
{'external-id-relationship': 'SELF',
'external-id-type': 'other-id',
'external-id-url': {'value': 'http://inspireheptest.cern.ch/record/20'},
'external-id-value': '20'
},...
]
putcode: putcode of the given work.
Returns: the Inspire recid mathcing the work.
"""
for identifier in external_identifiers:
id_type = identifier.get('external-id-type')
if not id_type or id_type.lower() != 'other-id':
continue
id_url = inspire_service_orcid_utils.smartget(identifier, 'external-id-url.value', '')
if not re.match(r'.*inspire.*', id_url, re.I):
continue
id_value = identifier.get('external-id-value')
if not id_value:
continue
# recid found.
return id_value
# The recid was not found in the external_identifiers.
# Thus we call get_bulk_works_details_iter().
putcodes_recid = list(self._get_putcodes_and_recids_iter([putcode]))
if putcodes_recid:
return putcodes_recid[0][1]
def test_invalid_token(self):
client = OrcidClient('invalidtoken', self.orcid)
response = client.get_all_works_summary()
with pytest.raises(exceptions.TokenInvalidException):
response.raise_for_result()
assert not response.ok
