def test_filteroverdo_handles_exceptions_in_non_dicts():
record = {
'$schema': 'hep.json',
'titles': None,
} # synthetic data
with pytest.raises(DoJsonError) as exc:
record2marcxml(record)
assert 'Error in rule "246" for field "titles"' in str(exc.value)
def dumps_etree(pid, record, **kwargs):
"""Dump MARC21 compatible record.
:param pid: The :class:`invenio_pidstore.models.PersistentIdentifier`
instance.
:param record: The :class:`invenio_records.api.Record` instance.
:returns: A LXML Element instance.
"""
r = record['_source']
# adding legacy version (controlfield 005)
acquisition_date = parse_date(r['acquisition_source']['date'])
r['legacy_version'] = acquisition_date.strftime("%Y%m%d%H%M%S.0")
# adding number of pages (datafield 300)
page_nr = get_value(r, 'page_nr[0]')
if page_nr:
r['number_of_pages'] = page_nr
# create and add download url
if 'urls' not in r and '_files' in r:
files = []
for f in r['_files']:
url = 'http://%s/api/files/%s/%s' % (current_app.config.get('SERVER_NAME'), f['bucket'], f['key'])
files.append({
'value': url,
'description': f.get('filetype', '')
})
r['urls'] = files
return etree.fromstring(record2marcxml(r))
def _send_robotupload(obj, eng):
combined_callback_url = ''
if callback_url:
combined_callback_url = os.path.join(
current_app.config["SERVER_NAME"], callback_url)
if not combined_callback_url.startswith('http'):
combined_callback_url = "https://{0}".format(
combined_callback_url)
if extra_data_key is not None:
data = obj.extra_data.get(extra_data_key) or {}
else:
data = obj.data
marcxml = record2marcxml(data)
if current_app.debug:
# Log what we are sending
LOGGER.debug(
"Going to robotupload mode:%s to url:%s:\n%s\n",
mode,
url,
marcxml,
)
if not in_production_mode():
obj.log.debug(
"Going to robotupload %s to %s:\n%s\n",
mode,
url,
marcxml,
)
obj.log.debug("Base object data:\n%s", pformat(data))
return
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=combined_callback_url,
mode=mode,
nonce=obj.id,
priority=5,
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error(
"Your IP is not in "
"app.config_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host: %s", result.text)
txt = "Error while submitting robotupload: {0}".format(result.text)
raise Exception(txt)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
if callback_url:
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def serialize_search(self,
pid_fetcher,
search_result,
links=None,
item_links_factory=None):
"""Serialize a search result as MARCXML."""
result = [
record2marcxml(el['_source'])
for el in search_result['hits']['hits']
]
return MARCXML_TEMPLATE.format(''.join(result))
def serialize_search(self, pid_fetcher, search_result, links=None, item_links_factory=None):
"""Serialize a search result as MARCXML."""
result = [record2marcxml(el['_source']) for el in search_result['hits']['hits']]
return MARCXML_TEMPLATE.format(''.join(result))
def serialize(self, pid, record, links_factory=None):
"""Serialize a single record as MARCXML."""
return MARCXML_TEMPLATE.format(record2marcxml(record))
def _send_robotupload(obj, eng):
is_update = obj.extra_data.get('is-update')
is_authors = eng.workflow_definition.data_type == 'authors'
if not is_authors and is_update and not current_app.config.get(
'FEATURE_FLAG_ENABLE_UPDATE_TO_LEGACY', False):
obj.log.info(
'skipping upload to legacy, feature flag ``FEATURE_FLAG_ENABLE_UPDATE_TO_LEGACY`` is disabled.'
)
return
combined_callback_url = ''
if callback_url:
combined_callback_url = os.path.join(
current_app.config["SERVER_NAME"], callback_url)
if not combined_callback_url.startswith('http'):
combined_callback_url = "https://{0}".format(
combined_callback_url)
if extra_data_key is not None:
data = obj.extra_data.get(extra_data_key) or {}
else:
data = obj.data
if not current_app.config.get(
'FEATURE_FLAG_ENABLE_SENDING_REFERENCES_TO_LEGACY'):
data = copy(data)
data.pop('references', None)
marcxml = record2marcxml(data)
if current_app.debug:
# Log what we are sending
LOGGER.debug(
"Going to robotupload mode:%s to url:%s:\n%s\n",
mode,
url,
marcxml,
)
if not in_production_mode():
obj.log.debug(
"Going to robotupload %s to %s:\n%s\n",
mode,
url,
marcxml,
)
obj.log.debug("Base object data:\n%s", pformat(data))
return
result = make_robotupload_marcxml(
url=url,
marcxml=marcxml,
callback_url=combined_callback_url,
mode=mode,
nonce=obj.id,
priority=5,
)
if "[INFO]" not in result.text:
if "cannot use the service" in result.text:
# IP not in the list
obj.log.error(
"Your IP is not in "
"app.config_BATCHUPLOADER_WEB_ROBOT_RIGHTS "
"on host: %s", result.text)
txt = "Error while submitting robotupload: {0}".format(result.text)
raise Exception(txt)
else:
obj.log.info("Robotupload sent!")
obj.log.info(result.text)
if callback_url:
eng.halt("Waiting for robotupload: {0}".format(result.text))
obj.log.info("end of upload")
def serialize(self, pid, record, links_factory=None):
"""Serialize a single record as MARCXML."""
return MARCXML_TEMPLATE.format(record2marcxml(record))
