def test_raise_ioerror_on_503(self, mock_get):
""":func:`.metadata.retrieve` raises IOError when unvailable."""
response = mock.MagicMock()
type(response).json = mock.MagicMock(return_value=None)
response.status_code = 503
mock_get.return_value = response
with self.assertRaises(IOError):
metadata.retrieve('1234.5678v3')
def test_raise_ioerror_on_sslerror(self, mock_get):
""":func:`.metadata.retrieve` raises IOError when SSL fails."""
from requests.exceptions import SSLError
mock_get.side_effect = SSLError
with self.assertRaises(IOError):
try:
metadata.retrieve('1234.5678v3')
except Exception as e:
if type(e) is SSLError:
self.fail('Should not return dependency exception')
raise
def _get_metadata(self, arxiv_id: str) -> DocMeta:
"""
Retrieve metadata from the :mod:`.metadata` service.
Parameters
----------
arxiv_id : str
Am arXiv identifier, with or without a version affix.
Returns
-------
:class:`.DocMeta`
Metadata for the arXiv paper.
Raises
------
DocumentFailed
Indexing of the document failed. This may have no bearing on the
success of subsequent papers.
IndexingFailed
Indexing of the document failed in a way that indicates recovery
is unlikely for subsequent papers.
"""
logger.debug(f'{arxiv_id}: get metadata')
try:
logger.debug(f'{arxiv_id}: requesting metadata')
docmeta: DocMeta = metadata.retrieve(arxiv_id)
except metadata.ConnectionFailed as e:
# The metadata service will retry bad responses, but not connection
# errors. Sometimes it just takes another try, so why not.
logger.warning(f'{arxiv_id}: first attempt failed, retrying')
try:
docmeta = metadata.retrieve(arxiv_id)
except metadata.ConnectionFailed as e:
# Things really are looking bad. There is no need to keep
# trying with subsequent records, so let's abort entirely.
logger.error(f'{arxiv_id}: second attempt failed, giving up')
raise IndexingFailed(
'Indexing failed; metadata endpoint could not be reached.'
) from e
except metadata.RequestFailed as e:
logger.error(f'{arxiv_id}: request failed')
raise DocumentFailed('Request to metadata service failed') from e
except metadata.BadResponse as e:
logger.error(f'{arxiv_id}: bad response from metadata service')
raise DocumentFailed('Bad response from metadata service') from e
except Exception as e:
logger.error(f'{arxiv_id}: unhandled error, metadata service: {e}')
raise IndexingFailed('Unhandled exception') from e
return docmeta
def test_response_is_not_json(self, mock_get):
""":func:`.metadata.retrieve` raises IOError when not valid JSON."""
from json.decoder import JSONDecodeError
response = mock.MagicMock()
# Ideally we would pass the exception itself as a side_effect, but it
# doesn't have the expected signature.
def raise_decodeerror(*args, **kwargs):
raise JSONDecodeError('Nope', 'Nope', 0)
type(response).json = mock.MagicMock(side_effect=raise_decodeerror)
response.status_code = 200
mock_get.return_value = response
with self.assertRaises(IOError):
metadata.retrieve('1234.5678v3')