def test_raise_missing_info(app):
"""Check that the proper exception is raised if name or url is missing."""
from invenio_oaiharvester.errors import NameOrUrlMissing
with app.app_context():
with pytest.raises(NameOrUrlMissing):
list_records()
with pytest.raises(NameOrUrlMissing):
get_records([])
def test_raise_missing_info(app):
"""Check that the proper exception is raised if name or url is missing."""
from invenio_oaiharvester.errors import NameOrUrlMissing
with app.app_context():
with pytest.raises(NameOrUrlMissing):
list_records()
with pytest.raises(NameOrUrlMissing):
get_records([])
def test_raise_wrong_date(app):
"""Check harvesting of records from multiple setspecs."""
with app.app_context():
with pytest.raises(WrongDateCombination):
list_records(metadata_prefix='arXiv',
from_date='2015-01-18',
until_date='2015-01-17',
url='http://export.arxiv.org/oai2',
name=None,
setspecs='physics:hep-lat')
def test_raise_wrong_date(app):
"""Check harvesting of records from multiple setspecs."""
with app.app_context():
with pytest.raises(WrongDateCombination):
list_records(
metadata_prefix='arXiv',
from_date='2015-01-18',
until_date='2015-01-17',
url='http://export.arxiv.org/oai2',
name=None,
setspecs='physics:hep-lat'
)
def test_list_records(app, sample_list_xml, sample_list_xml_cs):
"""Check harvesting of records from multiple setspecs."""
responses.add(
responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=cs.*'),
body=sample_list_xml_cs,
content_type='text/xml'
)
responses.add(
responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=physics.*'),
body=sample_list_xml,
content_type='text/xml'
)
with app.app_context():
_, records = list_records(
metadata_prefix='arXiv',
from_date='2015-01-15',
until_date='2015-01-20',
url='http://export.arxiv.org/oai2',
name=None,
setspecs='cs physics'
)
# 46 cs + 150 physics - 6 dupes == 190
assert len(records) == 190
def test_model_based_harvesting_list(app, sample_config, sample_list_xml):
"""Test harvesting using model."""
from invenio_oaiharvester.utils import get_oaiharvest_object
responses.add(responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=physics.*'),
body=sample_list_xml,
content_type='text/xml')
with app.app_context():
source = get_oaiharvest_object(sample_config)
last_updated = source.lastrun
time.sleep(0.1) # to allow for date checking to work better
_, records = list_records(name=sample_config)
assert len(records) == 150
assert last_updated < get_oaiharvest_object(sample_config).lastrun
def test_list_no_records(app, sample_empty_set):
"""Check harvesting of records from multiple setspecs."""
responses.add(responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=physics.*'),
body=sample_empty_set,
content_type='text/xml')
with app.app_context():
_, records = list_records(metadata_prefix='arXiv',
from_date='2015-01-17',
until_date='2015-01-17',
url='http://export.arxiv.org/oai2',
name=None,
setspecs='physics:hep-lat')
assert not records
def test_model_based_harvesting_list(app, sample_config, sample_list_xml):
"""Test harvesting using model."""
from invenio_oaiharvester.utils import get_oaiharvest_object
responses.add(
responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=physics.*'),
body=sample_list_xml,
content_type='text/xml'
)
with app.app_context():
source = get_oaiharvest_object(sample_config)
last_updated = source.lastrun
time.sleep(0.1) # to allow for date checking to work better
_, records = list_records(name=sample_config)
assert len(records) == 150
assert last_updated < get_oaiharvest_object(sample_config).lastrun
def test_list_no_records(app, sample_empty_set):
"""Check harvesting of records from multiple setspecs."""
responses.add(
responses.GET,
re.compile(r'http?://export.arxiv.org/oai2.*set=physics.*'),
body=sample_empty_set,
content_type='text/xml'
)
with app.app_context():
_, records = list_records(
metadata_prefix='arXiv',
from_date='2015-01-17',
until_date='2015-01-17',
url='http://export.arxiv.org/oai2',
name=None,
setspecs='physics:hep-lat'
)
assert not records
