def test_record2marcxml_handles_repeated_subfields():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'authors': [
{
'affiliations': [
{'value': 'SISSA, Trieste'},
{'value': 'Meudon Observ.'},
],
'full_name': 'Puy, Denis',
},
],
}
expected = (
b'<record>\n'
b' <datafield tag="100" ind1=" " ind2=" ">\n'
b' <subfield code="a">Puy, Denis</subfield>\n'
b' <subfield code="u">SISSA, Trieste</subfield>\n'
b' <subfield code="u">Meudon Observ.</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_oai_get_single_identifier_for_arxiv_set(inspire_app,
celery_app_with_context,
celery_session_worker):
data = {
"arxiv_eprints": [{
"value": "2009.01484"
}],
"report_numbers": [{
"value": "CERN-TH-2020-136"
}],
}
record_data = faker.record("lit", data)
record = LiteratureRecord.create(record_data)
record_marcxml = record2marcxml(record)
db.session.commit()
set_name = inspire_app.config["OAI_SET_CERN_ARXIV"]
oaiset = OAISet(spec=f"{set_name}", name="Test", description="Test")
db.session.add(oaiset)
db.session.commit()
sleep(2)
with inspire_app.test_client() as client:
response = client.get(
f"/api/oai2d?verb=GetRecord&metadataPrefix=marcxml&identifier=oai:inspirehep.net:{record['control_number']}"
)
assert record_marcxml in response.data
def test_oai_get_single_identifier_for_CDS_set(inspire_app, clean_celery_session):
data = {"_export_to": {"CDS": True}}
record_data = faker.record("lit", data)
record = LiteratureRecord.create(record_data)
record_uuid = record.id
record_marcxml = record2marcxml(record)
db.session.commit()
def assert_the_record_is_indexed():
current_search.flush_and_refresh("*")
result = es_search("records-hep")
uuids = get_value(result, "hits.hits._id")
assert str(record_uuid) in uuids
retry_until_pass(assert_the_record_is_indexed)
set_name = inspire_app.config["OAI_SET_CDS"]
oaiset = OAISet(spec=f"{set_name}", name="Test", description="Test")
db.session.add(oaiset)
db.session.commit()
with inspire_app.test_client() as client:
response = client.get(
f"/api/oai2d?verb=GetRecord&metadataPrefix=marcxml&identifier=oai:inspirehep.net:{record['control_number']}"
)
assert record_marcxml in response.data
def test_oai_with_for_arxiv_set(inspire_app, clean_celery_session):
data = {
"arxiv_eprints": [{"value": "2009.01484"}],
"report_numbers": [{"value": "CERN-TH-2020-136"}],
}
record_data = faker.record("lit", data)
record = LiteratureRecord.create(record_data)
record_uuid = record.id
record_marcxml = record2marcxml(record)
db.session.commit()
def assert_the_record_is_indexed():
current_search.flush_and_refresh("*")
result = es_search("records-hep")
uuids = get_value(result, "hits.hits._id")
assert str(record_uuid) in uuids
retry_until_pass(assert_the_record_is_indexed)
set_name = inspire_app.config["OAI_SET_CERN_ARXIV"]
oaiset = OAISet(spec=f"{set_name}", name="Test", description="Test")
db.session.add(oaiset)
db.session.commit()
with inspire_app.test_client() as client:
response = client.get(
f"/api/oai2d?verb=ListRecords&metadataPrefix=marcxml&set={set_name}"
)
assert record_marcxml in response.data
def test_record2marcxml_supports_relative_urls():
record = {
'$schema': '/schemas/records/hep.json',
'control_number': 4328,
}
expected = (b'<record>\n'
b' <controlfield tag="001">4328</controlfield>\n'
b'</record>\n')
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_supports_authors():
record = {
'$schema': 'http://localhost:5000/schemas/records/authors.json',
'control_number': 1010819,
}
expected = (b'<record>\n'
b' <controlfield tag="001">1010819</controlfield>\n'
b'</record>\n')
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_generates_controlfields():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'control_number': 4328,
}
expected = (b'<record>\n'
b' <controlfield tag="001">4328</controlfield>\n'
b'</record>\n')
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_handles_numbers():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'publication_info': [
{'year': 1975},
],
}
expected = (
b'<record>\n'
b' <datafield tag="773" ind1=" " ind2=" ">\n'
b' <subfield code="y">1975</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_handles_unicode():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'authors': [
{'full_name': u'Kätlne, J.'},
],
}
expected = (
b'<record>\n'
b' <datafield tag="100" ind1=" " ind2=" ">\n'
b' <subfield code="a">K\xc3\xa4tlne, J.</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_generates_datafields():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'authors': [
{'full_name': 'Glashow, S.L.'},
],
}
expected = (
b'<record>\n'
b' <datafield tag="100" ind1=" " ind2=" ">\n'
b' <subfield code="a">Glashow, S.L.</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_generates_indices():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'inspire_categories': [
{'term': 'Accelerators'},
],
}
expected = (
b'<record>\n'
b' <datafield tag="650" ind1="1" ind2="7">\n'
b' <subfield code="2">INSPIRE</subfield>\n'
b' <subfield code="a">Accelerators</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_handles_repeated_fields():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'_collections': [
'Literature',
'HAL Hidden',
],
}
expected = (b'<record>\n'
b' <datafield tag="980" ind1=" " ind2=" ">\n'
b' <subfield code="a">HEP</subfield>\n'
b' </datafield>\n'
b' <datafield tag="980" ind1=" " ind2=" ">\n'
b' <subfield code="a">HALhidden</subfield>\n'
b' </datafield>\n'
b'</record>\n')
result = record2marcxml(record)
assert expected == result
def test_oai_with_for_cds_set(inspire_app, celery_app_with_context,
celery_session_worker):
data = {"_export_to": {"CDS": True}}
record_data = faker.record("lit", data)
record = LiteratureRecord.create(record_data)
record_marcxml = record2marcxml(record)
db.session.commit()
set_name = inspire_app.config["OAI_SET_CDS"]
oaiset = OAISet(spec=f"{set_name}", name="Test", description="Test")
db.session.add(oaiset)
db.session.commit()
sleep(2)
with inspire_app.test_client() as client:
response = client.get(
f"/api/oai2d?verb=ListRecords&metadataPrefix=marcxml&set={set_name}"
)
assert record_marcxml in response.data
def test_record2marcxml_strips_control_characters():
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'abstracts': [
{
'source': 'submitter',
'value': u'A common feature shared by many quantum gravity models is modi\u001Ccations of two-point functions at energy scales around the Planck scale.',
},
],
} # holdingpen/812647
expected = (
b'<record>\n'
b' <datafield tag="520" ind1=" " ind2=" ">\n'
b' <subfield code="9">submitter</subfield>\n'
b' <subfield code="a">A common feature shared by many quantum gravity models is modications of two-point functions at energy scales around the Planck scale.</subfield>\n'
b' </datafield>\n'
b'</record>\n'
)
result = record2marcxml(record)
assert expected == result
def test_record2marcxml_raises_when_rules_were_not_implemented():
record = {'$schema': 'http://localhost:5000/schemas/records/data.json'}
with pytest.raises(NotImplementedError) as excinfo:
record2marcxml(record)
assert 'missing' in str(excinfo.value)
def record_json_to_marcxml(pid, record, **kwargs):
"""Converts record to marcxml for OAI."""
return ET.fromstring(record2marcxml(record["_source"]))
