def test_index_record(inspire_app, celery_app_with_context,
celery_session_worker):
models_committed.disconnect(index_after_commit)
records = [
create_record_async("lit"),
create_record_async("aut"),
create_record_async("job"),
create_record_async("jou"),
create_record_async("exp"),
create_record_async("con"),
create_record_async("dat"),
create_record_async("ins"),
]
uuids = [record.id for record in records]
task = index_records.delay(uuids)
results = task.get(timeout=5)
uuids = [str(uuid) for uuid in uuids]
assert results == uuids
for record in records:
result = InspireSearch.get_record_data_from_es(record)
assert record["control_number"] == result["control_number"]
models_committed.connect(index_after_commit)
def test_match_references(inspire_app, cli, clean_celery_session):
cited_data = {
"document_type": ["article"],
"dois": [{
"value": "10.1371/journal.pone.0188398"
}],
}
cited_record = create_record_async("lit", data=cited_data)
cited_record.index(
delay=False) # reference-matcher requires cited record to be indexed
citer_data = {
"references": [{
"reference": {
"dois": ["10.1371/journal.pone.0188398"]
}
}]
}
citer_record_1 = create_record_async("lit", data=citer_data)
citer_record_2 = create_record_async("lit", data=citer_data)
citer_record_3 = create_record_async("lit", data=citer_data)
citer_ids = [citer_record_1.id, citer_record_2.id, citer_record_3.id]
record_data = create_record_async("dat")
record_data_uuids = record_data.id
def assert_all_records_are_indexed():
current_search.flush_and_refresh("*")
result = es_search("records-hep")
uuids = get_value(result, "hits.hits._id")
for uuid in citer_ids:
assert str(uuid) in uuids
result = es_search("records-data")
uuids = get_value(result, "hits.hits._id")
assert str(record_data_uuids) in uuids
retry_until_pass(assert_all_records_are_indexed)
result = cli.invoke(["match", "references", "-bs", 2])
assert result.exit_code == 0
for citer_id in citer_ids:
updated_citer_record = LiteratureRecord.get_record(citer_id)
assert (get_value(updated_citer_record,
"references[0].record") == cited_record["self"])
def test_match_references(inspire_app, cli, celery_app_with_context,
celery_session_worker):
cited_data = {
"document_type": ["article"],
"dois": [{
"value": "10.1371/journal.pone.0188398"
}],
}
cited_record = create_record_async("lit", data=cited_data)
cited_record.index(
delay=False) # reference-matcher requires cited record to be indexed
citer_data = {
"references": [{
"reference": {
"dois": ["10.1371/journal.pone.0188398"]
}
}]
}
citer_record_1 = create_record_async("lit", data=citer_data)
citer_record_2 = create_record_async("lit", data=citer_data)
citer_record_3 = create_record_async("lit", data=citer_data)
citer_ids = [citer_record_1.id, citer_record_2.id, citer_record_3.id]
create_record_async("dat")
result = cli.invoke(["match", "references", "-bs", 2])
assert result.exit_code == 0
for citer_id in citer_ids:
updated_citer_record = LiteratureRecord.get_record(citer_id)
assert (get_value(updated_citer_record,
"references[0].record") == cited_record["self"])
def test_process_references_in_records_with_different_type_of_records_doesnt_throw_an_exception(
inspire_app, celery_app_with_context, celery_session_worker):
# disconnect this signal so records don't get indexed
models_committed.disconnect(index_after_commit)
cited_record_1 = LiteratureRecord.create(faker.record("lit"))
cited_record_2 = LiteratureRecord.create(faker.record("lit"))
data_citing_record_1 = faker.record(
"lit", literature_citations=[cited_record_1["control_number"]])
citing_record_1 = LiteratureRecord.create(data_citing_record_1)
data_citing_record_2 = faker.record(
"lit", literature_citations=[cited_record_2["control_number"]])
citing_record_2 = LiteratureRecord.create(data_citing_record_2)
db.session.commit()
records = [
create_record_async("aut"),
create_record_async("job"),
create_record_async("jou"),
create_record_async("exp"),
create_record_async("con"),
create_record_async("dat"),
create_record_async("ins"),
]
# reconnect signal before we call process_references_in_records
models_committed.connect(index_after_commit)
uuids = [record.id
for record in records] + [citing_record_1.id, citing_record_2.id]
task = process_references_in_records.delay(uuids)
results = task.get(timeout=5)
uuids = [str(uuid) for uuid in uuids]
assert results == uuids
result_cited_record_1 = InspireSearch.get_record_data_from_es(
cited_record_1)
expected_result_cited_record_1_citation_count = 1
assert (expected_result_cited_record_1_citation_count ==
result_cited_record_1["citation_count"])
result_cited_record_2 = InspireSearch.get_record_data_from_es(
cited_record_2)
expected_result_cited_record_2_citation_count = 1
assert (expected_result_cited_record_2_citation_count ==
result_cited_record_2["citation_count"])
def record_with_two_revisions(inspire_app, clean_celery_session):
record_data = {
"$schema": "http://localhost:5000/schemas/records/hep.json",
"control_number": 111,
"document_type": ["article"],
"titles": [{"title": "record rev0"}],
"self": {"$ref": "http://localhost:5000/api/literature/1243"},
"_collections": ["Literature"],
}
record = create_record_async("lit", data=record_data)
record_data["titles"][0]["title"] = "record rev1"
record.update(record_data)
db.session.commit()
def test_index_record_deletes_a_deleted_record(inspire_app,
clean_celery_session):
record_to_delete = create_record_async("lit")
record_to_delete_control_number = record_to_delete["control_number"]
record_to_delete = InspireRecord.get_record_by_pid_value(
record_to_delete_control_number, "lit")
record_to_delete.delete()
db.session.commit()
uuids = [record_to_delete.id]
task = index_records.delay(uuids)
results = task.get(timeout=5)
uuids = [str(uuid) for uuid in uuids]
assert results == uuids
with pytest.raises(TransportError):
InspireSearch.get_record_data_from_es(record_to_delete)
def test_update_relations_recalculate_citations_with_different_type_of_records_doesnt_throw_an_exception(
inspire_app, celery_app_with_context, celery_session_worker):
data_cited = faker.record("lit", with_control_number=True)
record_cited = InspireRecord.create(data_cited,
disable_relations_update=True)
db.session.commit()
record_cited_control_number = record_cited["control_number"]
data_citing = faker.record(
"lit",
literature_citations=[record_cited_control_number],
with_control_number=True,
)
record_citing = InspireRecord.create(data_citing,
disable_relations_update=True)
db.session.commit()
records = [
create_record_async("aut"),
create_record_async("job"),
create_record_async("jou"),
create_record_async("exp"),
create_record_async("con"),
create_record_async("dat"),
create_record_async("ins"),
]
uuids = [record.id
for record in records] + [record_cited.id, record_citing.id]
task = update_relations.delay(uuids)
results = task.get(timeout=5)
uuids = [str(uuid) for uuid in uuids]
assert results == uuids
result_record_cited = RecordCitations.query.filter_by(
cited_id=record_cited.id).one()
assert record_citing.id == result_record_cited.citer_id
record_cited = InspireRecord.get_record_by_pid_value(
record_cited_control_number, "lit")
record_cited_citation_count = 1
assert record_cited_citation_count == record_cited.citation_count