def test_authors_create_with_invalid_data(inspire_app):
data = faker.record("aut", with_control_number=True)
data["invalid_key"] = "should throw an error"
record_control_number = str(data["control_number"])
with pytest.raises(ValidationError):
AuthorsRecord.create(data)
record_pid = PersistentIdentifier.query.filter_by(
pid_value=record_control_number).one_or_none()
assert record_pid is None
def test_authors_create_with_existing_control_number(inspire_app):
data = faker.record("aut", with_control_number=True)
existing_object_uuid = uuid.uuid4()
create_pidstore(
object_uuid=existing_object_uuid,
pid_type="aut",
pid_value=data["control_number"],
)
with pytest.raises(PIDAlreadyExists):
AuthorsRecord.create(data)
def test_aut_record_appear_in_es_when_created(inspire_app,
celery_app_with_context,
celery_session_worker):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
expected_id = str(rec.id)
steps = [
{
"step": current_search.flush_and_refresh,
"args": ["records-authors"]
},
{
"step": es_search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total.value",
"expected_result": 1,
},
},
{
"step": es_search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.hits[0]._id",
"expected_result": expected_id,
},
},
]
retry_until_matched(steps)
def test_aut_record_removed_form_es_when_deleted(inspire_app, clean_celery_session):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
def assert_record():
current_search.flush_and_refresh("records-authors")
result = es_search("records-authors")
result_total = get_value(result, "hits.total.value")
expected_total = 1
assert expected_total == result_total
retry_until_pass(assert_record)
rec.delete()
db.session.commit()
def assert_record():
current_search.flush_and_refresh("records-authors")
result = es_search("records-authors")
result_total = get_value(result, "hits.total.value")
expected_total = 0
assert expected_total == result_total
retry_until_pass(assert_record)
def test_aut_record_update_when_changed(app, celery_app_with_context,
celery_session_worker,
retry_until_matched):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
expected_death_date = "1900-01-01"
data["death_date"] = expected_death_date
rec.update(data)
db.session.commit()
steps = [
{
"step": es.indices.refresh,
"args": ["records-authors"]
},
{
"step": es.search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total",
"expected_result": 1
},
},
]
resp = retry_until_matched(steps)["hits"]["hits"]
assert resp[0]["_source"]["death_date"] == expected_death_date
def test_literature_create_or_update_with_existing_record(inspire_app):
data = faker.record("aut", with_control_number=True)
record = AuthorsRecord.create(data)
assert data["control_number"] == record["control_number"]
data_update = {
"name": {
"name_variants": ["UPDATED"],
"preferred_name": "UPDATED",
"value": "UPDATED",
}
}
data.update(data_update)
record_updated = AuthorsRecord.create_or_update(data)
control_number = str(record_updated["control_number"])
assert record["control_number"] == record_updated["control_number"]
record_updated_db = RecordMetadata.query.filter_by(
id=record_updated.id).one()
assert data == record_updated_db.json
record_updated_pid = PersistentIdentifier.query.filter_by(
pid_type="aut", pid_value=str(control_number)).one()
assert record_updated.model.id == record_updated_pid.object_uuid
assert control_number == record_updated_pid.pid_value
def test_aut_record_update_when_changed(inspire_app, celery_app_with_context,
celery_session_worker):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
expected_death_date = "1900-01-01"
data["death_date"] = expected_death_date
data["control_number"] = rec["control_number"]
rec.update(data)
db.session.commit()
steps = [
{
"step": current_search.flush_and_refresh,
"args": ["records-authors"]
},
{
"step": es_search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total.value",
"expected_result": 1,
},
},
{
"step": es_search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.hits[0]._source.death_date",
"expected_result": expected_death_date,
},
},
]
retry_until_matched(steps)["hits"]["hits"]
def test_session_still_works_after_integrity_error_on_bai_provider(
inspire_app):
data = faker.record("aut")
with override_config(FEATURE_FLAG_ENABLE_BAI_PROVIDER=True):
AuthorsRecord.create(data=data)
db.session.commit()
with mock.patch(
"inspirehep.pidstore.providers.bai.InspireBAIProvider.next_bai_number"
) as next_bai_mock:
next_bai_mock.side_effect = [1, 2]
AuthorsRecord.create(data=data)
assert next_bai_mock.call_count == 2
db.session.commit()
expected_pid_count = 2
assert (PersistentIdentifier.query.filter_by(
pid_type="bai").count() == expected_pid_count)
def test_create_record_from_db_depending_on_its_pid_type(base_app, db, es):
data = faker.record("aut")
record = InspireRecord.create(data)
assert type(record) == AuthorsRecord
assert record.pid_type == "aut"
record = AuthorsRecord.create(data)
assert type(record) == AuthorsRecord
assert record.pid_type == "aut"
def test_create_record_from_db_depending_on_its_pid_type(inspire_app):
data = faker.record("aut")
record = InspireRecord.create(data)
assert isinstance(record, AuthorsRecord)
assert record.pid_type == "aut"
record = AuthorsRecord.create(data)
assert isinstance(record, AuthorsRecord)
assert record.pid_type == "aut"
def test_indexer_updates_advisor_when_student_name_changes(
inspire_app, clean_celery_session
):
advisor_data = faker.record("aut")
advisor = AuthorsRecord.create(advisor_data)
db.session.commit()
current_search.flush_and_refresh("records-authors")
student_data = faker.record(
"aut",
data={
"advisors": [
{
"name": advisor["name"]["value"],
"record": advisor["self"],
"degree_type": "phd",
}
]
},
)
student = AuthorsRecord.create(student_data)
db.session.commit()
def assert_record():
current_search.flush_and_refresh("records-authors")
records_from_es = AuthorsSearch().query_from_iq("").execute()
assert len(records_from_es.hits) == 2
retry_until_pass(assert_record, 3)
student["name"]["preferred_name"] = "Test Student"
student.update(dict(student))
db.session.commit()
expected_student_name = "Test Student"
def assert_record():
current_search.flush_and_refresh("records-authors")
record_from_es = AuthorsSearch().get_record_data_from_es(advisor)
assert record_from_es["students"][0]["name"] == expected_student_name
retry_until_pass(assert_record, retry_interval=3)
def test_regression_get_linked_author_records_uuids_if_author_changed_name_does_not_return_none_for_author_which_name_didnt_change(
app, clean_celery_session
):
author_data = faker.record("aut")
author = AuthorsRecord.create(author_data)
db.session.commit()
data = dict(author)
data["birth_date"] = "1950-01-01"
author.update(data)
db.session.commit()
new_author = AuthorsRecord.get_record_by_pid_value(author["control_number"])
assert set() == new_author.get_linked_author_records_uuids_if_author_changed_name()
def test_get_linked_advisors_when_name_changes(inspire_app):
data_advisor = faker.record("aut")
advisor = AuthorsRecord.create(data_advisor)
db.session.commit()
assert not advisor.get_linked_advisors_when_name_changes()
student_data = faker.record(
"aut",
data={
"advisors": [{
"name": advisor["name"]["value"],
"record": advisor["self"],
"degree_type": "phd",
}]
},
)
student = AuthorsRecord.create(student_data)
db.session.commit()
assert student.get_linked_advisors_when_name_changes() == set(
[str(advisor.id)])
student["ids"] = [{"value": "0000-0002-1558-1309", "schema": "ORCID"}]
student.update(dict(student))
db.session.commit()
assert not student.get_linked_advisors_when_name_changes()
student["name"]["preferred_name"] = "Test Author"
student.update(dict(student))
db.session.commit()
assert student.get_linked_advisors_when_name_changes() == set(
[str(advisor.id)])
del student["name"]["preferred_name"]
student.update(dict(student))
db.session.commit()
assert student.get_linked_advisors_when_name_changes() == set(
[str(advisor.id)])
def test_aut_record_appear_in_es_when_created(inspire_app, clean_celery_session):
data = faker.record("aut")
record = AuthorsRecord.create(data)
db.session.commit()
expected_control_number = record["control_number"]
def assert_record():
current_search.flush_and_refresh("records-authors")
record_from_es = AuthorsSearch().get_record_data_from_es(record)
assert expected_control_number == record_from_es["control_number"]
retry_until_pass(assert_record)
def test_indexer_deletes_record_from_es(inspire_app, datadir):
def assert_record_is_deleted_from_es():
current_search.flush_and_refresh("records-authors")
expected_records_count = 0
record_lit_es = AuthorsSearch().get_record(str(record.id)).execute().hits
assert expected_records_count == len(record_lit_es)
record = AuthorsRecord.create(faker.record("aut"))
db.session.commit()
record.delete()
db.session.commit()
retry_until_pass(assert_record_is_deleted_from_es)
def test_authors_create(inspire_app):
data = faker.record("aut")
record = AuthorsRecord.create(data)
control_number = str(record["control_number"])
record_db = RecordMetadata.query.filter_by(id=record.id).one()
assert record == record_db.json
record_pid = PersistentIdentifier.query.filter_by(
pid_type="aut", pid_value=str(control_number)).one()
assert record.model.id == record_pid.object_uuid
assert control_number == record_pid.pid_value
def test_process_references_in_records_process_author_records(
mock_batch_index, inspire_app, clean_celery_session):
author_record = AuthorsRecord.create(faker.record("aut"))
lit_record = LiteratureRecord.create(
faker.record(
"lit",
data={
"authors": [{
"full_name": author_record["name"]["value"],
"record": author_record["self"],
}]
},
))
lit_record_2 = LiteratureRecord.create(
faker.record(
"lit",
data={
"authors": [{
"full_name": author_record["name"]["value"],
"record": author_record["self"],
}]
},
))
db.session.commit()
def assert_records_in_es():
lit_record_from_es = InspireSearch.get_record_data_from_es(lit_record)
lit_record_from_es_2 = InspireSearch.get_record_data_from_es(
lit_record_2)
aut_record_from_es = InspireSearch.get_record_data_from_es(
author_record)
assert lit_record_from_es and aut_record_from_es and lit_record_from_es_2
retry_until_pass(assert_records_in_es, retry_interval=5)
models_committed.disconnect(index_after_commit)
author_record["name"]["value"] = "Another Name"
author_record.update(dict(author_record))
db.session.commit()
# reconnect signal before we call process_references_in_records
models_committed.connect(index_after_commit)
task = process_references_in_records.delay([author_record.id])
task.get(timeout=5)
assert sorted(mock_batch_index.mock_calls[0][1][0]) == sorted(
[str(lit_record.id), str(lit_record_2.id)])
def test_aut_record_update_when_changed(inspire_app, clean_celery_session):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
expected_death_date = "1900-01-01"
data["death_date"] = expected_death_date
data["control_number"] = rec["control_number"]
rec.update(data)
db.session.commit()
def assert_record():
current_search.flush_and_refresh("records-authors")
record_from_es = AuthorsSearch().get_record_data_from_es(rec)
assert expected_death_date == record_from_es["death_date"]
retry_until_pass(assert_record)
def test_indexer_updates_authors_papers_when_name_changes(
inspire_app, clean_celery_session
):
author_data = faker.record("aut")
author = AuthorsRecord.create(author_data)
db.session.commit()
current_search.flush_and_refresh("records-authors")
author_cn = author["control_number"]
lit_data = {
"authors": [
{
"record": {
"$ref": f"https://labs.inspirehep.net/api/authors/{author_cn}"
},
"full_name": author["name"]["value"],
}
]
}
lit_data = faker.record("lit", data=lit_data)
lit_1 = LiteratureRecord.create(lit_data)
db.session.commit()
expected_facet_author_name = f"{author['control_number']}_{author['name']['value']}"
def assert_record():
current_search.flush_and_refresh("records-hep")
record_from_es = LiteratureSearch().get_record_data_from_es(lit_1)
assert expected_facet_author_name == record_from_es["facet_author_name"][0]
retry_until_pass(assert_record)
data = dict(author)
data["name"]["value"] = "Some other name"
author.update(data)
db.session.commit()
expected_facet_author_name = f"{author['control_number']}_Some other name"
def assert_record():
current_search.flush_and_refresh("records-hep")
record_from_es = LiteratureSearch().get_record_data_from_es(lit_1)
assert expected_facet_author_name == record_from_es["facet_author_name"][0]
retry_until_pass(assert_record)
def test_aut_record_removed_form_es_when_deleted(app, celery_app_with_context,
celery_session_worker,
retry_until_matched):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
steps = [
{
"step": es.indices.refresh,
"args": ["records-authors"]
},
{
"step": es.search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total",
"expected_result": 1
},
},
]
retry_until_matched(steps)
rec.delete()
db.session.commit()
steps = [
{
"step": es.indices.refresh,
"args": ["records-authors"]
},
{
"step": es.search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total",
"expected_result": 0
},
},
]
retry_until_matched(steps)
def test_aut_record_appear_in_es_when_created(app, celery_app_with_context,
celery_session_worker,
retry_until_matched):
data = faker.record("aut")
rec = AuthorsRecord.create(data)
db.session.commit()
steps = [
{
"step": es.indices.refresh,
"args": ["records-authors"]
},
{
"step": es.search,
"args": ["records-authors"],
"expected_result": {
"expected_key": "hits.total",
"expected_result": 1
},
},
]
response = retry_until_matched(steps)
assert response["hits"]["hits"][0]["_id"] == str(rec.id)
def test_indexer_updates_authors_papers_when_name_changes(
inspire_app, celery_app_with_context, celery_session_worker):
author_data = faker.record("aut")
author = AuthorsRecord.create(author_data)
db.session.commit()
current_search.flush_and_refresh("records-authors")
author_cn = author["control_number"]
lit_data = {
"authors": [{
"record": {
"$ref": f"https://labs.inspirehep.net/api/authors/{author_cn}"
},
"full_name": author["name"]["value"],
}]
}
lit_data = faker.record("lit", data=lit_data)
lit_1 = LiteratureRecord.create(lit_data)
db.session.commit()
expected_hits = 1
expected_facet_author_name_count = 1
expected_facet_author_name = f"{author['control_number']}_{author['name']['value']}"
steps = [
{
"step": current_search.flush_and_refresh,
"args": ["*"]
},
{
"step": es_search,
"args": ["records-hep"],
"expected_result": {
"expected_key": "hits.total.value",
"expected_result": expected_hits,
},
},
{
"expected_key": "hits.hits[0]._source.facet_author_name[0]",
"expected_result": expected_facet_author_name,
},
]
results = retry_until_matched(steps, timeout=45)
assert (len(results["hits"]["hits"][0]["_source"]["facet_author_name"]) ==
expected_facet_author_name_count)
data = dict(author)
data["name"]["value"] = "Some other name"
author.update(data)
db.session.commit()
expected_facet_author_name = f"{author['control_number']}_Some other name"
steps = [
{
"step": current_search.flush_and_refresh,
"args": ["*"]
},
{
"step": es_search,
"args": ["records-hep"],
"expected_result": {
"expected_key": "hits.total.value",
"expected_result": expected_hits,
},
},
{
"expected_key": "hits.hits[0]._source.facet_author_name[0]",
"expected_result": expected_facet_author_name,
},
]
results = retry_until_matched(steps, timeout=45)
assert (len(results["hits"]["hits"][0]["_source"]["facet_author_name"]) ==
expected_facet_author_name_count)
def test_if_bai_is_processed_on_authors_record_creation(inspire_app):
data = faker.record("aut", other_pids=["bai"])
bai = data["ids"][0]["value"]
rec = AuthorsRecord.create(data)
assert (PersistentIdentifier.query.filter_by(pid_type="bai",
pid_value=bai).count() == 1)
def test_process_references_in_records_process_self_citations(
mock_batch_index, inspire_app, clean_celery_session,
enable_self_citations):
author_record = AuthorsRecord.create(
faker.record(
"aut",
data={
"name": {
"value": "'t Hooft, Gerardus",
"name_variants": ["'t Hooft, Gerard", "Hooft, Gerard T."],
"preferred_name": "Gerardus 't Hooft",
},
"ids": [
{
"value": "INSPIRE-00060582",
"schema": "INSPIRE ID"
},
{
"value": "G.tHooft.1",
"schema": "INSPIRE BAI"
},
],
},
))
author_record_2 = AuthorsRecord.create(
faker.record(
"aut",
data={
"name": {
"value": "'t Hooft, Gerardus Marcus",
"preferred_name": "Gerardus Marcus 't Hooft",
},
"ids": [
{
"value": "INSPIRE-00060583",
"schema": "INSPIRE ID"
},
{
"value": "G.tHooft.2",
"schema": "INSPIRE BAI"
},
],
},
))
lit_record = LiteratureRecord.create(
faker.record(
"lit",
data={
"authors": [{
"ids": [
{
"value": "INSPIRE-00060582",
"schema": "INSPIRE ID"
},
{
"value": "G.tHooft.1",
"schema": "INSPIRE BAI"
},
],
"full_name":
author_record["name"]["value"],
"record":
author_record["self"],
}]
},
))
lit_record_2 = LiteratureRecord.create(
faker.record(
"lit",
literature_citations=[lit_record["control_number"]],
data={
"authors": [{
"ids": [
{
"value": "INSPIRE-00060583",
"schema": "INSPIRE ID"
},
{
"value": "G.tHooft.2",
"schema": "INSPIRE BAI"
},
],
"full_name":
author_record_2["name"]["value"],
"record":
author_record_2["self"],
}]
},
))
db.session.commit()
def assert_records_in_es():
lit_record_from_es = InspireSearch.get_record_data_from_es(lit_record)
lit_record_from_es_2 = InspireSearch.get_record_data_from_es(
lit_record_2)
aut_record_from_es = InspireSearch.get_record_data_from_es(
author_record)
assert lit_record_from_es and aut_record_from_es and lit_record_from_es_2
retry_until_pass(assert_records_in_es, retry_interval=5)
models_committed.disconnect(index_after_commit)
lit_record["authors"].append({
"ids": [
{
"value": "INSPIRE-00060583",
"schema": "INSPIRE ID"
},
{
"value": "G.tHooft.2",
"schema": "INSPIRE BAI"
},
],
"full_name":
author_record_2["name"]["value"],
"record":
author_record_2["self"],
})
lit_record.update(dict(lit_record))
db.session.commit()
# reconnect signal before we call process_references_in_records
models_committed.connect(index_after_commit)
task = process_references_in_records.delay([lit_record.id])
task.get(timeout=5)
assert sorted(mock_batch_index.mock_calls[0][1][0]) == sorted(
[lit_record_2.id])
def test_disambiguation_races_assign(override_config, inspire_app,
clean_celery_session,
enable_disambiguation):
cataloger = create_user(role="cataloger")
with override_config(FEATURE_FLAG_ENABLE_BAI_PROVIDER=True,
FEATURE_FLAG_ENABLE_BAI_CREATION=True):
author_record_data = faker.record("aut")
author_record_data.update({
"name": {
"value": "Michael F. A'Hearn"
},
"ids": [{
"schema": "INSPIRE BAI",
"value": "M.F.A.Hearn.1"
}],
})
author_record = AuthorsRecord.create(author_record_data)
lit_data = faker.record("lit")
lit_data.update({
"authors": [{
"ids": [{
"value": "M.F.A.Hearn.1",
"schema": "INSPIRE BAI"
}],
"uuid":
"ce061c1e-866a-422d-9982-652183bae814",
"full_name":
"A'Hearn, M.F.",
"signature_block":
"HARNm",
"curated_relation":
True,
"record":
author_record["self"],
}]
})
lit_record = LiteratureRecord.create(lit_data)
db.session.commit()
with inspire_app.test_client() as client:
login_user_via_session(client, email=cataloger.email)
client.post(
"/api/assign/author",
data=orjson.dumps({
"literature_recids": [lit_record["control_number"]],
"from_author_recid":
author_record["control_number"],
}),
content_type="application/json",
)
def assert_disambiguation_on_record_update():
literature_record_from_es = InspireSearch.get_record_data_from_es(
lit_record)
assert (get_values_for_schema(
literature_record_from_es["authors"][0]["ids"],
"INSPIRE BAI")[0] != "M.F.A.Hearn.1")
retry_until_pass(assert_disambiguation_on_record_update,
retry_interval=2)
