def test_reindex_only_one_record(inspire_app, clean_celery_session, cli):
generate_records(count=1, data={"control_number": 3})
result = cli.invoke(["index", "reindex", "-id", "lit", "3", "-q", ""])
expected_message = "Successfully reindexed record ('lit', '3')"
assert result.exit_code == 0
assert expected_message in result.output
def test_reindex_record_lit_fails_with_invalid_field_content(
inspire_app, clean_celery_session, cli):
invalid_field = {"keywords": ["17B05"]}
with patch("inspirehep.indexer.base.InspireRecordIndexer"):
with patch("inspirehep.records.api.base.schema_validate"):
generate_records(count=1, data=invalid_field, skip_validation=True)
check_n_records_reindex_for_pidtype(inspire_app, cli, "lit", n_fail=1)
def test_reindex_record_lit_fails_with_invalid_record(inspire_app,
clean_celery_session,
cli):
broken_field = {"_desy_bookkeeping": {"date": '"2013-01-14_final'}}
with patch("inspirehep.indexer.base.InspireRecordIndexer"):
with patch("inspirehep.records.api.base.schema_validate"):
generate_records(count=1, data=broken_field, skip_validation=True)
check_n_records_reindex_for_pidtype(inspire_app, cli, "lit", n_fail=1)
def test_reindex_records_lit_using_multiple_batches(inspire_app,
clean_celery_session, cli):
generate_records(count=5)
generate_records(data={"deleted": True})
result = cli.invoke(
["index", "reindex", "-p", "lit", "-bs", "5", "-q", ""])
assert result.exit_code == 0
assert "5 succeeded" in result.output
assert "0 failed" in result.output
def test_get_query_records_to_index_indexes_deleted_record_too(
inspire_app, clean_celery_session):
generate_records(count=1)
generate_records(count=1, data={"deleted": True})
pids = ["lit"]
query = get_query_records_to_index(pids)
expected_count = 2
result_count = query.count()
assert result_count == expected_count
def test_reindex_records_lit_one_fails_and_two_ok(
inspire_app, clean_celery_session, cli
):
invalid_field = {"keywords": ["17B05"]}
generate_records(count=2)
with patch("inspirehep.indexer.base.InspireRecordIndexer"):
with patch("inspirehep.records.api.base.get_validation_errors"):
generate_records(count=1, data=invalid_field, skip_validation=True)
check_n_records_reindex_for_pidtype(inspire_app, cli, "lit", n_success=2, n_fail=1)
def test_get_query_records_to_index_only_lit(inspire_app, clean_celery_session):
generate_records(count=2)
generate_records(count=2, record_type=AuthorsRecord)
pids = ["lit"]
query = get_query_records_to_index(pids)
expected_count = 2
result_count = query.count()
assert result_count == expected_count
uuids = [str(item[0]) for item in query.all()]
deleted_records = _get_deleted_records_by_uuids(uuids)
assert deleted_records == []
def test_get_query_records_to_index_ok_different_pids(inspire_app,
celery_app_with_context,
celery_session_worker):
generate_records(count=2)
generate_records(count=2, record_type=AuthorsRecord)
pids = ["lit", "aut"]
query = get_query_records_to_index(pids)
expected_count = 4
result_count = query.count()
assert result_count == expected_count
uuids = [str(item[0]) for item in query.all()]
deleted_records = _get_deleted_records_by_uuids(uuids)
assert deleted_records == []
def test_reindex_records_data_records(inspire_app, celery_app_with_context,
celery_session_worker, cli):
generate_records(count=3, record_type=DataRecord)
check_n_records_reindex_for_pidtype(inspire_app, cli, "dat", n_success=3)
def test_reindex_records_data_records(inspire_app, clean_celery_session, cli):
generate_records(count=3, record_type=DataRecord)
check_n_records_reindex_for_pidtype(inspire_app, cli, "dat", n_success=3)
