def test_get_entities_from_tags_where_sys_entity_between_entities(
kwik_e_mart_nlp, query, tags):
"""Tests the behavior when a system entity is between two entities"""
processed_query = kwik_e_mart_nlp.create_query(query)
res_entity = taggers.get_entities_from_tags(processed_query, tags)
assert len(res_entity) == 4
def test_get_entities_from_tags_where_entity_truncated_by_new_entity(
kwik_e_mart_nlp, query, tags):
"""Test the behavior when a new entity is directly after another entity"""
processed_query = kwik_e_mart_nlp.create_query(query)
res_entity = taggers.get_entities_from_tags(processed_query, tags)
assert len(res_entity) == 2
def test_get_entities_from_tags_with_multi_token_entities(
kwik_e_mart_nlp, query, tags):
"""Tests the behavior with multi token entities"""
processed_query = kwik_e_mart_nlp.create_query(query)
res_entity = taggers.get_entities_from_tags(processed_query, tags)
assert len(res_entity) == 2
def test_get_entities_from_tags_where_entities_end_with_query_end(
kwik_e_mart_nlp, query, tags):
"""Tests the behavior when the entity is at the end of a query"""
processed_query = kwik_e_mart_nlp.create_query(query)
res_entity = taggers.get_entities_from_tags(processed_query, tags)
assert len(res_entity) == 1
def test_get_entities_from_tags_where_tag_idx_not_in_sys_candidate(
kwik_e_mart_nlp, query, tags):
"""Tests the behavior when the system entity tag index is outside
the system candidates spans"""
processed_query = kwik_e_mart_nlp.create_query(query)
res_entity = taggers.get_entities_from_tags(processed_query, tags)
assert res_entity == ()
def test_get_entities_from_tags_where_tag_idx_in_sys_candidate(
kwik_e_mart_nlp, query, tags, expected_time):
"""Tests the behavior when the system entity tag index is
within the system candidates spans"""
processed_query = kwik_e_mart_nlp.create_query(query, time_zone="UTC")
res_entity = taggers.get_entities_from_tags(processed_query, tags)
if res_entity[0].to_dict()["value"]["grain"] == "minute":
assert res_entity[0].to_dict(
)["value"]["value"][MINUTE_GRAIN_INDEX:] in set(expected_time)
else:
assert res_entity[0].to_dict()["value"]["value"] in set(expected_time)
