예제 #1
0
def test_reduce_entities():
    entities_cleaned = preprocess.entity_cleaning(input_entities)
    entities_with_page_number = preprocess.replace_page_number(
        entities_cleaned, "9")
    mapped_entities = preprocess.map_entities(entities_with_page_number)
    reduced_entities = preprocess.reduce_entities(mapped_entities)
    assert reduced_entities == reduce_
예제 #2
0
def test_get_entity_type():
    entities_to_assert = []
    entity_types = [
        "YEAR",
        "PERSON",
        "LOCATION",
        "ORGANIZATION",
        "COMMERCIAL_ITEM",
        "EVENT",
        "TITLE",
        "DATE",
        "QUANTITY",
        "OTHER",
        "ZIPCODE",
        "STATE",
        "ROUTE",
        "CITY",
    ]
    entities_cleaned = preprocess.entity_cleaning(input_entities)
    entities_with_page_number = preprocess.replace_page_number(
        entities_cleaned, "9")
    mapped_entities = preprocess.map_entities(entities_with_page_number)
    reduced_entities = preprocess.reduce_entities(mapped_entities)
    for entity in reduced_entities.values():
        entities_to_assert.append(
            formatter.get_entity_type(entity["entity_type"], entity))
    for i in entities_to_assert:
        assert i in entity_types
예제 #3
0
def validate(in_req, out_exp):
    cleaned = preprocess.entity_cleaning(in_req)
    with_page_number = preprocess.replace_page_number(cleaned, "9")
    mapped_entities = preprocess.map_entities(with_page_number)
    reduced_entities = preprocess.reduce_entities(mapped_entities)
    groupped_entities = formatter.group_entities(reduced_entities)
    translated_entities = formatter.translate_entities(groupped_entities, "9")
    if "PERSON NAME" in translated_entities.keys():
        translated_entities = formatter.add_full_name_entities(
            translated_entities)
    assert translated_entities == out_exp
예제 #4
0
def test_entity_aggregation():
    cleaned = preprocess.entity_cleaning(input_entities)
    with_page_number = preprocess.replace_page_number(cleaned, "9")
    mapped_entities = preprocess.map_entities(with_page_number)
    assert mapped_entities == aggregate_by_id
    reduced_entities = preprocess.reduce_entities(mapped_entities)
    assert reduced_entities == reduce_
    groupped_entities = formatter.group_entities(reduced_entities)
    assert groupped_entities == group_by_type
    translate_entities = formatter.translate_entities(groupped_entities, "9")
    assert translate_entities == output_aggregation
예제 #5
0
def format_entities(entities, page):
    washed_entities = preprocess.entity_cleaning(entities)
    entities_with_page_number = preprocess.replace_page_number(
        washed_entities, page)
    mapped_entities = preprocess.map_entities(entities_with_page_number)
    reduced_entities = preprocess.reduce_entities(mapped_entities)
    groupped_entities_by_type = group_entities(reduced_entities)
    translated_entities = translate_entities(groupped_entities_by_type, page)
    if "PERSON NAME" in translated_entities.keys():
        translated_entities = add_full_name_entities(translated_entities)
    logger.info(f"Translated Entities = {translated_entities}")
    return {"statusCode": 200, "body": translated_entities}
예제 #6
0
def test_page_number_replace():
    replaced_page_number = preprocess.replace_page_number(
        cleaned_entities_input, "9")
    assert replaced_page_number == input_entities_page_9
    output_page = formatter.format_entities(input_entities, "9")
    assert output_page["body"] == output_page_9["body"]
예제 #7
0
def test_map_entities():
    entities_cleaned = preprocess.entity_cleaning(input_entities)
    entities_with_page_number = preprocess.replace_page_number(
        entities_cleaned, "9")
    mapped_entities = preprocess.map_entities(entities_with_page_number)
    assert mapped_out == mapped_entities