poyee-data-warehouse/dw_base/spark/udf/customs/test/clean_crawler_data_test.py

import pytest
from typing import Set
from dw_base.spark.udf.customs.clean_crawler_data import get_regex_match, clean_germany_company_name


@pytest.mark.parametrize("company_name, expected", [
    ('Sécheron SA', {'é'}),
    ('Sécheron SA\\u0022ss',{'é','\\u0022'}),
    ('GEHS GRÜN ENERGİE HEIZUNG UND SANİTÂR',{'Ü','Â'})
])
def test_get_regex_match(company_name: str, expected: Set[str]):
    result = get_regex_match(company_name)
    assert result == expected


@pytest.mark.parametrize("company_name, expected", [
    ('Beiersdorf Indústria Comércio', 'Beiersdorf Indústria Comércio'),
    ('GPS Prüftechnik Rhein/Main GmbH', 'GPS Prüftechnik Rhein/Main GmbH')
])
def test_clean_germany_company_name(company_name: str, expected: str):
    result = clean_germany_company_name(company_name)
    assert result == expected