poyee-data-warehouse/dw_base/spark/udf/test/test_common_clean.py

import pytest
from dw_base.spark.udf.test.common_clean import clean_pre_join

@pytest.mark.parametrize("name, expected", [
    ('MS ABC Ltd.', 'ABC'),
    ('MS ABC I PRIVATE LIMITED', 'ABC'),
    ("M S ABC COMPANY PRIVATE L", 'ABC'),
    ('ABC Ltd.', 'ABC'),
    ('ABC P LTD', 'ABC'),
    ('ABC PRIVATE LIMITED', 'ABC'),
    ('ABC LIMITED', 'ABC'),
    ('ABC INC', 'ABC'),
    ('ABC AAA', 'ABC AAA'),
    ('ABC CO LIMITED', 'ABC'),
    ('ABC AND CO LIMITED', 'ABC AND CO LIMITED'),
    ('ABC COLTD', 'ABC'),

])
def test_clean_pre_join(name, expected):
    assert clean_pre_join(name) == expected