def test_simple(os_mock, stages_mock, dataset_mock):
# given
dataset_mock.create = Mock(spec=dataset_mock, return_value=dataset_mock)
# when
preprocess_corpus(PATH_TO_DATASET_STUB, DEFAULT_PREP_CONFIG)
# then
dataset_mock.create.assert_called_with(
PATH_TO_DATASET_STUB,
DEFAULT_PREP_CONFIG,
None,
None,
overriden_path_to_prep_dataset=PATH_TO_CUR_DIR_STUB)
stages_mock.run_until_preprocessing.assert_called_with(dataset_mock, None)
def test_output(stages_mock, dataset_mock):
# given
dataset_mock.create = Mock(spec=dataset_mock, return_value=dataset_mock)
# when
preprocess_corpus(PATH_TO_DATASET_STUB,
DEFAULT_PREP_CONFIG,
output_path=PATH_TO_OUTPUT_STUB)
# then
dataset_mock.create.assert_called_with(
PATH_TO_DATASET_STUB,
DEFAULT_PREP_CONFIG,
None,
None,
overriden_path_to_prep_dataset=PATH_TO_OUTPUT_STUB,
suppress_caching=False)
stages_mock.run_until_preprocessing.assert_called_with(dataset_mock, None)