def test_mok_store_invalid_params(data_type, extension):
"""
test storing data with invalid file format
"""
moker = Mok(data_type=data_type)
with pytest.raises(ValueError):
moker.store(extension=extension,
domain=extension if data_type == "specifc" else None)
def test_mok_batch_invalid():
"""
test generating random files fails with valueerror if
data_type is invalid
"""
moker = Mok(data_type="random")
moker.data_type = "specific"
with pytest.raises(ValueError):
moker.store()
def test_mok_store_random_extension(
domain,
test_data_dir # pylint: disable=redefined-outer-name
):
"""
test storing data with invalid file format
"""
moker = Mok(data_type="specific")
file_path = moker.store(domain=domain, data_dir=test_data_dir)
assert pathlib.Path(file_path).exists()
def test_mok_store_invalid_extension(
domain,
test_data_dir # pylint: disable=redefined-outer-name
):
"""
test storing data with invalid file format
"""
moker = Mok(data_type="specific")
with pytest.raises(ValueError):
moker.store(domain=domain, data_dir=test_data_dir, extension="invalid")
def test_mok_store_no_file_name(
domain,
extension,
test_data_dir # pylint: disable=redefined-outer-name
):
"""
test generating various domain specific datasets
"""
moker = Mok(data_type="specific")
assert not test_data_dir.joinpath(
f"{moker.default_fn}.{extension}").exists()
moker.store(data_dir=test_data_dir, extension=extension, domain=domain)
assert test_data_dir.joinpath(f"{moker.default_fn}.{extension}").exists()
def test_mok_batch(num_files, test_data_dir): # pylint: disable=redefined-outer-name
"""
test generating random files in a batch
"""
moker = Mok(data_type="random")
file_paths = moker.batch(data_dir=str(test_data_dir), num_files=num_files)
for file_path in file_paths:
assert pathlib.Path(file_path).exists()
assert pathlib.Path(file_path).suffix.replace(".", "") in [
"json",
"csv",
"parquet",
"xlsx",
]
def test_mok_batch_default_dir(num_files):
"""
test generating random files in a batch
"""
moker = Mok(data_type="random")
file_paths = moker.batch(num_files=num_files)
for file_path in file_paths:
assert pathlib.Path(file_path).exists()
assert pathlib.Path(file_path).suffix.replace(".", "") in [
"json",
"csv",
"parquet",
"xlsx",
]
shutil.rmtree(moker.default_dir)
def test_mok_store_specific(
domain,
extension,
test_data_dir, # pylint: disable=redefined-outer-name
):
"""
test generating various domain specific datasets
"""
file_name = test_data_dir.joinpath(f"{domain}.{extension}")
assert not file_name.exists()
moker = Mok(data_type="specific")
moker.store(
file_name=str(file_name),
data_dir=test_data_dir,
extension=extension,
domain=domain,
)
assert pathlib.Path(str(file_name) + f".{extension}").exists()
def test_mok_init_invalid(data_type):
"""
test initializing data generator with invalid data types
"""
with pytest.raises(ValueError):
Mok(data_type=data_type)
def test_mok_init_valid(data_type):
"""
test initializing data generator with valid data types
"""
moker = Mok(data_type=data_type)
assert moker.data_type == data_type