def test_serialize_incremental_chunk_size_pandas_to_recarray(input_df_descr):
if not is_test_data_serializable(input_df_descr):
return
df = _mixed_test_data()[input_df_descr][0]
expectation = _mixed_test_data()[input_df_descr][1]
if not isinstance(expectation, tuple) and issubclass(
expectation, Exception):
for div in (1, 4, 8):
chunk_size = div * 8 * 1024**2
with pytest.raises(expectation):
incr_ser = IncrementalPandasToRecArraySerializer(
df_serializer, df, chunk_size=chunk_size)
[chunk for chunk, _, _, _ in incr_ser.generator_bytes()]
return
for div in (1, 4, 8):
chunk_size = div * 8 * 1024**2
if input_df_descr is not None and len(expectation) > 0:
row_size = int(expectation[0].dtype.itemsize)
chunk_size = NON_HOMOGENEOUS_DTYPE_PATCH_SIZE_ROWS * row_size / div
incr_ser = IncrementalPandasToRecArraySerializer(df_serializer,
df,
chunk_size=chunk_size)
chunk_bytes = [chunk for chunk, _, _, _ in incr_ser.generator_bytes()]
matching = expectation[0].tostring() == b''.join(chunk_bytes)
assert matching
assert expectation[1] == incr_ser.dtype
def test_shape(input_df_descr):
if not is_test_data_serializable(input_df_descr):
return
df = _mixed_test_data()[input_df_descr][0]
expectation = _mixed_test_data()[input_df_descr][1]
incr_ser = IncrementalPandasToRecArraySerializer(df_serializer,
df,
chunk_size=_CHUNK_SIZE)
if not isinstance(expectation, tuple) and issubclass(
expectation, Exception):
with pytest.raises(expectation):
[chunk for chunk, _, _, _ in incr_ser.shape]
else:
assert incr_ser.shape == expectation[0].shape
def test_serialize_pandas_to_recarray(input_df_descr):
if not is_test_data_serializable(input_df_descr):
return
df = _mixed_test_data()[input_df_descr][0]
expectation = _mixed_test_data()[input_df_descr][1]
incr_ser = IncrementalPandasToRecArraySerializer(df_serializer,
df,
chunk_size=_CHUNK_SIZE)
if not isinstance(expectation, tuple) and issubclass(
expectation, Exception):
with pytest.raises(expectation):
[chunk for chunk, _, _, _ in incr_ser.generator_bytes()]
else:
incr_ser_data, incr_ser_dtype = incr_ser.serialize()
matching = expectation[0].tostring() == incr_ser_data.tostring()
assert matching
assert expectation[1] == incr_ser_dtype
