def test_pipeline_error(time):
"""Test exceptions at pipeline level"""
# test fit
df = _test_df()
func = lambda df: df['num1'] == df['num3']
pipeline = PdPipeline([ColByFrameFunc("Equality", func), ColDrop("B")])
with pytest.raises(PipelineApplicationError):
pipeline.fit(df, verbose=True, time=time)
# test transform
df = _test_df()
with pytest.raises(PipelineApplicationError):
pipeline.transform(df, verbose=True, time=time)
# test fit_transform
df = _test_df()
with pytest.raises(PipelineApplicationError):
pipeline.fit_transform(df, verbose=True, time=time)
def test_two_stage_pipeline_stage(time):
"""Testing something."""
drop_num1 = SilentDropStage('num1')
drop_num2 = SilentDropStage('num2')
pipeline = PdPipeline([drop_num1, drop_num2])
assert len(pipeline) == 2
df = _test_df()
res_df = pipeline.apply(df, verbose=True)
assert 'num1' not in res_df.columns
assert 'num2' not in res_df.columns
assert 'char' in res_df.columns
str(pipeline)
pipeline.memory_report()
# test fit
df = _test_df()
res_df = pipeline.fit(df, verbose=True, time=time)
for x in ['num1', 'num2', 'char']:
assert x in res_df.columns
# test transform
df = _test_df()
res_df = pipeline.transform(df, verbose=True, time=time)
assert 'num1' not in res_df.columns
assert 'num2' not in res_df.columns
assert 'char' in res_df.columns
# test fit_transform
df = _test_df()
res_df = pipeline.fit_transform(df, verbose=True, time=time)
# test get_transformer
trs = lambda pipline: pipeline[:1] # noqa: E731
pipeline = PdPipeline([drop_num1, drop_num2], transformer_getter=trs)
transformer = pipeline.get_transformer()
res_df = transformer(df, verbose=True)
assert 'num1' not in res_df.columns
assert 'num2' in res_df.columns
assert 'char' in res_df.columns