def test_pipeline_stage_to_pipeline_addition():
"""Testing something."""
drop_num1 = SilentDropStage('num1')
drop_num2 = SilentDropStage('num2')
pipeline = PdPipeline([drop_num1])
assert len(pipeline) == 1
pipeline = drop_num2 + pipeline
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
def test_pipeline_slice():
"""Testing something."""
drop_num1 = SilentDropStage('num1')
drop_num2 = SilentDropStage('num2')
drop_char = SilentDropStage('char')
pipeline = PdPipeline([drop_num1, drop_num2, drop_char])
assert len(pipeline) == 3
pipeline = pipeline[0:2]
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
def test_pipeline_slice_by_name():
"""Testing something."""
drop_num1 = SilentDropStage('num1', name='dropNum1')
drop_num2 = SilentDropStage('num2', name='dropNum2')
drop_char = SilentDropStage('char', name='dropChar')
pipeline = PdPipeline([drop_num1, drop_num2, drop_char])
assert len(pipeline) == 3
pipeline = pipeline[['dropNum1', 'dropNum2']]
assert len(pipeline) == 2
assert pipeline['dropNum1'] == drop_num1
with pytest.raises(ValueError) as e:
pipeline['dropChar']
assert str(e.value) == "'dropChar' is not exist."
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
def test_pipeline_index():
"""Testing something."""
df = _test_df()
drop_num1 = SilentDropStage('num1')
drop_num2 = SilentDropStage('num2')
drop_char = SilentDropStage('char')
pipeline = PdPipeline([drop_num1, drop_num2, drop_char])
assert len(pipeline) == 3
assert pipeline[0] == drop_num1
assert 'num1' not in pipeline[0](df).columns
assert pipeline[1] == drop_num2
assert 'num2' not in pipeline[1](df).columns
assert pipeline[2] == drop_char
assert 'char' not in pipeline[2](df).columns
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():
"""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)
# test fit
df = _test_df()
res_df = pipeline.fit(df, verbose=True)
for x in ['num1', 'num2', 'char']:
assert x in res_df.columns
# test transform
df = _test_df()
res_df = pipeline.transform(df, verbose=True)
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)
# 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
def test_pipeline_to_int_addition():
"""Testing something."""
pipeline = PdPipeline([SilentDropStage('num1')])
with pytest.raises(TypeError):
res = pipeline + 43
assert not isinstance(res, PdPipeline)