def _test_refine_spec_apply(self, in_table: pd.DataFrame, column: str,
spec: RefineSpec,
expected_out: pd.DataFrame=pd.DataFrame(),
expected_error: str='') -> None:
"""Render and assert the output is as expected."""
result = ProcessResult.coerce(spec.apply(in_table, column))
# Sanitize result+expected, so if sanitize changes these tests may
# break (which is what we want).
result.sanitize_in_place()
expected = ProcessResult(expected_out, expected_error)
expected.sanitize_in_place()
self.assertEqual(result.error, expected.error)
assert_frame_equal(result.dataframe, expected.dataframe)
def test_refine_rename_nan_category(self):
# Get explicit, because pandas botches the comparison:
#
# >>> self._test_refine_spec_apply(
# pd.DataFrame({'A': [np.nan]}, dtype='category'),
# 'A', RefineSpec({'b': 'c'}),
# pd.DataFrame({'A': [np.nan]}, dtype='category')
# )
# Attribute "dtype" are different
# [left]: CategoricalDtype(categories=[], ordered=False)
# [right]: CategoricalDtype(categories=[], ordered=False)
spec = RefineSpec({'b': 'c'})
dtype1 = pd.DataFrame({'A': []}, dtype='category')['A'].dtype
result = spec.apply(pd.DataFrame({'A': []}, dtype='category'), 'A')
dtype2 = result['A'].dtype
self.assertEqual(0, len(result))
self.assertEqual(0, len(result['A'].cat.categories))
def _test_refine_spec_apply(self,
in_table: pd.DataFrame,
column: str,
spec: RefineSpec,
expected_out: pd.DataFrame = pd.DataFrame(),
expected_error: str = '') -> None:
"""Render and assert the output is as expected."""
result = spec.apply(in_table, column)
if not expected_out.empty and expected_error:
table, error = result
self.assertEqual(error, expected_error)
assert_frame_equal(table, expected_out)
elif expected_error:
self.assertEqual(result, expected_error)
else:
assert_frame_equal(result, expected_out)