def test_rename_double_metadata(self):
non_sf = SensitiveFrame(columns=['empty'],
sensitive_data=['empty'],
quasi_identifiers=['empty'])
test_sf = non_sf.rename(columns={'empty': 'void'})
assert test_sf.columns.tolist() == ['void']
assert test_sf.quasi_identifiers == ['void']
assert test_sf.sensitive_data == ['void']
def test_get_and_set_double_metadata(self, sample_sf_two):
test_sf = SensitiveFrame()
test_sf['birth_year'] = sample_sf_two['born']
expected_df = DataFrame([2335, 2340], columns=['birth_year'])
assert_frame_equal(test_sf, expected_df)
assert test_sf.quasi_identifiers == ['birth_year']
assert test_sf.sensitive_data == ['birth_year']
def test_simple_partition(self, middle_df):
test_mp = MondrianPartitioner(2)
quasi_identifiers = ['middle_values']
test_sf = SensitiveFrame(middle_df, quasi_identifiers=quasi_identifiers)
actual_partitions = test_mp.partition_sf(test_sf)
expected_left_df = DataFrame.from_dict({0: 1, 1: 1}, orient='index')
expected_left_df.columns = ['middle_values']
expected_right_df = DataFrame.from_dict({2: 3, 3: 5, 4: 5}, orient='index')
expected_right_df.columns = ['middle_values']
assert len(actual_partitions) == 2
assert actual_partitions[0].quasi_identifiers == quasi_identifiers
assert actual_partitions[1].quasi_identifiers == quasi_identifiers
assert_frame_equal(actual_partitions[0], expected_left_df)
assert_frame_equal(actual_partitions[1], expected_right_df)
def test_partition_along_widest_quasi_identifiers(self, sample_df):
test_mp = MondrianPartitioner(2)
quasi_identifiers = ['slim_values', 'wide_values']
test_sf = SensitiveFrame(sample_df, quasi_identifiers=quasi_identifiers)
actual_partitions = test_mp.partition_sf(test_sf)
expected_left_df = DataFrame.from_dict({0: [1, 1], 1: [1, 1]}, orient='index')
expected_left_df.columns = quasi_identifiers
expected_right_df = DataFrame.from_dict({2: [1, 1000], 3: [2, 1000], 4: [2, 1000]}, orient='index')
expected_right_df.columns = quasi_identifiers
assert len(actual_partitions) == 2
assert sorted(actual_partitions[0].quasi_identifiers) == quasi_identifiers
assert sorted(actual_partitions[1].quasi_identifiers) == quasi_identifiers
assert_frame_equal(actual_partitions[0], expected_left_df)
assert_frame_equal(actual_partitions[1], expected_right_df)
def test_oversized_partition(self, middle_df):
test_mp = MondrianPartitioner(10)
test_sf = SensitiveFrame(middle_df, quasi_identifiers=['middle_values'])
assert test_mp.partition_sf(test_sf) == [test_sf]
def test_init_cleans_sf_metadata(self, sample_sf):
test_sf = SensitiveFrame(sample_sf)
assert test_sf.quasi_identifiers == []
assert test_sf.sensitive_data == []
def test_init_with_df_keeps_data(self, sample_df):
test_sf = SensitiveFrame(sample_df)
expected_df = sample_df
assert_frame_equal(test_sf, expected_df)
def test_init_selects_metadata(self):
test_sf = SensitiveFrame(columns=['a'], quasi_identifiers=['a', 'b'])
assert test_sf.quasi_identifiers == ['a']
def test_init_sets_multiple_metadata(self):
test_sf = SensitiveFrame(columns=['a', 'b'],
sensitive_data=['a'],
quasi_identifiers=['b'])
assert test_sf.sensitive_data == ['a']
assert test_sf.quasi_identifiers == ['b']
def test_init_sets_sensitive_data(self):
test_sf = SensitiveFrame(columns=['c', 'd'], sensitive_data=['c', 'd'])
assert test_sf.sensitive_data == ['c', 'd']
def sample_sf_three():
return SensitiveFrame([['william', 'red'], ['laren', 'red']],
columns=['given_name', 'uniform'],
quasi_identifiers=['uniform'],
sensitive_data=['given_name'])
def sample_sf_two():
return SensitiveFrame([['starfleet', 2335], ['starfleet', 2340]],
columns=['organization', 'born'],
quasi_identifiers=['born'],
sensitive_data=['born'])
def test_get_and_set_from_df(self, sample_df):
test_sf = SensitiveFrame()
test_sf['grade'] = sample_df['rank']
expected_df = DataFrame(['commander', 'lieutenant'], columns=['grade'])
assert_frame_equal(test_sf, expected_df)
assert test_sf.quasi_identifiers == []
def test_get_and_set_non_metadata(self, sample_sf):
test_sf = SensitiveFrame()
test_sf['person'] = sample_sf['name']
expected_df = DataFrame(['riker', 'ro'], columns=['person'])
assert_frame_equal(test_sf, expected_df)
assert test_sf.sensitive_data == []
def sample_sf():
return SensitiveFrame([['riker', 'commander', 'man', 'human'],
['ro', 'ensign', 'woman', 'bejoran']],
columns=['name', 'rank', 'gender', 'species'],
quasi_identifiers=['rank', 'gender'],
sensitive_data=['species'])
def sample_right_sf():
return SensitiveFrame(
[['riker', 'starfleet', 2335], ['ro', 'starfleet', 2340]],
columns=['name', 'organization', 'born'],
quasi_identifiers=['name', 'born'],
sensitive_data=['born'])
def test_join_non_metadata(self):
non_sf = SensitiveFrame(columns=['empty'])
test_sf = SensitiveFrame(columns=['void']).join(non_sf)
assert test_sf.quasi_identifiers == []
assert test_sf.sensitive_data == []