def test_add_bb_feature_invalid_df_columns(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
def bb_fn(ltuple, rtuple):
return 1.0
add_blackbox_feature(pd.DataFrame(), 'test', bb_fn)
def test_add_bb_feature_invalid_df_columns(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
def bb_fn(ltuple, rtuple):
return 1.0
add_blackbox_feature(pd.DataFrame(), 'test', bb_fn)
def test_add_bb_feature_name_already_present(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = create_feature_table()
len1 = len(feature_table)
def bb_fn(ltuple, rtuple):
return 1.0
add_blackbox_feature(feature_table, 'test', bb_fn)
add_blackbox_feature(feature_table, 'test', bb_fn)
def test_add_bb_feature_name_already_present(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = create_feature_table()
len1 = len(feature_table)
def bb_fn(ltuple, rtuple):
return 1.0
add_blackbox_feature(feature_table, 'test', bb_fn)
add_blackbox_feature(feature_table, 'test', bb_fn)
def test_add_bb_feature_valid_2(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = create_feature_table()
def bb_fn(ltuple, rtuple):
return 1.0
len1 = len(feature_table)
add_blackbox_feature(feature_table, 'test', bb_fn)
len2 = len(feature_table)
self.assertEqual(len1+1, len2)
self.assertEqual(feature_table.ix[len(feature_table)-1, 'function'](A.ix[1], B.ix[2]), 1.0)
def test_add_bb_feature_valid_1(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = get_features_for_matching(A, B, validate_inferred_attr_types=False)
def bb_fn(ltuple, rtuple):
return 1.0
len1 = len(feature_table)
add_blackbox_feature(feature_table, 'test', bb_fn)
len2 = len(feature_table)
self.assertEqual(len1+1, len2)
self.assertEqual(feature_table.ix[len(feature_table)-1, 'function'](A.ix[1], B.ix[2]), 1.0)
def test_add_bb_feature_valid_2(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = create_feature_table()
def bb_fn(ltuple, rtuple):
return 1.0
len1 = len(feature_table)
add_blackbox_feature(feature_table, 'test', bb_fn)
len2 = len(feature_table)
self.assertEqual(len1 + 1, len2)
self.assertEqual(
feature_table.ix[len(feature_table) - 1, 'function'](A.ix[1],
B.ix[2]), 1.0)
def test_add_bb_feature_valid_1(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = get_features_for_matching(
A, B, validate_inferred_attr_types=False)
def bb_fn(ltuple, rtuple):
return 1.0
len1 = len(feature_table)
add_blackbox_feature(feature_table, 'test', bb_fn)
len2 = len(feature_table)
self.assertEqual(len1 + 1, len2)
self.assertEqual(
feature_table.ix[len(feature_table) - 1, 'function'](A.ix[1],
B.ix[2]), 1.0)
def test_add_bb_feature_with_attrs(self):
A = read_csv_metadata(path_a)
B = read_csv_metadata(path_b, key='ID')
feature_table = get_features_for_matching(
A, B, validate_inferred_attr_types=False)
def bb_fn(ltuple, rtuple):
return min(len(ltuple['name']), len(rtuple['name']))
len1 = len(feature_table)
attrs = {'left_attribute': 'name', 'right_attribute': 'name'}
add_blackbox_feature(feature_table, 'bb_attr_test', bb_fn, **attrs)
len2 = len(feature_table)
self.assertEqual(len1 + 1, len2)
added_feature = feature_table.iloc[len(feature_table) - 1]
self.assertEqual(added_feature.feature_name, 'bb_attr_test')
self.assertEqual(added_feature.left_attribute, 'name')
self.assertEqual(added_feature.right_attribute, 'name')
self.assertEqual(added_feature.simfunction, None)
def test_add_bb_feature_invalid_feature_dict_type(self):
add_blackbox_feature(pd.DataFrame(), "", None)
def test_add_bb_feature_invalid_feat_name_type(self):
add_blackbox_feature(pd.DataFrame(), None, dict())
def test_add_bb_feature_invalid_df(self):
add_blackbox_feature(None, 'test', dict())
def test_add_bb_feature_invalid_feature_dict_type(self):
add_blackbox_feature(pd.DataFrame(), "", None)
def test_add_bb_feature_invalid_feat_name_type(self):
add_blackbox_feature(pd.DataFrame(), None, dict())
def test_add_bb_feature_invalid_df(self):
add_blackbox_feature(None, 'test', dict())
