add gokart.testing.assert_frame_equal (#256)

* add gokart.testing.assert_frame_equal

* fix " -> '

* fix isort

* fix review

* change assert_frame_equal -> assert_frame_contents_equal

* bug fix

gokart/testing/__init__.py

@@ -1 +1,2 @@
 from gokart.testing.check_if_run_with_empty_data_frame import test_run, try_to_run_test_for_empty_data_frame
+from gokart.testing.pandas_assert import assert_frame_contents_equal

gokart/testing/pandas_assert.py

@@ -0,0 +1,30 @@
+import pandas as pd
+
+
+def assert_frame_contents_equal(actual: pd.DataFrame, expected: pd.DataFrame, **kwargs):
+    """
+    Assert that two DataFrames are equal.
+    This function is mostly same as pandas.testing.assert_frame_equal, however
+    - this fuction ignores the order of index and columns.
+    - this function fails when duplicated index or columns are found.
+
+    Parameters
+    ----------
+    - actual, expected: pd.DataFrame
+        DataFrames to be compared.
+    - kwargs: Any
+        Parameters passed to pandas.testing.assert_frame_equal.
+    """
+    assert isinstance(actual, pd.DataFrame), 'actual is not a DataFrame'
+    assert isinstance(expected, pd.DataFrame), 'expected is not a DataFrame'
+
+    assert actual.index.is_unique, 'actual index is not unique'
+    assert expected.index.is_unique, 'expected index is not unique'
+    assert actual.columns.is_unique, 'actual columns is not unique'
+    assert expected.columns.is_unique, 'expected columns is not unique'
+
+    assert set(actual.columns) == set(expected.columns), 'columns are not equal'
+    assert set(actual.index) == set(expected.index), 'indexes are not equal'
+
+    expected_reindexed = expected.reindex(actual.index)[actual.columns]
+    pd.testing.assert_frame_equal(actual, expected_reindexed, **kwargs)

test/testing/test_pandas_assert.py

@@ -0,0 +1,37 @@
+import unittest
+
+import pandas as pd
+
+import gokart
+
+
+class TestPandasAssert(unittest.TestCase):
+    def test_assert_frame_contents_equal(self):
+        expected = pd.DataFrame(data=dict(f1=[1, 2, 3], f3=[111, 222, 333], f2=[4, 5, 6]), index=[0, 1, 2])
+        resulted = pd.DataFrame(data=dict(f2=[5, 4, 6], f1=[2, 1, 3], f3=[222, 111, 333]), index=[1, 0, 2])
+
+        gokart.testing.assert_frame_contents_equal(resulted, expected)
+
+    def test_assert_frame_contents_equal_with_small_error(self):
+        expected = pd.DataFrame(data=dict(f1=[1.0001, 2.0001, 3.0001], f3=[111, 222, 333], f2=[4, 5, 6]), index=[0, 1, 2])
+        resulted = pd.DataFrame(data=dict(f2=[5, 4, 6], f1=[2.0002, 1.0002, 3.0002], f3=[222, 111, 333]), index=[1, 0, 2])
+
+        gokart.testing.assert_frame_contents_equal(resulted, expected, atol=1e-1)
+
+    def test_assert_frame_contents_equal_with_duplicated_columns(self):
+        expected = pd.DataFrame(data=dict(f1=[1, 2, 3], f3=[111, 222, 333], f2=[4, 5, 6]), index=[0, 1, 2])
+        expected.columns = ['f1', 'f1', 'f2']
+        resulted = pd.DataFrame(data=dict(f2=[5, 4, 6], f1=[2, 1, 3], f3=[222, 111, 333]), index=[1, 0, 2])
+        resulted.columns = ['f2', 'f1', 'f1']
+
+        with self.assertRaises(AssertionError):
+            gokart.testing.assert_frame_contents_equal(resulted, expected)
+
+    def test_assert_frame_contents_equal_with_duplicated_indexes(self):
+        expected = pd.DataFrame(data=dict(f1=[1, 2, 3], f3=[111, 222, 333], f2=[4, 5, 6]), index=[0, 1, 2])
+        expected.index = [0, 1, 1]
+        resulted = pd.DataFrame(data=dict(f2=[5, 4, 6], f1=[2, 1, 3], f3=[222, 111, 333]), index=[1, 0, 2])
+        expected.index = [1, 0, 1]
+
+        with self.assertRaises(AssertionError):
+            gokart.testing.assert_frame_contents_equal(resulted, expected)