how to use assert_frame_equal in unittest

Question

New to unittest package. I'm trying to verify the DataFrame returned by a function through the following code. Even though I hardcoded the inputs of assert_frame_equal to be equal (pd.DataFrame([0,0,0,0])), the unittest still fails. Anyone would like to explain why it happens?

import unittest
from pandas.util.testing import assert_frame_equal
class TestSplitWeight(unittest.TestCase):
    def test_allZero(self):
        #splitWeight(pd.DataFrame([0,0,0,0]),10)
        self.assert_frame_equal(pd.DataFrame([0,0,0,0]),pd.DataFrame([0,0,0,0]))

suite = unittest.TestLoader().loadTestsFromTestCase(TestSplitWeight)
unittest.TextTestRunner(verbosity=2).run(suite)

Error: AttributeError: 'TestSplitWeight' object has no attribute 'assert_frame_equal'

score 25 · Answer 1 · edited Oct 08 '19 at 10:35

25

alecxe answer is incomplete, you can indeed use pandas' assert_frame_equal() with unittest.TestCase, using unittest.TestCase.addTypeEqualityFunc

import unittest
import pandas as pd
import pandas.testing as pd_testing

class TestSplitWeight(unittest.TestCase):
    def assertDataframeEqual(self, a, b, msg):
        try:
            pd_testing.assert_frame_equal(a, b)
        except AssertionError as e:
            raise self.failureException(msg) from e

    def setUp(self):
        self.addTypeEqualityFunc(pd.DataFrame, self.assertDataframeEqual)

    def test_allZero(self):
        self.assertEqual(pd.DataFrame([0,0,0,0]), pd.DataFrame([0,0,0,0]))

edited Oct 08 '19 at 10:35

Tom Jelen

2,559
1
24
23

answered Jan 24 '19 at 10:14

Léo Germond

720
8
18

I have duplicate your code and I got the following error: `TypeError: addTypeEqualityFunc() missing 1 required positional argument: 'function'`. Do you know why? – Dadou Jul 20 '20 at 17:06
What is the purpose of msg here? Can the message of assert_frame_equal be somehow passed to assertEqual? – Fato39 Feb 02 '21 at 15:17
1

@fato39 the msg parameter allows to pass-on an additional message, additionaly the raise...from construct used there wraps the original pandas exception into a unittest-compatible failureException. – Léo Germond Jun 08 '21 at 08:40

alecxe · Accepted Answer · 2016-08-08T22:58:29.823

assert_frame_equal() is coming from the pandas.util.testing package, not from the unittest.TestCase class. Replace:

self.assert_frame_equal(pd.DataFrame([0,0,0,0]),pd.DataFrame([0,0,0,0]))

with:

assert_frame_equal(pd.DataFrame([0,0,0,0]), pd.DataFrame([0,0,0,0]))

When you had self.assert_frame_equal, it tried to find assert_frame_equal attribute on the unittest.TestCase instance, and, since there is not assert_frame_equal attribute or method exposed on an unittest.TestCase class, it raised an AttributeError.

score -1 · Answer 3 · answered Dec 03 '21 at 12:39

-1

If you import this, the problem is gone.

from pandas.testing import assert_frame_equal  # <-- for testing dataframes

answered Dec 03 '21 at 12:39

banana_99

591
5
15

how to use assert_frame_equal in unittest

3 Answers3