Calling unittest.assertDictEqual for medium-size dictionaries takes too long #99151

boaza · 2022-11-06T07:56:01Z

Calling assertDictEqual(d1, d2) / takes forever, even for medium-size dictionaries. To reproduce:

from unittest import TestCase
from random import randint

    def test_assert_dict(self):
        r = 10000000
        num = 10000
        d1 = dict((randint(0, r), randint(0, r)) for _ in range(num))
        d2 = dict((randint(0, r), randint(0, r)) for _ in range(num))
        self.assertDictEqual(d1, d2)

Probably related to issue #63416.

Tested on Python 3.10, Windows 11

Linked PRs

gh-99151: Improve performance and error readability of unittest.TestCase.assertDictEqual #126923

ronaldoussoren · 2022-12-22T21:07:05Z

See also #51180 about slowness in difflib.ndiff (used by assertDictEquals to report about differences).

One avenue to explore would be to pick one of the other diff functions: unified_diff appears to be a lot faster, although this does change the output.

merlinz01 · 2024-11-16T23:32:13Z

I think a simple difflib.ndiff over pprint is a bad choice for 'assertDictEqual`. The output it gives is often very confusing, like this:

E       AssertionError: {'second': 2, 'seventh': 7, 'sixth': 6, 'fo[53 chars]': 3} != {'first': 1, 'seventh': 7, 'ninth': 9, 'sec[66 chars]': 3}
E         {'eighth': 8,
E          'fifth': 5,
E          'first': 1,
E          'fourth': 4,
E       +  'ninth': 9,
E          'second': 2,
E          'seventh': 7,
E       -  'sixth': 6,
E       -  'third': 3}
E       ?            ^
E       
E       +  'third': 3,
E       ?            ^
E       
E       +  'zeroth': 0}

I put together a better implementation which gives much nicer output, handles nested structures, and is WAY faster than the current implementation for the test given.

E   AssertionError: {'second': 2, 'seventh': 7, 'sixth': 6, 'fo[53 chars]': 3} != {'first': 1, 'seventh': 7, 'ninth': 9, 'sec[66 chars]': 3}
E   
E   Diff:
E   {
E       'eighth': 8,
E       'fifth': 5,
E       'first': 1,
E       'fourth': 4,
E       'second': 2,
E       'seventh': 7,
E       'third': 3,
E     - 'sixth': 6,
E     + 'ninth': 9,
E     + 'zeroth': 0,
E   }

I would be happy to contribute my code if that is desired.

boaza added the type-bug An unexpected behavior, bug, or error label Nov 6, 2022

ronaldoussoren added the stdlib Python modules in the Lib dir label Dec 22, 2022

bedevere-app bot mentioned this issue Nov 17, 2024

gh-99151: Improve performance and error readability of unittest.TestCase.assertDictEqual #126923

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

Calling unittest.assertDictEqual for medium-size dictionaries takes too long #99151

Calling unittest.assertDictEqual for medium-size dictionaries takes too long #99151

Uh oh!

Uh oh!

Uh oh!

Calling unittest.assertDictEqual for medium-size dictionaries takes too long #99151

Calling unittest.assertDictEqual for medium-size dictionaries takes too long #99151

Comments

Uh oh!

Linked PRs

Uh oh!

Uh oh!