Any ideas how to improve readability of test output on failed matches?

[bluenote10]

First of all thanks for this nice projects, look very useful and well written!

I'm wondering if you are aware of any tricks to make the output of a failed match within pytest easier to read. Here is a contrived example, simulating a large-ish payload that gets matched against some matcher.

from anys import AnyMatch

LONG_TEXT_EXAMPLE = """
Example of a long multi-line text.
Example of a long multi-line text.
Example of a long multi-line text.
"""

def test_example():
    payload = {
        "foo": LONG_TEXT_EXAMPLE,
        "bar": LONG_TEXT_EXAMPLE,
        "baz": LONG_TEXT_EXAMPLE,
    }
    assert payload == {
        "foo": LONG_TEXT_EXAMPLE,
        "bar": AnyMatch("(.*)looong(.*)"), # doesn't match; mistake on purpose
        "baz": LONG_TEXT_EXAMPLE,
    }

Running a plain pytest shows that the matching has failed, but the output doesn't make it obvious that it was this bar matcher:

>       assert payload == {"foo": "asjdfhlaksfdh", "bar": AnyMatch("(.*)looong(.*)")}
E       AssertionError: assert {'bar': '\nEx...sjdfhlaksfdh'} == {'bar': AnyMa...sjdfhlaksfdh'}
E         Omitting 1 identical items, use -vv to show
E         Differing items:
E         {'bar': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n'} != {'bar': AnyMatch('(.*)looong(.*)')}
E         Left contains 1 more item:
E         {'baz': '\n'
E                 'Example of a long multi-line text.\n'
E                 'Example of a long multi-line text.\n'...
E         
E         ...Full output truncated (3 lines hidden), use '-vv' to show

test_test.py:17: AssertionError
=============================================================================================================================================== short test summary info ===============================================================================================================================================
FAILED test_test.py::test_example - AssertionError: assert {'bar': '\nEx...sjdfhlaksfdh'} == {'bar': AnyMa...sjdfhlaksfdh'}
================================================================================================================================================== 1 failed in 0.06s ==================================================================================================================================================

Using pytest -vv contains the information in a sense, but the usage of the + and - in the diff still confuse me a little bit.

>       assert payload == {"foo": "asjdfhlaksfdh", "bar": AnyMatch("(.*)looong(.*)")}
E       AssertionError: assert {'foo': 'asjdfhlaksfdh', 'bar': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n', 'baz': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n'} == {'foo': 'asjdfhlaksfdh', 'bar': AnyMatch('(.*)looong(.*)')}
E         Common items:
E         {'foo': 'asjdfhlaksfdh'}
E         Differing items:
E         {'bar': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n'} != {'bar': AnyMatch('(.*)looong(.*)')}
E         Left contains 1 more item:
E         {'baz': '\n'
E                 'Example of a long multi-line text.\n'
E                 'Example of a long multi-line text.\n'
E                 'Example of a long multi-line text.\n'}
E         Full diff:
E           {
E         -  'bar': AnyMatch('(.*)looong(.*)'),
E         +  'bar': '\n'
E         +         'Example of a long multi-line text.\n'
E         +         'Example of a long multi-line text.\n'
E         +         'Example of a long multi-line text.\n',
E         +  'baz': '\n'
E         +         'Example of a long multi-line text.\n'
E         +         'Example of a long multi-line text.\n'
E         +         'Example of a long multi-line text.\n',
E            'foo': 'asjdfhlaksfdh',
E           }

test_test.py:17: AssertionError
=============================================================================================================================================== short test summary info ===============================================================================================================================================
FAILED test_test.py::test_example - AssertionError: assert {'foo': 'asjdfhlaksfdh', 'bar': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n', 'baz': '\nExample of a long multi-line text.\nExample of a long multi-line text.\nExample of a long multi-line text.\n'} == {'foo': 'asjd...
================================================================================================================================================== 1 failed in 0.06s ==================================================================================================================================================

In the end this is probably just how pytest works after an == comparison has failed, but maybe you have some other ideas how to improve the readability.

What if the library would offer its own assert_eq(a, b) helper (drawing some inspiration here from the great rust-pretty-assertions in Rust)? Then it could either:

• format the diff itself, with an understanding of the matchers, but that could get tough with all the pretty printing.
• the __eq__ implementation of the matchers could -- as a side effect (dirty...) -- collect all failed matchers in a local context. Then the assert_eq function could simply run a == b and after the comparison is done, it could look into that "local matcher errors collection" if it contains errors and display a more specific explanation why certain matchers have failed.