Add RichExceptionChainRepr for richer exception highlighting

[joshuadavidthomas]

Yanked most of __rich_console__ from Rich's Traceback class and adapted for pytest's ExceptionChainRepr.

I ignored locals for the time being, which I assume I can grab from reprlocals on each ReprEntry.

There's no testing other than the eye test and the code could stand to be cleaned up a bit as well. But! It's working.

[nicoddemus]

This looks amazing 😍 great work!

About testing: not sure how we can approach testing in pytest-rich in general TBH. Any ideas?

[joshuadavidthomas]

Thanks! 😊

As far as testing, I'm a bit stumped too.

A quick glance through Rich's test suite, most of the traceback tests are testing the print_exception method, so they aren't of any use to us. It does look like Console offers begin_capture and end_capture methods that may come in handy.

Might be worth pinging Will to see if he has any suggestions.

[joshuadavidthomas]

For the time being, could we use a mark on failing tests so that you could manually test the terminal output for tracebacks without breaking CI?

[joshuadavidthomas]

Running this PR on my main Django app, I get this error:

──────────────────────── forsythia/lease/clubs/tests/test_views.py::TestClubList::test_post ────────────────────────
Traceback (most recent call last):
  File "/usr/local/bin/pytest", line 8, in <module>
    sys.exit(console_main())
  File "/usr/local/lib/python3.10/site-packages/_pytest/config/__init__.py", line 188, in console_main
    code = main()
  File "/usr/local/lib/python3.10/site-packages/_pytest/config/__init__.py", line 165, in main
    ret: Union[ExitCode, int] = config.hook.pytest_cmdline_main(
  File "/usr/local/lib/python3.10/site-packages/pluggy/_hooks.py", line 265, in __call__
    return self._hookexec(self.name, self.get_hookimpls(), kwargs, firstresult)
  File "/usr/local/lib/python3.10/site-packages/pluggy/_manager.py", line 80, in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
  File "/usr/local/lib/python3.10/site-packages/pluggy/_callers.py", line 60, in _multicall
    return outcome.get_result()
  File "/usr/local/lib/python3.10/site-packages/pluggy/_result.py", line 60, in get_result
    raise ex[1].with_traceback(ex[2])
  File "/usr/local/lib/python3.10/site-packages/pluggy/_callers.py", line 39, in _multicall
    res = hook_impl.function(*args)
  File "/usr/local/lib/python3.10/site-packages/_pytest/main.py", line 315, in pytest_cmdline_main
    return wrap_session(config, _main)
  File "/usr/local/lib/python3.10/site-packages/_pytest/main.py", line 303, in wrap_session
    config.hook.pytest_sessionfinish(
  File "/usr/local/lib/python3.10/site-packages/pluggy/_hooks.py", line 265, in __call__
    return self._hookexec(self.name, self.get_hookimpls(), kwargs, firstresult)
  File "/usr/local/lib/python3.10/site-packages/pluggy/_manager.py", line 80, in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
  File "/usr/local/lib/python3.10/site-packages/pluggy/_callers.py", line 60, in _multicall
    return outcome.get_result()
  File "/usr/local/lib/python3.10/site-packages/pluggy/_result.py", line 60, in get_result
    raise ex[1].with_traceback(ex[2])
  File "/usr/local/lib/python3.10/site-packages/pluggy/_callers.py", line 39, in _multicall
    res = hook_impl.function(*args)
  File "/usr/local/lib/python3.10/site-packages/pytest_rich.py", line 241, in pytest_sessionfinish
    self.console.print(tb)
  File "/usr/local/lib/python3.10/site-packages/rich/console.py", line 1616, in print
    extend(render(renderable, render_options))
  File "/usr/local/lib/python3.10/site-packages/rich/console.py", line 1254, in render
    for render_output in iter_render:
  File "/usr/local/lib/python3.10/site-packages/pytest_rich.py", line 315, in __rich_console__
    repr_highlighter(
  File "/usr/local/lib/python3.10/site-packages/rich/highlighter.py", line 36, in __call__
    raise TypeError(f"str or Text instance required, not {text!r}")
TypeError: str or Text instance required, not None

Running without pytest-rich:

________________________________________ TestClubList.test_post[data0-201] _________________________________________

self = <forsythia.lease.clubs.tests.test_views.TestClubList object at 0x7f3ded0d7040>
api_client = <rest_framework.test.APIClient object at 0x7f3debb5dc30>
data = {'clubInfo': 'test club 1', 'description': 'test club 1', 'name': 'test club 1'}, expected_status_code = 201

    @pytest.mark.parametrize(
        "data,expected_status_code",
        [
            (
                {
                    "name": "test club 1",
                    "description": "test club 1",
                    "clubInfo": "test club 1",
                },
                status.HTTP_201_CREATED,
            ),
            (
                {
                    "name": "test club 2",
                },
                status.HTTP_201_CREATED,
            ),
            ({}, status.HTTP_400_BAD_REQUEST),
        ],
    )
    def test_post(self, api_client, data, expected_status_code):
        response = api_client.post(self.endpoint, data=data)

>       assert response.status_code != expected_status_code
E       assert 201 != 201
E        +  where 201 = <Response status_code=201, "application/json">.status_code

forsythia/lease/clubs/tests/test_views.py:57: AssertionError

Guess my implementation of _grab_err_msg needs some updating..

[nicoddemus]

Guess my implementation of _grab_err_msg needs some updating..

Yeah I guess somewhere there's a None that needs to be handled as text.

[joshuadavidthomas]

First pass at dealing with the error messages.

The original _grab_err_msg function couldn't deal with a plain assertion without a message, e.g. assert 0 vs assert 0, "Some failure". Fixed that and moved it to within the render_chain method for now.

Here are a few screenshots to show progress so far:

[joshuadavidthomas]

From my previously mentioned Django app, it works and looks a lot better:

[joshuadavidthomas]

This PR may be in a good spot to merge.

There are still improvments to be done:

• I'm not 100% in love with how the error messages are displayed, could explore a richer way to output them
• The code for RichExceptionChainRepr could probably stand to be cleaned up and refactored a bit, _render_chain is doing a lot of heavy lifting that could be split out (rendering error header, code, and message)
• Along with the refactoring, documentation in docstrings would be useful for future contributors
• Testing, though that is its own beast...

I think those all should go in as separate issues with their own PRs, in order to avoid this PR's scope from expanding beyond richer exception highlighting.

[joshuadavidthomas]

Additional stray thoughts:

• May want to rename the header from Traceback to something else (Failed Test Report?), since it's not an actual traceback it's a bit misleading for the header to say it is
• Still need to include the reprlocals information from the ReprEntry somewhere

So maybe I was fibbing a bit and this PR isn't ready to merge just yet.

[nicoddemus]

This PR may be in a good spot to merge.

Agreed, want to mark it as Ready for Review then?

[joshuadavidthomas]

Final tweaks, ready for review!

Comparisions

Built-in pytest TerminalReporter:

pytest-rich RichReporter:

Built-in pytest TerminalReporter:

pytest-rich RichReporter:

[joshuadavidthomas]

My goal was this PR was to get a good starting point. Here are some areas for improvement:

• pytest's TerminalReporter shows the entire test function being called, whereas since I adapted from rich's Traceback, RichReporter only shows the last 3 and next 3 lines from the failed line.

• TerminalReporter puts the error messages right after the failed lines, whereas RichReporter puts them at the bottom. Switching away from rich's Syntax object would make this easier, as Syntax is apparently not designed to be composed in pieces as detailed by Will here. Though we could slice it after the fact and put the errors in that way, relevant comment from Will.

[nicoddemus]

This looks awesome!

I didn't review the code in detail, but the results do look great.

Could you please also update the screenshot in assets/? Thanks!

[nicoddemus]

Btw your detailed comments for follow ups should probably be moved to their own issues so we don't lose track of them. 👍

[joshuadavidthomas]

Added new issues on where to go from here:

#16
#17

[fruch]

When I ever thought about pytest+rich, this is the feature I had in mind.