0.5.12: regression EdgeCaseError: astAssign.targets[0] is of type <class 'ast.Subscript'>

[ssbarnea]

While trying to update from 0.5.9 to 0.5.12, i faced a runtime exception on https://github.com/ansible/ansible-dev-tools repository.

Traceback (most recent call last):
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/bin/pydoclint", line 8, in <module>
    sys.exit(main())
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/click/core.py", line 1157, in __call__
    return self.main(*args, **kwargs)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/click/core.py", line 1078, in main
    rv = self.invoke(ctx)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/click/core.py", line 1434, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/click/core.py", line 783, in invoke
    return __callback(*args, **kwargs)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/click/decorators.py", line 33, in new_func
    return f(get_current_context(), *args, **kwargs)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/main.py", line 410, in main
    violationsInAllFiles: dict[str, list[Violation]] = _checkPaths(
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/main.py", line 585, in _checkPaths
    violationsInThisFile: list[Violation] = _checkFile(
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/main.py", line 673, in _checkFile
    visitor.visit(tree)
  File "/Users/ssbarnea/.config/uv/python/cpython-3.10.15-macos-aarch64-none/lib/python3.10/ast.py", line 418, in visit
    return visitor(node)
  File "/Users/ssbarnea/.config/uv/python/cpython-3.10.15-macos-aarch64-none/lib/python3.10/ast.py", line 426, in generic_visit
    self.visit(item)
  File "/Users/ssbarnea/.config/uv/python/cpython-3.10.15-macos-aarch64-none/lib/python3.10/ast.py", line 418, in visit
    return visitor(node)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/visitor.py", line 106, in visit_ClassDef
    checkClassAttributesAgainstClassDocstring(
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/utils/visitor_helper.py", line 46, in checkClassAttributesAgainstClassDocstring
    actualArgs: ArgList = extractClassAttributesFromNode(
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/utils/visitor_helper.py", line 178, in extractClassAttributesFromNode
    atl.extend(ArgList.fromAstAssign(itm).infoList)
  File "/Users/ssbarnea/.cache/pre-commit/repo9uz5ky6k/py_env-3.10/lib/python3.10/site-packages/pydoclint/utils/arg.py", line 234, in fromAstAssign
    raise EdgeCaseError(
pydoclint.utils.edge_case_error.EdgeCaseError: astAssign.targets[0] is of type <class 'ast.Subscript'>

I am still trying to figure out what is causing it as the output is not really very useful.

[ssbarnea]

Update: it seems this this problem originates from another place, probably a dependency of pydoclint that was recently released is causing it. It does reproduce with last releases and multiple versions of python too.

[jsh9]

Thanks for reporting! Let me try to reproduce this error with your repo.

[ssbarnea]

Locally I no longer can run pydoclint due to the raised bug and I tried on two different machines: macos and ubuntu, both arm64 and using python installed using mise. Still, as you probably seen, on CI it works. I am starting to suspect something related to either pre-commit, uv or mise being involved here.

[jsh9]

Hi @ssbarnea , I was unable to reproduce this issue locally either. Could you try it in a fresh local env? Or could you find out which line of code in the repo caused this issue? (If you look further than "Traceback (most recent call last)" there may be such info?)

[mvadari]

I am also recently running into this issue, despite not changing my pydoclint version (0.5.7). It is fine in CI and when I run pydoclint but fails in flake8.

[jsh9]

I am also recently running into this issue, despite not changing my pydoclint version (0.5.7). It is fine in CI and when I run pydoclint but fails in flake8.

I probably need more information to reproduce this issue, because when I ran ansible-dev-tools locally, I saw no errors.

[mvadari]

I am also recently running into this issue, despite not changing my pydoclint version (0.5.7). It is fine in CI and when I run pydoclint but fails in flake8.

I probably need more information to reproduce this issue, because when I ran ansible-dev-tools locally, I saw no errors.

This is the PR where I had the issue: XRPLF/xrpl-py#749

Reproduction:

poetry install
poetry run flake8

Running on Mac M1

[jsh9]

Hi @mvadari ,

I tried your steps, and I still couldn't see the issue. poetry run flake8 seemed to run without issues. (I tried in Python 3.9, 3.10, 3.11, 3.12, and 3.13.)

So the issue may be some dependencies in your existing environment clashes with pydoclint. And the reason why things work in CI (as well as on my computer) is that every time pydoclint is run in a fresh environment. (When I run it on your repo on my computer, it's in fresh environments also.)

[ssbarnea]

I am unable to reproduce on linux but on macos m1 seems to happen with any python blend i tried. I kinda doubt is its dependencies as it runs via pre-commit with only these:

❯ pip freeze
click==8.1.7
docstring_parser_fork==0.0.9
pydoclint @ file:///Users/ssbarnea/.cache/pre-commit/repo05xijrt5

I was failing on a file that was not even tracked under git, part of a test venv,

foo/lib/python3.13/site-packages/pip/_internal/cli/cmdoptions.py

I removed the folder and now is passing, still I am sure that the issue is real. Sadly, too late as I removed the folder.

[mvadari]

Hi @mvadari ,

I tried your steps, and I still couldn't see the issue. poetry run flake8 seemed to run without issues. (I tried in Python 3.9, 3.10, 3.11, 3.12, and 3.13.)

So the issue may be some dependencies in your existing environment clashes with pydoclint. And the reason why things work in CI (as well as on my computer) is that every time pydoclint is run in a fresh environment. (When I run it on your repo on my computer, it's in fresh environments also.)

@jsh9 I had this issue with a fresh install too (I also thought it might be some other dependency).

Seems like based on @ssbarnea's comment that it's at least a Mac (if not M1)-specific issue.

[jsh9]

OK, so I found the issue thanks to @ssbarnea 's clue (foo/lib/python3.13/site-packages/pip/_internal/cli/cmdoptions.py). It's this file: https://github.com/pypa/pip/blob/3b91f42e461de3f23e9bed46a8c5695435f930fb/src/pip/_internal/cli/cmdoptions.py and it's indeed causing that issue in pydoclint.

I have a PR to fix it.

That said, an separate (orthogonal) issue is: why on your local machines would your test framework (tox or poetry) inspect unrelated files. I don't have an answer for this. Maybe it's a regression from some external dependencies. (Btw my computer also has Apple M chips, so maybe OS isn't the culprit here.)

[jsh9]

The PR is merged, and the change will be shipped with the new version (0.5.13), which will come out in a few hours. (I have another thing to include into the new version.)

[jsh9]

0.5.13 is published to pypi.

[mvadari]

The fix works for me now, thanks!