fix: catch recursion in Alias.resolved

[tlambert03]

hey @pawamoy

I'm sure this is probably your least favorite bug at this point 😂 ... but I ran into an Alias resolution recursion error, similar to #83

Full traceback:

Traceback (most recent call last):
File "/Users/talley/mambaforge/envs/mmcore/bin/mkdocs", line 8, in <module>
sys.exit(cli())
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/click/core.py", line 1130, in __call__returnself.main(*args, **kwargs)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/click/core.py", line 1055, in main
rv =self.invoke(ctx)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/click/core.py", line 1657, in invokereturn _process_result(sub_ctx.command.invoke(sub_ctx))
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/click/core.py", line 1404, in invokereturn ctx.invoke(self.callback, **ctx.params)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/click/core.py", line 760, in invokereturn __callback(*args, **kwargs)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/__main__.py", line 234, in serve_command
serve.serve(dev_addr=dev_addr, livereload=livereload, watch=watch, **kwargs)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/commands/serve.py", line 83, in serve
builder(config)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/commands/serve.py", line 76, in builder
build(config, live_server=live_server, dirty=dirty)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/commands/build.py", line 308, in build
_populate_page(file.page, config, files, dirty)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/commands/build.py", line 181, in _populate_page
page.render(config, files)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocs/structure/pages.py", line 270, in renderself.content = md.convert(self.markdown)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/markdown/core.py", line 264, in convert
root =self.parser.parseDocument(self.lines).getroot()
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/markdown/blockparser.py", line 90, in parseDocumentself.parseChunk(self.root, '\n'.join(lines))
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/markdown/blockparser.py", line 105, in parseChunkself.parseBlocks(parent, text.split('\n\n'))
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/markdown/blockparser.py", line 123, in parseBlocksif processor.run(parent, blocks) isnotFalse:
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocstrings/extension.py", line 121, in run
html, handler, data =self._process_block(identifier, block, heading_level)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocstrings/extension.py", line 195, in _process_block
data: CollectorItem = handler.collect(identifier, options)
File "/Users/talley/mambaforge/envs/mmcore/lib/python3.10/site-packages/mkdocstrings_handlers/python/handler.py", line 199, in collect
unresolved, iterations = loader.resolve_aliases(implicit=False, external=False)
File "/Users/talley/dev/forks/griffe/src/griffe/loader.py", line 217, in resolve_aliases
next_resolved, next_unresolved =self.resolve_module_aliases(module, implicit, external)
File "/Users/talley/dev/forks/griffe/src/griffe/loader.py", line 357, in resolve_module_aliases
sub_resolved, sub_unresolved =self.resolve_module_aliases(
File "/Users/talley/dev/forks/griffe/src/griffe/loader.py", line 338, in resolve_module_aliases
member.resolve_target() # type:ignore[union-attr]
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 967, in resolve_targetself._target.aliases[self.path] =self# type:ignore[union-attr]# we just set the target
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 827, in __getattr__
attr =getattr(self.target, name)
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 935, in targetifnotself.resolved:
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 976, in resolvedreturnself._target isnotNoneand (notself._target.is_alias orself._target.resolved) # type:ignore[union-attr]
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 976, in resolvedreturnself._target isnotNoneand (notself._target.is_alias orself._target.resolved) # type:ignore[union-attr]
File "/Users/talley/dev/forks/griffe/src/griffe/dataclasses.py", line 976, in resolvedreturnself._target isnotNoneand (notself._target.is_alias orself._target.resolved) # type:ignore[union-attr]
[Previous line repeated 971 more times]
RecursionError: maximum recursion depth exceeded

In this case, it does appear to be a legitimate Cyclic Alias... that is self._target._target is self.

For what it's worth, the class I was trying to document with mkdocstrings is from pymmcore (pip install pymmcore)

::: pymmcore.CMMCore

... and that class is a SWIG wrapper with a .pyi type stub providing the signatures and docstring. And the cyclic aliases look like:

<Alias('CMMCore', 'pymmcore.pymmcore_swig.CMMCore')>
<Alias('CMMCore', 'pymmcore._pymmcore_swig.CMMCore')>
<Alias('CMMCore', 'pymmcore.pymmcore_swig.CMMCore')>
<Alias('CMMCore', 'pymmcore._pymmcore_swig.CMMCore')>
...

I opened this as a PR rather than an issue, since this does fix the docs building (the page looks exactly how I'd expect it to)... but I'm not sure what the actual best fix is. Feel free to close or suggest a better approach here for me to implement

[pawamoy]

least favorite bug

Not agaaaaaaaaaaaaaaaaaaain 😂
Yeah I hate this error, but that's entirely my fault 👐
I'm working on fixing the design (implicitness triggering alias errors all the time), but I struggle to find something explicit
that would also not require wrapping every attribute access in a try/except or contextlib.suppress block 🤔

For now I'll continue to apply bandages on this poor project 😢

Thanks for the report and PR!

Catching recursion errors is like the absolute last resort haha.

I read SWIG wrapper + type stubs, I understand compiled module?
I recall having dealt with a similar case in Numpy. Here's the special case:

griffe/src/griffe/agents/inspector.py

Lines 126 to 127 in 25060f6

	orparent_module_name=="numpy.core._multiarray_umath"
	andchild_module_name=="numpy.core.multiarray"

Maybe a similar situation here.

[tlambert03]

Catching recursion errors is like the absolute last resort haha.

yep. I changed this PR to an attribute check, and that also fixes mkdocs build for me. But if you'd prefer to close it and look into a different fix, I won't be offended in the least :) and can open a bug issue as a reminder.

I read SWIG wrapper + type stubs, I understand compiled module?

yep, compiled module.

Maybe a similar situation here.

yeah I think so (pymmcore.pymmcore_swig.CMMCore -> pymmcore._pymmcore_swig.CMMCore -> pymmcore.pymmcore_swig.CMMCore)

[pawamoy]

Closing in favor of 9a2a711 🙂

[tlambert03]

wow 😮 ... special casing it huh? don't think it's gonna happen again?

[pawamoy]

It will probably happen again, but I want to improve the alias resolution mechanism, and maybe provide a way for registering cycles in this new "cyclic relationships" map, so that project suffering from it can fix it themselves.