How to specify numpydoc_validation_checks from CLI

[HealthyPear]

From the docs it is not clear if this is possible and my first tries failed.

If I do python -m numpydoc -hI see,

Implementing `python -m numpydoc` functionality.

positional arguments:
  import_path           e.g. numpy.ndarray

options:
  -h, --help            show this help message and exit
  -c CONFIG, --config CONFIG
                        key=val where val will be parsed by literal_eval, e.g. -c use_plots=True. Multiple -c can be used.
  --validate            validate the object and report errors

and from the docs I see e.g. the numpydoc_use_plots configuration option mentioned in the help example.

I would like to specify the equivalent of e.g. numpydoc_validation_checks = {"all", "GL01", "GL02", "GL05"} - is this possible?

If not it should be explained, if yes I think there should be more examples of usage.

I tried e.g.,

python -m numpydoc -c numpydoc_validation_checks=all -c numpydoc_validation_checks=GL01 -c numpydoc_validation_checks=GL02 -c numpydoc_validation_checks=GL05 --validate test_module.good_docstring

and

python -m numpydoc -c numpydoc_validation_checks{"all", "GL01", "GL02", "GL05"} --validate test_module.good_docstring

[echedey-ls]

The problem here is that no config is being passed onto the validator when --validate is called.
See this line in __main__.py.

[HealthyPear]

Oh ok if so it seems an easy fix, I can try and make a PR if you want

[echedey-ls]

I don't maintain this project, but I'm looking into how to do something similar. You'll find that the function validate does not have arguments of which rules to exclude/include. I will stop working on this for now, but I would look into how sphinx is filtering the output, in case you want to continue.

[rossbar]

See also #467 - there is indeed a difference in the (now) 3 interfaces to the validation functionality (cli, pre-commit, and sphinx-build). #476 should help, but there will still be room for improvement. Ultimately I think there's an opportunity to refactor the validation to make it more modular - one of the big disadvantages currently is that all the validation checks run regardless of configuration - the configuration just dictates what's reported.