Filter out non-Python source files

[tjpnz]

I utilize CodeSpell to enforce US spelling as part of an internal Python coding standard. Most of the code bases it checks live inside mono repos where there'll be a lot of CPP source files and configuration. When I invoke CodeSpell it will find issues across all files, but I would like the ability for it to only check Python files.

What I've tried

Filtering out patterns using skip - not really practical since it expects me to know which files might exist in a repo ahead of time. I did notice that CodeSpell uses fnmatch under the hood which claims to support negative matches using !, but I wasn't able get this to work for matching non *.py patterns.

Passing a list of Python files - messy since I need to do a glob first and then filter out the Python files I want to skip (tests, files under .venv etc.)

What I would like to see

A global option to restrict checks to *.py files. It would respect both skip and any Python files I pass to the command.

---

Any pointers on how this could be achieved currently would be welcome! If it's not I would be open to pushing a PR if an approach could be agreed on.

[DimitriPapadopoulos]

Skipping Python files as such shouldn't be built in codespell: indeed it is not specific to Python projects.

So we need to find a way that doesn't imply specific options or code.

[julianstirling]

I suppose this could be achieved with a keyword like only:

[codespell]

only = *.py

This would allow @tjpnz to choose only python files, but seems general enough to be useful even in conjunction with skip:

[codespell]

only = *.md

# Skip CHANGELOG as it includes all PR titles, and many have spelling errors.
skip = CHANGELOG.md

[Valentina198955]

I utilize CodeSpell to enforce US spelling as part of an internal Python coding standard. Most of the code bases it checks live inside mono repos where there'll be a lot of CPP source files and configuration. When I invoke CodeSpell it will find issues across all files, but I would like the ability for it to only check Python files.

What I've tried

Filtering out patterns using skip - not really practical since it expects me to know which files might exist in a repo ahead of time. I did notice that CodeSpell uses fnmatch under the hood which claims to support negative matches using !, but I wasn't able get this to work for matching non *.py patterns.

Passing a list of Python files - messy since I need to do a glob first and then filter out the Python files I want to skip (tests, files under .venv etc.)

What I would like to see

A global option to restrict checks to *.py files. It would respect both skip and any Python files I pass to the command.

Any pointers on how this could be achieved currently would be welcome! If it's not I would be open to pushing a PR if an approach could be agreed on.