Add typos action

[PodushkaPIR]

• Add base typos-check in CI and put tool.typos in pyproject.toml
• Fix suggestion from typos

As it turns out, typos does not yet support layered configurations for working with specific directories and files, although there issues about this. Therefore, I had to create one general configuration file for the entire project. In this regard, some rules in the pyproject.toml file may seem strange.

If it is better to add some files to the ignore list or get it from, I am open for suggestions.

[BUYT-1]

Looks good, though I'd prefer a more granular approach. Ideally, even more granular than file-by-file basis, whitelisting each false positive explicitly. I'm thinking extend-ignore-re with enough context that it will never reasonably appear in another text or that it will always be a typo (e.g. \"mispelling\" is a misspelling). We can always add stuff to whitelists.

[BUYT-1]

Why are these ignored in all project files?

[PodushkaPIR]

Because there is only one config and the rules from the config should apply to all files in a good way.

[BUYT-1]

https://github.com/crate-ci/typos/blob/master/docs/reference.md:

Configuration is read from the following (in precedence order)

• Command line arguments
• File specified via --config PATH
• Search parents of specified file / directory for one of typos.toml, _typos.toml, .typos.toml, Cargo.toml, or pyproject.toml.
  • In pyproject.toml, the below fields must be under the [tool.typos] section. If this section does not
    exist, the config file will be skipped.
  • In Cargo.toml, the below fields must be under either [workspace.metadata.typos] or [package.metadata.typos]

Does this action apply only the first one it detects from the list or all of them but in order?

[PodushkaPIR]

I tested and action applied only one config which was detected (if I understood the question correctly)

[BUYT-1]

Why is this word excluded? "identificator" is not a real word, "identifier" and "identification" are.

[PodushkaPIR]

Yes! I will fix this.

[BUYT-1]

I think it would be better to only ignore words per-file/directory, not project-wide. Let's say, ignore "ND" for src/core/algorithms/nd/*, ignore "nd" for the Python ND example, and other ND-related places, and don't ignore it in others (extend-ignore-words-re seems to not ignore case, so that can be used).

[PodushkaPIR]

Okay, I'll do that. However, since typos can't apply specific rules to specific files and directories, I think it needs to use a matrix in CI with different scenarios.

[PodushkaPIR]

Would it be a good idea to create different files for specific scenarios (e.g. typos_nd.toml, typos_other.toml, etc)?

[BUYT-1]

Why are typos ignored in README files?

[BUYT-1]

This is a tough one. Would it be too hard to dump notebook cell text to files (.py and .md so the action understands) and check those instead?

[PodushkaPIR]

Yes, you're right.

[BUYT-1]

I think this should have a more granular whitelist instead.

[BUYT-1]

"nd" instead of "and" is a legitimate typo, though

[BUYT-1]

If it's too much of a pain to do fully, maybe we can merge as is and deal with the stuff I've pointed out later.

[PodushkaPIR]

If it's too much of a pain to do fully, maybe we can merge as is and deal with the stuff I've pointed out later.

No, it shouldn't be difficult, I'm a little busy right now, sorry(
I'll try to merge all your points