feature_request(ignoring): ignoring words based on their context

[Kristinita]

Related issues: #410, #2491, #3673.

1. Summary

It would be nice if Codespell will allow ignoring words based on their context.

2. Justification of the need of the feature

This was discussed in several issues of this repository. It would be nice to have behavior where Codespell ignore words solely in specific cases but continues to return warnings for all another cases.

3. Example of desired behavior

Example YAML configuration file with keys context and extensions. Examples in this file are false positives from my project.

# [OVERVIEW] Example of a Codespell configuration file for ignoring words based on their context:
#
#
# [KEYS] Both keys are optional, but at least one of them must exists.
#
# 1. “context” — the list of contexts in which the word is ignored. List items are case-sensitive.
# If the key doesn’t exist, Codespell will ignore the word in any context.
#
# 2. “extensions” — the list of extensions for which the word is ignored.
# If the key doesn’t exist, Codespell will ignore the word for any extension.


#############
# Example 1 #
#############
# [PURPOSE] I want to get Codespell warnings for the “colour” word
# except the case where this word is used in the phrase “Colour Contrast Checker”.
#
# [INFO] “Colour Contrast Checker” is the name of the online application:
# https://github.com/Pushedskydiver/Colour-Contrast-Checker
colour:
  context:
  - Colour Contrast Checker


#############
# Example 2 #
#############
# [PURPOSE] I want to get Codespell warnings for the “isnt” typo,
# but not for the CoffeeScript built-in operator “isnt”:
# https://coffeescript.org/#operators
isnt:
  extensions:
  - coffee


#############
# Example 3 #
#############
# [PURPOSE] I want to get Codespell warnings for the “nd” typo,
# but not for the class “.nd” from the Pygments library in my CSS and Stylus files:
#
# [NOTE] I preferred more informative class names than “.nd”, but Pygments developers have the another opinion:
# https://github.com/richleland/pygments-css/blob/146834e1f9476d517c10cee5f15381047866d93d/monokai.css#L34
nd:
  context:
  - .nd
  extensions:
  - css
  - styl


#############
# Example 4 #
#############
# [PURPOSE] I want to get Codespell warnings for the “whos” typo,
# but not for the IPython built-in magic command “%whos”:
# https://ipython.readthedocs.io/en/stable/interactive/magics.html#magic-whos
whos:
  context:
  - "%whos"
  extensions:
  - py

If this file exists, Codespell will ignore colour, isnt, nd and whos for contexts provided in the file.

It would be nice if Codespell will support something like this configuration file.

4. Alternatives of context+extension ignoring

4.1. Inline ignoring

I’m not sure that codespell:ignore comments is the best solution for users.

1. If a user gets the same false positive multiple times, he needs to add codespell:ignore comments multiple times. If a configuration file as in my example would be allowed, it would be enough for a user to make one record in a configuration.
2. I try not to use inline ignoring at all and prefer ignoring via the configuration if it’s possible. In my opinion, it would be nice if text files contains solely texts and code files contains solely code without records especially for checkers/linters/validators.
3. Unfortunately, currently Codespell by default doesn’t provide an opportunity ignoring words based on their context. But I hope that someday it will become possible, and users can make pull requests based on their personal configurations for cases like %whos from my example.

4.2. Per-file ignoring

In my opinion, it would be nice, that the pull request #3675 allows ignoring based on extensions, but I’m not sure that ignoring per file is the best idea. The file may contain both false positives and real typos, and if a user activate the option --per-file-ignores, he may skip real typos. The context key from my example would allow more accurate ignoring.

Thanks.

[DimitriPapadopoulos]

How about using option --exclude-file?