Feature Request: Regex lists

[Repiteo]

Seeing as this project is already diverging from pre-commit on several foundational levels, I want to propose an addition that is no longer planned for the OG project: regex lists. This functionality was originally considered for the pre-commit project¹, but was ultimately shelved for concerns of mixed syntax. Seeing how far tools like intellisense and schema validation has come in that time, as well as list support already being integrated for the prek-exclusive glob patterns, I think it warrants revisiting.

This would most prominently replace verbose regex, which often doesn't play nice with version control:

exclude = '''(?x)(
  ^thirdparty/|
  -(dll|dylib|so)_wrap\.[ch]$|
  ^platform/android/java/editor/src/main/java/com/android/|
  ^platform/android/java/lib/src/main/java/com/google/
)'''

The above syntax could instead be implemented as:

exclude = [
  "^thirdparty/",
  '-(dll|dylib|so)_wrap\.[ch]$',
  "^platform/android/java/editor/src/main/java/com/android/",
  "^platform/android/java/lib/src/main/java/com/google/",
]

Footnotes

1. Support files / exclude as both regex / list of regexes pre-commit/pre-commit#149 ↩

[j178]

I don’t think we need to add a separate regex-list form for now.

prek already supports glob and glob-list syntax for file path matching: https://prek.j178.dev/reference/configuration/#top-level-files

For file paths, glob patterns are usually a better fit than regex: they map directly to path segments, avoid regex escaping/anchoring details, and the existing glob-list form already covers the common “match any of these paths” case.

Regex strings are still supported for compatibility and for cases that need regex-specific features. Unless there is a concrete case that cannot be expressed well with the existing regex string, verbose regex, or glob-list syntax, I’d prefer not to add another config shape for regex lists.