[aggregate] Support for non-plaintext file formats #1405
Description
Relevant issues
Supersedes #277, #342, #422, #523, #524, #525, #533, #542, #555, #567, #803, #945, #971, #1355.
Brief
Formats requested so far include (La)TeX, reST, markdown, emails, Python, math blocks, XML, and perhaps programming languages in general.
The problem lies in the execution: supporting these would require incorporating parsers and AST handling for each filetype into proselint itself. That would impose a substantial maintenance burden, bloat the program, and is far beyond proselint's intended scope.
If anybody has suggestions or reference materials on the matter, please do let me know.
Suggestions
In general
It has been suggested in #393 and #443 that the typography checks should be disabled by default to make working with other file formats easier. Alternatively, proselint could disable it by default specifically for non-plaintext files.
A solid general solution would be an external tool that parses these formats and multiplexes linters over relevant AST nodes, similar to how treefmt works with entire files of given types. That would allow proselint, and all other linters, to format segments of files that are relevant to them specifically; be it text, code, or other blocks. I am not aware of any prior art on that front, but it is an interesting consideration. Just in case, though, I know of Rust AST handlers for (La)TeX and markdown that could be integrated after porting.
Markdown
We could consider adding the single and triple backticks used in markdown code blocks to the ignored quote spans regex, which would remedy problems such as in #971.
(La)TeX
@tani (thank you) has suggested external tools that can accomplish this.