Raw file language analysis doesn't load gitattributes data properly

[Nixinova]

With the recent change to using fileContent instead of cloning the repo, gitattributes data is no longer parsed, so vendored files are now affecting my language results. For example this file which is marked vendored has put 6% Beef into my stats. We need to think of a way to incorporate gitattributes data into this language analysis.

[Nixinova]

Maybe I need to also have an attributesContent option?

[lowlighter]

Yeah I think it may be the most suitable solution 👍

However the recent change should only impact the code snippet plugin I think, the languages plugin hasn't been updated yet (it stills clones repositories for both recent/most used stats, along with the .gitattributes)

[Nixinova]

Oh, does it? Hm. Then I have no idea where Beef is coming from here. I assume this problem is on my end then.

[Nixinova]

Found the issue maybe. But not classifying samples/ as vendored seems to be a bug that has been in linguist-js for ages, no idea why it only showed up now...

[Nixinova]

Well that was a rabbit hole Nixinova/LinguistJS@161a40a. Should be fixed in 2.3.2

[lowlighter]

Nice, thanks a lot for looking into it 😄 !