Language filters should use more than extension (including hints from file) when determining language of a file

[mike-r-mclaughlin]

Provided by: https://github.com/sourcegraph/accounts/issues/6716 [Recording: https://drive.google.com/file/d/14ZdQ3U9pKCsbkonyk0KkqWsHUH2g_D-R/view]

Feedback

A user ran a query that resulted in 109 results. All were YAML files. However in the filters section, the "languages" showed "MiniYAML 108". There was nothing indicating what the other 1 result was (until we filtered out MiniYAML to see what it thought it was). User asked: What is the difference between MiniYAML and YAML?

Feedback:

• Make it easier to figure out how Sourcegraph is determining what language a file is?(Q: Can Sourcegraph parse a hint at the top of the file?)
• Could Sourcegraph use something similar to what GItHub uses (linguist) for language detection?

/cc @sourcegraph/search-core

[jtibshirani]

We've made a lot of improvements here recently, including better handling for ambiguous extensions like .yaml.

We're also working to make language filters more precise by looking at the contents of a file, not just the filename: https://github.com/sourcegraph/sourcegraph/issues/60341. To keep things simple, I'm going to close out this issue in favor of that tracking issue.