Incorrect and seemingly unrelated language detection
scourgeofgrozny opened this issue · 2 comments
Version: 2022 Community 17.7.3
Issue:
Let me preface this by saying that I haven't opened any kind of Ezhil (.n) nor GDScript file (.gd) ever, not even before installing the WakaTime VS extension.
.gitignore and .sln files I've opened on Visual Studio for some reason are being shown as random languages, namely GDScript and a Tamil script programming language. Furthermore, they comprise a small percentage of time I've spent on the languages chart despite being 0 seconds. Is this a known phenomenon? I'm not super well-versed in how the language recognition/time-spent analysis works for Wakatime, but let me know if any other kind of information would be useful.
We recently added more involved language detection where we analyze the file contents looking for symbols matching a language syntax. It used to only use file extension, but we had some unknown languages. Now we need to make sure we're not matching languages too eagerly.
Duplicate of #943 which was just fixed.