Inconsistent hearing impaired/non-hearing impaired flagging about bazarr HOT 2 OPEN

Yeah, I've been aware of that for a while but, probably out of laziness, I never fixed it.

I think that, while downloading a subtitles file, we should always validate if it's HI or not. The file should be named accordingly, not using the language known by the provider. guess_external_subtitles should then return the same value on each execution. Make sense?

from bazarr.

It looks like the subtitle is saved to file based on what the provider thinks the language is. If it thinks it's English but not hearing impaired it'll get saved as *.en.srt for example

After that, guess_external_subtitles is called which reads the file and finds it to be hearing impaired. It is then saved to the database as being hearing impaired.

On re-scan, the language of an external subtitle is initially set based on its filename, which in this case is English non-hearing impaired as it's still just *.en.srt.

When guess_external_subtitles is called again, it does not read the file to change the language to hearing impaired because of the previously_indexed_subtitles_to_exclude argument. This makes it forever treated as non-hearing impaired.

My questions here are:

1. Should the provider be trusted and the file always treated as non-hearing impaired? Or should the file scan during guess_external_subtitles be consistent from then on and it always being treated as hearing impaired?
2. Should the result of guess_external_subtitles change the filename to match what it thinks it is?

from bazarr.