Append language to subtitle name if missing, handle arbitrary extensions

[joelpurra]

• May add the putio api language code.
  • Ignored if the language code is not a valid language.
  • Only included if the filename does not already include the same language code, as detectable by Kodi.
  • Appends the language code to the file's name, before the file's extension (Subtitle filenames have language code appended after .srt file extension #22).
• Handles arbitrary subtitle extensions, including:
  • .srt/application/x-subrip: current default.
  • .vtt/text/vtt: available in the putio api.

Based on original code in #23 by @berkanteber.

Closes #22.

See

• Subtitle filenames have language code appended after .srt file extension #22
• https://app.swaggerhub.com/apis-docs/putio/putio/2.8.13#/files/get_files__id__subtitles__key_
• https://en.wikipedia.org/wiki/SubRip
• https://en.wikipedia.org/wiki/WebVTT
• improve subtitle filename rename #23
• 8efb911

[joelpurra]

@berkanteber: sorry, code got stuck in "testing" since last year.

As mentioned, it's based on your code so you can see what I meant in my code review. (To be honest, I've forgotten most of the discussion from then.) Hope you find the code agreeable. Let me know if you would prefer a standalone pull request.

Also tested to never include any language code in the filename, but as you said kodi's language detection isn't always working. Ended up checking if the filename is already included the language code (the way it would be detected by kodi), and appending it if not.

[berkanteber]

Hey, sorry for late response. Just looked at what we've discussed earlier. Changed my mind in couple of things, but not all.

• Unknown languages: These are most probably folder subtitles (very few cases for other 2). I now think it's acceptable if we don't add und to these. If they are in proper format, they'll work OK, in edge cases users can rename from put.io.
• Known languages: I think we shouldn't use any extra logic here and always append.
  • I don't know what happens if filename is foo.eng.HI.srt and language_code is eng. If we don't append again, do we have English or Hindi?
  • We send full name or 3-letter codes, which misses 2 letter codes. If filename is foo.en.srt and language_code is eng, we still append it "unnecessarily". Converting these codes are also not an option unless Kodi has some builtin conversion table.
  • It's not desirable to aways think about "what happens if..." in such cases.
  • Kodi will always strip language from name, so there won't be duplicates in name. There may be in filesystem, but I think the important thing is how it is in Kodi. Other than easy fixes like .srt always at the end, we shouldn't think about it.
• Extensions: Doesn't matter much since all are .srt anyway. (All our subtitle formats are srt, we can specify to download as webvtt, in which case we'll convert on the fly. Since we don't specify here, all returns .srt. We could also explicitly request srt.) But agree, not hardcoding .srt looks better.

[joelpurra]

• Known languages: I think we shouldn't use any extra logic here and always append.

Deliberately choosing to duplicate the language identifier makes little sense:

- Video.eng.srt
+ Video.eng.eng.srt

https://github.com/putdotio/putio-kodi/pull/24/files#diff-a6b15b003d6e376a2b0b6757c2c12a4b3b160bc424a981d638475a176c430b36R217

-if subtitle_lang != None and subtitle_lang.lower() not in re.split(r"[. -]", subtitle_name.lower()):
+if subtitle_lang != None:
     subtitle_parts.append(subtitle_lang)

Guess this is the change you suggest. I'd say the logic is simple enough to keep, but sure, if you insist.

[berkanteber]

Checked a sample from DB.

Out of 108,678 subtitles:

• 8,983 matches for language code (2-letter) in name parts (we don't return these from API)
• 7,377 matches for language code (3-letter) in name parts
• 2,993 matches for language name in name parts

So, duplicates would be < 10%.

I don't know what happens if filename is foo.eng.HI.srt and language_code is eng. If we don't append again, do we have English or Hindi?

Checked this and Kodi takes the last one, so foo.eng.HI.srt would be identified incorrectly. We would need to handle all these special cases or only take the last part. If we only take the last part, it would be 7,849 (3-letter or full) matches, making it 7.2%.

Let's have it only for checking the last part then.

[berkanteber]

We can do:

subtitle_parts = [subtitle_name]
if ...:
   subtitle_parts.append(lang_code)
subtitle_parts.append(subtitle_ext)
subtitle_fullname = '.'.join(subtitle_parts)

[berkanteber]

Let's check only last part (if has parts).

Also, we can check subtitle['language'].

Does returning also the 2-letter code makes sense? If we do, what should it be called (without touching others) in your opinion?