ass to srt issue about pysubs2

[josscmorera]

Works great thank a lot, all good for now so I proceed to close this issue. :)

[tkarabela]

Hi, I've replicated your result with current release (1.0.0, 7159a89):

python -m pysubs2 --to srt original.ass

and I got the same output as you did (5.srt).

• In the original ASS file, there are three kinds of lines: (1), (2) and (3), where only (1) should ideally be written to the SRT file.
• (2) is being skipped in the current version, so that's good.
• As for (3), I'm not sure how pysubs should tell the difference between (1) and (3)? Both are dialogue lines with some override tags.

---

(1) Regular dialogue, possibly with override tags

These are handled OK, tags are stripped.

Dialogue: 0,0:24:06.04,0:24:11.01,Signs,,0,0,0,,{\fnAdobe Garamond Pro\b0\bord4\blur4\c&H060706&\3c&H060706&\fs73\pos(960,654.143)}Warrior

2635
00:24:06,040 --> 00:24:11,010
Warrior

(2) ASS drawing tags, {\p}

These are handled OK, skipped for SRT output (thanks to #31).

Dialogue: 13,0:00:20.65,0:00:20.85,OP-E,,0,0,0,,{\an5\pos(926.695,148.500)\fad(50,50)\bord0\1c&HFFFFFF&\blur2\alpha&H53&\p1}m 0 5 b 0 5 0 0 5 0 5 0 10 0 10 5 10 5 10 10 5 10 5 10 0 10 0 5

(nothing)

(3) Rendered karaoke lines

These are written to the output SRT, which is unfortunate.

Dialogue: 10,0:00:07.24,0:00:10.03,OP-E,,0,0,0,,{\an5\pos(655.758,142.500)\1c&H4E4AFB&\3c&H4E4AFB&\blur2.4\bord2.5\fad(300,300)}I

1
00:00:07,240 --> 00:00:10,030
I

[josscmorera]

would it be possible to force-delete the lines that gives out only 1 letter?

[tkarabela]

would it be possible to force-delete the lines that gives out only 1 letter?

Sure :) I'd make it optional though, so that no visible subtitles are deleted by default. Would adding a new commandline parameter, something like --clean, work for you? Ie.:

python -m pysubs2 --to srt --clean original.ass

[josscmorera]

that would be really awesome :)

[tkarabela]

I've released version 1.1.0 of the library, which includes --clean :) You can test it and then we can close the issue.

[josscmorera]

I just tested it and works amazingly, while testing it I found this:

00:25:35,120 --> 00:25:43,810
涙ã�ªã‚“ã�¦è¦‹ã�›ã�ªã�„ã‚“ã�

Not sure if this could be related but let me know if yes or not to close this issue
Thank you
Archive 21.zip

[tkarabela]

Oh, I see. That's not related to --clean, it's the character encoding. Please try:

python -m pysubs2 --to srt --clean --input-enc utf8 original.ass

That should fix the problem. I will look into this if it could be handled better in the CLI for some future release.

[toprak]

@tkarabela it does not work on me as expected on library mode.
it adds karaokes that are only 2 letter

def extract_and_convert_subs(mkv_file, sub_index, untranslated_path):
base_file = mkv_file.with_suffix('')
extracted_sub_file = untranslated_path / f"{base_file.name}.ass"

# Extract the subtitle in its original format
subprocess.run(['ffmpeg', '-hide_banner', '-loglevel', 'error', '-i', str(mkv_file), '-map', f'0:s:{sub_index}', f'{extracted_sub_file}'], check=True)

# Check if the subtitle format is ASS and convert it to SRT if necessary
with open(extracted_sub_file, 'r', encoding='utf-8') as f:
    first_line = f.readline()

if first_line.startswith('[Script Info]'):
    ass_subs = pysubs2.load(str(extracted_sub_file), encoding="utf-8")
    srt_file = untranslated_path / f"{base_file.name}.srt"
    ass_subs.remove_miscellaneous_events()
    ass_subs.save(str(srt_file), format_='srt', encoding="utf-8")
    os.remove(str(extracted_sub_file))
else:
    extracted_sub_file.rename(untranslated_path / f"{base_file.name}.srt")