Fixing the right-to-left language issue

[Theamoz]

For example, all symbols instead of the end of the line go to the beginning of the line.

And in this example, where an English word was added to the text, the entire sentence became scrambled and ruined.

I have successfully resolved this issue by implementing this code. It is strongly recommended to incorporate it into your application to ensure proper functionality for right-to-left languages such as Arabic and Persian

import pysrt
from collections import Counter
import unicodedata as ud

def dominant_strong_direction(s):
    count = Counter([ud.bidirectional(c) for c in list(s)])
    rtl_count = count['R'] + count['AL'] + count['RLE'] + count["RLI"]
    ltr_count = count['L'] + count['LRE'] + count["LRI"]
    return "rtl" if rtl_count > ltr_count else "ltr"

def convert_srt_to_rtl(input_file, output_file):
    subs = pysrt.open(input_file)
    for sub in subs:
        if dominant_strong_direction(sub.text) == "rtl":
            sub.text = "\u202b" + sub.text + "\u202c"
    subs.save(output_file, encoding='utf-8')

if __name__ == "__main__":
    input_file = "subtitle.srt"
    output_file = "subtitle_rtl.srt"
    convert_srt_to_rtl(input_file, output_file)

These are two examples that show the issue has been successfully resolved and also works well for all right-to-left languages

[machinewrapped]

Thanks, I'm not too familiar with RTL languages but it looks like you have a solution that works. I'll see if I can integrate it into the standard flow. Is it just the exported subtitles that have the issue or does RTL text display incorrectly in the GUI too?

[Theamoz]

Both the subtitles and the GUI itself have issues, which are effectively resolved with the solution I provided.

[machinewrapped]

Which player are you using, by the way?

I think I've got this working - here's a before and after of some translated text in PotPlayer.

Before:

After:

It doesn't look completely wrong in either case to my untrained eye, but I guess the quotation mark and period are not where they're supposed to be in the first image?

[Theamoz]

I have tried several different players, but none of them have made a difference.
Yeah, you're right, in your example, the only problem is the dots. But if you add an English word to it, the whole sentence gets messed up.

[machinewrapped]

Support for adding RTL tags to the output is in v0.8.4:
https://github.com/machinewrapped/gpt-subtrans/releases/tag/v0.8.4

Changes at the GUI level would be a bit more complicated. In the test Arabic translation I did the GUI seems to match how the subtitles are displayed in PotPlayer with the RTL markers added, even with mixed content. If you can provide a sample subtitle which displays incorrectly in the GUI I can look at it in a future release.

[Theamoz]

thanks I'm going to test it

[Theamoz]

Hey bro, this works really well, awesome
Also, the GUI is fine, and I was wrong
It’s all good.