Add sub-text-all property

[karelrooted]

Currently there has property sub-text https://mpv.io/manual/master/#command-interface-sub-text to observe the text of the current sub. It would be nice to have a similar property for the entire subtitle, possibly called sub-text-all

the usage of this feature may be to get the entire embedded subtitle and then translate it and add a translated subtitle track on the fly

Expected behavior of the wanted feature

add a sub-text-all property to return the current entire subtitle as text

[po5]

Not possible for many files to get all subtitle lines without parsing through the entire file, as the data can be scattered across the file near each video/audio data chunk. You'll notice that ffmpeg can extract some subtitles near instantly and others will take a while.

[llyyr]

the usage of this feature may be get the entire embedded subtitle and then translate it and add a translated subtitle track

This is completely out of scope for mpv, you're looking to use something like Aegisub for .ass files, and if it's not an .ass file then you should convert it with ffmpeg.

[karelrooted]

the usage of this feature may be get the entire embedded subtitle and then translate it and add a translated subtitle track

This is completely out of scope for mpv, you're looking to use something like Aegisub for .ass files, and if it's not an .ass file then you should convert it with ffmpeg.

when try to add bilingual subtitle track on the fly with embedded subtitle, currently have to use ffmpeg to get the entire subtitle, it seems unoptimized when mpv is already loading the file

[Dudemanguy]

mpv has the file loaded, but it won't read the entire thing at once of course. It only reads what is necessary for playback. For mpv to do what you want, it would have to read the entire file from beginning to end again which is no different from just using ffmpeg to read the whole thing.

[karelrooted]

mpv has the file loaded, but it won't read the entire thing at once of course. It only reads what is necessary for playback. For mpv to do what you want, it would have to read the entire file from beginning to end again which is no different from just using ffmpeg to read the whole thing.

copy that

[christoph-heinrich]

What if the property only returns the subtitle lines that mpv already knows about?
That would make my script much simpler, faster and more reliable. Currently it walks through the subtitle lines with sub-step and then reads sub-start, sub-end and sub-text.

[Dudemanguy]

mpv only holds onto a couple of subtitle packets at a time. Anything older gets discarded, so I don't think there's any way for this other than reading through all the packets again which I'm not sure we want to do.

[christoph-heinrich]

Well we could just not discard them 🙃 I don't think the memory overhead for some text would be a problem.

[Lypheo]

Anything older gets discarded

Where? Afaict all events are buffered unless the ASS_Track is flushed which only happens when we reset the sub track.

[Dudemanguy]

I thought libass internally discarded stuff after a while. But sure I could be wrong. Feel free to send a patch if it's not so bad.

[karelrooted]

this commit added a sub-text-all, secondary-sub-text-all property

I'm not a experience c coder, so the code may need some clean up to conform to mpv's code standard, but it works on my local bilingual player (not production tested though, because I'm the only user who use it)