Multiple feature ideas for an extractor - how to issue them

[mpeter50]

I recently started to use the TwitchVod extractor regularly, and I have a few improvements in mind. I think I would be able to make the changes, but probably not too soon.
Is it ok if I open a new issue for every one of these at once? Or should I rather open a single issue, and make a pull request per improvement?

[pukkandan]

sorry, I totally missed this.

Without knowing what you have in mind, it is difficult to say whether they are best suited as a single PR or multiple ones. In general, if the improvements are trivial, feel free to bundle them into a single PR.

That is, assuming u will be making PRs. If you are only requesting the feature, always open one issue per feature. Otherwise it becomes impossible to track.

[pukkandan]

You can bundle the first 2 into a single PR.

Partial download is already possible using ffmpeg. See #686. There is also a feature request to add a proper CLI argument for it, #52. There are some edge cases that needs to be taken care of when implementing this. Hence why it has been pending for so long

Or did you have a non-ffmpeg based implementation of this in mind?

[mpeter50]

On the partial download I didn't have much of an idea, but, it depends.
What do you mean by a non-ffmpeg implementation? Without ffmpeg as the downloader, or without using ffmpeg at all?

My idea for it is to see how TwitchLeecher does it (a tool that I have been using for a long time for the purpose), and try to replicate it here.
TL only downloads the segmnets that contain the requested timeframe. After that, it merges the .ts files into a big .ts file, and then it converts it to an .mp4 one. It contains some other steps, like fine-cropping at the end, and it's a bit buggy because usually there's no video data in the first few seconds, so it's a still image of the first frame, but without knowing the source of this problem I think this might be solved by downloading one more segment in the beginning.

[pukkandan]

Without ffmpeg as the downloader. ffmpeg downloads are very slow compared to native and so I'd prefer to avoid it when possible

Using it later in the process is fine. Also, any implementation shouldn't be twitch-specific ofc (it is fine if it is hls specific though)

[mpeter50]

I have started looking through common.py, and I see that extract() checks for a live_chat key in the subtitles dict.
Should I store the stream's live chat there? Asking because it's not mentioned at the infodict explanation.

[pukkandan]

Yes. Save it as a subtitle with language = live_chat. It's not in the docs since the only other extractor that supports livechat is youtube