Add --live option - Githubissues

deldesir commented 2 months ago

Description

This pull request introduces a new --live flag to enable the download of live videos. This flag adjusts the match filters to include live streams, which were previously excluded by default.

Motivation and Context

The primary motivation for this change is to allow users to download live videos directly, a feature that was not supported due to default match filters excluding live content.

Some context: https://github.com/iiab/calibre-web/issues/188

How has this been tested?

The changes were tested by:

Implementing the --live flag and ensuring it correctly modifies the match filters.
Running the updated script to download live videos from various sources to confirm the flag's effectiveness.
Verifying that non-live videos are still downloadable without the --live flag.
Ensuring no regression in existing functionalities.
Tested on Ubuntu 24.04

Screenshots (if appropriate):

Types of changes

[x] Bug fix (non-breaking change which fixes an issue)
[x] New feature (non-breaking change which adds functionality)
[ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)

Checklist:

[x] My code follows the code style of this project.
[ ] My change requires a change to the documentation.
[ ] I have updated the documentation accordingly.
[x] I have added tests to cover my changes.
[x] All new and existing tests passed.

holta commented 2 months ago

@deldesir what happens when someone accidentally tries to download a "genuinely live" YouTube video — that's actually being recorded live at that moment — e.g. a video from a nature scientist's webcam, that might never end?

(As distinct from "formerly live" YouTube videos, which people are currently / unfortunately being blocked from downloading!)

deldesir commented 2 months ago

It will be downloaded from the current time. This is the default behavior. To download it from the start --live-from-start must be passed. For finished streams, the whole video will be downloaded though.

holta commented 2 months ago

It will be downloaded from the current time.

Are you saying the download will never end?

This is the default behavior. To download it from the start --live-from-start must be passed.

Interesting!

For finished streams, the whole video will be downloaded though.

Certainly that's the priority use case needing a fix:

My concern is that educators/parents who accidentally try to download "genuinely live" YouTube videos, should be able to gracefully recover:

Either we should inform them of their mistake, or
Download partial video using --live-from-start and/or
Give thoughtful consideration to educators/parents who accidentally download years-and-years of a nature scientist's continuous webcam video (probably not what the educator/parent wants — at least in many cases?)

deldesir commented 2 months ago

From yt-dlp README:

--live-from-start: "Download livestreams from the start. Currently only supported for YouTube (Experimental)"
--no-live-from-start: "Download livestreams from the current time (default)"

holta commented 2 months ago

It will be downloaded from the current time.

Are you saying the download will never end?

I had (wrongly) assumed YouTube could record videos of nearly endless duration, e.g. from a wilderness scientist's webcam.

In fact, YouTube video recordings seem to currently be limited to "12 hours long, or 128 GB" according to https://www.quora.com/Whats-the-maximum-video-length-on-YouTube-or-is-it-dependent-on-video-size

So, to clarify my question: let's say we're 1-hour into a 12-hour-long "actually live" tennis match, when a teacher/parent suddenly then asks Calibre-Web to start downloading the tennis match video — will that single video download take 11 hours to complete?
Regardless, can we use the --live-from-start flag by default? (Even if it downloads the 1st hour of the tennis match very rapidly, and then drags on very slowly while downloading the final 11 hours ???)

holta commented 2 months ago

Just FYI I'm trying to understand the true high-level objectives here.

So @deldesir is helping to explain some of those underlying motivations (and needs) in this conversation here: