Added an option to export channels thumbnails for Jellyfin

[InterN0te]

Added an option to download channel thumbnails to poster.jpg and season-poster.jpg for Youtube channels :

This uses BeautifulSoup to find the thumbnail URL and only works for Youtube channels (playlists don't have thumbnails)

These 2 images are then used by Jellyfin:

I didn't manage to hide this option while adding a Playlist but it is ignore if it is

[meeb]

Thanks! This looks reasonable to add poster support for Jellyfin. I'll comment on inline.

[meeb]

Likely this needs nicer error handling. This will bubble up as a 500 probably if someone ticks the tickbox on a playlist.

[InterN0te]

This function is only called if the tickbox is checked AND the source is not a playlist, so it cannot be triggered until it is called from somewhere it should not

[meeb]

Ah right, yes I see. Could probably be a Django-flavoured exception at least then?

[meeb]

This is the only concern I have with the commit. How stable is using requests.get to load pages from YouTube? This needs confirmation it's reliable, as well as wrapping in some exception catches.

[meeb]

Oh and the HTTP status code probably needs a check.

[meeb]

This likely needs to be wrapped in a KeyError catch to politely handle YouTube changing their JSON format.

[meeb]

This reminded me, I believe yt-dlp extracts the posters/banners for channels and adds them to the metadata for media items. There are usually very large images in the "thumbnails" metadata key with resolutions of 1920x1080. Are these the same as the posters fetched above? Could these be used instead?

Where possible it would be highly preferable to use yt-dlp's well maintained upstream for features.

[meeb]

And wrap this to catch ValueError and TypeError in case the JSON found is invalid.

[InterN0te]

You're right, YT-DLP can do it. I switch to use this

Test :

import json
import yt_dlp

URL = 'https://www.youtube.com/playlist?list=PLyHWd4LxGoS8cdwh0unZXqVA3AqHoeDPU'

ydl_opts = {'extract_flat': True}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
    channel_info = ydl.extract_info(URL, download=False)
    
    banner_url = None
    avatar_url = None
    for thumbnail in channel_info['thumbnails']:
        if thumbnail['id'] == 'banner_uncropped':
            banner_url = thumbnail['url']
        if thumbnail['id'] == 'avatar_uncropped':
            avatar_url = thumbnail['url']
        if banner_url != None and avatar_url != None:
            break

print(banner_url)
print(avatar_url)

[InterN0te]

It is now perfect :

I still have to look for the Django-flavoured exception on playlist images request
I have no clue on how to do it. Is it just a declaration in the file tubesync/common/errors.py ?

[meeb]

One final comment here, this info is almost certainly in the metadata stored in Media.metadata - you can likely just load it straight out of the JSON without having to invoke yt-dlp at all.

[meeb]

I still have to look for the Django-flavoured exception on playlist images request I have no clue on how to do it. Is it just a declaration in the file tubesync/common/errors.py ?

Some exceptions defined in Django are handled better (auto-modify HTTP status codes, logging, etc), that's all I meant. Raise a SuspiciousOperation exception rather than a plain Exception:

https://docs.djangoproject.com/en/4.2/ref/exceptions/#suspiciousoperation

[meeb]

I'll merge this for now, thanks for the feature PR, most appreciated. I'll rework your function that uses yt-dlp again to use the already downloaded metadata at a later date.