Closed Tydoki closed 4 years ago
I second this, it seems to be arbitrary which videos are available. e.g. its not always the most recently uploaded. e.g. I have a script that runs daily. On one of the channels, assuming there are 10 new videos on this channel for today, video 4 and 7 were the only 2 to be recognized as existing, let alone downloaded. This first happened 7/24. Script is ran via cron at 2340 EDT nightly. Videos were still not found the following afternoon (7/25)
I tried to look into the HTML (assuming that is how the videos are found based on what i saw in the YDL source) to find if there are any differences between the video links that were found vs those that were not, and I could not see anything.
Update: Running the script manually today(7/26), only some of the videos from 7/24 are populating when running the script, most are not. This is true for the 15 (lol) channels I monitor daily. I am basing my findings on the found video id's when the channel page is initially downloaded to scan for matching date/filter preferences. Playlist pages (only using one at this time) seem to be populating as expected.
More Updates: I think I see what is happening. Using OP's channel example and cross checked against 3 of mine:
The given link is:
https://www.youtube.com/c/OfflineTVgg/videos
What YDL is using is:
https://www.youtube.com/playlist?list=UUDK9qD5DAQML-pzrtA7A4oA
which is the channel's UPLOADS playlist. for some reason YouTube, or the creator(s), are not populating EVERY video for the day. I can only hypothesize it has something to do with a change in the publish/recommendation algorithms.
Searching for a workaround, will post back.
The "offending" code seems to be in youtube.py per the comment `
# Workaround by extracting as a playlist if managed to obtain channel playlist URL
# otherwise fallback on channel by page extraction
on lines [2956-2958](https://github.com/ytdl-org/youtube-dl/blob/a115e07594ccb7749ca108c889978510c7df126e/youtube_dl/extractor/youtube.py#L2956) this behavior is intentional. I am trying to find the specific place where it makes the change (i'm still barely an intermediate in python) to the playlist vs the channel page. I am looking at adding a key
--force-channel-page` and then a bool statement to skip the playlist page swap if 'True'
This way it will only do it if the user chooses and not by default.
edit: added links to comment and module
Update: I'm probably wrong, sigh.
None of today's videos seem to have been added, but all of the videos up until yesterday seem to have been appropriately populated. This is annoying.. lol
The "offending" code seems to be in youtube.py per the comment
Channel by page listing is restricted to 35 pages of 30 items, i.e. 1050 videos total (see #5778) # Workaround by extracting as a playlist if managed to obtain channel playlist URL # otherwise fallback on channel by page extraction
on lines 2956-2958
yeah probably not the issue since the channel I used has only like 120 videos ...
Yeah the latest video was finally added to the upload playlist for the channel I used (OfflineTV) like 6-22 hours later Problem still exists though from testing on channels with new uploads
This isn't really a duplicate of https://github.com/ytdl-org/youtube-dl/issues/26056
But the problem is probably caused by the same issue.
Actually, I'm pretty sure that it's something wrong on YouTube's end. I was on the verge of submitting this bug report myself, but then I found that, if you go to the channel's 'Videos' tab, and then click the 'play all', you get a playlist that starts playing their actually most recent video, but the playlist on the side shows that it thinks you're currently playing the most recent video that youtube-dl sees, and labels it #1 on the list, and the actually newest isn't there at all, even though it's playing it right there. I'm used to waiting 20 minutes to an hour for uploaded/published videos to be listed on the playlists, but this has been pretty different from that, since about Thursday or Friday. I did submit it as a 'feedback' (the closest thing I can find to a bug report there) to YouTube.
Everything was 'normal' as of 2340 last night, as all my podcasts were available this morning ;-)
On Mon, Jul 27, 2020 at 7:26 AM J. Randall Owens notifications@github.com wrote:
Actually, I'm pretty sure that it's something wrong on YouTube's end. I was on the verge of submitting this bug report myself, but then I found that, if you go to the channel's 'Videos' tab, and then click the 'play all', you get a playlist that starts playing their actually most recent video, but the playlist on the side shows that it thinks you're currently playing the most recent video that youtube-dl sees, and labels it #1 https://github.com/ytdl-org/youtube-dl/issues/1 on the list, and the actually newest isn't there at all, even though it's playing it right there. I'm used to waiting 20 minutes to an hour for uploaded/published videos to be listed on the playlists, but this has been pretty different from that, since about Thursday or Friday. I did submit it as a 'feedback' (the closest thing I can find to a bug report there) to YouTube.
ā You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ytdl-org/youtube-dl/issues/26116#issuecomment-664339813, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGI5SH2TPOQ67UCQ3BQTXOLR5VP57ANCNFSM4PHVJ5WQ .
Issue seems fixed? I checked like 10 channels with new videos and their videos seem to be on the upload playlist.
https://github.com/ytdl-org/youtube-dl/issues/26056 seems to still not be fixed though. Maybe it wasn't related.
This was not a duplicate but it seems fixed.
Checklist
Verbose log
Description
When downloading whole YouTube channel the latest uploaded video does not get downloaded. I have tested this on a few channels where the latest uploaded video has ranged from 1 hour ago to 7 hours ago.
The image provided is to compare the latest video uploaded on the channel to the first video youtube-dl decides to check in the command/verbose log posted above.