Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upGitHub is where the world builds software
Millions of developers and companies build, ship, and maintain their software on GitHub — the largest and most advanced development platform in the world.
Please print the URLs of failing downloads #12368
Open
Comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I'm filing this as an issue to try to get some direction, because it has some similarity to #11053 where I implemented a solution to a related but more narrow problem, and it was rejected. So I don't want to code up stuff to have it rejected.
While thinking about #12364, I wondered how many SNL videos exhibited that failure. So I saved the set of links to a file and fed the file to
youtube-dl -s -i -a snlinksto see what succeded and what failed. To my surprise,youtube-dldoes not clearly identify which URL it is downloading when it prints a message, be it informational, warning, or [potentially fatal] error. For instance:It's notable to me that in the case of success, it prints 5 lines of information across two InfoExtractors, and none of the information would be as useful as the URL. (I would support each of those lines printing the URL along with them, although I gather that is disfavored, given #11053).
It also seems wrong to me that a separate line is printed for SMIL, m3u8, and JSON download information in the absence of
--verbose. I don't think most users know or need to see that, although it does sort of service as a useful but extremely crude progress indication.Some might argue that printing the
video_idis sufficient, because it is often extracted from the URL and so can correlate. I disagree.video_ids are opaque and while most users know what a URL is, I don't think very many understand what avideo_idis or that that is whatyoutube-dlis printing. In the above example, the initial3477503id is associated withhttp://www.nbc.com/saturday-night-live/video/creating-saturday-night-live-steve-higgins-makes-sound-effects-for-gym-class/3477503and of course it matches the last path element. But that's not necessarily clear if you don't know to look for it, and not user-friendly. It also breaks down in cases where thevideo_iddoes not correlate with the URL; or cases where all thevideo_ids are the same, like how they are all "story" in #12099 (comment). And in this particular case, of NBC SNL,http://www.nbc.com/saturday-night-live/video/creating-saturday-night-live-steve-higgins-makes-sound-effects-for-gym-class/3477503fails but
http://www.nbc.com/saturday-night-live/video/creating-saturday-night-live-steve-higgins-makes-sound-effects-for-gym-class/3477503?snl=0works and they have the same
video_id.Some might argue that URLs should only be displayed with
--verbose. Again, that seems wrong to me, since other much less useful information (SMIL, &c. progress) is shown without it. But even with--verbosethey are not shown. For instance:It's true that the URLs are printed in the batch listing initially, but again it's not sufficiently helpful. It requires the user to perform correlation (probably manually), and I could repeat all the reasons from the non-verbose case.
Thank you.