Hacker News new | ask | show | jobs
by jaspergilley 2070 days ago
You can download the most recent version of the YouTube-DL code from Wayback Machine here: https://web.archive.org/web/20201018144703/https://github.co...

Fuck the RIAA

8 comments

You can only download a snapshot from that page. Here's a tarball with the entire git repo in case anyone's interested: https://transfer.sh/Hj9dD/youtube-dl-git.tar.gz (52MiB)

Here's Debian's mirror btw, somewhat out of date and using the one-commit-per-upstream-release development model (in the upstream branch), so lacking a lot of history: https://salsa.debian.org/debian/youtube-dl

The gzipped tarball, was that as of latest commit? just curious if we can consider it the definitive final copy
Youtube-dlc has merged a lot of youtube-dl's outstanding PRs, i remember -dlc being 350 commits/7 behind -dl some 2 or 3 days ago. So definitely base the fork off *-dlc
There's also that copy in the Arctic. I don't think that Github will remove that one.

Edit: On a serious note, though. I just realised that Internet Archive apparently doesn't archive most PRs? (https://web.archive.org/web/20201018122643if_/https://github...) If so, that's a real shame, I had an open PR on youtube_dl and even though I still have the code locally, I would've liked to keep the PR conversation as it had some really helpful feedback and a bunch of people that were potentially interested in my feature.

That doesn't matter. As development stops, youtube-dl will stop being able to download videos in a matter of months (at best). This is an extremely brittle system that relies on youtube-dl being one step ahead of Youtube at all times. Without further development, the next update to Youtube's site will break the tool.

It was really a matter of time. I'm sure most of us were already amazed it lasted this long - and it did because it remained a relatively obscure tool for years.

It's also the case that the development of subversive/illegal tools requires a certain amount of subterfuge and sneakiness that the youtube-dl developers apparently weren't ready to engage in. While this is understandable, it also underscores the fact that such tools live on borrowed time (as does, I would argue, general-purpose computing itself).

While they may have made collaborative development more difficult for the moment, what hasn't changed is the strong motivation from a wide range of technical people to be able to download whatever they can watch in a browser. Something will rise from the ashes.
I can still install it via pip:

    pip3 install --user --upgrade youtube-dl
So I guess they haven't issued a take down to PyPy yet.
*PyPi. PyPy is an alternate implementation of python.
youtube-dl is in every linux repo, so the source code is widely mirrored and is not going anywhere.
they need to move development elsewhere though

youtube-dl becomes obsolete very quickly as endpoints make changes on their settings and formats, notably youtube.com themselves

just throw it up on Gitee good luck ever getting through the Chinese court system on a copyright issue
wow this straight up sounds like the best solution by far since who knows if microsoft or google is willing to put up a fight
That is actually one of the best solutions
I can still find youtube-dl on gitlab, so it's still out there.
ok so I might just be stupid but how does that link let me clone the repo?.....
Here's a git mirror of the entire repo history up the the latest commit hash that I'm aware of. Compare against the commit hash on Google Webcache [1] for safety:

    git clone https://code.hackerspace.pl/q3k/youtube-dl
Clone, verify and rehost.

[1] - https://webcache.googleusercontent.com/search?q=cache:o7ilf8...

As another comment pointed[0], there is a mirror on gitee: https://gitee.com/mirrors/youtube-downloader

[0] https://news.ycombinator.com/item?id=24872999

https://web.archive.org/web/20201018130325/https://codeload.... is a link to zip of master from commit 4eda10499e8db831167062b0e0dbc7d10d34c1f9, which was around 18/10/2020
IPFS CID of that zip file: QmaCpz7YkLGaAQxa3kRHGQhhd23RQFHuKPzhHaa2pC3i53
Wow I didn't think they'd host ZIP files of every commit of every repo.

Where the hell do they have that much hard disk space, and who is funding them?

We are funding them together: https://archive.org/donate/
The copyright trolls are encircling Archive.org ever since they briefly offered the National Emergency Library. They need all the good vibes we can send them, please donate!
I give them money every month because their archives are essential to preserving our history. Not to mention, they have really saved my bacon before when I really needed to read some technical blog published a decade ago that has since been lost due to the website being reorganized or even being completely taken down.

Even though it takes a lot of bandwidth, I think preserving the source code on GitHub is an invaluable service and hope they continue to do so. I didn't know they were saving the zip file of every release, but I can understand why. It gives you point in time snapshots more easily and who knows how easy it will be to use a git repo from today 100 years from now. ZIP is a much simpler format if it needs to be reimplemented and is used all over the place.

Internet Archive seems to be heavily promoting Protocol Labs tech and File Coin specifically on their Twitter feed. I have donated small sums and ran a couple fundraisers via Facebook which benefited Internet Archive. The promo for File Coin is new. They pitch it as building a decentralized AWS, which I'm not sure if File Coin gets you 100% of the way there, but it's a start?
In a large datacenter. People like you and me, some institutional grants here and there...
It doesn't contain git commit history. Clone a fresh mirror off here https://gitea.eponym.info/Mirrors/youtube-dl.git
You can use the download zip button instead of cloning.
click code, download zip
I mostly see this video "https://www.youtube.com/watch?v=BaW_jenozKcj", which should not be copyrighted.

Does anyone see the video links in the README claimed by RIAA?

Unfortunately, the claimed videos are in some tests in youtube_dl/extractor/youtube.py, e.g.

        # JS player signature function name containing $
        {
            'url': 'https://www.youtube.com/watch?v=nfWlot6h_JM',
            'info_dict': {
                'id': 'nfWlot6h_JM',
                'ext': 'm4a',
                'title': 'Taylor Swift - Shake It Off',
                'description': 'md5:307195cd21ff7fa352270fe884570ef0',
                'duration': 242,
                'uploader': 'TaylorSwiftVEVO',
                'uploader_id': 'TaylorSwiftVEVO',
                'upload_date': '20140818',
            },
            'params': {
                'youtube_include_dash_manifest': True,
                'format': '141/bestaudio[ext=m4a]',
            },
        },