Youtube History Snarfer

What?

Grab the list of youtube video IDs that a user has watched. Requires authentication as that user.

Why?

I have reasons. Sure, you could use the great youtube-dl to do it, but that's >118K lines of code that you'd be allowing to authenticate to youtube's servers. Here I've pulled out just the relevant code, which makes for only ~100 lines to audit if you worry about such things. Of course, there's no reason to suspect youtube-dl of doing anything shady, but some people (me) sleep better at night this way.

How?

ythistory.py -j cookiefile >videolist.txt

ythistory uses the Netscape cookies.txt format to store cookies. If you don't already have your youtube auth cookies, here's how to get them in Chrome:

1. Log in to youtube, and open a new tab.
2. Hit F12 and switch to the Network panel.
3. visit https://youtube.com/feed/history
4. Right-click on the first resource that shows up (history) and select "Copy -> Copy as cURL"
5. In the command you've just copied, you'll have a section that is -H 'cookie: and then a bunch of key=value; pairs. Copy just the key/value pairs.
6. Run ythistory.py -j cookiefile -c "COOKIES" --max 1, pasting the copied key/value pairs instead of COOKIES (but keep the double quotes).
7. cookiefile now has your cookies saved as well, no need to pass them on the command line anymore.

If you only want to incrementally list video ids after a certain point, use --since and a list of videos that delimit the oldest watching history you want to include (the history doesn't track when a video was watched, so I do simple pattern-matching). Eg. to only get the video ids since you finished watching Clickspring's clock build in one sitting, you could pass the last four videos in reverse order:

ythistory.py -j cookiefile --since "J3ZGlpDa-0g,NsnLVYwqESM,T28sGA597IE,R9m4X_R9HPs"

Giving ythistory a list of videos makes increases confidence that it has found the point you intended, in case you watch the same video, or even a handful of videos, again more recently.

Who?

I mean, I'm patrickyeon, but this is mostly just re-packaging work done by some subset of the youtube-dl contributors.

In the spirit of youtube-dl, this code is released into public domain. Do what you want with it.

Help?

I guess if you want. Bug reports and bugfix PRs accepted, new features highly unlikely to be accepted (the point is to expose auth capabilities to a minimum amount of code).