yap: A Python repository from Benjins

YAP - Youtube Annotation Preservation

NOTE: Youtube has removed access to all annotations as of January 15. Any downloads will result in an XML file w/o legacy annotations.

YouTube has decided to outlaw ~~fun~~annotations on their platform. This project consists of tools for people to ensure that their hard work is not lost.

Here are the main parts to this project:

anno.py: A python script ~~that builds off of youtube-dl~~ to download annotations ~~and metadata~~, but not the video files or any other data for videos. It also parses out any links to other videos in those annotations, and downloads them. This is useful for interactive videos based on annotations, which may have dozens of videos linking to each other in a tree/graph like fashion.
render/: HTML/Javascript that can render the annotations on top of a YouTube embed. Used to ensure that annotations are downloaded properly.
experimental/: Some experimental scripts that won't work out of the box, but that I've found to be much more efficient, as they use asynchronous IO (the dl_annotations.py script pipes into the write_to_db.py script).
crawl_user.py: A python script for finding all the public video URLS of a particular uploader (OR playlist...naming, I know).
crawl_related.py: A python script that given a set of youtube videos, produces their related videos from the sidebar.
crawl_search.py: A python script that will deliver the videos that youtube returns in its search query for the given terms.

See here for a live demo of some recovered annotations re-rendered over their source videos.

There are some other resources that may be useful:

Some examples of videos that use annotations in various ways:

TODO for rendering layer:

Benjins/yap

YAP - Youtube Annotation Preservation

NOTE: Youtube has removed access to all annotations as of January 15. Any downloads will result in an XML file w/o legacy annotations.