Angel Beats! YouTube OP Reupload

See the previous page for some more information on what this page refers to.

Download the entire comment database with

yt-dlp --skip-download --write-comments --dump-json zIFV8UUs1-c | jq -Mcr '[.comments.[] | {"id": .id, "parent": .parent, "text": .text, "author": .author, "author_id": .author_id, "timestamp": .timestamp}]'

For safety, run jq after yt-dlp has written the file to disk by omitting the pipe and --dump-json.

Piping through jq keeps only comments and their content, omitting useless properties like author thumbnail URL and like count. Over the 40 thousand comments to be archived, this ends up more than halving the file size, even with gzip.

The file is then put into tar to create incrementals with

tar -zcf date.tgz -g snapshot.snar info.json

You must rename the output JSON file to info.json. date should be yyyymmdd. Do not rename snapshot.snar.

You may need to pass --no-check-device to tar.

snapshot.snar (you need this for creation only)

To update to a given position, you first need to download a level 0 backup and then all files leading up to and including the date you intend to update to. From there you just have to extract every single tar file in order with

tar -x date.tgz info.json --listed-incremental=/dev/null

I recommend scripting this if it takes little time.

2024

August 24 (4.6M, level 0)