What happened? Was there any interest or the OCR side?

[coolaj86]

I'm curious as to what happened with this. It looks like you did an excellent job of architecture and documentation.

Update

I got the main Tracks database down to 800 KB (270 KB gzipped) using CSV as

ID  Title   Tags    Posted  Length  Bitrate FileName    FileSize    YouTube

I think I might do another 2 like that for Games + Songs and Artists / Composers and still keep the total data under 1mb, and provide a small snipped of JavaScript that fetches the 3 files and links them by numerical ID on the client side along with hard-coded numeric ids for Tags and the list of Mirrors.

Implementation Ideas / Notes

(for people who end up on this repo like I did)

Since there's less than 10,000 remixes + albums, and it's highly unlikely that there will be 100,000 items within our lifetimes, it would probably be simpler to

• ship the entire database as a single JSON file for GET / filter / etc with per-IP rate limit (to encourage proper caching)
  • could be brute-force optimally precompressed w/ gzip, zstd, and brotli
  • could also use single-digit ids for relationships
    (like a typical db rather than a typical api, though I don't like this idea as much)
  • possibly use permanent, browser-level caching for ID ranges e.g. 1-2000, 2001-4000, and dynamic for newer IDs
  • could also use multiple CSVs rather than a single JSON... maybe
• POST by ID for atomic updates
• have a GET that only hands back updates since a last_updated_at parameter

Scraping

I'm going to give this a shot myself with a little help from Grok to save on the tedious HTML parsing.

• https://ocremix.org/remixes/?&offset=0&sort=datedesc gives
  • the approximate number of OCR ids
  • the actual last OCR id
  • offset increments by 30 in sorted mode
  • in non-sorted mode there is a duplicate row for the Game, followed by remixes for the game
• Remixes
  • start at https://ocremix.org/remix/OCR00001
  • end at https://ocremix.org/remix/OCR04851
  • about 5% no longer exist (maybe pulled for copyright or never finished submission?)
  • all seem to have md5sum hash ids, and direct download links which match the torrent names
• Albums
  • many IDs between https://ocremix.org/album/1 and https://ocremix.org/album/100 happen to be valid
  • other IDs are random (?)
  • must crawl https://ocremix.org/albums/?&offset=0&sort=datedesc to get all album ids
  • some have direct downloads, others have dedicated pages
  • !! not all remixes from /albums appear in the /remixes

[psarando]

Thanks!

At the time I started this project, I had reached out to OCR to see if they would be open to a public API, and they did have plans to publish one eventually, so I let them know I would be available to volunteer if they needed assistance. DJP let me know at that time that they were looking into Symphony v2, which is why this project uses that framework.

This project was not only meant as a demo of what I could hopefully contribute to OCR, but also just for myself, as a way to see if I still had some PHP+MySQL chops (and to compare with the APIs I was working on at my job), but I didn't hear back from them after I made this.

A couple of months before DJP stepped down, he mentioned on their Discord that they had adopted Symphony and were on v6, along with API Platform, and that they still wanted to release some sort of API. Darkesword has also mentioned wanting to still release one at some point, but I think they have higher priorities for now.

I haven't looked at this project (or PHP) for a long time (and I haven't used it with my personal projects), but it's really outdated at this point, so I'm going to archive it at this point (which I've been meaning to do since dependabot opened PRs).

about 5% no longer exist (maybe pulled for copyright or never finished submission?)

By the way, remixes have been removed from the site (never for copyright, as far as I know), and they keep a Changelog with the reasons they were removed.