Skip to content

Corpus snapshot — 2026-07-01

Choose a tag to compare

View all tags
@alltechdev alltechdev released this 02 Jul 03:08
corpus-20260701
51c4013

Built corpus snapshot (server-side SQLite). Restore: gunzip -c corpus-20260701.db.gz > data/corpus.db, then npm run api.

Contents: 70,392 tracks, 1,619 artists, 3,546 albums, 10,220 singles, 1,835 artist playlists, 10,633 community playlists (whitelisted-only at serve time).

New in this snapshot — full detail metadata, all measured:

  • Release dates: 100% of tracks carry an accurate releaseDate (14,521 per-track /player dates for videos + standalone; album audio inherits its album's real date). 13,648/13,766 releases dated via /player (99.1%; the remainder are trackless album stubs).
  • Durations: 70,392/70,392 (100%).
  • Play counts: 39,096 tracks (where YouTube shows them; drives the /artist Top-songs sort).
  • Read-time album aggregates (type / trackCount / totalDurationSec) and releaseDate on every result endpoint.

rejected-artists.json: non-whitelisted artist channels found inside the community playlists, ranked by dropped-track count (whitelist-review candidates); unchanged since 2026-06-29.

Assets 4
Loading