Files
minstrel/internal/db/queries/system_playlists.sql
T
bvandeusen 2e7b81fdfe fix(playlists): robust For-You seed + deep fill; young-library mix fallbacks
For-You silently vanished after ~7 days of not listening (seed query
required a non-skip play in the last 7 days) and capped at ~40 on
self-hosted libraries with no ListenBrainz similarity data.

- PickTopPlayedTracksForUser: tiered seed — last 30d top plays, else
  all-time top plays, else liked tracks. For-You only disappears now
  if the account has zero plays AND zero likes.
- produceForYou uses deeper candidate source limits (raised random/
  tag/similar K) so it reaches ~100 even with empty lb_similar /
  similar_artists; richer when LB enrichment is present.
- Rediscover: tiered — 6-month-dormant, else ≥5-play 30-day-dormant.
- On This Day: floor 60→30 days, window ±7→±10 doy; still skips
  cleanly (no rows → no playlist) on insufficient history.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-16 12:48:20 -04:00

210 lines
7.3 KiB
SQL

-- M7 #352 slice 2: system-generated playlist queries.
-- name: ListActiveUsersForSystemPlaylists :many
-- Active = had a play in the last 7 days. The cron iterates this list.
SELECT u.id FROM users u
WHERE EXISTS (
SELECT 1 FROM play_events pe
WHERE pe.user_id = u.id
AND pe.started_at > now() - INTERVAL '7 days'
);
-- name: GetSystemPlaylistRun :one
SELECT user_id, last_run_at, last_run_date, in_flight, last_error
FROM system_playlist_runs
WHERE user_id = $1;
-- name: TryClaimSystemPlaylistRun :one
-- Atomic test-and-set on in_flight. Returns the row if we won the claim,
-- or pgx.ErrNoRows if someone else is already building.
INSERT INTO system_playlist_runs (user_id, in_flight)
VALUES ($1, true)
ON CONFLICT (user_id) DO UPDATE
SET in_flight = true
WHERE system_playlist_runs.in_flight = false
RETURNING user_id, last_run_at, last_run_date, in_flight, last_error;
-- name: FinishSystemPlaylistRun :exec
-- Mark a successful build complete.
UPDATE system_playlist_runs
SET last_run_at = $2,
last_run_date = $3,
in_flight = false,
last_error = NULL
WHERE user_id = $1;
-- name: FailSystemPlaylistRun :exec
-- Mark a failed build complete (in_flight=false, error captured).
UPDATE system_playlist_runs
SET in_flight = false,
last_error = $2
WHERE user_id = $1;
-- name: ClearStaleSystemPlaylistInFlight :exec
-- Cron startup recovery: clear any in_flight=true rows from a previously
-- crashed process. Safe because at startup, by definition, nothing is
-- still building.
UPDATE system_playlist_runs SET in_flight = false WHERE in_flight = true;
-- name: PickSeedArtists :many
-- Top-5 most-engaged distinct artist candidates in the user's last 7
-- days. The Go-side picker (pickSeedArtistsForDay) shuffles these
-- daily-deterministically and takes the first 3 so the set of
-- "Songs like X" mixes rotates day-to-day.
-- Score = unskipped-play count + 5 if user has liked the artist.
WITH plays AS (
SELECT t.artist_id,
COUNT(*) FILTER (WHERE pe.was_skipped = false) AS play_count
FROM play_events pe
JOIN tracks t ON t.id = pe.track_id
WHERE pe.user_id = $1
AND pe.started_at > now() - INTERVAL '7 days'
AND t.artist_id IS NOT NULL
GROUP BY t.artist_id
),
liked AS (
SELECT artist_id FROM general_likes_artists WHERE user_id = $1
)
SELECT p.artist_id,
(p.play_count + CASE WHEN l.artist_id IS NOT NULL THEN 5 ELSE 0 END)::bigint AS score
FROM plays p
LEFT JOIN liked l ON l.artist_id = p.artist_id
ORDER BY score DESC, p.artist_id
LIMIT 5;
-- name: PickTopPlayedTracksForUser :many
-- For-You candidate seeds, tiered so For-You never silently vanishes:
-- tier 0 top non-skip plays in the last 30 days
-- tier 1 (only if tier 0 empty) all-time top non-skip plays
-- tier 2 (only if tiers 0+1 empty) liked tracks
-- Returns up to 5 ids; tie-break by track_id for determinism. The
-- Go-side picker (pickForYouSeedForDay) rotates one per day via
-- userIDHash. Widened from a hard 7-day window, which made For-You
-- disappear after a week of not listening and never recover on a
-- self-hosted library with sparse history.
WITH recent AS (
SELECT t.id, COUNT(*) AS c, 0 AS tier
FROM play_events pe
JOIN tracks t ON t.id = pe.track_id
WHERE pe.user_id = $1
AND pe.started_at > now() - INTERVAL '30 days'
AND pe.was_skipped = false
GROUP BY t.id
),
alltime AS (
SELECT t.id, COUNT(*) AS c, 1 AS tier
FROM play_events pe
JOIN tracks t ON t.id = pe.track_id
WHERE pe.user_id = $1
AND pe.was_skipped = false
GROUP BY t.id
),
liked AS (
SELECT gl.track_id AS id, 0::bigint AS c, 2 AS tier
FROM general_likes gl
WHERE gl.user_id = $1
),
chosen AS (
SELECT id, c, tier FROM recent
UNION ALL
SELECT id, c, tier FROM alltime
WHERE NOT EXISTS (SELECT 1 FROM recent)
UNION ALL
SELECT id, c, tier FROM liked
WHERE NOT EXISTS (SELECT 1 FROM recent)
AND NOT EXISTS (SELECT 1 FROM alltime)
)
SELECT id
FROM chosen
ORDER BY tier, c DESC, id
LIMIT 5;
-- name: PickTopPlayedTrackForArtistByUser :one
-- "Songs like X" seed selection. Returns the user's most-played non-skipped
-- track by artist X in the last 7 days. Falls back to the artist's most-
-- recent album's first track if no plays exist.
-- Cast to uuid so sqlc infers pgtype.UUID rather than interface{}.
SELECT COALESCE(
(SELECT t.id
FROM play_events pe
JOIN tracks t ON t.id = pe.track_id
WHERE pe.user_id = $1
AND t.artist_id = $2
AND pe.started_at > now() - INTERVAL '7 days'
AND pe.was_skipped = false
GROUP BY t.id
ORDER BY COUNT(*) DESC, t.id
LIMIT 1),
(SELECT t.id
FROM tracks t
JOIN albums a ON a.id = t.album_id
WHERE t.artist_id = $2
ORDER BY a.release_date DESC NULLS LAST,
t.disc_number NULLS LAST,
t.track_number NULLS LAST,
t.id
LIMIT 1)
)::uuid AS id;
-- name: PickTopAlbumCoverForArtistByUser :one
-- "Songs like X" cover. Most-played album by user; tie-break by most-recent
-- release. NULL if the artist has no albums or all covers are NULL.
-- Note: albums uses cover_art_path (not cover_path).
SELECT a.cover_art_path
FROM albums a
LEFT JOIN tracks t ON t.album_id = a.id
LEFT JOIN play_events pe
ON pe.track_id = t.id
AND pe.user_id = $1
AND pe.was_skipped = false
WHERE a.artist_id = $2
AND a.cover_art_path IS NOT NULL
GROUP BY a.id, a.release_date, a.cover_art_path
ORDER BY COUNT(pe.id) DESC, a.release_date DESC NULLS LAST, a.id
LIMIT 1;
-- name: DeleteSystemPlaylistsForUser :exec
-- Atomic-replace step. CASCADE on playlist_tracks.playlist_id handles
-- the join rows.
DELETE FROM playlists WHERE user_id = $1 AND kind = 'system';
-- name: ListPlaylistsByUserAndKind :many
-- Used in integration tests; production filter applied in Go
-- (api.handleListPlaylists fetches via Service.List and filters by Kind in
-- memory so it can also surface other users' public playlists alongside
-- the caller's own kind-filtered ones).
-- Sorted by updated_at DESC.
SELECT id, user_id, name, description, is_public, kind, system_variant,
seed_artist_id, cover_path, track_count, duration_sec,
created_at, updated_at
FROM playlists
WHERE user_id = $1
AND ($2::text = 'all' OR kind = $2)
ORDER BY updated_at DESC;
-- name: CreateSystemPlaylist :one
-- Inserts a system-generated playlist. Used by BuildSystemPlaylists.
-- For 'for_you' variant, pass seed_artist_id as NULL (zero-value pgtype.UUID
-- with Valid=false).
INSERT INTO playlists (
user_id, name, description, is_public,
kind, system_variant, seed_artist_id, cover_path
) VALUES ($1, $2, '', false, 'system', $3, $4, $5)
RETURNING *;
-- name: GetSystemPlaylistByVariantForUser :one
-- Returns the user's system playlist for the given variant
-- ('for_you' / 'songs_like_artist' / 'discover'), or pgx.ErrNoRows
-- when the user hasn't built that variant yet. For
-- 'songs_like_artist' (which can have multiple rows per user, one
-- per seed), this returns whichever LIMIT 1 picks — callers that
-- need all instances should use a different query.
SELECT p.id, p.user_id, p.name, p.kind, p.system_variant,
p.seed_artist_id, p.cover_path, p.track_count
FROM playlists p
WHERE p.user_id = $1
AND p.kind = 'system'
AND p.system_variant = $2
ORDER BY p.created_at DESC
LIMIT 1;