feat(server/playlists): Discover system playlist

Adds the third system playlist variant: 'discover'. Surfaces 100
tracks the operator has not played and not liked, biased toward
artists they rarely play (< 10 plays) and complemented by tracks
liked by other users plus a random unheard sample. Cold start
collapses to all-random; single-user servers redistribute the
cross-user-likes allocation equally across the other two buckets.

The build runs as a fourth phase inside BuildSystemPlaylists
alongside For You and the seed-artist mixes. Same daily-deterministic
md5(track_id || dateStr) ordering as the other variants. Per-album
(<=2) and per-artist (<=3) caps prevent collapse onto a single
prolific artist. Buckets interleave round-robin so the playlist
order doesn't front-load one bucket's flavour.

Post-commit collage generation (already wired) gives Discover the
same 4-cell cover treatment as the other system playlists — no
new collage code needed.

Tests cover the slot-redistribution table (all-available,
cold-start, single-user, partial-dormant, fully-empty), the
per-album/per-artist caps, the round-robin interleave, and a
DB-backed cold-start integration test that asserts a Discover
playlist lands with non-zero tracks.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
2026-05-07 08:35:54 -04:00
parent a98a106f6f
commit f77a0699ec
4 changed files with 469 additions and 0 deletions
+259
View File
@@ -0,0 +1,259 @@
package playlists
import (
"context"
"fmt"
"github.com/jackc/pgx/v5/pgtype"
"git.fabledsword.com/bvandeusen/minstrel/internal/db/dbq"
)
const (
discoverTotalSlots = 100
discoverDormantSlots = 40
discoverCrossUserSlots = 30
discoverRandomSlots = 30
discoverMaxTracksPerAlbum = 2
discoverMaxTracksPerArtist = 3
)
// discoverTrack is the common shape used by the bucket allocator. The
// three sqlc-generated row types collapse into this internal struct so
// downstream functions don't need to be generic over them.
type discoverTrack struct {
ID pgtype.UUID
AlbumID pgtype.UUID
ArtistID pgtype.UUID
}
// buildDiscoverCandidates assembles the Discover playlist track list.
// Pulls from three buckets, applies per-album/per-artist caps, then
// redistributes any deficit equally across the remaining buckets.
//
// Returns up to discoverTotalSlots track IDs in the order they should
// appear in the playlist (round-robin interleaved across buckets).
// Empty slice when the library has no eligible tracks at all.
func buildDiscoverCandidates(ctx context.Context, q *dbq.Queries, userID pgtype.UUID, dateStr string) ([]rankedCandidate, error) {
dormantRows, err := q.ListDormantArtistTracksForDiscover(ctx, dbq.ListDormantArtistTracksForDiscoverParams{
UserID: userID, Column2: dateStr,
})
if err != nil {
return nil, fmt.Errorf("dormant bucket: %w", err)
}
crossUserRows, err := q.ListCrossUserLikedTracksForDiscover(ctx, dbq.ListCrossUserLikedTracksForDiscoverParams{
UserID: userID, Column2: dateStr,
})
if err != nil {
return nil, fmt.Errorf("cross-user bucket: %w", err)
}
randomRows, err := q.ListRandomUnheardTracksForDiscover(ctx, dbq.ListRandomUnheardTracksForDiscoverParams{
UserID: userID, Column2: dateStr,
})
if err != nil {
return nil, fmt.Errorf("random bucket: %w", err)
}
// Adapt sqlc-generated rows into the internal struct, then apply
// per-album / per-artist caps.
dormantPool := capByAlbumAndArtist(dormantRowsToTracks(dormantRows))
crossUserPool := capByAlbumAndArtist(crossUserRowsToTracks(crossUserRows))
randomPool := capByAlbumAndArtist(randomRowsToTracks(randomRows))
// Allocate slots with redistribution.
allocations := redistributeSlots([]bucketRequest{
{want: discoverDormantSlots, available: len(dormantPool)},
{want: discoverCrossUserSlots, available: len(crossUserPool)},
{want: discoverRandomSlots, available: len(randomPool)},
})
// Take the head of each bucket's capped pool.
dormantTake := dormantPool[:allocations[0]]
crossUserTake := crossUserPool[:allocations[1]]
randomTake := randomPool[:allocations[2]]
// Round-robin interleave so the playlist doesn't front-load one
// flavour.
out := interleaveBuckets(dormantTake, crossUserTake, randomTake)
// Convert to rankedCandidate (the type insertSystemPlaylist accepts).
// Score is unused for Discover — the daily-deterministic md5 ordering
// already gave us the ranking.
ranked := make([]rankedCandidate, 0, len(out))
for _, t := range out {
ranked = append(ranked, rankedCandidate{TrackID: t.ID})
}
return ranked, nil
}
func dormantRowsToTracks(rows []dbq.ListDormantArtistTracksForDiscoverRow) []discoverTrack {
out := make([]discoverTrack, len(rows))
for i, r := range rows {
out[i] = discoverTrack{ID: r.ID, AlbumID: r.AlbumID, ArtistID: r.ArtistID}
}
return out
}
func crossUserRowsToTracks(rows []dbq.ListCrossUserLikedTracksForDiscoverRow) []discoverTrack {
out := make([]discoverTrack, len(rows))
for i, r := range rows {
out[i] = discoverTrack{ID: r.ID, AlbumID: r.AlbumID, ArtistID: r.ArtistID}
}
return out
}
func randomRowsToTracks(rows []dbq.ListRandomUnheardTracksForDiscoverRow) []discoverTrack {
out := make([]discoverTrack, len(rows))
for i, r := range rows {
out[i] = discoverTrack{ID: r.ID, AlbumID: r.AlbumID, ArtistID: r.ArtistID}
}
return out
}
// capByAlbumAndArtist trims a bucket's pool so no single album appears
// more than discoverMaxTracksPerAlbum times and no single artist
// appears more than discoverMaxTracksPerArtist times. Preserves input
// order (which is the daily-deterministic md5 sort).
//
// pgtype.UUID's Bytes field is [16]byte, which IS comparable as a map
// key.
func capByAlbumAndArtist(rows []discoverTrack) []discoverTrack {
albumCount := map[pgtype.UUID]int{}
artistCount := map[pgtype.UUID]int{}
out := make([]discoverTrack, 0, len(rows))
for _, r := range rows {
if albumCount[r.AlbumID] >= discoverMaxTracksPerAlbum {
continue
}
if artistCount[r.ArtistID] >= discoverMaxTracksPerArtist {
continue
}
albumCount[r.AlbumID]++
artistCount[r.ArtistID]++
out = append(out, r)
}
return out
}
// bucketRequest captures one bucket's desired and available count.
type bucketRequest struct {
want int
available int
}
// redistributeSlots returns the final per-bucket allocation given each
// bucket's want and available count. When a bucket can't fill its want,
// the deficit splits equally between the OTHER buckets that still have
// supply. Sum of returned allocations <= sum of `want`.
//
// Algorithm:
// 1. Initial pass: allocate min(want, available) to each bucket;
// compute supply = available - allocated.
// 2. For each deficit bucket (allocated < want), split the deficit
// equally across peers that still have supply > 0.
// 3. Repeat until no further movement (deficit redistribution
// can create new deficits if the recipient was already at its
// `want`; we cap each bucket at its `available` after every pass).
//
// For 3 buckets, two redistribution passes are sufficient in practice;
// we cap at 4 passes defensively.
func redistributeSlots(buckets []bucketRequest) []int {
n := len(buckets)
final := make([]int, n)
supply := make([]int, n)
for i, b := range buckets {
final[i] = b.want
if final[i] > b.available {
final[i] = b.available
}
supply[i] = b.available - final[i]
}
for pass := 0; pass < 4; pass++ {
anyMoved := false
for i, b := range buckets {
deficit := b.want - final[i]
if deficit <= 0 {
continue
}
// Find peers with supply.
peers := make([]int, 0, n-1)
for j := 0; j < n; j++ {
if j == i {
continue
}
if supply[j] > 0 {
peers = append(peers, j)
}
}
if len(peers) == 0 {
continue
}
share := deficit / len(peers)
rem := deficit % len(peers)
for k, j := range peers {
take := share
if k < rem {
take++
}
if take > supply[j] {
take = supply[j]
}
if take > 0 {
final[j] += take
supply[j] -= take
anyMoved = true
}
}
}
if !anyMoved {
break
}
}
// Final clamp: total <= discoverTotalSlots.
sum := 0
for _, f := range final {
sum += f
}
for sum > discoverTotalSlots {
// Trim the largest bucket. Should be rare.
maxVal, idx := -1, -1
for i, f := range final {
if f > maxVal {
maxVal, idx = f, i
}
}
if idx < 0 {
break
}
final[idx]--
sum--
}
return final
}
// interleaveBuckets round-robins items from the buckets in order. The
// first item of each bucket appears before the second of any bucket.
func interleaveBuckets(buckets ...[]discoverTrack) []discoverTrack {
total := 0
for _, b := range buckets {
total += len(b)
}
out := make([]discoverTrack, 0, total)
indices := make([]int, len(buckets))
for {
anyAppended := false
for bi, b := range buckets {
if indices[bi] < len(b) {
out = append(out, b[indices[bi]])
indices[bi]++
anyAppended = true
}
}
if !anyAppended {
break
}
}
return out
}