Files
minstrel/internal/taste/profile_test.go
T
bvandeusen 7d18a3c808
test-go / test (push) Successful in 30s
test-go / integration (push) Successful in 4m43s
feat(taste): fold enriched folksonomy tags into the profile (#1490 Step 3)
The taste recompute's tag facet now unions the cached track_tags
(MusicBrainz/Last.fm folksonomy tags) alongside raw ID3 genre, so a coarse
"Rock" gains "post-punk / shoegaze / melancholic".

- taste_profile.sql: ListPlayEngagementInputsForUser +
  ListLikedTrackTasteInputsForUser now return track_id to key the
  enriched-tag lookup.
- accumulate(): for each play, fold its track's enriched tags weighted by
  engagement × tag.weight × EnrichedTagScale; for each liked track, by the
  tag-like bonus × tag.weight × scale. A track with no cached tags
  contributes genre only (graceful).
- New Config.EnrichedTagScale (default 0.5) — enriched tags augment the
  ID3 signal without swamping it; 0 = genre-only. Flows through
  recsettings.TasteConfig() (starts from DefaultConfig). Promoting it into
  the admin tuning lab is a small follow-up.

Unit-tested the pure foldEnrichedTags helper (overlap accumulation +
scale=0 disable).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-13 22:09:31 -04:00

124 lines
3.4 KiB
Go

package taste
import (
"testing"
"github.com/jackc/pgx/v5/pgtype"
"git.fabledsword.com/bvandeusen/minstrel/internal/db/dbq"
)
func strLess(a, b string) bool { return a < b }
func TestFoldEnrichedTags(t *testing.T) {
tagW := map[string]float64{"rock": 1.0} // pre-existing ID3 genre weight
tags := []dbq.TrackTag{
{Tag: "shoegaze", Weight: 1.0},
{Tag: "melancholic", Weight: 0.5},
{Tag: "rock", Weight: 0.4}, // overlaps genre → accumulates
}
// base=2 (engagement), scale=0.5 → contribution = 2 * weight * 0.5 = weight.
foldEnrichedTags(tagW, tags, 2.0, 0.5)
if got := tagW["shoegaze"]; got != 1.0 {
t.Errorf("shoegaze = %v, want 1.0", got)
}
if got := tagW["melancholic"]; got != 0.5 {
t.Errorf("melancholic = %v, want 0.5", got)
}
if got := tagW["rock"]; got != 1.4 { // 1.0 genre + 2*0.4*0.5
t.Errorf("rock = %v, want 1.4", got)
}
// scale=0 disables the enriched contribution entirely.
base := map[string]float64{"rock": 1.0}
foldEnrichedTags(base, tags, 2.0, 0)
if len(base) != 1 || base["rock"] != 1.0 {
t.Errorf("scale=0 should leave tagW untouched; got %v", base)
}
}
func TestRankWeighted_DropsEpsilonAndSortsDesc(t *testing.T) {
m := map[string]float64{"a": 3.0, "b": 1.0, "tiny": 0.01, "d": -2.0}
got := rankWeighted(m, 0.05, 10, 10, strLess)
if len(got) != 3 {
t.Fatalf("len = %d, want 3 (tiny dropped by epsilon)", len(got))
}
wantOrder := []string{"a", "b", "d"} // 3, 1, -2 descending
for i, w := range wantOrder {
if got[i].Key != w {
t.Errorf("pos %d = %q, want %q", i, got[i].Key, w)
}
}
}
func TestRankWeighted_CapsTopAndBottom(t *testing.T) {
m := map[string]float64{"e5": 5, "e4": 4, "e3": 3, "e2": 2, "e1": 1}
got := rankWeighted(m, 0.0, 1, 1, strLess) // top 1 + bottom 1
if len(got) != 2 {
t.Fatalf("len = %d, want 2", len(got))
}
if got[0].Key != "e5" || got[1].Key != "e1" {
t.Errorf("got [%s,%s], want [e5,e1]", got[0].Key, got[1].Key)
}
}
func TestRankWeighted_DeterministicTieBreak(t *testing.T) {
m := map[string]float64{"z": 1.0, "a": 1.0, "m": 1.0}
got := rankWeighted(m, 0.0, 10, 10, strLess)
// Equal weights → ordered by keyLess (string asc): a, m, z.
want := []string{"a", "m", "z"}
for i, w := range want {
if got[i].Key != w {
t.Errorf("tie-break pos %d = %q, want %q", i, got[i].Key, w)
}
}
}
func TestApplyFloor_ClampsNegatives(t *testing.T) {
m := map[string]float64{"a": -5, "b": 2, "c": -1}
applyFloor(m, -3)
if m["a"] != -3 {
t.Errorf("a = %.1f, want -3 (floored)", m["a"])
}
if m["b"] != 2 {
t.Errorf("b = %.1f, want 2 (untouched)", m["b"])
}
if m["c"] != -1 {
t.Errorf("c = %.1f, want -1 (above floor)", m["c"])
}
}
func TestSplitGenres(t *testing.T) {
mk := func(s string) *string { return &s }
cases := []struct {
in *string
want []string
}{
{nil, nil},
{mk(""), nil},
{mk("Rock"), []string{"Rock"}},
{mk(" Rock "), []string{"Rock"}},
{mk("Rock; Pop, Jazz"), []string{"Rock", "Pop", "Jazz"}},
}
for _, c := range cases {
got := splitGenres(c.in)
if len(got) != len(c.want) {
t.Errorf("splitGenres(%v) len = %d, want %d", c.in, len(got), len(c.want))
continue
}
for i := range got {
if got[i] != c.want[i] {
t.Errorf("splitGenres(%v)[%d] = %q, want %q", c.in, i, got[i], c.want[i])
}
}
}
}
func TestUuidLess_Orders(t *testing.T) {
a := pgtype.UUID{Bytes: [16]byte{1}, Valid: true}
b := pgtype.UUID{Bytes: [16]byte{2}, Valid: true}
if !uuidLess(a, b) || uuidLess(b, a) {
t.Error("uuidLess byte ordering wrong")
}
}