Files
minstrel/internal/api/suggestions.go
T
bvandeusenandClaude Opus 5 799dab029a
test-go / test (push) Successful in 1m0s
test-go / integration (push) Failing after 4m55s
feat(discover): rank suggestions by taste-tag overlap — #2377 (server)
The payoff slice. Until now a candidate's only claim on a slot was "some
artist you play is adjacent to it in a similarity graph" — a fact that says
nothing about whether the music sounds like anything you like. Now the
candidate's own folksonomy tags (cached by slice 5) are compared against
the user's taste-profile tags, so the deck ranks on taste and can say WHY.

The blend is MULTIPLICATIVE — score × (1 + weight × overlap) — and that
choice carries the whole safety argument:

  - An untagged candidate has overlap 0, so its score is EXACTLY unchanged.
    Tag coverage is permanently partial (#2376); it must cost a candidate
    nothing, not sink it (rule #131).
  - Nothing can leapfrog on tags alone. An additive term with a large
    weight would let a near-zero-similarity artist outrank a strong match
    for sharing one popular tag, which reads as noise.
  - Weight 0 restores pure similarity order bit-for-bit, so the operator's
    knob has a real off position.

overlap = Σ(shared) candWeight × normalizedTasteWeight ÷ Σ(all) candWeight.
Normalizing the taste side by the user's strongest tag makes the score
comparable across users (taste weights accumulate with listening, so a
heavy listener's raw numbers dwarf a new user's while meaning the same
thing). Dividing by the candidate's own mass makes it comparable across
candidates, so a densely-tagged artist can't win on tag count alone.

Applied to the whole over-fetched pool BEFORE selectSuggestions, so the
rotation and diversity rules operate on blended scores — boosting only the
twelve already chosen by similarity would leave the re-ranking undone.

A query failure is returned, NOT degraded past. Graceful degradation is
for expected absence (no taste profile, no cached tags) and both are
handled explicitly as empty inputs; swallowing a real error would hide a
broken DB behind a subtly worse ranking that nothing reports.

Migration 0051 adds a FOURTH tuning scope rather than columns on
taste_tuning, because snooze_days lives here too and a snooze must never
be read as taste signal (#2374) — filing it under 'taste' would put it one
careless join from the leak that design forbids. Expanding
recommendation_tuning_audit's CHECK is in the same migration per rule #36,
and a test asserts the audit row lands, which is what would catch its
absence.

snooze_days moves out of a Go constant onto the tuning card (rule #25),
closing the deferral from #2374.

Tag-overlap tests use deliberately SKEWED fixtures: an evenly-matching pool
cannot exercise a re-ranking, since every candidate gets the same
multiplier and the order is unchanged whether the blend works or not.

Admin UI + client attribution follow in this batch — rule #27.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-02 20:31:12 -04:00

298 lines
9.7 KiB
Go

package api
import (
"context"
"encoding/json"
"errors"
"io"
"net/http"
"strconv"
"strings"
"sync"
"github.com/go-chi/chi/v5"
"github.com/jackc/pgx/v5/pgtype"
"git.fabledsword.com/bvandeusen/minstrel/internal/apierror"
"git.fabledsword.com/bvandeusen/minstrel/internal/db/dbq"
"git.fabledsword.com/bvandeusen/minstrel/internal/lidarr"
"git.fabledsword.com/bvandeusen/minstrel/internal/recommendation"
)
// suggestionView is the wire shape returned by GET /api/discover/suggestions.
type suggestionView struct {
MBID string `json:"mbid"`
Name string `json:"name"`
Score float64 `json:"score"`
Attribution []seedContributionView `json:"attribution"`
// MatchedTags are the candidate's tags that overlap the user's taste
// profile, strongest first (#2377) — the "matches: shoegaze, melancholic"
// line. Omitted when empty, which is common: tag coverage for
// out-of-library artists is permanently partial (#2376), and the card
// falls back to the seed attribution it has always shown.
MatchedTags []string `json:"matched_tags,omitempty"`
// ImageURL is resolved on-demand from Lidarr (out-of-library
// artists have no local art row). Omitted when Lidarr is disabled
// or has no match — the client falls back to a placeholder. Not
// cached: a remote URL Lidarr surfaced, fetched by the browser.
ImageURL string `json:"image_url,omitempty"`
}
type seedContributionView struct {
ArtistID pgtype.UUID `json:"artist_id"`
Name string `json:"name"`
Contribution float64 `json:"contribution"`
IsLiked bool `json:"is_liked"`
PlayCount int64 `json:"play_count"`
}
// handleListSuggestions implements GET /api/discover/suggestions.
//
// Query params:
// - limit (default 12, capped at 50)
// - half_life_days (default 30)
//
// Returns 200 with a JSON array (possibly empty). Read-only; no admin gate.
func (h *handlers) handleListSuggestions(w http.ResponseWriter, r *http.Request) {
user, ok := requireUser(w, r)
if !ok {
return
}
limit := 12
if v := r.URL.Query().Get("limit"); v != "" {
n, err := strconv.Atoi(v)
if err != nil || n < 1 {
writeErr(w, apierror.BadRequest("bad_request", "invalid limit"))
return
}
limit = n
}
halfLife := 30.0
if v := r.URL.Query().Get("half_life_days"); v != "" {
f, err := strconv.ParseFloat(v, 64)
if err != nil || f <= 0 {
writeErr(w, apierror.BadRequest("bad_request", "invalid half_life_days"))
return
}
halfLife = f
}
// Read the tuned weight per request so an admin change takes effect on the
// next refresh, no restart (rule #25).
tagWeight := h.recSettings.Discover().TagOverlapWeight
suggestions, err := recommendation.SuggestArtists(
r.Context(), h.pool, user.ID, halfLife, limit, tagWeight)
if err != nil {
h.logger.Error("api: list suggestions", "err", err)
writeErr(w, apierror.InternalMsg("failed to load suggestions", err))
return
}
out := make([]suggestionView, 0, len(suggestions))
for _, s := range suggestions {
attr := make([]seedContributionView, 0, len(s.Attribution))
for _, a := range s.Attribution {
attr = append(attr, seedContributionView{
ArtistID: a.ArtistID,
Name: a.Name,
Contribution: a.Contribution,
IsLiked: a.IsLiked,
PlayCount: a.PlayCount,
})
}
out = append(out, suggestionView{
MBID: s.MBID, Name: s.Name, Score: s.Score, Attribution: attr,
MatchedTags: s.MatchedTags,
})
}
h.resolveSuggestionArt(r.Context(), out)
writeJSON(w, http.StatusOK, out)
}
// maxSnoozeDays caps a client-supplied duration. The DEFAULT is not here: it's
// a DB-backed knob on the admin tuning card (rule #25), read per request via
// recSettings.Discover().SnoozeDays. See #2377.
const maxSnoozeDays = 365.0
// snoozeRequest is the POST body. Both fields are optional in the JSON sense
// (an absent body snoozes for the default), but Name is required in practice:
// candidates are out-of-library, so the server has no artists row to resolve a
// display name from and the un-snooze list would have nothing to show. The
// client always has it — it just rendered the card.
type snoozeRequest struct {
Name string `json:"name"`
Days float64 `json:"days"`
}
// snoozeView is one row of GET /api/discover/snoozes.
type snoozeView struct {
MBID string `json:"mbid"`
Name string `json:"name"`
SnoozedUntil pgtype.Timestamptz `json:"snoozed_until"`
CreatedAt pgtype.Timestamptz `json:"created_at"`
}
// handleSnoozeSuggestion implements
// POST /api/discover/suggestions/{mbid}/snooze.
//
// Parks a candidate for `days` (default 90, capped at 365). Idempotent:
// snoozing an already-snoozed candidate extends it rather than conflicting.
//
// This is NOT negative feedback. It records no verdict on the artist and is
// never read by internal/taste — see 0049_suggestion_snoozes.up.sql for the
// rule #101 reasoning. Returns 204.
func (h *handlers) handleSnoozeSuggestion(w http.ResponseWriter, r *http.Request) {
user, ok := requireUser(w, r)
if !ok {
return
}
mbid := strings.TrimSpace(chi.URLParam(r, "mbid"))
if mbid == "" {
writeErr(w, apierror.BadRequest("invalid_id", "missing mbid"))
return
}
// An empty body is a valid "snooze this for the default period", so EOF
// is not an error here — decodeBody would reject it as a malformed body.
var body snoozeRequest
if err := json.NewDecoder(r.Body).Decode(&body); err != nil && !errors.Is(err, io.EOF) {
writeErr(w, apierror.BadRequest("invalid_body", ""))
return
}
name := strings.TrimSpace(body.Name)
if name == "" {
writeErr(w, apierror.BadRequest("invalid_body", "name is required"))
return
}
days := body.Days
if days <= 0 {
days = h.recSettings.Discover().SnoozeDays
}
if days > maxSnoozeDays {
// Clamp rather than reject: a client asking for longer than we allow
// still means "park this", and failing the write would leave the card
// sitting there as if the tap did nothing.
days = maxSnoozeDays
}
q := dbq.New(h.pool)
if err := q.SnoozeSuggestion(r.Context(), dbq.SnoozeSuggestionParams{
UserID: user.ID,
CandidateMbid: mbid,
CandidateName: name,
Column4: days,
}); err != nil {
h.logger.Error("api: snooze suggestion", "err", err)
writeErr(w, apierror.InternalMsg("failed to snooze suggestion", err))
return
}
w.WriteHeader(http.StatusNoContent)
}
// handleUnsnoozeSuggestion implements
// DELETE /api/discover/suggestions/{mbid}/snooze.
//
// Brings a parked candidate back immediately. 404s an MBID this user never
// snoozed, so the client can tell "undone" from "there was nothing there".
func (h *handlers) handleUnsnoozeSuggestion(w http.ResponseWriter, r *http.Request) {
user, ok := requireUser(w, r)
if !ok {
return
}
mbid := strings.TrimSpace(chi.URLParam(r, "mbid"))
if mbid == "" {
writeErr(w, apierror.BadRequest("invalid_id", "missing mbid"))
return
}
q := dbq.New(h.pool)
rows, err := q.UnsnoozeSuggestion(r.Context(), dbq.UnsnoozeSuggestionParams{
UserID: user.ID,
CandidateMbid: mbid,
})
if err != nil {
h.logger.Error("api: unsnooze suggestion", "err", err)
writeErr(w, apierror.InternalMsg("failed to unsnooze suggestion", err))
return
}
if rows == 0 {
writeErr(w, apierror.NotFound("snooze"))
return
}
w.WriteHeader(http.StatusNoContent)
}
// handleListSuggestionSnoozes implements GET /api/discover/snoozes.
//
// The un-snooze surface needs this: a parked candidate is by definition
// absent from the suggestion deck, so without a list there is no way to
// reach the DELETE above. Scoped to the caller (rule #47). Expired rows are
// already filtered by the query — the hourly gc sweep only reclaims space.
func (h *handlers) handleListSuggestionSnoozes(w http.ResponseWriter, r *http.Request) {
user, ok := requireUser(w, r)
if !ok {
return
}
rows, err := dbq.New(h.pool).ListActiveSuggestionSnoozes(r.Context(), user.ID)
if err != nil {
h.logger.Error("api: list suggestion snoozes", "err", err)
writeErr(w, apierror.InternalMsg("failed to load snoozes", err))
return
}
out := make([]snoozeView, 0, len(rows))
for _, row := range rows {
out = append(out, snoozeView{
MBID: row.CandidateMbid,
Name: row.CandidateName,
SnoozedUntil: row.SnoozedUntil,
CreatedAt: row.CreatedAt,
})
}
writeJSON(w, http.StatusOK, out)
}
// resolveSuggestionArt fills ImageURL on-demand from Lidarr's artist
// lookup, matched by MBID (foreignArtistId). Best-effort and cache-free:
// Lidarr is the only source — when it's disabled, unreachable, or has
// no match for a candidate, that entry keeps an empty ImageURL and the
// client renders its placeholder. Lookups run with bounded concurrency
// so a full Discover page doesn't serialize ~12 round-trips. Never
// fails the request; the suggestions list is the contract, art is a
// nicety.
func (h *handlers) resolveSuggestionArt(ctx context.Context, views []suggestionView) {
if len(views) == 0 {
return
}
cfg, err := h.lidarrCfg.Get(ctx)
if err != nil || !cfg.Enabled || cfg.BaseURL == "" || cfg.APIKey == "" {
return // Lidarr off / unconfigured → placeholders only
}
client := lidarr.NewClient(cfg.BaseURL, cfg.APIKey)
const maxConcurrent = 6
sem := make(chan struct{}, maxConcurrent)
var wg sync.WaitGroup
for i := range views {
if views[i].MBID == "" || views[i].Name == "" {
continue
}
wg.Add(1)
sem <- struct{}{}
go func(idx int) {
defer wg.Done()
defer func() { <-sem }()
results, lerr := client.LookupArtist(ctx, views[idx].Name)
if lerr != nil {
return // best-effort: no art on lookup failure
}
for _, res := range results {
if res.MBID == views[idx].MBID && res.ImageURL != "" {
// Distinct slice index per goroutine → race-free.
views[idx].ImageURL = res.ImageURL
return
}
}
}(i)
}
wg.Wait()
}