Files
minstrel/internal/dbtest/reset.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

130 lines
5.3 KiB
Go

// Package dbtest provides shared helpers for integration tests that need
// a clean Postgres state without disturbing the operator's admin user.
//
// Background: when integration tests run against the same Postgres
// instance an operator uses for local dev (a common dev-laptop setup),
// a TRUNCATE that includes the users table wipes the operator's admin
// login between every test run. To avoid that, tests should:
//
// 1. Call ResetDB instead of issuing TRUNCATE statements directly. It
// truncates every data table EXCEPT users, then deletes only those
// users whose username starts with TestUserPrefix.
// 2. Always create test user rows with a username that begins with
// TestUserPrefix; otherwise those rows survive across runs and
// leak into other tests.
//
// One test legitimately needs to wipe the entire users table:
// internal/auth/bootstrap_test.go, which exercises first-time admin
// bootstrap. That file does its own TRUNCATE and does not use this
// helper.
package dbtest
import (
"context"
"strings"
"testing"
"github.com/jackc/pgx/v5/pgxpool"
)
// TestUserPrefix is the required username prefix for any user row
// created by an integration test. ResetDB removes every user matching
// this prefix and leaves all others intact.
const TestUserPrefix = "test-"
// dataTables is the union of every non-users table touched by any
// integration test in the repo. Truncating the union is harmless for
// callers that only care about a subset; truncating tables that don't
// exist would error, but every name here corresponds to a migration
// that has shipped.
var dataTables = []string{
"artist_similarity",
"track_similarity",
"artist_similarity_unmatched", // M5c
"scrobble_queue",
"contextual_likes",
"general_likes_albums",
"general_likes_artists",
"general_likes",
"play_events",
"skip_events",
"play_sessions",
"sessions",
"lidarr_quarantine_actions",
"lidarr_quarantine",
// #2374. Keyed by (user_id, candidate_mbid) with no FK to artists —
// candidates are out-of-library — so the CASCADE from artists/users
// does NOT reach it for a leftover row whose user survived. Truncate
// explicitly or a stale snooze silently hides a candidate from the
// next test's suggestion assertions.
"suggestion_snoozes",
// #2376. Same reasoning: keyed by candidate MBID with no FK anywhere,
// so nothing cascades to them. A leftover tag row would make a
// candidate look enriched to the next test, and a leftover state row
// would make it look already-settled and thus ineligible.
"candidate_artist_tags",
"candidate_artist_tag_state",
"playlist_tracks",
"playlists",
"library_changes", // M7 #357 — must reset to keep cursor isolated per test
// SettingsService.reconcile() idempotently re-UpsertProviderSettings
// for every registered provider at boot, so truncating this is the
// correct per-test reset (clears test-modified enabled/api_key rows).
// cover_art_sources_meta is NOT truncated — boot only READS it
// (never recreates the singleton, seeded once by 0018); ResetDB
// resets its counter via UPDATE below instead.
"cover_art_provider_settings",
// Same reasoning for the tag-enrichment settings (#1490): truncate the
// per-provider rows (reconcile re-seeds registered ones), and reset the
// tag_sources_meta counter via UPDATE below rather than truncating it.
"tag_provider_settings",
// recsettings.New reconciles shipped defaults on every construction
// (#1250), so truncating gives each test pristine tuning values.
"recommendation_weight_profiles",
"taste_tuning",
// #2377. Same reasoning as taste_tuning above: recsettings.New re-seeds
// shipped defaults on every construction, so truncating gives each test
// pristine Discover knobs rather than whatever a previous test tuned.
"discover_tuning",
"recommendation_tuning_audit",
"tracks",
"albums",
"artists",
}
// ResetDB clears all data tables and removes any user whose username
// begins with TestUserPrefix. Users without that prefix (notably the
// operator's admin row) are left alone. Calls t.Fatalf on error.
func ResetDB(t *testing.T, pool *pgxpool.Pool) {
t.Helper()
ctx := context.Background()
stmt := "TRUNCATE " + strings.Join(dataTables, ", ") + " RESTART IDENTITY CASCADE"
if _, err := pool.Exec(ctx, stmt); err != nil {
t.Fatalf("dbtest.ResetDB truncate: %v", err)
}
if _, err := pool.Exec(ctx,
"DELETE FROM users WHERE username LIKE $1",
TestUserPrefix+"%",
); err != nil {
t.Fatalf("dbtest.ResetDB delete test users: %v", err)
}
// Reset the monotonic cover-art source-version counter to its
// post-migration seeded value. Truncating the row would break
// SettingsService boot, which reads (never recreates) this
// singleton; an UPDATE keeps the row while clearing cross-test
// version accumulation (CurrentVersion=4 want 1, key-only-bump).
if _, err := pool.Exec(ctx,
"UPDATE cover_art_sources_meta SET current_version = 1",
); err != nil {
t.Fatalf("dbtest.ResetDB reset cover-art version: %v", err)
}
// Same for the tag-sources counter (#1490) — clears cross-test version
// accumulation that would spuriously report version_bumped on a
// key-only change.
if _, err := pool.Exec(ctx,
"UPDATE tag_sources_meta SET current_version = 1, last_registered_providers_hash = ''",
); err != nil {
t.Fatalf("dbtest.ResetDB reset tag-sources version: %v", err)
}
}