feat(library): match own-state on artist MBID, not just name

Own-state ("in library") relied on an exact artist-NAME match, so an owned
album whose name differed from the MB canonical (punctuation/locale/feat.)
showed as not-owned even when followed. Capture the MusicBrainz artist MBID
on LibraryItem and prefer it for matching.

- New LibraryItem.artistMbid (migration add_library_artist_mbid). Populated at
  import (pipeline) + scan, both of which already hold the resolved target;
  the pipeline also now stores rgMbid (it previously didn't). COALESCE on
  conflict backfills MBIDs on re-scan without clobbering.
- /api/artists "in library, not followed" matches by artistMbid first, name
  case-insensitively as fallback (rows predating the column). Library route
  prefers the item's own artistMbid over the release join.

Last.fm/discover own-state still use name-match (they work) — a smaller
follow-up. worker 233, web 186 tests.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

web/prisma/migrations/20260714195144_add_library_artist_mbid/migration.sql

@@ -0,0 +1,2 @@
+-- AlterTable
+ALTER TABLE "LibraryItem" ADD COLUMN     "artistMbid" TEXT;

web/prisma/schema.prisma

@@ -89,6 +89,10 @@ model LibraryItem {
   // MusicBrainz release-group MBID when known (resolved at manual import) — gives the album
   // cover art without depending on a matching MonitoredRelease row.
   rgMbid       String?
+  // MusicBrainz ARTIST MBID when known (captured at import + scan). Own-state ("in library")
+  // prefers this over the fragile exact artist-name match; null rows fall back to name-match
+  // (re-scan to populate).
+  artistMbid   String?
   // The album's on-disk audio filenames ("## Title.ext"), captured at import + scan, so the
   // Library modal can show the REAL tracks on disk (revealing incomplete/edition mismatches)
   // instead of MusicBrainz's canonical listing. Empty for items imported before this existed

web/src/app/api/artists/route.test.ts

@@ -98,6 +98,16 @@ describe("artists collection API", () => {
     expect(body.ownedNotFollowed).not.toContain("followed artist"); // case-insensitive match
   });
 
+  it("GET matches owned artists to the watched roster by MBID despite a name mismatch", async () => {
+    // followed under the canonical MB name, but the library row's name differs (feat./locale)
+    await prisma.watchedArtist.create({ data: { mbid: "mb-beyonce", name: "Beyoncé" } });
+    await prisma.libraryItem.create({
+      data: { artist: "Beyonce", album: "Lemonade", artistMbid: "mb-beyonce", path: "/l", source: "scan", format: "FLAC", qualityClass: 2 },
+    });
+    const body = await (await GET(listReq())).json();
+    expect(body.ownedNotFollowed).not.toContain("Beyonce"); // matched by MBID → not "unfollowed"
+  });
+
   it("GET ?mbid= returns the follow state for a single artist", async () => {
     await prisma.watchedArtist.create({ data: { mbid: "followed-1", name: "Followed" } });
     expect((await (await GET(listReq("?mbid=followed-1"))).json()).followed).toBe(true);

web/src/app/api/artists/route.ts

@@ -20,16 +20,21 @@ export async function GET(request: Request) {
         },
       },
     }),
-    prisma.libraryItem.findMany({ select: { artist: true }, distinct: ["artist"] }),
+    prisma.libraryItem.findMany({ select: { artist: true, artistMbid: true } }),
   ]);
 
-  // Artists you own music by (LibraryItem) but don't follow yet — name-matched
+  // Artists you own music by (LibraryItem) but don't follow yet. Match on the MusicBrainz
-  // case-insensitively against the watched roster (the have-via-scan name-match caveat).
+  // ARTIST MBID first (robust to name punctuation/locale/feat. variants), falling back to a
+  // case-insensitive name match for library rows that predate the artistMbid column.
+  const followedMbids = new Set(artists.map((a) => a.mbid));
   const followedNames = new Set(artists.map((a) => a.name.trim().toLowerCase()));
-  const ownedNotFollowed = libraryArtists
+  const isFollowed = (r: { artist: string; artistMbid: string | null }) =>
-    .map((r) => r.artist)
+    (r.artistMbid && followedMbids.has(r.artistMbid)) || followedNames.has(r.artist.trim().toLowerCase());
-    .filter((name) => name.trim() && !followedNames.has(name.trim().toLowerCase()))
+  const ownedNotFollowed = [
-    .sort((a, b) => a.localeCompare(b));
+    ...new Set(
+      libraryArtists.filter((r) => r.artist.trim() && !isFollowed(r)).map((r) => r.artist),
+    ),
+  ].sort((a, b) => a.localeCompare(b));
 
   return Response.json({
     artists: artists.map((a) => {

worker/lyra_worker/pipeline.py

@@ -245,15 +245,17 @@ def _import(conn: psycopg.Connection, job_id: str, target: MBTarget,
     with conn.cursor() as cur:
         cur.execute(
             'INSERT INTO "LibraryItem" (id, "requestId", artist, album, path, source, '
-            'format, "qualityClass", "trackNames", "importedAt") '
+            'format, "qualityClass", "trackNames", "rgMbid", "artistMbid", "importedAt") '
-            "VALUES (gen_random_uuid()::text, %s, %s, %s, %s, %s, %s, %s, %s, now()) "
+            "VALUES (gen_random_uuid()::text, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, now()) "
             'ON CONFLICT (artist, album) DO UPDATE SET '
             '  "requestId" = EXCLUDED."requestId", path = EXCLUDED.path, source = EXCLUDED.source, '
             '  format = EXCLUDED.format, "qualityClass" = EXCLUDED."qualityClass", '
-            '  "trackNames" = EXCLUDED."trackNames", "importedAt" = now() '
+            '  "trackNames" = EXCLUDED."trackNames", "rgMbid" = COALESCE(EXCLUDED."rgMbid", "LibraryItem"."rgMbid"), '
+            '  "artistMbid" = COALESCE(EXCLUDED."artistMbid", "LibraryItem"."artistMbid"), "importedAt" = now() '
             'WHERE EXCLUDED."qualityClass" > "LibraryItem"."qualityClass"',
             (request_id, target.artist, target.album, path, winner.source,
-             winner.quality.fmt, quality_class(winner.quality), track_names),
+             winner.quality.fmt, quality_class(winner.quality), track_names,
+             target.rg_mbid or None, target.artist_mbid or None),
         )
         cur.execute(
             "UPDATE \"Request\" SET status = 'completed' WHERE id = %s", (request_id,)

worker/lyra_worker/scan.py

@@ -84,13 +84,16 @@ def _record_owned(conn, target, quality_class: int, fmt: str, path: str, watched
         track_names = list_audio_files(path)  # capture the real on-disk tracks
         cur.execute(
             'INSERT INTO "LibraryItem" (id, "requestId", artist, album, path, source, format, '
-            '"qualityClass", "trackNames", "importedAt") '
+            '"qualityClass", "trackNames", "rgMbid", "artistMbid", "importedAt") '
-            "VALUES (gen_random_uuid()::text, NULL, %s, %s, %s, 'scan', %s, %s, %s, now()) "
+            "VALUES (gen_random_uuid()::text, NULL, %s, %s, %s, 'scan', %s, %s, %s, %s, %s, now()) "
-            # refresh trackNames on re-scan (populates items imported before this column);
+            # refresh trackNames + MBIDs on re-scan (populates items recorded before these
-            # xmax=0 ⇒ this was an INSERT (a genuinely new library item) vs an UPDATE.
+            # columns); xmax=0 ⇒ this was an INSERT (a genuinely new library item) vs an UPDATE.
-            'ON CONFLICT (artist, album) DO UPDATE SET "trackNames" = EXCLUDED."trackNames" '
+            'ON CONFLICT (artist, album) DO UPDATE SET "trackNames" = EXCLUDED."trackNames", '
+            '  "rgMbid" = COALESCE(EXCLUDED."rgMbid", "LibraryItem"."rgMbid"), '
+            '  "artistMbid" = COALESCE(EXCLUDED."artistMbid", "LibraryItem"."artistMbid") '
             "RETURNING (xmax = 0) AS inserted",
-            (target.artist, target.album, path, fmt, quality_class, track_names),
+            (target.artist, target.album, path, fmt, quality_class, track_names,
+             target.rg_mbid or None, target.artist_mbid or None),
         )
         newly = cur.fetchone()[0]  # a new LibraryItem means this album wasn't recorded before
     conn.commit()

worker/tests/test_scan.py

@@ -49,9 +49,9 @@ def test_scan_records_have_monitored_and_follows_artist(conn, tmp_path):
     assert (result.imported, result.skipped) == (1, 0)
     assert _counts(conn) == {"LibraryItem": 1, "MonitoredRelease": 1, "WatchedArtist": 1}
     with conn.cursor() as cur:
-        cur.execute('SELECT source, "qualityClass" FROM "LibraryItem" WHERE artist=%s AND album=%s',
+        cur.execute('SELECT source, "qualityClass", "artistMbid", "rgMbid" FROM "LibraryItem" '
-                    ("John Mayer", "Continuum"))
+                    'WHERE artist=%s AND album=%s', ("John Mayer", "Continuum"))
-        assert cur.fetchone() == ("scan", 3)
+        assert cur.fetchone() == ("scan", 3, "a1", "rg1")  # MBIDs captured for own-state matching
         cur.execute('SELECT monitored, state, "currentQualityClass" FROM "MonitoredRelease" WHERE "rgMbid"=%s',
                     ("rg1",))
         assert cur.fetchone() == (True, "fulfilled", 3)