Plan smarter Lidarr matching via exact MBID lookup

Drop fuzzy difflib scoring: MusicBrainz resolves track->album release-group
MBID, Lidarr album/lookup?term=mbid:<id> returns the exact album. Live-verified
against the user's Lidarr.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/superpowers/plans/2026-06-08-smarter-lidarr-matching.md

@@ -0,0 +1,514 @@
+# Smarter Lidarr Matching Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Make `musicfetch.lidarr_search` resolve a Shazam-style `Artist - Track` to the correct album by asking MusicBrainz for the studio album's release-group MBID, then doing an **exact** Lidarr lookup (`album/lookup?term=mbid:<MBID>`) — so the noninteractive API picks the real album (Daft Punk *Discovery*) instead of junk (Pignickel novelty), with **no fuzzy ranking system**.
+
+**Architecture:** All changes are in the single-file `musicfetch` binary (the shared search used by both the CLI picker and the REST API). New helpers `_split_query` and `musicbrainz_best_album`, plus a rewritten `lidarr_search` with small lookup helpers and tiered fallbacks. Tests import the binary as a module via the existing `server.mf` loader (which registers it in `sys.modules` as `musicfetch_core`).
+
+**Tech Stack:** Python 3.10+, stdlib `time`, `requests` (already a dep), pytest with `monkeypatch`. No new dependencies. Live-validated against MusicBrainz + the user's Lidarr 3.1.0 — `album/lookup?term=mbid:48117b90-a16e-34ca-a514-19c702df1158` returns exactly `Discovery — Daft Punk`.
+
+---
+
+## Context for the implementer
+
+`musicfetch` is an executable Python file (no `.py` ext) at the repo root. Relevant existing pieces:
+- `Hit` dataclass: fields `source, kind, title, artist, album, year, thumbnail, payload`.
+- `_album_to_hit(album)` → `Hit(source="lidarr", kind="album", ..., payload={"album": album})`. The raw Lidarr album dict carries `foreignAlbumId` (MusicBrainz release-group MBID) and `releaseDate`.
+- `_artist_to_hit(artist)` → `Hit(source="lidarr", kind="artist", ...)`.
+- `lidarr_get(path, params=None, timeout=15)` → GET helper, raises on HTTP error.
+- `API_KEY`, `dbg(...)`, `err(...)`, module-level `requests`, `from requests.exceptions import RequestException, Timeout`.
+- Current `lidarr_search(query, limit)` at lines ~129-162 trusts `/api/v1/search` ordering then falls back to `/album/lookup` + `/artist/lookup`. **This is what we replace.**
+
+**Why MusicBrainz is still required:** Lidarr has no track-search endpoint; `album/lookup` only matches albums/artists. Shazam gives `Artist - Track`, and the track name won't match the album title in Lidarr. MusicBrainz recording search maps track → album, and gives us the release-group MBID that Lidarr's `mbid:` lookup resolves exactly. No scoring needed.
+
+**Don't break callers:** `lidarr_search(query, limit) -> list[Hit]` signature stays identical. `build_combined_hits` and the API depend on it returning `[]` on failure (so the YouTube fallback works).
+
+**Tests access the binary like this** (top of each new test module):
+```python
+import server.mf  # noqa: F401 — loads musicfetch and registers musicfetch_core in sys.modules
+import musicfetch_core as mf
+```
+Set `mf.API_KEY` via `monkeypatch.setattr(mf, "API_KEY", "testkey")` where needed.
+
+**One import to add** to the top imports block of `musicfetch` (Task 2): `import time`.
+
+---
+
+### Task 1: Query splitter `_split_query`
+
+**Files:**
+- Modify: `musicfetch` (add `_split_query` just above `lidarr_search`)
+- Test: `tests/test_lidarr_match.py`
+
+- [ ] **Step 1: Write the failing test**
+
+Create `tests/test_lidarr_match.py`:
+```python
+import server.mf  # noqa: F401 — loads musicfetch, registers musicfetch_core in sys.modules
+import musicfetch_core as mf
+
+
+def test_split_query_with_dash():
+    assert mf._split_query("Daft Punk - Discovery") == ("Daft Punk", "Discovery")
+
+
+def test_split_query_no_dash():
+    assert mf._split_query("Daft Punk") == ("Daft Punk", None)
+
+
+def test_split_query_splits_on_first_dash_only():
+    assert mf._split_query("A - B - C") == ("A", "B - C")
+
+
+def test_split_query_strips_whitespace():
+    assert mf._split_query("  Daft Punk  -  Discovery  ") == ("Daft Punk", "Discovery")
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run: `pytest tests/test_lidarr_match.py -v`
+Expected: FAIL — `AttributeError: module 'musicfetch_core' has no attribute '_split_query'`
+
+- [ ] **Step 3: Add the implementation**
+
+In `musicfetch`, immediately above `def lidarr_search(`:
+```python
+def _split_query(query: str) -> tuple[str, Optional[str]]:
+    """Split a Shazam-style 'Artist - Track' on the first ' - '.
+    Returns (artist, track) or (term, None) when there is no separator."""
+    if " - " in query:
+        left, right = query.split(" - ", 1)
+        return left.strip(), right.strip()
+    return query.strip(), None
+```
+
+- [ ] **Step 4: Run test to verify it passes**
+
+Run: `pytest tests/test_lidarr_match.py -v`
+Expected: PASS (4 passed)
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add musicfetch tests/test_lidarr_match.py
+git commit -m "feat(lidarr): add Artist - Track query splitter"
+```
+
+---
+
+### Task 2: MusicBrainz track→album resolver
+
+**Files:**
+- Modify: `musicfetch` (add `import time` to top imports; add MB constants + `_mb_rate_limit`, `_mb_artist_credit`, `musicbrainz_best_album` above `lidarr_search`)
+- Test: `tests/test_musicbrainz.py`
+
+The release-group selection prefers studio albums (`primary-type == "Album"` with no `secondary-types`), choosing the earliest dated one, skipping Single/Compilation/Live. Verified live: for "Daft Punk / Harder Better Faster Stronger" MB returns a Single, Compilations, Live albums, and the studio **Discovery** (mbid `48117b90-a16e-34ca-a514-19c702df1158`).
+
+- [ ] **Step 1: Write the failing test**
+
+Create `tests/test_musicbrainz.py`:
+```python
+import server.mf  # noqa: F401
+import musicfetch_core as mf
+
+
+class _FakeResp:
+    def __init__(self, payload):
+        self._payload = payload
+    def raise_for_status(self):
+        pass
+    def json(self):
+        return self._payload
+
+
+# Trimmed real-shaped MB recording response.
+MB_PAYLOAD = {
+    "recordings": [
+        {
+            "artist-credit": [{"name": "Daft Punk"}],
+            "releases": [
+                {"date": "2001",
+                 "release-group": {"id": "single-mbid", "title": "Harder, Better, Faster, Stronger",
+                                   "primary-type": "Single", "secondary-types": []}},
+                {"date": "2002",
+                 "release-group": {"id": "comp-mbid", "title": "Musique, Vol. 1",
+                                   "primary-type": "Album", "secondary-types": ["Compilation"]}},
+                {"date": "2001",
+                 "release-group": {"id": "48117b90-a16e-34ca-a514-19c702df1158",
+                                   "title": "Discovery", "primary-type": "Album",
+                                   "secondary-types": []}},
+            ],
+        }
+    ]
+}
+
+
+def test_picks_studio_album_over_single_and_comp(monkeypatch):
+    monkeypatch.setattr(mf.requests, "get", lambda *a, **k: _FakeResp(MB_PAYLOAD))
+    monkeypatch.setattr(mf.time, "sleep", lambda *_: None)
+    out = mf.musicbrainz_best_album("Daft Punk", "Harder Better Faster Stronger")
+    assert out["album_title"] == "Discovery"
+    assert out["artist"] == "Daft Punk"
+    assert out["year"] == "2001"
+    assert out["rg_mbid"] == "48117b90-a16e-34ca-a514-19c702df1158"
+
+
+def test_returns_none_on_empty(monkeypatch):
+    monkeypatch.setattr(mf.requests, "get", lambda *a, **k: _FakeResp({"recordings": []}))
+    monkeypatch.setattr(mf.time, "sleep", lambda *_: None)
+    assert mf.musicbrainz_best_album("Nobody", "Nothing") is None
+
+
+def test_returns_none_on_exception(monkeypatch):
+    def boom(*a, **k):
+        raise mf.requests.exceptions.RequestException("network down")
+    monkeypatch.setattr(mf.requests, "get", boom)
+    monkeypatch.setattr(mf.time, "sleep", lambda *_: None)
+    assert mf.musicbrainz_best_album("Daft Punk", "Discovery") is None
+
+
+def test_falls_back_to_any_releasegroup_when_no_studio(monkeypatch):
+    payload = {"recordings": [{"artist-credit": [{"name": "X"}], "releases": [
+        {"date": "2010", "release-group": {"id": "live1", "title": "Live Thing",
+                                           "primary-type": "Album", "secondary-types": ["Live"]}},
+    ]}]}
+    monkeypatch.setattr(mf.requests, "get", lambda *a, **k: _FakeResp(payload))
+    monkeypatch.setattr(mf.time, "sleep", lambda *_: None)
+    out = mf.musicbrainz_best_album("X", "Y")
+    assert out["album_title"] == "Live Thing"
+
+
+def test_first_artist_credit_only(monkeypatch):
+    payload = {"recordings": [{"artist-credit": [{"name": "SLVMLORD"}, {"name": "Travis Bradley"}],
+                              "releases": [{"date": "2025",
+                                            "release-group": {"id": "x", "title": "Album X",
+                                                              "primary-type": "Album",
+                                                              "secondary-types": []}}]}]}
+    monkeypatch.setattr(mf.requests, "get", lambda *a, **k: _FakeResp(payload))
+    monkeypatch.setattr(mf.time, "sleep", lambda *_: None)
+    out = mf.musicbrainz_best_album("SLVMLORD", "Under My Skin")
+    assert out["artist"] == "SLVMLORD"
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run: `pytest tests/test_musicbrainz.py -v`
+Expected: FAIL — `AttributeError: ... 'musicbrainz_best_album'`
+
+- [ ] **Step 3: Add the implementation**
+
+Add `import time` to the top imports block of `musicfetch` (with `import json`, `import os`, etc.). Then add above `lidarr_search`:
+```python
+MUSICBRAINZ_URL = "https://musicbrainz.org/ws/2"
+MB_HEADERS = {"User-Agent": "musicfetch/2.0 (https://github.com/; personal music fetcher)"}
+_mb_last_call = 0.0
+
+
+def _mb_rate_limit():
+    """Courtesy ~1 req/sec to MusicBrainz."""
+    global _mb_last_call
+    elapsed = time.time() - _mb_last_call
+    if elapsed < 1.0:
+        time.sleep(1.0 - elapsed)
+    _mb_last_call = time.time()
+
+
+def _mb_artist_credit(credit) -> str:
+    """First credited artist name only (ignore featured/secondary)."""
+    if credit and isinstance(credit, list) and isinstance(credit[0], dict):
+        return credit[0].get("name") or (credit[0].get("artist") or {}).get("name", "")
+    return ""
+
+
+def musicbrainz_best_album(artist: str, track: str, timeout: int = 8) -> Optional[dict]:
+    """Resolve 'artist - track' to its best studio album via MusicBrainz.
+    Returns {album_title, artist, year, rg_mbid} or None. Never raises."""
+    query = f'artist:"{artist}" AND recording:"{track}"'
+    try:
+        _mb_rate_limit()
+        resp = requests.get(
+            f"{MUSICBRAINZ_URL}/recording",
+            params={"query": query, "fmt": "json", "limit": 10},
+            headers=MB_HEADERS, timeout=timeout,
+        )
+        resp.raise_for_status()
+        data = resp.json()
+    except Exception as e:  # noqa: BLE001 — degrade to fallback on any failure
+        dbg(f"MusicBrainz lookup failed: {e}")
+        return None
+
+    # candidate = (is_studio, date_sortkey, title, artist, year, mbid)
+    candidates = []
+    for rec in data.get("recordings", []):
+        rec_artist = _mb_artist_credit(rec.get("artist-credit"))
+        for rel in rec.get("releases", []):
+            rg = rel.get("release-group") or {}
+            title = rg.get("title") or rel.get("title") or ""
+            if not title:
+                continue
+            mbid = rg.get("id") or ""
+            primary = rg.get("primary-type") or ""
+            secondary = rg.get("secondary-types") or []
+            date = rel.get("date") or rg.get("first-release-date") or ""
+            is_studio = primary == "Album" and not secondary
+            candidates.append((is_studio, date or "9999", title, rec_artist, date[:4], mbid))
+
+    if not candidates:
+        return None
+    pool = [c for c in candidates if c[0]] or candidates
+    pool.sort(key=lambda c: c[1])  # earliest date first
+    _, _, title, art, year, mbid = pool[0]
+    dbg(f"MusicBrainz resolved '{artist} - {track}' -> '{title}' ({year}) mbid={mbid}")
+    return {"album_title": title, "artist": art or artist, "year": year, "rg_mbid": mbid}
+```
+
+- [ ] **Step 4: Run test to verify it passes**
+
+Run: `pytest tests/test_musicbrainz.py -v`
+Expected: PASS (5 passed)
+
+- [ ] **Step 5: Commit**
+
+```bash
+git add musicfetch tests/test_musicbrainz.py
+git commit -m "feat(lidarr): MusicBrainz track-to-album resolver"
+```
+
+---
+
+### Task 3: Rewrite `lidarr_search` for MBID-exact lookup
+
+**Files:**
+- Modify: `musicfetch` (replace `lidarr_search`; add `_lidarr_album_candidates`, `_lidarr_artist_candidates`, `_fallback_lookup`, `_universal_search`)
+- Test: `tests/test_lidarr_search.py`
+
+- [ ] **Step 1: Write the failing test**
+
+Create `tests/test_lidarr_search.py`:
+```python
+import server.mf  # noqa: F401
+import musicfetch_core as mf
+
+DISCOVERY_MBID = "48117b90-a16e-34ca-a514-19c702df1158"
+
+DISCOVERY_ALBUM = {"title": "Discovery", "artist": {"artistName": "Daft Punk"},
+                   "releaseDate": "2001-01-01", "foreignAlbumId": DISCOVERY_MBID}
+
+
+def test_artist_track_uses_mbid_exact_lookup(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "testkey")
+    monkeypatch.setattr(mf, "musicbrainz_best_album",
+                        lambda artist, track: {"album_title": "Discovery", "artist": "Daft Punk",
+                                               "year": "2001", "rg_mbid": DISCOVERY_MBID})
+    seen = {}
+
+    def fake_get(path, params=None, timeout=15):
+        seen["term"] = (params or {}).get("term")
+        if path == "/api/v1/album/lookup" and seen["term"] == f"mbid:{DISCOVERY_MBID}":
+            return [DISCOVERY_ALBUM]
+        return []
+    monkeypatch.setattr(mf, "lidarr_get", fake_get)
+
+    hits = mf.lidarr_search("Daft Punk - Harder Better Faster Stronger", 10)
+    assert seen["term"] == f"mbid:{DISCOVERY_MBID}"   # exact MBID lookup, not fuzzy
+    assert hits[0].album == "Discovery"
+    assert hits[0].artist == "Daft Punk"
+    assert hits[0].payload["album"]["foreignAlbumId"] == DISCOVERY_MBID
+
+
+def test_year_enriched_from_musicbrainz(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "testkey")
+    monkeypatch.setattr(mf, "musicbrainz_best_album",
+                        lambda artist, track: {"album_title": "Discovery", "artist": "Daft Punk",
+                                               "year": "2001", "rg_mbid": DISCOVERY_MBID})
+    no_year = [{"title": "Discovery", "artist": {"artistName": "Daft Punk"},
+                "releaseDate": "", "foreignAlbumId": DISCOVERY_MBID}]
+    monkeypatch.setattr(mf, "lidarr_get",
+                        lambda path, params=None, timeout=15: no_year if path == "/api/v1/album/lookup" else [])
+    hits = mf.lidarr_search("Daft Punk - Discovery", 10)
+    assert hits[0].year == "2001"
+
+
+def test_no_api_key_returns_empty(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "")
+    assert mf.lidarr_search("Daft Punk - Discovery", 10) == []
+
+
+def test_mb_miss_falls_back_to_lookup(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "testkey")
+    monkeypatch.setattr(mf, "musicbrainz_best_album", lambda artist, track: None)
+    monkeypatch.setattr(mf, "lidarr_get",
+                        lambda path, params=None, timeout=15: [DISCOVERY_ALBUM] if path == "/api/v1/album/lookup" else [])
+    hits = mf.lidarr_search("Daft Punk - Discovery", 10)
+    assert hits[0].album == "Discovery"
+
+
+def test_single_term_is_artist_first(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "testkey")
+
+    def fake_get(path, params=None, timeout=15):
+        if path == "/api/v1/artist/lookup":
+            return [{"artistName": "Daft Punk"}]
+        if path == "/api/v1/album/lookup":
+            return [DISCOVERY_ALBUM]
+        return []
+    monkeypatch.setattr(mf, "lidarr_get", fake_get)
+    hits = mf.lidarr_search("Daft Punk", 10)
+    assert hits[0].kind == "artist"           # bare term -> artist first
+    assert hits[0].artist == "Daft Punk"
+
+
+def test_last_resort_universal_search(monkeypatch):
+    monkeypatch.setattr(mf, "API_KEY", "testkey")
+    monkeypatch.setattr(mf, "musicbrainz_best_album", lambda artist, track: None)
+
+    def fake_get(path, params=None, timeout=15):
+        if path == "/api/v1/search":
+            return [{"album": DISCOVERY_ALBUM}]
+        return []  # album/lookup + artist/lookup empty
+    monkeypatch.setattr(mf, "lidarr_get", fake_get)
+    hits = mf.lidarr_search("Daft Punk - Discovery", 10)
+    assert hits and hits[0].album == "Discovery"
+```
+
+- [ ] **Step 2: Run test to verify it fails**
+
+Run: `pytest tests/test_lidarr_search.py -v`
+Expected: FAIL (current `lidarr_search` ignores MB / `mbid:` lookup)
+
+- [ ] **Step 3: Replace `lidarr_search` and add helpers**
+
+In `musicfetch`, replace the entire existing `def lidarr_search(...)` body (lines ~129-162) with the following, adding the helpers below it:
+```python
+def lidarr_search(query: str, limit: int) -> list[Hit]:
+    """Return Lidarr hits, best match first. Resolves 'Artist - Track' to an
+    album's MusicBrainz release-group MBID, then does an exact Lidarr lookup
+    (term=mbid:<id>) — no fuzzy ranking. Falls back so it never raises and
+    returns [] only on total failure / missing key."""
+    if not API_KEY:
+        err("LIDARR_API_KEY not set — skipping Lidarr search.")
+        return []
+
+    artist, right = _split_query(query)
+
+    if right:
+        mb = musicbrainz_best_album(artist, right)
+        if mb and mb["rg_mbid"]:
+            hits = _lidarr_album_candidates(f"mbid:{mb['rg_mbid']}")
+            for h in hits:
+                if not h.year and mb["year"]:
+                    h.year = mb["year"]
+            if hits:
+                return hits[:limit]
+        # MusicBrainz miss / no exact album → plain lookup (album-first: a dash
+        # query named an album/track).
+        return _fallback_lookup(query, limit, artist_first=False)
+
+    # Bare term is most often an artist.
+    return _fallback_lookup(query, limit, artist_first=True)
+
+
+def _lidarr_album_candidates(term: str) -> list[Hit]:
+    try:
+        return [_album_to_hit(a) for a in lidarr_get("/api/v1/album/lookup", params={"term": term})]
+    except RequestException as e:
+        dbg(f"album/lookup failed: {e}")
+        return []
+
+
+def _lidarr_artist_candidates(term: str) -> list[Hit]:
+    try:
+        return [_artist_to_hit(a) for a in lidarr_get("/api/v1/artist/lookup", params={"term": term})]
+    except RequestException as e:
+        dbg(f"artist/lookup failed: {e}")
+        return []
+
+
+def _fallback_lookup(query: str, limit: int, artist_first: bool) -> list[Hit]:
+    """Plain album + artist lookups (no scoring); /search as last resort."""
+    albums = _lidarr_album_candidates(query)
+    artists = _lidarr_artist_candidates(query)
+    hits = (artists + albums) if artist_first else (albums + artists)
+    if hits:
+        return hits[:limit]
+    return _universal_search(query, limit)
+
+
+def _universal_search(query: str, limit: int) -> list[Hit]:
+    """Last resort: Lidarr's fuzzy /search (unranked)."""
+    hits: list[Hit] = []
+    try:
+        for item in lidarr_get("/api/v1/search", params={"term": query}):
+            if item.get("album"):
+                hits.append(_album_to_hit(item["album"]))
+            elif item.get("artist"):
+                hits.append(_artist_to_hit(item["artist"]))
+    except RequestException as e:
+        dbg(f"/api/v1/search failed: {e}")
+    return hits[:limit]
+```
+
+- [ ] **Step 4: Run tests to verify they pass**
+
+Run: `pytest tests/test_lidarr_search.py -v`
+Expected: PASS (6 passed)
+
+- [ ] **Step 5: Run the full suite**
+
+Run: `pytest -q`
+Expected: all green (prior 27 + new split/musicbrainz/lidarr-search tests), and `python3 -m py_compile musicfetch` clean.
+
+- [ ] **Step 6: Commit**
+
+```bash
+git add musicfetch tests/test_lidarr_search.py
+git commit -m "feat(lidarr): exact MBID album lookup via MusicBrainz resolution"
+```
+
+---
+
+### Task 4: Live verification against the user's Lidarr
+
+**Files:** none (manual verification by the controller, not a subagent).
+
+- [ ] **Step 1: Read-only check — `lidarr_search` resolves the real album**
+
+No mutation; confirms the MB → `mbid:` exact lookup end-to-end:
+```bash
+cd /home/zhering/Documents/musicfetch
+env LIDARR_URL=http://10.2.1.16:8686 LIDARR_API_KEY=49cf02acb4c7436b842df2150056d468 \
+  python3 -c "import server.mf, musicfetch_core as mf; \
+  hits=mf.lidarr_search('Daft Punk - Harder Better Faster Stronger', 5); \
+  print([(h.artist, h.album, h.payload['album'].get('foreignAlbumId')) for h in hits[:3]])"
+```
+Expected: first hit `('Daft Punk', 'Discovery', '48117b90-a16e-34ca-a514-19c702df1158')`.
+
+- [ ] **Step 2: Spot-check a second track** (different artist), e.g.:
+```bash
+env LIDARR_URL=http://10.2.1.16:8686 LIDARR_API_KEY=49cf02acb4c7436b842df2150056d468 \
+  python3 -c "import server.mf, musicfetch_core as mf; \
+  print([(h.artist,h.album) for h in mf.lidarr_search('Tame Impala - The Less I Know The Better',3)])"
+```
+Expected: top hit is the album containing that track (e.g. *Currents*), not a single/compilation.
+
+- [ ] **Step 3: (Optional, mutating) full /fetch** — only with user approval, since it adds the artist+album to their Lidarr. Start the API (`env MUSICFETCH_API_KEY=… LIDARR_URL=http://10.2.1.16:8686 LIDARR_API_KEY=… MUSICFETCH_ROOT=/media/music python3 -m uvicorn server.app:app --port 6769`), `POST /fetch?q=...&source=lidarr`, observe job + Lidarr UI, then clean up any added test artist via `DELETE /api/v1/artist/<id>?deleteFiles=false`.
+
+---
+
+## Self-Review
+
+**Spec coverage:**
+- Shared `lidarr_search` rewrite, same signature → Task 3. ✅
+- MusicBrainz resolver w/ studio release-group selection + first-artist credit → Task 2. ✅
+- `mbid:` exact Lidarr lookup (no fuzzy scoring) → Task 3. ✅
+- Query split → Task 1. ✅
+- Fallback tiers (MB miss → `_fallback_lookup` → `/api/v1/search`; returns [] on total failure / no key) → Task 3 (`test_mb_miss_falls_back_to_lookup`, `test_last_resort_universal_search`, `test_no_api_key_returns_empty`). ✅
+- Year enrichment from MB → Task 3 (`test_year_enriched_from_musicbrainz`). ✅
+- YouTube-fallback preserved (signature unchanged; `[]` on failure) → guaranteed + `test_no_api_key_returns_empty`. ✅
+- Single-term artist-first ordering → Task 3 (`test_single_term_is_artist_first`). ✅
+- Out-of-scope (difflib scoring removed; metadata/quality-profile hardening raised separately) intentionally excluded.
+
+**Placeholder scan:** None — all code and test bodies complete; real MBID/JSON baked in.
+
+**Type consistency:** `lidarr_search(query, limit) -> list[Hit]` unchanged. `musicbrainz_best_album` returns `{album_title, artist, year, rg_mbid}` — keys identical across Task 2 (definition) and Task 3 (consumes `mb["rg_mbid"]`, `mb["year"]`) and tests. `_split_query -> (str, Optional[str])` consistent. `_lidarr_album_candidates`/`_lidarr_artist_candidates`/`_fallback_lookup(query, limit, artist_first)`/`_universal_search(query, limit)` signatures consistent between Task 3 definition and call sites. `_album_to_hit` payload `{"album": {...}}` with `foreignAlbumId` matches the assertions in Task 3.

docs/superpowers/specs/2026-06-08-smarter-lidarr-matching-design.md

@@ -31,7 +31,12 @@ good matches first. Resolve `Artist - Track` → album via MusicBrainz.
 - **Track-first semantics** (`Artist - Track`): the right side is treated as a
   track to resolve to its album. (YouTube path already handles exact tracks; this
   makes Lidarr the accurate album/discography source.)
-- **Scoring** with stdlib `difflib` (no new dependency).
+- **No fuzzy scoring.** Live-verified that Lidarr's `album/lookup` accepts a
+  direct MusicBrainz id: `term=mbid:<release-group-mbid>` (also `term=lidarr:<mbid>`)
+  returns **exactly one** album. So we resolve the album's MBID via MusicBrainz and
+  ask Lidarr for that exact MBID — no difflib, no ranking heuristics. The only
+  selection is deterministic release-group type-filtering inside the MusicBrainz
+  step (prefer studio Album over single/comp/live).
 - **YouTube fallback preserved** exactly as today (see below).
 
 ## Architecture
@@ -43,28 +48,26 @@ musicfetch
 ├── _split_query(query) -> (left, right|None)        # split on first " - "
 ├── musicbrainz_best_album(artist, track) -> dict|None
 │       # MB recording search -> best release-group {album_title, artist, year, rg_mbid}
-├── _similar(a, b) -> float                           # difflib ratio, casefolded
+├── _lidarr_album_candidates(term) / _lidarr_artist_candidates(term) -> list[Hit]
-├── _score_album_hit(hit, want_artist, want_album, rg_mbid) -> float
+├── _universal_search(query, limit) -> list[Hit]     # /api/v1/search last resort
-└── lidarr_search(query, limit) -> list[Hit]          # REWRITTEN: scored, best-first
+└── lidarr_search(query, limit) -> list[Hit]         # REWRITTEN: MBID-exact + fallbacks
 ```
 
 ### Data flow
 
 1. **`Artist - Track` query:**
-   a. `musicbrainz_best_album(artist, track)` → album candidate (title, artist,
+   a. `musicbrainz_best_album(artist, track)` → `{album_title, artist, year, rg_mbid}`.
-      year, release-group MBID).
+   b. Lidarr `GET /api/v1/album/lookup?term=mbid:<rg_mbid>` → 0 or 1 exact album → `Hit`.
-   b. Lidarr `GET /api/v1/album/lookup?term="<artist> <album>"` → map to `Hit`s.
+   c. Enrich `Hit.year` from MB when the Lidarr hit lacks one. Return it.
-   c. Score each: `0.7*_similar(want_artist, hit.artist) + 0.3*_similar(want_album,
+2. **Single-term query (no ` - `):** `_fallback_lookup` — artist-first concatenation
-      hit.album)`, plus a strong bonus (e.g. +0.5) if `hit.payload.album.foreignAlbumId
+   of `/artist/lookup` + `/album/lookup` for the raw term (a bare term is most often
-      == rg_mbid`. Sort desc.
+   an artist). No scoring; the interactive picker / noninteractive top-pick consume
-   d. Enrich `Hit.year` from MB when the Lidarr hit lacks one.
+   the order.
-2. **Single-term query (no ` - `):** Lidarr `/album/lookup` + `/artist/lookup`
+3. **Fallbacks (never regress):** if MusicBrainz misses or the exact MBID lookup
-   with the raw term; score each against the whole query (artist hits scored on
+   returns nothing, use `_fallback_lookup(query)` (album-first there, since a dash
-   artist name, album hits on artist+album); merge, sort desc.
+   query named an album/track). If `/album/lookup` and `/artist/lookup` both yield
-3. **Fallbacks (never regress):** if MB times out / returns nothing, skip to step
+   nothing, fall back to the existing `/api/v1/search`. `lidarr_search` returns `[]`
-   2 using `(artist, track)` recombined as the term. If `/album/lookup` and
+   only when everything fails or the key is missing.
-   `/artist/lookup` both fail, fall back to the existing `/api/v1/search` path.
-   `lidarr_search` returns `[]` only when everything fails or the key is missing.
 
 ### MusicBrainz client details
 
@@ -95,27 +98,26 @@ This feature does not alter fallback behavior:
 - All MB and Lidarr HTTP calls are wrapped; exceptions/timeouts are caught and
   degrade to the next fallback tier. `lidarr_search` never raises.
 - Empty/garbled MB JSON → treated as no match.
-- Existing `DEBUG` logging extended to show MB query, chosen release-group, and
+- Existing `DEBUG` logging extended to show MB query and chosen release-group.
-  top scored candidates.
 
 ## Testing
 
 Unit tests (mock `requests`, no live network):
 - `musicbrainz_best_album`: from canned MB JSON, picks studio Album over a single
-  and a compilation; picks earliest among Albums; returns `None` on empty.
+  and a compilation; picks earliest among Albums; falls back to any release-group
-- `_similar` / `_score_album_hit`: real *Discovery* by Daft Punk outscores the
+  when no studio exists; returns `None` on empty/exception.
-  *Pignickel* novelty for query `Daft Punk - Discovery`-style candidates; MBID
-  match bonus dominates.
 - `_split_query`: `"A - B"` → `("A","B")`; no dash → `("A", None)`; only first
   ` - ` splits.
-- Fallback: MB failure → Lidarr lookup path; lookup failure → `/search`.
+- `lidarr_search`: `Artist - Track` resolves via MB then does an `mbid:` exact
+  lookup returning the real album (year enriched from MB); MB miss → fallback
+  lookup; fallback empty → `/api/v1/search`; no key → `[]`.
 
 Manual live check (end of implementation): with the API pointed at the user's
-Lidarr (`10.2.1.16:8686`), `POST /fetch?q=Daft Punk - Harder Better Faster
+Lidarr (`10.2.1.16:8686`), `lidarr_search("Daft Punk - Harder Better Faster
-Stronger&source=lidarr` resolves to **Discovery** by Daft Punk (not a single,
+Stronger")` resolves to **Discovery** by Daft Punk (the exact MBID
-compilation, or novelty), and the interactive-release flow proceeds.
+`48117b90-a16e-34ca-a514-19c702df1158`), not a single/compilation/novelty.
 
 ## Out of Scope (YAGNI)
 
-Caching MB responses, multi-track/album disambiguation UI, configurable scoring
+Caching MB responses, multi-track/album disambiguation UI, fuzzy similarity
-weights, fuzzy artist aliasing beyond difflib, MB cover-art lookup.
+scoring (eliminated by the `mbid:` exact lookup), MB cover-art lookup.