video enrichment: pull everything TMDB/TVDB offer + backfill gaps only

Enrichment now harvests the full detail payload (same call, no extra requests):
- TMDB: tagline, genres, rating (vote_average), runtime, status, first/last air
  date (shows), release date + runtime (movies) — on top of overview/backdrop/ids.
- TVDB: switches to /series/<id>/extended for overview + genres.

enrichment_apply now uses BACKFILL semantics: metadata columns are written via
COALESCE(NULLIF(col,''), ?) so enrichment only fills fields the media server
left empty — it never clobbers server-provided data. Genres backfill to the
normalised link tables only when the item has none yet. Whitelist expanded for
the new columns.

Seam tests: backfill-only (server overview/genres kept, gaps filled), genre
backfill when empty, TMDB full-metadata extraction.

core/video/enrichment/clients.py

@@ -83,10 +83,24 @@ class TMDBClient:
                 meta["backdrop_url"] = self.IMG + dr["backdrop_path"]
             ext = dr.get("external_ids") or {}
             meta["imdb_id"] = ext.get("imdb_id") or dr.get("imdb_id")
+            # Everything TMDB offers (same call) — the worker backfills only the
+            # gaps the server left.
+            meta["tagline"] = dr.get("tagline")
+            meta["status"] = dr.get("status")
+            if dr.get("vote_average"):
+                meta["rating"] = dr.get("vote_average")
+            gs = [g.get("name") for g in (dr.get("genres") or []) if g.get("name")]
+            if gs:
+                meta["genres"] = gs
             if kind == "movie":
                 meta["release_date"] = dr.get("release_date")
+                meta["runtime_minutes"] = dr.get("runtime")
             else:
-                meta["status"] = dr.get("status")
+                meta["first_air_date"] = dr.get("first_air_date")
+                meta["last_air_date"] = dr.get("last_air_date")
+                ert = dr.get("episode_run_time") or []
+                if ert:
+                    meta["runtime_minutes"] = ert[0]
                 meta["tvdb_id"] = _int(ext.get("tvdb_id"))
         except Exception:
             logger.exception("TMDB details fetch failed for %s", title or tmdb_id)
@@ -157,10 +171,15 @@ class TVDBClient:
             tvdb_id = _int(top.get("tvdb_id") or top.get("id"))
             meta["overview"] = top.get("overview")
         else:
-            # Known id from the server → fetch the series directly for its overview.
+            # Known id from the server → fetch the extended record (overview +
+            # genres + everything TVDB offers).
             try:
-                dr = self._authed_get("/series/" + str(tvdb_id))
+                dr = self._authed_get("/series/" + str(tvdb_id) + "/extended")
-                meta["overview"] = ((dr or {}).get("data") or {}).get("overview")
+                sd = (dr or {}).get("data") or {}
+                meta["overview"] = sd.get("overview")
+                gs = [g.get("name") for g in (sd.get("genres") or []) if g.get("name")]
+                if gs:
+                    meta["genres"] = gs
             except Exception:
                 logger.exception("TVDB details fetch failed for %s", title or tvdb_id)
         if tvdb_id is None:

database/video_database.py

@@ -63,8 +63,10 @@ _ENRICH = {
 # arbitrary keys; backfill semantics applied by the caller).
 _ENRICH_META_COLS = {
     "movies": {"overview", "backdrop_url", "release_date", "status", "content_rating",
-               "runtime_minutes", "studio", "imdb_id", "tmdb_id"},
+               "runtime_minutes", "studio", "tagline", "rating", "rating_critic",
+               "imdb_id", "tmdb_id"},
     "shows": {"overview", "backdrop_url", "status", "network", "content_rating",
+              "tagline", "rating", "first_air_date", "last_air_date",
               "imdb_id", "tmdb_id", "tvdb_id"},
 }
 
@@ -222,7 +224,9 @@ class VideoDatabase:
                     continue
                 if not include_ids and col in id_cols:
                     continue
-                sets.append(f"{col}=?")
+                # BACKFILL: only fill a column the server left empty — enrichment
+                # fills gaps, it never clobbers data the media server provided.
+                sets.append(f"{col}=COALESCE(NULLIF({col}, ''), ?)")
                 params.append(val)
             params.append(item_id)
             return f"UPDATE {tbl} SET {', '.join(sets)} WHERE id=?", params
@@ -232,12 +236,21 @@ class VideoDatabase:
             sql, params = build(True)
             try:
                 conn.execute(sql, params)
-                conn.commit()
             except sqlite3.IntegrityError:
                 conn.rollback()
                 sql, params = build(False)   # keep existing id, just record status/metadata
                 conn.execute(sql, params)
-                conn.commit()
+            # Genres backfill — only when the item has none yet (enrichment fills
+            # the gap the server didn't). Written to the normalised link tables.
+            genres = (metadata or {}).get("genres")
+            link = {"movies": ("movie_genres", "movie_id"),
+                    "shows": ("show_genres", "show_id")}.get(tbl)
+            if matched and genres and link:
+                lt, oc = link
+                has = conn.execute(f"SELECT 1 FROM {lt} WHERE {oc}=? LIMIT 1", (item_id,)).fetchone()
+                if not has:
+                    self._set_genres(conn, lt, oc, item_id, genres)
+            conn.commit()
         finally:
             conn.close()
 

tests/test_video_database.py

@@ -492,6 +492,27 @@ def test_enrichment_breakdown_unmatched_retry(db):
     assert db.enrichment_breakdown("tmdb")["movie"]["pending"] == 1
 
 
+def test_enrichment_backfills_only_gaps_never_clobbers_server(db):
+    # Server gave overview + a genre; enrichment must fill the EMPTY fields
+    # (tagline/rating) but leave the server's overview + genres untouched.
+    mid = db.upsert_movie("plex", {"server_id": "m1", "title": "Dune",
+                                   "overview": "server overview", "genres": ["Sci-Fi"]})
+    db.enrichment_apply("tmdb", "movie", mid, matched=True, external_id=438631, metadata={
+        "overview": "tmdb overview", "tagline": "Fear is the mind-killer",
+        "rating": 8.4, "genres": ["Drama"]})
+    d = db.movie_detail(mid)
+    assert d["overview"] == "server overview"     # NOT clobbered
+    assert d["tagline"] == "Fear is the mind-killer" and d["rating"] == 8.4   # gaps filled
+    assert d["genres"] == ["Sci-Fi"]              # had genres → enrichment left them
+
+
+def test_enrichment_backfills_genres_when_item_has_none(db):
+    mid = db.upsert_movie("plex", {"server_id": "m1", "title": "X"})   # no genres
+    db.enrichment_apply("tmdb", "movie", mid, matched=True, external_id=1,
+                        metadata={"genres": ["Drama", "Comedy"]})
+    assert db.movie_detail(mid)["genres"] == ["Comedy", "Drama"]
+
+
 def test_error_status_is_distinct_and_retryable_in_ui(db):
     a = db.upsert_movie("plex", {"server_id": "m1", "title": "A"})
     db.enrichment_apply("tmdb", "movie", a, matched=False, error=True)

tests/test_video_enrichment.py

@@ -83,6 +83,21 @@ def test_tmdb_client_with_known_id_skips_the_search(monkeypatch):
     assert not any("/search/" in u for u in urls)
 
 
+def test_tmdb_pulls_full_metadata(monkeypatch):
+    # Everything TMDB offers comes from the one detail call.
+    class _Resp:
+        def __init__(self, b): self._b = b
+        def raise_for_status(self): pass
+        def json(self): return self._b
+    detail = {"overview": "O", "tagline": "Fear", "vote_average": 8.4, "runtime": 155,
+              "genres": [{"name": "Sci-Fi"}, {"name": "Drama"}], "status": "Released",
+              "backdrop_path": "/b.jpg", "external_ids": {"imdb_id": "tt1"}}
+    monkeypatch.setitem(sys.modules, "requests", types.SimpleNamespace(get=lambda u, **k: _Resp(detail)))
+    m = TMDBClient("KEY").match("movie", "Dune", 2021, known_id=438631)["metadata"]
+    assert m["tagline"] == "Fear" and m["rating"] == 8.4 and m["runtime_minutes"] == 155
+    assert m["genres"] == ["Sci-Fi", "Drama"] and m["status"] == "Released" and m["imdb_id"] == "tt1"
+
+
 def test_tmdb_client_raises_on_rate_limit(monkeypatch):
     # A 429 must raise (→ worker records 'error'), not silently return no-match.
     class _Resp429: