youtube scans: dedup against downloaded videos so they don't re-download

a completed youtube download is removed from the wishlist (correct), but the
channel/playlist scans' downloaded_ids seam was stubbed to [] — so the next scan,
not seeing it wishlisted and not knowing it was downloaded, re-added it (still in
the channel's recent uploads / last-N net) → re-download loop for the last-N
videos every scan.

fix: db.downloaded_youtube_video_ids() pulls completed youtube grabs from the
permanent download history; both scans' _default_downloaded_ids now use it. so a
downloaded video stays gone. tested (db method + the scans already exclude
downloaded_ids in their pure logic).

core/automation/handlers/video_scan_watchlist_channels.py

@@ -143,9 +143,10 @@ def _default_dismissed_ids(channel_id: Any) -> List[Any]:
 
 
 def _default_downloaded_ids(channel_id: Any) -> List[Any]:
-    # The permanent download-history ledger has no YouTube grabs until the fulfillment
-    # engine runs; empty for now, so the scan is correct the day those rows appear.
-    return []
+    # Already-downloaded YouTube videos (global; a video id is unique). A completed download
+    # leaves the wishlist, so without this the scan would re-add + re-download it.
+    from api.video import get_video_db
+    return get_video_db().downloaded_youtube_video_ids()
 
 
 def _default_add_videos(channel: Dict[str, Any], videos: List[Dict[str, Any]]) -> int:

core/automation/handlers/video_scan_watchlist_playlists.py

@@ -86,7 +86,10 @@ def _default_wishlisted_ids(playlist_id: Any) -> List[Any]:
 
 
 def _default_downloaded_ids(playlist_id: Any) -> List[Any]:
-    return []   # populated once the download-history ledger carries youtube grabs
+    # Already-downloaded YouTube videos (global). A completed download leaves the wishlist,
+    # so without this the scan would re-add + re-download it.
+    from api.video import get_video_db
+    return get_video_db().downloaded_youtube_video_ids()
 
 
 def _default_dismissed_ids(playlist_id: Any) -> List[Any]:

database/video_database.py

@@ -1459,6 +1459,19 @@ class VideoDatabase:
         finally:
             conn.close()
 
+    def downloaded_youtube_video_ids(self) -> list:
+        """YouTube video ids already successfully downloaded (from the permanent history).
+        The watchlist scans exclude these — a completed video is removed from the wishlist
+        on download, so without this it would be re-added (it's still in the channel's recent
+        uploads / last-N net) and re-downloaded on every scan."""
+        conn = self._get_connection()
+        try:
+            return [r["media_id"] for r in conn.execute(
+                "SELECT DISTINCT media_id FROM video_download_history "
+                "WHERE source='youtube' AND outcome='completed' AND media_id IS NOT NULL")]
+        finally:
+            conn.close()
+
     def media_tmdb_id(self, kind: str, media_id) -> tuple:
         """(tmdb_id, imdb_id) for a library movie/show row — used to resolve sidecar /
         subtitle metadata for an owned re-grab (whose media_id is the library id, not a

tests/test_video_process_youtube_wishlist.py

@@ -176,6 +176,17 @@ def test_youtube_wishlist_to_download_shape(db):
     assert top["video_title"] == "Second" and top["published_at"] == "2024-05-01"
 
 
+def test_downloaded_youtube_video_ids_only_completed_youtube(db):
+    # the dedup the scans use so a downloaded video (removed from the wishlist) isn't re-added
+    db.record_download_history({"id": 1, "kind": "youtube", "source": "youtube",
+                                "media_id": "v1", "status": "completed", "dest_path": "/a.mp4"})
+    db.record_download_history({"id": 2, "kind": "youtube", "source": "youtube",
+                                "media_id": "v2", "status": "failed"})          # failed → not counted
+    db.record_download_history({"id": 3, "kind": "movie", "source": "soulseek",
+                                "media_id": "99", "status": "completed"})        # not youtube
+    assert set(db.downloaded_youtube_video_ids()) == {"v1"}
+
+
 def test_count_and_claim_queue(db):
     a = db.add_video_download({"kind": "youtube", "source": "youtube", "media_id": "v1",
                                "title": "A", "status": "queued"})