Deduplicate tracks in ListeningStats queries

When multiple records share MusicBrainz release IDs, the left joins in
ListeningStats fan out — producing duplicate rows per scrobbled track.
Fix by grouping release subqueries on release_id with MIN(record_id)
before joining, so each release maps to exactly one record.

lib/music_library/listening_stats.ex

@@ -269,9 +269,9 @@ defmodule MusicLibrary.ListeningStats do
         distinct: true
 
     from t in Track,
-      left_join: cr in subquery(Collection.collected_releases_query()),
+      left_join: cr in subquery(unique_collected_releases_query()),
       on: cr.release_id == fragment("? ->> '$.musicbrainz_id'", t.album),
-      left_join: wr in subquery(Wishlist.wishlisted_releases_query()),
+      left_join: wr in subquery(unique_wishlisted_releases_query()),
       on: wr.release_id == fragment("? ->> '$.musicbrainz_id'", t.album),
       left_join: ar in subquery(all_artists_query),
       on: wr.record_id == ar.record_id or cr.record_id == ar.record_id,
@@ -286,9 +286,9 @@ defmodule MusicLibrary.ListeningStats do
 
   defp top_albums_base_query do
     from t in Track,
-      left_join: cr in subquery(Collection.collected_releases_query()),
+      left_join: cr in subquery(unique_collected_releases_query()),
       on: cr.release_id == fragment("? ->> '$.musicbrainz_id'", t.album),
-      left_join: wr in subquery(Wishlist.wishlisted_releases_query()),
+      left_join: wr in subquery(unique_wishlisted_releases_query()),
       on: wr.release_id == fragment("? ->> '$.musicbrainz_id'", t.album),
       where: fragment("json_extract(album, '$.title') != ''"),
       group_by: [
@@ -325,6 +325,26 @@ defmodule MusicLibrary.ListeningStats do
       order_by: [desc: count(t.scrobbled_at_uts, :distinct)]
   end
 
+  defp unique_collected_releases_query do
+    from rr in subquery(Collection.collected_releases_query()),
+      group_by: rr.release_id,
+      select: %{
+        record_id: min(rr.record_id),
+        cover_hash: min(rr.cover_hash),
+        release_id: rr.release_id
+      }
+  end
+
+  defp unique_wishlisted_releases_query do
+    from rr in subquery(Wishlist.wishlisted_releases_query()),
+      group_by: rr.release_id,
+      select: %{
+        record_id: min(rr.record_id),
+        cover_hash: min(rr.cover_hash),
+        release_id: rr.release_id
+      }
+  end
+
   defp cutoff_timestamp(days, opts) do
     current_time = Keyword.get_lazy(opts, :current_time, &DateTime.utc_now/0)
     timezone = Keyword.get(opts, :timezone, &MusicLibrary.default_timezone/0)

test/music_library/listening_stats_test.exs

@@ -286,6 +286,82 @@ defmodule MusicLibrary.ListeningStatsTest do
     end
   end
 
+  describe "deduplication when multiple records share release IDs" do
+    setup do
+      # Create two collected records that share the same MusicBrainz release IDs
+      # (both use the default marbles fixture with identical musicbrainz_data).
+      # This produces two rows per release_id in record_releases, which previously
+      # caused fan-out in ListeningStats joins.
+      record_a = RecordsFixtures.record(%{title: "Marbles Copy A"})
+      record_b = RecordsFixtures.record(%{title: "Marbles Copy B"})
+
+      # Pick a release_id shared by both records
+      shared_release_id = "d3f9b9e2-73f5-4b47-a2a7-2c2199aad608"
+      now = System.system_time(:second)
+
+      track_fixture(%{
+        title: "The Invisible Man",
+        album_title: "Marbles",
+        album_musicbrainz_id: shared_release_id,
+        artist_name: "Marillion",
+        scrobbled_at_uts: now - 100
+      })
+
+      track_fixture(%{
+        title: "You're Gone",
+        album_title: "Marbles",
+        album_musicbrainz_id: shared_release_id,
+        artist_name: "Marillion",
+        scrobbled_at_uts: now - 200
+      })
+
+      # The lexicographically smaller record_id should consistently win (MIN)
+      expected_record_id = Enum.min([record_a.id, record_b.id])
+
+      %{
+        expected_record_id: expected_record_id,
+        shared_release_id: shared_release_id
+      }
+    end
+
+    test "list_tracks returns one row per track", %{expected_record_id: expected_record_id} do
+      tracks = ListeningStats.list_tracks()
+
+      titles = Enum.map(tracks, & &1.track.title)
+      assert length(titles) == length(Enum.uniq(titles)), "list_tracks returned duplicate rows"
+
+      # Both tracks should map to the same deterministic record_id
+      for result <- tracks do
+        assert result.collected_record_id == expected_record_id
+      end
+    end
+
+    test "recent_activity returns one row per track", %{expected_record_id: expected_record_id} do
+      %{recent_tracks: recent_tracks} =
+        ListeningStats.recent_activity("Etc/UTC", 20)
+
+      titles = Enum.map(recent_tracks, & &1.track.title)
+
+      assert length(titles) == length(Enum.uniq(titles)),
+             "recent_activity returned duplicate rows"
+
+      for result <- recent_tracks do
+        assert result.collected_record_id == expected_record_id
+      end
+    end
+
+    test "get_top_albums returns one entry per album", %{expected_record_id: expected_record_id} do
+      results = ListeningStats.get_top_albums(limit: 10)
+
+      marbles_entries = Enum.filter(results, fn r -> r.album_title == "Marbles" end)
+      assert length(marbles_entries) == 1, "get_top_albums returned duplicate album entries"
+
+      [entry] = marbles_entries
+      assert entry.play_count == 2
+      assert entry.collected_record_id == expected_record_id
+    end
+  end
+
   describe "get_top_artists/1" do
     test "counts tracks with missing artist_infos records" do
       artist_info =