Shrink collection summary to fit OpenAI token limit

Deduplicate records by musicbrainz_id (merging formats)
and cap genres to 3 per record, reducing the summary from
~32K to ~24K estimated tokens.

lib/music_library/chats/collection_chat.ex

@@ -8,9 +8,10 @@ defmodule MusicLibrary.Chats.CollectionChat do
   alias MusicLibrary.Chats.Prompt
 
   @impl true
-  @spec stream_response([map()], String.t(), (String.t() -> any())) :: :ok | {:error, term()}
-  def stream_response(messages, collection_summary, callback) do
-    instructions = build_instructions(collection_summary)
+  @spec stream_response([map()], {String.t(), non_neg_integer()}, (String.t() -> any())) ::
+          :ok | {:error, term()}
+  def stream_response(messages, {summary, record_count}, callback) do
+    instructions = build_instructions(summary, record_count)
 
     OpenAI.chat_stream(messages,
       on_chunk: callback,
@@ -19,9 +20,7 @@ defmodule MusicLibrary.Chats.CollectionChat do
     )
   end
 
-  defp build_instructions(collection_summary) do
-    record_count = count_records(collection_summary)
-
+  defp build_instructions(collection_summary, record_count) do
     Prompt.build("""
     Answer questions about the user's music collection. \
     Use the provided collection catalog as your primary reference. \
@@ -31,12 +30,4 @@ defmodule MusicLibrary.Chats.CollectionChat do
     #{collection_summary}\
     """)
   end
-
-  defp count_records(""), do: 0
-
-  defp count_records(summary) do
-    summary
-    |> String.split("\n")
-    |> length()
-  end
 end

lib/music_library/collection.ex

@@ -191,7 +191,9 @@ defmodule MusicLibrary.Collection do
     |> MapSet.new()
   end
 
-  @spec collection_summary() :: String.t()
+  @max_genres_per_record 3
+
+  @spec collection_summary() :: {String.t(), non_neg_integer()}
   def collection_summary do
     records =
       from(r in Record,
@@ -201,16 +203,31 @@ defmodule MusicLibrary.Collection do
       )
       |> Repo.all()
 
-    records
-    |> Enum.map_join("\n", &format_record_line/1)
+    record_count = length(records)
+
+    summary =
+      records
+      |> Enum.group_by(& &1.musicbrainz_id)
+      |> Enum.map(fn {_id, group} -> format_group(group) end)
+      |> Enum.sort()
+      |> Enum.join("\n")
+
+    {summary, record_count}
   end
 
-  defp format_record_line(record) do
+  defp format_group(records) do
+    record = hd(records)
     artist_names = Record.artist_names(record)
-    genres = record.genres || []
+    formats = records |> Enum.map(& &1.format) |> Enum.uniq() |> Enum.join("/")
+
+    genres =
+      records
+      |> Enum.flat_map(&(&1.genres || []))
+      |> Enum.uniq()
+      |> Enum.take(@max_genres_per_record)
 
     base =
-      "#{artist_names} - #{record.title} (#{record.release_date || "Unknown"}, #{record.format}, #{record.type})"
+      "#{artist_names} - #{record.title} (#{record.release_date || "Unknown"}, #{formats}, #{record.type})"
 
     if genres == [] do
       base

test/music_library/chats/collection_chat_test.exs

@@ -31,7 +31,7 @@ defmodule MusicLibrary.Chats.CollectionChatTest do
     summary =
       "Radiohead - OK Computer (1997-06-16, cd, album) [alternative rock, art rock]\nPink Floyd - The Wall (1979-11-30, vinyl, album) [progressive rock]"
 
-    assert :ok = CollectionChat.stream_response([], summary, fn _chunk -> :ok end)
+    assert :ok = CollectionChat.stream_response([], {summary, 2}, fn _chunk -> :ok end)
 
     assert_receive {:captured_instructions, instructions}
     assert instructions =~ summary
@@ -43,7 +43,7 @@ defmodule MusicLibrary.Chats.CollectionChatTest do
   test "stream_response handles empty collection" do
     stub_and_capture_instructions(self())
 
-    assert :ok = CollectionChat.stream_response([], "", fn _chunk -> :ok end)
+    assert :ok = CollectionChat.stream_response([], {"", 0}, fn _chunk -> :ok end)
 
     assert_receive {:captured_instructions, instructions}
     assert instructions =~ "0 records"

test/music_library/collection_test.exs

@@ -298,11 +298,11 @@ defmodule MusicLibrary.CollectionTest do
   end
 
   describe "collection_summary/0" do
-    test "returns empty string when collection is empty" do
-      assert Collection.collection_summary() == ""
+    test "returns empty string and zero count when collection is empty" do
+      assert Collection.collection_summary() == {"", 0}
     end
 
-    test "returns one line per collected record" do
+    test "returns one line per collected record with record count" do
       record_with_artist("Radiohead", %{
         title: "OK Computer",
         format: :cd,
@@ -321,9 +321,10 @@ defmodule MusicLibrary.CollectionTest do
         purchased_at: ~U[2024-01-02 00:00:00Z]
       })
 
-      summary = Collection.collection_summary()
+      {summary, record_count} = Collection.collection_summary()
       lines = String.split(summary, "\n")
 
+      assert record_count == 2
       assert length(lines) == 2
 
       assert Enum.any?(lines, fn line ->
@@ -339,6 +340,54 @@ defmodule MusicLibrary.CollectionTest do
              end)
     end
 
+    test "deduplicates records with same musicbrainz_id and merges formats" do
+      mbid = Ecto.UUID.generate()
+
+      record_with_artist("AC/DC", %{
+        title: "Highway to Hell",
+        format: :cd,
+        type: :album,
+        genres: ["hard rock", "rock"],
+        release_date: "1979-07-27",
+        musicbrainz_id: mbid,
+        purchased_at: ~U[2024-01-01 00:00:00Z]
+      })
+
+      record_with_artist("AC/DC", %{
+        title: "Highway to Hell",
+        format: :vinyl,
+        type: :album,
+        genres: ["hard rock", "rock"],
+        release_date: "1979-07-27",
+        musicbrainz_id: mbid,
+        purchased_at: ~U[2024-01-02 00:00:00Z]
+      })
+
+      {summary, record_count} = Collection.collection_summary()
+      lines = String.split(summary, "\n")
+
+      assert record_count == 2
+      assert length(lines) == 1
+      assert hd(lines) =~ "cd/vinyl"
+    end
+
+    test "caps genres to 3 per record" do
+      record_with_artist("ABC", %{
+        title: "The Lexicon of Love",
+        format: :vinyl,
+        type: :album,
+        genres: ~w[dance dance-pop disco electronic new-wave pop synth-pop],
+        release_date: "1982-06-21",
+        purchased_at: ~U[2024-01-01 00:00:00Z]
+      })
+
+      {summary, _count} = Collection.collection_summary()
+      [genres_str] = Regex.run(~r/\[(.+)\]/, summary, capture: :all_but_first)
+      genre_count = genres_str |> String.split(", ") |> length()
+
+      assert genre_count == 3
+    end
+
     test "omits genre brackets when genres are empty" do
       record_with_artist("Radiohead", %{
         title: "OK Computer",
@@ -349,7 +398,7 @@ defmodule MusicLibrary.CollectionTest do
         purchased_at: ~U[2024-01-01 00:00:00Z]
       })
 
-      summary = Collection.collection_summary()
+      {summary, _count} = Collection.collection_summary()
       refute summary =~ "["
       refute summary =~ "]"
     end
@@ -365,7 +414,8 @@ defmodule MusicLibrary.CollectionTest do
         purchased_at: nil
       })
 
-      summary = Collection.collection_summary()
+      {summary, record_count} = Collection.collection_summary()
+      assert record_count == 1
       assert summary =~ "OK Computer"
       refute summary =~ "Wish You Were Here"
     end