attempt to improve analysis

api/src/party/__tests__/question-utils.test.ts

@@ -1,6 +1,12 @@
 import { describe, expect, it } from "vitest";
 
-import { getQuestionRange, getReleaseYearRange } from "../question-utils";
+import type { PartyAnalytics } from "../question-utils";
+import {
+	getQuestionRange,
+	getReleaseYearRange,
+	getTopClusterArtists,
+	getTopClusterTracks,
+} from "../question-utils";
 
 describe("question range helpers", () => {
 	it("normalizes inverted generic ranges", () => {
@@ -25,3 +31,34 @@ describe("question range helpers", () => {
 		expect(range.max).toBeLessThanOrEqual(2026);
 	});
 });
+
+describe("analytics cluster helpers", () => {
+	it("returns usable tracks and artists across all story clusters", () => {
+		const analytics: PartyAnalytics = {
+			storyClusters: [
+				{
+					tracks: [
+						{ name: "Shared Track", artists: [{ name: "Shared Artist" }] },
+					],
+					artists: [{ name: "Shared Artist" }],
+				},
+				{
+					tracks: [
+						{ name: "Solo Track", artists: [{ name: "Solo Artist" }] },
+						{ name: "" },
+					],
+					artists: [{ name: "Solo Artist" }, { name: "" }],
+				},
+			],
+		};
+
+		expect(getTopClusterTracks(analytics).map((track) => track.name)).toEqual([
+			"Shared Track",
+			"Solo Track",
+		]);
+		expect(getTopClusterArtists(analytics)).toEqual([
+			"Shared Artist",
+			"Solo Artist",
+		]);
+	});
+});

api/src/party/question-utils.ts

@@ -39,6 +39,7 @@ export type AnalyticsTrack = {
 	albumName?: string;
 	memberScores?: { userId: string; score: number }[];
 };
+type AnalyticsArtist = { name: string };
 type QuestionLike = {
 	text: string;
 	questionKey?: string;
@@ -180,17 +181,13 @@ function normalizeQuestionKey(value: string): string {
 }
 
 export function getTopClusterArtists(analytics: PartyAnalytics): string[] {
-	return (analytics?.storyClusters?.[0]?.artists ?? []).map(
-		(artist) => artist.name,
-	);
+	return getAllClusterArtists(analytics).map((artist) => artist.name);
 }
 
 export function getTopClusterTracks(
 	analytics: PartyAnalytics,
 ): AnalyticsTrack[] {
-	return (analytics?.storyClusters?.[0]?.tracks ?? []).filter((track) =>
-		isUsableText(track.name),
-	);
+	return getAllClusterTracks(analytics);
 }
 
 export function pickRelevantTrack(
@@ -338,9 +335,9 @@ async function collectSongCandidates({
 	);
 	push(peopleSong);
 
-	const topClusterTracks = [...(analytics?.storyClusters?.[0]?.tracks ?? [])]
-		.filter((track) => isUsableText(track.name))
-		.sort((a, b) => getTrackScore(b) - getTrackScore(a));
+	const topClusterTracks = getAllClusterTracks(analytics).sort(
+		(a, b) => getTrackScore(b) - getTrackScore(a),
+	);
 
 	for (const track of topClusterTracks) {
 		const song = await resolveQuestionSong(db, analytics, {
@@ -400,12 +397,9 @@ async function resolveSongFromMentionedPeople(
 
 	if (userIds.length === 0) return null;
 
-	const tracks = [...(analytics?.storyClusters?.[0]?.tracks ?? [])]
-		.filter((track) => isUsableText(track.name))
-		.sort(
-			(a, b) =>
-				getMemberTrackScore(b, userIds) - getMemberTrackScore(a, userIds),
-		);
+	const tracks = getAllClusterTracks(analytics).sort(
+		(a, b) => getMemberTrackScore(b, userIds) - getMemberTrackScore(a, userIds),
+	);
 
 	for (const track of tracks) {
 		const song = await resolveQuestionSong(db, analytics, {
@@ -484,6 +478,43 @@ function getMemberTrackScore(
 	}, 0);
 }
 
+function getAllClusterTracks(analytics: PartyAnalytics): AnalyticsTrack[] {
+	const tracks: AnalyticsTrack[] = [];
+	const seen = new Set<string>();
+
+	for (const cluster of analytics?.storyClusters ?? []) {
+		for (const track of cluster.tracks ?? []) {
+			if (!isUsableText(track.name)) continue;
+			const key = [
+				track.name,
+				track.albumName ?? "",
+				track.artists?.map((artist) => artist.name).join("|") ?? "",
+			].join("::");
+			if (seen.has(key)) continue;
+			seen.add(key);
+			tracks.push(track);
+		}
+	}
+
+	return tracks;
+}
+
+function getAllClusterArtists(analytics: PartyAnalytics): AnalyticsArtist[] {
+	const artists: AnalyticsArtist[] = [];
+	const seen = new Set<string>();
+
+	for (const cluster of analytics?.storyClusters ?? []) {
+		for (const artist of cluster.artists ?? []) {
+			if (!isUsableText(artist.name)) continue;
+			if (seen.has(artist.name)) continue;
+			seen.add(artist.name);
+			artists.push(artist);
+		}
+	}
+
+	return artists;
+}
+
 export function isUsableText(
 	value: string | null | undefined,
 ): value is string {

api/src/workflows/__tests__/party-analysis.test.ts

@@ -239,6 +239,40 @@ describe("PartyAnalysisWorkflow", () => {
 	});
 
 	describe("analyzeParty - story clusters", () => {
+		it("keeps names for entities that only appear for later members", async () => {
+			const { partyId, userIdB } = await seedPartyWithTwoSimilarUsers();
+			const uniqueAlbum = await createAlbum("User B Unique Album");
+			const uniqueArtist = await createArtist("User B Unique Artist");
+			const uniqueTrack = await createTrack(
+				"User B Unique Track",
+				uniqueAlbum.id,
+				[uniqueArtist.id],
+			);
+			await addTopTrack(userIdB, uniqueTrack.id, 2);
+			await addTopArtist(userIdB, uniqueArtist.id, 2);
+
+			const result = await partyAnalysisWorkflow.analyzeParty(partyId);
+			const allTracks = result.storyClusters.flatMap(
+				(cluster) => cluster.tracks,
+			);
+			const allArtists = result.storyClusters.flatMap(
+				(cluster) => cluster.artists,
+			);
+
+			expect(allTracks).toContainEqual(
+				expect.objectContaining({
+					id: uniqueTrack.id,
+					name: "User B Unique Track",
+				}),
+			);
+			expect(allArtists).toContainEqual(
+				expect.objectContaining({
+					id: uniqueArtist.id,
+					name: "User B Unique Artist",
+				}),
+			);
+		});
+
 		it("sorts clusters with all-member cluster first", async () => {
 			const { partyId, sharedTrackId } = await seedPartyWithTwoSimilarUsers();
 

api/src/workflows/party-analysis.ts

@@ -469,8 +469,9 @@ export class PartyAnalysisWorkflow extends ConfiguredInstance {
 		const result = new Map<string, TrackEntityScore>();
 
 		for (const [trackId, memberScores] of entityMap) {
-			const firstData = memberData.values().next().value;
-			const trackInfo = firstData?.tracks.get(trackId) ?? {
+			const trackInfo = Array.from(memberData.values())
+				.find((data) => data.tracks.has(trackId))
+				?.tracks.get(trackId) ?? {
 				name: "",
 				artists: [],
 				albumName: undefined,
@@ -525,8 +526,9 @@ export class PartyAnalysisWorkflow extends ConfiguredInstance {
 				}
 			}
 
-			const firstData = memberData.values().next().value;
-			const artistInfo = firstData?.artists.get(artistId) ?? { name: "" };
+			const artistInfo = Array.from(memberData.values())
+				.find((data) => data.artists.has(artistId))
+				?.artists.get(artistId) ?? { name: "" };
 
 			result.set(artistId, {
 				id: artistId,
@@ -869,7 +871,8 @@ export class PartyAnalysisWorkflow extends ConfiguredInstance {
 		genres.sort(
 			(a, b) =>
 				b.memberCount - a.memberCount ||
-				b.memberScores.length - a.memberScores.length,
+				b.memberScores.reduce((total, member) => total + member.score, 0) -
+					a.memberScores.reduce((total, member) => total + member.score, 0),
 		);
 
 		// Return top genres that are shared by at least 2 members