Add discover features to YouTube app for enhanced channel analysis

- Introduced new API routes for generating channel DNA profiles, analyzing content evolution, and identifying hidden gems.
- Implemented UI components for displaying channel trivia, evolution timelines, and rabbit hole connections.
- Enhanced the discover view to integrate various analysis features, providing users with a comprehensive overview of channel performance and personality.
- Updated command palette and context rail to include navigation options for the new discover features.

Made-with: Cursor

Author: joaaogui

apps/youtube/src/app/api/discover/dna/[channelId]/route.ts

@@ -0,0 +1,118 @@
+import { db } from "@/db";
+import { transcripts, videos } from "@/db/schema";
+import { getModel } from "@/lib/ai-providers";
+import { auth } from "@/lib/auth";
+import { createTaggedLogger } from "@/lib/logger";
+import {
+  checkRateLimit,
+  getClientIp,
+  RATE_LIMITS,
+} from "@/lib/rate-limit";
+import { withErrorHandling } from "@/lib/route-handler";
+import {
+  corsHeaders,
+  mergeHeaders,
+  optionsResponse,
+  rateLimitExceededResponse,
+  withRateLimitHeaders,
+} from "@data-projects/shared";
+import { generateText } from "ai";
+import { eq } from "drizzle-orm";
+
+const log = createTaggedLogger("discover-dna");
+
+const SYSTEM_PROMPT = `You are an expert at analyzing YouTube creator personalities based on their video transcripts.
+
+Given transcript excerpts from a creator's videos, build a personality profile. Identify:
+- Their humor style (if any)
+- Recurring phrases or catchphrases they use frequently
+- Presentation style (conversational, educational, energetic, calm, etc.)
+- Favorite topics they keep coming back to
+- Unique verbal or stylistic traits
+
+Return valid JSON only, with this structure:
+{
+  "traits": [
+    {
+      "category": "Category name (e.g. 'Humor', 'Presentation', 'Expertise')",
+      "value": "Brief description (e.g. 'Dry, self-deprecating wit')",
+      "examples": ["Example quote or behavior"]
+    }
+  ],
+  "catchphrases": ["phrase 1", "phrase 2"],
+  "style": "2-3 sentence style summary",
+  "summary": "One paragraph personality overview"
+}
+
+Be specific and reference actual content from the transcripts. Find 4-6 traits.`;
+
+export async function OPTIONS() {
+  return optionsResponse(corsHeaders);
+}
+
+export const POST = withErrorHandling("discover-dna", async (request, { params }) => {
+  const session = await auth();
+  if (!session) {
+    return Response.json({ error: "Not authenticated" }, { status: 401, headers: corsHeaders });
+  }
+
+  const clientIp = getClientIp(request);
+  const rateLimitResult = checkRateLimit(`discover-dna:${clientIp}`, RATE_LIMITS.aiQuery);
+  if (!rateLimitResult.success) {
+    return rateLimitExceededResponse(rateLimitResult, "Too many requests", corsHeaders);
+  }
+
+  const { channelId } = await params;
+
+  const rows = await db
+    .select({
+      title: videos.title,
+      excerpt: transcripts.excerpt,
+      fullText: transcripts.fullText,
+    })
+    .from(transcripts)
+    .innerJoin(videos, eq(transcripts.videoId, videos.id))
+    .where(eq(videos.channelId, channelId));
+
+  const withText = rows.filter((r) => r.excerpt || r.fullText);
+
+  if (withText.length < 3) {
+    return Response.json(
+      { error: "Not enough transcripts to analyze (minimum 3). Sync transcripts first." },
+      { status: 400, headers: corsHeaders },
+    );
+  }
+
+  const sampled = withText.length > 30
+    ? withText.sort(() => Math.random() - 0.5).slice(0, 30)
+    : withText;
+
+  const excerpts = sampled.map((r) => {
+    const text = r.fullText
+      ? r.fullText.slice(0, 500)
+      : r.excerpt ?? "";
+    return `## "${r.title}"\n${text}`;
+  });
+
+  log.info({ channelId, transcriptCount: sampled.length }, "Generating DNA profile");
+
+  const { text } = await generateText({
+    model: getModel(),
+    system: SYSTEM_PROMPT,
+    messages: [{ role: "user", content: excerpts.join("\n\n---\n\n") }],
+    temperature: 0.4,
+    maxOutputTokens: 2000,
+  });
+
+  const jsonMatch = text.match(/\{[\s\S]*\}/);
+  if (!jsonMatch) {
+    log.error({ channelId }, "Failed to parse AI response for DNA");
+    return Response.json({ error: "Failed to generate profile" }, { status: 500, headers: corsHeaders });
+  }
+
+  const result = JSON.parse(jsonMatch[0]);
+
+  return Response.json(result, {
+    headers: mergeHeaders(corsHeaders, withRateLimitHeaders(rateLimitResult)),
+  });
+});

apps/youtube/src/app/api/discover/evolution/[channelId]/route.ts

@@ -0,0 +1,132 @@
+import { db } from "@/db";
+import { videos } from "@/db/schema";
+import { getModel } from "@/lib/ai-providers";
+import { auth } from "@/lib/auth";
+import { createTaggedLogger } from "@/lib/logger";
+import {
+  checkRateLimit,
+  getClientIp,
+  RATE_LIMITS,
+} from "@/lib/rate-limit";
+import { withErrorHandling } from "@/lib/route-handler";
+import {
+  corsHeaders,
+  mergeHeaders,
+  optionsResponse,
+  rateLimitExceededResponse,
+  withRateLimitHeaders,
+} from "@data-projects/shared";
+import { generateText } from "ai";
+import { eq } from "drizzle-orm";
+
+const log = createTaggedLogger("discover-evolution");
+
+const SYSTEM_PROMPT = `You are an expert YouTube analyst. Given a list of videos grouped by time period, analyze how the creator's content evolved.
+
+For each era, describe:
+- What topics dominated
+- How the style or format changed
+- Key growth or decline signals
+
+Return valid JSON only, with this structure:
+{
+  "eras": [
+    {
+      "period": "Era name (e.g. 'The Early Days')",
+      "startDate": "YYYY-MM",
+      "endDate": "YYYY-MM",
+      "topics": ["topic1", "topic2"],
+      "style": "Brief style description",
+      "description": "2-3 sentence narrative of this period",
+      "videoCount": 15
+    }
+  ],
+  "summary": "One paragraph overall evolution summary"
+}
+
+Keep era names creative and descriptive. Identify 3-6 distinct eras based on natural shifts you observe.`;
+
+function groupIntoEras(
+  vids: { title: string; publishedAt: Date; views: number; topics: string[] | null; duration: number }[],
+): string {
+  const sorted = [...vids].sort(
+    (a, b) => a.publishedAt.getTime() - b.publishedAt.getTime(),
+  );
+
+  const eraSize = Math.max(1, Math.ceil(sorted.length / 5));
+  const groups: string[] = [];
+
+  for (let i = 0; i < sorted.length; i += eraSize) {
+    const slice = sorted.slice(i, i + eraSize);
+    const start = slice[0].publishedAt.toISOString().slice(0, 7);
+    const end = slice.at(-1)!.publishedAt.toISOString().slice(0, 7);
+    const lines = slice.map(
+      (v) =>
+        `- "${v.title}" (${v.publishedAt.toISOString().slice(0, 10)}, ${v.views.toLocaleString()} views, ${Math.round(v.duration / 60)}min) [${(v.topics ?? []).join(", ")}]`,
+    );
+    groups.push(`### ${start} to ${end} (${slice.length} videos)\n${lines.join("\n")}`);
+  }
+
+  return groups.join("\n\n");
+}
+
+export async function OPTIONS() {
+  return optionsResponse(corsHeaders);
+}
+
+export const POST = withErrorHandling("discover-evolution", async (request, { params }) => {
+  const session = await auth();
+  if (!session) {
+    return Response.json({ error: "Not authenticated" }, { status: 401, headers: corsHeaders });
+  }
+
+  const clientIp = getClientIp(request);
+  const rateLimitResult = checkRateLimit(`discover-evolution:${clientIp}`, RATE_LIMITS.aiQuery);
+  if (!rateLimitResult.success) {
+    return rateLimitExceededResponse(rateLimitResult, "Too many requests", corsHeaders);
+  }
+
+  const { channelId } = await params;
+
+  const channelVideos = await db
+    .select({
+      title: videos.title,
+      publishedAt: videos.publishedAt,
+      views: videos.views,
+      topics: videos.topics,
+      duration: videos.duration,
+    })
+    .from(videos)
+    .where(eq(videos.channelId, channelId));
+
+  if (channelVideos.length < 5) {
+    return Response.json(
+      { error: "Not enough videos to analyze evolution (minimum 5)" },
+      { status: 400, headers: corsHeaders },
+    );
+  }
+
+  const grouped = groupIntoEras(channelVideos);
+
+  log.info({ channelId, videoCount: channelVideos.length }, "Generating evolution analysis");
+
+  const { text } = await generateText({
+    model: getModel(),
+    system: SYSTEM_PROMPT,
+    messages: [{ role: "user", content: grouped }],
+    temperature: 0.3,
+    maxOutputTokens: 2000,
+  });
+
+  const jsonMatch = text.match(/\{[\s\S]*\}/);
+  if (!jsonMatch) {
+    log.error({ channelId }, "Failed to parse AI response for evolution");
+    return Response.json({ error: "Failed to generate analysis" }, { status: 500, headers: corsHeaders });
+  }
+
+  const result = JSON.parse(jsonMatch[0]);
+
+  return Response.json(result, {
+    headers: mergeHeaders(corsHeaders, withRateLimitHeaders(rateLimitResult)),
+  });
+});