server.ts

/**
 * Custom dev server — wraps Next.js and handles /api/voice WebSocket upgrades
 * directly on the HTTP server, which Next.js API routes can't do.
 *
 * Usage: tsx watch server.ts (via `pnpm dev`)
 */

import { writeFile } from "node:fs/promises";
import { createServer } from "node:http";
import { parse } from "node:url";
import next from "next";
import { WebSocketServer } from "ws";
import { createFrameAnalyzer } from "./src/lib/gemini-session";

const dev = process.env.NODE_ENV !== "production";
const hostname = "localhost";
const port = Number(process.env.PORT) || 3000;

const app = next({ dev, hostname, port });
const handle = app.getRequestHandler();

app.prepare().then(() => {
	const server = createServer(async (req, res) => {
		await handle(req, res, parse(req.url ?? "/", true));
	});

	const wss = new WebSocketServer({ noServer: true });

	server.on("upgrade", (req, socket, head) => {
		const { pathname } = parse(req.url ?? "/", true);
		if (pathname === "/api/voice") {
			wss.handleUpgrade(req, socket, head, (ws) => {
				handleVoiceConnection(ws);
			});
		}
		// Other upgrade requests (e.g. Next.js HMR) are left alone
	});

	server.listen(port, () => {
		console.log(`> Ready on http://${hostname}:${port}`);
	});
});

function handleVoiceConnection(ws: import("ws").WebSocket) {
	console.log("[voice-api] new WebSocket connection");

	const apiKey = process.env.GEMINI_API_KEY;
	if (!apiKey) {
		console.error("[voice-api] GEMINI_API_KEY missing from env");
		ws.close(1011, "GEMINI_API_KEY not configured");
		return;
	}

	const analyzer = createFrameAnalyzer(apiKey);
	let frameCount = 0;
	let analyzing = false;

	ws.on("message", async (data: Buffer | ArrayBuffer) => {
		// Skip if still processing the previous frame
		if (analyzing) {
			return;
		}

		const buf = Buffer.isBuffer(data) ? data : Buffer.from(data);
		const arrayBuffer = buf.buffer.slice(
			buf.byteOffset,
			buf.byteOffset + buf.byteLength
		) as ArrayBuffer;

		frameCount++;
		if (frameCount === 1) {
			writeFile("debug-frame.jpg", buf).then(() =>
				console.log(
					`[voice-api] wrote debug-frame.jpg (${buf.byteLength} bytes)`
				)
			);
		}
		if (frameCount <= 3 || frameCount % 10 === 0) {
			console.log(
				`[voice-api] analyzing frame #${frameCount} (${buf.byteLength} bytes)`
			);
		}

		analyzing = true;
		try {
			const event = await analyzer.analyzeFrame(arrayBuffer);
			if (ws.readyState === ws.OPEN) {
				if (event) {
					console.log("[dialogue]", JSON.stringify(event));
					ws.send(JSON.stringify(event));
				} else {
					ws.send(JSON.stringify({ type: "no_dialog" }));
				}
			}
		} catch (err) {
			console.error("[voice-api] Gemini error:", err);
		} finally {
			analyzing = false;
		}
	});

	ws.on("close", () => {
		console.log(
			`[voice-api] client disconnected after ${frameCount} frames`
		);
	});

	ws.on("error", (err) => {
		console.error("[voice-api] WebSocket error:", err);
	});
}