import express, { type Express } from "express";
import type { Request, Response } from "express";
import { createServer } from "http";
import path from "path";
import { fileURLToPath } from "url";
import { getConfig } from "./config.js";
import { registry } from "./registry.js";
import { adminRoutes, adminWebSocket } from "./admin/index.js";
import { createEndpointHandler } from "./middleware/index.js";
import { ALL_PROVIDERS } from "./providers/index.js";
const __dirname = path.dirname(fileURLToPath(import.meta.url));
const app: Express = express();
app.use(express.json());
const envConfig = getConfig();
// Initialize registry from environment config
registry.initFromEnv(envConfig);
// Register all providers
for (const provider of ALL_PROVIDERS) {
registry.registerProvider(provider);
}
// =============================================================================
// Middleware
// =============================================================================
// Logging middleware (disabled by default since dashboard shows all requests)
const VERBOSE_LOGGING = process.env.VERBOSE === "true";
if (VERBOSE_LOGGING) {
app.use((req, res, next) => {
const start = Date.now();
res.on("finish", () => {
const duration = Date.now() - start;
if (
!req.path.startsWith("/api") &&
!req.path.startsWith("/dashboard") &&
req.path !== "/ws"
) {
console.log(
`${new Date().toISOString()} ${req.method} ${req.path} ${res.statusCode} ${duration}ms`,
);
}
});
next();
});
}
// =============================================================================
// Dashboard
// =============================================================================
// Simple rate limiter for dashboard routes (prevents abuse of file system access)
const dashboardRateLimiter = (() => {
const requests = new Map<string, number[]>();
const windowMs = 60000; // 1 minute
const maxRequests = 100; // 100 requests per minute per IP
return (req: Request, res: Response, next: () => void) => {
const ip = req.ip || req.socket.remoteAddress || "unknown";
const now = Date.now();
const timestamps = requests.get(ip) || [];
// Remove old timestamps outside the window
const validTimestamps = timestamps.filter((t) => now - t < windowMs);
if (validTimestamps.length >= maxRequests) {
res.status(429).json({ error: "Too many requests" });
return;
}
validTimestamps.push(now);
requests.set(ip, validTimestamps);
next();
};
})();
// Serve static dashboard files in production
const dashboardPath = path.join(__dirname, "../dashboard/dist");
app.use("/dashboard", dashboardRateLimiter, express.static(dashboardPath));
// Serve index.html for all dashboard routes (SPA support)
app.get(
"/dashboard/*",
dashboardRateLimiter,
(_req: Request, res: Response) => {
res.sendFile(path.join(dashboardPath, "index.html"));
},
);
// Redirect root /dashboard to /dashboard/
app.get("/dashboard", (_req: Request, res: Response) => {
res.redirect("/dashboard/");
});
// =============================================================================
// Health & Models
// =============================================================================
app.get("/health", (_req: Request, res: Response) => {
res.json({ status: "ok", timestamp: new Date().toISOString() });
});
app.get("/v1/models", (_req: Request, res: Response) => {
res.json({
object: "list",
data: [
{
id: "gpt-4o",
object: "model",
created: 1715367049,
owned_by: "mock-server",
},
{
id: "gpt-4o-mini",
object: "model",
created: 1715367049,
owned_by: "mock-server",
},
{
id: "gpt-4-turbo",
object: "model",
created: 1715367049,
owned_by: "mock-server",
},
{
id: "gpt-3.5-turbo",
object: "model",
created: 1715367049,
owned_by: "mock-server",
},
],
});
});
// =============================================================================
// Admin Routes
// =============================================================================
app.use("/api", adminRoutes);
// =============================================================================
// LLM Provider Endpoints
// =============================================================================
// Register all provider routes automatically
for (const provider of ALL_PROVIDERS) {
const handler = createEndpointHandler(provider);
if (provider.method === "POST") {
app.post(provider.routePath, handler);
} else if (provider.method === "GET") {
app.get(provider.routePath, handler);
}
}
// =============================================================================
// Server Lifecycle
// =============================================================================
let httpServer: ReturnType<typeof createServer> | null = null;
export function startServer(): Promise<void> {
return new Promise((resolve) => {
httpServer = createServer(app);
// Attach WebSocket server
adminWebSocket.attach(httpServer);
httpServer.listen(envConfig.port, () => {
console.log(`
🚀 Mock LLM Server running at http://localhost:${envConfig.port}
Dashboard: http://localhost:${envConfig.port}/dashboard
Endpoints:
POST /v1/chat/completions (OpenAI)
POST /v1/responses (OpenAI Responses API)
POST /v1/messages (Anthropic)
POST /v1beta/models/* (Gemini v1beta)
POST /v1/models/* (Gemini v1)
Set VERBOSE=true for request logging
`);
resolve();
});
});
}
export function stopServer(): Promise<void> {
return new Promise((resolve) => {
adminWebSocket.close();
if (httpServer) {
httpServer.close(() => {
httpServer = null;
resolve();
});
} else {
resolve();
}
});
}
// Legacy exports for test compatibility
export function resetRateLimitState(): void {
registry.resetAllRateLimiters();
}
export { app };
// Start server if run directly
const isMain =
process.argv[1]?.endsWith("server.ts") ||
process.argv[1]?.endsWith("server.js");
if (isMain) {
startServer();
}
