Backend Monorepo

A Lerna and NPM Workspaces powered monorepo for managing multiple Express.js/TypeScript backend microservices. Currently contains the ai-chat application, which utilizes MongoDB, Clerk, and ImageKit.

Live Demo (AI Thread): https://i-faaza.com/app/ai-thread

🚀 Prerequisites

Make sure you have the following installed on your machine:

• Node.js (v18 or higher recommended)
• pnpm (v9 or higher)
• Redis (v7 or higher) - Required for Quota Management/Rate Limiting.
• A running MongoDB instance or a MongoDB Atlas connection string.
• A Clerk account for authentication.
• An ImageKit account for media handling.

🛠️ Installation

1. Clone the repository (if you haven't already):

   # navigate to your directory
   cd be-multiapp

2. Install dependencies: Run this from the root directory. PNPM Workspaces will automatically link packages:

   pnpm install

3. Set up Environment Variables: Depending on how you plan to run the application, configure your environment variables:

   Option A: Using Docker Compose (Recommended) Create a docker-compose.override.yml file in the root directory for your local development secrets. This file is ignored by Git.

   version: '3.8'
   services:
     api:
       environment:
         - IMAGE_KIT_ENDPOINT=your_imagekit_endpoint
         - IMAGE_KIT_PUBLIC_KEY=your_imagekit_public_key
         - IMAGE_KIT_PRIVATE_KEY=your_imagekit_private_key
         - CLIENT_URL=http://localhost:5173
         - MONGO=your_mongodb_connection_string
         - CLERK_PUBLISHABLE_KEY=your_clerk_publishable_key
         - CLERK_SECRET_KEY=your_clerk_secret_key
         - REDIS_PRIVATE_URL=redis://localhost:6379
         - GEMINI_API_KEY=your_gemini_api_key

   Option B: Running locally with npm Create a .env file in the root directory. You can use the existing env file as a template (cp env .env), then populate it:

   # ImageKit credentials
   IMAGE_KIT_ENDPOINT=your_imagekit_endpoint
   IMAGE_KIT_PUBLIC_KEY=your_imagekit_public_key
   IMAGE_KIT_PRIVATE_KEY=your_imagekit_private_key
   
   # Frontend Client URL (for CORS)
   CLIENT_URL=http://localhost:5173
   
   # MongoDB Connection String
   MONGO=your_mongodb_connection_string
   
   # Clerk Authentication credentials
   CLERK_PUBLISHABLE_KEY=your_clerk_publishable_key
   CLERK_SECRET_KEY=your_clerk_secret_key
   
   # Redis Connection
   REDIS_PRIVATE_URL=redis://127.0.0.1:6379
   
   # Gemini API Key
   GEMINI_API_KEY=your_gemini_api_key

💻 Running the Application

Development Mode

To run ai-chat in development mode:

pnpm --filter ai-chat dev

(Or cd apps/ai-chat && pnpm dev)

Note: Pastikan Redis Server sudah berjalan di lokal (atau melalui Docker terpisah) sebelum menjalankan aplikasi.

The server should now be running.

Production Mode

To build and run the application for production locally:

1. Build all apps using Lerna:

   npx lerna run build

2. Start the compiled application:

   npm start --workspace=apps/ai-chat

Using Docker

You can also run the application using Docker. This ensures a consistent environment and simplifies deployment.

We use Docker Compose overrides to manage environments without relying on a .env file.

Local Development:

1. Edit the docker-compose.override.yml file to include your local environment variables.
2. Run Docker Compose. It will automatically merge the override file with the base docker-compose.yml:

   docker-compose up -d

   To rebuild the image if you make changes, run docker-compose up --build -d.

Production Deployment:

1. Edit the docker-compose.prod.yml file with your actual production credentials.
2. Tell Docker Compose to use both the base file and the production file explicitly:

   docker-compose -f docker-compose.yml -f docker-compose.prod.yml up -d

☁️ Railway Deployment (Recommended)

1. Add a Redis service in your Railway project.
2. In your API service settings, add a variable: REDIS_PRIVATE_URL with value ${{Redis.REDIS_PRIVATE_URL}}.
3. Railway will handle the connection automatically.

To stop the container in either environment:

docker-compose down

📚 Tech Stack

• Framework: Express.js
• Language: TypeScript
• Database: MongoDB with Mongoose
• Caching/Rate Limiting: Redis via ioredis
• Authentication: Clerk (@clerk/express)
• Other: cors, dotenv, pnpm