Leafra - AI-Powered PDF RAG System

Leafra is a Next.js application that provides a Retrieval-Augmented Generation (RAG) system for PDF documents. Users can upload PDFs, ask questions about them, and receive AI-powered responses based on the document content.

Features

• 📄 PDF Upload and Processing
• 🤖 AI-Powered Chat Interface
• 🔍 Vector Search using Pinecone
• 🔐 Authentication (Email/Password + OAuth)
• 📊 Real-time Chat Streaming
• 🗄️ PostgreSQL Database
• ⚡ Background Job Processing with BullMQ

Architecture

Tech Stack

• Framework: Next.js 15 (App Router)
• Database: PostgreSQL with Drizzle ORM
• Vector Database: Pinecone
• AI/ML: TogetherAI (Embeddings & Chat)
• Queue: BullMQ with Redis (Upstash)
• File Upload: UploadThing
• Authentication: Better-auth
• UI: React, Tailwind CSS, shadcn/ui

Application Flow

1. Authentication Flow: User signs up/in → Better-auth → Session management
2. Chat Creation: User creates chat → Stored in DB → Chat page loads
3. PDF Upload: UploadThing → BullMQ queue → Worker processes → Pinecone embeddings
4. Chat Flow: User message → API route → Save to DB → Query Pinecone → Stream AI response → Save response

Getting Started

Prerequisites

• Node.js 18+ or Bun
• PostgreSQL database
• Pinecone account
• TogetherAI account
• Upstash Redis account (optional but recommended)
• UploadThing account

Installation

1. Clone the repository:

git clone <repository-url>
cd leafra

2. Install dependencies:

npm install
# or
bun install

3. Set up environment variables: Create a .env.local file in the root directory with the following variables:

# Database
DATABASE_URL=postgresql://username:password@host:port/database

# Pinecone
PINECONE_API_KEY=your-pinecone-api-key

# TogetherAI
TOGETHER_AI_API_KEY=your-together-ai-api-key
TOGETHER_AI_MODEL=your-model-name

# Redis (Upstash) - Optional
UPSTASH_REDIS_REST_URL=https://your-redis-instance.upstash.io
UPSTASH_REDIS_REST_TOKEN=your-redis-token

# OAuth Providers (Optional)
GOOGLE_CLIENT_ID=your-google-client-id
GOOGLE_CLIENT_SECRET=your-google-client-secret
GITHUB_CLIENT_ID=your-github-client-id
GITHUB_CLIENT_SECRET=your-github-client-secret
DISCORD_CLIENT_ID=your-discord-client-id
DISCORD_CLIENT_SECRET=your-discord-client-secret

# Next.js
NEXT_PUBLIC_BASE_URL=http://localhost:3000
NODE_ENV=development

4. Set up the database:

# Run migrations
npm run db:push
# or with drizzle-kit
npx drizzle-kit push

5. Start the development server:

npm run dev
# or
bun dev

6. Start the worker (in a separate terminal):

# Compile worker
tsc --watch ./lib/worker.ts

# Run worker (in another terminal)
bunx nodemon ./lib/worker.js
# or
bun run ./lib/worker.js

Open http://localhost:3000 in your browser.

Project Structure

leafra/
├── app/                    # Next.js App Router
│   ├── api/               # API routes
│   │   ├── auth/          # Authentication endpoints
│   │   ├── chat/          # Chat API
│   │   ├── messages/      # Messages API
│   │   └── uploadthing/   # File upload
│   ├── actions/           # Server actions
│   ├── chat/              # Chat pages
│   ├── dashboard/         # Dashboard page
│   └── layout.tsx         # Root layout
├── components/             # React components
│   ├── custom/            # Custom components
│   ├── ui/                # shadcn/ui components
│   └── shared/            # Shared components
├── lib/                   # Utility libraries
│   ├── auth.ts            # Authentication config
│   ├── db.ts              # Database connection
│   ├── db/
│   │   └── schema.ts      # Database schema
│   ├── env.ts             # Environment validation
│   ├── logger.ts          # Logging utility
│   ├── api-response.ts    # API response helpers
│   ├── integrations/     # Third-party integrations
│   │   ├── pinecone.ts   # Pinecone client
│   │   └── redis.ts      # Redis client
│   └── worker.ts          # Background worker
├── types/                 # TypeScript types
└── middleware.ts          # Next.js middleware

API Documentation

Authentication

All API routes (except /api/auth/*) require authentication via session cookie.

Endpoints

POST /api/chat

Send a message to the chat and get an AI response.

Request Body:

{
  "messages": [
    {
      "role": "user",
      "content": "What is this document about?"
    }
  ],
  "chatId": "uuid-string"
}

Response: Streaming text response

GET /api/messages?chatId=uuid

Get all messages for a chat.

Query Parameters:

• chatId (required): UUID of the chat

Response:

[
  {
    "id": 1,
    "chatId": "uuid",
    "content": "Message content",
    "role": "user",
    "createdAt": "2024-01-01T00:00:00Z"
  }
]

Development

Available Scripts

• npm run dev - Start development server
• npm run build - Build for production
• npm run start - Start production server
• npm run lint - Run ESLint
• npm run typecheck - Run TypeScript type checking
• npm run dev:worker - Watch and run worker

Code Quality

• TypeScript for type safety
• ESLint for code linting
• Zod for runtime validation
• Structured logging
• Standardized error responses

Deployment

Environment Variables

Ensure all required environment variables are set in your production environment. See the .env.example file for reference.

Database Migrations

Run database migrations before deploying:

npx drizzle-kit push

Worker Process

The worker process must be running separately in production. Consider using:

• PM2
• Docker containers
• Cloud functions
• Kubernetes jobs

Recommended Platforms

• Vercel - For Next.js hosting
• Railway - For PostgreSQL and worker
• Upstash - For Redis
• Pinecone - For vector database

Security

• All API routes are authenticated
• Environment variables are validated at startup
• Input validation using Zod
• SQL injection protection via Drizzle ORM
• Session-based authentication

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run tests and linting
5. Submit a pull request

License

[Add your license here]

Support

For issues and questions, please open an issue on GitHub.