🧠 RAGify - Lightweight RAG Service with FastAPI, FAISS, and OpenAI

RAGify is a minimal and powerful Retrieval-Augmented Generation (RAG) service built with FastAPI, FAISS, and OpenAI API. It allows you to upload PDF or CSV documents, store them as vector embeddings locally using FAISS, and chat with them using OpenAI's GPT models.

---

🚀 Features

• Upload and parse PDF and CSV files.
• Embeds documents using local SentenceTransformer (MiniLM-L6-v2) or OpenAI Embeddings.
• Stores vectors efficiently using FAISS local index.
• Chat with your documents using OpenAI GPT models (like GPT-3.5-Turbo).
• Real-time async chat endpoint streams tokens as they are generated.
• Lightweight and privacy-friendly (documents never leave your machine).
• Duplicate file uploads are automatically detected using SHA256 hashing.
• Simple to run locally or in Docker.

---

🛠️ Installation

🔧 Running Locally (Recommended for development)

1. Clone the repository:

git clone https://github.com/HagaiHen/rag-service.git
cd rag-service

2. Create .env file in the root:

OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxx
MONGO_URI=mongodb+srv://<user>:<password>@cluster.mongodb.net
MONGO_DB=rag_service

3. Install dependencies:

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

4. Run the FastAPI server:

uvicorn app.main:app --reload

5. Visit:

• Swagger Docs: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

🐳 Running with Docker

1. Build the image:

docker build -t rag-service .

2. Run the container:

docker run --env-file .env -p 8000:8000 ragify

💡 API Usage Example

1. Upload a File
   • POST /upload
   • Body: form-data
   • file: (upload your .pdf or .csv)
   

2. Chat with your data
   • POST /chat
   • Query Params:
   • user_input: "What is this document about?"
   • session_id: "my-session-id"
   • user_id: "my-user"
   
   • POST /async_chat
   • Same query params as /chat, but streams the response token-by-token

3. Retrieve chat history
   • GET /history
   • Query Params:
   • user_id: "my-user"
   • session_id: "my-session-id" (optional)
   Omit the session_id to retrieve history for all sessions of the user.

Chat history is cached in memory and also stored in MongoDB per user and session.
If the service restarts or the cache is empty, history is loaded from the database. Cached sessions are automatically removed after 15 minutes of inactivity.

---

📜 License

This project is licensed under the MIT License. Feel free to use, modify, and distribute it.

---

🤝 Contributing

Contributions are welcome! 1. Fork the repo 2. Create your feature branch (git checkout -b feature/awesome-feature) 3. Commit your changes (git commit -m 'Add amazing feature') 4. Push to the branch (git push origin feature/awesome-feature) 5. Open a Pull Request