RAG Chatbot with User Management

A Retrieval-Augmented Generation (RAG) chatbot system with user authentication, PDF management, and vector database integration using FastAPI, LangChain, and ChromaDB.

Features

• User Management: Admin and regular user authentication
• PDF Management: Upload, ingest, and delete PDFs per user
• Vector Database: ChromaDB integration with metadata preservation
• Chat Interface: Interactive chat client with authentication
• Azure Deployment: Docker-based deployment on Azure VM

Quick Start

Prerequisites

• Python 3.8+
• OpenAI API key
• SQLite (included)
• ChromaDB (included)

Local Setup

1. Clone and install

   git clone https://github.com/kittysoftpaw0510/RAG_chatbot
   cd RAG_chatbot
   pip install -r requirements.txt

2. Environment setup

   mkdir data
   # Copy your PDF files into 'data' directory

3. Create .env file

   OPENAI_API_KEY=sk-proj-xxx
   ADMIN_USERNAME=admin
   ADMIN_PASSWORD=your_admin_password
   

4. Start the server

   uvicorn main:app --host 0.0.0.0 --port 8000

5. Use the chat client

   python Client/chat_client.py

User Management

Authentication

• Admin Users: Authenticated using environment variables
• Regular Users: Stored in SQLite database
• Session Management: Credentials required for each operation

User Operations

• Admin Functions:

  • View all users
  • Reset user passwords
  • Manage all PDFs
  • Clear vector database

• Regular User Functions:

  • Upload PDFs
  • Ingest PDFs manually
  • Chat with their documents
  • Delete their own PDFs

PDF Management

Upload and Ingestion

1. Upload PDFs: Files are stored in user-specific directories
2. Manual Ingestion: Process PDFs to create vector embeddings
3. Metadata Preservation: User information stored with embeddings

File Operations

• Upload PDFs to user directory
• Ingest PDFs to vector database
• Delete PDFs (files + database records + embeddings)
• Clear all vector embeddings

API Endpoints

Authentication

• POST /login - User authentication
• POST /admin/login - Admin authentication

User Management

• GET /users - List all users (admin only)
• POST /users - Create new user
• PUT /users/{user_id}/password - Reset user password (admin only)

PDF Management

• POST /upload/{user_id} - Upload PDF for user
• POST /ingest/{user_id} - Ingest PDFs for user
• DELETE /pdfs/{user_id} - Delete all PDFs for user
• DELETE /pdfs/{user_id}/{filename} - Delete specific PDF

Chat

• POST /chat - Chat with RAG system

Azure Deployment

Prerequisites

• Azure CLI
• Docker
• SSH access

Deployment Steps

1. Azure CLI setup

   az login
   az group create --name rag-chatbot-rg --location eastus

2. Create VM

   az vm create --resource-group rag-chatbot-rg --name rag-vm \
     --image UbuntuLTS --admin-username azureuser \
     --generate-ssh-keys --size Standard_DS2_v2

3. Configure VM

   az vm open-port --port 8000 --resource-group rag-chatbot-rg --name rag-vm

4. Deploy with Docker

   docker build -t ragbot .
   docker run -d -p 8000:8000 --name rag_container ragbot

Testing Azure Deployment

python Client/test_multi_chat.py

Project Structure

RAG_chatbot/
├── Client/
│   ├── admin_tools.py
│   ├── chat_client.py
│   ├── test_multi_chat.py
│   ├── user_tools.py
├── data/
│   ├── public/
│   │   └── ... (public PDFs)
│   ├── user_1/
│   │   └── ... (user_1's PDFs)
│   └── ... (other user folders)
├── Dockerfile
├── main.py
├── README.md
├── requirements.txt
├── routes/
│   ├── admin/
│   │   ├── admin_auth.py
│   │   ├── chat_manage.py
│   │   ├── data_manage.py
│   │   ├── user_manage.py
│   │   └── vectordb_manage.py
│   └── user/
│       ├── chat_manage.py
│       ├── data_manage.py
│       ├── user_auth.py
│       ├── user_manage.py
│       └── vectordb_manage.py
└── utils/
    ├── ingest.py
    ├── llm.py
    ├── vectordb.py
    └── sqlitedb.py

Configuration

Environment Variables

• OPENAI_API_KEY: Your OpenAI API key
• ADMIN_USERNAME: Admin username (default: admin)
• ADMIN_PASSWORD: Admin password

Database

• SQLite: User management and PDF metadata
• ChromaDB: Vector embeddings storage

Troubleshooting

Common Issues

1. Authentication Errors

   • Verify .env file exists with correct credentials
   • Check user exists in SQLite database

2. PDF Upload Issues

   • Ensure data directory exists
   • Check file permissions

3. Vector Database Issues

   • Clear chroma directory if corrupted
   • Re-ingest PDFs after clearing

4. Azure Deployment

   • Verify port 8000 is open
   • Check Docker container is running
   • Ensure environment variables are set

Useful Commands

# Clear vector database
python Client/memory_management.py --clear-vectors

# List all users
python Client/user_management.py --list-users

# Delete user PDFs
python Client/data_management.py --delete-user-pdfs USER_ID

# Admin operations
python Client/admin_tools.py --reset-password USER_ID

# User operations
python Client/user_tools.py --create-user USERNAME PASSWORD

Development

Adding New Features

1. Update API endpoints in main.py
2. Add corresponding client functions
3. Update database schema if needed
4. Test with multiple users

Testing

• Use test_multi_chat.py for Azure deployment testing
• Test user isolation and permissions
• Verify PDF operations work correctly

---

For detailed deployment instructions and troubleshooting, see the original documentation sections below.