RAG Chat App

A Retrieval-Augmented Generation (RAG) chatbot that allows you to ask questions about your documents and get intelligent answers with source citations.

Demo link

https://demo-rag-chat.streamlit.app/

Features

• Document Processing: Automatically loads and processes text documents from the data folder
• Smart Search: Uses vector embeddings to find relevant document sections
• AI-Powered Answers: Leverages OpenAI's GPT models to generate contextual responses
• Source Citations: Shows which documents were used to generate each answer
• Interactive UI: Web-based chat interface built with Streamlit
• Persistent Storage: Saves processed documents in a vector database for fast retrieval

Project Structure

rag_chat_app/
├── src/                         # Source code modules
│   ├── interfaces.py           # Component interfaces
│   ├── config.py               # Configuration management
│   ├── logger.py               # Logging setup
│   ├── document_loader.py      # Text file loading
│   ├── document_processor.py   # Document chunking
│   ├── vector_store.py         # Vector database operations
│   ├── rag_pipeline.py         # Question-answering pipeline
│   └── factory.py              # Component initialization
├── data/                       # Your documents go here
├── tests.py                    # Test suite
├── main.py                     # Streamlit web app
├── requirements.txt            # Python dependencies
└── README.md                   # This file

Quick Start

1. Install dependencies:

   pip install -r requirements.txt

2. Set OpenAI API key:

   # Create .env file with your API key
   echo "OPENAI_API_KEY=your_key_here" > .env

3. Run tests to verify setup:

   python tests.py

4. Run the application:

   streamlit run main.py

How It Works

1. Document Loading: The app scans the data/ folder for text files and loads them
2. Text Processing: Documents are split into smaller chunks for better search accuracy
3. Vector Embeddings: Each chunk is converted to vector embeddings using OpenAI's API
4. Storage: Embeddings are stored in a ChromaDB vector database
5. Query Processing: When you ask a question, it finds the most relevant document chunks
6. Answer Generation: The relevant context is sent to GPT to generate a comprehensive answer

Configuration

Set up your environment with these variables:

• OPENAI_API_KEY - Required: Your OpenAI API key
• OPENAI_MODEL - Optional: GPT model to use (default: gpt-3.5-turbo)
• CHUNK_SIZE - Optional: Size of document chunks (default: 1000)
• CHUNK_OVERLAP - Optional: Overlap between chunks (default: 200)
• SEARCH_K - Optional: Number of relevant chunks to retrieve (default: 3)

Testing

Verify everything works correctly:

python tests.py

Adding Your Documents

Simply place your text files in the data/ folder. Supported formats:

• .txt files
• The app will automatically process them when you run it

Deployment to Streamlit Cloud

Step 1: Push to GitHub

1. Remove your .env file from the repo (it's already in .gitignore)
2. Push your code to a public GitHub repository

Step 2: Deploy on Streamlit Cloud

1. Go to share.streamlit.io
2. Connect your GitHub repository
3. Set main file path: main.py

Step 3: Add Secrets

In your Streamlit Cloud app:

1. Go to Settings → Secrets
2. Add your configuration:

OPENAI_API_KEY = "sk-your-actual-api-key-here"

# Optional: Override settings for production
OPENAI_MODEL = "gpt-4"
CHUNK_SIZE = "300"
SEARCH_K = "1"

Step 4: Deploy

Your app will automatically deploy and be publicly accessible while keeping your API key secure!

Troubleshooting Deployment

ChromaDB SQLite3 Issue

If you see an SQLite3 version error during deployment, the app includes an automatic fix:

• pysqlite3-binary dependency provides a newer SQLite3 version
• Automatic SQLite3 patching for Streamlit Cloud compatibility
• ChromaDB will work seamlessly without any additional configuration needed

The deployment should work without any manual intervention!