RAG Demo - Document Q&A System

A Retrieval-Augmented Generation (RAG) system built with LangChain, ChromaDB, and Anthropic Claude LLM API. Features intelligent document indexing, incremental updates, and conversational AI.

Architecture Overview

🎯 Features

Core Capabilities

• 📚 Smart Document Indexing - PDF, TXT, and Markdown support with incremental updates
• 🔍 Semantic Search - Vector-based similarity search with ChromaDB
• 🤖 AI-Powered Answers - Contextual responses using Anthropic Claude
• ⚡ Incremental Updates - Only re-index modified documents for efficiency
• 🗃️ File Tracking - Built-in metadata tracking and change detection

Other Features

• 🎛️ Centralized Configuration - Easy customization via config files
• 🛡️ Error Handling - Error handling and logging
• 🏗️ Modular Architecture - Clean separation for maintainability

📊 Workflow Diagrams

Ingestion Workflow

graph LR
    A[📁 Source Documents] --> B[📊 File Tracker]
    B --> C{🔄 File Changed?}
    C -->|No| D[✅ Skip Processing]
    C -->|Yes| E{📄 File Type Check}
    
    E -->|PDF| F[📄 PDF Loader]
    E -->|TXT| G[📝 Text Loader]
    E -->|MD| H[📓 Markdown Loader]
    
    F --> I[🔍 Text Extraction]
    G --> I
    H --> I
    
    I --> J[📏 Text Chunking]
    J --> K[🧠 Embedding Generation]
    K --> L[💾 ChromaDB Storage]
    
    style A fill:#e1f5fe
    style L fill:#c8e6c9
    style B fill:#fff3e0
    style D fill:#f3e5f5

Loading

RAG Query Workflow

graph LR
    A[❓ User Query] --> B[🧠 Query Embedding]
    B --> C[🔍 Vector Search]
    C --> D[💾 ChromaDB Retrieval]
    
    D --> E[📚 Relevant Documents]
    E --> F[🔄 Context Assembly]
    
    F --> G[📝 Prompt Construction]
    
    G --> H[🤖 Anthropic Claude API]
    H --> I[💬 AI Response]
    I --> J[✨ Final Answer]
    
    style A fill:#e3f2fd
    style E fill:#f1f8e9
    style H fill:#fce4ec
    style J fill:#e8f5e8

Loading

🛠️ Tech Stack

• Python - Core programming language for the entire application
• LangChain - Framework for building LLM-powered applications and RAG pipelines
• ChromaDB - Vector database for storing and retrieving document embeddings. ChromaDB for its simplicity and ease of use in local deployments (Other alternatives you could consider - Elasticsearch, MongoDB, Pinecone)
• Anthropic Claude API - Large language model for generating contextual responses (Other alternatives you could consider - Ollama (for running locally), OpenAI etc.)
• PyPDF2 - PDF document processing and text extraction
• Sentence Transformers - Text embedding generation for semantic search

🏗️ Project Structure

rag-demo/
├── README.md                # This file
├── requirements.txt         # Python dependencies
├── .env                     # Environment variables (API keys)
├── setup.sh                 # Virtual environment setup script
├── common/                  # Shared utilities and configuration
│   ├── __init__.py
│   ├── config.py            # Centralized configuration
│   ├── utils.py             # Utility functions
│   └── vector_store.py      # ChromaDB wrapper
├── indexer/                 # Document indexing module
│   ├── __init__.py
│   ├── main.py              # Indexing CLI and logic
│   ├── document_loader.py   # PDF/text document loaders
│   ├── text_processor.py    # Text chunking and processing
│   └── file_tracker.py      # File change tracking
├── search/                  # Search and query module
│   ├── __init__.py
│   ├── main.py              # Search CLI and interface
│   └── rag_chain.py         # Claude AI integration
├── source_documents/        # Source documents directory
│   ├── *.pdf                # PDF documents
│   └── *.txt                # Text files
└── chroma_db/               # Vector database (auto-created)
    └── ...                  # ChromaDB persistence files

🚀 Quick Start

Prerequisites

• Python 3.9+
• Anthropic API key (Get one here) - Note: This is a paid API, monitor your usage

Installation

1. Clone and setup:

   git clone <repository-url>
   cd rag-langchain-demo

2. Start up script:

   chmod +x setup.sh && ./setup.sh

3. Configure API key:

   echo "ANTHROPIC_API_KEY=your_api_key_here" > .env

4. Index your documents:

   python -m indexer.main directory source_documents/

   Example Indexer Commands:

   # Index all your documents
   python -m indexer.main directory source_documents/
   
   # Index a single file
   python -m indexer.main file source_documents/sample_document.txt
   
   # Check index status
   python -m indexer.main status
   
   # Clear the index
   python -m indexer.main clear
   
   # Force re-index all documents
   python -m indexer.main force source_documents/

5. Start querying:

   python -m search.main ask "What is machine learning?"

   Example Search Commands:

   python -m search.main ask "What is machine learning?"
   python -m search.main ask "What is price of CHLORINATOR"

Database Inspection

SQLite Commands:

sqlite3 chroma_db/chroma.sqlite3

# Inside SQLite:
.tables              # Show all tables
.schema              # Show table schemas
SELECT name FROM sqlite_master WHERE type='table' order by name;  # List tables
.quit                # Exit

🐛 Troubleshooting

Common Issues

1. API Key Errors

# Error: Missing ANTHROPIC_API_KEY environment variable
echo "ANTHROPIC_API_KEY=your_key" > .env

2. Import Errors

# Ensure virtual environment is activated
source rag-env/bin/activate
pip install -r requirements.txt

3. ChromaDB Issues

# Clear database if corrupted
rm -rf chroma_db/
python -m indexer.main directory source_documents/

4. No Search Results

# Check if documents are indexed
python -m indexer.main status

Debug Mode

Enable detailed logging by modifying common/config.py:

APP_CONFIG = {
    "debug_mode": True,
    "log_level": "DEBUG"
}