DocRag – Technical Documentation Intelligence Platform

A production-ready RAG (Retrieval-Augmented Generation) platform for technical documentation, specializing in API documentation discovery and code example generation. Built with FastAPI (Async), Google Gemini, and ChromaDB.

Note: This project has been migrated from Flask/OpenAI to a high-performance FastAPI/Gemini architecture.

🚀 Key Features

• ⚡ High-Performance Architecture: Fully async implementation using FastAPI, httpx, and asyncpg.
• 🧠 Advanced RAG Engine: Hybrid search using Google Gemini Embeddings and ChromaDB.
• 🤖 Intelligent Discovery: Automatically crawls and indexes documentation (Sitemaps, OpenAPI/Swagger, HTML).
• 📝 Code Generation: Uses Gemini 1.5 Flash to generate accurate code examples in 8+ languages (Python, Node.js, cURL, etc.).
• 🛡️ Production Ready: Features rate limiting, security headers, JSON logging, and Docker/Railway compatibility.

🏗️ Technology Stack

• Backend: Python 3.11+, FastAPI (Async)
• AI/LLM: Google Gemini (gemini-1.5-flash), google-genai SDK
• Vector DB: ChromaDB (Persistent)
• Database: PostgreSQL (Metadata & History) via SQLAlchemy 2.0 (Async)
• Processing: BeautifulSoup4, Trafilatura (Content Extraction)

📂 Repository Layout

.
├── app/
│   ├── api/                # FastAPI Routes & Dependencies
│   ├── core/               # RAG Engine & Logging
│   ├── db/                 # Database Models & Connections
│   ├── services/           # Document Processor & Crawler
│   └── utils/              # Validators & Helpers
├── scripts/
│   ├── ingest.py           # Document Ingestion Script
│   ├── reset_chroma.py     # Utility to clear Vector DB
│   └── fix_collation.sql   # Maintenance scripts
├── config.py               # Central Configuration
├── fastapi_app.py          # Application Entrypoint
├── Procfile                # Railway Deployment Config
└── pyproject.toml          # Dependencies

🛠️ Quickstart (Local Development)

1. Prerequisites

• Python 3.11+
• Google Gemini API Key
• PostgreSQL (Optional, defaults to SQLite for local)

2. Setup

# Clone and enter directory
git clone <repo-url>
cd rag-documentation-assistant

# Create virtual environment
python -m venv .venv
# Windows:
.venv\Scripts\Activate.ps1
# Linux/Mac:
source .venv/bin/activate

# Install dependencies
pip install -r requirements.txt

3. Configuration

Create a .env file based on .env.example:

# AI Provider
AI_PROVIDER=gemini
GEMINI_API_KEY=your_gemini_key_here

# Database (Default: SQLite)
DATABASE_URL=sqlite:///docrag.db

# Vector DB
CHROMA_DB_PATH=./chroma_db

4. Initialize & Run

# Initialize DB (Automatic on startup) and Start Server
uvicorn fastapi_app:app --reload

App will be ready at http://localhost:8000.

5. Ingest Documentation

To populate the vector database with documentation:

# Discover APIs automatically
API_DISCOVERY_ENABLED=true python scripts/ingest.py --discover-apis

# OR Ingest a specific URL
python scripts/ingest.py --url https://fastapi.tiangolo.com/

🔧 Maintenance

Resetting Embeddings

If you face "dimension mismatch" errors or want to re-index:

1. Stop the running server (Ctrl+C).
2. Run the reset script:

   python scripts/reset_chroma.py

3. Restart server and re-ingest.

Database Migrations

We use Alembic for schema migrations (if using Postgres):

alembic upgrade head

🚀 Deployment (Railway)

The project is configured for Railway out of the box using Procfile.

1. Push to GitHub.
2. Connect repository to Railway.
3. Add GEMINI_API_KEY and other variables in Railway Service Settings.
4. Attach a PostgreSQL plugin (optional but recommended for production).
5. Deploy!

📄 License

MIT