RAG System for Codebases

A production-grade RAG system combining dense vector search and sparse keyword retrieval for deep code understanding.

Features

• Hybrid Retrieval: Combines Qdrant (Vector) and Redis (Keyword/BM25) with RRF fusion.
• Deep Indexing: AST-based parsing for chunking of Go code.
• Hierarchical Context: Understanding from file to function level.
• LLM Integration: Works seamlessly with local Ollama models.
• Automated Docs: Swagger/OpenAPI documentation auto-generated.

Infrastructure Setup

The system uses a hybrid deployment model for the best local development experience:

• Infrastructure: Qdrant and Redis run in Docker containers (via Docker Compose).
• Application: The Go server runs natively on your host for full filesystem access.

Prerequisites

• Go 1.22+
• Docker & Docker Compose
• Ollama running locally

Quick Start

1. Start Infrastructure (Qdrant & Redis)

   docker-compose up -d

2. Build Server

   go build -o rag-server cmd/rag-server/main.go

3. Run Server

   ./rag-server

   The API will be available at http://localhost:8080.

API Usage

Indexing a Codebase

You can index any folder on your local machine.

curl -X POST http://localhost:8080/api/index \
  -H "Content-Type: application/json" \
  -d '{"path": "/Users/guru/projects/my-app"}'

Querying

Ask natural language questions about your code.

curl -X POST http://localhost:8080/api/query \
  -H "Content-Type: application/json" \
  -d '{
    "query": "How is authentication handled?",
    "max_results": 5
  }'

API Documentation

Swagger UI is available at: http://localhost:8080/swagger/index.html

Regenerate docs:

swag init -g cmd/rag-server/main.go

Configuration

Configuration is managed via .env file.

# Ollama
OLLAMA_URL=http://localhost:11434
EMBEDDING_MODEL=all-minilm
LLM_MODEL=llama3.2:1b

# Databases
VECTOR_STORE_URL=http://localhost:6333
REDIS_URL=localhost:6379

# Hybrid Search Tuning
HYBRID_ENABLED=true
HYBRID_VECTOR_WEIGHT=0.7
FUSION_STRATEGY=rrf

Project Structure

├── cmd/rag-server/      # Application entrypoint
├── internal/
│   ├── api/             # HTTP handlers & middleware
│   ├── indexing/        # AST parsing & chunking logic
│   ├── retrieval/       # Hybrid search & ranking engine
│   ├── vectorstore/     # Qdrant integration
│   └── domain/          # Core data models
├── docs/                # Generated Swagger docs
└── docker-compose.yml   # Infrastructure orchestration