DNFileRAG.Infrastructure 1.0.0

DNFileRAG

A .NET 9-powered, real-time file-driven RAG (Retrieval-Augmented Generation) engine that auto-ingests documents and serves fast, contextual query responses via API.

Installation

NuGet Package

dotnet add package DNFileRAG.Infrastructure

From Source

git clone https://github.com/MLidstrom/DNFileRAG.git
cd DNFileRAG
dotnet build

Features

• Real-time Document Ingestion - Automatically watches directories and indexes new/modified documents
• Multiple Document Formats - Supports PDF, DOCX, TXT, MD, and HTML files
• Flexible Embedding Providers - OpenAI, Azure OpenAI, or Ollama (local)
• Multiple LLM Providers - OpenAI, Azure OpenAI, Anthropic, or Ollama (local)
• Vector Search - Powered by Qdrant for fast semantic search
• REST API - Simple API for queries and document management
• Clean Architecture - Modular, testable, and extensible design

Quick Start

Prerequisites

• .NET 9 SDK
• Qdrant vector database (Docker recommended)
• One of: OpenAI API key, Azure OpenAI, or Ollama for local inference

1. Start Qdrant

docker run -p 6333:6333 -p 6334:6334 qdrant/qdrant

2. Clone and Build

git clone https://github.com/MLidstrom/DNFileRAG.git
cd DNFileRAG
dotnet build

3. Configure

Edit src/DNFileRAG/appsettings.json:

{
  "Qdrant": {
    "Host": "localhost",
    "Port": 6333,
    "CollectionName": "DNFileRAG",
    "VectorSize": 1536
  },
  "Embedding": {
    "Provider": "Ollama",
    "Ollama": {
      "BaseUrl": "http://localhost:11434",
      "Model": "nomic-embed-text"
    }
  },
  "Llm": {
    "Provider": "Ollama",
    "Ollama": {
      "BaseUrl": "http://localhost:11434",
      "Model": "llama3.2"
    }
  }
}

4. Run

cd src/DNFileRAG
dotnet run

The API will be available at http://localhost:8181

API Endpoints

Query Example

curl -X POST http://localhost:8181/api/query \
  -H "Content-Type: application/json" \
  -d '{"query": "What is the main topic of the documents?"}'

Project Structure

DNFileRAG/
├── src/
│   ├── DNFileRAG/                 # Web API host
│   ├── DNFileRAG.Core/            # Domain models & interfaces
│   └── DNFileRAG.Infrastructure/  # External service implementations
│       ├── Embeddings/            # Embedding providers
│       ├── Llm/                   # LLM providers
│       ├── Parsers/               # Document parsers
│       ├── Services/              # Core services
│       └── VectorStore/           # Qdrant integration
└── tests/
    └── DNFileRAG.Tests/           # Unit tests

Configuration

Embedding Providers

LLM Providers

Local Development with Ollama

For fully local development without API keys:

1. Install Ollama
2. Pull required models:

   ollama pull mxbai-embed-large  # Recommended: 1024 dims, stable
   ollama pull llama3.2
   

3. Set providers to Ollama in configuration
4. Configure vector size to match embedding model:

Query Guardrails

DNFileRAG includes relevance score filtering to prevent off-topic queries:

{
  "Rag": {
    "MinRelevanceScore": 0.6
  }
}

Queries with no documents above the threshold return a "no relevant information" response without calling the LLM, saving cost and preventing hallucination

Docker

docker-compose up -d

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Qdrant - Vector database
• Serilog - Structured logging
• PdfPig - PDF parsing
• DocumentFormat.OpenXml - DOCX parsing