ntrx 🛰️ – RTK Correction Data Dissemination

Welcome to ntrx!
A high-performance, asynchronous, Redis-native NTRIP caster written in Python using FastAPI and asyncio.

---

🚀 Key Features

• Redis-Native Architecture: Uses Redis Pub/Sub for Inter-Process Communication (IPC).
  • Live Position Streaming: Client NMEA positions are streamed to Redis (ntripcaster:positions).
  • Kill Switch: Admin control via Redis (ntripcaster:control) to actively disconnect users.
• Modern Python Implementation:
  • Built with Python 3.12+, FastAPI, and AsyncIO.
  • Strict Typing: Uses Pydantic models for all data exchange and state management.
  • Clean Code: Adheres to SOLID principles, low cyclomatic complexity, and class-based design.
• Scalable: Stateless design allows horizontal scaling of API instances.

---

🧩 Architecture

See ARCHITECTURE.md for detailed diagrams and schema information.

• NTRIP Caster: Handles source/client connections, parses NMEA, and publishes events to Redis.
• FastAPI Server: Provides a REST/WebSocket API for monitoring and control (Kill Switch).
• Redis: Central message bus for positions, control commands, and shared state.

---

⚡ Quick Start

1. Prerequisites

• Python 3.12+
• Redis Server (optional — required for live state, position streaming, and kill switch)

2. Installation

git clone https://github.com/ndjuric/ntrx.git
cd ntrx
python -m venv venv
source venv/bin/activate
pip install -e .  # Install in editable mode

3. Configuration

Create a .env file in the project root:

REDIS_HOST=127.0.0.1
REDIS_PASSWORD=changeme
FASTAPI_HOST=0.0.0.0
FASTAPI_PORT=8000
LOG_MAX_SIZE_MB=10
LOG_MAX_BACKUP_COUNT=5

Edit storage/ntripcaster.json to define your Source and Client credentials.

4. Running the System

For full functionality (live state, position streaming, kill switch), run Redis alongside the Caster and API. The Caster and API can also run without Redis in degraded mode — core NTRIP data relay will work normally.

Option A: Individual Processes

1. Start Redis: docker-compose up -d
2. Start API: python -m ntrx api
3. Start Caster: python -m ntrx ntripcaster

Option B: CLI Helper

python -m ntrx --help

Note: Redis is optional. If Redis is not running, the caster and API will start in degraded mode — core NTRIP functionality works, but live state publishing, position streaming, and the control channel (kill switch) will be disabled. The program automatically detects docker-compose.yml in the project root and suggests docker-compose up -d if Redis is unreachable.

TODO – Operational Modes: Two modes are planned:

1. Standalone (current) – mountpoints and state are kept in-memory; Redis is optional (IPC only).
2. Thin Client – mountpoints and all state are stored in Redis; the caster becomes stateless, enabling horizontal scaling behind a load balancer.

---

📡 API Endpoints

• GET /state – Returns the current system state (connected sources/clients).
• POST /api/kill/{username} – Disconnects a generic user (Source or Client) immediately.
• WS /ws – Real-time state updates via WebSocket.
• GET / & GET /debug/routes - Debug utilities.

---

🧪 Testing & Verification

A robust integration test suite is included to verify the entire pipeline (Socket -> Caster -> Redis -> API).

# Verify connection, streaming, and kill switch
python src/ntrx/tests/integration_test.py

---

🤝 Contributing

• Code must use Pydantic models for any data structure.
• All IPC must go through Redis.
• Cyclomatic complexity of methods must be kept under 10.
• 100% adherence to Class-Based structure.

---

📄 License

MIT License