go-chat

A multi-service, cloud-native chat system prototype built to explore gRPC, WebSockets, multi-language service orchestration, and real-time messaging infrastructure.

Overview

This system includes:

• chat-service: Core backend (Go) using gRPC for message processing and broadcasting
• chat-gateway: WebSocket bridge (Node.js) between frontend clients and gRPC
• chat-ai: AI responder (Python) powered by Hugging Face transformers, listening via gRPC stream
• chat-client: Lightweight HTML/Vue.js frontend for testing gateway connection

Features

• Bi-directional chat flow using gRPC and WebSocket
• Modular, multi-language architecture (Go, Node.js, Python, Vue.js)
• AI-generated responses using a Hugging Face transformer model
• Isolated development environments using devcontainers and Dockerfiles
• Unified Makefile and concurrently-powered orchestration
• Docker Compose configuration for full stack deployment

Development

Each service has its own Makefile for local dev. You can run everything at once with:

make run-all

Or run them individually:

make -C chat-service run
make -C chat-gateway run
make -C chat-ai run
make -C chat-client run

Testing

Basic unit tests are scaffolded for each service. Run them using:

make -C chat-service test
make -C chat-ai test

Docker Compose

To bring up all services:

docker-compose up --build

Services will be available at:

• chat-service: gRPC at localhost:8080
• chat-gateway: WebSocket server at localhost:8081
• chat-client: Vue.js test page served at localhost:3000

---

Full System Overview (Extended)

Services

• chat-gateway: Node.js WebSocket bridge that proxies messages between frontend clients and chat-service
• chat-client: Simple static Vue.js test page served over Python HTTP, connects via WebSocket to the gateway
• chat-ai: Python gRPC stream listener that connects to chat-service and replies to user messages using a Hugging Face model

Development Workflow

Run all services concurrently:

make run-all

This uses a custom script powered by concurrently to orchestrate startup of all services.

Each service also has an individual Makefile to support:

• make run
• make test
• make proto
• make lint (where applicable)

Docker Support

To run the full system using Docker:

docker-compose up --build

The following ports will be exposed:

• 8080: gRPC service (chat-service)
• 8081: WebSocket gateway (chat-gateway)
• 3000: Static client HTML page (chat-client)

All containers share a virtual network and reference chat-service by name.

---

Status

All services scaffolded with basic unit tests and development Makefiles. The AI responder connects successfully to the gRPC stream. Logging, proto generation, and local dev tools are implemented.

Roadmap

• Add multi-room support to enable isolated conversations per room (e.g., general, support, random)
• Externalize configuration for ports, model selection, and service hosts via environment variables or .env files
• Harden service loops with retries, timeouts, and graceful reconnection for all gRPC streaming clients
• Enhance AI response quality with context history, prompt tuning, or model upgrade (e.g., replace DialoGPT)
• Integrate logging frameworks with consistent formats, timestamps, and log levels across all services

---

License

MIT