penr-oz-gpt-chat

GPT Chat Client leveraging Neural Network Service

• Based on ng-video-lecture, nanoGPT and nanochat
• Using the Neural Network service

API Contract

See API.md for the full request/response contracts between the React client, Express proxy, and Neural Network service.

Quickstart Guide

1. Clone the Repository:

   git clone https://github.com/derinworks/penr-oz-gpt-chat.git
   cd penr-oz-gpt-chat

2. Setup:

   • Install dependencies:

     npm install

   • Configure environment by copying the example file and editing it:

     cp .env.example .env

     Update .env with your Neural Network service URL:

     VITE_PREDICTION_SERVER_URL=http://localhost:8000

3. Neural Network Service:

   • Follow instructions on Quick Start Guide
   • Deployed remotely then use a .env file as such to configure url:

   VITE_PREDICTION_SERVER_URL=http://???:8000

4. Run:

   • Launch React client (Vite dev server only):

     npm start

     App running at http://localhost:3000
   • Build for production (type-check + Vite bundle):

     npm run build

   • Start the proxy server (Express on port 3001):

     npm run server

   • Run client and server together (recommended for development):

     npm run dev

     React client at http://localhost:3000, proxy server at http://localhost:3001 (auto-restarts on server file changes)
   • Preview the production build:

     npm run preview

5. Testing:

   • Run all unit tests:

     npm test

   • Run tests with coverage report (requires ≥ 80% line/function coverage on core modules):

     npm run test:coverage

   • Tests live alongside source files (src/api.test.ts) and in the server directory (server/index.test.ts).
   • The test stack uses Vitest with Supertest for HTTP assertions and built-in vi.fn() mocks for dependency isolation.