Skip to content

project-structure.md

Latest commit

 

History

History
41 lines (37 loc) · 2.35 KB

project-structure.md

File metadata and controls

41 lines (37 loc) · 2.35 KB
Raw

Project structure

A multi-module Gradle build.

CodeForge/
├── app/
│   ├── shared/          Compose Multiplatform UI, ViewModel, API client (commonMain)
│   │                    + platform code: Monaco bridge (web), plain editor (Android, Desktop)
│   ├── webApp/          WebAssembly entry point, index.html, monaco-bridge.js
│   ├── androidApp/      Android entry point
│   └── desktopApp/      Desktop entry point
└── backend/
    ├── src/main/kotlin/
    │   ├── config/      AppConfig: settings from the environment or the private secrets file
    │   ├── routes/      problems, submissions, Codeforces import
    │   ├── service/     JudgeService (builds the job, maps results to verdicts), Codeforces scraper, IST times
    │   ├── worker/      JudgeWorkers: claim, lease heartbeat, reaper, Redis wake-ups
    │   ├── sandbox/     Sandbox, DockerSandbox, ProcessSandbox
    │   ├── queue/       Redis connection and wake-up push
    │   ├── models/      Exposed tables
    │   └── database/    connection, schema, seed data
    ├── judge-sandbox/
    │   ├── harness/judge.py   compile once, run tests with limits, measure, decide status
    │   ├── test/run_tests.py  attack suite for both sandbox modes
    │   └── Dockerfile         sandbox image: g++, OpenJDK 21, Python 3
    ├── deploy/start.sh  Render entry point: moves secrets out of the environment
    ├── Dockerfile       Render image (process mode) and a docker-mode target
    └── docker-compose.yml   PostgreSQL, Redis and the backend with the Docker sandbox

Client architecture

  • JudgeViewModel holds the state: imported problem and its real limits, code per language, judging flag, the latest verdict, metrics and per-test results, history.
  • Importing a problem resets results, metrics and test nodes. Executing keeps the previous result on screen until the new verdict arrives.
  • The UI is built from small panels (ForgePanel) on one theme (ForgeTheme, ForgeType): verdict stamp, heat line, test-case nodes, metrics with limit bars, event log, results.
  • On the web, CodeTextArea places the Monaco editor (an HTML element) over an empty Compose box and keeps it aligned; it hides the editor while a dropdown is open over it.