AutoGrade — Automated Code Grading Platform

A modern, automated code grading platform built in Python. It executes user-submitted code in an isolated sandbox with resource constraints, runs test cases concurrently in a background thread pool, and features a sleek, dark-themed Single Page Application (SPA) web frontend.

---

Features

• 🖥️ Sleek Web Frontend: A premium, dark-mode browser interface to view problems, write Python code with line numbers, run tests, see live status updates, and create custom coding problems.
• ⚙️ Concurrently Graded Submissions: Submissions are queued using a thread-pool executor so the Flask API remains responsive.
• 🔒 Sandboxed Execution: Subprocesses are spawned in isolated temporary directories with blocklists for dangerous imports (such as subprocess, socket, requests, os.system, and ctypes).
• 📁 Unified SQLite Database: Flat-file database layout storing problems, submissions, and test results with automatic seeding of sample problems.
• 🐳 Docker Support: Exposes the application instantly through a lightweight container configuration.

---

Architecture

autograde/
├── app.py                  # Flask app factory + entry point
├── api/
│   ├── database.py         # Unified SQLite setup, seeding, & query models
│   └── routes.py           # REST API endpoints (submit, status, problems)
├── grader/
│   └── engine.py           # Sandboxed code runner & concurrent grader
├── static/
│   ├── index.html          # Frontend SPA structure
│   ├── index.css           # Premium dark-theme stylesheet
│   └── app.js              # Frontend API integration & reactive state
├── tests/
│   └── test_autograde.py   # Full test suite (pytest)
├── Dockerfile
├── docker-compose.yml
├── requirements.txt
└── .gitignore

---

Setup & Run

1. Local Development

Make sure you have Flask installed:

pip install -r requirements.txt

Start the server:

python app.py

Open http://127.0.0.1:5000/ in your web browser to access the frontend workspace!

2. Docker Container

Build and spin up the container:

docker-compose up --build

The application will be accessible at http://127.0.0.1:5000/.

---

API Reference

Create Problem

POST /api/problems

{
  "id": "two-sum",
  "title": "Sum of Two Numbers",
  "description": "Read two numbers and print their sum.",
  "test_cases": [
    { "name": "Basic Case", "input": "3\n4", "expected_output": "7" },
    { "name": "Zeros Case", "input": "0\n0", "expected_output": "0" }
  ],
  "time_limit_ms": 3000,
  "memory_limit_mb": 64
}

Submit Solution

POST /api/submit

{
  "problem_id": "two-sum",
  "code": "a = int(input())\nb = int(input())\nprint(a + b)"
}

Response (202 Accepted):

{
  "submission_id": "528574c8-32a4-4cc9-b7b5-27a372cc6d23",
  "status": "pending"
}

Poll Submission Status

GET /api/status/<submission_id>

{
  "id": "528574c8-32a4-4cc9-b7b5-27a372cc6d23",
  "status": "completed",
  "summary": { "total": 2, "passed": 2, "failed": 0, "score": 100.0 },
  "results": [
    {
      "test_name": "Basic Case",
      "passed": true,
      "expected_output": "7",
      "actual_output": "7",
      "execution_time_ms": 32.5,
      "stderr": "",
      "error_type": null
    }
  ]
}

Other Endpoints

Method	Endpoint	Description
GET	/api/problems	List all coding problems
GET	/api/problems/<id>	Fetch a specific problem
GET	/api/submissions	Retrieve recent submissions
POST	/api/batch	Batch submit (up to 20 solutions)
GET	/api/health	Service health status

---

Sandbox Security Limits

Each submission runs in a separate process with:

• Time Limits: Configurable timeout (default 5s) enforced using subprocess communications.
• Resource Constraints: Skip limits on Windows, and virtual memory/CPU constraints (resource.setrlimit) on POSIX/Linux platforms.
• Imports Restrictions: Restricts dangerous modules like socket, subprocess, requests, urllib, os.system, and ctypes inside the sandbox environment.

---

Running Tests

Run the pytest suite to verify grader reliability and sandboxing rules:

pytest tests/ -v