Spaces:
Sleeping
Sleeping
| # server/app.py | |
| from fastapi import FastAPI, HTTPException | |
| from openenv.core.env_server import create_fastapi_app | |
| from ..models import Action, Observation, TaskInfo | |
| from .environment import CodeDebugEnvironment | |
| from .tasks import TASK_REGISTRY | |
| from .grader import grade | |
| # Core OpenEnv app (provides /reset, /step, /state, /ws, /health) | |
| app = create_fastapi_app(CodeDebugEnvironment, Action, Observation) | |
| # ── Additional required hackathon endpoints ──────────────────────────── | |
| from fastapi.responses import HTMLResponse | |
| def home(): | |
| return """ | |
| <!DOCTYPE html> | |
| <html> | |
| <head> | |
| <title>Code Debug Env | OpenEnv</title> | |
| <style> | |
| @import url('https://fonts.googleapis.com/css2?family=Inter:wght@400;600;800&display=swap'); | |
| body { | |
| font-family: 'Inter', sans-serif; | |
| background: linear-gradient(135deg, #0f172a 0%, #1e1b4b 100%); | |
| color: #f8fafc; | |
| display: flex; | |
| flex-direction: column; | |
| align-items: center; | |
| justify-content: center; | |
| min-height: 100vh; | |
| margin: 0; | |
| text-align: center; | |
| overflow-x: hidden; | |
| } | |
| .container { | |
| background: rgba(30, 41, 59, 0.7); | |
| backdrop-filter: blur(12px); | |
| padding: 3rem; | |
| border-radius: 24px; | |
| border: 1px solid rgba(255, 255, 255, 0.1); | |
| box-shadow: 0 25px 50px -12px rgba(0, 0, 0, 0.5); | |
| max-width: 600px; | |
| animation: fadeIn 0.8s ease-out; | |
| } | |
| @keyframes fadeIn { from { opacity: 0; transform: translateY(20px); } to { opacity: 1; transform: translateY(0); } } | |
| h1 { font-size: 3rem; margin-bottom: 0.5rem; background: linear-gradient(to right, #60a5fa, #a78bfa); -webkit-background-clip: text; -webkit-text-fill-color: transparent; } | |
| p { color: #94a3b8; font-size: 1.1rem; line-height: 1.6; } | |
| .badge { background: #334155; padding: 4px 12px; border-radius: 20px; font-size: 0.8rem; vertical-align: middle; border: 1px solid #475569; } | |
| .grid { display: grid; grid-template-columns: 1fr 1fr; gap: 1rem; margin-top: 2rem; } | |
| .btn { | |
| display: block; | |
| padding: 1rem; | |
| border-radius: 12px; | |
| background: rgba(255, 255, 255, 0.05); | |
| border: 1px solid rgba(255, 255, 255, 0.1); | |
| color: #fff; | |
| text-decoration: none; | |
| transition: all 0.2s; | |
| font-weight: 600; | |
| position: relative; | |
| overflow: hidden; | |
| } | |
| .btn:hover { background: rgba(255, 255, 255, 0.1); border-color: #60a5fa; transform: translateY(-2px); } | |
| .btn.loading { pointer-events: none; opacity: 0.7; } | |
| .loading-text { display: none; } | |
| .btn.loading .loading-text { display: inline-block; animation: pulse 1.5s infinite; } | |
| .btn.loading span:not(.loading-text) { display: none; } | |
| @keyframes pulse { 0% { opacity: 0.4; } 50% { opacity: 1; } 100% { opacity: 0.4; } } | |
| footer { margin-top: 2rem; color: #64748b; font-size: 0.9rem; } | |
| </style> | |
| </head> | |
| <body> | |
| <div class="container" id="main-card"> | |
| <h1>Code Debug Env</h1> | |
| <p>A production-grade <strong>OpenEnv</strong> for training frontier reasoning agents on code repair tasks. <span class="badge">v1.1.0</span></p> | |
| <div class="grid"> | |
| <a href="/health" class="btn">📡 Server Health</a> | |
| <a href="/tasks" class="btn">📋 Task Registry</a> | |
| <a id="baseline-btn" href="/baseline" class="btn"> | |
| <span>🤖 Run Baseline</span> | |
| <span class="loading-text">⏳ Evaluating (~2 mins)...</span> | |
| </a> | |
| <a href="https://huggingface.co/spaces/luciferai-devil/code-debug-env" class="btn">📖 Documentation</a> | |
| </div> | |
| <footer>Built for Meta & PyTorch × Scaler Hackathon</footer> | |
| </div> | |
| <script> | |
| document.getElementById('baseline-btn').addEventListener('click', function(e) { | |
| this.classList.add('loading'); | |
| }); | |
| </script> | |
| </body> | |
| </html> | |
| """ | |
| def list_tasks() -> list[TaskInfo]: | |
| """Return all tasks with their action schema.""" | |
| return [ | |
| TaskInfo( | |
| task_id=tid, | |
| difficulty=task["difficulty"], | |
| description=task["description"], | |
| action_schema=Action.model_json_schema(), | |
| ) | |
| for tid, task in TASK_REGISTRY.items() | |
| ] | |
| def get_grader_score(task_id: str, submitted_code: str) -> dict: | |
| """ | |
| Grade a submission directly (for testing / evaluation). | |
| Returns: { score: float, passed: int, total: int, test_results: list } | |
| """ | |
| if task_id not in TASK_REGISTRY: | |
| raise HTTPException(status_code=404, detail=f"Unknown task_id: {task_id}") | |
| task = TASK_REGISTRY[task_id] | |
| result = grade(submitted_code, task_id, task["test_suite"]) | |
| return { | |
| "task_id": task_id, | |
| "score": result["score"], | |
| "passed": result["passed"], | |
| "total": result["total"], | |
| "test_results": [r.model_dump() for r in result["test_results"]], | |
| } | |
| def run_baseline() -> dict: | |
| """ | |
| Run the baseline agent on all tasks and return scores. | |
| This endpoint triggers the baseline inference script. | |
| """ | |
| import subprocess, sys, json, re, os | |
| try: | |
| # Get absolute path to the baseline script | |
| base_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) | |
| baseline_script = os.path.join(base_dir, "baseline", "run_baseline.py") | |
| result = subprocess.run( | |
| [sys.executable, baseline_script, "--output", "json"], | |
| capture_output=True, text=True, timeout=120, | |
| ) | |
| # Robustly find JSON in potentially noisy stdout | |
| stdout = result.stdout.strip() | |
| stderr = result.stderr.strip() | |
| match = re.search(r'(\{.*\})', stdout, re.DOTALL) | |
| if match: | |
| try: | |
| return json.loads(match.group(1)) | |
| except Exception as j_err: | |
| raise ValueError(f"JSON Decode Error: {j_err}. Raw Match: {match.group(1)}") | |
| raise ValueError(f"No JSON found. Stdout: {stdout[:100]}. Stderr: {stderr[:100]}. ReturnCode: {result.returncode}") | |
| except Exception as e: | |
| raise HTTPException(status_code=500, detail=str(e)) | |
| def main(): | |
| """Entry point for the server.""" | |
| import uvicorn | |
| uvicorn.run(app, host="0.0.0.0", port=8000) | |
| if __name__ == "__main__": | |
| main() | |