Spaces:

arya89
/

openops

No application file

File size: 17,481 Bytes

"""
OpenOps FastAPI Server
Provides REST API endpoints for the incident management environment
"""

import sys
from pathlib import Path

# Add parent directory to path
sys.path.insert(0, str(Path(__file__).parent.parent))

from fastapi import FastAPI, HTTPException, Query
from fastapi.responses import JSONResponse, HTMLResponse
from pydantic import BaseModel
from typing import Dict, Any, Optional
import uvicorn

from server.my_env_environment import MyEnvEnvironment
from models import IncidentAction, IncidentObservation


# FastAPI app
app = FastAPI(
    title="OpenOps API",
    description="Production Incident Management Environment API",
    version="1.0.0"
)

# Global environment instance
env_instance: Optional[MyEnvEnvironment] = None


# Request Models
class StepRequest(BaseModel):
    action_id: int
    task_id: Optional[int] = 1


# ============================================================
# LANDING PAGE WITH STATUS
# ============================================================

@app.get("/", response_class=HTMLResponse)
async def root():
    """Landing page with environment status and documentation."""
    return """
    <!DOCTYPE html>
    <html lang="en">
    <head>
        <meta charset="UTF-8">
        <meta name="viewport" content="width=device-width, initial-scale=1.0">
        <title>OpenOps - AI Incident Commander</title>
        <style>
            * { margin: 0; padding: 0; box-sizing: border-box; }
            body {
                font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
                background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
                min-height: 100vh;
                padding: 20px;
            }
            .container {
                max-width: 900px;
                margin: 0 auto;
                background: white;
                border-radius: 12px;
                box-shadow: 0 20px 60px rgba(0,0,0,0.3);
                overflow: hidden;
            }
            .header {
                background: linear-gradient(135deg, #f093fb 0%, #f5576c 100%);
                color: white;
                padding: 40px 30px;
                text-align: center;
            }
            .header h1 {
                font-size: 2.5em;
                margin-bottom: 10px;
                text-shadow: 2px 2px 4px rgba(0,0,0,0.2);
            }
            .status {
                display: inline-block;
                background: rgba(255,255,255,0.2);
                padding: 8px 20px;
                border-radius: 20px;
                font-size: 0.9em;
                margin-top: 10px;
            }
            .content {
                padding: 40px 30px;
            }
            .section {
                margin-bottom: 30px;
            }
            .section h2 {
                color: #333;
                margin-bottom: 15px;
                font-size: 1.5em;
                border-left: 4px solid #667eea;
                padding-left: 15px;
            }
            .endpoint {
                background: #f8f9fa;
                border-left: 4px solid #667eea;
                padding: 15px 20px;
                margin: 10px 0;
                border-radius: 4px;
                transition: all 0.3s;
            }
            .endpoint:hover {
                background: #e9ecef;
                transform: translateX(5px);
            }
            .endpoint strong {
                color: #667eea;
                display: block;
                margin-bottom: 5px;
            }
            .endpoint code {
                background: white;
                padding: 2px 8px;
                border-radius: 3px;
                font-family: 'Courier New', monospace;
                color: #e83e8c;
            }
            .stats {
                display: grid;
                grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
                gap: 20px;
                margin-top: 20px;
            }
            .stat-card {
                background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
                color: white;
                padding: 20px;
                border-radius: 8px;
                text-align: center;
            }
            .stat-card h3 {
                font-size: 2em;
                margin-bottom: 5px;
            }
            .stat-card p {
                opacity: 0.9;
                font-size: 0.9em;
            }
            .features {
                display: grid;
                grid-template-columns: repeat(auto-fit, minmax(250px, 1fr));
                gap: 15px;
                margin-top: 20px;
            }
            .feature {
                background: #f8f9fa;
                padding: 20px;
                border-radius: 8px;
                border-left: 4px solid #28a745;
            }
            .feature h3 {
                color: #28a745;
                margin-bottom: 10px;
                font-size: 1.1em;
            }
            .button {
                display: inline-block;
                background: #667eea;
                color: white;
                padding: 12px 30px;
                text-decoration: none;
                border-radius: 6px;
                margin: 10px 10px 10px 0;
                transition: all 0.3s;
                font-weight: 600;
            }
            .button:hover {
                background: #764ba2;
                transform: translateY(-2px);
                box-shadow: 0 5px 15px rgba(102, 126, 234, 0.4);
            }
            .footer {
                background: #f8f9fa;
                padding: 20px;
                text-align: center;
                color: #6c757d;
                font-size: 0.9em;
            }
            .badge {
                display: inline-block;
                background: #28a745;
                color: white;
                padding: 4px 12px;
                border-radius: 12px;
                font-size: 0.85em;
                margin-left: 10px;
            }
        </style>
    </head>
    <body>
        <div class="container">
            <div class="header">
                <h1>🚨 OpenOps</h1>
                <p style="font-size: 1.2em; margin-top: 10px;">AI Incident Commander Environment</p>
                <div class="status">✅ Status: Running</div>
            </div>
            
            <div class="content">
                <div class="section">
                    <h2>📊 Performance Metrics</h2>
                    <div class="stats">
                        <div class="stat-card">
                            <h3>0.85</h3>
                            <p>Task 1 Score<br>(Easy)</p>
                        </div>
                        <div class="stat-card">
                            <h3>0.85</h3>
                            <p>Task 2 Score<br>(Medium)</p>
                        </div>
                        <div class="stat-card">
                            <h3>1.00</h3>
                            <p>Task 3 Score<br>(Hard)</p>
                        </div>
                        <div class="stat-card">
                            <h3>0.90</h3>
                            <p>Average Score<br>⭐</p>
                        </div>
                    </div>
                </div>

                <div class="section">
                    <h2>🎯 Environment Overview</h2>
                    <p style="line-height: 1.8; color: #555;">
                        OpenOps simulates realistic production incidents where AI agents must investigate alerts, 
                        identify root causes, execute mitigation actions, communicate with stakeholders, 
                        and resolve incidents to minimize revenue loss.
                    </p>
                </div>

                <div class="section">
                    <h2>✨ Key Features</h2>
                    <div class="features">
                        <div class="feature">
                            <h3>🔍 Smart Detection</h3>
                            <p>Intelligent incident classification based on alerts and logs</p>
                        </div>
                        <div class="feature">
                            <h3>📈 Progressive Difficulty</h3>
                            <p>Three tasks from simple crashes to cascading failures</p>
                        </div>
                        <div class="feature">
                            <h3>⚡ Rule-Based Agent</h3>
                            <p>Reliable baseline with optimized playbooks</p>
                        </div>
                        <div class="feature">
                            <h3>🎮 21 Actions</h3>
                            <p>Investigate, mitigate, communicate, and resolve</p>
                        </div>
                    </div>
                </div>

                <div class="section">
                    <h2>🔌 API Endpoints</h2>
                    
                    <div class="endpoint">
                        <strong>API Documentation</strong>
                        <code>GET /docs</code>
                        <p style="margin-top: 8px; color: #666;">Interactive Swagger UI with all endpoints</p>
                    </div>
                    
                    <div class="endpoint">
                        <strong>Reset Environment</strong>
                        <code>POST /reset?task_id=1</code>
                        <p style="margin-top: 8px; color: #666;">Initialize a new incident scenario (task_id: 1, 2, or 3)</p>
                    </div>
                    
                    <div class="endpoint">
                        <strong>Execute Action</strong>
                        <code>POST /step</code>
                        <p style="margin-top: 8px; color: #666;">Take an action and receive observation</p>
                    </div>
                    
                    <div class="endpoint">
                        <strong>Get State</strong>
                        <code>GET /state</code>
                        <p style="margin-top: 8px; color: #666;">Retrieve current environment state</p>
                    </div>
                    
                    <div class="endpoint">
                        <strong>List Actions</strong>
                        <code>GET /actions</code>
                        <p style="margin-top: 8px; color: #666;">Get all available actions</p>
                    </div>
                </div>

                <div class="section">
                    <h2>🚀 Quick Start</h2>
                    <a href="/docs" class="button">📚 View API Docs</a>
                    <a href="https://github.com/arya89/OpenOps" class="button">💻 GitHub</a>
                    <a href="https://huggingface.co/spaces/arya89/openops-incident-commander" class="button">🤗 HuggingFace</a>
                </div>

                <div class="section">
                    <h2>📖 Three Progressive Tasks</h2>
                    <div class="endpoint">
                        <strong>Task 1: Simple API Crash <span class="badge">Easy</span></strong>
                        <p style="margin-top: 8px; color: #666;">API service down due to OOM → Inspect logs → Restart → Resolve</p>
                    </div>
                    <div class="endpoint">
                        <strong>Task 2: Bad Deployment <span class="badge">Medium</span></strong>
                        <p style="margin-top: 8px; color: #666;">Database deployment broke queries → Rollback → Notify team → Resolve</p>
                    </div>
                    <div class="endpoint">
                        <strong>Task 3: Cascading Failure <span class="badge">Hard</span></strong>
                        <p style="margin-top: 8px; color: #666;">DB overload → API timeouts → Scale DB → Restart API → Communicate → Resolve</p>
                    </div>
                </div>
            </div>
            
            <div class="footer">
                <p><strong>OpenOps v1.0.0</strong></p>
                <p>Built for OpenEnv AI Hackathon 2026 | Meta PyTorch x Scaler</p>
                <p style="margin-top: 10px;">Created by arya89 | Production-ready incident management environment</p>
            </div>
        </div>
    </body>
    </html>
    """


@app.get("/health")
async def health():
    """Detailed health check."""
    return {
        "status": "healthy",
        "environment_loaded": env_instance is not None,
        "current_task": env_instance.task_id if env_instance else None,
        "version": "1.0.0"
    }


# ============================================================
# ENVIRONMENT API ENDPOINTS
# ============================================================

@app.post("/reset")
async def reset(task_id: int = Query(default=1, ge=1, le=3)) -> Dict[str, Any]:
    """
    Reset the environment for a specific task.
    OpenEnv standard endpoint.
    
    Args:
        task_id: Task difficulty (1=easy, 2=medium, 3=hard)
    
    Returns:
        Initial observation after reset
    """
    global env_instance
    
    try:
        # Create new environment instance
        env_instance = MyEnvEnvironment()
        obs = env_instance.reset(task_id=task_id)
        
        # Return observation in OpenEnv format
        return obs.model_dump()
        
    except Exception as e:
        raise HTTPException(
            status_code=500,
            detail=f"Failed to reset environment: {str(e)}"
        )


@app.post("/step")
async def step(request: StepRequest) -> Dict[str, Any]:
    """
    Execute an action in the environment.
    
    Args:
        request: StepRequest with action_id and task_id
    
    Returns:
        Observation after taking the action
    """
    global env_instance
    
    try:
        # Check if environment is initialized
        if env_instance is None:
            raise HTTPException(
                status_code=400,
                detail="Environment not initialized. Call /reset first."
            )
        
        # Validate action_id
        if request.action_id < 0 or request.action_id > 20:
            raise HTTPException(
                status_code=400,
                detail=f"Invalid action_id: {request.action_id}. Must be 0-20."
            )
        
        # Create action
        action = IncidentAction(
            action_id=request.action_id,
            task_id=request.task_id if request.task_id else env_instance.task_id
        )
        
        # Execute step
        obs = env_instance.step(action)
        
        # Return observation in OpenEnv format
        return obs.model_dump()
        
    except Exception as e:
        raise HTTPException(
            status_code=500,
            detail=f"Failed to execute step: {str(e)}"
        )


@app.get("/state")
async def get_state() -> Dict[str, Any]:
    """
    Get current environment state.
    
    Returns:
        Current state of the environment
    """
    global env_instance
    
    try:
        if env_instance is None:
            raise HTTPException(
                status_code=400,
                detail="Environment not initialized. Call /reset first."
            )
        
        state = env_instance.state
        
        return {
            "state": state.model_dump() if hasattr(state, 'model_dump') else state,
            "task_id": env_instance.task_id,
            "total_reward": env_instance.total_reward,
            "incident_resolved": env_instance.incident_resolved,
            "time_elapsed": env_instance.time_elapsed
        }
        
    except Exception as e:
        raise HTTPException(
            status_code=500,
            detail=f"Failed to get state: {str(e)}"
        )


@app.get("/actions")
async def get_actions() -> Dict[str, Any]:
    """
    Get list of available actions.
    
    Returns:
        Dictionary of action IDs and names
    """
    try:
        temp_env = MyEnvEnvironment()
        
        return {
            "actions": temp_env.ACTION_NAMES,
            "total_actions": len(temp_env.ACTION_NAMES)
        }
        
    except Exception as e:
        raise HTTPException(
            status_code=500,
            detail=f"Failed to get actions: {str(e)}"
        )


# ============================================================
# ERROR HANDLERS
# ============================================================

@app.exception_handler(HTTPException)
async def http_exception_handler(request, exc):
    """Handle HTTP exceptions."""
    return JSONResponse(
        status_code=exc.status_code,
        content={
            "error": exc.detail,
            "status_code": exc.status_code
        }
    )


@app.exception_handler(Exception)
async def general_exception_handler(request, exc):
    """Handle general exceptions."""
    return JSONResponse(
        status_code=500,
        content={
            "error": "Internal server error",
            "detail": str(exc)
        }
    )


# ============================================================
# MAIN ENTRY POINT (Required by OpenEnv)
# ============================================================

def main():
    """
    Main entry point for the server.
    Required by OpenEnv validation.
    """
    uvicorn.run(
        app,
        host="0.0.0.0",
        port=7860,
        log_level="info"
    )


if __name__ == "__main__":
    main()