hf_demo.py

# hf_demo.py – ARF v4 API with Memory
from fastapi import FastAPI, HTTPException
from fastapi.middleware.cors import CORSMiddleware
import gradio as gr
import numpy as np
from datetime import datetime

# ARF v4 imports
from agentic_reliability_framework.core.governance.risk_engine import RiskEngine
from agentic_reliability_framework.runtime.memory import create_faiss_index, RAGGraphMemory
from agentic_reliability_framework.runtime.memory.constants import MemoryConstants

app = FastAPI(title="ARF v4 API with Memory")

# Enable CORS for your frontend
app.add_middleware(
    CORSMiddleware,
    allow_origins=["https://arf-frontend-sandy.vercel.app"],
    allow_methods=["*"],
)

# ---------------------------------------------------------------------------
# Initialize ARF components
# ---------------------------------------------------------------------------
risk_engine = RiskEngine()

# Create FAISS index and memory (using default dimension from constants)
faiss_index = create_faiss_index(dim=MemoryConstants.VECTOR_DIM)
memory = RAGGraphMemory(faiss_index)

# ---------------------------------------------------------------------------
# API Endpoints
# ---------------------------------------------------------------------------

@app.get("/")
async def root():
    return {
        "service": "ARF OSS API",
        "version": "4.0.0",
        "status": "operational",
        "memory_stats": memory.get_graph_stats() if memory.has_historical_data() else "empty"
    }

@app.get("/health")
async def health():
    return {"status": "ok", "version": "4.0.0"}

@app.get("/api/v1/get_risk")
async def get_risk():
    risk_score = risk_engine.get_current_risk()
    return {
        "system_risk": risk_score.mean,
        "status": "critical" if risk_score.mean > 0.8 else "normal"
    }

@app.post("/api/v1/incident")
async def store_incident(event_data: dict, analysis: dict):
    try:
        incident_id = memory.store_incident(event_data, analysis)
        return {"incident_id": incident_id}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/api/v1/memory/similar")
async def find_similar_incidents(action: str, k: int = 5):
    class DummyEvent:
        def __init__(self, action):
            self.component = "user_action"
            self.latency_p99 = 0.0
            self.error_rate = 0.0
            self.throughput = 0
            self.cpu_util = 0.0
            self.memory_util = 0.0
            self.timestamp = datetime.now()
            self.severity = "low"
    event = DummyEvent(action)
    analysis = {"action": action}
    similar = memory.find_similar(event, analysis, k=k)
    results = []
    for node in similar:
        results.append({
            "incident_id": node.incident_id,
            "component": node.component,
            "severity": node.severity,
            "timestamp": node.timestamp,
            "metrics": node.metrics,
            "agent_analysis": node.agent_analysis,
            "similarity_score": node.metadata.get("similarity_score", 0.0)
        })
    return {"similar": results, "count": len(results)}

@app.get("/api/v1/memory/stats")
async def memory_stats():
    return memory.get_graph_stats()

# Optional Gradio interface
iface = gr.Interface(
    fn=lambda: f"ARF v4 - Current risk: {risk_engine.get_current_risk().mean:.2f}",
    inputs=[],
    outputs="text",
    title="ARF v4 Demo"
)
app = gr.mount_gradio_app(app, iface, path="/")

# ============== MAIN ENTRY POINT ==============
if __name__ == "__main__":
    # Launch the Gradio interface – this will start the server
    # Hugging Face automatically provides the correct port
    iface.launch(server_name="0.0.0.0")