app.py

from fastapi import FastAPI, Request
import os
import requests
from pydantic import BaseModel
from dotenv import load_dotenv
import openai

# ✅ Load environment variables (from Hugging Face secrets)
load_dotenv()

# ✅ Initialize FastAPI app
app = FastAPI(title="AI Feedback Engine")

# ✅ Read secrets from environment variables
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
PULSE_API_URL = os.getenv("PULSE_API_URL")
PULSE_API_KEY = os.getenv("PULSE_API_KEY")

# ✅ Configure OpenAI
openai.api_key = OPENAI_API_KEY

# ✅ Pydantic model for chatbot message
class Message(BaseModel):
    text: str

@app.get("/")
def home():
    return {"message": "🚀 AI Feedback Engine is running!"}

@app.post("/auto_feedback")
async def auto_feedback(msg: Message):
    try:
        user_input = msg.text

        # Step 1️⃣: Generate AI feedback + recommendation
        ai_prompt = f"""
        You are an HR feedback assistant.
        A user said: "{user_input}"
        Generate:
        1. A short, professional feedback (1–2 sentences)
        2. A practical recommendation for improvement.
        Return as JSON with keys: 'feedback' and 'recommendation'.
        """

        completion = openai.ChatCompletion.create(
            model="gpt-3.5-turbo",
            messages=[{"role": "system", "content": ai_prompt}]
        )

        ai_text = completion.choices[0].message["content"]

        # Step 2️⃣: Send to Pulse Survey API
        pulse_response = requests.post(
            f"{PULSE_API_URL}/pulse-survey-answers/store",
            headers={"Authorization": f"Bearer {PULSE_API_KEY}"},
            json={"question": user_input, "answer": ai_text},
            timeout=10
        )

        # Step 3️⃣: Return structured result to chatbot
        return {
            "status": "success",
            "user_input": user_input,
            "ai_response": ai_text,
            "pulse_status": pulse_response.status_code,
        }

    except Exception as e:
        return {"status": "error", "message": str(e)}

# ✅ This part ensures it runs locally too (optional)
if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=7860)