app.py

import os
import json
from typing import Any, Dict

import requests
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
import dotenv
import gradio as gr

dotenv.load_dotenv()

# ======================================================
# ENVIRONMENT
# ======================================================
HF_TOKEN = os.getenv("HF_TOKEN")
AUTH_COOKIE_TOKEN = os.getenv("AUTH_COOKIE_TOKEN")
BASE_API = os.getenv("BASE_API", "https://catering-lac.vercel.app")

MODEL_NAME = "meta-llama/Llama-2-7b-chat-hf"
HF_API_URL = f"https://router.huggingface.co/hf-inference/models/{MODEL_NAME}"

print("===== ENV CHECK =====")
print("HF_TOKEN exists:", HF_TOKEN is not None)
print("AUTH_COOKIE_TOKEN exists:", AUTH_COOKIE_TOKEN is not None)
print("BASE_API:", BASE_API)
print("MODEL_NAME:", MODEL_NAME)
print("HF_API_URL:", HF_API_URL)
print("======================")


# ======================================================
# BACKEND HELPERS
# ======================================================
def backend_headers() -> Dict[str, str]:
    return {
        "Content-Type": "application/json",
        "Cookie": f"auth-token={AUTH_COOKIE_TOKEN}"
    }

def http_get(url: str):
    res = requests.get(url, headers=backend_headers(), timeout=30)
    if res.status_code >= 400:
        raise RuntimeError(f"GET FAILED {res.status_code}: {res.text}")
    return res.json()


# ======================================================
# LLM CALL
# ======================================================
SYSTEM_PROMPT = """
You generate catering recommendation plans.

Rules:
- Stay under or close to budget.
- Use ONLY menu items provided.
- Output VALID JSON only.
"""

def call_llm(quote, menu_items, categories):
    payload = {
        "quote_request": quote,
        "menu_items": menu_items,
        "categories": categories
    }

    user_prompt = json.dumps(payload, indent=2)

    body = {
        "inputs": f"{SYSTEM_PROMPT}\n\nUSER:\n{user_prompt}",
        "parameters": {
            "temperature": 0.2,
            "max_new_tokens": 700
        }
    }

    headers = {
        "Authorization": f"Bearer {HF_TOKEN}",
        "Content-Type": "application/json"
    }

    response = requests.post(HF_API_URL, headers=headers, json=body, timeout=90)

    if response.status_code == 404:
        raise RuntimeError("Model not available on router")

    data = response.json()

    if isinstance(data, list):
        txt = data[0].get("generated_text", "").strip()
    else:
        txt = data.get("generated_text", "").strip()

    if txt.startswith("```"):
        txt = txt.replace("```json", "").replace("```", "").strip()

    return json.loads(txt)


# ======================================================
# FASTAPI BACKEND
# ======================================================
app = FastAPI(title="Catering AI Generator")

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"]
)

class GenerateRequest(BaseModel):
    quote_id: str

class GenerateResponse(BaseModel):
    quote_id: str
    plan: Dict[str, Any]
    persisted: bool


@app.post("/generate-plan", response_model=GenerateResponse)
def generate_plan_api(body: GenerateRequest):
    quote = http_get(f"{BASE_API}/api/admin/catering-requests/{body.quote_id}")
    menu_items = http_get(f"{BASE_API}/api/menu/items")
    categories = http_get(f"{BASE_API}/api/menu/categories")

    plan = call_llm(quote, menu_items, categories)

    return GenerateResponse(
        quote_id=body.quote_id,
        plan=plan,
        persisted=False
    )


# ======================================================
# GRADIO UI (for HuggingFace)
# ======================================================
def gradio_generate(quote_id):
    try:
        quote = http_get(f"{BASE_API}/api/admin/catering-requests/{quote_id}")
        menu_items = http_get(f"{BASE_API}/api/menu/items")
        categories = http_get(f"{BASE_API}/api/menu/categories")

        plan = call_llm(quote, menu_items, categories)
        return plan

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


ui = gr.Interface(
    fn=gradio_generate,
    inputs=gr.Textbox(label="Enter Quote ID"),
    outputs="json",
    title="Catering AI Generator (Llama-2-7B)",
    description="Generates a catering plan using AI."
)


# HuggingFace Spaces runs Gradio by default
def main():
    ui.launch(server_name="0.0.0.0", server_port=7860)


if __name__ == "__main__":
    main()