app.py

import gradio as gr
from fastapi import FastAPI, Request
import uvicorn
from sentence_transformers import SentenceTransformer, util
import torch
import requests
import threading

# 🔐 Supabase setup
SUPABASE_URL = "https://olbjfxlclotxtnpjvpfj.supabase.co"
SUPABASE_KEY = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJzdXBhYmFzZSIsInJlZiI6Im9sYmpmeGxjbG90eHRucGp2cGZqIiwicm9sZSI6ImFub24iLCJpYXQiOjE3NTIyMzYwMDEsImV4cCI6MjA2NzgxMjAwMX0.7q_o5DCFEAAysnWXMChH4MI5qNhIVc4OgpT5JvgYxc0"

model = SentenceTransformer('all-MiniLM-L6-v2')

def get_faq_from_supabase(uid):
    url = f"{SUPABASE_URL}/rest/v1/faq_texts?uid=eq.{uid}"
    headers = {
        "apikey": SUPABASE_KEY,
        "Authorization": f"Bearer {SUPABASE_KEY}",
        "Content-Type": "application/json"
    }
    try:
        r = requests.get(url, headers=headers)
        r.raise_for_status()
        data = r.json()
        return [{"q": d["question"], "a": d["answer"]} for d in data]
    except Exception as e:
        print("❌ Supabase error:", e)
        return []

def chatbot(uid, question):
    faqs = get_faq_from_supabase(uid)
    if not faqs:
        return "Tidak ada data FAQ."

    questions = [f["q"] for f in faqs]
    answers = [f["a"] for f in faqs]

    embeddings = model.encode(questions, convert_to_tensor=True)
    query_embedding = model.encode(question, convert_to_tensor=True)

    scores = util.pytorch_cos_sim(query_embedding, embeddings)
    best_idx = torch.argmax(scores).item()
    return answers[best_idx]

# 🌐 FastAPI app
app = FastAPI()

@app.post("/predict")
async def predict(request: Request):
    payload = await request.json()
    uid, question = payload.get("data", [None, None])
    if not uid or not question:
        return {"error": "UID dan pertanyaan diperlukan."}
    
    answer = chatbot(uid, question)
    return {"data": [answer]}

# 🌈 Gradio UI (opsional)
def launch_gradio():
    demo = gr.Interface(
        fn=chatbot,
        inputs=["text", "text"],
        outputs="text",
        title="Biruu Chatbot",
        examples=[["uid123", "Apakah bisa bayar di tempat?"]],
        allow_flagging="never"
    )
    demo.launch(share=True)

# ▶️ Jalankan Gradio di thread terpisah agar tidak bentrok dengan FastAPI
threading.Thread(target=launch_gradio).start()

if _name_ == "_main_":
    uvicorn.run("app:app", host="0.0.0.0", port=7860)