import gradio as gr
import torch
import numpy as np
from chronos import Chronos2Pipeline

# --- MODELİ YÜKLE ---
print("🚀 Chronos-2 Modeli Yükleniyor...")
try:
    pipeline = Chronos2Pipeline.from_pretrained(
        "amazon/chronos-2",
        device_map="cpu",
        dtype=torch.float32,
    )
    print("✅ Model Başarıyla Yüklendi!")
except Exception as e:
    print(f"❌ Model Yükleme Hatası: {e}")
    pipeline = None

def predict(context_str, prediction_length):
    if pipeline is None: return "Error: Model yüklenemedi."

    try:
        clean_s = context_str.strip()
        if not clean_s: return "Error: Veri boş."
        data_list = [float(x) for x in clean_s.split(',')]
        
        # Tensor Oluştur
        context_tensor = torch.tensor(data_list).unsqueeze(0).unsqueeze(0)
        
        # Tahmin Yap
        forecast = pipeline.predict(context_tensor, int(prediction_length))
        
        # --- ÇIKTI FORMATI: Medyan | Alt Sınır | Üst Sınır ---
        # 0.5: Medyan (Beklenen)
        # 0.1: Alt Sınır (Kötü Senaryo)
        # 0.9: Üst Sınır (İyi Senaryo)
        
        median_price = forecast[0].quantile(0.5).item()
        lower_price = forecast[0].quantile(0.1).item()
        upper_price = forecast[0].quantile(0.9).item()
        
        return f"{median_price}|{lower_price}|{upper_price}"

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

iface = gr.Interface(fn=predict, inputs=["text", "number"], outputs="text")
iface.launch()