import gradio as gr
import yfinance as yf
import plotly.graph_objects as go
import pandas as pd
import numpy as np
from sklearn.linear_model import LinearRegression

def forecast_crypto(coin, currency, window_days):
    # 1. Ambil Data Kripto Dasar (Selalu dalam USD)
    base_ticker = f"{coin}-USD"
    data = yf.download(base_ticker, period="1y", interval="1d")
    
    if data.empty:
        return None, f"⚠️ Gagal mengambil data untuk {base_ticker}."
    
    if isinstance(data.columns, pd.MultiIndex):
        data.columns = data.columns.get_level_values(0)
        
    # 2. DATA ENGINEERING: Konversi Mata Uang Historis
    if currency != "USD":
        fx_ticker = f"USD{currency}=X"
        fx_data = yf.download(fx_ticker, period="1y", interval="1d")
        
        if fx_data.empty:
            return None, f"⚠️ Gagal mengambil data kurs {fx_ticker}."
            
        if isinstance(fx_data.columns, pd.MultiIndex):
            fx_data.columns = fx_data.columns.get_level_values(0)
            
        # Pasar kripto buka 24/7, tapi pasar Forex tutup saat akhir pekan.
        # Kita harus menyelaraskan (reindex) dan mengisi hari yang kosong (ffill/bfill)
        fx_close = fx_data['Close'].reindex(data.index).ffill().bfill()
        
        # Kalikan semua harga USD dengan kurs lokal di hari tersebut
        for col in ['Open', 'High', 'Low', 'Close']:
            if col in data.columns:
                data[col] = data[col] * fx_close

    # 3. Analisis Tren Historis
    data['Trend (SMA)'] = data['Close'].rolling(window=int(window_days)).mean()
    
    # 4. MACHINE LEARNING: Prediksi Harga Besok
    # Ambil tren X hari terakhir untuk dipelajari mesin
    df_pred = data[['Close']].dropna().tail(int(window_days))
    
    if len(df_pred) < 3:
        return None, "⚠️ Data tidak cukup untuk melakukan prediksi."

    # Siapkan data untuk model ML
    X = np.arange(len(df_pred)).reshape(-1, 1)
    y = df_pred['Close'].values
    
    model = LinearRegression()
    model.fit(X, y)
    
    # Prediksi 1 hari ke depan
    next_day_X = np.array([[len(df_pred)]])
    predicted_price = model.predict(next_day_X)[0]
    
    # 5. Visualisasi Interaktif Plotly
    fig = go.Figure()
    
    # Garis Harga Asli
    fig.add_trace(go.Scatter(x=data.index, y=data['Close'], 
                             mode='lines', name='Harga Aktual',
                             line=dict(color='#2980b9', width=2)))
    
    # Garis Tren (SMA)
    fig.add_trace(go.Scatter(x=data.index, y=data['Trend (SMA)'], 
                             mode='lines', name=f'Garis Tren ({window_days} Hari)',
                             line=dict(color='#e74c3c', width=2, dash='dash')))
    
    fig.update_layout(
        title=f"📈 AI Market Forecasting: {coin} to {currency}",
        xaxis_title="Tanggal",
        yaxis_title=f"Harga ({currency})",
        template="plotly_white",
        hovermode="x unified",
        legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1)
    )
    
    # 6. Ringkasan Eksekutif & Hasil Prediksi
    last_price = float(data['Close'].iloc[-1])
    current_trend = float(data['Trend (SMA)'].iloc[-1])
    
    status = "🟢 Naik (Bullish)" if last_price > current_trend else "🔴 Turun (Bearish)"
    
    # Format mata uang
    simbol = "Rp" if currency == "IDR" else ("$" if currency == "USD" else ("€" if currency == "EUR" else "£"))
    
    summary = f"""
    ### 📊 Market & Prediction Summary ({coin}/{currency})
    * **Harga Penutupan Terakhir:** **{simbol} {last_price:,.2f}**
    * **Status Tren Pasar:** **{status}**
    * 🤖 **AI Prediksi Harga Besok:** **{simbol} {predicted_price:,.2f}**
    
    *(Catatan: Konversi mata uang dilakukan secara real-time menggunakan data pasar valuta asing. Prediksi dihitung menggunakan Machine Learning berdasarkan momentum {window_days} hari terakhir).*
    """
    
    return fig, summary

# --- UI GRADIO ---
with gr.Blocks(theme=gr.themes.Soft(), title="CryptoProphet AI") as app:
    gr.Markdown("# 🤖 CryptoProphet: AI Market Forecasting")
    gr.Markdown("Sistem analitik prediktif yang menarik data *real-time* via API, melakukan konversi kurs otomatis, dan menggunakan **Machine Learning** untuk meramalkan harga kripto di hari berikutnya.")
    
    with gr.Row():
        with gr.Column(scale=1):
            in_coin = gr.Dropdown(choices=["BTC", "ETH", "SOL", "BNB", "DOGE", "XRP", "ADA"], 
                                  value="BTC", label="Pilih Koin Kripto")
            in_currency = gr.Dropdown(choices=["USD", "IDR", "EUR", "GBP"], 
                                      value="IDR", label="Pilih Mata Uang (Fiat)")
            in_days = gr.Slider(minimum=7, maximum=90, value=30, step=1, label="Jendela Prediksi ML (Hari)")
            btn = gr.Button("🔄 Jalankan Prediksi AI", variant="primary")
        
        with gr.Column(scale=2):
            out_summary = gr.Markdown("### 📊 Memuat data prediksi AI...")
            
    with gr.Row():
        out_plot = gr.Plot()
        
    btn.click(forecast_crypto, inputs=[in_coin, in_currency, in_days], outputs=[out_plot, out_summary])
    app.load(forecast_crypto, inputs=[in_coin, in_currency, in_days], outputs=[out_plot, out_summary])

app.launch()