app.py

import streamlit as st
import pandas as pd
import numpy as np
import plotly.express as px
import plotly.graph_objects as go
from datetime import datetime
import requests
from bs4 import BeautifulSoup
from fpdf import FPDF
import base64

def create_pdf_report(country, sector, forecast, gap, status, news_list):
    pdf = FPDF()
    pdf.add_page()
    pdf.set_font("helvetica", "B", 16)
    
    # Başlık
    pdf.cell(0, 10, text="ClimateVision 2030 - Strategic Report", new_x="LMARGIN", new_y="NEXT", align="C")
    
    # Mevcut veriler (Executive Summary & Compliance)
    pdf.set_font("helvetica", "B", 14)
    pdf.cell(0, 10, text="1. Executive Summary", new_x="LMARGIN", new_y="NEXT")
    pdf.set_font("helvetica", "", 12)
    pdf.multi_cell(0, 8, text=f"Country: {country} | Sector: {sector}\nForecast: {forecast} MtCO2e", new_x="LMARGIN", new_y="NEXT")
    
    pdf.set_font("helvetica", "B", 14)
    pdf.cell(0, 10, text="2. Compliance Audit", new_x="LMARGIN", new_y="NEXT")
    pdf.set_font("helvetica", "", 12)
    pdf.cell(0, 8, text=f"Status: {status}", new_x="LMARGIN", new_y="NEXT")
    pdf.cell(0, 8, text=f"Mitigation Gap: {gap} MtCO2e", new_x="LMARGIN", new_y="NEXT")

    # --- Section 3: Strategic Insights ---
    pdf.ln(5)
    pdf.set_font("helvetica", "B", 14)
    pdf.cell(0, 10, text="3. Strategic Insights & News Alignment", new_x="LMARGIN", new_y="NEXT")
    
    for article in news_list:
        pdf.set_font("helvetica", "B", 11)
        # multi_cell öncesi 'w=0' ve 'new_x/y' ayarlarını netleştiriyoruz
        pdf.multi_cell(0, 8, text=f"Source: {article['source']} - {article['title']}", new_x="LMARGIN", new_y="NEXT")
        
        pdf.set_font("helvetica", "I", 10)
        # Hata buradaydı: w=0 kullanarak tüm genişliği almasını ve satır sonu yapmasını sağlıyoruz
        pdf.multi_cell(0, 6, text=f"Model Insight: {article['model_comment']}", new_x="LMARGIN", new_y="NEXT")
        pdf.ln(3)
        
    return bytes(pdf.output())


# --- CONFIGURATION ---
st.set_page_config(
    page_title="ClimateVision 2030 | Strategic Decision Intelligence",
    page_icon="🌍",
    layout="wide",
    initial_sidebar_state="expanded"
)

# --- CUSTOM UI STYLING (Senior UI/UX) ---
st.markdown("""
    <style>
    .main { background-color: #f8f9fa; }
    .stMetric { background-color: #ffffff; padding: 15px; border-radius: 10px; box-shadow: 0 2px 4px rgba(0,0,0,0.05); }
    .status-badge { padding: 5px 12px; border-radius: 20px; font-weight: bold; font-size: 14px; }
    .paris-compliant { background-color: #d4edda; color: #155724; }
    .risk-alert { background-color: #fff3cd; color: #856404; }
    .non-compliant { background-color: #f8d7da; color: #721c24; }
    </style>
    """, unsafe_allow_html=True)

# --- SIDEBAR NAVIGATION ---
def sidebar_navigation():
    # Yer tutucu yerine kendi profesyonel görselini ekle
    try:
        st.sidebar.image("logo.png", use_container_width=True)
    except:
        # Görsel yüklenemezse şık bir yazı göster (Fallback)
        st.sidebar.title("🌍 CLIMATE VISION 2030")
    
    st.sidebar.markdown("---")
    # ... (Diğer kodlar aynı kalacak)
    
    page = st.sidebar.radio(
        "Strategic Pillars",
        ["🏠 Strategic Overview", 
         "🔮 2030 Projection Engine", 
         "⚖️ Paris GAP Analysis", 
         "🧪 What-If Scenario Lab", 
         "📈 Model X-Ray (XAI)"]
    )
    
    st.sidebar.markdown("---")
    st.sidebar.info("**Asset Note:** Model Status: Production Ready (v1.2.4)")
    st.sidebar.caption(f"Last Intelligence Sync: {datetime.now().strftime('%Y-%m-%d')}")
    
    return page

# --- PLACEHOLDER FUNCTIONS FOR PAGES ---
def show_overview():
    st.title("🏠 Strategic Overview")
    st.subheader("Global Emission Landscape & BAU Momentum")
    st.markdown("""
    *Executive Summary:* This module analyzes historical trajectories (1970-2024) and 
    identifies **Business-as-Usual (BAU)** trends across global economies.
    """)
    # GIS Map and Global KPIs will be here in Step 2.
    st.info("Global Map and KPI metrics loading...")

def show_overview():
    # --- PAGE HEADER ---
    st.title("🏠 Strategic Overview")
    st.markdown("""
    <p style='font-size: 1.2rem; color: #555;'>
    Analyze the <b>Global Atmospheric Load</b> and historical emission trajectories. 
    This module identifies structural trends and the <b>Business-as-Usual (BAU)</b> momentum 
    required for high-level policy auditing.
    </p>
    """, unsafe_allow_html=True)

    # --- TOP LEVEL METRICS (KPIs) ---
    col1, col2, col3, col4 = st.columns(4)
    with col1:
        st.metric(label="Global Emission Load (2024)", value="54.2 GtCO2e", delta="1.2% vs Prev Year")
    with col2:
        st.metric(label="BAU Momentum", value="Increasing", delta="Critical", delta_color="inverse")
    with col3:
        st.metric(label="Decoupling Index", value="0.42", help="Measures the separation of GDP growth from emission growth.")
    with col4:
        st.metric(label="Atmospheric Tipping Point", value="~7 Years", help="Estimated time until 1.5°C carbon budget is exhausted.")

    st.markdown("---")

    # --- GLOBAL GIS MAP (CHOROPLETH) ---
    st.subheader("🌍 Global Emission Intensity & Risk Mapping")
    
    # Mock Data for GIS (Replace with your actual 'df_last_year' data)
    map_data = pd.DataFrame({
        'Country': ['USA', 'CHN', 'IND', 'DEU', 'TUR', 'BRA', 'RUS'],
        'Emission': [5000, 12000, 3000, 700, 500, 1000, 1600],
        'Risk_Score': [75, 90, 65, 40, 55, 30, 80]
    })

    fig_map = px.choropleth(
        map_data,
        locations="Country",
        locationmode='ISO-3',
        color="Emission",
        hover_name="Country",
        hover_data=["Risk_Score"],
        color_continuous_scale=px.colors.sequential.YlOrRd,
        labels={'Emission': 'MtCO2e'}
    )
    
    fig_map.update_layout(
        margin={"r":0,"t":0,"l":0,"b":0},
        geo=dict(showframe=False, showcoastlines=True, projection_type='equirectangular'),
        paper_bgcolor='rgba(0,0,0,0)',
        plot_bgcolor='rgba(0,0,0,0)',
    )
    
    st.plotly_chart(fig_map, use_container_width=True)

    # --- STRATEGIC INSIGHTS SECTION ---
    col_a, col_b = st.columns([1, 1])

    with col_a:
        st.subheader("📈 Macro-Economic Decoupling Analysis")
        st.markdown("""
        The **Decoupling Index** indicates how much a country's economic growth (GDP) has 
        separated from its greenhouse gas emissions. 
        - **Absolute Decoupling:** Emissions fall as GDP rises (Goal).
        - **Relative Decoupling:** Emissions rise slower than GDP.
        """)
        # Placeholder for a Decoupling Chart
        chart_data = pd.DataFrame(np.random.randn(20, 2), columns=['GDP Trend', 'Emission Trend'])
        st.line_chart(chart_data)

    with col_b:
        st.subheader("🚨 Priority Tipping Points")
        st.error("**High Risk Sector:** Power Industry (Decarbonization lag identified)")
        st.warning("**Target Gap:** Global 2030 targets require a 45% reduction in CO2 vs 2010 levels.")
        st.success("**Emerging Opportunity:** Rapid acceleration in Renewables in EU/China.")

    st.markdown("---")
    st.caption("Data Source: EDGAR (Emissions Database for Global Atmospheric Research) v8.0 | Verified by ClimateVision Engine")

def show_projection():
    st.title("🔮 2030 Projection Engine")
    st.subheader("Hybrid Intelligence: Prophet Trend + LSTM Residual Correction")
    # Live filters and Prediction graph will be here in Step 3.

    import joblib # Prophet modelleri için
# from tensorflow.keras.models import load_model # LSTM için (Korumaya alarak yorum satırı yaptım)

def show_projection():
    st.title("🔮 2030 Projection Engine")
    st.markdown("""
    <p style='font-size: 1.1rem;'>
    This engine utilizes <b>Hybrid Intelligence</b>: 
    <b>Prophet</b> for long-term trend decomposition and <b>LSTM (RNN)</b> for non-linear residual correction. 
    Generating high-fidelity atmospheric trajectories for 2030.
    </p>
    """, unsafe_allow_html=True)

    # --- MODEL LOADING (CACHED) ---
    @st.cache_resource
    def load_hybrid_models():
        # Gerçek projende: 
        # prophet_model = joblib.load('models/prophet_v1.pkl')
        # lstm_model = load_model('models/lstm_v1.keras')
        return "Models Loaded Successfully"

    model_status = load_hybrid_models()

    # --- SELECTION BAR ---
    st.markdown("### 🛠️ Configuration & Inference")
    col1, col2, col3 = st.columns([2, 2, 1])
    
    with col1:
        country = st.selectbox("Target Economy (Country/Region)", 
                             ["Global Total", "European Union", "USA", "China", "Turkey", "India"])
    with col2:
        sector = st.selectbox("Economic Sector", 
                            ["All Sectors", "Power Industry", "Transport", "Industrial Combustion", "Buildings", "Agriculture"])
    with col3:
        st.write("") # Boşluk
        predict_btn = st.button("🔥 Generate 2030 Projection", use_container_width=True)

    if predict_btn:
        with st.spinner(f"Inference Mode: Analyzing {country} - {sector} trajectory..."):
            # --- MOCK DATA GENERATION (Gerçek modellerini buraya bağlayacaksın) ---
            years = np.arange(2010, 2031)
            historical_data = np.random.uniform(450, 500, size=15) # 2010-2024
            
            # Prophet Trend
            prophet_trend = np.linspace(500, 540, 6) # 2025-2030
            # LSTM Residual Correction (Hafif dalgalanma ekler)
            lstm_correction = np.random.normal(0, 5, 6)
            hybrid_forecast = prophet_trend + lstm_correction
            
            # Confidence Interval Calculation
            upper_bound = hybrid_forecast * 1.05
            lower_bound = hybrid_forecast * 0.95

            # --- VISUALIZATION (Plotly) ---
            fig = go.Figure()

            # Historical Line
            fig.add_trace(go.Scatter(x=years[:15], y=historical_data, name="Historical Data",
                                     line=dict(color='#2c3e50', width=3)))

            # Confidence Interval (Shadow)
            fig.add_trace(go.Scatter(
                x=years[14:], y=upper_bound, mode='lines', line=dict(width=0), showlegend=False))
            fig.add_trace(go.Scatter(
                x=years[14:], y=lower_bound, mode='lines', line=dict(width=0),
                fill='toself', fillcolor='rgba(46, 204, 113, 0.2)', name="95% Confidence Interval"))

            # Forecast Line
            fig.add_trace(go.Scatter(x=years[14:], y=np.concatenate([[historical_data[-1]], hybrid_forecast]),
                                     name="Hybrid AI Forecast (2030)",
                                     line=dict(color='#2ecc71', width=4, dash='dash')))

            fig.update_layout(
                title=f"Atmospheric Emission Trajectory: {country} ({sector})",
                xaxis_title="Timeline", yaxis_title="MtCO2e",
                hovermode="x unified", template="plotly_white",
                legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1)
            )

            st.plotly_chart(fig, use_container_width=True)

            # --- INSIGHT CARDS ---
            c1, c2, c3 = st.columns(3)
            with c1:
                st.success(f"**2030 Point Estimate:** {hybrid_forecast[-1]:.2f} MtCO2e")
            with c2:
                growth_rate = ((hybrid_forecast[-1] - historical_data[-1]) / historical_data[-1]) * 100
                st.metric("Estimated Growth vs 2024", f"{growth_rate:.1f}%", delta_color="inverse")
            with c3:
                st.warning("**Model Confidence:** 92.4% (Based on Historical Variance)")

    else:
        st.info("Select a country and sector, then click the button to trigger the inference engine.")

    st.markdown("---")
    st.caption("Note: Hybrid models are retrained monthly to incorporate the latest atmospheric readings.")

def show_gap_analysis():
    st.title("⚖️ Paris GAP Analysis")
    st.markdown("""
    **The Audit Layer:** Comparing 2030 Hybrid AI Forecasts against Nationally Determined Contributions (NDCs). 
    This section identifies the *Policy Gap* required to maintain the 1.5°C trajectory.
    """)

    # --- SIMULATED DATA & LOGIC ---
    # Gerçek projede bir önceki sayfadaki 'hybrid_forecast' değerini session_state ile buraya taşıyabilirsin.
    forecast_2030 = 540.0  # Örnek tahmin
    paris_target = 380.0   # 2010 seviyelerine göre %45 azaltım hedefi (Örnek)
    gap = forecast_2030 - paris_target
    gap_percentage = (gap / forecast_2030) * 100

    # --- STATUS BADGES ---
    st.markdown("### 🛡️ Compliance Audit Status")
    if gap <= 0:
        st.markdown('<span class="status-badge paris-compliant">✅ PARIS COMPLIANT</span>', unsafe_allow_html=True)
    elif 0 < gap < 50:
        st.markdown('<span class="status-badge risk-alert">⚠️ AT RISK</span>', unsafe_allow_html=True)
    else:
        st.markdown('<span class="status-badge non-compliant">🚨 NON-COMPLIANT</span>', unsafe_allow_html=True)

    # --- GAUGE CHART & METRICS ---
    col1, col2 = st.columns([1, 1])
    
    with col1:
        fig_gauge = go.Figure(go.Indicator(
            mode = "gauge+number",
            value = gap,
            domain = {'x': [0, 1], 'y': [0, 1]},
            title = {'text': "Reduction Gap (MtCO2e)"},
            gauge = {
                'axis': {'range': [None, 300]},
                'bar': {'color': "#e74c3c"},
                'steps': [
                    {'range': [0, 50], 'color': "#fff3cd"},
                    {'range': [50, 300], 'color': "#f8d7da"}
                ],
                'threshold': {'line': {'color': "black", 'width': 4}, 'thickness': 0.75, 'value': 250}
            }
        ))
        st.plotly_chart(fig_gauge, use_container_width=True)

    with col2:
        st.write("### Strategic Audit Summary")
        st.metric("Total Mitigation Gap", f"{gap:.1f} MtCO2e", f"{gap_percentage:.1f}% Reduction Needed", delta_color="inverse")
        st.info(f"""
        **Insight:** To bridge this gap, the selected economy must accelerate its 
        decarbonization rate by **2.4x** compared to the historical BAU trend.
        """)

def show_what_if_lab():
    st.title("🧪 What-If Scenario Laboratory")
    st.subheader("Policy Intervention Simulation")
    
    # --- SIDEBAR OR TOP PANEL SLIDERS ---
    with st.expander("🛠️ Intervention Control Panel", expanded=True):
        c1, c2, c3 = st.columns(3)
        with c1:
            renewables = st.slider("Renewable Energy Acceleration (%)", 0, 100, 20)
        with c2:
            carbon_tax = st.slider("Carbon Tax Increase ($/ton)", 0, 250, 50)
        with c3:
            tech_leap = st.select_slider("Technological Leap (CCUS)", options=["None", "Low", "Moderate", "Aggressive"])

    # --- SIMULATION LOGIC ---
    # Müdahalelerin tahmini etkisini hesaplayan basit bir fonksiyon
    reduction_impact = (renewables * 0.5) + (carbon_tax * 0.2) + (30 if tech_leap == "Aggressive" else 10)
    base_forecast_2030 = 540.0
    simulated_2030 = base_forecast_2030 - reduction_impact
    
    # Prosperity Index calculation (Logic: Growth vs. Sustainability)
    prosperity_score = (100 - (simulated_2030 / 10)) + (renewables * 0.1)

    # --- COMPARISON CHART ---
    fig_sim = go.Figure()
    fig_sim.add_trace(go.Bar(x=['BAU Forecast', 'Post-Intervention'], 
                             y=[base_forecast_2030, simulated_2030],
                             marker_color=['#95a5a6', '#2ecc71']))
    fig_sim.update_layout(title="Policy Impact Assessment (2030 Projection)")
    
    st.plotly_chart(fig_sim, use_container_width=True)

    # --- GREEN PROSPERITY INDEX ---
    st.markdown("---")
    st.subheader("🍃 Green Prosperity Index (GPI)")
    st.progress(min(max(prosperity_score/100, 0.0), 1.0))
    st.write(f"The simulated policies result in a Prosperity Score of **{prosperity_score:.1f}/100**.")

def show_xai():
    st.title("📈 Model X-Ray (Explainable AI)")
    st.markdown("""
    **Transparency Layer:** This module provides an 'X-Ray' view of our Hybrid Intelligence. 
    By analyzing model residuals and feature dominance, we ensure that every 2030 projection is 
    statistically grounded and explainable.
    """)

    # News Data
    news_items = [
            {
                "title": "EU Tightens Carbon Credit Framework for 2030",
                "summary": "The European Commission announced a stricter framework for carbon credits by 2030 to normalize the Emissions Trading System (ETS).",
                "source": "Reuters",
                "search_query": "Reuters EU Carbon Credit Framework 2030",
                "sentiment": "positive",
                "alignment_score": 92,
                "model_comment": "This policy change aligns 92% with our 'Low Emission' scenario and carbon price surge projections."
            },
            {
                "title": "Global Supply Chain Disruptions Impacting Solar Parts",
                "summary": "Global logistics crises are causing significant delays in solar panel component shipments, affecting renewable targets.",
                "source": "Bloomberg",
                "search_query": "Bloomberg Solar Supply Chain Disruptions 2030",
                "sentiment": "negative",
                "alignment_score": 45,
                "model_comment": "Caution: Supply chain risks may exert downward pressure on our 2030 renewable capacity forecasts."
            }
        ]

    tab1, tab2, tab3 = st.tabs(["🔍 Diagnostic Intelligence", "🧬 Feature Dominance", "📰 Policy News Agent"])

    with tab1:
        st.subheader("Model Röntgeni: Residuals Analysis")
        st.info("Visualizing how the LSTM layer corrected the Prophet baseline residuals.")
        
        # Simulated Residuals Plot
        res_x = np.linspace(0, 100, 100)
        res_y = np.random.normal(0, 2, 100) # Gaussian noise centered at zero
        
        fig_res = px.scatter(x=res_x, y=res_y, labels={'x': 'Inference Timeline', 'y': 'Error Variance (Residuals)'},
                             title="Hybrid Model Residual Distribution", opacity=0.6)
        fig_res.add_hline(y=0, line_dash="dash", line_color="red")
        fig_res.update_traces(marker=dict(color='#34495e'))
        st.plotly_chart(fig_res, use_container_width=True)
        
        st.write("""
        **Strategic Insight:** The residuals are randomly distributed around zero, confirming that 
        the **LSTM residual correction** successfully captured the non-linear variances that 
        Prophet's trend baseline missed.
        """)

    with tab2:
        st.subheader("Inference Drivers: Global Feature Importance")
        
        # Mock Feature Importance (Based on Project Logic)
        importance_data = pd.DataFrame({
            'Feature': ['Historical Momentum', 'Energy Sector Intensity', 'GDP Decoupling Rate', 'CH4 Concentration', 'Land Use Changes'],
            'Impact Score': [0.45, 0.25, 0.15, 0.10, 0.05]
        }).sort_values(by='Impact Score', ascending=True)

        fig_imp = px.bar(importance_data, x='Impact Score', y='Feature', orientation='h',
                         title="Feature Dominance in 2030 Projections",
                         color_discrete_sequence=['#2ecc71'])
        st.plotly_chart(fig_imp, use_container_width=True)
        
        st.write("> **Asset Note:** 'Historical Momentum' remains the primary driver, followed closely by 'Energy Sector Intensity'.")

    with tab3:
        st.subheader("📰 Strategic News Agent (Scraped Intelligence)")
        st.info("This module analyzes real-time policy news to validate our 2030 projections.")

       

        # Bu döngü ve içindekiler MUTLAKA 'with tab3' altında girintili olmalı
        for article in news_items:
            icon = "🟢" if article["sentiment"] == "positive" else "🔴"
            status = "SUPPORTIVE" if article["sentiment"] == "positive" else "RISK FACTOR"
            reliable_link = f"https://www.google.com/search?q={article['search_query'].replace(' ', '+')}"

            with st.expander(f"{icon} {article['source']}: {article['title']}"):
                col1, col2 = st.columns([2, 1])
                with col1:
                    st.write(f"**Summary:** {article['summary']}")
                    st.link_button("Verify Source on Google News", reliable_link)
                with col2:
                    st.metric("Model Alignment", f"{article['alignment_score']}%")
                    st.caption(f"**Status:** {status}")
                
                # Model Insight'ı her haberin içine (expander altına) koyuyoruz
                st.divider()
                st.markdown(f"🔍 **Model Insight:** {article['model_comment']}")


    # --- FINAL REPORTING EXPORT (ACTIVE VERSION) ---
    st.markdown("---")
    st.subheader("📄 Decision Support Report")
    
    # Rapor için gerekli güncel verileri hazırla
    # Not: Gerçek verileri yukarıdaki analizlerden çekebilirsin
    report_data = {
        "country": "Selected Nation",
        "sector": "All Sectors",
        "forecast": 540.25,
        "gap": 160.25,
        "status": "DANGER: NON-COMPLIANT"
    }
    
    st.caption("Strategic reports include 2030 projections, GAP analysis, and explainability audits.")

    # Bu kısmı Tab'ların dışına, en alta koyuyoruz
    pdf_bytes = create_pdf_report(
        report_data["country"], 
        report_data["sector"], 
        report_data["forecast"], 
        report_data["gap"], 
        report_data["status"],
        news_items  # <--- Tab 3'te tanımladığın haber listesini buraya ekledik
    )

    st.download_button(
        label="📥 Download Executive Summary (PDF)",
        data=pdf_bytes,
        file_name=f"ClimateVision_Full_Report_{datetime.now().strftime('%Y%m%d')}.pdf",
        mime="application/pdf",
        width="stretch"
    )

# --- MAIN APP LOGIC ---
def main():
    selected_page = sidebar_navigation()
    
    if selected_page == "🏠 Strategic Overview":
        show_overview()
    elif selected_page == "🔮 2030 Projection Engine":
        show_projection()
    elif selected_page == "⚖️ Paris GAP Analysis":
        show_gap_analysis()
    elif selected_page == "🧪 What-If Scenario Lab":
        show_what_if_lab()
    elif selected_page == "📈 Model X-Ray (XAI)":
        show_xai()

if __name__ == "__main__":
    main()