src/streamlit_app.py

import streamlit as st
import pandas as pd
import requests
import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots
import numpy as np
import io

# 设置页面配置
st.set_page_config(
    page_title="台湾上市公司碳排放数据分析",
    page_icon="🌱",
    layout="wide",
    initial_sidebar_state="expanded"
)

# 缓存数据下载函数
@st.cache_data
def download_and_process_data():
    """下载并处理碳排放数据"""
    try:
        url = "https://mopsfin.twse.com.tw/opendata/t187ap46_O_1.csv"
        response = requests.get(url, timeout=30)
        response.raise_for_status()
        
        # 使用StringIO读取CSV数据
        df = pd.read_csv(io.StringIO(response.content.decode('utf-8-sig')))
        
        # 移除空值
        df = df.dropna()
        
        # 检查关键栏位
        scope1_col = "範疇一排放量(公噸CO2e)"
        scope2_col = "範疇二排放量(公噸CO2e)"
        company_col = "公司代號"
        
        # 寻找公司代号相关栏位
        company_cols = [col for col in df.columns if "公司" in col or "代號" in col or "股票" in col]
        emission_cols = [col for col in df.columns if "排放" in col]
        
        # 自动找到正确的栏位名称
        if company_col not in df.columns and company_cols:
            company_col = company_cols[0]
        
        if scope1_col not in df.columns:
            scope1_candidates = [col for col in emission_cols if "範疇一" in col or "Scope1" in col]
            if scope1_candidates:
                scope1_col = scope1_candidates[0]
        
        if scope2_col not in df.columns:
            scope2_candidates = [col for col in emission_cols if "範疇二" in col or "Scope2" in col]
            if scope2_candidates:
                scope2_col = scope2_candidates[0]
        
        # 确保数值栏位为数字格式
        if scope1_col in df.columns:
            df[scope1_col] = pd.to_numeric(df[scope1_col], errors='coerce')
        if scope2_col in df.columns:
            df[scope2_col] = pd.to_numeric(df[scope2_col], errors='coerce')
        
        # 移除转换后的空值
        required_cols = [col for col in [scope1_col, scope2_col, company_col] if col in df.columns]
        df = df.dropna(subset=required_cols)
        
        return df, company_col, scope1_col, scope2_col, True
        
    except Exception as e:
        st.error(f"数据下载失败: {str(e)}")
        return None, None, None, None, False

# 创建旭日图
def create_sunburst_chart(df, company_col, scope1_col, scope2_col, num_companies):
    """创建旭日图"""
    if not all(col in df.columns for col in [company_col, scope1_col, scope2_col]):
        return None
    
    sunburst_data = []
    df_top = df.nlargest(num_companies, scope1_col)
    
    for _, row in df_top.iterrows():
        company = str(row[company_col])
        scope1 = row[scope1_col]
        scope2 = row[scope2_col]
        
        sunburst_data.extend([
            dict(ids=f"公司-{company}", labels=f"公司 {company}", parents="", values=scope1 + scope2),
            dict(ids=f"範疇一-{company}", labels=f"範疇一: {scope1:.0f}", parents=f"公司-{company}", values=scope1),
            dict(ids=f"範疇二-{company}", labels=f"範疇二: {scope2:.0f}", parents=f"公司-{company}", values=scope2)
        ])
    
    fig = go.Figure(go.Sunburst(
        ids=[d['ids'] for d in sunburst_data],
        labels=[d['labels'] for d in sunburst_data],
        parents=[d['parents'] for d in sunburst_data],
        values=[d['values'] for d in sunburst_data],
        branchvalues="total",
        hovertemplate='<b>%{label}</b><br>排放量: %{value:.0f} 公噸CO2e<extra></extra>',
        maxdepth=3
    ))
    
    fig.update_layout(
        title=f"碳排放量旭日图 (前{num_companies}家公司)",
        font_size=12,
        height=700
    )
    
    return fig

# 创建双层圆饼图
def create_nested_pie_chart(df, company_col, scope1_col, scope2_col, num_companies):
    """创建双层圆饼图"""
    if not all(col in df.columns for col in [company_col, scope1_col, scope2_col]):
        return None
    
    df_top = df.nlargest(num_companies, scope1_col)
    
    fig = make_subplots(
        rows=1, cols=2,
        specs=[[{"type": "pie"}, {"type": "pie"}]],
        subplot_titles=("範疇一排放量", "範疇二排放量")
    )
    
    # 範疇一圆饼图
    fig.add_trace(go.Pie(
        labels=df_top[company_col],
        values=df_top[scope1_col],
        name="範疇一",
        hovertemplate='<b>%{label}</b><br>範疇一排放量: %{value:.0f} 公噸CO2e<br>占比: %{percent}<extra></extra>',
        textinfo='label+percent',
        textposition='auto'
    ), row=1, col=1)
    
    # 範疇二圆饼图
    fig.add_trace(go.Pie(
        labels=df_top[company_col],
        values=df_top[scope2_col],
        name="範疇二",
        hovertemplate='<b>%{label}</b><br>範疇二排放量: %{value:.0f} 公噸CO2e<br>占比: %{percent}<extra></extra>',
        textinfo='label+percent',
        textposition='auto'
    ), row=1, col=2)
    
    fig.update_layout(
        title_text=f"碳排放量圆饼图比较 (前{num_companies}家公司)",
        showlegend=True,
        height=600
    )
    
    return fig

# 创建散点图
def create_scatter_plot(df, company_col, scope1_col, scope2_col):
    """创建散点图"""
    if not all(col in df.columns for col in [company_col, scope1_col, scope2_col]):
        return None
    
    fig = px.scatter(
        df,
        x=scope1_col,
        y=scope2_col,
        hover_data=[company_col],
        title="範疇一 vs 範疇二排放量散点图",
        labels={
            scope1_col: "範疇一排放量 (公噸CO2e)",
            scope2_col: "範疇二排放量 (公噸CO2e)"
        },
        hover_name=company_col
    )
    
    fig.update_layout(height=600)
    return fig

# 创建综合旭日图
def create_comprehensive_sunburst(df, company_col, scope1_col, scope2_col):
    """创建综合旭日图"""
    if not all(col in df.columns for col in [company_col, scope1_col, scope2_col]):
        return None
    
    df_copy = df.copy()
    df_copy['total_emission'] = df_copy[scope1_col] + df_copy[scope2_col]
    df_copy['emission_level'] = pd.cut(df_copy['total_emission'],
                                bins=[0, 1000, 5000, 20000, float('inf')],
                                labels=['低排放(<1K)', '中排放(1K-5K)', '高排放(5K-20K)', '超高排放(>20K)'])
    
    sunburst_data = []
    
    for level in df_copy['emission_level'].unique():
        if pd.isna(level):
            continue
        level_companies = df_copy[df_copy['emission_level'] == level].nlargest(8, 'total_emission')
        
        for _, row in level_companies.iterrows():
            company = str(row[company_col])
            scope1 = row[scope1_col]
            scope2 = row[scope2_col]
            total = scope1 + scope2
            
            sunburst_data.extend([
                dict(ids=str(level), labels=str(level), parents="", values=total),
                dict(ids=f"{level}-{company}", labels=f"{company}", parents=str(level), values=total),
                dict(ids=f"{level}-{company}-範疇一", labels=f"範疇一({scope1:.0f})",
                     parents=f"{level}-{company}", values=scope1),
                dict(ids=f"{level}-{company}-範疇二", labels=f"範疇二({scope2:.0f})",
                     parents=f"{level}-{company}", values=scope2)
            ])
    
    fig = go.Figure(go.Sunburst(
        ids=[d['ids'] for d in sunburst_data],
        labels=[d['labels'] for d in sunburst_data],
        parents=[d['parents'] for d in sunburst_data],
        values=[d['values'] for d in sunburst_data],
        branchvalues="total",
        hovertemplate='<b>%{label}</b><br>排放量: %{value:.0f} 公噸CO2e<extra></extra>',
        maxdepth=4
    ))
    
    fig.update_layout(
        title="分级碳排放量旭日图",
        font_size=10,
        height=700
    )
    
    return fig

# 主应用
def main():
    st.title("🌱 台湾上市公司碳排放数据分析")
    st.markdown("---")
    
    # 侧边栏设置
    st.sidebar.header("📊 图表设置")
    
    # 下载数据
    with st.spinner("正在下载和处理数据..."):
        df, company_col, scope1_col, scope2_col, success = download_and_process_data()
    
    if not success or df is None:
        st.error("无法获取数据，请稍后再试。")
        return
    
    # 显示数据概览
    st.success(f"✅ 成功加载 {len(df)} 家公司的碳排放数据")
    
    # 数据统计信息
    col1, col2, col3, col4 = st.columns(4)
    
    with col1:
        st.metric("总公司数", len(df))
    with col2:
        if scope1_col in df.columns:
            st.metric("範疇一平均排放量", f"{df[scope1_col].mean():.0f} 公噸")
    with col3:
        if scope2_col in df.columns:
            st.metric("範疇二平均排放量", f"{df[scope2_col].mean():.0f} 公噸")
    with col4:
        if all(col in df.columns for col in [scope1_col, scope2_col]):
            total_emission = df[scope1_col].sum() + df[scope2_col].sum()
            st.metric("总排放量", f"{total_emission:.0f} 公噸")
    
    st.markdown("---")
    
    # 侧边栏控制选项
    chart_type = st.sidebar.selectbox(
        "选择图表类型",
        ["旭日图", "双层圆饼图", "散点图", "综合旭日图", "全部图表"]
    )
    
    num_companies = st.sidebar.slider(
        "显示公司数量 (适用于旭日图和圆饼图)",
        min_value=5,
        max_value=min(30, len(df)),
        value=15,
        step=1
    )
    
    # 显示原始数据选项
    if st.sidebar.checkbox("显示原始数据"):
        st.subheader("📋 原始数据预览")
        st.dataframe(df.head(20), use_container_width=True)
        st.markdown("---")
    
    # 根据选择显示图表
    if chart_type == "旭日图" or chart_type == "全部图表":
        st.subheader("🌞 碳排放量旭日图")
        fig1 = create_sunburst_chart(df, company_col, scope1_col, scope2_col, num_companies)
        if fig1:
            st.plotly_chart(fig1, use_container_width=True)
        else:
            st.error("无法创建旭日图，缺少必要数据字段")
    
    if chart_type == "双层圆饼图" or chart_type == "全部图表":
        st.subheader("🥧 双层圆饼图")
        fig2 = create_nested_pie_chart(df, company_col, scope1_col, scope2_col, num_companies)
        if fig2:
            st.plotly_chart(fig2, use_container_width=True)
        else:
            st.error("无法创建圆饼图，缺少必要数据字段")
    
    if chart_type == "散点图" or chart_type == "全部图表":
        st.subheader("📈 範疇一 vs 範疇二散点图")
        fig3 = create_scatter_plot(df, company_col, scope1_col, scope2_col)
        if fig3:
            st.plotly_chart(fig3, use_container_width=True)
        else:
            st.error("无法创建散点图，缺少必要数据字段")
    
    if chart_type == "综合旭日图" or chart_type == "全部图表":
        st.subheader("🎯 分级碳排放量旭日图")
        fig4 = create_comprehensive_sunburst(df, company_col, scope1_col, scope2_col)
        if fig4:
            st.plotly_chart(fig4, use_container_width=True)
        else:
            st.error("无法创建综合旭日图，缺少必要数据字段")
    
    # 详细统计信息
    if st.sidebar.checkbox("显示详细统计"):
        st.subheader("📊 详细统计信息")
        
        if all(col in df.columns for col in [scope1_col, scope2_col]):
            col1, col2 = st.columns(2)
            
            with col1:
                st.write("**範疇一排放量统计:**")
                st.write(f"• 平均值: {df[scope1_col].mean():.2f} 公噸CO2e")
                st.write(f"• 中位数: {df[scope1_col].median():.2f} 公噸CO2e")
                st.write(f"• 最大值: {df[scope1_col].max():.2f} 公噸CO2e")
                st.write(f"• 最小值: {df[scope1_col].min():.2f} 公噸CO2e")
            
            with col2:
                st.write("**範疇二排放量统计:**")
                st.write(f"• 平均值: {df[scope2_col].mean():.2f} 公噸CO2e")
                st.write(f"• 中位数: {df[scope2_col].median():.2f} 公噸CO2e")
                st.write(f"• 最大值: {df[scope2_col].max():.2f} 公噸CO2e")
                st.write(f"• 最小值: {df[scope2_col].min():.2f} 公噸CO2e")
    
    # 页脚
    st.markdown("---")
    st.markdown(
        """
        <div style='text-align: center; color: gray;'>
        <p>数据来源: <a href='https://mopsfin.twse.com.tw/' target='_blank'>台湾证券交易所公开资讯观测站</a></p>
        <p>© 2024 碳排放数据分析应用</p>
        </div>
        """,
        unsafe_allow_html=True
    )

if __name__ == "__main__":
    main()