"""
CardioAI 语音助手
提供语音交互式心血管疾病风险评估
"""

import streamlit as st
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.metrics import roc_auc_score
import plotly.express as px
import plotly.graph_objects as go


# ============================================
# 数据加载函数 (带缓存)
# ============================================
@st.cache_data
def load_data(file_path: str) -> pd.DataFrame:
    """加载 Excel 数据文件"""
    df = pd.read_excel(file_path)
    return df


# ============================================
# 数据预处理函数
# ============================================
@st.cache_data
def preprocess_data(df: pd.DataFrame) -> tuple:
    """
    数据预处理

    Returns:
        (model, scaler, feature_columns)
    """
    df_clean = df.copy()

    # 年龄转换
    df_clean['age_years'] = (df_clean['age'] / 365).round().astype(int)

    # 计算 BMI
    df_clean['bmi'] = df_clean['weight'] / ((df_clean['height'] / 100) ** 2)

    # 删除异常值
    df_clean = df_clean[df_clean['ap_hi'] > df_clean['ap_lo']]
    df_clean = df_clean[(df_clean['ap_hi'] >= 90) & (df_clean['ap_hi'] <= 250)]
    df_clean = df_clean[(df_clean['ap_lo'] >= 60) & (df_clean['ap_lo'] <= 150)]

    # 特征工程
    df_clean['bp_diff'] = df_clean['ap_hi'] - df_clean['ap_lo']
    df_clean['map'] = df_clean['ap_lo'] + (df_clean['bp_diff'] / 3)

    # 特征列
    feature_columns = [
        'age_years', 'gender', 'ap_hi', 'ap_lo',
        'cholesterol', 'gluc', 'bmi', 'smoke', 'alco', 'active'
    ]

    X = df_clean[feature_columns]
    y = df_clean['cardio']

    # 分割数据
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.2, random_state=42, stratify=y
    )

    # 标准化
    scaler = StandardScaler()
    X_train_scaled = scaler.fit_transform(X_train)
    X_test_scaled = scaler.transform(X_test)

    # 训练模型
    model = GradientBoostingClassifier(
        n_estimators=100,
        max_depth=5,
        learning_rate=0.1,
        random_state=42
    )
    model.fit(X_train_scaled, y_train)

    # 验证模型
    y_prob = model.predict_proba(X_test_scaled)[:, 1]
    roc_auc = roc_auc_score(y_test, y_prob)

    return model, scaler, feature_columns, roc_auc


# ============================================
# 预测函数
# ============================================
def predict_risk(model, scaler, input_data, feature_columns):
    """预测心血管疾病风险"""
    input_df = pd.DataFrame([input_data])[feature_columns]
    input_scaled = scaler.transform(input_df)
    probability = model.predict_proba(input_scaled)[0]
    return probability


# ============================================
# 语音指令处理
# ============================================
def process_voice_command(command: str, patient_data: dict, model, scaler, feature_columns) -> str:
    """
    处理语音指令

    Args:
        command: 用户语音指令
        patient_data: 患者数据
        model: 训练好的模型
        scaler: 标准化器
        feature_columns: 特征列

    Returns:
        响应文本
    """
    command = command.lower()

    # 预测风险
    probability = predict_risk(model, scaler, patient_data, feature_columns)
    risk_percent = probability[1] * 100
    no_risk_percent = probability[0] * 100

    if "风险" in command or "患病" in command or "probability" in command:
        if risk_percent > 50:
            return f"根据您提供的信息，您患有心血管疾病的风险为 {risk_percent:.1f}%，风险较高，建议尽快就医检查。"
        else:
            return f"根据您提供的信息，您患有心血管疾病的风险为 {risk_percent:.1f}%，风险较低，请继续保持健康的生活方式。"

    elif "正常" in command or "健康" in command or "healthy" in command:
        return f"您的健康概率为 {no_risk_percent:.1f}%，患病风险为 {risk_percent:.1f}%。"

    elif "年龄" in command or "age" in command:
        return f"您的年龄是 {patient_data['age_years']} 岁。"

    elif "血压" in command or "blood pressure" in command:
        return f"您的血压信息：收缩压 {patient_data['ap_hi']} mmHg，舒张压 {patient_data['ap_lo']} mmHg。"

    elif "bmi" in command or "体重" in command:
        return f"您的 BMI 为 {patient_data['bmi']:.1f}。"

    elif "胆固醇" in command or "cholesterol" in command:
        chol_map = {1: "正常", 2: "高于正常", 3: "远高于正常"}
        return f"您的胆固醇水平：{chol_map.get(patient_data['cholesterol'], '未知')}。"

    elif "血糖" in command or "glucose" in command:
        gluc_map = {1: "正常", 2: "高于正常", 3: "远高于正常"}
        return f"您的血糖水平：{gluc_map.get(patient_data['gluc'], '未知')}。"

    elif "建议" in command or "advice" in command or "recommend" in command:
        if risk_percent > 60:
            return "建议您：1. 立即就医进行全面检查 2. 控制饮食，减少高盐高脂食物 3. 适度运动 4. 戒烟限酒 5. 保持规律作息。"
        elif risk_percent > 30:
            return "建议您：1. 定期体检监测 2. 保持健康饮食 3. 坚持适度运动 4. 控制体重 5. 避免吸烟饮酒。"
        else:
            return "建议您：1. 保持当前健康生活方式 2. 均衡饮食 3. 适量运动 4. 定期体检。"

    elif "帮助" in command or "help" in command:
        return """您可以询问以下信息：
- 患病风险
- 血压情况
- BMI 数值
- 胆固醇水平
- 血糖水平
- 健康建议
- 我的年龄

请告诉我您想了解什么？"""

    else:
        return f"抱歉，我没有理解您的指令。您可以问我关于患病风险、血压、BMI等问题。或者直接点击「开始语音评估」按钮进行心血管疾病风险评估。"


# ============================================
# Streamlit 页面配置
# ============================================
st.set_page_config(
    page_title="CardioAI 语音助手",
    page_icon="🎤",
    layout="wide"
)


# ============================================
# 初始化会话状态
# ============================================
if 'patient_data' not in st.session_state:
    st.session_state.patient_data = None
if 'voice_history' not in st.session_state:
    st.session_state.voice_history = []
if 'model_trained' not in st.session_state:
    st.session_state.model_trained = False


# ============================================
# 主程序
# ============================================
def main():
    """主程序入口"""

    # 页面标题
    st.title("🎤 CardioAI 智能语音助手")
    st.markdown("---")
    st.markdown("**您的私人心血管健康顾问**")

    # 数据路径
    DATA_PATH = "C:/Users/SAM/Desktop/sam_test/ai_code/aicodes/data/心血管疾病.xlsx"

    # 加载和训练模型
    try:
        df = load_data(DATA_PATH)
        model, scaler, feature_columns, roc_auc = preprocess_data(df)
        st.session_state.model_trained = True
        st.session_state.model = model
        st.session_state.scaler = scaler
        st.session_state.feature_columns = feature_columns
    except Exception as e:
        st.error(f"❌ 系统初始化失败: {e}")
        return

    # ============================================
    # 布局：侧边栏 + 主内容
    # ============================================
    st.sidebar.header("👤 患者信息录入")

    # 患者信息表单
    with st.sidebar.form("patient_form"):
        st.subheader("基本信息")

        age = st.number_input("年龄 (岁)", 20, 100, 50)
        gender = st.selectbox("性别", ["女性", "男性"])

        st.subheader("身体指标")

        ap_hi = st.number_input("收缩压 (mmHg)", 90, 250, 120)
        ap_lo = st.number_input("舒张压 (mmHg)", 60, 150, 80)
        height = st.number_input("身高 (cm)", 100, 220, 170)
        weight = st.number_input("体重 (kg)", 30, 200, 70)

        st.subheader("生化指标")

        cholesterol = st.selectbox("胆固醇水平", ["正常", "高于正常", "远高于正常"])
        gluc = st.selectbox("血糖水平", ["正常", "高于正常", "远高于正常"])

        st.subheader("生活方式")

        smoke = st.checkbox("吸烟")
        alco = st.checkbox("饮酒")
        active = st.checkbox("运动")

        submit = st.form_submit_button("确认信息", type="primary")

    if submit:
        # 构建患者数据
        gender_val = 1 if gender == "女性" else 2
        cholesterol_map = {"正常": 1, "高于正常": 2, "远高于正常": 3}
        gluc_map = {"正常": 1, "高于正常": 2, "远高于正常": 3}

        bmi = weight / ((height / 100) ** 2)

        st.session_state.patient_data = {
            'age_years': age,
            'gender': gender_val,
            'ap_hi': ap_hi,
            'ap_lo': ap_lo,
            'cholesterol': cholesterol_map[cholesterol],
            'gluc': gluc_map[gluc],
            'bmi': bmi,
            'smoke': 1 if smoke else 0,
            'alco': 1 if alco else 0,
            'active': 1 if active else 0
        }

        # 初始风险评估
        probability = predict_risk(
            st.session_state.model,
            st.session_state.scaler,
            st.session_state.patient_data,
            st.session_state.feature_columns
        )

        risk_msg = f"您好！我是 CardioAI 语音助手。根据您提供的信息，您的心血管疾病风险评估如下：患病风险为 {probability[1]*100:.1f}%，健康概率为 {probability[0]*100:.1f}%。您可以询问我任何关于您心血管健康的问题。"
        st.session_state.voice_history = [
            {"role": "assistant", "content": risk_msg}
        ]

        st.success("✅ 患者信息已录入，可以开始语音交互了！")

    # ============================================
    # 主内容区域
    # ============================================

    # 欢迎信息和风险评估展示
    st.markdown("### 🎯 心血管疾病风险评估")

    if st.session_state.patient_data:
        # 预测风险
        probability = predict_risk(
            st.session_state.model,
            st.session_state.scaler,
            st.session_state.patient_data,
            st.session_state.feature_columns
        )

        risk_percent = probability[1] * 100

        # 风险仪表盘
        col1, col2 = st.columns([1, 2])

        with col1:
            # 风险仪表图
            fig_gauge = go.Figure(go.Indicator(
                mode="gauge+number",
                value=risk_percent,
                domain={'x': [0, 1], 'y': [0, 1]},
                title={'text': "心血管疾病风险率"},
                gauge={
                    'axis': {'range': [0, 100]},
                    'bar': {'color': "#e74c3c" if risk_percent > 50 else "#2ecc71"},
                    'steps': [
                        {'range': [0, 30], 'color': "#2ecc71"},
                        {'range': [30, 60], 'color': "#f1c40f"},
                        {'range': [60, 100], 'color': "#e74c3c"}
                    ],
                    'threshold': {
                        'line': {'color': "black", 'width': 4},
                        'thickness': 0.75,
                        'value': risk_percent
                    }
                }
            ))
            fig_gauge.update_layout(height=200)
            st.plotly_chart(fig_gauge, use_container_width=True)

        with col2:
            # 风险等级和建议
            if risk_percent < 30:
                st.success("🟢 **低风险** - 请继续保持健康的生活方式")
                st.info("建议：均衡饮食、适量运动、定期体检")
            elif risk_percent < 60:
                st.warning("🟡 **中风险** - 建议关注心血管健康")
                st.info("建议：控制饮食、适度运动、定期监测血压")
            else:
                st.error("🔴 **高风险** - 建议及时就医")
                st.info("建议：尽快就医、严格控制饮食、戒烟限酒")

        st.markdown(f"**模型 ROC-AUC**: {roc_auc:.4f}")
    else:
        st.info("👈 请在左侧填写患者信息后开始评估")

    st.markdown("---")

    # ============================================
    # 语音交互区域
    # ============================================
    st.markdown("### 🎤 语音交互")

    # 快速指令按钮
    st.markdown("#### 快速指令")
    quick_commands = [
        ("查询风险", "我的患病风险是多少？"),
        ("查询血压", "我的血压情况如何？"),
        ("查询BMI", "我的BMI是多少？"),
        ("健康建议", "请给我一些健康建议"),
        ("帮助", "帮助")
    ]

    cols = st.columns(5)
    for i, (label, cmd) in enumerate(quick_commands):
        if cols[i].button(label):
            if st.session_state.patient_data:
                response = process_voice_command(
                    cmd,
                    st.session_state.patient_data,
                    st.session_state.model,
                    st.session_state.scaler,
                    st.session_state.feature_columns
                )
                st.session_state.voice_history.append(
                    {"role": "user", "content": cmd}
                )
                st.session_state.voice_history.append(
                    {"role": "assistant", "content": response}
                )
            else:
                st.warning("请先录入患者信息")

    st.markdown("---")

    # 自定义语音输入
    st.markdown("#### 自定义语音输入")

    # 模拟语音输入 (实际项目中可接入语音识别API)
    voice_input = st.text_input(
        "输入您的问题：",
        placeholder="例如：我的胆固醇水平如何？",
        key="voice_input"
    )

    if st.button("发送", type="primary") and voice_input:
        if st.session_state.patient_data:
            response = process_voice_command(
                voice_input,
                st.session_state.patient_data,
                st.session_state.model,
                st.session_state.scaler,
                st.session_state.feature_columns
            )
            st.session_state.voice_history.append(
                {"role": "user", "content": voice_input}
            )
            st.session_state.voice_history.append(
                {"role": "assistant", "content": response}
            )
        else:
            st.warning("请先录入患者信息")

    # ============================================
    # 对话历史
    # ============================================
    st.markdown("---")
    st.markdown("### 💬 对话历史")

    for msg in st.session_state.voice_history:
        if msg["role"] == "user":
            st.markdown(f"**👤 您**: {msg['content']}")
        else:
            st.markdown(f"**🤖 CardioAI**: {msg['content']}")
        st.markdown("")

    # 清空对话
    if st.button("清空对话历史"):
        st.session_state.voice_history = []
        st.rerun()

    st.markdown("---")

    # ============================================
    # 使用说明
    # ============================================
    with st.expander("📖 使用说明"):
        st.markdown("""
        ### CardioAI 语音助手使用指南

        1. **录入患者信息**: 在左侧边栏填写患者的基本信息、身体指标、生化指标和生活方式。

        2. **开始评估**: 点击「确认信息」按钮后，系统会自动进行心血管疾病风险评估。

        3. **语音交互**: 您可以：
           - 使用快速指令按钮快速查询
           - 输入自定义问题

        4. **支持的查询**:
           - 患病风险和健康概率
           - 血压、BMI、胆固醇、血糖等指标
           - 健康建议

        ### 风险等级说明
        - 🟢 **低风险** (0-30%): 继续保持健康生活方式
        - 🟡 **中风险** (30-60%): 建议关注并改善生活方式
        - 🔴 **高风险** (>60%): 建议及时就医检查
        """)


if __name__ == "__main__":
    main()