from langchain_deepseek import ChatDeepSeek
from langgraph.prebuilt import create_react_agent
from langchain_core.messages import HumanMessage, AIMessage

# ==================== 步骤1：模型初始化 ====================
model = ChatDeepSeek(model="deepseek-v4-flash")

# ==================== 步骤2：定义工具 ====================
def get_weather(city: str) -> str:
    """查询指定城市的天气"""
    weather_data = {
        "北京": "北京今日晴，温度22-30°C，空气质量优",
        "上海": "上海今日多云，温度25-32°C，湿度较大",
        "广州": "广州今日雷阵雨，温度26-33°C，出门请带伞",
        "深圳": "深圳今日晴转多云，温度24-31°C",
    }
    return weather_data.get(city, f"抱歉，天气查询工具暂不支持{city}")

def calculate(expression: str) -> str:
    """执行数学计算，支持加减乘除"""
    try:
        result = eval(expression)
        return f"计算结果：{expression} = {result}"
    except Exception as e:
        return f"计算出错：{str(e)}"

import datetime

def get_time_info(query: str = "当前") -> str:
    """获取当前时间信息"""
    now = datetime.datetime.now()
    return f"当前时间：{now.strftime('%Y年%m月%d日 %H:%M:%S')}，星期{['一','二','三','四','五','六','日'][now.weekday()]}"

def currency_convert(amount: float, from_currency: str, to_currency: str = "CNY") -> str:
    """货币转换（示例汇率，实际需接入实时API）"""
    rates = {"USD": 7.25, "EUR": 7.89, "JPY": 0.048, "CNY": 1.0}
    if from_currency not in rates or to_currency not in rates:
        return "不支持的货币类型"
    result = amount * rates[from_currency] / rates[to_currency]
    return f"{amount} {from_currency} = {result:.2f} {to_currency}"

def search_info(query: str) -> str:
    """搜索各类信息（示例数据）"""
    if "Python" in query:
        return "Python是一种广泛使用的高级编程语言，由Guido van Rossum于1991年发布。"
    elif "AI" in query or "人工智能" in query:
        return "人工智能是计算机科学的一个分支，致力于创造能够模拟人类智能的系统。"
    return f"关于'{query}'的搜索结果：暂无详细信息，请尝试更具体的搜索词。"

tools = [get_weather, calculate, get_time_info, currency_convert, search_info]

# ==================== 步骤3：创建 Agent（面向对象封装）============
from typing import List

class MultiFunctionAssistant:
    """多功能智能助手"""

    def __init__(self):
        self.model = model
        self.tools = tools
        self.system_prompt = (
            "你是一个多功能智能助手，可以帮助用户：\n"
            "1. 查询天气\n2. 数学计算\n3. 查询时间\n"
            "4. 货币转换\n5. 搜索信息\n"
            "请根据用户需求选择合适的工具。"
        )
        self.agent = create_react_agent(
            model=self.model,
            tools=self.tools,
            system_prompt=self.system_prompt
        )
        self.messages: List = []  # 消息历史列表

    def chat(self, user_input: str) -> str:
        """与助手对话，返回回答文本"""
        if not user_input.strip():
            return "请输入有效的问题"

        self.messages.append(HumanMessage(content=user_input))
        result = self.agent.invoke({"messages": self.messages})

        # 更新消息列表为完整的交互历史
        self.messages = result["messages"]

        # 提取最后一条AIMessage的内容作为回复
        for msg in reversed(self.messages):
            if isinstance(msg, AIMessage) and msg.content:
                return msg.content

        return "助手暂未生成回复"

    def reset(self):
        """重置对话历史"""
        self.messages = []
        print("对话已重置")

# ==================== 步骤4：测试 ====================
def main():
    assistant = MultiFunctionAssistant()

    # 方式一：预设问题批量测试
    test_questions = [
        "北京天气怎么样？",
        "帮我计算一下 (45 + 55) * 2.5",
        "现在几点了？",
        "100美元能兑换多少人民币？",
    ]
    for q in test_questions:
        print(f"\n👤 用户: {q}")
        reply = assistant.chat(q)
        print(f"🤖 助手: {reply}")

    # 重置对话
    assistant.reset()

    # 方式二：交互式循环
    print("\n===== 交互模式（输入 'exit' 退出，'reset' 重置）=====")
    while True:
        user_input = input("\n👤 请输入问题: ").strip()
        if user_input.lower() == "exit":
            print("再见！")
            break
        elif user_input.lower() == "reset":
            assistant.reset()
            continue
        reply = assistant.chat(user_input)
        print(f"🤖 助手: {reply}")

if __name__ == "__main__":
    main()