import requests
import re
import json
import pandas as pd
import time

# 1. 配置请求头（模拟浏览器，避免被反爬）
headers = {
    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
}

# 2. 获取当前台风列表
def get_current_typhoon_list():
    t = int(round(time.time() * 1000))  # 13位时间戳（网页动态加载需要）
    url = f"http://typhoon.nmc.cn/weatherservice/typhoon/jsons/list_default?t={t}&callback=typhoon_jsons_list_default"
    response = requests.get(url, headers=headers, verify=False)
    # 解析JSON数据（处理网页回调格式）
    data_str = re.match(".*?({.*}).*", response.text, re.S).group(1)
    data = json.loads(data_str)
    typhoon_list = data["typhoonList"]
    # 筛选当前活跃的台风（state为start）
    active_typhoon = []
    for typhoon in typhoon_list:
        if typhoon[7] == "start":  # state字段，start表示活跃，stop表示结束
            item = {
                "台风ID": typhoon[0],
                "英文名": typhoon[1],
                "中文名": typhoon[2],
                "编号": typhoon[4],
                "名字含义": typhoon[6]
            }
            active_typhoon.append(item)
    return active_typhoon

# 3. 获取单个台风的实时动态和路径数据
def get_typhoon_detail(typhoon_id):
    t = int(round(time.time() * 1000))
    url = f"http://typhoon.nmc.cn/weatherservice/typhoon/jsons/view_{typhoon_id}?t={t}&callback=typhoon_jsons_view_{typhoon_id}"
    response = requests.get(url, headers=headers, verify=False)
    data_str = re.match(".*?({.*}).*", response.text, re.S).group(1)
    data = json.loads(data_str)
    typhoon_data = data["typhoon"]
    # 解析实时动态
    detail = {
        "台风ID": typhoon_id,
        "当前状态": typhoon_data[3],
        "中心经度": typhoon_data[8][-1][4],
        "中心纬度": typhoon_data[8][-1][5],
        "中心气压": typhoon_data[8][-1][6],
        "最大风速": typhoon_data[8][-1][7],
        "更新时间": typhoon_data[8][-1][1]
    }
    return detail

# 4. 批量采集并保存
if __name__ == "__main__":
    active_typhoons = get_current_typhoon_list()
    if active_typhoons:
        all_detail = []
        for typhoon in active_typhoons:
            detail = get_typhoon_detail(typhoon["台风ID"])
            # 合并基础信息和详细信息
            all_detail.append({**typhoon, **detail})
            time.sleep(1)  # 延迟1秒，避免反爬
        # 保存数据
        df = pd.DataFrame(all_detail)
        df.to_csv("中央气象台台风实时数据.csv", index=False, encoding="utf-8-sig")
        print(f"共采集到{len(all_detail)}个活跃台风数据，已保存")
    else:
        print("当前无活跃台风")