interface IUser {
  id: string;
  nickname: string;
  avatarUrl: string;
}

interface IWeibo {
  id: string;
  content: string;
  imageUrls: string[];
  videoUrl?: string;
  audioUrl?: string;
  publishTime: Date;
  user: IUser;
}

interface IAudioInfo {
  url: string;
  title: string;
}

import axios from 'axios';
import cheerio from 'cheerio';
import { IWeibo, IUser, IAudioInfo } from './interfaces';

// 設(shè)置代理配置
const proxyConfig = {
  host: 'www.16yun.cn',
  port: '5445',
  auth: {
    username: '16QMSOML',
    password: '280651'
  }
};

// 獲取微博信息的函數(shù)
async function getWeiboInfo(weiboUrl: string): Promise<IWeibo | null> {
  try {
    const response = await axios.get(weiboUrl, {
      proxy: proxyConfig
    });
    const $ = cheerio.load(response.data);

    // 提取用戶信息
    const user: IUser = {
      id: $('#user_id').text(),
      nickname: $('#user_nickname').text(),
      avatarUrl: $('#user_avatar').attr('src'),
    };

    // 提取微博內(nèi)容
    const content = $('#weibo_content').text();

    // 提取圖片 URL
    const imageUrls = $('#weibo_images img').map((i, img) => $(img).attr('src')).get();

    // 提取視頻 URL
    const videoUrl = $('#weibo_video').attr('src');

    // 提取音頻信息
    const audioInfo = await getAudioInfo(weiboUrl);

    // 提取發(fā)布時(shí)間
    const publishTime = new Date($('#publish_time').text());

    return {
      id: $('#weibo_id').text(),
      content,
      imageUrls,
      videoUrl,
      audioUrl: audioInfo ? audioInfo.url : undefined,
      publishTime,
      user,
    };
  } catch (error) {
    console.error('獲取微博信息失敗:', error);
    return null;
  }
}

// 獲取音頻信息的函數(shù)
async function getAudioInfo(weiboUrl: string): Promise<IAudioInfo | null> {
  try {
    const response = await axios.get(weiboUrl, {
      proxy: proxyConfig
    });
    const $ = cheerio.load(response.data);
    const audioUrl = $('audio').attr('src');
    const audioTitle = $('audio').attr('title');
    if (audioUrl && audioTitle) {
      return { url: audioUrl, title: audioTitle };
    }
    return null;
  } catch (error) {
    console.error('獲取音頻信息失敗:', error);
    return null;
  }
}

// 示例用法
(async () => {
  try {
    const weiboUrl = 'https://weibo.com/1234567890/1234567890123456'; 
    const weiboInfo = await getWeiboInfo(weiboUrl);
    if (weiboInfo) {
      console.log('微博信息:', weiboInfo);
    } else {
      console.log('沒(méi)有找到微博信息');
    }
  } catch (error) {
    console.error('爬取微博失敗:', error);
  }
})();

npx ts-node crawler.ts