machete_hero/bot/Xigua.mjs

import Hero from '@ulixee/hero';
import configs from '../config.mjs';
import HeroBot from './HeroBot.mjs';

class Xigua extends HeroBot {

  async scrap(url) {
    let data = {};

    try {
        let options = {};

        if (this.heroServer) {
            options.connectionToCore = this.heroServer;
        }

        const profile = await this.init('xigua');
        if (profile) {
            options.userProfile = profile;
        }

        const hero = new Hero(options);
        await hero.goto(url, configs.heroBotOptions);

        //等待所有内容加载完成
        const tab = await hero.activeTab;
        await tab.waitForLoad('AllContentLoaded', {timeoutMs: configs.heroTabOptions.timeoutMs});

        //解析网页HTML数据
        const elems = await hero.detach( hero.document.querySelectorAll('meta') );
        let meta_name = '';
        for (const elem of elems) {
          meta_name = elem.getAttribute('name');
          if (!meta_name) {continue;}
          meta_name = meta_name.toLowerCase();
          if (meta_name.indexOf('og:image') > -1) {
            data.cover = elem.getAttribute('content');
          }else if (meta_name.indexOf('og:title') > -1) {
            data.title = elem.getAttribute('content');
          }
        }

        await hero.close();
    }catch(error) {
        console.error("Error got when request %s via hero: %s", url, error);
    };

    return data;
  }

}

export default Xigua;