微信公众号 话题链接提取

背景

1. 这个可以用来提取话题界面中的文章链接

const fs = require('fs');
const cheerio = require('cheerio');
let path = `./html`;
fs.readdir(path, function (err, files) {
  files.forEach((file) => {
    console.log(file);
    if (file.split('.')[1] === 'html') {
      fs.readFile(path + '/' + file, 'utf-8', (err, data) => {
        const $ = cheerio.load(data);

        // 1. 提取目录
        const writeStream = fs.createWriteStream(
          path + '/目录/' + file,
          'utf-8'
        );
        $(
          '#js_content_overlay > div.album.js_album_container.album-rich_media_area_primary_full > div > div.album__content.js_album_bd > ul li'
        ).each((index, ele) => {
          writeStream.write($(ele).attr('data-link'));
          writeStream.write('\n');
        });
        writeStream.end();
      });
    }
  });
});