cheerio 解析新浪博客目录列表
作者:互联网
正序目录
- 解析提取手机版新浪博客网页的目录,然后存成markdown文件
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] === 'txt') {
fs.readFile(path + '/' + file, 'utf-8', (err, data) => {
const $ = cheerio.load(data);
// const writeStream = fs.createWriteStream(path + '/' + file, 'utf-8');
const writeStream = fs.createWriteStream(
path + '/目录/' + '新浪博客目录.txt',
'utf-8'
);
$('#pl-home-bloglist > article > ul>li').each((data, ele) => {
let title = $(ele).find('h2').html();
let url = $(ele).find('a').attr('data-link');
writeStream.write('[');
writeStream.write(title);
writeStream.write(']');
writeStream.write('(');
writeStream.write(url);
writeStream.write(')');
writeStream.write('\n');
writeStream.write('\n');
console.log(title);
console.log(url);
});
writeStream.end();
});
}
});
});
倒序目录
- 和上面的效果一样,只不过是倒序的
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] === 'txt') {
fs.readFile(path + '/' + file, 'utf-8', (err, data) => {
const $ = cheerio.load(data);
// const writeStream = fs.createWriteStream(path + '/' + file, 'utf-8');
const writeStream = fs.createWriteStream(
path + '/目录/' + '新浪博客目录.txt',
'utf-8'
);
$($('#pl-home-bloglist > article > ul>li').get().reverse()).each(
(data, ele) => {
let title = $(ele).find('h2').html();
let url = $(ele).find('a').attr('data-link');
writeStream.write('[');
writeStream.write(title);
writeStream.write(']');
writeStream.write('(');
writeStream.write(url);
writeStream.write(')');
writeStream.write('\n');
writeStream.write('\n');
console.log(title);
console.log(url);
}
);
writeStream.end();
});
}
});
});
标签:fs,const,cheerio,列表,write,file,path,解析,writeStream 来源: https://www.cnblogs.com/zhuoss/p/15897938.html