node.js爬取中关村的在线电瓶车信息

var cheerio = require('cheerio');
var fetch = require('node-fetch');

// cheerio 是一个类似浏览器端的jQuery，用来解析HTML的
// fetch 用来发送请求

// 初始url
var url = "http:/https://img.qb5200.com/download-x/detail.zol.com.cn/convenienttravel/hangzhou/#list_merchant_loc"
// 由于每个a标签下是相对路径，故需要一个根地址来拼接，如下
var urlRoot = "http:/https://img.qb5200.com/download-x/detail.zol.com.cn" 
// 存放所有url，之所以用set，是为了防止有相同的而重复爬去
var urls = new Set()
// 存储所有数据
var data = [] 

// 这是得到每个页面的48个链接，并开始发送请求

function ad(arg){
// 参数 arg 先不管
// 本地化一下需要爬取的链接
let url2 = arg || url;
// 请求第一页该网页，拿到数据之后，复制给 app
var app = await fetch(url2).then(res=>res.text())
// 然后假装用jQuery解析了
var $ = cheerio.load(app)
// 获取当前页所有电瓶车的a标签
var ele = $("#J_PicMode a.pic")
// 存放已经爬取过的url，防止重复爬取
var old_urls = []
var urlapp = []
//拿到所有a标签地址之后，存在数组里面，等会儿要开始爬的
for (let i = 0; i < ele.length; i++) {
old_urls.push(fetch(urlRoot+$(ele[i]).attr('href')).then(res=>res.text()))
}
// 用把URL一块丢给promise处理
urlapp = await Promise.all(old_urls)
// 处理完成之后，循环加入jQuery😂
for (let i = 0; i < urlapp.length; i++) {
let $2 = cheerio.load(urlapp[i],{decodeEntities: false})
data.push({
name:$2(".product-model__name").text(),
price:$2(".price-type").text()
})
}
// 至此，一页的数据就爬完了
// console.log(data);

// 然后开始爬取下一页
var nextURL = $(".next").attr('href')
// 判断当前页是不是最后一页
if (nextURL){
let next = await fetch(urlRoot+nextURL).then(res=>res.text())
// 获取下一页的标签，拿到地址，走你
ad(urlRoot+nextURL)
}
return data
}
ad()

var cheerio = require('cheerio');
var fetch = require('node-fetch');
var url = "http:/https://img.qb5200.com/download-x/detail.zol.com.cn/convenienttravel/hangzhou/#list_merchant_loc"
var urlRoot = "http:/https://img.qb5200.com/download-x/detail.zol.com.cn"
// var url = "http://localhost:3222/app1"
var urls = new Set()
var data = [] 
async function ad(arg){
let url2 = arg || url;
var app = await fetch(url2).then(res=>res.text())
var $ = cheerio.load(app)
var ele = $("#J_PicMode a.pic")
var old_urls = []
var urlapp = []
for (let i = 0; i < ele.length; i++) {
old_urls.push(fetch(urlRoot+$(ele[i]).attr('href')).then(res=>res.text()))
}
urlapp = await Promise.all(old_urls)
for (let i = 0; i < urlapp.length; i++) {
let $2 = cheerio.load(urlapp[i],{decodeEntities: false})
data.push({
name:$2(".product-model__name").text(),
price:$2(".price-type").text()
})
}

var nextURL = $(".next").attr('href')
if (nextURL){
let next = await fetch(urlRoot+nextURL).then(res=>res.text())
ad(urlRoot+nextURL)
}
return data
}
ad()