node.js爬取中關(guān)村的在線電瓶車信息

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

// cheerio 是一個(gè)類似瀏覽器端的jQuery，用來解析HTML的
// fetch 用來發(fā)送請(qǐng)求

// 初始url
var url = "http://detail.zol.com.cn/convenienttravel/hangzhou/#list_merchant_loc"
// 由于每個(gè)a標(biāo)簽下是相對(duì)路徑，故需要一個(gè)根地址來拼接，如下
var urlRoot = "http://detail.zol.com.cn" 
// 存放所有url，之所以用set，是為了防止有相同的而重復(fù)爬去
var urls = new Set()
// 存儲(chǔ)所有數(shù)據(jù)
var data = [] 

// 這是得到每個(gè)頁面的48個(gè)鏈接，并開始發(fā)送請(qǐng)求

function ad(arg){
// 參數(shù) arg 先不管
// 本地化一下需要爬取的鏈接
let url2 = arg || url;
// 請(qǐng)求第一頁該網(wǎng)頁，拿到數(shù)據(jù)之后，復(fù)制給 app
var app = await fetch(url2).then(res=>res.text())
// 然后假裝用jQuery解析了
var $ = cheerio.load(app)
// 獲取當(dāng)前頁所有電瓶車的a標(biāo)簽
var ele = $("#J_PicMode a.pic")
// 存放已經(jīng)爬取過的url，防止重復(fù)爬取
var old_urls = []
var urlapp = []
//拿到所有a標(biāo)簽地址之后，存在數(shù)組里面，等會(huì)兒要開始爬的
for (let i = 0; i < ele.length; i++) {
old_urls.push(fetch(urlRoot+$(ele[i]).attr('href')).then(res=>res.text()))
}
// 用把URL一塊丟給promise處理
urlapp = await Promise.all(old_urls)
// 處理完成之后，循環(huán)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，一頁的數(shù)據(jù)就爬完了
// console.log(data);

// 然后開始爬取下一頁
var nextURL = $(".next").attr('href')
// 判斷當(dāng)前頁是不是最后一頁
if (nextURL){
let next = await fetch(urlRoot+nextURL).then(res=>res.text())
// 獲取下一頁的標(biāo)簽，拿到地址，走你
ad(urlRoot+nextURL)
}
return data
}
ad()

var cheerio = require('cheerio');
var fetch = require('node-fetch');
var url = "http://detail.zol.com.cn/convenienttravel/hangzhou/#list_merchant_loc"
var urlRoot = "http://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()