爬虫--音乐

一、代码如下

import re
import requests

class Yinyue(object):

    head = {
        'user-agent':'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.26 Safari/537.36 Core/1.63.6726.400 QQBrowser/10.2.2265.400'
    }

    def huoquyuming(self):
        url = 'http://www.yy8844.cn/'
        res = requests.get(url=url,headers=self.head)
        html = res.content.decode('gbk')
        return html

    def guolv(self,html):
        lianjie = []
        patt = re.compile(r'<a href="/ting/(.*?).shtml"')
        item = patt.findall(html)
        for i in item:
            lianjie.append('http://www.yy8844.cn/ting/{}.shtml'.format(i))
        return lianjie

    def save(self,lianjie):
        items = []
        for i in lianjie:
            res = requests.get(i)
            html = res.content.decode('gbk')
            patt = re.compile(r'var MusicId=(.*?);')
            item = patt.findall(html)
            items.append(item)
        for w,j in enumerate(items):
            url = 'http://96.ierge.cn/%d/%d/%d.mp3'%(int(j[0])/30000,int(j[0])/2000,int(j[0]))
            req = requests.get(url)
            html1 = req.content
            with open(r'C:\Users\wangyl\daima\music\{}.mp3'.format(w),'wb') as f:
                f.write(html1)

yinyue = Yinyue()
html = yinyue.huoquyuming()
lianjie = yinyue.guolv(html)
yinyue.save(lianjie)