第11章 链接验证

import requests,bs4,os
url = "http://www.baidu.com"
os.makedirs('xkcd',exist_ok=True)
res = requests.get(url)
res.raise_for_status()
resHtml = bs4.BeautifulSoup(res.text,'html.parser')
linkUrl = resHtml.select('a[href]')
for i in range(len(linkUrl)):
    vari = linkUrl[i].get('href')
    if vari.startswith("http"):
        sonUrl = vari
    elif vari.startswith("#"):
        continue
    else:
        sonUrl = "http://www.baidu.com"+ vari
    temp = requests.get(sonUrl)
    sonFile = open(os.path.join('xkcd',str(i)+'.html'),'wb')
    for chunk in temp.iter_content(100000):
        sonFile.write(chunk)
    sonFile.close()
    print("%s第%d个任务已完成"%(sonUrl,i))
print("任务已完成！")