#控制台终端执行命令 准备环境
pip install -i https://pypi.tuna.tsinghua.edu.cn/simple some-package lxml
#导入库
from urllib.request import urlopen
from lxml import etree
import os
#读取网页
html = urlopen('http://gp.qq.com/cp/a20190522gamedata/pc_list.shtml')
html_lxml = etree.HTML(html.read().decode('GBK'))
#匹配元素
list_index = html_lxml.xpath('//div[@id="section-container"]/ul')
#循环列表
for list in list_index:
#获取枪支分类
file_title = list.xpath('preceding-sibling::h3')[-1].xpath('string()')
print('正在下载' + file_title + '文件')
#创建文件夹(枪支分类)
if not os.path.exists(file_title):
os.mkdir(file_title)
#获取枪支
for href in list.xpath('./li/a'):
#读取图片链接
img_src = urlopen('https:'+href.xpath('./img/@src')[0]).read()
#获取枪支名称
img_title = href.xpath('./p/text()')[0]
print('下载文件:' + img_title)
#保存文件
with open(file_title+'/'+img_title+'.png', mode='wb') as ret:
ret.write(img_src)
print('下载成功')