黑板客 -- 爬虫闯关 -- 关卡01

爬虫闯关链接：http://www.heibanke.com/lesson/crawler_ex00

知识点：Python正则表达式，Web编程

参考代码：

#!/usr/bin/env python
# encoding: utf-8

import requests
import sys
import re
reload(sys)

sys.setdefaultencoding("utf-8")

str = ""
def attack():
	global str
	while True:
		resp = requests.get("http://www.heibanke.com/lesson/crawler_ex00/"+str)
		pattern = u"数字是?(\d+)".decode("utf8")
		word = re.findall(pattern, unicode(resp.content).decode("utf8"))
		print word
		if not word:
			break
		else:
			str = word[0]
			continue
	print resp.content

if __name__ == '__main__':
	attack()