【安全】requests和BeautifulSoup小试牛刀

 1 # -*- coding: UTF-8 -*-
 2 __author__ = ‘weimw‘
 3 
 4 import requests
 5 from BeautifulSoup import BeautifulSoup
 6 
 7 def request_ctf5():
 8     f = open("out.txt", "w")
 9     for i in range(1, 2000):
10         url = ‘http://ctf5.simplexue.com/8/index.php?id=‘ + str(i)
11         res = requests.get(url)
12         res_table = BeautifulSoup(res.content).find(‘table‘)
13         tr_1 = res_table.findAll(‘tr‘)[0]
14         try:
15             tr_2 = res_table.findAll(‘tr‘)[1]
16         except:
17             print >> f, "no id and content"
18             continue
19         id_t = tr_2.findAll(‘td‘)[0].text
20         content_t = tr_2.findAll(‘td‘)[1].text
21         if id_t and content_t:
22             print >> f, "id = %s, content = %s" % (id_t, content_t)
23     return
24 
25 if __name__ == ‘__main__‘:
26     request_ctf5()