[python] HDU自动登录提交代码程序

调了一个下午，被python的正则绊住了；在C#上运作好好的式子在python老是报错，原来python的断言式必须是固定长度的，像类似(?<=[^>].*?)的零宽度正回顾后发断言是不允许出现的，这点python做的没有C# 方便，弄得我只好分了好几个步骤来做.

 1 __author__ = 'wuminye'
 2 import time
 3 import urllib
 4 import urllib2
 5 import cookielib
 6 import re
 7 
 8 std = ['Queuing', 'Compiling', 'Running']
 9 
10 hostaddr = 'http://acm.hdu.edu.cn'
11 loginaddr = '/userloginex.php?action=login'
12 submaddr = '/submit.php?action=submit'
13 statusaddr = '/status.php'
14 
15 cj = cookielib.LWPCookieJar()
16 cookie_support = urllib2.HTTPCookieProcessor(cj)
17 #httpHandler = urllib2.HTTPHandler(debuglevel=1)
18 opener = urllib2.build_opener(cookie_support, urllib2.HTTPHandler)
19 urllib2.install_opener(opener)
20 header = {'User-Agent': 'IE10'}
21 
22 
23 def login(username, password):
24     """
25 
26     :rtype : object
27     """
28     postData = {'username': username,
29                 'userpass': password,
30                 'submit': 'Sign In'}
31     postData = urllib.urlencode(postData)
32     request = urllib2.Request(hostaddr + loginaddr, postData, header)
33     response = urllib2.urlopen(request, timeout=10)
34     if response.getcode() != 200 and response.getcode() != 302:
35         print 'login web error!'
36         return False
37     ans = response.read()
38     if ans.find("Sign Out") == -1:
39         return False
40     return True
41 
42 
43 def querystatus():
44     postData = {'user': 'wuminye',
45                 'lang': 0,
46                 'first': '',
47                 'pid': '',
48                 'status': 0}
49     postData = urllib.urlencode(postData)
50     request = urllib2.Request(hostaddr + statusaddr + '?' + postData, headers=header)
51     response = urllib2.urlopen(request, timeout=10)
52     if response.getcode() != 200 and response.getcode() != 302:
53         print 'query web error!'
54         return False
55     ans = response.read()
56     m = re.search('(?<=<td height=22px>).*?</td>(<td.*?</td>){8}(?=</tr>)', ans)
57     ans = '<td>' + m.group(0)
58     mylist = []
59     for i in range(0, 7):
60         m = re.search(r"(?<=<td).+?(?=</td>)", ans)
61         t = m.group()[1:]
62         n = t.find('>')
63         if n != -1:
64             t = t[n + 1:]
65         t = re.search('[^<]+', t)
66         mylist.append(t.group())
67         ans = ans[m.end() + 5:]
68     for i in range(0, 4):
69         ans = ans[ans.find('>') + 1:]
70     ans = re.search(r"[^<]+", ans).group()
71     mylist.append(ans)
72     return mylist
73 
74 
75 def subbmit(pid, lan, code):
76     postData = {'problemid': pid,
77                 'language': lan,
78                 'usercode': code,
79                 'submit': 'Submit'}
80     postData = urllib.urlencode(postData)
81     request = urllib2.Request(hostaddr + submaddr, postData, header)
82     response = urllib2.urlopen(request, timeout=10)
83     if response.getcode() != 200 and response.getcode() != 302:
84         print 'subbmit web error!'
85         return False
86     print 'Subbmit Done!'
87     ans = querystatus()
88     while ans[2] in std:
89         time.sleep(1)
90         print 'Status:', ans[2]
91         ans = querystatus()
92     print ans