python實現自動登錄人人網并采集信息的源碼

將寫代碼過程中經常用到的一些代碼片段做個收藏，如下代碼段是關于python實現自動登錄人人網并采集信息的的代碼，應該是對大家有用途。

#!/usr/bin/python

import sys
import re
import urllib2
import urllib
import cookielib

class Renren(object):

    def __init__(self):
        self.name = self.pwd = self.content = self.domain = self.origURL =  ''
        self.operate = ''#登錄進去的操作對象
        self.cj = cookielib.LWPCookieJar()
        try: 
            self.cj.revert('./renren.coockie') 
        except Exception,e:
            print e

        self.opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(self.cj))
        urllib2.install_opener(self.opener)

    def setinfo(self,username,password,domain,origURL):
        '''設置用戶登錄信息'''
        self.name = username
        self.pwd = password
        self.domain = domain
        self.origURL = origURL

    def login(self):
        '''登錄人人網'''
        params = {
            'domain':self.domain,
            'origURL':self.origURL,
            'email':self.name, 
            'password':self.pwd}
        print 'login.......'
        req = urllib2.Request( 
            urllib.urlencode(params)
        )

        self.file=urllib2.urlopen(req).read()        
        newsfeed = open('news.html','w')
        try:
            newsfeed.write(self.file)
        except Exception, e:
            newsfeed.close()
        self.operate = self.opener.open(req) 
        print type(self.operate)
        print self.operate.geturl()

        if self.operate.geturl(): 
            print 'Logged on successfully!'
            self.cj.save('./renren.coockie')
            self.__viewnewinfo()
        else:
            print 'Logged on error'

    def __viewnewinfo(self):
        '''查看好友的更新狀態'''
        self.__caiinfo()

    def __caiinfo(self):
        '''采集信息'''       
        apatten = re.compile('<h4.+>(.+)</h4>:')#匹配作者
        cpatten = re.compile('</a>(.+)s')#匹配內容  
        content = h4patten.findall(self.file)
        print len(content)   
        infocontent = self.operate.readlines()
        print type(infocontent)
        print 'friend newinfo:' 
        for i in infocontent:
            content = h4patten.findall(i)
            if len(content) != 0:
                for m in content:
                    username = apatten.findall(m)
                    info = cpatten.findall(m)
                    if len(username) !=0:
                        print username[0],'說:',info[0]
                        print '----------------------------------------------'
                    else:
                        continue

ren = Renren()
username = 'username'#你的人人網的帳號
password = 'password'#你的人人網的密碼
domain = 'www.renren.com'#人人網的地址
ren.setinfo(username,password,domain,origURL)
ren.login()