Python爬虫学习（简单的模拟登陆（二））

用cookie模拟新浪微博登录，可以对某人新浪微博是否更新进行实时的邮件监控

#-*-coding:utf8-*-

import smtplib
from email.mime.text import MIMEText
import requests
from lxml import etree
import os
import time
import sys
reload(sys)
sys.setdefaultencoding('utf-8')


#########################################################################################
class mailhelper(object):
    def __init__(self):

        self.mail_host="smtp.xxxx.com"  #设置服务器
        self.mail_user="xxxxxx"    #用户名
        self.mail_pass="xxxxxx"   #密码
        self.mail_postfix="xxxx.com"  #发件箱的后缀

    def send_mail(self,to_list,sub,content):
        me="xxoohelper"+"<"+self.mail_user+"@"+self.mail_postfix+">"
        msg = MIMEText(content,_subtype='plain',_charset='utf-8')
        msg['Subject'] = sub
        msg['From'] = me
        msg['To'] = ";".join(to_list)
        try:
            server = smtplib.SMTP()
            server.connect(self.mail_host)
            server.login(self.mail_user,self.mail_pass)
            server.sendmail(me, to_list, msg.as_string())
            server.close()
            return True
        except Exception, e:
            print str(e)
            return False
############################################################################################################3



class xxoohelper(object):
    def __init__(self):
        self.url = 'http://weibo.cn/xxxxxx' #请输入准备抓取的微博地址

    def getSource(self):
        html = requests.get(self.url).content
        return html

    def getData(self,html):
        data = {"Cookie":"需要抓取的Cookies"}
        return data

    def getContent(self,data):
        newhtml = requests.get(self.url,cookies = data).content
        new_selector = etree.HTML(newhtml)
        content = new_selector.xpath('//span[@class="ctt"]')
        newcontent = unicode(content[2].xpath('string(.)')).replace('http://','')
        sendtime = new_selector.xpath('//span[@class="ct"]/text()')[0]
        sendtext = newcontent + sendtime
        return sendtext

    def tosave(self,text):
        f= open('weibo.txt','a')
        f.write(text + '\n')
        f.close()

    def tocheck(self,data):
        if not os.path.exists('weibo.txt'):
            return True
        else:
            f = open('weibo.txt', 'r')
            existweibo = f.readlines()
            if data + '\n' in existweibo:
                return False
            else:
                return True

if __name__ == '__main__':
    mailto_list=['[email protected]'] #此处填写接收邮件的邮箱
    helper = xxoohelper()
    while True:
        source = helper.getSource()
        data = helper.getData(source)
        content = helper.getContent(data)
        if helper.tocheck(content):
            if mailhelper().send_mail(mailto_list,u"女神更新啦",content):
                print u"发送成功"
            else:
                print u"发送失败"
            helper.tosave(content)
            print content
        else:
            print u'pass'
        time.sleep(10)