Python实现登录人人网并抓取新鲜事的方法,python新鲜事,from sgmllib
文章由Byrx.net分享于2019-03-23 04:03:51
Python实现登录人人网并抓取新鲜事的方法,python新鲜事,from sgmllib
from sgmllib import SGMLParserimport sys,urllib2,urllib,cookielibclass spider(SGMLParser): def __init__(self,email,password): SGMLParser.__init__(self) self.h3=False self.h3_is_ready=False self.div=False self.h3_and_div=False self.a=False self.depth=0 self.names="" self.dic={} self.email=email self.password=password self.domain='renren.com' try: cookie=cookielib.CookieJar() cookieProc=urllib2.HTTPCookieProcessor(cookie) except: raise else: opener=urllib2.build_opener(cookieProc) urllib2.install_opener(opener) def login(self): url='http://www.renren.com/PLogin.do' postdata={ 'email':self.email, 'password':self.password, 'domain':self.domain } req=urllib2.Request( url, urllib.urlencode(postdata) ) self.file=urllib2.urlopen(req).read() #print self.file def start_h3(self,attrs): self.h3 = True def end_h3(self): self.h3=False self.h3_is_ready=True def start_a(self,attrs): if self.h3 or self.div: self.a=True def end_a(self): self.a=False def start_div(self,attrs): if self.h3_is_ready == False: return if self.div==True: self.depth += 1 for k,v in attrs: if k == 'class' and v == 'content': self.div=True; self.h3_and_div=True #h3 and div is connected def end_div(self): if self.depth == 0: self.div=False self.h3_and_div=False self.h3_is_ready=False self.names="" if self.div == True: self.depth-=1 def handle_data(self,text): #record the name if self.h3 and self.a: self.names+=text #record says if self.h3 and (self.a==False): if not text:pass else: self.dic.setdefault(self.names,[]).append(text) return if self.h3_and_div: self.dic.setdefault(self.names,[]).append(text) def show(self): type = sys.getfilesystemencoding() for key in self.dic: print ( (''.join(key)).replace(' ','')).decode('utf-8').encode(type), \ ( (''.join(self.dic[key])).replace(' ','')).decode('utf-8').encode(type)renrenspider=spider('your email','your password')renrenspider.login()renrenspider.feed(renrenspider.file)renrenspider.show()
相关内容
- 求a到b之间所有质数,求a质数,def Prime(b)
- Python汉字转拼音,python汉字拼音,使用字典和转换程序
- pygame tetromino 俄罗斯方块,pygametetromino,import pygam
- pil 放缩图像功能,pil缩图像功能,import Image
- 查单词的词典,单词词典,# -*- coding
- 抢购业务模块(redis+mysql),redismysql,class CatchC
- 下载全国城市经纬度,全国城市经纬度,import reim
- Python邮件服务,,#!/usr/bin/e
- python实现堆排序算法代码,python堆排序算法,#!/usr/bin/
- Python测试Sqlite代码,pythonsqlite代码,from sqlite3
评论关闭