Python实现登录人人网并抓取新鲜事的方法
发布时间:2020-12-17 17:27:38 所属栏目:Python 来源:网络整理
导读:今天PHP站长网 52php.cn把收集自互联网的代码分享给大家,仅供参考。 from sgmllib import SGMLParserimport sys,urllib2,urllib,cookielibclass spider(SGMLParser): def __init__(self,email,password): SGMLParser.__i
以下代码由PHP站长网 52php.cn收集自互联网 现在PHP站长网小编把它分享给大家,仅供参考 from sgmllib import SGMLParser import sys,urllib2,urllib,cookielib class 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() 以上内容由PHP站长网【52php.cn】收集整理供大家参考研究 如果以上内容对您有帮助,欢迎收藏、点赞、推荐、分享。 (编辑:李大同) 【声明】本站内容均来自网络,其相关言论仅代表作者个人观点,不代表本站立场。若无意侵犯到您的权利,请及时与联系站长删除相关内容! |
推荐文章
站长推荐
热点阅读