Python实现登录人人网并抓取新鲜事的方法

这里演示了Python登录人人网并抓取新鲜事的方法（抓取后的排版不太美观~~）

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.p=False
    self.h3_and_p=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.p:
      self.a=True
  def end_a(self):
    self.a=False
  def start_p(self,attrs):
    if self.h3_is_ready == False:
      return
    if self.p==True:
      self.depth += 1
    for k,v in attrs:
      if k == 'class' and v == 'content':
        self.p=True;
        self.h3_and_p=True  #h3 and p is connected
  def end_p(self):
    if self.depth == 0:
      self.p=False
      self.h3_and_p=False
      self.h3_is_ready=False
      self.names=""
    if self.p == 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_p:
      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()

希望本文所述对大家的Python程序设计有所帮助。