gistfile1.txt

#!/usr/bin/python
# - * - coding: UTF-8 - * -

# Copyright (C) 2010 Upsuper <[email protected]>
# License: AGPLv3

from threading import Thread, Lock
from httplib import HTTPConnection
import urllib
import cgi, json
import sys, re

# 常数设置
THREADS_NUM = 5             # 最大线程数
RENREN_USER = ''            # 人人网用户名
RENREN_PWD  = ''            # 人人网密码
USER_AGENT  = 'urenren 0.1' # 提交给人人网的 User-Agent

# JSON 输出函数
json_dump = lambda v: json.dump(v, sys.stdout)
# 编译匹配用正则表达式
parse_re = re.compile(r'<td><p><a\s(?:[^>]+\s)?href="[^\?]+\?id=(\d+)[^"]*">'
                      r'<img src="([^"]+)"(?:[^>]*)/></a></p>'
                      r'<a href="[^"]+">([^<]+)</a></td>')

class RequestThread(Thread):
  def __init__(self, id, sid, friends):
    Thread.__init__(self)
    self.__page = '/getfriends.do?curpage=%%d&id=%d&sid=%s' % (id, sid)
    self.__friends = friends
    self.__conn = HTTPConnection('3g.renren.com')

  def run(self):
    global curpage, stop_sign

    while not stop_sign:
      # 获取当前页面
      curpage_lock.acquire()
      page = curpage
      curpage += 1
      curpage_lock.release()

      # 连接获取数据
      conn = self.__conn
      conn.request('GET', self.__page % (page, ), None, {
        'User-Agent': USER_AGENT
      })
      data = conn.getresponse().read()

      # 处理获取的数据
      friend_iter = parse_re.finditer(data)
      t_stop_sign = True
      for f in friend_iter:
        t_stop_sign = False
        id = int(f.group(1))
        self.__friends[id] = (f.group(3), f.group(2))
      stop_sign = t_stop_sign

def readFriends(id):
  # 初始化多线程
  global curpage, curpage_lock, stop_sign
  curpage = 0
  curpage_lock = Lock()
  stop_sign = False
  threads = []
  friends = {}

  # 创建线程
  for i in xrange(THREADS_NUM):
    threads.append(RequestThread(id, sid, friends))
  # 开始执行线程
  for t in threads:
    t.start()
  # 等待线程结束
  for t in threads:
    t.join()

  return friends

def main():
  # 初始化CGI输出
  print 'Content-Type: text/plain'
  print

  # 获取ID信息
  form = cgi.FieldStorage()
  id1, id2 = 0, 0
  if form.has_key('id1'): id1 = int(form['id1'].value)
  if form.has_key('id2'): id2 = int(form['id2'].value)
  if not (id1 and id2):
    json_dump({'error': True})
    return

  # 登入人人网
  conn = HTTPConnection('3g.renren.com')
  conn.request('POST', '/login.do',
      urllib.urlencode({'email': RENREN_USER, 'password': RENREN_PWD}),
      {
        'Content-Type': 'application/x-www-form-urlencoded',
        'User-Agent': USER_AGENT,
      })
  response = conn.getresponse()
  data = response.read()
  conn.close()
  match = re.search(r'sid=([0-9a-f]+)', data, re.I)
  global sid
  sid = match.group(1)

  # 读取好友
  friends1 = readFriends(id1)
  friends2 = readFriends(id2)
  if len(friends1) > len(friends2):
    friends1, friends2 = friends2, friends1

  # 判断交集
  intersection = []
  for i in friends1.iterkeys():
    if i in friends2:
      intersection.append((i, friends2[i][0], friends2[i][1]))
  json_dump({'error': False, 'count': len(intersection), 'data': intersection})

if __name__ == '__main__':
  try:
    main()
  except:
    json_dump({'error': True})