# -*- coding: utf-8 -*-
import urllib, re, time, threading
from Queue import Queue
queue = Queue()
start = time.time()
f = open("user_th10.txt", "w+")
def Main():
    while True:
          if queue.empty():
            return
          site = queue.get_nowait()
          url = urllib.urlopen(site).read()
          pattern = r'profile.php\?id=\d+\">([^<]+)</a>'
          users = re.compile(pattern)
          users = re.findall(pattern, url)
          if len(users) == 0:
            break
          for user in users:
                f.write(user.decode('windows-1251').encode('utf-8')+'\n')
                print user

for count in range(1, 15, 1):
    queue.put('http://python.su/forum/userlist.php?username=&show_group=-1&sort_by=username&sort_dir=ASC&p='+str(count))
threads = []
for x in xrange(10):
    t = threading.Thread(target = Main)
    threads.append(t)
    t.start()
for t in threads:
    t.join()
f.write("Времени затрачено: %s" % (time.time() - start)+'\n')
f.close()

class Parser:
    def __init__(self, ):
        self.queue = Queue()
        self.t = threading.Thread()
        self.start_time = time.time()
        self.fileopen = filename

    def start_p(self, fileopen):
        while True:
            if self.queue.empty():
                return
            site = self.queue.get_nowait()
            url_site = urllib.urlopen(site).read()
            pattern = r'class=\"gen\">([^<]+)</a>'
            users = re.compile(pattern)
            users = re.findall(pattern, url_site)
            with open(fileopen, 'a') as f:
                for user in users:
                    f.write(user.decode('windows-1251').encode('utf-8')+'\n')
                    print user
            self.queue.task_done()
    def parser(self, url):
        for count in range(0, 1000, 50):
            self.queue.put(url+'memberlist.php?start='+str(count))
        threads = []
        for x in xrange(25):
            self.t = threading.Thread(target = self.start_p(self.fileopen))
            threads.append(self.t)
            self.t.start()
        for self.t in threads:
            self.t.join()