Пример #1
0
 def setUp(self):
     """set up"""
     config = {}
     mini_spider.read_conf("spider.conf", config)
     thread_safe_set = safe_utils.thread_safe(set)
     urlSet = thread_safe_set()
     queue = Queue.Queue(maxsize=-1)
     rlock = threading.RLock()
     self.spider = spider.MiniSpider(config, queue, urlSet, rlock)
Пример #2
0
                        t = spider.MiniSpider(config, queue, urlSet, rlock)
                        threads.append(t)
                else:
                    error = "config threadcount is not a positive integer,\
                            please check your config file, exiting ..."
                    log.log("error", error)
                    sys.exit(1)
                for i in range(len(threads)):
                    log.log("info", threads[i].getName() + " is starting")
                    threads[i].setDaemon(True)
                    threads[i].start()
                print "now spidering ,please wait ~~~"
                queue.join()
                info = "now all done , and exiting"
                log.log("info", info)
                print "spider done ~~"
            else:
                error = "config read failed!"
                log.log("error", error)
                sys.exit(1)


if __name__ == '__main__':
    config = {}
    #Bloom Filters 
    thread_safe_set = safe_utils.thread_safe(set)
    urlSet = thread_safe_set()
    queue = Queue.Queue(maxsize = -1)
    rlock = threading.RLock()
    main(config, queue, urlSet, rlock)