Ejemplo n.º 1
0
    def __init__(self, taskdb, projectdb, newtask_queue, status_queue,
                 out_queue, data_path='./data', resultdb=None):
        self.taskdb = taskdb
        self.projectdb = projectdb
        self.resultdb = resultdb
        self.newtask_queue = newtask_queue
        self.status_queue = status_queue
        self.out_queue = out_queue
        self.data_path = data_path

        self._send_buffer = deque()
        self._quit = False
        self._exceptions = 0
        self.projects = dict()
        self._force_update_project = False
        self._last_update_project = 0
        self._last_tick = int(time.time())
        self._postpone_request = []

        self._cnt = {
            "5m_time": counter.CounterManager(
                lambda: counter.TimebaseAverageEventCounter(30, 10)),
            "5m": counter.CounterManager(
                lambda: counter.TimebaseAverageWindowCounter(30, 10)),
            "1h": counter.CounterManager(
                lambda: counter.TimebaseAverageWindowCounter(60, 60)),
            "1d": counter.CounterManager(
                lambda: counter.TimebaseAverageWindowCounter(10 * 60, 24 * 6)),
            "all": counter.CounterManager(
                lambda: counter.TotalCounter()),
        }
        self._cnt['1h'].load(os.path.join(self.data_path, 'scheduler.1h'))
        self._cnt['1d'].load(os.path.join(self.data_path, 'scheduler.1d'))
        self._cnt['all'].load(os.path.join(self.data_path, 'scheduler.all'))
        self._last_dump_cnt = 0
Ejemplo n.º 2
0
    def __init__(self,
                 inqueue,
                 outqueue,
                 poolsize=100,
                 proxy=None,
                 async_mode=True):
        self.inqueue = inqueue
        self.outqueue = outqueue

        self.poolsize = poolsize
        self._running = False
        self._quit = False
        self.proxy = proxy
        self.async_mode = async_mode
        self.ioloop = tornado.ioloop.IOLoop()

        self.robots_txt_cache = {}

        # binding io_loop to http_client here
        if self.async_mode:
            self.http_client = MyCurlAsyncHTTPClient(max_clients=self.poolsize,
                                                     io_loop=self.ioloop)
        else:
            self.http_client = tornado.httpclient.HTTPClient(
                MyCurlAsyncHTTPClient, max_clients=self.poolsize)

        self._cnt = {
            '5m':
            counter.CounterManager(
                lambda: counter.TimebaseAverageWindowCounter(30, 10)),
            '1h':
            counter.CounterManager(
                lambda: counter.TimebaseAverageWindowCounter(60, 60)),
        }