def setUp(self):
self.clock = Clock()
self.request_queue = MemoryQueue()
self.response_queue = ResponseQueue()
self.dwn = Downloader(Settings(self.default_settings),
self.request_queue,
self.response_queue,
download_handler=MockDownloaderHandler(
Settings()),
clock=self.clock)
self.handler = self.dwn.download_handler
def __init__(self, download_handler, concurrency, delay, randomize_delay,
clock=None):
self.download_handler = download_handler
self.concurrency = concurrency
self.delay = delay
self.randomize_delay = randomize_delay
self.in_progress = set() # request waiting to be downloaded
self.transferring = set() # requests being downloaded (subset of `in_progress`)
self.last_download_time = 0
self.queue = MemoryQueue() # queue of (request, deferred)
# clock is used in unittests
self.clock = clock or reactor
self.delayed_processing = ScheduledCall(self._process, clock=self.clock)
def setup(self):
assert self.spider is not None, 'Spider is not set in Engine.'
# IMPORTANT: order of the following initializations is very important
# so please, think twice about any changes to it
# initialize logging
if self.settings.get_bool('LOG_ENABLED'):
log.start(self.settings['LOG_FILE'], self.settings['LOG_LEVEL'],
self.settings['LOG_STDOUT'],
self.settings['LOG_ENCODING'])
# initialize signals
self.signals = SignalManager(self)
#initialize stats
stats_cls = load_object(self.settings.get('STATS_CLASS'))
self.stats = stats_cls(self)
# initialize downloader
self.request_queue = PriorityQueue(lambda _: MemoryQueue())
self.response_queue = ResponseQueue(
self.settings.get_int('RESPONSE_ACTIVE_SIZE_LIMIT'))
self.downloader = Downloader(self.settings,
self.request_queue,
self.response_queue,
clock=self.clock)
# initialize extensions
self.extensions = ExtensionManager(self)
# initialize downloader pipeline
self.pipeline = PipelineManager(self)
self.initialized = True
# now that everything is ready, set the spider's engine
self.spider.set_engine(self)
def qfactory(priority):
return MemoryQueue()