def __init__(self, spider):
self.spider = spider
self.logger = spider.logger
self.scheduler = Scheduler(spider)
self.settings = spider.settings
max_request_size = self.settings["MAX_REQUEST_SIZE"]
self.dynamic = self.settings["DYNAMIC_CRAWL"]
if self.dynamic:
module_path = DRIVER_MODULE
module = import_module(module_path)
init_kwargs = self.settings['DRIVER_INIT_KWARGS']
self.driver = getattr(module,
self.settings.get('DRIVER').title())(**init_kwargs)
else:
self.driver = None
self.driver_sem = BoundedSemaphore(1)
self.downloader = Downloader(spider, self.driver, self.driver_sem)
self.pool = Pool(size=max_request_size)
评论列表
文章目录