class ResultFilter(object):
def __init__(self, results_class):
self._rf_worker = Worker()
self._results_class = results_class
self.thresholds = [20,1,1]
def ready_to_shutdown(self):
return self._rf_worker.ready_to_shutdown()
def submit_ranker_result(self, rank_result):
"""
This function is used by Ranker.
Takes [page_url, page_weight, [(link_soup,weight)]] and passes it to rf_worker
as
[page_url, page_weight, [(link_soup,weight)], self.thresholds, self._results_class]
"""
args = list(rank_result) + [self.thresholds, self._results_class]
rfw_task = WorkerTask(args, result_filter_routine, args[0])
self._rf_worker.add_task(rfw_task)
def get_result(self, filter_match = (lambda x: True)):
"""
Returns [page_url, page_weight, [(link,weight)]]
"""
completed_rfw_task = self._rf_worker.get_completed_task(filter_match)
if completed_rfw_task is None:
return None
return completed_rfw_task.result
def purge_tasks(self, filter_not_match):
"""
Removes all tasks, for which <filter_not_match> returns False.
"""
self._rf_worker.purge_tasks(filter_not_match)
