Beispiel #1
0
    def __init__(self, worker_id, task_queue, completed_queue):
        ConsumerThread.__init__(self, worker_id, task_queue, completed_queue)

        self._storage = DatabaseStorage()
        self._crawler = GenericSourceCrawler(Scraper([GenericSourceParser]), Scraper([RSSParser]))

        topic_scraper = Scraper([RSSParser, GenericFrontPageParser])
        article_scraper = Scraper([TheOnionParser, EngadgetParser, TechCrunchParser, TorontoStarParser, TheGlobeAndMailParser, GenericContentParser])

        sub_task = ArticleScrapingTask(article_scraper)
        self._task = TopicScrapingTask(topic_scraper, sub_task)
Beispiel #2
0
 def __init__(self, worker_id, task_queue, completed_queue):
     ConsumerThread.__init__(self, worker_id, task_queue, completed_queue)
     self._library = AlgorithmLibrary([DummyAlgorithm(), KeywordAlgorithm()])
Beispiel #3
0
 def __init__(self, worker_id, task_queue, completed_queue, algorithm):
     ConsumerThread.__init__(self, worker_id, task_queue, completed_queue)
     self._algorithm = algorithm()