Example #1
0
    def __process_goose(self):
        goose_config = Configuration()
        goose_config.browser_user_agent = 'Mozilla 5.0'
        goose_config.enable_image_fetching = True
        g = Goose(config=goose_config)
        try:
            article = g.extract(self.url)

            if article.top_image.src:
                self.images = self.get_all_images_from_example_src(
                    article.top_image.src)

        except (requests.exceptions.ConnectionError,
                requests.exceptions.ReadTimeout):
            return None
        return article
Example #2
0
 def getConfig(self):
     config = Configuration()
     config.enable_image_fetching = False
     return config