def test_extract_icon_url(self, get_icon_patch): def return_first_val(*a, **kwargs): return a[0] get_icon_patch.side_effect = return_first_val self.assertEqual('http://www.pariszigzag.fr/wp-content/themes' '/paris_zigzag_2016/favicon.ico', extract_icon_url(self.article)) self.assertEqual('https://s.ytimg.com/yts/img/favicon-vflz7uhzw.ico', extract_icon_url(self.article2))
def parse_webpage(self): result = {'site_link': self.page_response.url} icon_url = extract_icon_url(self.page_response) if icon_url: result['icon_url'] = icon_url links = list(extract_feed_links(self.page_response)) if links: result['link'] = links[0] if len(links) > 1: result['links'] = links result['title'] = extract_opg_prop(self.page_response, og_prop='og:site_name') if not result['title']: result['title'] = extract_title(self.page_response) return {key: value for key, value in result.items() if value}