def test_reject__ng(self): file = open(os.path.join(os.path.dirname(__file__), "..", "..", "filter_cases_ng.txt")) for line in file: url = line.strip() self.assertEqual( (url, True), (url, IronnewsUtility.reject(url))) file.close()
from google_news import GoogleNews from ironnews_utility import IronnewsUtility print "Content-Type: text/plain" print "" KeywordManager.initialize() keyword = KeywordManager.get() print keyword.encode("utf-8") articles = GoogleNews.search(keyword, 30) for article in articles: url = article["url"] title = article["title"] print "---" print url print title.encode("utf-8") if IronnewsUtility.reject(url): print "reject!" continue url2 = IronnewsUtility.get_canonical_url(url) if url2 != url: print "canonical! " + url2 ArticleManager.add(url2, title, Article.CATEGORY_RAIL) KeywordManager.update(keyword)