public void testHomePage() { Crawler crawler = new CorriereDellaSeraCrawler(); try { Collection<String> urls = crawler.retrieveArticleUrlsFromFeed(); urls.addAll(crawler.retrieveArticleUrlsFromHomePage()); Iterator<String> i = urls.iterator(); boolean result = true; while (i.hasNext()) { String s = i.next(); try { if (crawler.getArticle(s) != null) { System.out.println(" OK >> \t " + s); result &= true; } } catch (CrawlerCannotReadArticleException e) { System.err.println("ERROR >> \t " + s); result &= false; e.printStackTrace(); } catch (Exception e) { System.err.println("CRITICAL >> \t " + s); e.printStackTrace(); } } assertEquals(true, result); } catch (Exception e) { e.printStackTrace(); } }
public void testHomePageAndFeedLinks() { try { Crawler crawler = new LaRepubblicaCrawler(); Set<String> set = new HashSet<String>(); set.addAll(crawler.retrieveArticleUrlsFromFeed()); set.addAll(crawler.retrieveArticleUrlsFromHomePage()); Iterator<String> i = set.iterator(); boolean result = true; while (i.hasNext()) { String s = i.next(); try { if (crawler.getArticle(s) != null) { System.out.println(" OK >> \t " + s); result &= true; } } catch (CrawlerCannotReadArticleException e) { System.err.println("ERROR >> \t " + s); result &= false; e.printStackTrace(); } } assertEquals(true, result); } catch (Exception e) { e.printStackTrace(); } }