@Test public void testGetDomainName() throws Exception { URL url = null; url = new URL("http://nutch.apache.org"); assertEquals("apache.org", URLUtil.getDomainName(url)); url = new URL("http://en.wikipedia.org/wiki/Java_coffee"); assertEquals("wikipedia.org", URLUtil.getDomainName(url)); url = new URL("http://140.211.11.130/foundation/contributing.html"); assertEquals("140.211.11.130", URLUtil.getDomainName(url)); url = new URL("http://www.example.co.uk:8080/index.html"); assertEquals("example.co.uk", URLUtil.getDomainName(url)); url = new URL("http://com"); assertEquals("com", URLUtil.getDomainName(url)); url = new URL("http://www.example.co.uk.com"); assertEquals("uk.com", URLUtil.getDomainName(url)); // "nn" is not a tld url = new URL("http://example.com.nn"); assertEquals("nn", URLUtil.getDomainName(url)); url = new URL("http://"); assertEquals("", URLUtil.getDomainName(url)); url = new URL("http://www.edu.tr.xyz"); assertEquals("xyz", URLUtil.getDomainName(url)); url = new URL("http://www.example.c.se"); assertEquals("example.c.se", URLUtil.getDomainName(url)); // plc.co.im is listed as a domain suffix url = new URL("http://www.example.plc.co.im"); assertEquals("example.plc.co.im", URLUtil.getDomainName(url)); // 2000.hu is listed as a domain suffix url = new URL("http://www.example.2000.hu"); assertEquals("example.2000.hu", URLUtil.getDomainName(url)); // test non-ascii url = new URL("http://www.example.商業.tw"); assertEquals("example.商業.tw", URLUtil.getDomainName(url)); }