TeStereo report



@Test @Ignore("The file 'testXHTML_utf8.html' is not available fo testing") public void XtestParseUTF8() throws IOException, SAXException, TikaException {
  String path="/test-documents/testXHTML_utf8.html";
  Metadata metadata=new Metadata();
  String content=new Tika().parseToString(HtmlParserTest.class.getResourceAsStream(path),metadata);
  assertTrue("Did not contain expected text:" + "Title : Tilte with UTF-8 chars √∂√§√•",content.contains("Title : Tilte with UTF-8 chars √∂√§√•"));
  assertTrue("Did not contain expected text:" + "Content with UTF-8 chars",content.contains("Content with UTF-8 chars"));
  assertTrue("Did not contain expected text:" + "√•√§√∂",content.contains("√•√§√∂"));
}