final String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);
        logger.info(content);
        assertTrue(content.contains("テスト"));
    }

    public void test_getTika_xml_broken() throws UnsupportedEncodingException {
        final InputStream in = new ByteArrayInputStream("<?xml encoding=\"UTF-8\"/><hoge>テスト<br></hoge>".getBytes(Constants.UTF_8));