String data = "<!DOCTYPE html>\n<html><body><p>test <span>content</span></p></body></html>";
InputStream stream = new ByteArrayInputStream(data.getBytes("UTF-8"));
Writer writer = new StringWriter();
ContentHandler contentHandler = new BodyContentHandler(writer);
Metadata metadata = new Metadata();
Detector contentTypeDetector = new DefaultDetector(classLoader);
MediaType type = contentTypeDetector.detect(stream, metadata);
assertEquals(type.toString(), "text/html");
metadata.add(Metadata.CONTENT_TYPE, type.toString());
ParseContext parseCtx = new ParseContext();
parser.parse(stream, contentHandler, metadata, parseCtx);
writer.flush();