public void parseStreaming(InputStream is, OutputStream os, Context context) {
// HTML5 may violate XML correctness. It shouldn't, but hey, shit happens.
HtmlParser hp = new HtmlParser(XmlViolationPolicy.ALLOW);
// true streaming for max awesomeness
hp.setStreamabilityViolationPolicy(XmlViolationPolicy.FATAL);
handler.setOutput(os);
handler.setContext(context);
hp.setContentHandler(handler);
try {
hp.parse(new InputSource(is));
handler.end();
} catch (IOException | SAXException ex) {
Logger.getLogger(getClass()).error("HTML parsing failed", ex);
}
}
public Map<String, String> getFragmentParts(InputStream is, Context context) {
// HTML5 may violate XML correctness. It shouldn't, but hey, shit happens.
HtmlParser hp = new HtmlParser(XmlViolationPolicy.FATAL);
// true streaming for max awesomeness
hp.setStreamabilityViolationPolicy(XmlViolationPolicy.FATAL);
handler.setAction(Handler.Action.EXTRACT_INLINE_PARTS);
handler.setContext(context);
handler.doNotWriteDocumentStart();
hp.setContentHandler(handler);
try {
hp.parse(new InputSource(is));
} catch (IOException | SAXException ex) {
Logger.getLogger(getClass()).error("HTML parsing failed", ex);
}
return handler.getFragmentParts();
}
