I don't know about libxml, but normally you'd not need to do any I/O once you've gotten hold of the raw HTML - so there should be no need for callbacks. E.g. with cheerio you can parse HTML synchronously.