you'd write an annotation that let you specify a CSS selector, write a converter which read that annotation, used JSoup to parse the HTML, and then extracted the text of the selected element, and then the converter would delegate to the JSON converter to deserialize that text to the requested body type