There is a bunch of 'parse rendering' code in ParserOutput::getText() which has no place in (what we hope will be) a plain value object.
Further, getText() operates directly on the HTML as a string; in the Parsoid world it should be operating (and returning) a DOM tree instead.
Related methods: ParserOutput::addWrapperDivClass(), ::clearWrapperDivClass().
Perhaps the "postprocessing" step here belongs somewhere in the Content hierarchy?