I can't even decide whether the text (e.g. in a div or p element) should be stored as a unit with links stored as meta-data, or should the stretches of text before/after/between links be "chunks" and the links store their own content (like DOM)?

It's an old question, see the "piece-chain" structure Prof With chose for text-with-"looks" (aka "style") http://www.inf.ethz.ch/personal/wirth/ProjectOberon/PO.Syste... section 5.2, "Text Management".