@GeoffR: AFAICT, it currently doesn't even *add* those divs
.
So I went ahead and took a hatchet to the code (I've never used lxml, and I'm far from fluent in Python to begin with, so this may very well be awful, code-wise).
My highly scientific test with a single book would point towards success!
That said, I'm also using your (synced) kepub patch for that very issue, so
.
EDIT: It appears to be screwing up books using div's instead of p's... On the other hand, those books also don't suffer from the cut-off bug on my device... :?
EDITē: So, take two, with a crappy heuristic to, well, not do it on books that appear to use div tags to handle paragraphs...