'Easier' than with the old code base. poppler has code to handle text
flow and extents, unlike xpdf. This will let us write caret support and
stream the text to ATs. I don't think that the caret will be hard to
add to the code base. And simply reading the visible area without the
caret is even easier.
As I said, this is now a pretty straightforward project -- it just needs
someone to tackle it.
The 'accessibility' features of PDF docs themselves are triviablly
extractable too, but they really need the rest of the framework before
doing so. For example, the alt tag is pretty useless by itself.
Thanks,
-Jonathan