Extracting microdata is fairly tedious compared to microformats, due to
its generality. Should probably be included as a standard function of
the parser, or even better, as a separate module, which then can
concern itself with being fully standards compliant.