see also :
Telsa Gwynn's summary of the ScrollKeeper talk at GUADEC "Scrollkeeper is the middle layer: it abstracts all metadata handling into a library. It extracts data, stores it in a database, and provides an API for help browsers to talk to. ... It's a mixture of C, shell scripts, and libxml. You feed it documents and an OMF metadata file (in XML). It spits out normal and extended trees in XML for every locale."