More VTD-XML progress
I did some more hacking on my Python implementation the other night, and am starting to hack out a usable API.
While I haven’t enhanced the parser in any significant way, I have added the ability to select nodes via simple XPath expressions. It handles most (but not all) axes, the shortened axis identifers, and chaining of steps into a path. For example, I can evaluate “name/ancestor::label/@added” and get the indices of all the matching nodes in document order.
I’m now looking at two things - figuring out how to profile this beast, and looking at Python parser generators so that I can implement the entire XPath grammar. After that, I need to get a deeper understanding of Python iterators so I can improve the just-in-time evaluation of some things.