Actually, let me revise my rpevious comment. I think we should fake the new interface for now by adding a TreeEventBuilder that requires having its own TreeBuilder internally, instead of wrapping an arbitrary target. That way, we can avoid having to clean up the existing dependency of the C parser on the C-level TreeBuilder implementation.

Once this cleanup gets done, we can add the support for wrapping user defined targets as a feature.

So, basically, I propose to take the route that my TreeEventBuilder patch went, just with a simplified implementation. I'll see if I can cut down the patch to those essentials.

Would you be ok with something that uses the isinstance() hack in the parser?