How to track events and windows?

[Tomas carnecky]

I'm working on a small project [1] that needs to hook a few X11 (XEvent 
and friends) and GLX (glXSwapBuffers) functions. The idea is that when 
the user presses a predefined hotkey, the library saves the Display and 
Window XID and later any call to glXSwapBuffers with the correct 
arguments will cause that function to save the framebuffer and write it 
to a file. I first took the Display variable passed to the XEvent 
functions and "event->xkey.window" as the Window XID. This works for 
simple X11 applications that open only one connection to the server and 
use only one window (for both drawing and events).
But unfortunately there are apps that handle events in a more 
complicated way: libSDL, cedega, wine are just a few examples. They use 
two or more different windows and each app does it in a different way. 
That's why I thought I could query the window that currently has focus 
and take its XID. But too bad that the window that has input focus is 
not always the window that the apps uses for drawing. And what's even 
stranger, when I query the input focus window after I receive a 
key-event, then the focus window doesn't match the window from the 
event. That means, if I understand it correctly, that a window _without_ 
input focus received a key event. How is that possible?
A last way to get the Window XID would be to query the window that is 
under the mouse but that wouldn't work for apps that are not fullscreen 
and where the user has the mouse outside of the window.

Does anyone have an idea how I could properly get the Window XID?

thanks
tom

[1] http://www.neopsis.com/projects/yukon/