You will need to use widget geometry for this, otherwise it’s going to get really ugly.
The mouse event spits out screen space, too - you translated it into the local coordinates of the widget (you may need to do it more than once if you have a nested hierarchy - windows inside windows…)
Once you have local coords and know world coords → MapRange is your friend.
Anyway, I’m not overriding the mouse functions as I’m using Input Actions instead,
Perhaps you could explain the reasoning behind this choice. Sometimes there is a good one, other times it’s a dead end. Widgets are fine with processing input and this produces the necessary data for a lot of operations.