Specifically for panning and zooming, doesn't the OS translate those inputs to mouse events, like Windows does by default? Otherwise it is simply a matter of performing this translation at the backend level.