The easiest way I can think of would just be to have an array where the value at the index of the qwerty key they pressed is the azerty key they pressed. And don't forget about us dvorak users!
Well, the program I wrote should work with any keyboard layout that uses latin characters, so dvorak should be fine
The problem with having such an array is that you would have to ask the user "which keyboard layout are you using?", which does not feel very professional. (Or you'd have to find out from digging around in the windows registry. Which may be a nightmare considering Apple keyboards for example have their own Windows layouts!)
Also, I cited AZERTY and QWERTZ as examples because they are widespread, but you would also have to create an array for 'exotic' keyboard layouts... For example, the French equivalent of dvorak, "bépo" is completely different from anything out there. If each country has its own optimized keyboard layout, making an array for every one of them would be pretty difficult.
It would be elegantly solved if SFML itself provided a way to use "physical" scancodes.
Plus, it would solve the problem that some AZERTY users face with unusable keys in SFML such as é, à, !, etc., well, pretty much any symbol key. (And I believe QWERTZ users can't use ü either, swedish users can't use å, and so on.)
EDIT: I meant they can't use these keys with IsKeyPressed (I know you can use something to get unicode text that just got typed in, but this is not meant for 'in-game' use)