Scan Codes Not Unique?

[Apollys]

Since this issue, I've switched to using scan codes for all hotkey-related things. Things were going smoothly until...

D:\>python -m keyboard
{"event_type": "down", "scan_code": 82, "name": "insert", "time": 1526769257.152305, "is_keypad": false}
{"event_type": "up", "scan_code": 82, "name": "insert", "time": 1526769257.224113, "is_keypad": false}
{"event_type": "down", "scan_code": 82, "name": "0", "time": 1526769258.5705125, "is_keypad": true}
0{"event_type": "up", "scan_code": 82, "name": "0", "time": 1526769258.6413236, "is_keypad": true}

How do I distinguish between keys with the same "scan_code" but different "is_keypad" values in keyboard.add_hotkey() and keyboard.send()?

[boppreh]

In this case I blame the operating system for reporting two separate physical keys with the same scan code. Windows has "virtual key codes", and these keys do indeed report separate values. You could hack something yourself with that, but the concept is not completely portable.

I plan to add support for numpad keys in the future, so you could add_hotkey('numpad 5', ...). It could help in this case, but I'm afraid there's no good solution at the moment.

[Apollys]

Well obviously we can distinguish the two events in the context of adding a hotkey, because the events also carry the is_keypad boolean.

In terms of sending, I'm not too familiar but again there's a clear difference, so if we could just set the is_keypad boolean in the sent event then that would handle it. I don't know if this one is easy though.

[boppreh]

Yes, the feature I'm looking to add is the ability of hotkey listeners to check the numpad flag.

For sending, however, things are more complicated. The numpad flag is derived from a list of signals; it's not an OS information. If you are on Windows, I suggest taking a look into the code that sends events. It's just a couple of lines long, and you can tweak to provide your own virtual key code, which does differentiate the two inserts correctly.

I know it's not ideal, but it's an easy way to fix your problem.

[boppreh]

Ah, one more thing. Due to a variety of situations where Windows does something dumb (e.g. scan code = 0), the library is capable of sending raw virtual key codes if you provide a negative number for the scan code. So 'send(-72)' will send an event for virtual key code 72, not scan code. You just have to check what are the virtual key codes for the keys you want.

[Apollys]

I will look into that, thanks for the rapid feedback.