OnKeyDown: getting the charcater value from key

Dan Rathbun

the flag value is for me totally cryptic (in addition it is very different on keydown and keyup for the same key).

Oh yes it is quite clear that there are issues, and it IS a pain in the butt.

I remember vaguely someone explaining how the flag bit masks worked, and how to | and & them properly.

It would be nicve if the API creates a KeyFlag subclass of integer with methods like:
.shift? .control? .command? .alt?
... etc.

Anton_S

@fredo6 said:

But what I am trying to do is simplyget a natural character(like the digits [0-9)) and I really do not understand the logic.

@fredo6 said:

My goal was to see if the keys could be transformed into characters in order to trap the 0-9, =, +, -, *, / characters.

The tool onKeyDown method returns virtual key constant value, key repeat, key flags, and view instance.

Here is the list of Virtual Key Codes on Windows: http://msdn.microsoft.com/en-us/library/windows/desktop/dd375731(v=vs.85).aspx

key.chr will return the character of the value.
for instance 87.chr returns 'W'

You can filter the keys like this:

def onKeyDown(key, repeat, flags, view)
  if %w(1 2 3 4 5 6 7 8 9 0 = + - * /).include?(key.chr)
    # process code here
  else
    return
  end
end

How to convert numeric decimal to hexadecimal:
87.to_s(16) # => '57' or 0x57 which can be found in the virtual key codes list.

When viewing flags you might first want to convert them to hexadecimal.
The flags don't give you anything special about the key.
Here is what they give: http://msdn.microsoft.com/en-us/library/windows/desktop/ms646267(v=vs.85).aspx#_win32_Keystroke_Message_Flags

Anton_S

Apparently key.chr doesn't return proper characters for all values. For that, we might want to sett up a hash of values and characters.

CHARS = {
# Standard keys
0x30 => '0'
0x31 => '1'
0x32 => '2'
0x33 => '3'
0x34 => '4'
0x35 => '5'
0x36 => '6'
0x37 => '7'
0x38 => '8'
0x39 => '9'
0xBD => '-'
0xBB => '+'
0x0D => '=' # use enter key for =
# Numpads
0x60 => '0'
0x61 => '1'
0x62 => '2'
0x63 => '3'
0x64 => '4'
0x65 => '5'
0x66 => '6'
0x67 => '7'
0x68 => '8'
0x69 => '9'
0x6B => '+'
0x6D => '-'
0x6A => '*'
0x6F => '/'
}

So the filter set up would look like this now.

def onKeyDown(key, repeat, flags, view)
  key_char = CHARS[key]
  return unless key_char
  # process
end

This technique should work on Windows. I don't know about Macs though.

fredo6

Anton,

Thanks very much.
I have the virtual keycodes on Mac too.

The main issue is that these map physical keys, so it will be very difficult to have a reliable result for non-US keyboards for the the digits for instance, which may require a Shift (like on French keyboard).

And you are right that flags does not bring much info.

Fredo

tt_su

Yea, the Tool class return key codes, not character codes. And they are system dependant.

I'm not sure how you'd get the resulting character that would be produced, though should think there should be some system API for this. I'd have to search for that.

Dan Rathbun

@Anton: Would using ToUnicode() be better than ToAscii() ?

Anton_S

@dan rathbun said:

@Anton: Would using ToUnicode() be better than ToAscii() ?

I just tested and seen no difference. I think it doesn't matter much.

tt_su

@dan rathbun said:

@Anton: Would using ToUnicode() be better than ToAscii() ?

Indeed this appear to be the correct function to use:
http://msdn.microsoft.com/en-us/library/windows/desktop/ms646320(v=vs.85.aspx

@unknownuser said:

Translates the specified virtual-key code and keyboard state to the corresponding Unicode character or characters.

Though, it probably produce a UTF-16LE string, transposing to UTF-8 would probably be best for interoperability with the rest of the SketchUp environment.

Anton_S

Nah, both of them are encoded in US-ASCII.
adding char.force_encoding('UTF-8') should do the tick, aint it?

tt_su

@anton_s said:

Nah, both of them are encoded in US-ASCII.

You're talking about ToAscii?
I was looking at ToUnicode.

@anton_s said:

adding char.force_encoding('UTF-8') should do the tick, aint it?

I'd be careful with force_encoding - it should only be used if you have raw binary data representing a string or if you know the string is incorrectly encoded. Otherwise one should transpose the en encoding.
foo = bar.encode("UTF-8")

Anton_S

@tt_su said:

You're talking about ToAscii?
I was looking at ToUnicode.

Yes, the ToUnicode and ToAscii both return same formatted Ruby strings, but on the actual approach they don't. When I initially created a buffer char = 0.chr*2 it was already formatted in US-ASCII. The ToUnicodeand ToAscii simply fills the the specified buffer with ascii/unicode characters without changing the format of the Ruby string.

Changing char = 0.chr*2 to char= ' ' will initially create a UTF-8 formatted string, rather than the US-ASCII. The format remains unchanged when the ToUnicode/ToAscii functions fills the char string.

@tt_su said:

I'd be careful with force_encoding...

For some reason force_encoding even managed to crash SketchUp at some points when testing key_test.

Dan Rathbun

Ya'all need to be careful where testing string encoding, until MR1 comes out.
The correct default encoding is not properly set in MR0.

driven

@dan rathbun said:

The correct default encoding is not properly set in MR0.

if anyones prepared to admit it...

the other issue is why these things are so difficult on a mac...

apple doesn't support key logging and goes out of it's way to prevent it occurring...

apple does allow it IF it's for assistive devices to aid those with disabilities...

if you want to use 'assistive devices api' to do ordinary tasks, you need to connivence the 'admin' user to turn it on.

mac folk are reluctant to do this as it's system wide, and would enable malicious 'key loggers' to, key-log...

john

Dan Rathbun

Err.. found a Ruby Core bug.
Even after setting Encoding::default_internal="UTF-8"
the Integer#chr() method does not return a string in the default internal encoding.

.. however there is a workaround. Ruby 2.0's Integer#chr() method now takes an encoding argument, so, this:
char = 0.chr(Encoding::default_internal)*2
or this:
char = 0.chr("UTF-8")*2
WILL work.

Anton_S

@dan rathbun said:

char = 0.chr("UTF-8")*2

Nice find Dan I re-edited the Ruby code above.

Dan Rathbun

@anton_s said:

@dan rathbun said:

char = 0.chr("UTF-8")*2

Nice find Dan I re-edited the Ruby code above.

But you did not enclosed the encoding nickname string in quotes. That'll raise a NameError exception ("uninitialized constant UTF".)

Anton_S

Okay, I figured a way to get key via the Windows API. The code below will give you the key code of the exact character that your pressing. For instance,

Pressing 1 yields 1; Pressing Shift-1 returns !.
Pressing w yields w; Pressing Shift-w returns W.

I guess it should work properly on the Français keyboard as well. Test it out.

require 'Win32API'

class KeyTest

  GetKeyboardState = Win32API.new('User32', 'GetKeyboardState', 'P', 'L')
  ToAscii          = Win32API.new('User32', 'ToAscii', 'LLPPL', 'L')
  ToUnicode        = Win32API.new('User32', 'ToUnicode', 'LLPPLL', 'L')

  def onKeyDown(key, repeat, flags, view)
    buffer = 0.chr*256
    char = 0.chr('UTF-8')*2 # maintain UTF-8 encoding
    GetKeyboardState.call(buffer)
    res = ToUnicode.call(key, repeat, buffer, char, char.size, 0)
    #res = ToAscii.call(key, repeat, buffer, char, 0)
    return if res == 0
    char.strip!
    p char
    # process
  end

end

unless file_loaded?(__FILE__)
  UI.menu('Plugins').add_item('Key Test'){
    Sketchup.active_model.select_tool(KeyTest.new)
  }
end

You need Win32API.so for SU2013 and below though.

Place in the Plugins folder. Access: Plugins > Key Test

Anton_S

@dan rathbun said:

. That'll raise a NameError exception ("uninitialized constant UTF".)

Oh Fixed

tt_su

@dan rathbun said:

Err.. found a Ruby Core bug.
Even after setting Encoding::default_internal="UTF-8"
the Integer#chr() method does not return a string in the default internal encoding.

sigh "Unicode support" grumble kicks Ruby on the shin

Dan Rathbun

Yea did you look at the C code by clicking the method in the CHM ?
It has a kind of convoluted switch statement.