Windows Vista Speech Recognition and Cygwin/X

[Jon TURNEY]

Paul Loewenstein wrote:
> The XWin server  uses primarily hardware scan codes for interpreting the 
> windows WM_KEYDOWN and WM_SYSKEYDOWN  messages. Unfortunately, Vista 
> speech recognition (WSR) doesn't bother to fill in the scancode field in 
> the lParam entry.  Neither does SendKeys.SendWait(),  which is what 
> Vocola 3, a recently released supplement to WSR, uses to send input to 
> applications.
> 
> To work around these Microsoft bugs, I modified winkeybd.c as follows:
> 
> void
> winTranslateKey (WPARAM wParam, LPARAM lParam, int *piScanCode)
> {
> 
>  int        iKeyFixup = g_iKeyMap[wParam * WIN_KEYMAP_COLS + 1];
>  int        iKeyFixupEx = g_iKeyMap[wParam * WIN_KEYMAP_COLS + 2];
>  int        iParamScanCode = LOBYTE (HIWORD (lParam));
> 
> 
> to
> 
> void
> winTranslateKey (WPARAM wParam, LPARAM lParam, int *piScanCode)
> {
> 
>  HKL        dwhkl = GetKeyboardLayout(0);
> 
>  int        iKeyFixup = g_iKeyMap[wParam * WIN_KEYMAP_COLS + 1];
>  int        iKeyFixupEx = g_iKeyMap[wParam * WIN_KEYMAP_COLS + 2];
> 
>  int        iParamScanCode = MapVirtualKeyEx (wParam, 
> /*MAPVK_VK_TO_VSC*/0, dwhkl);
> 
> so that the scan code is regenerated from the keycode.

diff is your friend :-)

> An initial test appears to work. However, it is not clear to me whether 
> I have introduced some subtle bugs. For example, its correctness depends 
> on extended codes having the same scan code as the non-extended 
> equivalent on all keyboard layouts.

How about writing it in the following form, that should address that concern, 
if I've understood correctly (in all cases, assuming 0 is never a valid scancode)

$ quilt diff
Index: xorg-server-1.5.3/xserver/hw/xwin/winkeybd.c
===================================================================
--- xorg-server-1.5.3.orig/xserver/hw/xwin/winkeybd.c   2009-01-05 
23:50:04.000000000 +0000
+++ xorg-server-1.5.3/xserver/hw/xwin/winkeybd.c        2009-01-10 
21:55:40.000000000 +0000
@@ -80,6 +80,12 @@
    int          iKeyFixupEx = g_iKeyMap[wParam * WIN_KEYMAP_COLS + 2];
    int          iParamScanCode = LOBYTE (HIWORD (lParam));

+  /* WM_ key messages faked by speech recognizers don't have a scan code... */
+  if (iParamScanCode == 0)
+    {
+      iParamScanCode = MapVirtualKeyEx(wParam, /*MAPVK_VK_TO_VSC*/0, 
GetKeyboardLayout(0));
+    }
+
    /* Branch on special extended, special non-extended, or normal key */
    if ((HIWORD (lParam) & KF_EXTENDED) && iKeyFixupEx)
      *piScanCode = iKeyFixupEx;

> Can someone tell me if I've broken anything?  Or suggest a more robust 
> fix?  It would be nice to be able to release it rather than having a 
> privately distributed modification solely for Vista speech recognition.

Well, my keyboard seems to still work with the above change.  I don't think 
there will be a problem including a patch like this, provided we can be fairly 
confident it hasn't broken normal keyboards :-)


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://x.cygwin.com/docs/
FAQ:                   http://x.cygwin.com/docs/faq/