Jaeyoun Chung's mysterious patch that apparently makes Korean input

[sgt/putty] / window.c

diff --git a/window.c b/window.c
index 54665c9..a8fe7ab 100644 (file)
--- a/window.c
+++ b/window.c
@@ -2384,7 +2384,7 @@ static LRESULT CALLBACK WndProc(HWND hwnd, UINT message,
            HIMC hIMC;
            int n;
            char *buff;
-   
+
            if(osVersion.dwPlatformId == VER_PLATFORM_WIN32_WINDOWS || 
                osVersion.dwPlatformId == VER_PLATFORM_WIN32s) break; /* no Unicode */
 
@@ -2395,9 +2395,17 @@ static LRESULT CALLBACK WndProc(HWND hwnd, UINT message,
            n = ImmGetCompositionStringW(hIMC, GCS_RESULTSTR, NULL, 0);
 
            if (n > 0) {
+               int i;
                buff = (char*) smalloc(n);
                ImmGetCompositionStringW(hIMC, GCS_RESULTSTR, buff, n);
-               luni_send((unsigned short *)buff, n / 2, 1);
+               /*
+                * Jaeyoun Chung reports that Korean character
+                * input doesn't work correctly if we do a single
+                * luni_send() covering the whole of buff. So
+                * instead we luni_send the characters one by one.
+                */
+               for (i = 0; i < n; i += 2)
+                   luni_send((unsigned short *)(buff+i), 1, 1);
                free(buff);
            }
            ImmReleaseContext(hwnd, hIMC);