Enhanced proposed fix [6c0d7aec67]: unicode text input Windows 8. More places where Unicode surrogate handling could be improved.

3 files changed, 21 insertions, 51 deletions

diff --git a/win/tkWinFont.c b/win/tkWinFont.c
index 940bc10..a281a4f 100644
--- a/win/tkWinFont.c
+++ b/win/tkWinFont.c
@@ -26,9 +26,8 @@
  * Under Windows, a "font family" is uniquely identified by its face name.
  */
 
-#define FONTMAP_SHIFT	    10
-
-#define FONTMAP_PAGES	    	(1 << (sizeof(Tcl_UniChar)*8 - FONTMAP_SHIFT))
+#define FONTMAP_SHIFT		12
+#define FONTMAP_PAGES	    	(1 << (21 - FONTMAP_SHIFT))
 #define FONTMAP_BITSPERPAGE	(1 << FONTMAP_SHIFT)
 
 typedef struct FontFamily {
@@ -743,7 +742,7 @@ void
 TkpGetFontAttrsForChar(
     Tk_Window tkwin,		/* Window on the font's display */
     Tk_Font tkfont,		/* Font to query */
-    Tcl_UniChar c,		/* Character of interest */
+    int c,			/* Character of interest */
     TkFontAttributes *faPtr)	/* Output: Font attributes */
 {
     WinFont *fontPtr = (WinFont *) tkfont;
@@ -828,7 +827,7 @@ Tk_MeasureChars(
     HFONT oldFont;
     WinFont *fontPtr;
     int curX, moretomeasure;
-    Tcl_UniChar ch;
+    int ch;
     SIZE size;
     FontFamily *familyPtr;
     Tcl_DString runString;
@@ -859,7 +858,7 @@ Tk_MeasureChars(
     start = source;
     end = start + numBytes;
     for (p = start; p < end; ) {
-	next = p + Tcl_UtfToUniChar(p, &ch);
+	next = p + TkUtfToUniChar(p, &ch);
 	thisSubFontPtr = FindSubFontForChar(fontPtr, ch, &lastSubFontPtr);
 	if (thisSubFontPtr != lastSubFontPtr) {
 	    familyPtr = lastSubFontPtr->familyPtr;
@@ -921,7 +920,7 @@ Tk_MeasureChars(
 	familyPtr = lastSubFontPtr->familyPtr;
 	Tcl_DStringInit(&runString);
 	for (p = start; p < end; ) {
-	    next = p + Tcl_UtfToUniChar(p, &ch);
+	    next = p + TkUtfToUniChar(p, &ch);
 	    Tcl_UtfToExternal(NULL, familyPtr->encoding, p,
 		    (int) (next - p), 0, NULL, buf, sizeof(buf), NULL,
 		    &dstWrote, NULL);
@@ -970,13 +969,13 @@ Tk_MeasureChars(
 	 */
 
 	const char *lastWordBreak = NULL;
-	Tcl_UniChar ch2;
+	int ch2;
 
 	end = p;
 	p = source;
 	ch = ' ';
 	while (p < end) {
-	    next = p + Tcl_UtfToUniChar(p, &ch2);
+	    next = p + TkUtfToUniChar(p, &ch2);
 	    if ((ch != ' ') && (ch2 == ' ')) {
 		lastWordBreak = p;
 	    }
@@ -1443,7 +1442,7 @@ MultiFontTextOut(
 				 * string when drawing. */
     double angle)
 {
-    Tcl_UniChar ch;
+    int ch;
     SIZE size;
     HFONT oldFont;
     FontFamily *familyPtr;
@@ -1458,7 +1457,7 @@ MultiFontTextOut(
 
     end = source + numBytes;
     for (p = source; p < end; ) {
-	next = p + Tcl_UtfToUniChar(p, &ch);
+	next = p + TkUtfToUniChar(p, &ch);
 	thisSubFontPtr = FindSubFontForChar(fontPtr, ch, &lastSubFontPtr);
 	if (thisSubFontPtr != lastSubFontPtr) {
 	    if (p > source) {
diff --git a/win/tkWinKey.c b/win/tkWinKey.c
index 2698c4d..7fee101 100644
--- a/win/tkWinKey.c
+++ b/win/tkWinKey.c
@@ -88,6 +88,8 @@ TkpGetString(
 				 * result. */
 {
     XKeyEvent *keyEv = &eventPtr->xkey;
+	int len;
+	char buf[6];
 
     Tcl_DStringInit(dsPtr);
     if (keyEv->send_event == -1) {
@@ -95,41 +97,14 @@ TkpGetString(
 	    Tcl_ExternalToUtfDString(TkWinGetKeyInputEncoding(),
 		    keyEv->trans_chars, keyEv->nbytes, dsPtr);
 	}
-    } else if (keyEv->send_event == -2) {
-	/*
-	 * Special case for win2000 multi-lingal IME input. xkey.trans_chars[]
-	 * already contains a UNICODE char.
-	 */
-
-	int unichar;
-	char buf[XMaxTransChars];
-	int len;
-
-	unichar = keyEv->trans_chars[1] & 0xff;
-	unichar <<= 8;
-	unichar |= keyEv->trans_chars[0] & 0xff;
-
-	len = Tcl_UniCharToUtf((Tcl_UniChar) unichar, buf);
-
-	Tcl_DStringAppend(dsPtr, buf, len);
     } else if (keyEv->send_event == -3) {
 
-	char buf[XMaxTransChars];
-	int len;
-
 	/*
-	 * Special case for WM_UNICHAR.
+	 * Special case for WM_UNICHAR and win2000 multi-lingal IME input
 	 */
 
-	len = Tcl_UniCharToUtf(keyEv->keycode, buf);
-	if ((keyEv->keycode <= 0xffff) || (len == XMaxTransChars)) {
-	    Tcl_DStringAppend(dsPtr, buf, len);
-	} else {
-	    Tcl_UniCharToUtf(((keyEv->keycode - 0x10000) >> 10) | 0xd800, buf);
-	    Tcl_DStringAppend(dsPtr, buf, 3);
-	    Tcl_UniCharToUtf(((keyEv->keycode - 0x10000) & 0x3ff) | 0xdc00, buf);
-	    Tcl_DStringAppend(dsPtr, buf, 3);
-	}
+	len = TkUniCharToUtf(keyEv->keycode, buf);
+	Tcl_DStringAppend(dsPtr, buf, len);
     } else {
 	/*
 	 * This is an event generated from generic code. It has no nchars or
@@ -140,9 +115,6 @@ TkpGetString(
 
 	if (((keysym != NoSymbol) && (keysym > 0) && (keysym < 256))
 		|| (keysym == XK_Return) || (keysym == XK_Tab)) {
-	    char buf[XMaxTransChars];
-	    int len;
-
 	    len = Tcl_UniCharToUtf((Tcl_UniChar) (keysym & 255), buf);
 	    Tcl_DStringAppend(dsPtr, buf, len);
 	}
diff --git a/win/tkWinX.c b/win/tkWinX.c
index bb60a2d..987fbb5 100644
--- a/win/tkWinX.c
+++ b/win/tkWinX.c
@@ -1580,8 +1580,8 @@ HandleIMEComposition(
 	/*
 	 * Set up the fields pertinent to key event.
 	 *
-	 * We set send_event to the special value of -2, so that TkpGetString
-	 * in tkWinKey.c knows that trans_chars[] already contains a UNICODE
+	 * We set send_event to the special value of -3, so that TkpGetString
+	 * in tkWinKey.c knows that keycode already contains a UNICODE
 	 * char and there's no need to do encoding conversion.
 	 *
 	 * Note that the event *must* be zeroed out first; Tk plays cunning
@@ -1592,7 +1592,7 @@ HandleIMEComposition(
 
 	memset(&event, 0, sizeof(XEvent));
 	event.xkey.serial = winPtr->display->request++;
-	event.xkey.send_event = -2;
+	event.xkey.send_event = -3;
 	event.xkey.display = winPtr->display;
 	event.xkey.window = winPtr->window;
 	event.xkey.root = RootWindow(winPtr->display, winPtr->screenNum);
@@ -1600,8 +1600,7 @@ HandleIMEComposition(
 	event.xkey.state = TkWinGetModifierState();
 	event.xkey.time = TkpGetMS();
 	event.xkey.same_screen = True;
-	event.xkey.keycode = 0;
-	event.xkey.nbytes = 2;
+	event.xkey.nbytes = 0;
 
 	for (i=0; i<n; ) {
 	    /*
@@ -1609,8 +1608,8 @@ HandleIMEComposition(
 	     * UNICODE character in the composition.
 	     */
 
-	    event.xkey.trans_chars[0] = (char) buff[i++];
-	    event.xkey.trans_chars[1] = (char) buff[i++];
+	    event.xkey.keycode = ((unsigned char) buff[i++]) << 8;
+	    event.xkey.keycode += (unsigned char) buff[i++];
 
 	    event.type = KeyPress;
 	    Tk_QueueWindowEvent(&event, TCL_QUEUE_TAIL);