Fix non-BMP character entry through the inline IME.

3 files changed, 25 insertions, 22 deletions

diff --git a/src/UniConversion.cxx b/src/UniConversion.cxx
index dea069843..0d69d9969 100644
--- a/src/UniConversion.cxx
+++ b/src/UniConversion.cxx
@@ -17,7 +17,6 @@ using namespace Scintilla;
 namespace Scintilla {
 #endif
 
-enum { SURROGATE_LEAD_FIRST = 0xD800 };
 enum { SURROGATE_TRAIL_FIRST = 0xDC00 };
 enum { SURROGATE_TRAIL_LAST = 0xDFFF };
 enum { SUPPLEMENTAL_PLANE_FIRST = 0x10000 };
@@ -43,7 +42,7 @@ unsigned int UTF8Length(const wchar_t *uptr, unsigned int tlen) {
 }
 
 void UTF8FromUTF16(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned int len) {
-	int k = 0;
+	unsigned int k = 0;
 	for (unsigned int i = 0; i < tlen && uptr[i];) {
 		unsigned int uch = uptr[i];
 		if (uch < 0x80) {
@@ -67,7 +66,8 @@ void UTF8FromUTF16(const wchar_t *uptr, unsigned int tlen, char *putf, unsigned
 		}
 		i++;
 	}
-	putf[len] = '\0';
+	if (k < len)
+		putf[k] = '\0';
 }
 
 unsigned int UTF8CharLength(unsigned char ch) {
diff --git a/src/UniConversion.h b/src/UniConversion.h
index 8c7ac4a27..08898cac3 100644
--- a/src/UniConversion.h
+++ b/src/UniConversion.h
@@ -55,6 +55,12 @@ inline bool UTF8IsNEL(const unsigned char *us) {
 	return (us[0] == 0xc2) && (us[1] == 0x85);
 }
 
+enum { SURROGATE_LEAD_FIRST = 0xD800 };
+enum { SURROGATE_LEAD_LAST = 0xDBFF };
+inline unsigned int UTF16CharLength(wchar_t uch) {
+	return ((uch >= SURROGATE_LEAD_FIRST) && (uch <= SURROGATE_LEAD_LAST)) ? 2 : 1;
+}
+
 #ifdef SCI_NAMESPACE
 }
 #endif
diff --git a/win32/ScintillaWin.cxx b/win32/ScintillaWin.cxx
index f4e43bb54..95caf83ad 100644
--- a/win32/ScintillaWin.cxx
+++ b/win32/ScintillaWin.cxx
@@ -979,22 +979,20 @@ sptr_t ScintillaWin::HandleCompositionInline(uptr_t, sptr_t lParam) {
 
 		bool tmpRecordingMacro = recordingMacro;
 		recordingMacro = false;
-		for (unsigned int i = 0; i < wcsLen; i++) {
-			wchar_t uniChar[1] = { 0 };
+		for (size_t i = 0; i < wcsLen; ) {
+			const size_t ucWidth = UTF16CharLength(wcs[i]);
+			const std::wstring uniChar(wcs+i, ucWidth);
 			char oneChar[UTF8MaxBytes + 1] = "\0\0\0\0"; // Maximum 4 bytes in utf8
 			unsigned int oneCharLen = 0;
 
-			uniChar[0] = wcs[i];
-
 			if (IsUnicodeMode()) {
-				oneCharLen = UTF8Length(uniChar, 1);
-				UTF8FromUTF16(uniChar, 1, oneChar, oneCharLen);
-				oneChar[oneCharLen] = '\0';
+				oneCharLen = UTF8Length(uniChar.c_str(), static_cast<unsigned int>(uniChar.length()));
+				UTF8FromUTF16(uniChar.c_str(), static_cast<unsigned int>(uniChar.length()), oneChar, oneCharLen);
 			} else {
 				oneCharLen = ::WideCharToMultiByte(InputCodePage(), 0,
-					uniChar, 1, oneChar, sizeof(oneChar)-1, 0, 0);
-				oneChar[oneCharLen] = '\0';
+					uniChar.c_str(), static_cast<unsigned int>(uniChar.length()), oneChar, sizeof(oneChar)-1, 0, 0);
 			}
+			oneChar[oneCharLen] = '\0';
 
 			// Display a character.
 			AddCharUTF(oneChar, oneCharLen);
@@ -1018,6 +1016,7 @@ sptr_t ScintillaWin::HandleCompositionInline(uptr_t, sptr_t lParam) {
 				break;
 			}
 			DrawImeIndicator(indicator, oneCharLen);
+			i += ucWidth;
 		}
 		recordingMacro = tmpRecordingMacro;
 
@@ -1032,23 +1031,22 @@ sptr_t ScintillaWin::HandleCompositionInline(uptr_t, sptr_t lParam) {
 			(hIMC, GCS_RESULTSTR, wcs, maxLenInputIME);
 		unsigned int wcsLen = bytes / 2;
 
-		for (unsigned int i = 0; i < wcsLen; i++) {
-			wchar_t uniChar[1] = { 0 };
+		for (size_t i = 0; i < wcsLen;) {
+			const size_t ucWidth = UTF16CharLength(wcs[i]);
+			const std::wstring uniChar(wcs+i, ucWidth);
 			char oneChar[UTF8MaxBytes+1] = "\0\0\0\0"; // Maximum 4 bytes in UTF-8.
 			unsigned int oneCharLen = 0;
 
-			uniChar[0] = wcs[i];
-
 			if (IsUnicodeMode()) {
-				oneCharLen = UTF8Length(uniChar, 1);
-				UTF8FromUTF16(uniChar, 1, oneChar, oneCharLen);
-				oneChar[oneCharLen] = '\0';
+				oneCharLen = UTF8Length(uniChar.c_str(), static_cast<unsigned int>(uniChar.length()));
+				UTF8FromUTF16(uniChar.c_str(), static_cast<unsigned int>(uniChar.length()), oneChar, oneCharLen);
 			} else {
 				oneCharLen = ::WideCharToMultiByte(InputCodePage(), 0,
-					uniChar, 1, oneChar, sizeof(oneChar)-1, 0, 0);
-				oneChar[oneCharLen] = '\0';
+					uniChar.c_str(), static_cast<unsigned int>(uniChar.length()), oneChar, sizeof(oneChar)-1, 0, 0);
 			}
+			oneChar[oneCharLen] = '\0';
 			AddCharUTF(oneChar, oneCharLen);
+			i += ucWidth;
 		}
 	}
 	SetCandidateWindowPos();
@@ -2224,7 +2222,6 @@ void ScintillaWin::Paste() {
 
 					unsigned int mlen = UTF8Length(&uptr[0], ulen);
 					std::vector<char> putf(mlen+1);
-						// CP_UTF8 not available on Windows 95, so use UTF8FromUTF16()
 					UTF8FromUTF16(&uptr[0], ulen, &putf[0], mlen);
 
 					InsertPasteShape(&putf[0], mlen, pasteShape);