TIP #548 implementation: Support `wchar_t` conversion functions and deprecate `Tcl_WinUtfToTChar()` and `Tcl_WinTCharToUtf()`

11 files changed, 196 insertions, 139 deletions

diff --git a/generic/tcl.decls b/generic/tcl.decls
index a3647d7..910c29e 100644
--- a/generic/tcl.decls
+++ b/generic/tcl.decls
@@ -1198,7 +1198,7 @@ declare 335 {
     int Tcl_UtfToTitle(char *src)
 }
 declare 336 {
-    int Tcl_UtfToUniChar(const char *src, Tcl_UniChar *chPtr)
+    int Tcl_UtfToChar16(const char *src, unsigned short *chPtr)
 }
 declare 337 {
     int Tcl_UtfToUpper(char *src)
@@ -1253,11 +1253,11 @@ declare 353 {
 	    unsigned long numChars)
 }
 declare 354 {
-    char *Tcl_UniCharToUtfDString(const Tcl_UniChar *uniStr,
+    char *Tcl_Char16ToUtfDString(const unsigned short *uniStr,
 	    int uniLength, Tcl_DString *dsPtr)
 }
 declare 355 {
-    Tcl_UniChar *Tcl_UtfToUniCharDString(const char *src,
+    unsigned short *Tcl_UtfToChar16DString(const char *src,
 	    int length, Tcl_DString *dsPtr)
 }
 declare 356 {
@@ -2390,6 +2390,19 @@ declare 645 {
 	    int endValue, int *indexPtr)
 }
 
+# TIP #548
+declare 646 {
+    int Tcl_UtfToUniChar(const char *src, int *chPtr)
+}
+declare 647 {
+    char *Tcl_UniCharToUtfDString(const int *uniStr,
+	    int uniLength, Tcl_DString *dsPtr)
+}
+declare 648 {
+    int *Tcl_UtfToUniCharDString(const char *src,
+	    int length, Tcl_DString *dsPtr)
+}
+
 # ----- BASELINE -- FOR -- 8.7.0 ----- #
 
 ##############################################################################
diff --git a/generic/tcl.h b/generic/tcl.h
index a5d5799..ff4cd8d 100644
--- a/generic/tcl.h
+++ b/generic/tcl.h
@@ -2154,7 +2154,7 @@ typedef struct Tcl_EncodingType {
 
 #if TCL_UTF_MAX > 4
     /*
-     * unsigned int isn't 100% accurate as it should be a strict 4-byte value
+     * int isn't 100% accurate as it should be a strict 4-byte value
      * (perhaps wchar_t). 64-bit systems may have troubles. The size of this
      * value must be reflected correctly in regcustom.h and
      * in tclEncoding.c.
@@ -2162,7 +2162,7 @@ typedef struct Tcl_EncodingType {
      * XXX: string rep that Tcl_UniChar represents.  Changing the size
      * XXX: of Tcl_UniChar is /not/ supported.
      */
-typedef unsigned int Tcl_UniChar;
+typedef int Tcl_UniChar;
 #else
 typedef unsigned short Tcl_UniChar;
 #endif
diff --git a/generic/tclDecls.h b/generic/tclDecls.h
index 3f39cd5..eddd385 100644
--- a/generic/tclDecls.h
+++ b/generic/tclDecls.h
@@ -1027,7 +1027,8 @@ EXTERN int		Tcl_UtfToLower(char *src);
 /* 335 */
 EXTERN int		Tcl_UtfToTitle(char *src);
 /* 336 */
-EXTERN int		Tcl_UtfToUniChar(const char *src, Tcl_UniChar *chPtr);
+EXTERN int		Tcl_UtfToChar16(const char *src,
+				unsigned short *chPtr);
 /* 337 */
 EXTERN int		Tcl_UtfToUpper(char *src);
 /* 338 */
@@ -1068,10 +1069,10 @@ EXTERN int		Tcl_UniCharNcmp(const Tcl_UniChar *ucs,
 				const Tcl_UniChar *uct,
 				unsigned long numChars);
 /* 354 */
-EXTERN char *		Tcl_UniCharToUtfDString(const Tcl_UniChar *uniStr,
+EXTERN char *		Tcl_Char16ToUtfDString(const unsigned short *uniStr,
 				int uniLength, Tcl_DString *dsPtr);
 /* 355 */
-EXTERN Tcl_UniChar *	Tcl_UtfToUniCharDString(const char *src, int length,
+EXTERN unsigned short *	 Tcl_UtfToChar16DString(const char *src, int length,
 				Tcl_DString *dsPtr);
 /* 356 */
 EXTERN Tcl_RegExp	Tcl_GetRegExpFromObj(Tcl_Interp *interp,
@@ -1904,6 +1905,14 @@ EXTERN int		Tcl_LinkArray(Tcl_Interp *interp,
 /* 645 */
 EXTERN int		Tcl_GetIntForIndex(Tcl_Interp *interp,
 				Tcl_Obj *objPtr, int endValue, int *indexPtr);
+/* 646 */
+EXTERN int		Tcl_UtfToUniChar(const char *src, int *chPtr);
+/* 647 */
+EXTERN char *		Tcl_UniCharToUtfDString(const int *uniStr,
+				int uniLength, Tcl_DString *dsPtr);
+/* 648 */
+EXTERN int *		Tcl_UtfToUniCharDString(const char *src, int length,
+				Tcl_DString *dsPtr);
 
 typedef struct {
     const struct TclPlatStubs *tclPlatStubs;
@@ -2275,7 +2284,7 @@ typedef struct TclStubs {
     char * (*tcl_UtfToExternalDString) (Tcl_Encoding encoding, const char *src, int srcLen, Tcl_DString *dsPtr); /* 333 */
     int (*tcl_UtfToLower) (char *src); /* 334 */
     int (*tcl_UtfToTitle) (char *src); /* 335 */
-    int (*tcl_UtfToUniChar) (const char *src, Tcl_UniChar *chPtr); /* 336 */
+    int (*tcl_UtfToChar16) (const char *src, unsigned short *chPtr); /* 336 */
     int (*tcl_UtfToUpper) (char *src); /* 337 */
     int (*tcl_WriteChars) (Tcl_Channel chan, const char *src, int srcLen); /* 338 */
     int (*tcl_WriteObj) (Tcl_Channel chan, Tcl_Obj *objPtr); /* 339 */
@@ -2293,8 +2302,8 @@ typedef struct TclStubs {
     int (*tcl_UniCharIsWordChar) (int ch); /* 351 */
     int (*tcl_UniCharLen) (const Tcl_UniChar *uniStr); /* 352 */
     int (*tcl_UniCharNcmp) (const Tcl_UniChar *ucs, const Tcl_UniChar *uct, unsigned long numChars); /* 353 */
-    char * (*tcl_UniCharToUtfDString) (const Tcl_UniChar *uniStr, int uniLength, Tcl_DString *dsPtr); /* 354 */
-    Tcl_UniChar * (*tcl_UtfToUniCharDString) (const char *src, int length, Tcl_DString *dsPtr); /* 355 */
+    char * (*tcl_Char16ToUtfDString) (const unsigned short *uniStr, int uniLength, Tcl_DString *dsPtr); /* 354 */
+    unsigned short * (*tcl_UtfToChar16DString) (const char *src, int length, Tcl_DString *dsPtr); /* 355 */
     Tcl_RegExp (*tcl_GetRegExpFromObj) (Tcl_Interp *interp, Tcl_Obj *patObj, int flags); /* 356 */
     TCL_DEPRECATED_API("Use Tcl_EvalTokensStandard") Tcl_Obj * (*tcl_EvalTokens) (Tcl_Interp *interp, Tcl_Token *tokenPtr, int count); /* 357 */
     void (*tcl_FreeParse) (Tcl_Parse *parsePtr); /* 358 */
@@ -2585,6 +2594,9 @@ typedef struct TclStubs {
     int (*tcl_IsShared) (Tcl_Obj *objPtr); /* 643 */
     int (*tcl_LinkArray) (Tcl_Interp *interp, const char *varName, void *addr, int type, int size); /* 644 */
     int (*tcl_GetIntForIndex) (Tcl_Interp *interp, Tcl_Obj *objPtr, int endValue, int *indexPtr); /* 645 */
+    int (*tcl_UtfToUniChar) (const char *src, int *chPtr); /* 646 */
+    char * (*tcl_UniCharToUtfDString) (const int *uniStr, int uniLength, Tcl_DString *dsPtr); /* 647 */
+    int * (*tcl_UtfToUniCharDString) (const char *src, int length, Tcl_DString *dsPtr); /* 648 */
 } TclStubs;
 
 extern const TclStubs *tclStubsPtr;
@@ -3287,8 +3299,8 @@ extern const TclStubs *tclStubsPtr;
 	(tclStubsPtr->tcl_UtfToLower) /* 334 */
 #define Tcl_UtfToTitle \
 	(tclStubsPtr->tcl_UtfToTitle) /* 335 */
-#define Tcl_UtfToUniChar \
-	(tclStubsPtr->tcl_UtfToUniChar) /* 336 */
+#define Tcl_UtfToChar16 \
+	(tclStubsPtr->tcl_UtfToChar16) /* 336 */
 #define Tcl_UtfToUpper \
 	(tclStubsPtr->tcl_UtfToUpper) /* 337 */
 #define Tcl_WriteChars \
@@ -3323,10 +3335,10 @@ extern const TclStubs *tclStubsPtr;
 	(tclStubsPtr->tcl_UniCharLen) /* 352 */
 #define Tcl_UniCharNcmp \
 	(tclStubsPtr->tcl_UniCharNcmp) /* 353 */
-#define Tcl_UniCharToUtfDString \
-	(tclStubsPtr->tcl_UniCharToUtfDString) /* 354 */
-#define Tcl_UtfToUniCharDString \
-	(tclStubsPtr->tcl_UtfToUniCharDString) /* 355 */
+#define Tcl_Char16ToUtfDString \
+	(tclStubsPtr->tcl_Char16ToUtfDString) /* 354 */
+#define Tcl_UtfToChar16DString \
+	(tclStubsPtr->tcl_UtfToChar16DString) /* 355 */
 #define Tcl_GetRegExpFromObj \
 	(tclStubsPtr->tcl_GetRegExpFromObj) /* 356 */
 #define Tcl_EvalTokens \
@@ -3907,6 +3919,12 @@ extern const TclStubs *tclStubsPtr;
 	(tclStubsPtr->tcl_LinkArray) /* 644 */
 #define Tcl_GetIntForIndex \
 	(tclStubsPtr->tcl_GetIntForIndex) /* 645 */
+#define Tcl_UtfToUniChar \
+	(tclStubsPtr->tcl_UtfToUniChar) /* 646 */
+#define Tcl_UniCharToUtfDString \
+	(tclStubsPtr->tcl_UniCharToUtfDString) /* 647 */
+#define Tcl_UtfToUniCharDString \
+	(tclStubsPtr->tcl_UtfToUniCharDString) /* 648 */
 
 #endif /* defined(USE_TCL_STUBS) */
 
@@ -4092,6 +4110,36 @@ extern const TclStubs *tclStubsPtr;
 #undef Tcl_StringMatch
 #define Tcl_StringMatch(str, pattern) Tcl_StringCaseMatch((str), (pattern), 0)
 
+#if TCL_UTF_MAX <= 4
+#   undef Tcl_UniCharToUtfDString
+#   define Tcl_UniCharToUtfDString Tcl_Char16ToUtfDString
+#   undef Tcl_UtfToUniCharDString
+#   define Tcl_UtfToUniCharDString Tcl_UtfToChar16DString
+#   undef Tcl_UtfToUniChar
+#   define Tcl_UtfToUniChar Tcl_UtfToChar16
+#endif
+#if defined(USE_TCL_STUBS)
+#   define Tcl_WCharToUtfDString (sizeof(wchar_t) != sizeof(short) \
+		? (char *(*)(const wchar_t *, int, Tcl_DString *))tclStubsPtr->tcl_UniCharToUtfDString \
+		: (char *(*)(const wchar_t *, int, Tcl_DString *))Tcl_Char16ToUtfDString)
+#   define Tcl_UtfToWCharDString (sizeof(wchar_t) != sizeof(short) \
+		? (wchar_t *(*)(const char *, int, Tcl_DString *))tclStubsPtr->tcl_UtfToUniCharDString \
+		: (wchar_t *(*)(const char *, int, Tcl_DString *))Tcl_UtfToChar16DString)
+#   define Tcl_UtfToWChar (sizeof(wchar_t) != sizeof(short) \
+		? (int (*)(const char *, wchar_t *))tclStubsPtr->tcl_UtfToChar16 \
+		: (int (*)(const char *, wchar_t *))Tcl_UtfToUniChar)
+#else
+#   define Tcl_WCharToUtfDString (sizeof(wchar_t) != sizeof(short) \
+		? (char *(*)(const wchar_t *, int, Tcl_DString *))Tcl_UniCharToUtfDString \
+		: (char *(*)(const wchar_t *, int, Tcl_DString *))Tcl_Char16ToUtfDString)
+#   define Tcl_UtfToWCharDString (sizeof(wchar_t) != sizeof(short) \
+		? (wchar_t *(*)(const char *, int, Tcl_DString *))Tcl_UtfToUniCharDString \
+		: (wchar_t *(*)(const char *, int, Tcl_DString *))Tcl_UtfToChar16DString)
+#   define Tcl_UtfToWChar (sizeof(wchar_t) != sizeof(short) \
+		? (int (*)(const char *, wchar_t *))Tcl_UtfToChar16 \
+		: (int (*)(const char *, wchar_t *))Tcl_UtfToUniChar)
+#endif
+
 /*
  * Deprecated Tcl procedures:
  */
diff --git a/generic/tclIOSock.c b/generic/tclIOSock.c
index 12e2900..adf729a 100644
--- a/generic/tclIOSock.c
+++ b/generic/tclIOSock.c
@@ -30,11 +30,12 @@ gai_strerror(
     ThreadSpecificData *tsdPtr = TCL_TSD_INIT(&dataKey);
 
     if (tsdPtr->initialized) {
-	Tcl_DStringFree(&tsdPtr->errorMsg);
+	Tcl_DStringSetLength(&tsdPtr->errorMsg, 0);
     } else {
+	Tcl_DStringInit(&tsdPtr->errorMsg);
 	tsdPtr->initialized = 1;
     }
-    Tcl_WinTCharToUtf(gai_strerrorW(code), -1, &tsdPtr->errorMsg);
+    Tcl_WCharToUtfDString(gai_strerrorW(code), -1, &tsdPtr->errorMsg);
     return Tcl_DStringValue(&tsdPtr->errorMsg);
 }
 #endif
diff --git a/generic/tclInt.h b/generic/tclInt.h
index 3db3f24..caa9f7a 100644
--- a/generic/tclInt.h
+++ b/generic/tclInt.h
@@ -3243,17 +3243,6 @@ MODULE_SCOPE const char*TclGetCommandTypeName(Tcl_Command command);
 MODULE_SCOPE void	TclRegisterCommandTypeName(
 			    Tcl_ObjCmdProc *implementationProc,
 			    const char *nameStr);
-#if (TCL_UTF_MAX > 4) && (defined(__CYGWIN__) || defined(_WIN32))
-MODULE_SCOPE int TclUtfToWChar(const char *src, WCHAR *chPtr);
-MODULE_SCOPE char *	TclWCharToUtfDString(const WCHAR *uniStr,
-			    int uniLength, Tcl_DString *dsPtr);
-MODULE_SCOPE WCHAR * TclUtfToWCharDString(const char *src,
-			    int length, Tcl_DString *dsPtr);
-#else
-#   define TclUtfToWChar TclUtfToUniChar
-#   define TclWCharToUtfDString Tcl_UniCharToUtfDString
-#   define TclUtfToWCharDString Tcl_UtfToUniCharDString
-#endif
 MODULE_SCOPE int	TclUtfCmp(const char *cs, const char *ct);
 MODULE_SCOPE int	TclUtfCasecmp(const char *cs, const char *ct);
 MODULE_SCOPE int	TclUtfCount(int ch);
@@ -4631,10 +4620,17 @@ MODULE_SCOPE void	TclDbInitNewObj(Tcl_Obj *objPtr, const char *file,
  *----------------------------------------------------------------
  */
 
+#if TCL_UTF_MAX > 4
 #define TclUtfToUniChar(str, chPtr) \
 	((((unsigned char) *(str)) < 0x80) ?		\
 	    ((*(chPtr) = (unsigned char) *(str)), 1)	\
 	    : Tcl_UtfToUniChar(str, chPtr))
+#else
+#define TclUtfToUniChar(str, chPtr) \
+	((((unsigned char) *(str)) < 0x80) ?		\
+	    ((*(chPtr) = (unsigned char) *(str)), 1)	\
+	    : Tcl_UtfToChar16(str, chPtr))
+#endif
 
 /*
  *----------------------------------------------------------------
diff --git a/generic/tclMain.c b/generic/tclMain.c
index 4b8fa8c..4a66793 100644
--- a/generic/tclMain.c
+++ b/generic/tclMain.c
@@ -70,10 +70,8 @@ NewNativeObj(
     Tcl_DString ds;
 
 #ifdef UNICODE
-    if (length > 0) {
-	length *= sizeof(WCHAR);
-    }
-    Tcl_WinTCharToUtf(string, length, &ds);
+    Tcl_DStringInit(&ds);
+    Tcl_WCharToUtfDString(string, length, &ds);
 #else
     Tcl_ExternalToUtfDString(NULL, (char *) string, length, &ds);
 #endif
diff --git a/generic/tclPlatDecls.h b/generic/tclPlatDecls.h
index abc8ee8..354d752 100644
--- a/generic/tclPlatDecls.h
+++ b/generic/tclPlatDecls.h
@@ -117,6 +117,16 @@ extern const TclPlatStubs *tclPlatStubsPtr;
 #undef TCL_STORAGE_CLASS
 #define TCL_STORAGE_CLASS DLLIMPORT
 
-#endif /* _TCLPLATDECLS */
-
+#if defined(USE_TCL_STUBS) && (defined(_WIN32) || defined(__CYGWIN__))\
+	&& (defined(TCL_NO_DEPRECATED) || TCL_MAJOR_VERSION > 8)
+#undef Tcl_WinUtfToTChar
+#undef Tcl_WinTCharToUtf
+#ifdef _WIN32
+#define Tcl_WinUtfToTChar(string, len, dsPtr) (Tcl_DStringInit(dsPtr), \
+		(TCHAR *)Tcl_UtfToChar16DString((string), (len), (dsPtr)))
+#define Tcl_WinTCharToUtf(string, len, dsPtr) (Tcl_DStringInit(dsPtr), \
+		(char *)Tcl_Char16ToUtfDString((string), ((((len) + 2) >> 1) - 1), (dsPtr)))
+#endif
+#endif
 
+#endif /* _TCLPLATDECLS */
diff --git a/generic/tclStubInit.c b/generic/tclStubInit.c
index 5e918f5..1c3f094 100644
--- a/generic/tclStubInit.c
+++ b/generic/tclStubInit.c
@@ -60,6 +60,9 @@
 #undef TclBNInitBignumFromLong
 #undef Tcl_BackgroundError
 #define TclStaticPackage Tcl_StaticPackage
+#undef Tcl_UniCharToUtfDString
+#undef Tcl_UtfToUniCharDString
+#undef Tcl_UtfToUniChar
 
 #undef TclBN_mp_tc_and
 #undef TclBN_mp_tc_or
@@ -245,6 +248,8 @@ TclpGetPid(Tcl_Pid pid)
     return (int) (size_t) pid;
 }
 
+#if !defined(TCL_NO_DEPRECATED) && TCL_MAJOR_VERSION < 9
+#undef Tcl_WinUtfToTChar
 char *
 Tcl_WinUtfToTChar(
     const char *string,
@@ -252,12 +257,9 @@ Tcl_WinUtfToTChar(
     Tcl_DString *dsPtr)
 {
     Tcl_DStringInit(dsPtr);
-    if (!string) {
-	return NULL;
-    }
-    return (char *)TclUtfToWCharDString(string, len, dsPtr);
+    return (char *)Tcl_UtfToChar16DString(string, len, dsPtr);
 }
-
+#undef Tcl_WinTCharToUtf
 char *
 Tcl_WinTCharToUtf(
     const char *string,
@@ -265,16 +267,9 @@ Tcl_WinTCharToUtf(
     Tcl_DString *dsPtr)
 {
     Tcl_DStringInit(dsPtr);
-    if (!string) {
-	return NULL;
-    }
-    if (len < 0) {
-	len = wcslen((wchar_t *)string);
-    } else {
-	len /= 2;
-    }
-    return TclWCharToUtfDString((const WCHAR *)string, len, dsPtr);
+    return Tcl_Char16ToUtfDString((const unsigned short *)string, len >> 1, dsPtr);
 }
+#endif /* !defined(TCL_NO_DEPRECATED) */
 
 #if defined(TCL_WIDE_INT_IS_LONG)
 /* On Cygwin64, long is 64-bit while on Win64 long is 32-bit. Therefore
@@ -479,6 +474,11 @@ tellOld(
 }
 #endif /* !TCL_NO_DEPRECATED */
 
+#if defined(TCL_NO_DEPRECATED) || TCL_MAJOR_VERSION > 8
+#define Tcl_WinUtfToTChar 0
+#define Tcl_WinTCharToUtf 0
+#endif
+
 /*
  * WARNING: The contents of this file is automatically generated by the
  * tools/genStubs.tcl script. Any modifications to the function declarations
@@ -1332,7 +1332,7 @@ const TclStubs tclStubs = {
     Tcl_UtfToExternalDString, /* 333 */
     Tcl_UtfToLower, /* 334 */
     Tcl_UtfToTitle, /* 335 */
-    Tcl_UtfToUniChar, /* 336 */
+    Tcl_UtfToChar16, /* 336 */
     Tcl_UtfToUpper, /* 337 */
     Tcl_WriteChars, /* 338 */
     Tcl_WriteObj, /* 339 */
@@ -1350,8 +1350,8 @@ const TclStubs tclStubs = {
     Tcl_UniCharIsWordChar, /* 351 */
     Tcl_UniCharLen, /* 352 */
     Tcl_UniCharNcmp, /* 353 */
-    Tcl_UniCharToUtfDString, /* 354 */
-    Tcl_UtfToUniCharDString, /* 355 */
+    Tcl_Char16ToUtfDString, /* 354 */
+    Tcl_UtfToChar16DString, /* 355 */
     Tcl_GetRegExpFromObj, /* 356 */
     Tcl_EvalTokens, /* 357 */
     Tcl_FreeParse, /* 358 */
@@ -1642,6 +1642,9 @@ const TclStubs tclStubs = {
     Tcl_IsShared, /* 643 */
     Tcl_LinkArray, /* 644 */
     Tcl_GetIntForIndex, /* 645 */
+    Tcl_UtfToUniChar, /* 646 */
+    Tcl_UniCharToUtfDString, /* 647 */
+    Tcl_UtfToUniCharDString, /* 648 */
 };
 
 /* !END!: Do not edit above this line. */
diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index 6c39d1c..320d7aa 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -221,22 +221,33 @@ three:
  *---------------------------------------------------------------------------
  */
 
+#undef Tcl_UniCharToUtfDString
 char *
 Tcl_UniCharToUtfDString(
-    const Tcl_UniChar *uniStr,	/* Unicode string to convert to UTF-8. */
-    int uniLength,		/* Length of Unicode string in Tcl_UniChars
-				 * (must be >= 0). */
+    const int *uniStr,	/* Unicode string to convert to UTF-8. */
+    int uniLength,		/* Length of Unicode string. */
     Tcl_DString *dsPtr)		/* UTF-8 representation of string is appended
 				 * to this previously initialized DString. */
 {
-    const Tcl_UniChar *w, *wEnd;
+    const int *w, *wEnd;
     char *p, *string;
-    int oldLength, len = 1;
+    int oldLength;
 
     /*
      * UTF-8 string length in bytes will be <= Unicode string length * 4.
      */
 
+    if (uniStr == NULL) {
+	return NULL;
+    }
+    if (uniLength < 0) {
+	uniLength = 0;
+	w = uniStr;
+	while (*w != '\0') {
+	    uniLength++;
+	    w++;
+	}
+    }
     oldLength = Tcl_DStringLength(dsPtr);
     Tcl_DStringSetLength(dsPtr, oldLength + (uniLength + 1) * 4);
     string = Tcl_DStringValue(dsPtr) + oldLength;
@@ -244,45 +255,43 @@ Tcl_UniCharToUtfDString(
     p = string;
     wEnd = uniStr + uniLength;
     for (w = uniStr; w < wEnd; ) {
-	if (!len && ((*w & 0xFC00) != 0xDC00)) {
-	    /* Special case for handling high surrogates. */
-	    p += Tcl_UniCharToUtf(-1, p);
-	}
-	len = Tcl_UniCharToUtf(*w, p);
-	p += len;
-	if ((*w >= 0xD800) && (len < 3)) {
-	    len = 0; /* Indication that high surrogate was found */
-	}
+	p += Tcl_UniCharToUtf(*w, p);
 	w++;
     }
-    if (!len) {
-	/* Special case for handling high surrogates. */
-	p += Tcl_UniCharToUtf(-1, p);
-    }
     Tcl_DStringSetLength(dsPtr, oldLength + (p - string));
 
     return string;
 }
 
-#if (TCL_UTF_MAX > 4) && (defined(__CYGWIN__) || defined(_WIN32))
 char *
-TclWCharToUtfDString(
-    const WCHAR *uniStr,	/* WCHAR string to convert to UTF-8. */
-    int uniLength,		/* Length of WCHAR string in Tcl_UniChars
-				 * (must be >= 0). */
+Tcl_Char16ToUtfDString(
+    const unsigned short *uniStr,/* Utf-16 string to convert to UTF-8. */
+    int uniLength,		/* Length of Utf-16 string. */
     Tcl_DString *dsPtr)		/* UTF-8 representation of string is appended
 				 * to this previously initialized DString. */
 {
-    const WCHAR *w, *wEnd;
+    const unsigned short *w, *wEnd;
     char *p, *string;
     int oldLength, len = 1;
 
     /*
-     * UTF-8 string length in bytes will be <= Unicode string length * 4.
+     * UTF-8 string length in bytes will be <= Utf16 string length * 3.
      */
 
+    if (uniStr == NULL) {
+	return NULL;
+    }
+    if (uniLength < 0) {
+
+	uniLength = 0;
+	w = uniStr;
+	while (*w != '\0') {
+	    uniLength++;
+	    w++;
+	}
+    }
     oldLength = Tcl_DStringLength(dsPtr);
-    Tcl_DStringSetLength(dsPtr, oldLength + (uniLength + 1) * 4);
+    Tcl_DStringSetLength(dsPtr, oldLength + (uniLength + 1) * 3);
     string = Tcl_DStringValue(dsPtr) + oldLength;
 
     p = string;
@@ -307,7 +316,6 @@ TclWCharToUtfDString(
 
     return string;
 }
-#endif
 /*
  *---------------------------------------------------------------------------
  *
@@ -350,13 +358,14 @@ static const unsigned short cp1252[32] = {
    0x2DC, 0x2122, 0x0161, 0x203A, 0x0153,   0x9D, 0x017E, 0x0178
 };
 
+#undef Tcl_UtfToUniChar
 int
 Tcl_UtfToUniChar(
     const char *src,	/* The UTF-8 string. */
-    Tcl_UniChar *chPtr)/* Filled with the Tcl_UniChar represented by
+    int *chPtr)/* Filled with the unsigned int represented by
 				 * the UTF-8 string. */
 {
-    Tcl_UniChar byte;
+    int byte;
 
     /*
      * Unroll 1 to 4 byte UTF-8 sequences.
@@ -372,20 +381,6 @@ Tcl_UtfToUniChar(
 	 * characters representing themselves.
 	 */
 
-#if TCL_UTF_MAX <= 4
-	/* If *chPtr contains a high surrogate (produced by a previous
-	 * Tcl_UtfToUniChar() call) and the next 3 bytes are UTF-8 continuation
-	 * bytes, then we must produce a follow-up low surrogate. We only
-	 * do that if the high surrogate matches the bits we encounter.
-	 */
-	if ((byte >= 0x80)
-		&& (((((byte - 0x10) << 2) & 0xFC) | 0xD800) == (*chPtr & 0xFCFC))
-		&& ((src[1] & 0xF0) == (((*chPtr << 4) & 0x30) | 0x80))
-		&& ((src[2] & 0xC0) == 0x80)) {
-	    *chPtr = ((src[1] & 0x0F) << 6) + (src[2] & 0x3F) + 0xDC00;
-	    return 3;
-	}
-#endif
 	if ((unsigned)(byte-0x80) < (unsigned)0x20) {
 	    *chPtr = cp1252[byte-0x80];
 	} else {
@@ -431,23 +426,11 @@ Tcl_UtfToUniChar(
 	    /*
 	     * Four-byte-character lead byte followed by three trail bytes.
 	     */
-#if TCL_UTF_MAX <= 4
-	    Tcl_UniChar high = (((byte & 0x07) << 8) | ((src[1] & 0x3F) << 2)
-		    | ((src[2] & 0x3F) >> 4)) - 0x40;
-	    if (high >= 0x400) {
-		/* out of range, < 0x10000 or > 0x10ffff */
-	    } else {
-		/* produce high surrogate, advance source pointer */
-		*chPtr = 0xD800 + high;
-		return 1;
-	    }
-#else
 	    *chPtr = (((byte & 0x07) << 18) | ((src[1] & 0x3F) << 12)
 		    | ((src[2] & 0x3F) << 6) | (src[3] & 0x3F));
 	    if ((unsigned)(*chPtr - 0x10000) <= 0xFFFFF) {
 		return 4;
 	    }
-#endif
 	}
 
 	/*
@@ -460,14 +443,13 @@ Tcl_UtfToUniChar(
     return 1;
 }
 
-#if (TCL_UTF_MAX > 4) && (defined(__CYGWIN__) || defined(_WIN32))
 int
-TclUtfToWChar(
+Tcl_UtfToChar16(
     const char *src,	/* The UTF-8 string. */
-    WCHAR *chPtr)/* Filled with the WCHAR represented by
+    unsigned short *chPtr)/* Filled with the unsigned short represented by
 				 * the UTF-8 string. */
 {
-    WCHAR byte;
+    unsigned short byte;
 
     /*
      * Unroll 1 to 4 byte UTF-8 sequences.
@@ -540,7 +522,7 @@ TclUtfToWChar(
 	    /*
 	     * Four-byte-character lead byte followed by three trail bytes.
 	     */
-	    WCHAR high = (((byte & 0x07) << 8) | ((src[1] & 0x3F) << 2)
+	    unsigned short high = (((byte & 0x07) << 8) | ((src[1] & 0x3F) << 2)
 		    | ((src[2] & 0x3F) >> 4)) - 0x40;
 	    if (high >= 0x400) {
 		/* out of range, < 0x10000 or > 0x10ffff */
@@ -560,7 +542,6 @@ TclUtfToWChar(
     *chPtr = byte;
     return 1;
 }
-#endif
 
 /*
  *---------------------------------------------------------------------------
@@ -580,7 +561,8 @@ TclUtfToWChar(
  *---------------------------------------------------------------------------
  */
 
-Tcl_UniChar *
+#undef Tcl_UtfToUniCharDString
+int *
 Tcl_UtfToUniCharDString(
     const char *src,		/* UTF-8 string to convert to Unicode. */
     int length,			/* Length of UTF-8 string in bytes, or -1 for
@@ -589,10 +571,13 @@ Tcl_UtfToUniCharDString(
 				 * appended to this previously initialized
 				 * DString. */
 {
-    Tcl_UniChar ch = 0, *w, *wString;
+    int ch = 0, *w, *wString;
     const char *p, *end;
     int oldLength;
 
+    if (src == NULL) {
+	return NULL;
+    }
     if (length < 0) {
 	length = strlen(src);
     }
@@ -605,20 +590,20 @@ Tcl_UtfToUniCharDString(
     oldLength = Tcl_DStringLength(dsPtr);
 
     Tcl_DStringSetLength(dsPtr,
-	    oldLength + (int) ((length + 1) * sizeof(Tcl_UniChar)));
-    wString = (Tcl_UniChar *) (Tcl_DStringValue(dsPtr) + oldLength);
+	    oldLength + ((length + 1) * sizeof(int)));
+    wString = (int *) (Tcl_DStringValue(dsPtr) + oldLength);
 
     w = wString;
     p = src;
     end = src + length - 4;
     while (p < end) {
-	p += TclUtfToUniChar(p, &ch);
+	p += Tcl_UtfToUniChar(p, &ch);
 	*w++ = ch;
     }
     end += 4;
     while (p < end) {
 	if (Tcl_UtfCharComplete(p, end-p)) {
-	    p += TclUtfToUniChar(p, &ch);
+	    p += Tcl_UtfToUniChar(p, &ch);
 	} else {
 	    ch = UCHAR(*p++);
 	}
@@ -631,9 +616,8 @@ Tcl_UtfToUniCharDString(
     return wString;
 }
 
-#if (TCL_UTF_MAX > 4) && (defined(__CYGWIN__) || defined(_WIN32))
-WCHAR *
-TclUtfToWCharDString(
+unsigned short *
+Tcl_UtfToChar16DString(
     const char *src,		/* UTF-8 string to convert to Unicode. */
     int length,			/* Length of UTF-8 string in bytes, or -1 for
 				 * strlen(). */
@@ -641,10 +625,14 @@ TclUtfToWCharDString(
 				 * appended to this previously initialized
 				 * DString. */
 {
-    WCHAR ch = 0, *w, *wString;
+    unsigned short ch = 0;
+    unsigned short *w, *wString;
     const char *p, *end;
     int oldLength;
 
+    if (src == NULL) {
+	return NULL;
+    }
     if (length < 0) {
 	length = strlen(src);
     }
@@ -657,20 +645,20 @@ TclUtfToWCharDString(
     oldLength = Tcl_DStringLength(dsPtr);
 
     Tcl_DStringSetLength(dsPtr,
-	    oldLength + (int) ((length + 1) * sizeof(WCHAR)));
-    wString = (WCHAR *) (Tcl_DStringValue(dsPtr) + oldLength);
+	    oldLength + ((length + 1) * sizeof(unsigned short)));
+    wString = (unsigned short *) (Tcl_DStringValue(dsPtr) + oldLength);
 
     w = wString;
     p = src;
     end = src + length - 4;
     while (p < end) {
-	p += TclUtfToWChar(p, &ch);
+	p += Tcl_UtfToChar16(p, &ch);
 	*w++ = ch;
     }
     end += 4;
     while (p < end) {
 	if (Tcl_UtfCharComplete(p, end-p)) {
-	    p += TclUtfToWChar(p, &ch);
+	    p += Tcl_UtfToChar16(p, &ch);
 	} else {
 	    ch = UCHAR(*p++);
 	}
@@ -682,7 +670,6 @@ TclUtfToWCharDString(
 
     return wString;
 }
-#endif
 /*
  *---------------------------------------------------------------------------
  *
@@ -2151,7 +2138,7 @@ Tcl_UniCharCaseMatch(
 		if ((p != '[') && (p != '?') && (p != '\\')) {
 		    if (nocase) {
 			while (*uniStr && (p != *uniStr)
-				&& (p != (Tcl_UniChar)Tcl_UniCharToLower(*uniStr))) {
+				&& (p != Tcl_UniCharToLower(*uniStr))) {
 			    uniStr++;
 			}
 		    } else {
@@ -2191,13 +2178,13 @@ Tcl_UniCharCaseMatch(
 	    Tcl_UniChar startChar, endChar;
 
 	    uniPattern++;
-	    ch1 = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*uniStr) : *uniStr);
+	    ch1 = (nocase ? Tcl_UniCharToLower(*uniStr) : *uniStr);
 	    uniStr++;
 	    while (1) {
 		if ((*uniPattern == ']') || (*uniPattern == 0)) {
 		    return 0;
 		}
-		startChar = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*uniPattern)
+		startChar = (nocase ? Tcl_UniCharToLower(*uniPattern)
 			: *uniPattern);
 		uniPattern++;
 		if (*uniPattern == '-') {
@@ -2205,7 +2192,7 @@ Tcl_UniCharCaseMatch(
 		    if (*uniPattern == 0) {
 			return 0;
 		    }
-		    endChar = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*uniPattern)
+		    endChar = (nocase ? Tcl_UniCharToLower(*uniPattern)
 			    : *uniPattern);
 		    uniPattern++;
 		    if (((startChar <= ch1) && (ch1 <= endChar))
@@ -2343,7 +2330,7 @@ TclUniCharMatch(
 		if ((p != '[') && (p != '?') && (p != '\\')) {
 		    if (nocase) {
 			while ((string < stringEnd) && (p != *string)
-				&& (p != (Tcl_UniChar)Tcl_UniCharToLower(*string))) {
+				&& (p != Tcl_UniCharToLower(*string))) {
 			    string++;
 			}
 		    } else {
@@ -2384,20 +2371,20 @@ TclUniCharMatch(
 	    Tcl_UniChar ch1, startChar, endChar;
 
 	    pattern++;
-	    ch1 = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*string) : *string);
+	    ch1 = (nocase ? Tcl_UniCharToLower(*string) : *string);
 	    string++;
 	    while (1) {
 		if ((*pattern == ']') || (pattern == patternEnd)) {
 		    return 0;
 		}
-		startChar = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*pattern) : *pattern);
+		startChar = (nocase ? Tcl_UniCharToLower(*pattern) : *pattern);
 		pattern++;
 		if (*pattern == '-') {
 		    pattern++;
 		    if (pattern == patternEnd) {
 			return 0;
 		    }
-		    endChar = (nocase ? (Tcl_UniChar)Tcl_UniCharToLower(*pattern)
+		    endChar = (nocase ? Tcl_UniCharToLower(*pattern)
 			    : *pattern);
 		    pattern++;
 		    if (((startChar <= ch1) && (ch1 <= endChar))
diff --git a/generic/tclUtil.c b/generic/tclUtil.c
index c5c816f..0e4bb18 100644
--- a/generic/tclUtil.c
+++ b/generic/tclUtil.c
@@ -2304,7 +2304,7 @@ Tcl_StringCaseMatch(
 		    if (nocase) {
 			while (*str) {
 			    charLen = TclUtfToUniChar(str, &ch1);
-			    if (ch2==ch1 || ch2==(Tcl_UniChar)Tcl_UniCharToLower(ch1)) {
+			    if (ch2==ch1 || ch2==Tcl_UniCharToLower(ch1)) {
 				break;
 			    }
 			    str += charLen;
diff --git a/generic/tclZipfs.c b/generic/tclZipfs.c
index d842289..d59d893 100644
--- a/generic/tclZipfs.c
+++ b/generic/tclZipfs.c
@@ -4927,7 +4927,8 @@ TclZipfs_AppHook(
 #ifdef _WIN32
 	Tcl_DString ds;
 
-	archive = Tcl_WinTCharToUtf((*argvPtr)[1], -1, &ds);
+	Tcl_DStringInit(&ds);
+	archive = Tcl_WCharToUtfDString((*argvPtr)[1], -1, &ds);
 #else /* !_WIN32 */
 	archive = (*argvPtr)[1];
 #endif /* _WIN32 */