merge changes from pspjuth that optimize conversion from unichar to utf and add optimized versions for reading a word from byte codes.

4 files changed, 42 insertions, 11 deletions

diff --git a/generic/tclCompile.h b/generic/tclCompile.h
index b89346d..f9fddf3 100644
--- a/generic/tclCompile.h
+++ b/generic/tclCompile.h
@@ -1516,6 +1516,37 @@ MODULE_SCOPE int	TclPushProcCallFrame(ClientData clientData,
 		    (*((p)+3)))
 
 /*
+ * Override TclGetUInt4AtPtr or TclGetInt4AtPtr macros if
+ * a known better version exists.
+ */
+#ifdef WORDS_BIGENDIAN
+#define OVERRIDE_INT4(i) (i)
+#elif defined(__GNUC__) && (__GNUC__ > 4) || ((__GNUC__ == 4) && (__GNUC_MINOR__ >= 3))
+#define OVERRIDE_INT4(i) __builtin_bswap32(i)
+#elif defined(_MSC_VER) && _MSC_VER>=1300
+#define OVERRIDE_INT4(i) _byteswap_ulong(i)
+#endif
+
+#ifdef OVERRIDE_INT4
+#undef TclGetUInt4AtPtr
+static inline unsigned int
+TclGetUInt4AtPtr(const unsigned char *p)
+{
+    uint32_t i;
+    memcpy(&i,p,4);
+    return OVERRIDE_INT4(i);
+}
+#undef TclGetInt4AtPtr
+static inline signed int
+TclGetInt4AtPtr(const unsigned char *p)
+{
+    int32_t i;
+    memcpy(&i,p,4);
+    return OVERRIDE_INT4(i);
+}
+#endif /* OVERRIDE_INT4 */
+
+/*
  * Macros used to compute the minimum and maximum of two integers. The ANSI C
  * "prototypes" for these macros are:
  *
diff --git a/generic/tclInt.h b/generic/tclInt.h
index 50eb370..6c39558 100644
--- a/generic/tclInt.h
+++ b/generic/tclInt.h
@@ -3133,6 +3133,7 @@ MODULE_SCOPE int	TclTrimLeft(const char *bytes, int numBytes,
 MODULE_SCOPE int	TclTrimRight(const char *bytes, int numBytes,
 			    const char *trim, int numTrim);
 MODULE_SCOPE int	TclUtfCasecmp(const char *cs, const char *ct);
+MODULE_SCOPE int	TclUtfCount(Tcl_UniChar ch);
 MODULE_SCOPE Tcl_Obj *	TclpNativeToNormalized(ClientData clientData);
 MODULE_SCOPE Tcl_Obj *	TclpFilesystemPathType(Tcl_Obj *pathPtr);
 MODULE_SCOPE int	TclpDlopen(Tcl_Interp *interp, Tcl_Obj *pathPtr,
diff --git a/generic/tclStringObj.c b/generic/tclStringObj.c
index 8d70d20..1974079 100644
--- a/generic/tclStringObj.c
+++ b/generic/tclStringObj.c
@@ -3097,7 +3097,7 @@ ExtendStringRepWithUnicode(
      */
 
     int i, origLength, size = 0;
-    char *dst, buf[TCL_UTF_MAX];
+    char *dst;
     String *stringPtr = GET_STRING(objPtr);
 
     if (numChars < 0) {
@@ -3123,7 +3123,7 @@ ExtendStringRepWithUnicode(
     }
 
     for (i = 0; i < numChars && size >= 0; i++) {
-	size += Tcl_UniCharToUtf((int) unicode[i], buf);
+	size += TclUtfCount(unicode[i]);
     }
     if (size < 0) {
 	Tcl_Panic("max size for a Tcl value (%d bytes) exceeded", INT_MAX);
diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index b878149..f585c03 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -89,12 +89,11 @@ static const unsigned char totalBytes[256] = {
  * Functions used only in this module.
  */
 
-static int		UtfCount(int ch);
 
 /*
  *---------------------------------------------------------------------------
  *
- * UtfCount --
+ * TclUtfCount --
  *
  *	Find the number of bytes in the Utf character "ch".
  *
@@ -107,9 +106,9 @@ static int		UtfCount(int ch);
  *---------------------------------------------------------------------------
  */
 
-INLINE static int
-UtfCount(
-    int ch)			/* The Tcl_UniChar whose size is returned. */
+INLINE int
+TclUtfCount(
+    Tcl_UniChar ch)		/* The Tcl_UniChar whose size is returned. */
 {
     if ((ch > 0) && (ch < UNICODE_SELF)) {
 	return 1;
@@ -829,7 +828,7 @@ Tcl_UtfToUpper(
 	 * char to dst if its size is <= the original char.
 	 */
 
-	if (bytes < UtfCount(upChar)) {
+	if (bytes < TclUtfCount(upChar)) {
 	    memcpy(dst, src, (size_t) bytes);
 	    dst += bytes;
 	} else {
@@ -882,7 +881,7 @@ Tcl_UtfToLower(
 	 * char to dst if its size is <= the original char.
 	 */
 
-	if (bytes < UtfCount(lowChar)) {
+	if (bytes < TclUtfCount(lowChar)) {
 	    memcpy(dst, src, (size_t) bytes);
 	    dst += bytes;
 	} else {
@@ -932,7 +931,7 @@ Tcl_UtfToTitle(
 	bytes = TclUtfToUniChar(src, &ch);
 	titleChar = Tcl_UniCharToTitle(ch);
 
-	if (bytes < UtfCount(titleChar)) {
+	if (bytes < TclUtfCount(titleChar)) {
 	    memcpy(dst, src, (size_t) bytes);
 	    dst += bytes;
 	} else {
@@ -944,7 +943,7 @@ Tcl_UtfToTitle(
 	bytes = TclUtfToUniChar(src, &ch);
 	lowChar = Tcl_UniCharToLower(ch);
 
-	if (bytes < UtfCount(lowChar)) {
+	if (bytes < TclUtfCount(lowChar)) {
 	    memcpy(dst, src, (size_t) bytes);
 	    dst += bytes;
 	} else {