Feature-complete

1 files changed, 56 insertions, 1 deletions

diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index 4dd1e09..eda317f 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -799,7 +799,7 @@ Tcl_UtfCharComplete(
  */
 
 int
-Tcl_NumUtfChars(
+TclNumUtfChars(
     const char *src,	/* The UTF-8 string to measure. */
     int length)		/* The length of the string in bytes, or -1
 			 * for strlen(string). */
@@ -850,6 +850,61 @@ Tcl_NumUtfChars(
     return i;
 }
 
+#if TCL_UTF_MAX > 3
+#undef Tcl_NumUtfChars
+int
+Tcl_NumUtfChars(
+    const char *src,	/* The UTF-8 string to measure. */
+    int length)		/* The length of the string in bytes, or -1
+			 * for strlen(string). */
+{
+    unsigned short ch = 0;
+    int i = 0;
+
+    if (length < 0) {
+	/* string is NUL-terminated, so TclUtfToUniChar calls are safe. */
+	while ((*src != '\0') && (i < INT_MAX)) {
+	    src += Tcl_UtfToChar16(src, &ch);
+	    i++;
+	}
+    } else {
+	/* Will return value between 0 and length. No overflow checks. */
+
+	/* Pointer to the end of string. Never read endPtr[0] */
+	const char *endPtr = src + length;
+	/* Pointer to last byte where optimization still can be used */
+	const char *optPtr = endPtr - 4;
+
+	/*
+	 * Optimize away the call in this loop. Justified because...
+	 * when (src <= optPtr), (endPtr - src) >= (endPtr - optPtr)
+	 * By initialization above (endPtr - optPtr) = TCL_UTF_MAX
+	 * So (endPtr - src) >= TCL_UTF_MAX, and passing that to
+	 * Tcl_UtfCharComplete we know will cause return of 1.
+	 */
+	while (src <= optPtr
+		/* && Tcl_UtfCharComplete(src, endPtr - src) */ ) {
+	    src += Tcl_UtfToChar16(src, &ch);
+	    i++;
+	}
+	/* Loop over the remaining string where call must happen */
+	while (src < endPtr) {
+	    if (Tcl_UtfCharComplete(src, endPtr - src)) {
+		src += Tcl_UtfToChar16(src, &ch);
+	    } else {
+		/*
+		 * src points to incomplete UTF-8 sequence
+		 * Treat first byte as character and count it
+		 */
+		src++;
+	    }
+	    i++;
+	}
+    }
+    return i;
+}
+#endif
+
 /*
  *---------------------------------------------------------------------------
  *