merge 8.5

2 files changed, 18 insertions, 14 deletions

diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index 46a18e4..4103eff 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -631,24 +631,25 @@ Tcl_NumUtfChars(
 	    i++;
 	}
 	/* Loop over the remaining string where call must happen */
-	while ((src < endPtr) && Tcl_UtfCharComplete(src, endPtr - src)) {
+	while (src < endPtr) {
+	    if (Tcl_UtfCharComplete(src, endPtr - src)) {
 #if TCL_UTF_MAX < 4
-	    if (((unsigned)UCHAR(*src) - 0xF0) < 5) {
-		/* treat F0 - F4 as single character */
-		ch = 0;
-		src++;
-	    } else
+		if (((unsigned)UCHAR(*src) - 0xF0) < 5) {
+		    /* treat F0 - F4 as single character */
+		    ch = 0;
+		    src++;
+		} else
 #endif
-	    src += TclUtfToUniChar(src, &ch);
+		src += TclUtfToUniChar(src, &ch);
+	    } else {
+		/*
+		 * src points to incomplete UTF-8 sequence 
+		 * Treat first byte as character and count it
+		 */
+		src++;
+	    }
 	    i++;
 	}
-	if (src < endPtr) {
-	    /*
-	     * String ends in an incomplete UTF-8 sequence.
-	     * Count every byte in it.
-	     */
-	    i += endPtr - src;
-	}
     }
     return i;
 }
diff --git a/tests/utf.test b/tests/utf.test
index 0acbb76..0e5aba6 100644
--- a/tests/utf.test
+++ b/tests/utf.test
@@ -179,6 +179,9 @@ test utf-4.12.1 {Tcl_NumUtfChars: #4-byte UTF-8 character} {testnumutfchars test
 test utf-4.12.2 {Tcl_NumUtfChars: #4-byte UTF-8 character} {testnumutfchars testbytestring utf16} {
     testnumutfchars [testbytestring \xF0\x9F\x92\xA9] end
 } 2
+test utf-4.13 {Tcl_NumUtfChars: end of string} {testnumutfchars testbytestring} { 
+    testnumutfchars foobar[testbytestring \xF2\xC2\xA0] end
+} 8
 
 test utf-5.1 {Tcl_UtfFindFirst} {testfindfirst testbytestring} {
     testfindfirst [testbytestring abcbc] 98