Merge 8.6

3 files changed, 41 insertions, 3 deletions

diff --git a/generic/tkInt.h b/generic/tkInt.h
index 9e83a93..a4c2891 100644
--- a/generic/tkInt.h
+++ b/generic/tkInt.h
@@ -1396,9 +1396,11 @@ MODULE_SCOPE void	TkUnixSetXftClipRegion(Region clipRegion);
 #if TCL_UTF_MAX > 4
 #   define TkUtfToUniChar (size_t)Tcl_UtfToUniChar
 #   define TkUniCharToUtf (size_t)Tcl_UniCharToUtf
+#   define TkUtfPrev Tcl_UtfPrev
 #else
     MODULE_SCOPE size_t TkUtfToUniChar(const char *, int *);
     MODULE_SCOPE size_t TkUniCharToUtf(int, char *);
+    MODULE_SCOPE const char *TkUtfPrev(const char *, const char *);
 #endif
 
 #if TCL_MAJOR_VERSION > 8
diff --git a/generic/tkTextIndex.c b/generic/tkTextIndex.c
index 1db1208..70a53ae 100644
--- a/generic/tkTextIndex.c
+++ b/generic/tkTextIndex.c
@@ -436,7 +436,7 @@ TkTextMakeByteIndex(
 		 */
 
 		start = segPtr->body.chars + (byteIndex - index);
-		p = Tcl_UtfPrev(start, segPtr->body.chars);
+		p = TkUtfPrev(start, segPtr->body.chars);
 		p += TkUtfToUniChar(p, &ch);
 		indexPtr->byteIndex += p - start;
 	    }
@@ -2125,7 +2125,7 @@ TkTextIndexBackChars(
 	    if (segPtr->typePtr == &tkTextCharType) {
 		start = segPtr->body.chars;
 		end = segPtr->body.chars + segSize;
-		for (p = end; ; p = Tcl_UtfPrev(p, start)) {
+		for (p = end; ; p = TkUtfPrev(p, start)) {
 		    if (charCount == 0) {
 			dstPtr->byteIndex -= (end - p);
 			goto backwardCharDone;
@@ -2366,7 +2366,7 @@ StartEnd(
 		}
 		if (offset > 0) {
 		    chSize = (segPtr->body.chars + offset
-			    - Tcl_UtfPrev(segPtr->body.chars + offset,
+			    - TkUtfPrev(segPtr->body.chars + offset,
 			    segPtr->body.chars));
 		}
 		firstChar = 0;
diff --git a/generic/tkUtil.c b/generic/tkUtil.c
index f899cf2..01155c9 100644
--- a/generic/tkUtil.c
+++ b/generic/tkUtil.c
@@ -1266,7 +1266,43 @@ size_t TkUniCharToUtf(int ch, char *buf)
     }
     return Tcl_UniCharToUtf(ch, buf);
 }
+/*
+ *---------------------------------------------------------------------------
+ *
+ * TkUtfPrev --
+ *
+ *	Almost the same as Tcl_UtfPrev.
+ *	This function is capable of jumping over a upper/lower surrogate pair.
+ *	So, might jump back up to 6 bytes.
+ *
+ * Results:
+ *	pointer to the first byte of the current UTF-8 character. A surrogate
+ *	pair is also handled as being a single entity.
+ *
+ * Side effects:
+ *	None.
+ *
+ *---------------------------------------------------------------------------
+ */
 
+const char *
+TkUtfPrev(
+    const char *src,	/* The UTF-8 string. */
+    const char *start)		/* Start position of string */
+{
+    const char *p = Tcl_UtfPrev(src, start);
+    const char *first = Tcl_UtfPrev(p, start);
+    int ch;
+
+#if TCL_UTF_MAX == 3
+    if ((src - start > 3) && ((src[-1] & 0xC0) == 0x80) && ((src[-2] & 0xC0) == 0x80)
+	    && ((src[-3] & 0xC0) == 0x80) && (UCHAR(src[-4]) >= 0xF0)) {
+	return src - 4;
+    }
+#endif
+
+    return (first + TkUtfToUniChar(first, &ch) >= src) ? first : p ;
+}
 
 #endif