Merge 8.7

2 files changed, 45 insertions, 10 deletions

diff --git a/generic/tclStringObj.c b/generic/tclStringObj.c
index bb718ce..4b44517 100644
--- a/generic/tclStringObj.c
+++ b/generic/tclStringObj.c
@@ -3809,6 +3809,9 @@ TclStringReverse(
     String *stringPtr;
     Tcl_UniChar ch = 0;
     int inPlace = flags & TCL_STRING_IN_PLACE;
+#if TCL_UTF_MAX < 4
+    int needFlip = 0;
+#endif
 
     if (TclIsPureByteArray(objPtr)) {
 	int numBytes;
@@ -3827,10 +3830,9 @@ TclStringReverse(
     if (stringPtr->hasUnicode) {
 	Tcl_UniChar *from = Tcl_GetUnicode(objPtr);
 	Tcl_UniChar *src = from + stringPtr->numChars;
+	Tcl_UniChar *to;
 
 	if (!inPlace || Tcl_IsShared(objPtr)) {
-	    Tcl_UniChar *to;
-
 	    /*
 	     * Create a non-empty, pure unicode value, so we can coax
 	     * Tcl_SetObjLength into growing the unicode rep buffer.
@@ -3840,19 +3842,54 @@ TclStringReverse(
 	    Tcl_SetObjLength(objPtr, stringPtr->numChars);
 	    to = Tcl_GetUnicode(objPtr);
 	    while (--src >= from) {
+#if TCL_UTF_MAX < 4
+		ch = *src;
+		if ((ch & 0xF800) == 0xD800) {
+		    needFlip = 1;
+		}
+		*to++ = ch;
+#else
 		*to++ = *src;
+#endif
 	    }
 	} else {
 	    /*
 	     * Reversing in place.
 	     */
 
+#if TCL_UTF_MAX < 4
+	    to = src;
+#endif
 	    while (--src > from) {
 		ch = *src;
+#if TCL_UTF_MAX < 4
+		if ((ch & 0xF800) == 0xD800) {
+		    needFlip = 1;
+		}
+#endif
 		*src = *from;
 		*from++ = ch;
 	    }
 	}
+#if TCL_UTF_MAX < 4
+	if (needFlip) {
+	    /*
+	     * Flip back surrogate pairs.
+	     */
+
+	    from = to - stringPtr->numChars;
+	    while (--to >= from) {
+		ch = *to;
+		if ((ch & 0xFC00) == 0xD800) {
+		    if ((to-1 >= from) && ((to[-1] & 0xFC00) == 0xDC00)) {
+			to[0] = to[-1];
+			to[-1] = ch;
+			--to;
+		    }
+		}
+	    }
+	}
+#endif
     }
 
     if (objPtr->bytes) {
@@ -3876,8 +3913,8 @@ TclStringReverse(
 	     * Pass 1. Reverse the bytes of each multi-byte character.
 	     */
 
-	    int charCount = 0;
 	    int bytesLeft = numBytes;
+	    int chw;
 
 	    while (bytesLeft) {
 		/*
@@ -3886,18 +3923,16 @@ TclStringReverse(
 		 * skip calling Tcl_UtfCharComplete() here.
 		 */
 
-		int bytesInChar = TclUtfToUniChar(from, &ch);
+		int bytesInChar = TclUtfToUCS4(from, &chw);
 
 		ReverseBytes((unsigned char *)to, (unsigned char *)from,
 			bytesInChar);
 		to += bytesInChar;
 		from += bytesInChar;
 		bytesLeft -= bytesInChar;
-		charCount++;
 	    }
 
 	    from = to = objPtr->bytes;
-	    stringPtr->numChars = charCount;
 	}
 	/* Pass 2. Reverse all the bytes. */
 	ReverseBytes((unsigned char *)to, (unsigned char *)from, numBytes);
diff --git a/tests/string.test b/tests/string.test
index b01d059..0eaa3da 100644
--- a/tests/string.test
+++ b/tests/string.test
@@ -2086,19 +2086,19 @@ test string-24.15.$noComp {string reverse command - pure bytearray} {
     binary scan [run {tcl::string::reverse [binary format H* 010203]}] H* x
     set x
 } 030201
-test string-24.16.$noComp {string reverse command - surrogates} knownBug {
+test string-24.16.$noComp {string reverse command - surrogates} {
     run {string reverse \u0444bulb\uD83D\uDE02}
 } \uD83D\uDE02blub\u0444
-test string-24.17.$noComp {string reverse command - surrogates} knownBug {
+test string-24.17.$noComp {string reverse command - surrogates} {
     run {string reverse \uD83D\uDE02hello\uD83D\uDE02}
 } \uD83D\uDE02olleh\uD83D\uDE02
-test string-24.18.$noComp {string reverse command - surrogates} knownBug {
+test string-24.18.$noComp {string reverse command - surrogates} {
     set s \u0444bulb\uD83D\uDE02
     # shim shimmery ...
     string index $s 0
     run {string reverse $s}
 } \uD83D\uDE02blub\u0444
-test string-24.19.$noComp {string reverse command - surrogates} knownBug {
+test string-24.19.$noComp {string reverse command - surrogates} {
     set s \uD83D\uDE02hello\uD83D\uDE02
     # shim shimmery ...
     string index $s 0