(merge-mark core-8-6-branch).

Give state to UtfToUtfProc() and UnicodeToUtfProc() in which the last (intermediate) Tcl_UniChar can be remembered: This way those functions can produce 4-byte UTF-8 sequences built from surrogate pairs if Tcl is compiled with TCL_UTF_MAX=4. For TCL_UTF_MAX=3 and TCL_UTF_MAX=6 this has no effect.

1 files changed, 33 insertions, 24 deletions

diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index 7ab33f8..d430b32 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -2293,8 +2293,11 @@ UtfToUtfProc(
     const char *srcStart, *srcEnd, *srcClose;
     const char *dstStart, *dstEnd;
     int result, numChars, charLimit = INT_MAX;
-    Tcl_UniChar ch = 0;
+    Tcl_UniChar *chPtr = (Tcl_UniChar *) statePtr;
 
+    if (flags & TCL_ENCODING_START) {
+    	*statePtr = 0;
+    }
     result = TCL_OK;
 
     srcStart = src;
@@ -2346,12 +2349,12 @@ UtfToUtfProc(
 	     * incomplete char its bytes are made to represent themselves.
 	     */
 
-	    ch = (unsigned char) *src;
+	    *chPtr = (unsigned char) *src;
 	    src += 1;
-	    dst += Tcl_UniCharToUtf(ch, dst);
+	    dst += Tcl_UniCharToUtf(*chPtr, dst);
 	} else {
-	    src += TclUtfToUniChar(src, &ch);
-	    dst += Tcl_UniCharToUtf(ch, dst);
+	    src += TclUtfToUniChar(src, chPtr);
+	    dst += Tcl_UniCharToUtf(*chPtr, dst);
 	}
     }
 
@@ -2407,8 +2410,11 @@ UnicodeToUtfProc(
     const char *srcStart, *srcEnd;
     const char *dstEnd, *dstStart;
     int result, numChars, charLimit = INT_MAX;
-    Tcl_UniChar ch = 0;
+    Tcl_UniChar *chPtr = (Tcl_UniChar *) statePtr;
 
+    if (flags & TCL_ENCODING_START) {
+    	*statePtr = 0;
+    }
     if (flags & TCL_ENCODING_CHAR_LIMIT) {
 	charLimit = *dstCharsPtr;
     }
@@ -2436,11 +2442,11 @@ UnicodeToUtfProc(
 	 * Tcl_UniChar-size data.
 	 */
 
-	ch = *(Tcl_UniChar *)src;
-	if (ch && ch < 0x80) {
-	    *dst++ = (ch & 0xFF);
+	*chPtr = *(Tcl_UniChar *)src;
+	if (*chPtr && *chPtr < 0x80) {
+	    *dst++ = (*chPtr & 0xFF);
 	} else {
-	    dst += Tcl_UniCharToUtf(ch, dst);
+	    dst += Tcl_UniCharToUtf(*chPtr, dst);
 	}
 	src += sizeof(Tcl_UniChar);
     }
@@ -2497,8 +2503,11 @@ UtfToUnicodeProc(
 {
     const char *srcStart, *srcEnd, *srcClose, *dstStart, *dstEnd;
     int result, numChars;
-    Tcl_UniChar ch = 0;
+    Tcl_UniChar *chPtr = (Tcl_UniChar *) statePtr;
 
+    if (flags & TCL_ENCODING_START) {
+    	*statePtr = 0;
+    }
     srcStart = src;
     srcEnd = src + srcLen;
     srcClose = srcEnd;
@@ -2524,7 +2533,7 @@ UtfToUnicodeProc(
 	    result = TCL_CONVERT_NOSPACE;
 	    break;
 	}
-	src += TclUtfToUniChar(src, &ch);
+	src += TclUtfToUniChar(src, chPtr);
 
 	/*
 	 * Need to handle this in a way that won't cause misalignment by
@@ -2533,23 +2542,23 @@ UtfToUnicodeProc(
 
 #ifdef WORDS_BIGENDIAN
 #if TCL_UTF_MAX > 4
-	*dst++ = (ch >> 24);
-	*dst++ = ((ch >> 16) & 0xFF);
-	*dst++ = ((ch >> 8) & 0xFF);
-	*dst++ = (ch & 0xFF);
+	*dst++ = (*chPtr >> 24);
+	*dst++ = ((*chPtr >> 16) & 0xFF);
+	*dst++ = ((*chPtr >> 8) & 0xFF);
+	*dst++ = (*chPtr & 0xFF);
 #else
-	*dst++ = (ch >> 8);
-	*dst++ = (ch & 0xFF);
+	*dst++ = (*chPtr >> 8);
+	*dst++ = (*chPtr & 0xFF);
 #endif
 #else
 #if TCL_UTF_MAX > 4
-	*dst++ = (ch & 0xFF);
-	*dst++ = ((ch >> 8) & 0xFF);
-	*dst++ = ((ch >> 16) & 0xFF);
-	*dst++ = (ch >> 24);
+	*dst++ = (*chPtr & 0xFF);
+	*dst++ = ((*chPtr >> 8) & 0xFF);
+	*dst++ = ((*chPtr >> 16) & 0xFF);
+	*dst++ = (*chPtr >> 24);
 #else
-	*dst++ = (ch & 0xFF);
-	*dst++ = (ch >> 8);
+	*dst++ = (*chPtr & 0xFF);
+	*dst++ = (*chPtr >> 8);
 #endif
 #endif
     }