Fix [5607d6482c]: strict ucs-2 never implemented (TIP #346/#656)

2 files changed, 31 insertions, 13 deletions

diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index 1d3a3eb..d2b0efc 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -573,13 +573,13 @@ TclInitEncodingSubsystem(void)
     type.freeProc	= NULL;
     type.nullSize	= 2;
     type.encodingName   = "ucs-2le";
-    type.clientData	= INT2PTR(TCL_ENCODING_LE|TCL_ENCODING_NOCOMPLAIN);
+    type.clientData	= INT2PTR(TCL_ENCODING_LE);
     Tcl_CreateEncoding(&type);
     type.encodingName   = "ucs-2be";
-    type.clientData	= INT2PTR(TCL_ENCODING_NOCOMPLAIN);
+    type.clientData	= INT2PTR(0);
     Tcl_CreateEncoding(&type);
     type.encodingName   = "ucs-2";
-    type.clientData	= INT2PTR(isLe.c|TCL_ENCODING_NOCOMPLAIN);
+    type.clientData	= INT2PTR(isLe.c);
     Tcl_CreateEncoding(&type);
 
     type.toUtfProc	= Utf32ToUtfProc;
@@ -601,13 +601,13 @@ TclInitEncodingSubsystem(void)
     type.freeProc	= NULL;
     type.nullSize	= 2;
     type.encodingName   = "utf-16le";
-    type.clientData	= INT2PTR(TCL_ENCODING_LE);
+    type.clientData	= INT2PTR(TCL_ENCODING_LE|ENCODING_UTF);
     Tcl_CreateEncoding(&type);
     type.encodingName   = "utf-16be";
-    type.clientData	= INT2PTR(0);
+    type.clientData	= INT2PTR(ENCODING_UTF);
     Tcl_CreateEncoding(&type);
     type.encodingName   = "utf-16";
-    type.clientData	= INT2PTR(isLe.c);
+    type.clientData	= INT2PTR(isLe.c|ENCODING_UTF);
     Tcl_CreateEncoding(&type);
 
 #ifndef TCL_NO_DEPRECATED
@@ -2984,10 +2984,7 @@ UtfToUcs2Proc(
 				 * output buffer. */
 {
     const char *srcStart, *srcEnd, *srcClose, *dstStart, *dstEnd;
-    int result, numChars;
-#if TCL_UTF_MAX < 4
-    int len;
-#endif
+    int result, numChars, len;
     Tcl_UniChar ch = 0;
 
     flags |= PTR2INT(clientData);
@@ -3017,17 +3014,32 @@ UtfToUcs2Proc(
 	    break;
 	}
 #if TCL_UTF_MAX < 4
-	src += (len = TclUtfToUniChar(src, &ch));
+	len = TclUtfToUniChar(src, &ch);
 	if ((ch >= 0xD800) && (len < 3)) {
+	    if (STOPONERROR) {
+			result = TCL_CONVERT_UNKNOWN;
+			break;
+	    }
+	    src += len;
 	    src += TclUtfToUniChar(src, &ch);
 	    ch = 0xFFFD;
 	}
 #else
-	src += TclUtfToUniChar(src, &ch);
+	len = TclUtfToUniChar(src, &ch);
 	if (ch > 0xFFFF) {
+	    if (STOPONERROR) {
+			result = TCL_CONVERT_UNKNOWN;
+			break;
+	    }
 	    ch = 0xFFFD;
 	}
 #endif
+	if (STOPONERROR && ((ch & ~0x7FF) == 0xD800)) {
+	    result = TCL_CONVERT_SYNTAX;
+	    break;
+	}
+
+	src += len;
 
 	/*
 	 * Need to handle this in a way that won't cause misalignment by
diff --git a/tests/encoding.test b/tests/encoding.test
index 03f0273..83e75be 100644
--- a/tests/encoding.test
+++ b/tests/encoding.test
@@ -561,7 +561,7 @@ test encoding-16.9 {
 test encoding-17.1 {UtfToUtf16Proc} -body {
     encoding convertto utf-16 "\U460DC"
 } -result "\xD8\xD8\xDC\xDC"
-test encoding-17.2 {UtfToUcs2Proc} -body {
+test encoding-17.2 {UtfToUcs2Proc, invalid testcase, see [5607d6482c]} -constraints deprecated -body {
     encoding convertfrom utf-16 [encoding convertto ucs-2 "\U460DC"]
 } -result "\uFFFD"
 test encoding-17.3 {UtfToUtf16Proc} -body {
@@ -853,6 +853,12 @@ test encoding-24.42 {Parse invalid utf-8, fallback to cp1252 [885c86a9a0]} -body
 test encoding-24.43 {Parse invalid utf-8, fallback to cp1252 [885c86a9a0]} -body {
     encoding convertfrom -nocomplain utf-8 \x80
 } -result \u20AC
+test encoding-24.44 {Try to generate invalid ucs-2 with -strict} -body {
+    encoding convertto -strict ucs-2 \uD800
+} -returnCodes 1 -result {unexpected character at index 0: 'U+00D800'}
+test encoding-24.45 {Try to generate invalid ucs-2 with -strict} -body {
+    encoding convertto -strict ucs-2 \U10000
+} -returnCodes 1 -result {unexpected character at index 0: 'U+010000'}
 
 file delete [file join [temporaryDirectory] iso2022.txt]