Bug [5be203d6ca] - better fix and add test

2 files changed, 14 insertions, 14 deletions

diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index 42fcb7f..405c179 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -10,6 +10,7 @@
  */
 
 #include "tclInt.h"
+#include <assert.h>
 
 typedef size_t (LengthProc)(const char *src);
 
@@ -3397,16 +3398,13 @@ TableToUtfProc(
 	}
 	byte = *((unsigned char *) src);
 	if (prefixBytes[byte]) {
-	    src++;
-	    if (src >= srcEnd) {
+	    if (src >= srcEnd-1) {
+                /* Prefix byte but nothing after it */
 		if (!(flags & TCL_ENCODING_END)) {
-                    /* Suffix bytes expected, don't consume prefix */
-		    src--;
+                    /* More data to come */
 		    result = TCL_CONVERT_MULTIBYTE;
 		    break;
 		} else if (PROFILE_STRICT(flags)) {
-                    /* Truncation. Do not consume so error location correct */
-		    src--;
 		    result = TCL_CONVERT_SYNTAX;
 		    break;
 		} else if (PROFILE_REPLACE(flags)) {
@@ -3415,6 +3413,7 @@ TableToUtfProc(
 		    ch = (unsigned) byte;
 		}
 	    } else {
+                ++src;
 		ch = toUnicode[byte][*((unsigned char *)src)];
 	    }
 	} else {
@@ -3448,14 +3447,7 @@ TableToUtfProc(
 	src++;
     }
 
-    if (src > srcEnd) {
-        /*
-         * This can happen in the case of a trailing truncated byte sequence
-         * where src is incremented past the prefix byte which is the last.
-         * See bug [5be203d6ca]
-         */
-        src--;
-    }
+    assert(src <= srcEnd);
     *srcReadPtr = src - srcStart;
     *dstWrotePtr = dst - dstStart;
     *dstCharsPtr = numChars;
diff --git a/tests/utfext.test b/tests/utfext.test
index bc996c9..a6c9dd8 100644
--- a/tests/utfext.test
+++ b/tests/utfext.test
@@ -93,6 +93,14 @@ test xx-bufferoverflow {buffer overflow Tcl_ExternalToUtf} -body {
 # % testencoding Tcl_ExternalToUtf utf-8 abcdefgh {start end noterminate charlimit} {} 20 rv wv cv
 # nospace {} abcÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ
 
+test TableToUtf-bug-5be203d6ca {Bug 5be203d6ca - truncated prefix in table encoding} -body {
+    set src \x82\x4f\x82\x50\x82 
+    lassign [testencoding Tcl_ExternalToUtf shiftjis $src {start} 0 16 srcRead dstWritten charsWritten] buf 
+    set result [list [testencoding Tcl_ExternalToUtf shiftjis $src {start} 0 16 srcRead dstWritten charsWritten] $srcRead $dstWritten $charsWritten]
+    lappend result {*}[list [testencoding Tcl_ExternalToUtf shiftjis [string range $src $srcRead end] {end} 0 10 srcRead dstWritten charsWritten] $srcRead $dstWritten $charsWritten]
+} -result [list [list multibyte 0 \xEF\xBC\x90\xEF\xBC\x91\x00\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF] 4 6 2 [list ok 0 \xC2\x82\x00\xFF\xFF\xFF\xFF\xFF\xFF\xFF] 1 2 1]
+
+
 ::tcltest::cleanupTests
 return