From 5dab0fa5de0367d41b4ad28b9a66c3cd1bf818ba Mon Sep 17 00:00:00 2001
From: apnadkarni <apnmbx-wits@yahoo.com>
Date: Sun, 2 Jul 2023 04:55:47 +0000
Subject: Backport fix [66ffafd309]. DBCS infinite loop on invalid encoding

---
 generic/tclEncoding.c |  9 +++++----
 tests/chanio.test     |  2 +-
 tests/encoding.test   | 21 +++++++++++++++++++++
 tests/io.test         |  2 +-
 4 files changed, 28 insertions(+), 6 deletions(-)

diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index f9f93fe..d0f04a7 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -3400,19 +3400,19 @@ TableToUtfProc(
 	    src++;
 	    if (src >= srcEnd) {
 		if (!(flags & TCL_ENCODING_END)) {
+                    /* Suffix bytes expected, don't consume prefix */
 		    src--;
 		    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)) {
 		    ch = UNICODE_REPLACE_CHAR;
 		} else {
-		    src--; /* See bug [bdcb5126c0] */
-		    result = TCL_CONVERT_MULTIBYTE;
-		    break;
+		    ch = (unsigned) byte;
 		}
 	    } else {
 		ch = toUnicode[byte][*((unsigned char *)src)];
@@ -3421,6 +3421,7 @@ TableToUtfProc(
 	    ch = pageZero[byte];
 	}
 	if ((ch == 0) && (byte != 0)) {
+            /* Prefix+suffix pair is invalid */
 	    if (PROFILE_STRICT(flags)) {
 		result = TCL_CONVERT_SYNTAX;
 		break;
@@ -3431,7 +3432,7 @@ TableToUtfProc(
 	    if (PROFILE_REPLACE(flags)) {
 		ch = UNICODE_REPLACE_CHAR;
 	    } else {
-		ch = (Tcl_UniChar)byte;
+                ch = (Tcl_UniChar)byte;
 	    }
 	}
 
diff --git a/tests/chanio.test b/tests/chanio.test
index b80c926..c5d3aca 100644
--- a/tests/chanio.test
+++ b/tests/chanio.test
@@ -1104,7 +1104,7 @@ test chan-io-7.3 {FilterInputBytes: split up character at EOF} -setup {
     lappend x [chan gets $f line] $line
 } -cleanup {
     chan close $f
-} -result [list 15 "1234567890123０１" 18 0 1 -1 ""]
+} -result [list 16 "1234567890123０１\x82" 18 0 1 -1 ""]
 test chan-io-7.4 {FilterInputBytes: recover from split up character} -setup {
     variable x ""
 } -constraints {stdio fileevent} -body {
diff --git a/tests/encoding.test b/tests/encoding.test
index 4abc10e..8167357 100644
--- a/tests/encoding.test
+++ b/tests/encoding.test
@@ -1119,6 +1119,27 @@ test encoding-bug-6a3e2cb0f0-3 {Bug [6a3e2cb0f0] - invalid bytes in escape encod
     encoding convertfrom -profile replace iso2022-jp x\x1b\x7aaby
 } -result x\uFFFDy
 
+test encoding-bug-66ffafd309-1-tcl8 {Bug [66ffafd309] - truncated DBCS} -body {
+    encoding convertfrom -profile tcl8 gb12345 x
+} -result x
+test encoding-bug-66ffafd309-1-strict {Bug [66ffafd309] - truncated DBCS} -body {
+    encoding convertfrom -profile strict gb12345 x
+} -result {unexpected byte sequence starting at index 0: '\x78'} -returnCodes error
+test encoding-bug-66ffafd309-1-replace {Bug [66ffafd309] - truncated DBCS} -body {
+    encoding convertfrom -profile replace gb12345 x
+} -result \uFFFD
+test encoding-bug-66ffafd309-2-tcl8 {Bug [66ffafd309] - invalid DBCS} -body {
+    # Not truncated but invalid
+    encoding convertfrom -profile tcl8 jis0208 \x78\x79
+} -result \x78\x79
+test encoding-bug-66ffafd309-2-strict {Bug [66ffafd309] - invalid DBCS} -body {
+    # Not truncated but invalid
+    encoding convertfrom -profile strict jis0208 \x78\x79
+} -result {unexpected byte sequence starting at index 1: '\x79'} -returnCodes error
+test encoding-bug-66ffafd309-2-replace {Bug [66ffafd309] - invalid DBCS} -body {
+    # Not truncated but invalid
+    encoding convertfrom -profile replace jis0208 \x78\x79
+} -result \uFFFD\uFFFD
 # cleanup
 namespace delete ::tcl::test::encoding
 ::tcltest::cleanupTests
diff --git a/tests/io.test b/tests/io.test
index c0191ee..ca636ce 100644
--- a/tests/io.test
+++ b/tests/io.test
@@ -1136,7 +1136,7 @@ test io-7.3 {FilterInputBytes: split up character at EOF} {testchannel} {
     lappend x [gets $f line] $line
     close $f
     set x
-} [list 15 "1234567890123０１" 18 0 1 -1 ""]
+} [list 16 "1234567890123０１\x82" 18 0 1 -1 ""]
 test io-7.4 {FilterInputBytes: recover from split up character} {stdio fileevent} {
     set f [open "|[list [interpreter] $path(cat)]" w+]
     fconfigure $f -encoding binary -buffering none
-- 
cgit v0.12