Merge 9.0. Also fix replace profile handling of orphan surrogates

1 files changed, 61 insertions, 3 deletions

diff --git a/tests/encoding.test b/tests/encoding.test
index 8ed7c1d..db5680d 100644
--- a/tests/encoding.test
+++ b/tests/encoding.test
@@ -564,15 +564,27 @@ test encoding-16.18 {
 	return done
     } [namespace current]]
 } -result done
-test encoding-16.19 {UnicodeToUtfProc, bug [d19fe0a5b]} -body {
+test encoding-16.19 {Utf16ToUtfProc, bug [d19fe0a5b]} -body {
     encoding convertfrom utf-16 "\x41\x41\x41"
 } -result \u4141\uFFFD
-test encoding-16.20 {UnicodeToUtfProc, bug [d19fe0a5b]} -constraints deprecated -body {
+test encoding-16.20 {Utf16ToUtfProc, bug [d19fe0a5b]} -constraints deprecated -body {
     encoding convertfrom utf-16 "\xD8\xD8"
 } -result \uD8D8
-test encoding-16.21 {UnicodeToUtfProc, bug [d19fe0a5b]} -body {
+test encoding-16.21 {Utf16ToUtfProc, bug [d19fe0a5b]} -body {
     encoding convertfrom utf-32 "\x00\x00\x00\x00\x41\x41"
 } -result \x00\uFFFD
+test encoding-16.22 {Utf16ToUtfProc, strict, bug [db7a085bd9]} -body {
+    encoding convertfrom -profile strict utf-16le \x00\xD8
+} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\x00'}
+test encoding-16.23 {Utf16ToUtfProc, strict, bug [db7a085bd9]} -body {
+    encoding convertfrom -profile strict utf-16le \x00\xDC
+} -returnCodes 1 -result {unexpected byte sequence starting at index 0: '\x00'}
+test encoding-16.24 {Utf32ToUtfProc} -body {
+    encoding convertfrom utf-32 "\xFF\xFF\xFF\xFF"
+} -result \uFFFD
+test encoding-16.25 {Utf32ToUtfProc} -body {
+    encoding convertfrom utf-32 "\x01\x00\x00\x01"
+} -result \uFFFD
 
 test encoding-17.1 {UtfToUtf16Proc} -body {
     encoding convertto utf-16 "\U460DC"
@@ -604,6 +616,12 @@ test encoding-17.9 {Utf32ToUtfProc} -body {
 test encoding-17.10 {Utf32ToUtfProc} -body {
     encoding convertfrom -profile tcl8 utf-32 "\xFF\xFF\xFF\xFF"
 } -result \uFFFD
+test encoding-17.11 {Utf32ToUtfProc} -body {
+    encoding convertfrom -profile strict utf-32le "\x00\xD8\x00\x00"
+} -returnCodes error -result {unexpected byte sequence starting at index 0: '\x00'}
+test encoding-17.12 {Utf32ToUtfProc} -body {
+    encoding convertfrom -profile strict utf-32le "\x00\xDC\x00\x00"
+} -returnCodes error -result {unexpected byte sequence starting at index 0: '\x00'}
 
 test encoding-18.1 {TableToUtfProc on invalid input} -body {
 	list [catch {encoding convertto jis0208 \\} res] $res
@@ -1049,6 +1067,46 @@ test encoding-28.0 {all encodings load} -body {
 
 runtests
 
+test encoding-bug-183a1adcc0-1 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints {
+    testencoding
+} -body {
+    # Note - buffers are initialized to \xff
+    list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 1} result] $result
+} -result [list 0 [list nospace {} \xff]]
+
+test encoding-bug-183a1adcc0-2 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints {
+    testencoding
+} -body {
+    # Note - buffers are initialized to \xff
+    list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 0} result] $result
+} -result [list 0 [list nospace {} {}]]
+
+test encoding-bug-183a1adcc0-3 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints {
+    testencoding
+} -body {
+    # Note - buffers are initialized to \xff
+    list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 2} result] $result
+} -result [list 0 [list nospace {} \x00\x00]]
+
+test encoding-bug-183a1adcc0-4 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints {
+    testencoding
+} -body {
+    # Note - buffers are initialized to \xff
+    list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 3} result] $result
+} -result [list 0 [list nospace {} \x00\x00\xff]]
+
+test encoding-bug-183a1adcc0-5 {Bug [183a1adcc0] Buffer overflow Tcl_UtfToExternal} -constraints {
+    testencoding
+} -constraints {
+    knownBug
+} -body {
+    # The knownBug constraint is because test depends on TCL_UTF_MAX and
+    # also UtfToUtf16 assumes space required in destination buffer is
+    # sizeof(Tcl_UniChar) which is incorrect when TCL_UTF_MAX==4
+    # Note - buffers are initialized to \xff
+    list [catch {testencoding Tcl_UtfToExternal utf-16 A {start end} {} 4} result] $result
+} -result [list 0 [list ok {} [expr {$::tcl_platform(byteOrder) eq "littleEndian" ? "\x41\x00" : "\x00\x41"}]\x00\x00]]
+
 }
 
 test encoding-29.0 {get encoding nul terminator lengths} -constraints {