From 901525301c284507e65f39f5c68785ab9ec1eb16 Mon Sep 17 00:00:00 2001
From: dgp <dgp@users.sourceforge.net>
Date: Mon, 13 Apr 2020 00:09:43 +0000
Subject: added test case covering [c61818e4c9] - string trim for not valid
 utf-8 sequence (mistakenly considers NTS-zero char as a continuation of utf-8
 pair)

---
 tests/string.test | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/tests/string.test b/tests/string.test
index e1ae63a..f6eaaf0 100644
--- a/tests/string.test
+++ b/tests/string.test
@@ -1459,6 +1459,23 @@ test string-20.4 {string trimright} {
 test string-20.5 {string trimright} {
     string trimright ""
 } {}
+test string-20.6 {string trim on not valid utf-8 sequence (consider NTS as continuation char), bug [c61818e4c9]} -setup {
+    interp alias {} bytes {} encoding convertfrom identity
+} -body {
+    set result {}
+    set a [bytes \xc0\x80\x88]
+    set b foo$a
+    set m [list \u0000 U \x88 V [bytes \x88] W]
+    lappend result [string map $m $b]
+    lappend result [string map $m [string trimright $b x]]
+    lappend result [string map $m [string trimright $b \u0000]]
+    lappend result [string map $m [string trimleft $b fox]]
+    lappend result [string map $m [string trimleft $b fo\u0000]]
+    lappend result [string map $m [string trim $b fox]]
+    lappend result [string map $m [string trim $b fo\u0000]]
+} -result [list {*}[lrepeat 3 fooUV] {*}[lrepeat 2 UV V]] -cleanup {
+    interp alias {} bytes {}
+}
 
 test string-21.1 {string wordend} {
     list [catch {string wordend a} msg] $msg
-- 
cgit v0.12


From e4426ffac822281e598f797f1a787a4bd05c090b Mon Sep 17 00:00:00 2001
From: dgp <dgp@users.sourceforge.net>
Date: Mon, 13 Apr 2020 00:12:26 +0000
Subject: Convert test to not directly use identity encoding.

---
 tests/string.test | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)

diff --git a/tests/string.test b/tests/string.test
index f6eaaf0..72b2a49 100644
--- a/tests/string.test
+++ b/tests/string.test
@@ -1459,13 +1459,11 @@ test string-20.4 {string trimright} {
 test string-20.5 {string trimright} {
     string trimright ""
 } {}
-test string-20.6 {string trim on not valid utf-8 sequence (consider NTS as continuation char), bug [c61818e4c9]} -setup {
-    interp alias {} bytes {} encoding convertfrom identity
-} -body {
+test string-20.6 {string trim on not valid utf-8 sequence (consider NTS as continuation char), bug [c61818e4c9]} {
     set result {}
-    set a [bytes \xc0\x80\x88]
+    set a [bytestring \xc0\x80\x88]
     set b foo$a
-    set m [list \u0000 U \x88 V [bytes \x88] W]
+    set m [list \u0000 U \x88 V [bytestring \x88] W]
     lappend result [string map $m $b]
     lappend result [string map $m [string trimright $b x]]
     lappend result [string map $m [string trimright $b \u0000]]
@@ -1473,9 +1471,7 @@ test string-20.6 {string trim on not valid utf-8 sequence (consider NTS as conti
     lappend result [string map $m [string trimleft $b fo\u0000]]
     lappend result [string map $m [string trim $b fox]]
     lappend result [string map $m [string trim $b fo\u0000]]
-} -result [list {*}[lrepeat 3 fooUV] {*}[lrepeat 2 UV V]] -cleanup {
-    interp alias {} bytes {}
-}
+} [list {*}[lrepeat 3 fooUV] {*}[lrepeat 2 UV V]]
 
 test string-21.1 {string wordend} {
     list [catch {string wordend a} msg] $msg
-- 
cgit v0.12


From a8641c2da18720d07e3a0235703a9587b46d12ca Mon Sep 17 00:00:00 2001
From: dgp <dgp@users.sourceforge.net>
Date: Mon, 13 Apr 2020 01:40:18 +0000
Subject: Another test for [string trimright] that demonstrates its own
 failures, not those of Tcl_UtfPrev.

---
 tests/string.test | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/tests/string.test b/tests/string.test
index 72b2a49..05a0623 100644
--- a/tests/string.test
+++ b/tests/string.test
@@ -1472,6 +1472,21 @@ test string-20.6 {string trim on not valid utf-8 sequence (consider NTS as conti
     lappend result [string map $m [string trim $b fox]]
     lappend result [string map $m [string trim $b fo\u0000]]
 } [list {*}[lrepeat 3 fooUV] {*}[lrepeat 2 UV V]]
+test string-20.7 {[c61818e4c9] [string trimright] fails when UtfPrev is ok} {
+    set result {}
+    set a [bytestring \xE8\x80]
+    set b foo$a
+    set m [list \xE8 U \x80 V [bytestring \xE8] W [bytestring \x80] X]]
+    lappend result [string map $m $b]
+    lappend result [string map $m [string trimright $b x]]
+    lappend result [string map $m [string trimright $b \xE8]]
+    lappend result [string map $m [string trimright $b [bytestring \xE8]]]
+    lappend result [string map $m [string trimright $b \x80]]
+    lappend result [string map $m [string trimright $b [bytestring \x80]]]
+    lappend result [string map $m [string trimright $b \xE8\x80]]
+    lappend result [string map $m [string trimright $b [bytestring \xE8\x80]]]
+    lappend result [string map $m [string trimright $b \u0000]]
+} [list {*}[lrepeat 4 fooUV] {*}[lrepeat 2 fooU] {*}[lrepeat 2 foo] fooUV]
 
 test string-21.1 {string wordend} {
     list [catch {string wordend a} msg] $msg
-- 
cgit v0.12