From fc6f32f9ee409bf755b420263eb698e8e5d6ccca Mon Sep 17 00:00:00 2001
From: Chris Webb <chris@arachsys.com>
Date: Tue, 9 May 2023 22:56:55 +0200
Subject: [PATCH] Stop _ from tearing multibyte UTF-8 sequences

Fixes #4887

[ja: add test]
---
 src/normal.cc                               | 9 +++++----
 test/regression/4887-torn-utf8-sequence/cmd | 1 +
 test/regression/4887-torn-utf8-sequence/in  | 0
 test/regression/4887-torn-utf8-sequence/out | 1 +
 4 files changed, 7 insertions(+), 4 deletions(-)
 create mode 100644 test/regression/4887-torn-utf8-sequence/cmd
 create mode 100644 test/regression/4887-torn-utf8-sequence/in
 create mode 100644 test/regression/4887-torn-utf8-sequence/out

diff --git a/src/normal.cc b/src/normal.cc
index e3220521..2bc7fc17 100644
--- a/src/normal.cc
+++ b/src/normal.cc
@@ -1768,6 +1768,7 @@ void spaces_to_tabs(Context& context, NormalParams params)
 
 void trim_selections(Context& context, NormalParams)
 {
+    using Utf8It = utf8::iterator<BufferIterator>;
     auto& buffer = context.buffer();
     ScopedSelectionEdition selection_edition{context};
     auto& selections = context.selections();
@@ -1776,8 +1777,8 @@ void trim_selections(Context& context, NormalParams)
     for (int i = 0; i < (int)selections.size(); ++i)
     {
         auto& sel = selections[i];
-        auto beg = buffer.iterator_at(sel.min());
-        auto end = buffer.iterator_at(sel.max());
+        auto beg = Utf8It{buffer.iterator_at(sel.min()), buffer};
+        auto end = Utf8It{buffer.iterator_at(sel.max()), buffer};
         while (beg != end and is_blank(*beg))
             ++beg;
         while (beg != end and is_blank(*end))
@@ -1787,8 +1788,8 @@ void trim_selections(Context& context, NormalParams)
             to_remove.push_back(i);
         else
         {
-            sel.min() = beg.coord();
-            sel.max() = end.coord();
+            sel.min() = beg.base().coord();
+            sel.max() = end.base().coord();
         }
     }
 
diff --git a/test/regression/4887-torn-utf8-sequence/cmd b/test/regression/4887-torn-utf8-sequence/cmd
new file mode 100644
index 00000000..8a70e75f
--- /dev/null
+++ b/test/regression/4887-torn-utf8-sequence/cmd
@@ -0,0 +1 @@
+:exec %{<percent>cç<lt>esc>x_<semicolon>iab<lt>esc>}<ret>
diff --git a/test/regression/4887-torn-utf8-sequence/in b/test/regression/4887-torn-utf8-sequence/in
new file mode 100644
index 00000000..e69de29b
diff --git a/test/regression/4887-torn-utf8-sequence/out b/test/regression/4887-torn-utf8-sequence/out
new file mode 100644
index 00000000..155dd4fa
--- /dev/null
+++ b/test/regression/4887-torn-utf8-sequence/out
@@ -0,0 +1 @@
+abç