From fa99afe35eb5d8cf01d875e12b53165bf1104a60 Mon Sep 17 00:00:00 2001
From: bfredl <bjorn.linse@gmail.com>
Date: Wed, 4 Sep 2024 12:09:42 +0200
Subject: fix(multibyte): handle backspace of wide clusters in replace mode

Make utf_head_off more robust against invalid sequences
and embedded NUL chars
---
 test/functional/editor/mode_insert_spec.lua | 93 +++++++++++++++++++++++++++++
 test/unit/mbyte_spec.lua                    | 13 +++-
 2 files changed, 103 insertions(+), 3 deletions(-)

(limited to 'test')

diff --git a/test/functional/editor/mode_insert_spec.lua b/test/functional/editor/mode_insert_spec.lua
index fc1e6c4ee4..87d5c46134 100644
--- a/test/functional/editor/mode_insert_spec.lua
+++ b/test/functional/editor/mode_insert_spec.lua
@@ -351,4 +351,97 @@ describe('insert-mode', function()
       eq(2, api.nvim_win_get_cursor(0)[1])
     end)
   end)
+
+  it('backspace after replacing multibyte chars', function()
+    local screen = Screen.new(30, 3)
+    screen:attach()
+    api.nvim_buf_set_lines(0, 0, -1, true, { 'test ȧ̟̜̝̅̚m̆̉̐̐̇̈ å' })
+    feed('^Rabcdefghi')
+    screen:expect([[
+      abcdefghi^                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcdefgh^å                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcdefg^ å                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcdef^m̆̉̐̐̇̈ å                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcde^ȧ̟̜̝̅̚m̆̉̐̐̇̈ å                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcd^ ȧ̟̜̝̅̚m̆̉̐̐̇̈ å                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<esc>')
+
+    api.nvim_buf_set_lines(0, 0, -1, true, { 'wow 🧑‍🌾🏳️‍⚧️x' })
+    feed('^Rabcd')
+
+    screen:expect([[
+      abcd^🧑‍🌾🏳️‍⚧️x                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('e')
+    screen:expect([[
+      abcde^🏳️‍⚧️x                      |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('f')
+    screen:expect([[
+      abcdef^x                       |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcde^🏳️‍⚧️x                      |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abcd^🧑‍🌾🏳️‍⚧️x                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+
+    feed('<bs>')
+    screen:expect([[
+      abc^ 🧑‍🌾🏳️‍⚧️x                     |
+      {1:~                             }|
+      {5:-- REPLACE --}                 |
+    ]])
+  end)
 end)
diff --git a/test/unit/mbyte_spec.lua b/test/unit/mbyte_spec.lua
index 787a8862ae..62390c8794 100644
--- a/test/unit/mbyte_spec.lua
+++ b/test/unit/mbyte_spec.lua
@@ -4,7 +4,6 @@ local itp = t.gen_itp(it)
 local ffi = t.ffi
 local eq = t.eq
 local to_cstr = t.to_cstr
-local ok = t.ok
 
 local lib = t.cimport(
   './src/nvim/mbyte.h',
@@ -302,7 +301,10 @@ describe('mbyte', function()
       local mb_glyphs = {}
       while pos < len do
         local clen = lib.utfc_ptr2len(cstr + pos)
-        ok(clen > 0) -- otherwise we get stuck
+        if clen == 0 then
+          eq(0, string.byte(str, pos + 1)) -- only NUL bytes can has length zery
+          clen = 1 -- but skip it, otherwise we get stuck
+        end
         if clen > 1 then
           table.insert(mb_glyphs, string.sub(str, pos + 1, pos + clen))
         end
@@ -325,13 +327,18 @@ describe('mbyte', function()
     -- stylua doesn't like ZWJ chars..
     -- stylua: ignore start
     check('hej och hå 🧑‍🌾!', { 'å', '🧑‍🌾' })
-    -- emoji only (various kinds of combinations, use g8 to see them)
+
+    -- emoji (various kinds of combinations, use g8 to see them)
     check("🏳️‍⚧️🧑‍🌾❤️😂🏴‍☠️", {"🏳️‍⚧️", "🧑‍🌾", "❤️", "😂", "🏴‍☠️"})
     check('🏳️‍⚧️xy🧑‍🌾\r❤️😂å🏴‍☠️€', { '🏳️‍⚧️', '🧑‍🌾', '❤️', '😂', 'å', '🏴‍☠️', '€' })
+    check('🏳️‍⚧️\000🧑‍🌾\000❤️\000😂\000å\000🏴‍☠️\000€', { '🏳️‍⚧️', '🧑‍🌾', '❤️', '😂', 'å', '🏴‍☠️', '€' })
+    check('\195🏳️‍⚧️\198🧑‍🌾\165❤️\168\195😂\255🏴‍☠️\129€\165', { '🏳️‍⚧️', '🧑‍🌾', '❤️', '😂', '🏴‍☠️', '€' })
 
     check('🇦🅱️ 🇦🇽 🇦🇨🇦 🇲🇽🇹🇱',{'🇦', '🅱️', '🇦🇽', '🇦🇨', '🇦', '🇲🇽', '🇹🇱'})
     check('🏴󠁧󠁢󠁳󠁣󠁴󠁿🏴󠁧󠁢󠁷󠁬󠁳󠁿', {'🏴󠁧󠁢󠁳󠁣󠁴󠁿', '🏴󠁧󠁢󠁷󠁬󠁳󠁿'})
 
+    check('å\165ü\195aëq\168β\000\169本\255', {'å', 'ü', 'ë', 'β', '本'})
+
     lib.p_arshape = true -- default
     check('سلام', { 'س', 'لا', 'م' })
     lib.p_arshape = false
-- 
cgit