Fix macOS build: use CURSES_INCLUDE_DIRS and enable wide-char decls

Two latent defects broke the kte target on macOS once TerminalInputHandler
switched from getch() to get_wch().

include_directories() referenced CURSES_INCLUDE_DIR, a name FindCurses never
sets (it exports CURSES_INCLUDE_DIRS). The variable expanded to nothing, so
no curses include path was added and the compiler silently fell back to the
macOS SDK's narrow-char header on the default include path.

Both the SDK and Homebrew headers declare get_wch and friends only when
NCURSES_WIDECHAR is set, defaulting it to 0 absent a feature-test macro, so
the build failed with "use of undeclared identifier 'get_wch'". We always
request and link the wide library, so define NCURSES_WIDECHAR=1; redefining
it identically is benign on platforms whose headers already set it.

kte and kge now compile and link against libncursesw, and the app bundle
vendors it. Full test suite passes.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Buffer.cc

@@ -235,7 +235,8 @@ Buffer::Buffer(const Buffer &other)
 	edit_mode_          = other.edit_mode_;
 	edit_mode_detected_ = other.edit_mode_detected_;
 	version_            = other.version_;
-	syntax_enabled_     = other.syntax_enabled_;
+	syntax_enabled_        = other.syntax_enabled_;
+	syntax_user_override_  = other.syntax_user_override_;
 	filetype_           = other.filetype_;
 	// Fresh undo system for the copy
 	undo_tree_ = std::make_unique<UndoTree>();
@@ -286,7 +287,8 @@ Buffer::operator=(const Buffer &other)
 	edit_mode_          = other.edit_mode_;
 	edit_mode_detected_ = other.edit_mode_detected_;
 	version_            = other.version_;
-	syntax_enabled_     = other.syntax_enabled_;
+	syntax_enabled_        = other.syntax_enabled_;
+	syntax_user_override_  = other.syntax_user_override_;
 	filetype_           = other.filetype_;
 	// Recreate undo system for this instance
 	undo_tree_ = std::make_unique<UndoTree>();
@@ -327,6 +329,9 @@ Buffer::Buffer(Buffer &&other) noexcept
 	  mark_set_(other.mark_set_),
 	  mark_curx_(other.mark_curx_),
 	  mark_cury_(other.mark_cury_),
+	  visual_line_active_(other.visual_line_active_),
+	  visual_line_anchor_y_(other.visual_line_anchor_y_),
+	  visual_line_active_y_(other.visual_line_active_y_),
 	  undo_tree_(std::move(other.undo_tree_)),
 	  undo_sys_(std::move(other.undo_sys_))
 {
@@ -334,11 +339,18 @@ Buffer::Buffer(Buffer &&other) noexcept
 	edit_mode_          = other.edit_mode_;
 	edit_mode_detected_ = other.edit_mode_detected_;
 	version_            = other.version_;
-	syntax_enabled_     = other.syntax_enabled_;
+	syntax_enabled_        = other.syntax_enabled_;
+	syntax_user_override_  = other.syntax_user_override_;
 	filetype_           = std::move(other.filetype_);
 	highlighter_        = std::move(other.highlighter_);
 	content_            = std::move(other.content_);
 	rows_cache_dirty_   = other.rows_cache_dirty_;
+	on_disk_identity_   = other.on_disk_identity_;
+	// Non-owning: the recorder object itself is owned by SwapManager and outlives
+	// this move. The caller (Editor) is responsible for calling SwapManager::Rehome()
+	// so the journal's Buffer* key follows this object to its new address.
+	swap_rec_       = other.swap_rec_;
+	other.swap_rec_ = nullptr;
 	// Update UndoSystem's buffer reference to point to this object
 	if (undo_sys_) {
 		undo_sys_->UpdateBufferReference(*this);
@@ -353,32 +365,42 @@ Buffer::operator=(Buffer &&other) noexcept
 	if (this == &other)
 		return *this;
 
-	curx_           = other.curx_;
+	curx_                 = other.curx_;
-	cury_           = other.cury_;
+	cury_                 = other.cury_;
-	rx_             = other.rx_;
+	rx_                   = other.rx_;
-	nrows_          = other.nrows_;
+	nrows_                = other.nrows_;
-	rowoffs_        = other.rowoffs_;
+	rowoffs_              = other.rowoffs_;
-	coloffs_        = other.coloffs_;
+	coloffs_              = other.coloffs_;
-	rows_           = std::move(other.rows_);
+	rows_                 = std::move(other.rows_);
-	filename_       = std::move(other.filename_);
+	filename_             = std::move(other.filename_);
-	is_file_backed_ = other.is_file_backed_;
+	is_file_backed_       = other.is_file_backed_;
-	dirty_          = other.dirty_;
+	dirty_                = other.dirty_;
-	read_only_      = other.read_only_;
+	read_only_            = other.read_only_;
-	mark_set_       = other.mark_set_;
+	mark_set_             = other.mark_set_;
-	mark_curx_      = other.mark_curx_;
+	mark_curx_            = other.mark_curx_;
-	mark_cury_      = other.mark_cury_;
+	mark_cury_            = other.mark_cury_;
-	undo_tree_      = std::move(other.undo_tree_);
+	visual_line_active_   = other.visual_line_active_;
-	undo_sys_       = std::move(other.undo_sys_);
+	visual_line_anchor_y_ = other.visual_line_anchor_y_;
+	visual_line_active_y_ = other.visual_line_active_y_;
+	undo_tree_            = std::move(other.undo_tree_);
+	undo_sys_             = std::move(other.undo_sys_);
 
 	// Move edit mode + syntax/highlighting state
 	edit_mode_          = other.edit_mode_;
 	edit_mode_detected_ = other.edit_mode_detected_;
 	version_            = other.version_;
-	syntax_enabled_     = other.syntax_enabled_;
+	syntax_enabled_        = other.syntax_enabled_;
+	syntax_user_override_  = other.syntax_user_override_;
 	filetype_           = std::move(other.filetype_);
 	highlighter_        = std::move(other.highlighter_);
 	content_            = std::move(other.content_);
 	rows_cache_dirty_   = other.rows_cache_dirty_;
+	on_disk_identity_   = other.on_disk_identity_;
+	// Non-owning: the recorder object itself is owned by SwapManager and outlives
+	// this move. The caller (Editor) is responsible for calling SwapManager::Rehome()
+	// so the journal's Buffer* key follows this object to its new address.
+	swap_rec_       = other.swap_rec_;
+	other.swap_rec_ = nullptr;
 	// Update UndoSystem's buffer reference to point to this object
 	if (undo_sys_) {
 		undo_sys_->UpdateBufferReference(*this);

Buffer.h

@@ -548,6 +548,25 @@ public:
 	}
 
 
+	// Marks that the user explicitly set syntax state via a command (:syntax
+	// on/off, :set filetype=...), as opposed to it being auto-applied from
+	// GUIConfig. Frontends that re-apply the config-driven default every
+	// frame (e.g. ImGuiFrontend::apply_syntax_to_buffer) must check this
+	// first, mirroring EditModeDetected()'s "don't stomp a manual toggle"
+	// pattern - otherwise a manual :syntax off is silently undone on the very
+	// next frame.
+	void SetSyntaxUserOverride(bool on)
+	{
+		syntax_user_override_ = on;
+	}
+
+
+	[[nodiscard]] bool SyntaxUserOverride() const
+	{
+		return syntax_user_override_;
+	}
+
+
 	void SetFiletype(const std::string &ft)
 	{
 		filetype_ = ft;
@@ -673,6 +692,7 @@ private:
 	// Syntax/highlighting state
 	std::uint64_t version_ = 0; // increment on edits
 	bool syntax_enabled_   = true;
+	bool syntax_user_override_ = false; // true once user explicitly set syntax state via a command
 	std::string filetype_;
 	std::unique_ptr<kte::HighlighterEngine> highlighter_;
 	// Non-owning pointer to swap recorder managed by Editor/SwapManager

CMakeLists.txt

@@ -4,7 +4,7 @@ project(kte)
 include(GNUInstallDirs)
 
 set(CMAKE_CXX_STANDARD 20)
-set(KTE_VERSION "1.11.2")
+set(KTE_VERSION "1.12.0")
 
 # Default to terminal-only build to avoid SDL/OpenGL dependency by default.
 # Enable with -DBUILD_GUI=ON when SDL2/OpenGL/Freetype are available.
@@ -71,8 +71,34 @@ endif ()
 # NCurses for terminal mode
 set(CURSES_NEED_NCURSES TRUE)
 set(CURSES_NEED_WIDE TRUE)
+
+# macOS ships only a narrow-char ncurses in the SDK, and Homebrew's wide-char
+# build is keg-only, so FindCurses can't see it without a hint.
+if (APPLE)
+    find_program(BREW_EXECUTABLE brew)
+    if (BREW_EXECUTABLE)
+        execute_process(COMMAND "${BREW_EXECUTABLE}" --prefix ncurses
+                OUTPUT_VARIABLE HOMEBREW_NCURSES_PREFIX
+                OUTPUT_STRIP_TRAILING_WHITESPACE
+                ERROR_QUIET)
+        if (HOMEBREW_NCURSES_PREFIX AND IS_DIRECTORY "${HOMEBREW_NCURSES_PREFIX}")
+            message(STATUS "Using Homebrew ncursesw: ${HOMEBREW_NCURSES_PREFIX}")
+            list(PREPEND CMAKE_PREFIX_PATH "${HOMEBREW_NCURSES_PREFIX}")
+        endif ()
+    endif ()
+endif ()
+
 find_package(Curses REQUIRED)
-include_directories(${CURSES_INCLUDE_DIR})
+
+# FindCurses exports CURSES_INCLUDE_DIRS; the singular CURSES_INCLUDE_DIR was
+# never set, so this silently expanded to nothing and the compiler fell back
+# to whatever curses header happened to be on the default include path.
+include_directories(${CURSES_INCLUDE_DIRS})
+
+# The wide-char entry points we rely on (get_wch, etc.) are declared only when
+# NCURSES_WIDECHAR is set; the header defaults it to 0 unless a feature-test
+# macro says otherwise. We always link the wide library, so ask for them.
+add_compile_definitions(NCURSES_WIDECHAR=1)
 
 # On Alpine Linux, CMake's FindCurses looks in wrong paths
 # Manually find the correct ncurses library
@@ -344,6 +370,8 @@ if (BUILD_TESTS)
             tests/test_migration_coverage.cc
             tests/test_smart_newline.cc
             tests/test_reflow_undo.cc
+            tests/test_syntax_highlighting.cc
+            tests/test_paste_split.cc
 
             # minimal engine sources required by Buffer
             PieceTable.cc

Command.cc

@@ -159,6 +159,27 @@ ensure_at_least_one_line(Buffer &buf)
 }
 
 
+// RAII helper: brackets a multi-edit command so its individual undo nodes
+// undo/redo as a single atomic step, per the project's undo-grouping convention.
+struct UndoGroupGuard {
+	UndoSystem *u;
+
+
+	explicit UndoGroupGuard(UndoSystem *u_) : u(u_)
+	{
+		if (u)
+			u->BeginGroup();
+	}
+
+
+	~UndoGroupGuard()
+	{
+		if (u)
+			u->EndGroup();
+	}
+};
+
+
 // Determine if a command mutates the buffer contents (text edits)
 static bool
 is_mutating_command(CommandId id)
@@ -264,8 +285,12 @@ extract_region_text(const Buffer &buf, std::size_t sx, std::size_t sy, std::size
 
 
 // Helper: delete region and leave cursor at start (sx,sy). Adjust lines appropriately.
+// If `u` is non-null, each underlying mutation is recorded as its own undo node;
+// callers that want the whole region-delete to undo/redo as one step should
+// wrap the call in an UndoGroupGuard.
 static void
-delete_region(Buffer &buf, std::size_t sx, std::size_t sy, std::size_t ex, std::size_t ey)
+delete_region(Buffer &buf, std::size_t sx, std::size_t sy, std::size_t ex, std::size_t ey,
+              UndoSystem *u = nullptr)
 {
 	std::size_t nrows = buf.Nrows();
 	if (nrows == 0)
@@ -282,7 +307,14 @@ delete_region(Buffer &buf, std::size_t sx, std::size_t sy, std::size_t ex, std::
 		std::size_t xe   = std::min(ex, line.size());
 		if (xe < xs)
 			std::swap(xs, xe);
+		std::string deleted = std::string(line.substr(xs, xe - xs));
 		buf.delete_text(static_cast<int>(sy), static_cast<int>(xs), xe - xs);
+		if (u && !deleted.empty()) {
+			buf.SetCursor(xs, sy);
+			u->Begin(UndoType::Delete);
+			u->Append(std::string_view(deleted));
+			u->commit();
+		}
 	} else {
 		// Multi-line: delete from (sx,sy) to (ex,ey)
 		// Strategy:
@@ -303,11 +335,25 @@ delete_region(Buffer &buf, std::size_t sx, std::size_t sy, std::size_t ex, std::
 
 		// Delete tail of first line (from xs to end)
 		if (xs < first_line_len) {
+			std::string tail = std::string(rows[sy].substr(xs));
 			buf.delete_text(static_cast<int>(sy), static_cast<int>(xs), first_line_len - xs);
+			if (u && !tail.empty()) {
+				buf.SetCursor(xs, sy);
+				u->Begin(UndoType::Delete);
+				u->Append(std::string_view(tail));
+				u->commit();
+			}
 		}
 
 		// Delete lines from ey down to sy+1 (reverse order to preserve indices)
 		for (std::size_t i = ey; i > sy; --i) {
+			if (u) {
+				std::string row_text = static_cast<std::string>(buf.Rows()[i]);
+				buf.SetCursor(0, i);
+				u->Begin(UndoType::DeleteRow);
+				u->Append(std::string_view(row_text));
+				u->commit();
+			}
 			buf.delete_row(static_cast<int>(i));
 		}
 
@@ -317,6 +363,12 @@ delete_region(Buffer &buf, std::size_t sx, std::size_t sy, std::size_t ex, std::
 			const auto &rows_after = buf.Rows();
 			std::size_t line_len   = rows_after[sy].size();
 			buf.insert_text(static_cast<int>(sy), static_cast<int>(line_len), suffix);
+			if (u) {
+				buf.SetCursor(line_len, sy);
+				u->Begin(UndoType::Insert);
+				u->Append(std::string_view(suffix));
+				u->commit();
+			}
 		}
 	}
 	buf.SetCursor(sx, sy);
@@ -908,6 +960,10 @@ cmd_unknown_esc_command(CommandContext &ctx)
 static void
 apply_filetype(Buffer &buf, const std::string &ft)
 {
+	// Only reachable from explicit user commands (:syntax on, :set
+	// filetype=...) - mark so frontends that re-apply config-driven syntax
+	// defaults every frame (e.g. ImGuiFrontend) don't stomp this choice.
+	buf.SetSyntaxUserOverride(true);
 	buf.EnsureHighlighter();
 	auto *eng = buf.Highlighter();
 	if (!eng)
@@ -977,6 +1033,7 @@ cmd_syntax(CommandContext &ctx)
 	};
 	trim(arg);
 	if (arg == "on") {
+		b->SetSyntaxUserOverride(true);
 		b->SetSyntaxEnabled(true);
 		// If no highlighter but filetype is cpp by extension, set it
 		if (!b->Highlighter() || !b->Highlighter()->HasHighlighter()) {
@@ -984,6 +1041,7 @@ cmd_syntax(CommandContext &ctx)
 		}
 		ctx.editor.SetStatus("syntax: on");
 	} else if (arg == "off") {
+		b->SetSyntaxUserOverride(true);
 		b->SetSyntaxEnabled(false);
 		ctx.editor.SetStatus("syntax: off");
 	} else if (arg == "reload") {
@@ -2509,12 +2567,10 @@ cmd_newline(CommandContext &ctx)
 						}
 						pos = p + with.size();
 					} else {
-						// When replacing with empty, continue after the deletion point to avoid re-matching
+						// Replacing with empty leaves nothing inserted at the deletion
+						// point, so resume scanning from `p` itself (not p+1) to catch
+						// adjacent/overlapping matches, e.g. "aaaa" -> "" replacing "aa".
 						pos = p;
-						if (pos < static_cast<std::size_t>(buf->Rows()[y].size()))
-							++pos;
-						else
-							break;
 					}
 					++total;
 				}
@@ -2923,14 +2979,28 @@ cmd_newline(CommandContext &ctx)
 				return true;
 			}
 			std::size_t changed = 0;
+			UndoSystem *ru      = buf->Undo();
+			UndoGroupGuard rguard(ru);
 			// Iterate by index to allow modifications via PieceTable helpers
 			for (std::size_t y = 0; y < buf->Rows().size(); ++y) {
 				std::string before = static_cast<std::string>(buf->Rows()[y]);
 				std::string after  = std::regex_replace(before, rx, repl);
 				if (after != before) {
 					// Replace entire line y with 'after' using PieceTable ops
+					if (ru) {
+						buf->SetCursor(0, y);
+						ru->Begin(UndoType::DeleteRow);
+						ru->Append(std::string_view(before));
+						ru->commit();
+					}
 					buf->delete_row(static_cast<int>(y));
 					buf->insert_row(static_cast<int>(y), std::string_view(after));
+					if (ru) {
+						buf->SetCursor(0, y);
+						ru->Begin(UndoType::InsertRow);
+						ru->Append(std::string_view(after));
+						ru->commit();
+					}
 					++changed;
 				}
 			}
@@ -3003,18 +3073,28 @@ cmd_newline(CommandContext &ctx)
 		ensure_cursor_visible(ctx.editor, *buf);
 		return true;
 	}
+	UndoSystem *u = buf->Undo();
+	if (u && repeat > 1)
+		u->BeginGroup();
 	for (int i = 0; i < repeat; ++i) {
+		// Sync the buffer's cursor to the split point before Begin(), which
+		// records its node's row/col from the buffer's *current* cursor. Undo
+		// must reverse this exact split, not wherever the cursor ends up after
+		// the whole loop finishes.
+		buf->SetCursor(x, y);
+		if (u) {
+			u->Begin(UndoType::Newline);
+			u->commit();
+		}
 		buf->split_line(static_cast<int>(y), static_cast<int>(x));
 		// Move to start of next line
 		y += 1;
 		x = 0;
 	}
+	if (u && repeat > 1)
+		u->EndGroup();
 	buf->SetCursor(x, y);
 	buf->SetDirty(true);
-	if (auto *u = buf->Undo()) {
-		u->Begin(UndoType::Newline);
-		u->commit();
-	}
 	ensure_cursor_visible(ctx.editor, *buf);
 	return true;
 }
@@ -3198,7 +3278,9 @@ cmd_backspace(CommandContext &ctx)
 			x = prev_len;
 			buf->SetCursor(x, y);
 			if (u) {
-				u->Begin(UndoType::Newline);
+				// Forward action here is a join, not a split: JoinLines has the
+				// correct (inverted) apply() semantics, unlike Newline.
+				u->Begin(UndoType::JoinLines);
 				u->commit();
 			}
 		} else {
@@ -3277,7 +3359,9 @@ cmd_delete_char(CommandContext &ctx)
 		} else if (y + 1 < rows_view.size()) {
 			buf->join_lines(static_cast<int>(y));
 			if (u) {
-				u->Begin(UndoType::Newline);
+				// Forward action here is a join, not a split: JoinLines has the
+				// correct (inverted) apply() semantics, unlike Newline.
+				u->Begin(UndoType::JoinLines);
 				u->commit();
 			}
 		} else {
@@ -3350,18 +3434,32 @@ cmd_kill_to_eol(CommandContext &ctx)
 	std::size_t x = buf->Curx();
 	int repeat    = ctx.count > 0 ? ctx.count : 1;
 	std::string killed_total;
+	UndoSystem *u = buf->Undo();
+	UndoGroupGuard guard(u);
 	for (int i = 0; i < repeat; ++i) {
 		const auto &rows_view = buf->Rows();
 		if (y >= rows_view.size())
 			break;
 		if (x < rows_view[y].size()) {
 			// delete from cursor to end of line
-			killed_total    += rows_view[y].substr(x);
+			std::string seg = static_cast<std::string>(rows_view[y].substr(x));
+			killed_total    += seg;
 			std::size_t len = rows_view[y].size() - x;
 			buf->delete_text(static_cast<int>(y), static_cast<int>(x), len);
+			if (u) {
+				buf->SetCursor(x, y);
+				u->Begin(UndoType::Delete);
+				u->Append(std::string_view(seg));
+				u->commit();
+			}
 		} else if (y + 1 < rows_view.size()) {
 			// at EOL: delete the newline (join with next line)
 			killed_total += "\n";
+			if (u) {
+				buf->SetCursor(x, y);
+				u->Begin(UndoType::JoinLines);
+				u->commit();
+			}
 			buf->join_lines(static_cast<int>(y));
 		} else {
 			// nothing to delete
@@ -3395,20 +3493,37 @@ cmd_kill_line(CommandContext &ctx)
 	(void) x; // cursor x will be reset to 0
 	int repeat = ctx.count > 0 ? ctx.count : 1;
 	std::string killed_total;
+	UndoSystem *u = buf->Undo();
+	UndoGroupGuard guard(u);
 	for (int i = 0; i < repeat; ++i) {
 		const auto &rows_view = buf->Rows();
 		if (rows_view.empty())
 			break;
 		if (rows_view.size() == 1) {
 			// last remaining line: clear its contents
-			killed_total += static_cast<std::string>(rows_view[0]);
+			std::string content = static_cast<std::string>(rows_view[0]);
-			if (!rows_view[0].empty())
+			killed_total += content;
-				buf->delete_text(0, 0, rows_view[0].size());
+			if (!content.empty()) {
+				buf->delete_text(0, 0, content.size());
+				if (u) {
+					buf->SetCursor(0, 0);
+					u->Begin(UndoType::Delete);
+					u->Append(std::string_view(content));
+					u->commit();
+				}
+			}
 			y = 0;
 		} else if (y < rows_view.size()) {
 			// erase current line; keep y pointing at the next line
-			killed_total += static_cast<std::string>(rows_view[y]);
+			std::string content = static_cast<std::string>(rows_view[y]);
+			killed_total += content;
 			killed_total += "\n";
+			if (u) {
+				buf->SetCursor(0, y);
+				u->Begin(UndoType::DeleteRow);
+				u->Append(std::string_view(content));
+				u->commit();
+			}
 			buf->delete_row(static_cast<int>(y));
 			const auto &rows_after = buf->Rows();
 			if (y >= rows_after.size()) {
@@ -3622,7 +3737,10 @@ cmd_kill_region(CommandContext &ctx)
 		return false;
 	}
 	std::string text = extract_region_text(*buf, sx, sy, ex, ey);
-	delete_region(*buf, sx, sy, ex, ey);
+	{
+		UndoGroupGuard guard(buf->Undo());
+		delete_region(*buf, sx, sy, ex, ey, buf->Undo());
+	}
 	ensure_cursor_visible(ctx.editor, *buf);
 	if (!text.empty()) {
 		if (ctx.editor.KillChain())
@@ -4234,6 +4352,7 @@ cmd_delete_word_prev(CommandContext &ctx)
 	std::size_t x = buf->Curx();
 	int repeat    = ctx.count > 0 ? ctx.count : 1;
 	std::string killed_total;
+	UndoGroupGuard guard(buf->Undo());
 	for (int i = 0; i < repeat; ++i) {
 		if (y >= rows.size()) {
 			y = rows.empty() ? 0 : rows.size() - 1;
@@ -4275,7 +4394,7 @@ cmd_delete_word_prev(CommandContext &ctx)
 		}
 		// Now delete from (x, y) to (start_x, start_y) using PieceTable
 		std::string deleted = extract_region_text(*buf, x, y, start_x, start_y);
-		delete_region(*buf, x, y, start_x, start_y);
+		delete_region(*buf, x, y, start_x, start_y, buf->Undo());
 		// Prepend to killed_total (since we're deleting backwards)
 		killed_total = deleted + killed_total;
 	}
@@ -4307,6 +4426,7 @@ cmd_delete_word_next(CommandContext &ctx)
 	std::size_t x    = buf->Curx();
 	int repeat       = ctx.count > 0 ? ctx.count : 1;
 	std::string killed_total;
+	UndoGroupGuard guard(buf->Undo());
 	for (int i = 0; i < repeat; ++i) {
 		if (y >= rows.size())
 			break;
@@ -4346,7 +4466,7 @@ cmd_delete_word_next(CommandContext &ctx)
 		}
 		// Now delete from (start_x, start_y) to (x, y) using PieceTable
 		std::string deleted = extract_region_text(*buf, start_x, start_y, x, y);
-		delete_region(*buf, start_x, start_y, x, y);
+		delete_region(*buf, start_x, start_y, x, y, buf->Undo());
 		y            = start_y;
 		x            = start_x;
 		killed_total += deleted;
@@ -4380,8 +4500,16 @@ cmd_indent_region(CommandContext &ctx)
 		ctx.editor.SetStatus("No region to indent");
 		return false;
 	}
+	UndoSystem *u = buf->Undo();
+	UndoGroupGuard guard(u);
 	for (std::size_t y = sy; y <= ey && y < buf->Rows().size(); ++y) {
 		buf->insert_text(static_cast<int>(y), 0, std::string_view("\t"));
+		if (u) {
+			buf->SetCursor(0, y);
+			u->Begin(UndoType::Insert);
+			u->Append('\t');
+			u->commit();
+		}
 	}
 	buf->SetDirty(true);
 	buf->ClearMark();
@@ -4405,6 +4533,8 @@ cmd_unindent_region(CommandContext &ctx)
 		ctx.editor.SetStatus("No region to unindent");
 		return false;
 	}
+	UndoSystem *u = buf->Undo();
+	UndoGroupGuard guard(u);
 	for (std::size_t y = sy; y <= ey && y < buf->Rows().size(); ++y) {
 		const auto &rows_view = buf->Rows();
 		if (y >= rows_view.size())
@@ -4413,13 +4543,26 @@ cmd_unindent_region(CommandContext &ctx)
 		if (!line.empty()) {
 			if (line[0] == '\t') {
 				buf->delete_text(static_cast<int>(y), 0, 1);
+				if (u) {
+					buf->SetCursor(0, y);
+					u->Begin(UndoType::Delete);
+					u->Append('\t');
+					u->commit();
+				}
 			} else if (line[0] == ' ') {
 				std::size_t spaces = 0;
 				while (spaces < line.size() && spaces < 8 && line[spaces] == ' ') {
 					++spaces;
 				}
-				if (spaces > 0)
+				if (spaces > 0) {
 					buf->delete_text(static_cast<int>(y), 0, spaces);
+					if (u) {
+						buf->SetCursor(0, y);
+						u->Begin(UndoType::Delete);
+						u->Append(line.substr(0, spaces));
+						u->commit();
+					}
+				}
 			}
 		}
 	}
@@ -4436,25 +4579,8 @@ cmd_reflow_paragraph(CommandContext &ctx)
 	Buffer *buf = ctx.editor.CurrentBuffer();
 	if (!buf)
 		return false;
-	struct GroupGuard {
-		UndoSystem *u;
-
-
-		explicit GroupGuard(UndoSystem *u_) : u(u_)
-		{
-			if (u)
-				u->BeginGroup();
-		}
-
-
-		~GroupGuard()
-		{
-			if (u)
-				u->EndGroup();
-		}
-	};
 	// Reflow performs a multi-edit transformation; make it a single standalone undo/redo step.
-	GroupGuard guard(buf->Undo());
+	UndoGroupGuard guard(buf->Undo());
 	if (auto *u = buf->Undo())
 		u->commit();
 	ensure_at_least_one_line(*buf);

Editor.cc

@@ -165,9 +165,23 @@ std::size_t
 Editor::AddBuffer(const Buffer &buf)
 {
 	auto &bufs = Buffers();
+	// push_back may reallocate the vector's storage, moving every existing
+	// Buffer to a new address. Drain any in-flight swap records first so the
+	// writer thread never dereferences an address that's about to move, then
+	// rehome each attached buffer that actually moved.
+	if (swap_ && !bufs.empty())
+		swap_->Flush();
+	std::vector<Buffer *> old_addrs;
+	old_addrs.reserve(bufs.size());
+	for (auto &b: bufs)
+		old_addrs.push_back(&b);
 	bufs.push_back(buf);
-	// Attach swap recorder
 	if (swap_) {
+		for (std::size_t i = 0; i < old_addrs.size(); ++i) {
+			Buffer *new_addr = &bufs[i];
+			if (new_addr != old_addrs[i])
+				bufs[i].SetSwapRecorder(swap_->Rehome(old_addrs[i], new_addr));
+		}
 		swap_->Attach(&bufs.back());
 		bufs.back().SetSwapRecorder(swap_->RecorderFor(&bufs.back()));
 	}
@@ -182,8 +196,19 @@ std::size_t
 Editor::AddBuffer(Buffer &&buf)
 {
 	auto &bufs = Buffers();
+	if (swap_ && !bufs.empty())
+		swap_->Flush();
+	std::vector<Buffer *> old_addrs;
+	old_addrs.reserve(bufs.size());
+	for (auto &b: bufs)
+		old_addrs.push_back(&b);
 	bufs.push_back(std::move(buf));
 	if (swap_) {
+		for (std::size_t i = 0; i < old_addrs.size(); ++i) {
+			Buffer *new_addr = &bufs[i];
+			if (new_addr != old_addrs[i])
+				bufs[i].SetSwapRecorder(swap_->Rehome(old_addrs[i], new_addr));
+		}
 		swap_->Attach(&bufs.back());
 		bufs.back().SetSwapRecorder(swap_->RecorderFor(&bufs.back()));
 	}
@@ -495,8 +520,22 @@ Editor::CloseBuffer(std::size_t index)
 		// This prevents stale swap files from accumulating (e.g., when used as git editor).
 		swap_->Detach(&bufs[index], true);
 		bufs[index].SetSwapRecorder(nullptr);
+		// Drain in-flight records before the erase-shift below moves other
+		// buffers to new addresses (vector::erase never reallocates, but it does
+		// move-assign each trailing buffer into the previous slot).
+		swap_->Flush();
 	}
 	bufs.erase(bufs.begin() + static_cast<std::ptrdiff_t>(index));
+	if (swap_) {
+		// erase() shifts every buffer after `index` down by one slot in-place
+		// (no reallocation), so the buffer now at slot i used to live at slot
+		// i+1 (same underlying storage, since data() doesn't move on erase).
+		for (std::size_t i = index; i < bufs.size(); ++i) {
+			Buffer *new_addr = &bufs[i];
+			Buffer *old_addr = new_addr + 1;
+			bufs[i].SetSwapRecorder(swap_->Rehome(old_addr, new_addr));
+		}
+	}
 	if (bufs.empty()) {
 		curbuf_ = 0;
 	} else if (curbuf_ >= bufs.size()) {

ImGuiFrontend.cc

@@ -44,6 +44,12 @@ apply_syntax_to_buffer(Buffer *b, const GUIConfig &cfg)
 	if (!b->EditModeDetected() && !b->Filename().empty())
 		b->SetEditMode(DetectEditMode(b->Filename()));
 
+	// If the user explicitly set syntax state via a command (:syntax on/off,
+	// :set filetype=...), leave it alone - otherwise this runs every frame
+	// and silently undoes a manual ":syntax off" on the very next frame.
+	if (b->SyntaxUserOverride())
+		return;
+
 	// Writing mode disables syntax; otherwise follow the global config.
 	if (cfg.syntax && b->GetEditMode() != EditMode::Writing) {
 		b->SetSyntaxEnabled(true);
@@ -469,7 +475,19 @@ GUIFrontend::Step(Editor &ed, bool &running)
 
 		// Route input events to the correct window's input handler
 		if (target) {
-			target->input.ProcessSDLEvent(e);
+			Editor &tgt_ed = (target_idx == 0) ? ed : target->editor;
+			if (tgt_ed.FilePickerVisible()) {
+				// Modal: don't let keystrokes fall through as edit commands
+				// to the buffer underneath. Escape closes the picker;
+				// everything else (navigation, filtering, double-click) is
+				// handled by the picker's own ImGui widgets via
+				// ImGui_ImplSDL2_ProcessEvent above.
+				if (e.type == SDL_KEYDOWN && e.key.keysym.sym == SDLK_ESCAPE) {
+					tgt_ed.SetFilePickerVisible(false);
+				}
+			} else {
+				target->input.ProcessSDLEvent(e);
+			}
 		}
 	}
 

ImGuiInputHandler.cc

@@ -8,6 +8,20 @@
 #include "ImGuiInputHandler.h"
 #include "KKeymap.h"
 #include "Editor.h"
+#include "PasteSplit.h"
+
+// Verbose k-prefix suffix logging for debugging macOS/SDL key translation
+// issues. Default to off; enable by defining IMGUI_IH_DEBUG=1 at compile
+// time. Mirrors QtInputHandler.cc's QT_IH_DEBUG gate.
+#ifndef IMGUI_IH_DEBUG
+#define IMGUI_IH_DEBUG 0
+#endif
+
+#if IMGUI_IH_DEBUG
+#define IH_LOGF(...) do { std::fprintf(stderr, __VA_ARGS__); std::fflush(stderr); } while (0)
+#else
+#define IH_LOGF(...) ((void) 0)
+#endif
 
 
 static bool
@@ -182,18 +196,19 @@ map_key(const SDL_Keycode key,
 			k_ctrl_pending             = false;
 			CommandId id;
 			bool mapped = KLookupKCommand(ascii_key, pass_ctrl, id);
+#if IMGUI_IH_DEBUG
 			// Diagnostics for u/U
 			if (lower == 'u') {
 				char disp = (ascii_key >= 0x20 && ascii_key <= 0x7e)
 					            ? static_cast<char>(ascii_key)
 					            : '?';
-				std::fprintf(stderr,
+				IH_LOGF(
-				             "[kge] k-prefix suffix: sym=%d mods=0x%x ascii=%d '%c' ctrl2=%d pass_ctrl=%d mapped=%d id=%d\n",
+					"[kge] k-prefix suffix: sym=%d mods=0x%x ascii=%d '%c' ctrl2=%d pass_ctrl=%d mapped=%d id=%d\n",
-				             static_cast<int>(key), static_cast<unsigned int>(mod), ascii_key, disp,
+					static_cast<int>(key), static_cast<unsigned int>(mod), ascii_key, disp,
-				             ctrl2 ? 1 : 0, pass_ctrl ? 1 : 0, mapped ? 1 : 0,
+					ctrl2 ? 1 : 0, pass_ctrl ? 1 : 0, mapped ? 1 : 0,
-				             mapped ? static_cast<int>(id) : -1);
+					mapped ? static_cast<int>(id) : -1);
-				std::fflush(stderr);
 			}
+#endif
 			if (mapped) {
 				out = {true, id, "", 0};
 				if (ed)
@@ -376,32 +391,12 @@ ImGuiInputHandler::ProcessSDLEvent(const SDL_Event &e)
 			if (clip) {
 				std::string text(clip);
 				SDL_free(clip);
-				// Split on '\n' and enqueue as InsertText/Newline commands
+				// Turn line breaks (\n, \r\n, or bare \r) into Newline
+				// commands and the rest into InsertText; InsertText itself
+				// rejects any embedded '\r'/'\n'.
 				std::lock_guard<std::mutex> lk(mu_);
-				std::size_t start = 0;
+				for (const auto &cmd : SplitPasteIntoCommands(text))
-				while (start <= text.size()) {
+					q_.push(cmd);
-					std::size_t pos = text.find('\n', start);
-					std::string_view segment;
-					bool has_nl = (pos != std::string::npos);
-					if (has_nl) {
-						segment = std::string_view(text).substr(start, pos - start);
-					} else {
-						segment = std::string_view(text).substr(start);
-					}
-					if (!segment.empty()) {
-						MappedInput ins{
-							true, CommandId::InsertText, std::string(segment), 0
-						};
-						q_.push(ins);
-					}
-					if (has_nl) {
-						MappedInput nl{true, CommandId::Newline, std::string(), 0};
-						q_.push(nl);
-						start = pos + 1;
-					} else {
-						break;
-					}
-				}
 				// Suppress the corresponding TEXTINPUT that may follow
 				suppress_text_input_once_ = true;
 				return true; // consumed
@@ -524,15 +519,17 @@ ImGuiInputHandler::ProcessSDLEvent(const SDL_Event &e)
 					bool pass_ctrl  = k_ctrl_pending_;
 					k_ctrl_pending_ = false;
 					bool mapped     = KLookupKCommand(ascii_key, pass_ctrl, id);
+#if IMGUI_IH_DEBUG
 					// Diagnostics: log any k-prefix TEXTINPUT suffix mapping
-					char disp = (ascii_key >= 0x20 && ascii_key <= 0x7e)
+					{
-						            ? static_cast<char>(ascii_key)
+						char disp = (ascii_key >= 0x20 && ascii_key <= 0x7e)
-						            : '?';
+							            ? static_cast<char>(ascii_key)
-					std::fprintf(stderr,
+							            : '?';
-					             "[kge] k-prefix TEXTINPUT suffix: ascii=%d '%c' mapped=%d id=%d\n",
+						IH_LOGF("[kge] k-prefix TEXTINPUT suffix: ascii=%d '%c' mapped=%d id=%d\n",
-					             ascii_key, disp, mapped ? 1 : 0,
+						       ascii_key, disp, mapped ? 1 : 0,
-					             mapped ? static_cast<int>(id) : -1);
+						       mapped ? static_cast<int>(id) : -1);
-					std::fflush(stderr);
+					}
+#endif
 					if (mapped) {
 						mi = {true, id, "", 0};
 						if (ed_)

KKeymap.cc

@@ -1,6 +1,4 @@
-#include <iostream>
 #include <ncurses.h>
-#include <ostream>
 
 #include "KKeymap.h"
 
@@ -40,6 +38,15 @@ KLookupKCommand(const int ascii_key, const bool ctrl, CommandId &out) -> bool
 		out = CommandId::ToggleReadOnly; // C-k ' (toggle read-only)
 		return true;
 	}
+	if (ascii_key == 'E') {
+		// Explicitly rejected and kept distinct from 'e' (OpenFileStart): the
+		// switch below operates on the lowercased key, so it can't otherwise
+		// tell 'E' and 'e' apart. Return false and let the caller show its
+		// normal "unknown k-command" status-line message - curses already
+		// owns the terminal here, so writing to stderr would corrupt the
+		// screen.
+		return false;
+	}
 
 	switch (k_lower) {
 	case 'a':
@@ -66,9 +73,6 @@ KLookupKCommand(const int ascii_key, const bool ctrl, CommandId &out) -> bool
 	case 'e':
 		out = CommandId::OpenFileStart;
 		return true;
-	case 'E':
-		std::cerr << "E is not a valid command" << std::endl;
-		return false;
 	case 'f':
 		out = CommandId::FlushKillRing;
 		return true;

PasteSplit.h

@@ -0,0 +1,48 @@
+/*
+ * PasteSplit.h - split pasted/clipboard text into editor commands
+ *
+ * Pure logic (no SDL/GUI dependencies) so it can be unit tested and shared
+ * across frontends. Line breaks in the pasted text must become Newline
+ * commands; InsertText itself rejects embedded '\r'/'\n' (see Command.cc).
+ */
+#pragma once
+#include <string>
+#include <vector>
+
+#include "InputHandler.h"
+
+
+// Translate a block of pasted text into a sequence of editor commands.
+//
+// Any of "\n", "\r\n", or a bare "\r" is treated as a single line break and
+// emitted as a Newline command; the text between breaks becomes InsertText
+// commands. Empty text segments are skipped, but line breaks are always
+// emitted so blank lines round-trip correctly.
+inline std::vector<MappedInput> SplitPasteIntoCommands(const std::string &text)
+{
+	std::vector<MappedInput> out;
+	std::string segment;
+
+	auto flush_segment = [&]() {
+		if (!segment.empty()) {
+			out.push_back(MappedInput{true, CommandId::InsertText, segment, 0});
+			segment.clear();
+		}
+	};
+
+	for (std::size_t i = 0; i < text.size(); ++i) {
+		const char c = text[i];
+		if (c == '\n' || c == '\r') {
+			flush_segment();
+			out.push_back(MappedInput{true, CommandId::Newline, std::string(), 0});
+			// Collapse a "\r\n" pair into one line break.
+			if (c == '\r' && i + 1 < text.size() && text[i + 1] == '\n')
+				++i;
+		} else {
+			segment.push_back(c);
+		}
+	}
+	flush_segment();
+
+	return out;
+}

Swap.cc

@@ -268,6 +268,43 @@ SwapManager::RecorderFor(Buffer *buf)
 }
 
 
+SwapRecorder *
+SwapManager::Rehome(Buffer *old_addr, Buffer *new_addr)
+{
+	if (!old_addr || !new_addr || old_addr == new_addr)
+		return nullptr;
+	std::lock_guard<std::mutex> lg(mtx_);
+	SwapRecorder *result = nullptr;
+
+	auto jit = journals_.find(old_addr);
+	if (jit != journals_.end()) {
+		JournalCtx ctx = std::move(jit->second);
+		journals_.erase(jit);
+		journals_[new_addr] = std::move(ctx);
+	}
+
+	auto rit = recorders_.find(old_addr);
+	if (rit != recorders_.end()) {
+		recorders_.erase(rit);
+		// BufferRecorder binds a Buffer& at construction, so it can't be
+		// repointed in place; rebuild it against the buffer's new address.
+		auto rec              = std::make_unique<BufferRecorder>(*this, *new_addr);
+		result                = rec.get();
+		recorders_[new_addr] = std::move(rec);
+	}
+
+	// Defensive: any record still queued (not yet drained by the writer thread)
+	// for the old address must follow the buffer to its new location. Callers
+	// are expected to Flush() before rehoming so this should normally be a no-op.
+	for (auto &p: queue_) {
+		if (p.buf == old_addr)
+			p.buf = new_addr;
+	}
+
+	return result;
+}
+
+
 void
 SwapManager::Attach(Buffer *buf)
 {

Swap.h

@@ -94,6 +94,15 @@ public:
 	// Detach(buf) or SwapManager destruction.
 	SwapRecorder *RecorderFor(Buffer *buf);
 
+	// Re-key an attached buffer's journal/recorder entries after its Buffer object
+	// has moved to a new address (e.g. std::vector<Buffer> reallocation/erase-shift).
+	// Callers must ensure no swap records for old_addr are in flight (see Flush())
+	// before calling this, and must not call it with an address that isn't
+	// currently attached. Returns the recorder for new_addr (nullptr if old_addr
+	// wasn't attached); the caller is responsible for calling
+	// new_buf->SetSwapRecorder() with the result.
+	SwapRecorder *Rehome(Buffer *old_addr, Buffer *new_addr);
+
 	// Notify that the buffer's filename changed (e.g., SaveAs)
 	void NotifyFilenameChanged(Buffer &buf);
 

TerminalInputHandler.cc

@@ -1,4 +1,6 @@
 #include <cstdio>
+#include <climits>
+#include <cwchar>
 #include <ncurses.h>
 
 #include "TerminalInputHandler.h"
@@ -12,6 +14,22 @@ CTRL(char c)
 {
 	return c & 0x1F;
 }
+
+
+// Encode a single wide character in the process locale's multibyte encoding
+// (UTF-8, given main.cc's setlocale(LC_ALL, "")). Returns false if the
+// codepoint can't be represented, leaving `out` untouched.
+bool
+wchar_to_mb(wchar_t wc, std::string &out)
+{
+	std::mbstate_t state{};
+	char buf[MB_LEN_MAX];
+	std::size_t n = std::wcrtomb(buf, wc, &state);
+	if (n == static_cast<std::size_t>(-1))
+		return false;
+	out.assign(buf, n);
+	return true;
+}
 }
 
 TerminalInputHandler::TerminalInputHandler() = default;
@@ -21,6 +39,7 @@ TerminalInputHandler::~TerminalInputHandler() = default;
 
 static bool
 map_key_to_command(const int ch,
+                   const bool is_keycode,
                    bool &k_prefix,
                    bool &esc_meta,
                    bool &k_ctrl_pending,
@@ -28,9 +47,13 @@ map_key_to_command(const int ch,
                    Editor *ed,
                    MappedInput &out)
 {
-	// Handle special keys from ncurses
+	// Handle special keys from ncurses. These are only meaningful when
+	// get_wch() reported KEY_CODE_YES: a regular (possibly non-ASCII) wide
+	// character can numerically collide with a KEY_* constant otherwise
+	// (e.g. U+0107 'ć' equals KEY_BACKSPACE's value), which would wrongly
+	// swallow it as a special key instead of inserting it.
 	// These keys exit k-prefix mode if active (user pressed C-k then a special key).
-	switch (ch) {
+	switch (is_keycode ? ch : -1) {
 	case KEY_ENTER:
 		// Some terminals send KEY_ENTER distinct from '\n'/'\r'
 		k_prefix = false;
@@ -259,7 +282,7 @@ map_key_to_command(const int ch,
 		esc_meta      = false;
 		int ascii_key = ch;
 		// Handle ESC + BACKSPACE (meta-backspace, Alt-Backspace)
-		if (ch == KEY_BACKSPACE || ch == 127 || ch == CTRL('H')) {
+		if ((is_keycode && ch == KEY_BACKSPACE) || ch == 127 || ch == CTRL('H')) {
 			ascii_key = KEY_BACKSPACE; // normalized value for lookup
 		} else if (ch == ',') {
 			// Some terminals emit ',' when Shift state is lost after ESC; treat as '<'
@@ -281,7 +304,7 @@ map_key_to_command(const int ch,
 	}
 
 	// Backspace in ncurses can be KEY_BACKSPACE or 127
-	if (ch == KEY_BACKSPACE || ch == 127 || ch == CTRL('H')) {
+	if ((is_keycode && ch == KEY_BACKSPACE) || ch == 127 || ch == CTRL('H')) {
 		k_prefix       = false;
 		k_ctrl_pending = false;
 		out            = {true, CommandId::Backspace, "", 0};
@@ -297,12 +320,21 @@ map_key_to_command(const int ch,
 		return true;
 	}
 
-	// Printable ASCII
+	// Printable character: ASCII, or (for a regular decoded wide character,
-	if (ch >= 0x20 && ch <= 0x7E) {
+	// not a keycode) any other printable Unicode codepoint - e.g. accented
+	// Latin, Cyrillic, CJK, etc.
+	if (!is_keycode && ch >= 0x20 && ch != 0x7F) {
+		std::string mb;
+		if (ch <= 0x7E) {
+			mb.assign(1, static_cast<char>(ch));
+		} else if (!wchar_to_mb(static_cast<wchar_t>(ch), mb)) {
+			out.hasCommand = false;
+			return true;
+		}
 		out.hasCommand = true;
 		out.id         = CommandId::InsertText;
-		out.arg.assign(1, static_cast<char>(ch));
+		out.arg        = mb;
-		out.count = 0;
+		out.count      = 0;
 		return true;
 	}
 
@@ -314,12 +346,15 @@ map_key_to_command(const int ch,
 bool
 TerminalInputHandler::decode_(MappedInput &out)
 {
-	int ch = getch();
+	wint_t wch;
-	if (ch == ERR) {
+	int ret = get_wch(&wch);
+	if (ret == ERR) {
 		return false; // no input
 	}
+	const bool is_keycode = (ret == KEY_CODE_YES);
+	const int ch          = static_cast<int>(wch);
 	bool consumed = map_key_to_command(
-		ch,
+		ch, is_keycode,
 		k_prefix_, esc_meta_,
 		k_ctrl_pending_,
 		mouse_selecting_,

UndoNode.h

@@ -10,6 +10,12 @@ enum class UndoType : std::uint8_t {
 	Newline,
 	DeleteRow,
 	InsertRow,
+	// Inverse of Newline: forward = join_lines(row) [removes the newline at the
+	// end of `row`], backward = split_line(row, col) [recreates the original
+	// two lines]. Used by backspace-at-col-0 and delete-at-eol, whose forward
+	// action is a join, not a split — Newline's apply() semantics are the
+	// wrong direction for those.
+	JoinLines,
 };
 
 struct UndoNode {

UndoSystem.cc

@@ -37,7 +37,7 @@ UndoSystem::Begin(UndoType type)
 
 	// Some operations should always be standalone undo steps.
 	const bool always_standalone = (type == UndoType::Newline || type == UndoType::DeleteRow || type ==
-	                                UndoType::InsertRow);
+	                                UndoType::InsertRow || type == UndoType::JoinLines);
 	if (always_standalone) {
 		commit();
 	}
@@ -77,6 +77,7 @@ UndoSystem::Begin(UndoType type)
 			case UndoType::Newline:
 			case UndoType::DeleteRow:
 			case UndoType::InsertRow:
+			case UndoType::JoinLines:
 				break;
 			}
 		}
@@ -325,6 +326,16 @@ UndoSystem::apply(const UndoNode *node, int direction)
 			buf_->SetCursor(0, static_cast<std::size_t>(node->row));
 		}
 		break;
+	case UndoType::JoinLines:
+		// Mirror image of Newline: forward removes a newline, backward restores it.
+		if (direction > 0) {
+			buf_->join_lines(node->row);
+			buf_->SetCursor(static_cast<std::size_t>(node->col), static_cast<std::size_t>(node->row));
+		} else {
+			buf_->split_line(node->row, node->col);
+			buf_->SetCursor(0, static_cast<std::size_t>(node->row + 1));
+		}
+		break;
 	}
 }
 
@@ -424,6 +435,8 @@ UndoSystem::type_str(UndoType t)
 		return "DeleteRow";
 	case UndoType::InsertRow:
 		return "InsertRow";
+	case UndoType::JoinLines:
+		return "JoinLines";
 	}
 	return "?";
 }

UndoTree.cc

@@ -1,3 +1,24 @@
-// Placeholder translation unit for UndoTree struct definition.
+// Undo logic is implemented in UndoSystem; this file only owns node lifetime.
-// Undo logic is implemented in UndoSystem.
 #include "UndoTree.h"
+
+namespace {
+void
+free_node_graph(UndoNode *node)
+{
+	// Walk the sibling (redo-branch) list; for each node, recursively free its
+	// child subtree first, then the node itself.
+	while (node) {
+		UndoNode *next = node->next;
+		free_node_graph(node->child);
+		delete node;
+		node = next;
+	}
+}
+} // namespace
+
+
+UndoTree::~UndoTree()
+{
+	free_node_graph(root);
+	delete pending;
+}

UndoTree.h

@@ -7,4 +7,11 @@ struct UndoTree {
 	UndoNode *current = nullptr; // current state of buffer
 	UndoNode *saved   = nullptr; // points to node matching last save (for dirty flag)
 	UndoNode *pending = nullptr; // in-progress batch (detached)
+
+	// Frees the entire node graph (root's subtree/branches plus any detached
+	// pending node). `current`/`saved` are aliases into that same graph, not
+	// separately owned. Without this, closing/reloading/resetting a buffer
+	// (which replaces or destroys its UndoTree) leaks every UndoNode ever
+	// created for it.
+	~UndoTree();
 };

main.cc

@@ -292,6 +292,20 @@ main(int argc, char *argv[])
 			fe = std::make_unique<TerminalFrontend>();
 		}
 
+		// Guarantee Shutdown() runs on every exit path from here on (normal
+		// return, early return, or exception unwinding e.g. from Init/Execute/
+		// Step) so a crash never leaves the terminal in raw/cbreak mode.
+		struct FrontendShutdownGuard {
+			Frontend *fe;
+
+
+			~FrontendShutdownGuard()
+			{
+				if (fe)
+					fe->Shutdown();
+			}
+		} shutdown_guard{fe.get()};
+
 #if defined(KTE_BUILD_GUI) && defined(__APPLE__)
 		if (use_gui) {
 			/* likely using the .app, so need to cd */
@@ -320,8 +334,6 @@ main(int argc, char *argv[])
 			fe->Step(editor, running);
 		}
 
-		fe->Shutdown();
-
 		return 0;
 	} catch (const std::exception &e) {
 		std::string msg = std::string("Unhandled exception: ") + e.what();

syntax/GoHighlighter.cc

@@ -46,14 +46,45 @@ GoHighlighter::GoHighlighter()
 void
 GoHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const
 {
+	StatefulHighlighter::LineState prev;
+	(void) HighlightLineStateful(buf, row, prev, out);
+}
+
+
+StatefulHighlighter::LineState
+GoHighlighter::HighlightLineStateful(const Buffer &buf,
+                                     int row,
+                                     const LineState &prev,
+                                     std::vector<HighlightSpan> &out) const
+{
+	StatefulHighlighter::LineState state = prev;
 	if (row < 0 || static_cast<std::size_t>(row) >= buf.Nrows())
-		return;
+		return state;
 	std::string s = buf.GetLineString(static_cast<std::size_t>(row));
 	int n         = static_cast<int>(s.size());
 	int i         = 0;
 	int bol       = 0;
 	while (bol < n && (s[bol] == ' ' || s[bol] == '\t'))
 		++bol;
+
+	// Continue a multi-line block comment from the previous line.
+	if (state.in_block_comment) {
+		int j = i;
+		while (i + 1 < n) {
+			if (s[i] == '*' && s[i + 1] == '/') {
+				i += 2;
+				push(out, j, i, TokenKind::Comment);
+				state.in_block_comment = false;
+				break;
+			}
+			++i;
+		}
+		if (state.in_block_comment) {
+			push(out, j, n, TokenKind::Comment);
+			return state;
+		}
+	}
+
 	// line comment
 	while (i < n) {
 		char c = s[i];
@@ -82,7 +113,8 @@ GoHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightSp
 			}
 			if (!closed) {
 				push(out, i, n, TokenKind::Comment);
-				break;
+				state.in_block_comment = true;
+				return state;
 			} else {
 				push(out, i, j, TokenKind::Comment);
 				i = j;
@@ -152,5 +184,6 @@ GoHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightSp
 		push(out, i, i + 1, TokenKind::Default);
 		++i;
 	}
+	return state;
 }
 } // namespace kte

syntax/GoHighlighter.h

@@ -5,12 +5,17 @@
 #include <unordered_set>
 
 namespace kte {
-class GoHighlighter final : public LanguageHighlighter {
+class GoHighlighter final : public StatefulHighlighter {
 public:
 	GoHighlighter();
 
 	void HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const override;
 
+	LineState HighlightLineStateful(const Buffer &buf,
+	                                int row,
+	                                const LineState &prev,
+	                                std::vector<HighlightSpan> &out) const override;
+
 private:
 	std::unordered_set<std::string> kws_;
 	std::unordered_set<std::string> types_;

syntax/HighlighterEngine.cc

@@ -74,7 +74,22 @@ HighlighterEngine::GetLine(const Buffer &buf, int row, std::uint64_t buf_version
 
 	StatefulHighlighter::LineState prev_state;
 	int start_row = -1;
-	if (!state_cache_.empty()) {
+
+	// Fast path: state_last_contig_ tracks the highest row we've cached state
+	// for, per version. If that row is already below our target, it's
+	// necessarily the best anchor the O(n) scan below would have found, so we
+	// can skip the scan entirely. Re-validated against state_cache_ before
+	// use since InvalidateFrom()/SetHighlighter() may have raced/cleared it.
+	auto contig_it = state_last_contig_.find(buf_version);
+	if (contig_it != state_last_contig_.end() && contig_it->second < row) {
+		auto sc_it = state_cache_.find(contig_it->second);
+		if (sc_it != state_cache_.end() && sc_it->second.version == buf_version) {
+			start_row  = contig_it->second;
+			prev_state = sc_it->second.state;
+		}
+	}
+
+	if (start_row < 0 && !state_cache_.empty()) {
 		// linear search over map (unordered), track best candidate
 		int best = -1;
 		for (const auto &kv: state_cache_) {
@@ -109,6 +124,9 @@ HighlighterEngine::GetLine(const Buffer &buf, int row, std::uint64_t buf_version
 		se.state        = next_state;
 		state_cache_[r] = se;
 		cur_state       = next_state;
+		int &contig     = state_last_contig_[buf_version];
+		if (r > contig)
+			contig = r;
 	}
 
 	// Store in cache and return by value
@@ -139,6 +157,14 @@ HighlighterEngine::InvalidateFrom(int row)
 				++it;
 		}
 	}
+	// A version's tracked contiguous-state row is no longer valid once any
+	// row at or above it has been evicted from state_cache_ above.
+	for (auto it = state_last_contig_.begin(); it != state_last_contig_.end();) {
+		if (it->second >= row)
+			it = state_last_contig_.erase(it);
+		else
+			++it;
+	}
 }
 
 

syntax/PythonHighlighter.cc

@@ -55,31 +55,26 @@ PythonHighlighter::HighlightLineStateful(const Buffer &buf, int row, const LineS
 	std::string s = buf.GetLineString(static_cast<std::size_t>(row));
 	int n         = static_cast<int>(s.size());
 
+	int i = 0;
+
 	// Triple-quoted string continuation uses in_raw_string with raw_delim either "'''" or "\"\"\""
 	if (state.in_raw_string && (state.raw_delim == "'''" || state.raw_delim == "\"\"\"")) {
 		auto pos = s.find(state.raw_delim);
 		if (pos == std::string::npos) {
 			push(out, 0, n, TokenKind::String);
 			return state; // still inside
-		} else {
-			int end = static_cast<int>(pos + static_cast<int>(state.raw_delim.size()));
-			push(out, 0, end, TokenKind::String);
-			// remainder processed normally
-			s                   = s.substr(end);
-			n                   = static_cast<int>(s.size());
-			state.in_raw_string = false;
-			state.raw_delim.clear();
-			// Continue parsing remainder as a separate small loop
-			int base = end;
-			// original offset, but we already emitted to 'out' with base=0; following spans should be from 'end'
-			// For simplicity, mark rest as Default
-			if (n > 0)
-				push(out, base, base + n, TokenKind::Default);
-			return state;
 		}
+		int end = static_cast<int>(pos + static_cast<int>(state.raw_delim.size()));
+		push(out, 0, end, TokenKind::String);
+		state.in_raw_string = false;
+		state.raw_delim.clear();
+		// Resume the normal tokenizer at the closing delimiter's end, on the
+		// same (unmodified) `s`/`n`, so anything after it - including a new
+		// triple-quoted string opening on this same line - is re-scanned
+		// instead of being dumped into a single opaque Default span.
+		i = end;
 	}
 
-	int i = 0;
 	// Detect comment start '#', ignoring inside strings
 	while (i < n) {
 		char c = s[i];

syntax/RustHighlighter.cc

@@ -47,11 +47,42 @@ RustHighlighter::RustHighlighter()
 void
 RustHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const
 {
+	StatefulHighlighter::LineState prev;
+	(void) HighlightLineStateful(buf, row, prev, out);
+}
+
+
+StatefulHighlighter::LineState
+RustHighlighter::HighlightLineStateful(const Buffer &buf,
+                                       int row,
+                                       const LineState &prev,
+                                       std::vector<HighlightSpan> &out) const
+{
+	StatefulHighlighter::LineState state = prev;
 	if (row < 0 || static_cast<std::size_t>(row) >= buf.Nrows())
-		return;
+		return state;
 	std::string s = buf.GetLineString(static_cast<std::size_t>(row));
 	int n         = static_cast<int>(s.size());
 	int i         = 0;
+
+	// Continue a multi-line block comment from the previous line.
+	if (state.in_block_comment) {
+		int j = i;
+		while (i + 1 < n) {
+			if (s[i] == '*' && s[i + 1] == '/') {
+				i += 2;
+				push(out, j, i, TokenKind::Comment);
+				state.in_block_comment = false;
+				break;
+			}
+			++i;
+		}
+		if (state.in_block_comment) {
+			push(out, j, n, TokenKind::Comment);
+			return state;
+		}
+	}
+
 	while (i < n) {
 		char c = s[i];
 		if (c == ' ' || c == '\t') {
@@ -79,7 +110,8 @@ RustHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<Highlight
 			}
 			if (!closed) {
 				push(out, i, n, TokenKind::Comment);
-				break;
+				state.in_block_comment = true;
+				return state;
 			} else {
 				push(out, i, j, TokenKind::Comment);
 				i = j;
@@ -140,5 +172,6 @@ RustHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<Highlight
 		push(out, i, i + 1, TokenKind::Default);
 		++i;
 	}
+	return state;
 }
 } // namespace kte

syntax/RustHighlighter.h

@@ -5,12 +5,17 @@
 #include <unordered_set>
 
 namespace kte {
-class RustHighlighter final : public LanguageHighlighter {
+class RustHighlighter final : public StatefulHighlighter {
 public:
 	RustHighlighter();
 
 	void HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const override;
 
+	LineState HighlightLineStateful(const Buffer &buf,
+	                                int row,
+	                                const LineState &prev,
+	                                std::vector<HighlightSpan> &out) const override;
+
 private:
 	std::unordered_set<std::string> kws_;
 	std::unordered_set<std::string> types_;

syntax/SqlHighlighter.cc

@@ -47,12 +47,42 @@ SqlHighlighter::SqlHighlighter()
 void
 SqlHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const
 {
+	StatefulHighlighter::LineState prev;
+	(void) HighlightLineStateful(buf, row, prev, out);
+}
+
+
+StatefulHighlighter::LineState
+SqlHighlighter::HighlightLineStateful(const Buffer &buf,
+                                      int row,
+                                      const LineState &prev,
+                                      std::vector<HighlightSpan> &out) const
+{
+	StatefulHighlighter::LineState state = prev;
 	if (row < 0 || static_cast<std::size_t>(row) >= buf.Nrows())
-		return;
+		return state;
 	std::string s = buf.GetLineString(static_cast<std::size_t>(row));
 	int n         = static_cast<int>(s.size());
 	int i         = 0;
 
+	// Continue a multi-line block comment from the previous line.
+	if (state.in_block_comment) {
+		int j = i;
+		while (i + 1 < n) {
+			if (s[i] == '*' && s[i + 1] == '/') {
+				i += 2;
+				push(out, j, i, TokenKind::Comment);
+				state.in_block_comment = false;
+				break;
+			}
+			++i;
+		}
+		if (state.in_block_comment) {
+			push(out, j, n, TokenKind::Comment);
+			return state;
+		}
+	}
+
 	while (i < n) {
 		char c = s[i];
 		if (c == ' ' || c == '\t') {
@@ -68,7 +98,7 @@ SqlHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightS
 			push(out, i, n, TokenKind::Comment);
 			break;
 		}
-		// simple block comment on same line: /* ... */
+		// block comment: /* ... */ (may span multiple lines)
 		if (c == '/' && i + 1 < n && s[i + 1] == '*') {
 			int j       = i + 2;
 			bool closed = false;
@@ -82,7 +112,8 @@ SqlHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightS
 			}
 			if (!closed) {
 				push(out, i, n, TokenKind::Comment);
-				break;
+				state.in_block_comment = true;
+				return state;
 			} else {
 				push(out, i, j, TokenKind::Comment);
 				i = j;
@@ -151,5 +182,6 @@ SqlHighlighter::HighlightLine(const Buffer &buf, int row, std::vector<HighlightS
 		push(out, i, i + 1, TokenKind::Default);
 		++i;
 	}
+	return state;
 }
 } // namespace kte

syntax/SqlHighlighter.h

@@ -5,12 +5,17 @@
 #include <unordered_set>
 
 namespace kte {
-class SqlHighlighter final : public LanguageHighlighter {
+class SqlHighlighter final : public StatefulHighlighter {
 public:
 	SqlHighlighter();
 
 	void HighlightLine(const Buffer &buf, int row, std::vector<HighlightSpan> &out) const override;
 
+	LineState HighlightLineStateful(const Buffer &buf,
+	                                int row,
+	                                const LineState &prev,
+	                                std::vector<HighlightSpan> &out) const override;
+
 private:
 	std::unordered_set<std::string> kws_;
 	std::unordered_set<std::string> types_;

tests/test_command_semantics.cc

@@ -108,3 +108,26 @@ TEST(CommandSemantics_CopyRegion_And_KillRegion)
 	ASSERT_EQ(b.MarkSet(), false);
 	ASSERT_EQ(h.Text(), std::string("hello "));
 }
+
+
+TEST(CommandSemantics_Syntax_OnOff_SetsUserOverride)
+{
+	TestHarness h;
+	Editor &ed = h.EditorRef();
+	Buffer &b  = h.Buf();
+
+	// Before any explicit :syntax command, nothing has overridden the
+	// frontend's config-driven default.
+	ASSERT_EQ(b.SyntaxUserOverride(), false);
+
+	ASSERT_TRUE(Execute(ed, CommandId::Syntax, "off"));
+	ASSERT_EQ(b.SyntaxEnabled(), false);
+	// This flag is what a frontend's per-frame "apply config default" pass
+	// must check before re-enabling syntax, or a manual :syntax off gets
+	// silently undone on the next frame.
+	ASSERT_EQ(b.SyntaxUserOverride(), true);
+
+	ASSERT_TRUE(Execute(ed, CommandId::Syntax, "on"));
+	ASSERT_EQ(b.SyntaxEnabled(), true);
+	ASSERT_EQ(b.SyntaxUserOverride(), true);
+}

tests/test_kkeymap.cc

@@ -37,6 +37,21 @@ TEST(KKeymap_KPrefix_CanonicalChords)
 }
 
 
+TEST(KKeymap_KPrefix_UppercaseE_RejectedAndDistinctFromLowercaseE)
+{
+	CommandId id{};
+
+	// 'e' is a real binding (OpenFileStart).
+	ASSERT_TRUE(KLookupKCommand('e', false, id));
+	ASSERT_EQ(id, CommandId::OpenFileStart);
+
+	// 'E' must be rejected, not silently aliased to 'e' via case-insensitive
+	// lookup (the switch normalizes to lowercase, so this has to be special-
+	// cased before it).
+	ASSERT_EQ(KLookupKCommand('E', false, id), false);
+}
+
+
 TEST(KKeymap_CtrlChords_CanonicalChords)
 {
 	CommandId id{};

tests/test_migration_coverage.cc

@@ -414,17 +414,21 @@ TEST (Migration_EmptyBufferCheck_Pattern)
 
 TEST (Migration_SyntaxHighlighter_Pattern)
 {
-	// Test the pattern used in syntax highlighters
+	// Test the pattern used in syntax highlighters: loop Nrows(), fetch each
+	// line via GetLineString(row). Assert the pattern actually returns the
+	// right content, not just that it doesn't crash - see syntax highlighter
+	// correctness tests (test_syntax_highlighting.cc) for tokenization coverage.
 	Buffer buf;
 	buf.insert_text(0, 0, std::string("int main() {\n    return 0;\n}"));
 
+	ASSERT_EQ(buf.Nrows(), static_cast<std::size_t>(3));
+	const std::vector<std::string> expected = {"int main() {", "    return 0;", "}"};
 	for (std::size_t row = 0; row < buf.Nrows(); ++row) {
-		// This is the pattern used in all migrated highlighters
 		if (row >= buf.Nrows()) {
 			break; // Should never happen
 		}
 		std::string line = buf.GetLineString(row);
-		// Successfully accessed line - size() is always valid for std::string
+		ASSERT_EQ(line, expected[row]);
 	}
 }
 

tests/test_paste_split.cc

@@ -0,0 +1,104 @@
+// Tests for SplitPasteIntoCommands: pasted text must turn embedded line
+// breaks into Newline commands so InsertText never receives a '\r' or '\n'.
+#include "Test.h"
+
+#include "PasteSplit.h"
+
+
+namespace {
+
+// Assert that no InsertText command carries an embedded newline/carriage
+// return, mirroring the rejection in Command.cc's InsertText handler.
+void
+assert_no_newlines_in_inserts(const std::vector<MappedInput> &cmds)
+{
+	for (const auto &c : cmds) {
+		if (c.id == CommandId::InsertText) {
+			ASSERT_TRUE(c.arg.find('\n') == std::string::npos);
+			ASSERT_TRUE(c.arg.find('\r') == std::string::npos);
+		}
+	}
+}
+
+} // namespace
+
+
+TEST(PasteSplit_PlainTextNoNewline)
+{
+	auto cmds = SplitPasteIntoCommands("hello world");
+	ASSERT_EQ(cmds.size(), 1u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText);
+	ASSERT_TRUE(cmds[0].arg == "hello world");
+}
+
+
+TEST(PasteSplit_UnixNewlines)
+{
+	auto cmds = SplitPasteIntoCommands("a\nb");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 3u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText && cmds[0].arg == "a");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[2].id == CommandId::InsertText && cmds[2].arg == "b");
+}
+
+
+// Regression: CRLF clipboards left a '\r' in the InsertText segment, which
+// InsertText rejected with "InsertText arg must not contain newlines".
+TEST(PasteSplit_WindowsCRLF)
+{
+	auto cmds = SplitPasteIntoCommands("a\r\nb");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 3u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText && cmds[0].arg == "a");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[2].id == CommandId::InsertText && cmds[2].arg == "b");
+}
+
+
+// Regression: a bare '\r' (classic Mac / some macOS apps) must also break lines.
+TEST(PasteSplit_BareCR)
+{
+	auto cmds = SplitPasteIntoCommands("a\rb");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 3u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText && cmds[0].arg == "a");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[2].id == CommandId::InsertText && cmds[2].arg == "b");
+}
+
+
+TEST(PasteSplit_BlankLinesPreserved)
+{
+	// Two line breaks between "a" and "b" => one empty line.
+	auto cmds = SplitPasteIntoCommands("a\n\nb");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 4u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText && cmds[0].arg == "a");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[2].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[3].id == CommandId::InsertText && cmds[3].arg == "b");
+}
+
+
+TEST(PasteSplit_TrailingNewline)
+{
+	auto cmds = SplitPasteIntoCommands("abc\n");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 2u);
+	ASSERT_TRUE(cmds[0].id == CommandId::InsertText && cmds[0].arg == "abc");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+}
+
+
+TEST(PasteSplit_MultilineCRLF)
+{
+	auto cmds = SplitPasteIntoCommands("line1\r\nline2\r\nline3");
+	assert_no_newlines_in_inserts(cmds);
+	ASSERT_EQ(cmds.size(), 5u);
+	ASSERT_TRUE(cmds[0].arg == "line1");
+	ASSERT_TRUE(cmds[1].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[2].arg == "line2");
+	ASSERT_TRUE(cmds[3].id == CommandId::Newline);
+	ASSERT_TRUE(cmds[4].arg == "line3");
+}

tests/test_search_replace_flow.cc

@@ -101,6 +101,36 @@ TEST(SearchFlow_SearchReplace_EmptyFind_DoesNotMutateBuffer_And_ClearsState)
 }
 
 
+TEST(SearchFlow_SearchReplace_EmptyWith_ReplacesAdjacentOverlappingMatches)
+{
+	TestHarness h;
+	Editor &ed = h.EditorRef();
+	Buffer &b  = h.Buf();
+
+	// "aaaa" with "aa" -> "" must remove both non-overlapping occurrences, not
+	// just the first: after deleting the match at column 0, the next "aa" now
+	// sits at column 0 too (not column 1), so the scan must resume at the
+	// deletion point rather than one character past it.
+	b.insert_text(0, 0, "aaaa\n");
+	b.SetCursor(0, 0);
+
+	ASSERT_TRUE(h.Exec(CommandId::SearchReplace));
+	ASSERT_TRUE(ed.PromptActive());
+	ASSERT_EQ(ed.CurrentPromptKind(), Editor::PromptKind::ReplaceFind);
+
+	ASSERT_TRUE(h.Exec(CommandId::InsertText, "aa"));
+	ASSERT_TRUE(h.Exec(CommandId::Newline));
+	ASSERT_TRUE(ed.PromptActive());
+	ASSERT_EQ(ed.CurrentPromptKind(), Editor::PromptKind::ReplaceWith);
+
+	// Leave the replacement empty and accept.
+	ASSERT_TRUE(h.Exec(CommandId::Newline));
+
+	ASSERT_TRUE(!ed.PromptActive());
+	ASSERT_EQ(std::string(b.Rows()[0]), std::string(""));
+}
+
+
 TEST(SearchFlow_RegexFind_InvalidPattern_FailsSafely_And_ClearsStateOnEnter)
 {
 	TestHarness h;

tests/test_syntax_highlighting.cc

@@ -0,0 +1,167 @@
+// test_syntax_highlighting.cc - tokenization correctness for LanguageHighlighter
+// implementations, focused on multi-line state propagation (block comments,
+// triple-quoted strings) that stateless-looking per-line scans easily get wrong.
+#include "Test.h"
+#include "Buffer.h"
+#include "Highlight.h"
+#include "syntax/LanguageHighlighter.h"
+#include "syntax/GoHighlighter.h"
+#include "syntax/RustHighlighter.h"
+#include "syntax/SqlHighlighter.h"
+#include "syntax/PythonHighlighter.h"
+
+#include <string>
+#include <vector>
+
+using kte::HighlightSpan;
+using kte::StatefulHighlighter;
+using kte::TokenKind;
+
+namespace {
+bool
+has_kind(const std::vector<HighlightSpan> &spans, TokenKind k)
+{
+	for (const auto &sp: spans) {
+		if (sp.kind == k)
+			return true;
+	}
+	return false;
+}
+} // namespace
+
+
+TEST (Syntax_Go_MultiLineBlockComment_PropagatesAcrossLines)
+{
+	Buffer b;
+	b.replace_all_bytes("/* start\nmiddle line\nend */\nfunc foo() {}\n");
+
+	kte::GoHighlighter hl;
+	StatefulHighlighter::LineState state;
+	std::vector<HighlightSpan> spans;
+
+	// Line 0 opens an unclosed block comment; state must carry forward.
+	state = hl.HighlightLineStateful(b, 0, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+
+	// Line 1 is entirely inside the comment; the whole line must be Comment,
+	// not re-tokenized as code (this is the bug: without state propagation
+	// "middle" and "line" would come back as identifiers).
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 1, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+	ASSERT_TRUE(!spans.empty());
+	for (const auto &sp: spans) {
+		ASSERT_TRUE(sp.kind == TokenKind::Comment);
+	}
+
+	// Line 2 closes the comment.
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 2, state, spans);
+	ASSERT_TRUE(!state.in_block_comment);
+	ASSERT_TRUE(has_kind(spans, TokenKind::Comment));
+
+	// Line 3 is ordinary code again: "func" must be a Keyword, not a Comment.
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 3, state, spans);
+	ASSERT_TRUE(!state.in_block_comment);
+	ASSERT_TRUE(has_kind(spans, TokenKind::Keyword));
+	ASSERT_TRUE(!has_kind(spans, TokenKind::Comment));
+}
+
+
+TEST (Syntax_Rust_MultiLineBlockComment_PropagatesAcrossLines)
+{
+	Buffer b;
+	b.replace_all_bytes("/* start\nmiddle line\nend */\nfn foo() {}\n");
+
+	kte::RustHighlighter hl;
+	StatefulHighlighter::LineState state;
+	std::vector<HighlightSpan> spans;
+
+	state = hl.HighlightLineStateful(b, 0, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 1, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+	for (const auto &sp: spans) {
+		ASSERT_TRUE(sp.kind == TokenKind::Comment);
+	}
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 2, state, spans);
+	ASSERT_TRUE(!state.in_block_comment);
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 3, state, spans);
+	ASSERT_TRUE(has_kind(spans, TokenKind::Keyword));
+	ASSERT_TRUE(!has_kind(spans, TokenKind::Comment));
+}
+
+
+TEST (Syntax_Sql_MultiLineBlockComment_PropagatesAcrossLines)
+{
+	Buffer b;
+	b.replace_all_bytes("/* start\nmiddle line\nend */\nSELECT 1;\n");
+
+	kte::SqlHighlighter hl;
+	StatefulHighlighter::LineState state;
+	std::vector<HighlightSpan> spans;
+
+	state = hl.HighlightLineStateful(b, 0, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 1, state, spans);
+	ASSERT_TRUE(state.in_block_comment);
+	for (const auto &sp: spans) {
+		ASSERT_TRUE(sp.kind == TokenKind::Comment);
+	}
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 2, state, spans);
+	ASSERT_TRUE(!state.in_block_comment);
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 3, state, spans);
+	ASSERT_TRUE(has_kind(spans, TokenKind::Keyword));
+	ASSERT_TRUE(!has_kind(spans, TokenKind::Comment));
+}
+
+
+TEST (Syntax_Python_TripleQuote_ClosesAndReopensOnSameLine)
+{
+	Buffer b;
+	// Line 0 opens a triple-quoted string that stays open.
+	// Line 1 closes the first string ("end'''") and, on the SAME line, opens a
+	// second triple-quoted string ("'''start of") that stays open into line 2.
+	b.replace_all_bytes("x = '''abc\nend''' + '''start of\nnext string'''\n");
+
+	kte::PythonHighlighter hl;
+	StatefulHighlighter::LineState state;
+	std::vector<HighlightSpan> spans;
+
+	state = hl.HighlightLineStateful(b, 0, state, spans);
+	ASSERT_TRUE(state.in_raw_string);
+	ASSERT_EQ(state.raw_delim, std::string("'''"));
+
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 1, state, spans);
+	// The bug: without re-scanning the remainder after the closing '''  the
+	// second opening ''' on this line is never noticed, so state.in_raw_string
+	// would incorrectly come back false here.
+	ASSERT_TRUE(state.in_raw_string);
+	ASSERT_EQ(state.raw_delim, std::string("'''"));
+
+	// Line 2 is inside the second string; it must be highlighted as String,
+	// not as ordinary code.
+	spans.clear();
+	state = hl.HighlightLineStateful(b, 2, state, spans);
+	ASSERT_TRUE(!spans.empty());
+	bool all_string = true;
+	for (const auto &sp: spans) {
+		if (sp.kind != TokenKind::String)
+			all_string = false;
+	}
+	ASSERT_TRUE(all_string);
+}

tests/test_undo.cc

@@ -1198,6 +1198,312 @@ TEST (Undo_Command_RedoCountSelectsBranch)
 }
 
 
+TEST (Undo_Command_Newline_UndoRejoinsCorrectLines)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abcdef\nghijkl");
+	buf->SetCursor(3, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::Newline));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(3));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("def"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("ghijkl"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("ghijkl"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_Backspace_JoinUndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abc\ndef");
+	buf->SetCursor(0, 1);
+
+	ASSERT_TRUE(Execute(ed, CommandId::Backspace));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(1));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("def"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(1));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_DeleteChar_JoinUndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abc\ndef");
+	buf->SetCursor(3, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::DeleteChar));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(1));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("def"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(1));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_RegexReplaceAll_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("foo one\nfoo two\nbar three");
+	buf->SetCursor(0, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::RegexpReplace));
+	ASSERT_TRUE(ed.PromptActive());
+	ed.SetPromptText("foo");
+	ASSERT_TRUE(Execute(ed, CommandId::Newline));
+	ASSERT_TRUE(ed.PromptActive());
+	ed.SetPromptText("baz");
+	ASSERT_TRUE(Execute(ed, CommandId::Newline));
+
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("baz one"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("baz two"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("bar three"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("foo one"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("foo two"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("bar three"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("baz one"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("baz two"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("bar three"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_IndentUnindentRegion_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("one\ntwo\nthree");
+	buf->SetMark(0, 0);
+	buf->SetCursor(0, 2);
+
+	ASSERT_TRUE(Execute(ed, CommandId::IndentRegion));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("\tone"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("\ttwo"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("\tthree"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("one"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("two"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("three"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("\tone"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("\ttwo"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("\tthree"));
+
+	buf->SetMark(0, 0);
+	buf->SetCursor(0, 2);
+	ASSERT_TRUE(Execute(ed, CommandId::UnindentRegion));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("one"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("two"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("three"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("\tone"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("\ttwo"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("\tthree"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_KillToEol_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abcdef\nghijkl");
+	buf->SetCursor(3, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::KillToEOL));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abcdef"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("ghijkl"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_KillLine_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abc\ndef\nghi");
+	buf->SetCursor(0, 1);
+
+	ASSERT_TRUE(Execute(ed, CommandId::KillLine));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("ghi"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(3));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("def"));
+	ASSERT_EQ(std::string(buf->Rows()[2]), std::string("ghi"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(buf->Nrows(), static_cast<std::size_t>(2));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc"));
+	ASSERT_EQ(std::string(buf->Rows()[1]), std::string("ghi"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_KillRegion_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abc def ghi");
+	buf->SetMark(4, 0);
+	buf->SetCursor(8, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::KillRegion));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc ghi"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc def ghi"));
+
+	ASSERT_TRUE(Execute(ed, CommandId::Redo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc ghi"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
+TEST (Undo_Command_DeleteWordPrevNext_UndoRedo)
+{
+	InstallDefaultCommands();
+
+	Editor ed;
+	ed.SetDimensions(24, 80);
+
+	Buffer b;
+	ed.AddBuffer(std::move(b));
+	Buffer *buf = ed.CurrentBuffer();
+	ASSERT_TRUE(buf != nullptr);
+
+	buf->replace_all_bytes("abc def ghi");
+	buf->SetCursor(8, 0);
+
+	ASSERT_TRUE(Execute(ed, CommandId::DeleteWordPrev));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc ghi"));
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc def ghi"));
+
+	buf->SetCursor(4, 0);
+	ASSERT_TRUE(Execute(ed, CommandId::DeleteWordNext));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc ghi"));
+	ASSERT_TRUE(Execute(ed, CommandId::Undo));
+	ASSERT_EQ(std::string(buf->Rows()[0]), std::string("abc def ghi"));
+
+	validate_undo_tree(*buf->Undo());
+}
+
+
 TEST (Undo_InsertRow_UndoDeletesRow)
 {
 	Buffer b;