v1.2.0 with basic killring.

Print the core dump k-command correctly.

.gitignore

@@ -1,2 +1,4 @@
 *.log
 build
+ke
+*.txt

CMakeLists.txt

@@ -2,21 +2,29 @@ cmake_minimum_required(VERSION 3.15)
 project(ke C)  # Specify C language explicitly
 
 set(CMAKE_C_STANDARD 99)
-set(KE_VERSION "1.0.8")
+set(KE_VERSION "1.2.0")
 
 set(CMAKE_C_FLAGS "-Wall -Wextra -pedantic -Wshadow -Werror -std=c99 -g")
 set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -D_DEFAULT_SOURCE -D_XOPEN_SOURCE")
-set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fsanitize=address -fno-omit-frame-pointer")
 
+# Optionally enable AddressSanitizer (ASan)
+option(ENABLE_ASAN "Enable AddressSanitizer for builds" OFF)
+
+if (ENABLE_ASAN)
+    message(STATUS "ASan enabled")
+    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fsanitize=address -fno-omit-frame-pointer")
+    # Ensure the sanitizer is linked too (especially important on some platforms)
+    set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=address")
+endif()
+
+include(GNUInstallDirs)
 
 # Add executable
 add_executable(ke main.c)
-
-# Define KE_VERSION for use in C code (e.g., #define KE_VERSION)
 target_compile_definitions(ke PRIVATE KE_VERSION="ke version ${KE_VERSION}")
+install(TARGETS ke RUNTIME DESTINATION bin)
+install(FILES ke.1 TYPE MAN)
+
+install(TARGETS ke RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
+install(FILES ke.1 DESTINATION ${CMAKE_INSTALL_MANDIR}/man1)
 
-# Set output properties
-set_target_properties(ke PROPERTIES
-        FOLDER bin
-        RUNTIME_OUTPUT_DIRECTORY bin
-)

Makefile

@@ -0,0 +1,26 @@
+TARGET :=		ke
+KE_VERSION :=		devel
+DEST :=			$(HOME)/.local/bin/$(TARGET)
+
+CFLAGS :=	-Wall -Wextra -pedantic -Wshadow -Werror -std=c99 -g
+CFLAGS +=	-D_DEFAULT_SOURCE -D_XOPEN_SOURCE
+CFLAGS +=	-fsanitize=address -fno-omit-frame-pointer
+
+LDFLAGS :=	-fsanitize=address
+
+all: $(TARGET) test.txt
+
+$(TARGET): main.c
+	$(CC) $(CFLAGS) -o $(TARGET) $(LDFLAGS) main.c
+
+.PHONY: install
+#install: $(TARGET) 
+install:
+	cp $(TARGET) $(DEST)
+
+clean:
+	rm -f $(TARGET)
+
+.PHONY: test.txt
+test.txt:
+	cp test.txt.bak $@

README.txt

@@ -11,4 +11,8 @@ It should be available via homebrew, even:
     brew tap kisom/homebrew-tap
     brew install ke
 
-Released under an ISC license.
+To get verbose ASAN messages:
+
+    export LSAN_OPTIONS=verbosity=1:log_threads=1
+
+Released under an ISC license.

ke.1

@@ -26,6 +26,8 @@ saving a file can be done with either C-k s or C-k C-s.
 Delete from the cursor to the beginning of the line.
 .It C-k d
 Delete from the cursor to the end of the line.
+.It C-k C-d
+Delete the entire link.
 .It C-k e
 Edit a new file. Also C-k C-e.
 .It C-k f
@@ -40,7 +42,9 @@ exit the editor. Also C-k C-q.
 save the file, prompting for a filename if needed. Also C-k C-s.
 .It C-k x
 save the file and exit. Also C-k C-x.
-.It C-k \
+.It C-k y
+Yank the killring.
+.It C-k \[char92]
 Dump core.
 .El
 .Sh FIND

main.c

@@ -19,6 +19,8 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <locale.h>
+#include <wchar.h>
 #include <termios.h>
 #include <time.h>
 #include <unistd.h>
@@ -49,6 +51,13 @@
 
 #define INITIAL_CAPACITY	64
 
+
+#define KILLRING_NO_OP		0	/* don't touch the killring */
+#define KILLRING_APPEND		1	/* append deleted chars */
+#define KILLRING_PREPEND	2	/* prepend deleted chars */
+#define KILLING_SET		3	/* set killring to deleted char */
+#define KILLRING_FLUSH		4	/* clear the killring */
+
 int
 next_power_of_2(int n)
 {
@@ -122,12 +131,22 @@ void	 editor_set_status(const char *fmt, ...);
 void		 die(const char *s);
 int		 get_winsz(int *rows, int *cols);
 void		 goto_line(void);
+int		 cursor_at_eol(void);
+void		 find_next_word(void);
+void		 delete_next_word(void);
+void		 find_prev_word(void);
+void		 delete_prev_word(void);
 void		 delete_row(int at);
+void		 killring_start_with_char(unsigned char ch);
+void		 killring_append_char(unsigned char ch);
+void		 killring_prepend_char(unsigned char ch);
+void		 killring_flush(void);
+void		 killring_yank(void);
 void		 row_append_row(struct erow *row, char *s, int len);
 void		 row_insert_ch(struct erow *row, int at, int16_t c);
 void		 row_delete_ch(struct erow *row, int at);
 void		 insertch(int16_t c);
-void		 deletech(void);
+void		 deletech(uint8_t op);
 void		 open_file(const char *filename);
 char		*rows_to_buffer(int *buflen);
 int		 save_file(void);
@@ -189,6 +208,9 @@ struct editor_t {
 	int		 nrows;
 	int		 rowoffs, coloffs;
 	struct erow	*row;
+ struct erow	*killring;
+ int		 killing;   /* are we in a contiguous delete sequence? */
+ int		 suppress_killrow; /* internal flag: don't add to killring inside delete_row */
 	char		*filename;
 	int		 dirty;
 	int		 dirtyex;
@@ -204,6 +226,9 @@ struct editor_t {
 	.rowoffs = 0,
 	.coloffs = 0,
 	.row = NULL,
+ .killring = NULL,
+ .killing = 0,
+ .suppress_killrow = 0,
 	.filename = NULL,
 	.dirty = 0,
 	.dirtyex = 0,
@@ -231,6 +256,9 @@ init_editor(void)
 	editor.nrows = 0;
 	editor.rowoffs = editor.coloffs = 0;
 	editor.row = NULL;
+	editor.killring = NULL;
+	editor.killing = 0;
+	editor.suppress_killrow = 0;
 
 	editor.msg[0] = '\0';
 	editor.msgtm = 0;
@@ -256,6 +284,12 @@ reset_editor(void)
 		editor.filename = NULL;
 	}
 
+	if (editor.killring != NULL) {
+		erow_free(editor.killring);
+		free(editor.killring);
+		editor.killring = NULL;
+	}
+
 	init_editor();
 }
 
@@ -298,25 +332,64 @@ nibble_to_hex(char c)
 {
 	c &= 0xf;
 	if (c < 10) {
-		return c + 0x30;
+		return (char)('0' + c);
 	}
-	return c + 0x41;
+	return (char)('A' + (c - 10));
 }
 
 
 int
 erow_render_to_cursor(struct erow *row, int cx)
 {
-	int	rx = 0;
-	int	j;
+	int rx		= 0;
+	size_t j	= 0;
 
-	for (j = 0; j < cx; j++) {
-		if (row->line[j] == '\t') {
-			rx += (TAB_STOP-1) - (rx%TAB_STOP);
-		} else if (row->line[j] < 0x20) {
-			rx += 2;
+	wchar_t		wc;
+	mbstate_t	st;
+
+	memset(&st, 0, sizeof(st));
+
+	while (j < (size_t)cx && j < (size_t)row->size) {
+		unsigned char b = (unsigned char)row->line[j];
+		if (b == '\t') {
+			rx += (TAB_STOP - 1) - (rx % TAB_STOP);
+			rx++;
+			j++;
+			continue;
+		}
+		if (b < 0x20) {
+			/* render as \xx -> width 3 */
+			rx += 3;
+			j++;
+			continue;
+		}
+
+		size_t rem	= (size_t)row->size - j;
+		size_t n	= mbrtowc(&wc, &row->line[j], rem, &st);
+
+		if (n == (size_t)-2) {
+			/* incomplete sequence at end; treat one byte */
+			rx += 1;
+			j += 1;
+			memset(&st, 0, sizeof(st));
+		}
+		else if (n == (size_t)-1) {
+			/* invalid byte; consume one and reset state */
+			rx += 1;
+			j += 1;
+			memset(&st, 0, sizeof(st));
+		}
+		else if (n == 0) {
+			/* null character */
+			rx += 0;
+			j += 1;
+		}
+		else {
+			int w = wcwidth(wc);
+			if (w < 0) w = 1; /* non-printable -> treat as width 1 */
+			rx += w;
+			j += n;
 		}
-		rx++;
 	}
 
 	return rx;
@@ -326,23 +399,60 @@ erow_render_to_cursor(struct erow *row, int cx)
 int
 erow_cursor_to_render(struct erow *row, int rx)
 {
-	int	cur_rx = 0;
-	int	curx = 0;
+	int cur_rx	= 0;
+	size_t j	= 0;
 
-	for (curx = 0; curx < row->size; curx++) {
-		if (row->line[curx] == '\t') {
-			cur_rx += (TAB_STOP - 1) - (cur_rx % TAB_STOP);
-		} else if (row->line[curx] < 0x20) {
-			cur_rx += 2;
+	wchar_t		wc;
+	mbstate_t	st;
+
+	memset(&st, 0, sizeof(st));
+
+	while (j < (size_t)row->size) {
+		int w = 0;
+		size_t adv = 1;
+
+		unsigned char b = (unsigned char)row->line[j];
+		if (b == '\t') {
+			int add = (TAB_STOP - 1) - (cur_rx % TAB_STOP);
+			w = add + 1;
+			adv = 1;
+			/* tabs are single byte */
 		}
-		cur_rx++;
+		else if (b < 0x20) {
+			w = 3; /* "\\xx" */
+			adv = 1;
+		}
+		else {
+			size_t rem = (size_t)row->size - j;
+			size_t n = mbrtowc(&wc, &row->line[j], rem, &st);
 
-		if (cur_rx > rx) {
+			if (n == (size_t)-2 || n == (size_t)-1) {
+				/* invalid/incomplete */
+				w = 1;
+				adv = 1;
+				memset(&st, 0, sizeof(st));
+			}
+			else if (n == 0) {
+				w = 0;
+				adv = 1;
+			}
+			else {
+				int ww = wcwidth(wc);
+				if (ww < 0) ww = 1;
+				w = ww;
+				adv = n;
+			}
+		}
+
+		if (cur_rx + w > rx) {
 			break;
 		}
+
+		cur_rx += w;
+		j += adv;
 	}
 
-	return curx;
+	return (int)j;
 }
 
 
@@ -359,29 +469,35 @@ erow_update(struct erow *row)
 	for (j = 0; j < row->size; j++) {
 		if (row->line[j] == '\t') {
 			tabs++;
-		} else if (!isprint(row->line[j])) {
+		}
+		else if ((unsigned char)row->line[j] < 0x20) {
+			/* treat only ASCII control characters as non-printable */
 			ctrl++;
 		}
 	}
 
-	if (row->rsize) {
+	if (row->rsize || row->render != NULL) {
 		free(row->render);
 		row->rsize = 0;
 	}
 	row->render = NULL;
-	row->render = malloc(row->size + (tabs * (TAB_STOP-1)) + (ctrl * 3) + 1);
+	row->render = malloc(row->size + (tabs * (TAB_STOP - 1)) + (ctrl * 3) + 1);
 	assert(row->render != NULL);
 
 	for (j = 0; j < row->size; j++) {
 		if (row->line[j] == '\t') {
 			do {
 				row->render[i++] = ' ';
-			} while ((i % TAB_STOP) != 0);
-		} else if (!isprint(row->line[j])) {
+			}
+			while ((i % TAB_STOP) != 0);
+		}
+		else if ((unsigned char)row->line[j] < 0x20) {
 			row->render[i++] = '\\';
 			row->render[i++] = nibble_to_hex(row->line[j] >> 4);
 			row->render[i++] = nibble_to_hex(row->line[j] & 0x0f);
-		} else {
+		}
+		else {
+			/* leave UTF-8 multibyte bytes untouched so terminal can render */
 			row->render[i++] = row->line[j];
 		}
 	}
@@ -438,6 +554,105 @@ erow_insert(int at, char *s, int len)
 }
 
 
+void
+killring_flush(void)
+{
+	if (editor.killring != NULL) {
+		erow_free(editor.killring);
+		free(editor.killring);
+		editor.killring = NULL;
+	}
+}
+
+
+void
+killring_yank(void)
+{
+	if (editor.killring == NULL) {
+		return;
+	}
+	/*
+	 * Insert killring contents at the cursor without clearing the ring.
+	 * Interpret '\n' as an actual newline() rather than inserting a raw 0x0A
+	 * byte, so yanked content preserves lines correctly.
+	 */
+	for (int i = 0; i < editor.killring->size; i++) {
+		unsigned char ch = (unsigned char)editor.killring->line[i];
+		if (ch == '\n') {
+			newline();
+		}
+		else {
+			insertch(ch);
+		}
+	}
+}
+
+
+void
+killring_start_with_char(unsigned char ch)
+{
+	struct erow	*row = NULL;
+
+	if (editor.killring != NULL) {
+		erow_free(editor.killring);
+		free(editor.killring);
+		editor.killring = NULL;
+	}
+
+	editor.killring = malloc(sizeof(struct erow));
+	assert(editor.killring != NULL);
+	assert(erow_init(editor.killring, 0) == 0);
+
+	/* append one char to empty killring without affecting editor.dirty */
+	row = editor.killring;
+
+	row->line = realloc(row->line, row->size + 2);
+	assert(row->line != NULL);
+	row->line[row->size] = ch;
+	row->size++;
+	row->line[row->size] = '\0';
+	erow_update(row);
+}
+
+
+void
+killring_append_char(unsigned char ch)
+{
+	struct erow	*row = NULL;
+
+	if (editor.killring == NULL) {
+		killring_start_with_char(ch);
+		return;
+	}
+
+	row = editor.killring;
+	row->line = realloc(row->line, row->size + 2);
+	assert(row->line != NULL);
+	row->line[row->size] = ch;
+	row->size++;
+	row->line[row->size] = '\0';
+	erow_update(row);
+}
+
+
+void
+killring_prepend_char(unsigned char ch)
+{
+	if (editor.killring == NULL) {
+		killring_start_with_char(ch);
+		return;
+	}
+
+	struct erow* row = editor.killring;
+	row->line = realloc(row->line, row->size + 2);
+	assert(row->line != NULL);
+	memmove(&row->line[1], &row->line[0], row->size + 1);
+	row->line[0] = ch;
+	row->size++;
+	erow_update(row);
+}
+
+
 void
 erow_free(struct erow *row)
 {
@@ -463,6 +678,7 @@ die(const char *s)
 }
 
 
+
 /*
  * get_winsz uses the TIOCGWINSZ to get the window size.
  *
@@ -508,6 +724,123 @@ goto_line(void)
 }
 
 
+int
+cursor_at_eol(void)
+{
+	assert(editor.curx >= 0);
+	assert(editor.cury >= 0);
+	assert(editor.cury <= editor.nrows);
+	assert(editor.curx <= editor.row[editor.cury].size);
+
+	return editor.curx == editor.row[editor.cury].size;
+}
+
+
+void
+find_next_word(void)
+{
+	while (cursor_at_eol()) {
+		move_cursor(ARROW_RIGHT);
+	}
+	
+	if (isalnum(editor.row[editor.cury].line[editor.curx])) {
+		while (!isspace(editor.row[editor.cury].line[editor.curx]) && !cursor_at_eol()) {
+			move_cursor(ARROW_RIGHT);
+		}
+
+		return;
+	}
+
+	if (isspace(editor.row[editor.cury].line[editor.curx])) {
+		while (isspace(editor.row[editor.cury].line[editor.curx])) {
+			move_cursor(ARROW_RIGHT);
+		}
+
+		find_next_word();
+	}
+}
+
+
+void
+delete_next_word(void)
+{
+    while (cursor_at_eol()) {
+        move_cursor(ARROW_RIGHT);
+        deletech(KILLRING_APPEND);
+    }
+
+ if (isalnum(editor.row[editor.cury].line[editor.curx])) {
+        while (!isspace(editor.row[editor.cury].line[editor.curx]) && !cursor_at_eol()) {
+            move_cursor(ARROW_RIGHT);
+            deletech(KILLRING_APPEND);
+        }
+
+        return;
+    }
+
+    if (isspace(editor.row[editor.cury].line[editor.curx])) {
+        while (isspace(editor.row[editor.cury].line[editor.curx])) {
+            move_cursor(ARROW_RIGHT);
+            deletech(KILLRING_APPEND);
+        }
+
+        delete_next_word();
+    }
+}
+
+
+void
+find_prev_word(void)
+{
+	if (editor.cury == 0 && editor.curx == 0) {
+		return;
+	}
+
+	move_cursor(ARROW_LEFT);
+
+	while (cursor_at_eol() || isspace(editor.row[editor.cury].line[editor.curx])) {
+		if (editor.cury == 0 && editor.curx == 0) {
+			return;
+		}
+
+		move_cursor(ARROW_LEFT);
+	}
+
+	while (editor.curx > 0 && !isspace(editor.row[editor.cury].line[editor.curx - 1])) {
+		move_cursor(ARROW_LEFT);
+	}
+}
+
+void
+delete_prev_word(void)
+{
+	if (editor.cury == 0 && editor.curx == 0) {
+		return;
+	}
+
+	deletech(KILLRING_PREPEND);
+
+	while (editor.cury > 0 || editor.curx > 0) {
+		if (editor.curx == 0) {
+			deletech(KILLRING_PREPEND);
+			continue
+		}
+
+		if (!isspace(editor.row[editor.cury].line[editor.curx - 1])) {
+			break;
+		}
+
+		deletech(KILLRING_PREPEND);
+	}
+
+	while (editor.curx > 0) {
+		if (isspace(editor.row[editor.cury].line[editor.curx - 1])) {
+			break;
+		}
+		deletech(KILLRING_PREPEND);
+	}
+}
+
 void
 delete_row(int at)
 {
@@ -515,9 +848,43 @@ delete_row(int at)
 		return;
 	}
 
+	/*
+	 * Update killring with the deleted row's contents followed by a newline
+	 * unless this deletion is an internal merge triggered by deletech at
+	 * start-of-line. In that case, deletech will account for the single
+	 * newline itself and we must NOT also push the entire row here.
+	 */
+	if (!editor.suppress_killrow) {
+		struct erow* r = &editor.row[at];
+		/* Start or continue the kill sequence based on editor.killing */
+		if (r->size > 0) {
+			/* push raw bytes of the line */
+			if (!editor.killing) {
+				killring_start_with_char((unsigned char)r->line[0]);
+				for (int i = 1; i < r->size; i++) {
+					killring_append_char((unsigned char)r->line[i]);
+				}
+			} else {
+				for (int i = 0; i < r->size; i++) {
+					killring_append_char((unsigned char)r->line[i]);
+				}
+			}
+			killring_append_char('\n');
+			editor.killing = 1;
+		} else {
+			if (!editor.killing) {
+				killring_start_with_char('\n');
+			}
+			else {
+				killring_append_char('\n');
+			}
+			editor.killing = 1;
+		}
+	}
+
 	erow_free(&editor.row[at]);
-	memmove(&editor.row[at], &editor.row[at+1],
-	    sizeof(struct erow) * (editor.nrows - at - 1));
+	memmove(&editor.row[at], &editor.row[at + 1],
+	        sizeof(struct erow) * (editor.nrows - at - 1));
 	editor.nrows--;
 	editor.dirty++;
 }
@@ -563,6 +930,7 @@ row_delete_ch(struct erow *row, int at)
 	if (at < 0 || at >= row->size) {
 		return;
 	}
+
 	memmove(&row->line[at], &row->line[at+1], row->size-at);
 	row->size--;
 	erow_update(row);
@@ -582,16 +950,24 @@ insertch(int16_t c)
 		erow_insert(editor.nrows, "", 0);
 	}
 
-	row_insert_ch(&editor.row[editor.cury], editor.curx, (char)(c & 0xff));
+	/* Any insertion breaks a delete sequence for killring chaining. */
+	editor.killing = 0;
+	/* Ensure we pass a non-negative byte value to avoid assert(c > 0). */
+	row_insert_ch(&editor.row[editor.cury], editor.curx,
+	              (int16_t)(c & 0xff));
 	editor.curx++;
 	editor.dirty++;
 }
 
 
+/*
+ * deletech
+ */
 void
-deletech(void)
+deletech(uint8_t op)
 {
-	struct erow	*row = NULL;
+	struct erow* row = NULL;
+	unsigned char dch = 0;
 
 	if (editor.cury >= editor.nrows) {
 		return;
@@ -602,16 +978,60 @@ deletech(void)
 	}
 
 	row = &editor.row[editor.cury];
+
+	/* determine which character is being deleted for killring purposes */
+	if (editor.curx > 0) {
+		dch = (unsigned char)row->line[editor.curx - 1];
+	}
+	else {
+		/* deleting at start of line merges with previous line: treat as newline */
+		dch = '\n';
+	}
+
+	/* update buffer */
 	if (editor.curx > 0) {
 		row_delete_ch(row, editor.curx - 1);
 		editor.curx--;
-	} else {
+	}
+	else {
 		editor.curx = editor.row[editor.cury - 1].size;
 		row_append_row(&editor.row[editor.cury - 1], row->line,
-		    row->size);
+		               row->size);
+		int prev = editor.suppress_killrow;
+		editor.suppress_killrow = 1; /* prevent killring update on internal row delete */
 		delete_row(editor.cury);
+		editor.suppress_killrow = prev;
 		editor.cury--;
 	}
+
+	/* update killring if requested */
+	if (op == KILLRING_FLUSH) {
+		killring_flush();
+		editor.killing = 0;
+		return;
+	}
+
+	if (op == KILLRING_NO_OP) {
+		/* Do not modify killring or chaining state. */
+		return;
+	}
+
+	if (!editor.killing) {
+		killring_start_with_char(dch);
+		editor.killing = 1;
+		return;
+	}
+
+	if (op == KILLING_SET) {
+		killring_start_with_char(dch);
+		editor.killing = 1;
+	}
+	else if (op == KILLRING_APPEND) {
+		killring_append_char(dch);
+	}
+	else if (op == KILLRING_PREPEND) {
+		killring_prepend_char(dch);
+	}
 }
 
 
@@ -627,7 +1047,7 @@ open_file(const char *filename)
 
 	editor.filename = strdup(filename);
 	assert(editor.filename != NULL);
-	
+
 	editor.dirty = 0;
 	if ((fp = fopen(filename, "r")) == NULL) {
 		if (errno == ENOENT) {
@@ -779,13 +1199,17 @@ is_arrow_key(int16_t c)
 int16_t
 get_keypress(void)
 {
-	char	seq[3];
-	char	c = -1;
+	char		seq[3];
+	/* read raw byte so UTF-8 bytes (>=0x80) are not sign-extended */
+	unsigned char	uc = 0;
+	int16_t		c;
 
-	if (read(STDIN_FILENO, &c, 1) == -1) {
+	if (read(STDIN_FILENO, &uc, 1) == -1) {
 		die("get_keypress:read");
 	}
 
+	c = (int16_t)uc;
+
 	if (c == 0x1b) {
 		if (read(STDIN_FILENO, &seq[0], 1) != 1) return c;
 		if (read(STDIN_FILENO, &seq[1], 1) != 1) return c;
@@ -793,7 +1217,7 @@ get_keypress(void)
 		if (seq[0] == '[') {
 			if (seq[1] < 'A') {
 				if (read(STDIN_FILENO, &seq[2], 1) != 1)
-					 return c;
+					return c;
 				if (seq[2] == '~') {
 					switch (seq[1]) {
 					case '1': return HOME_KEY;
@@ -805,7 +1229,8 @@ get_keypress(void)
 					case '8': return END_KEY;
 					}
 				}
-			} else {
+			}
+			else {
 				switch (seq[1]) {
 				case 'A': return ARROW_UP;
 				case 'B': return ARROW_DOWN;
@@ -818,7 +1243,8 @@ get_keypress(void)
 					/* nada */ ;
 				}
 			}
-		} else if (seq[0] == 'O') {
+		}
+		else if (seq[0] == 'O') {
 			switch (seq[1]) {
 			case 'F': return END_KEY;
 			case 'H': return HOME_KEY;
@@ -865,14 +1291,14 @@ editor_prompt(char *prompt, void (*cb)(char *, int16_t))
 				}
 				return buf;
 			}
-		} else if (!iscntrl(c) && c < 128) {
+		} else if ((c == TAB_KEY) || (c >= 0x20 && c != 0x7f)) {
 			if (buflen == bufsz - 1) {
 				bufsz *= 2;
 				buf = realloc(buf, bufsz);
 				assert(buf != NULL);
 			}
 
-			buf[buflen++] = c;
+			buf[buflen++] = (char)(c & 0xff);
 			buf[buflen] = '\0';
 		}
 
@@ -984,8 +1410,8 @@ editor_openfile(void)
 void
 move_cursor(int16_t c)
 {
-	struct erow	*row;
-	int	 reps;
+	struct erow* row;
+	int reps;
 
 	row = (editor.cury >= editor.nrows) ? NULL : &editor.row[editor.cury];
 
@@ -1006,7 +1432,13 @@ move_cursor(int16_t c)
 	case CTRL_KEY('f'):
 		if (row && editor.curx < row->size) {
 			editor.curx++;
-		} else if (row && editor.curx == row->size) {
+			/* skip over UTF-8 continuation bytes */
+			while (row && editor.curx < row->size &&
+				((unsigned char)row->line[editor.curx] & 0xC0) == 0x80) {
+				editor.curx++;
+			}
+		}
+		else if (row && editor.curx == row->size) {
 			editor.cury++;
 			editor.curx = 0;
 		}
@@ -1015,17 +1447,30 @@ move_cursor(int16_t c)
 	case CTRL_KEY('b'):
 		if (editor.curx > 0) {
 			editor.curx--;
-		} else if (editor.cury > 0) {
+			/* move to the start byte if we landed on a continuation */
+			while (editor.curx > 0 &&
+				((unsigned char)row->line[editor.curx] & 0xC0) == 0x80) {
+				editor.curx--;
+			}
+		}
+		else if (editor.cury > 0) {
 			editor.cury--;
 			editor.curx = editor.row[editor.cury].size;
+			/* ensure at a codepoint boundary at end of previous line */
+			row = &editor.row[editor.cury];
+			while (editor.curx > 0 &&
+				((unsigned char)row->line[editor.curx] & 0xC0) == 0x80) {
+				editor.curx--;
+			}
 		}
 		break;
 	case PG_UP:
-	case PG_DN: {
+	case PG_DN:
 		if (c == PG_UP) {
 			editor.cury = editor.rowoffs;
-		} else if (c == PG_DN) {
-			editor.cury = editor.rowoffs + editor.rows-1;
+		}
+		else if (c == PG_DN) {
+			editor.cury = editor.rowoffs + editor.rows - 1;
 			if (editor.cury > editor.nrows) {
 				editor.cury = editor.nrows;
 			}
@@ -1037,7 +1482,6 @@ move_cursor(int16_t c)
 		}
 
 		break;
-	}
 
 	case HOME_KEY:
 	case CTRL_KEY('a'):
@@ -1055,8 +1499,7 @@ move_cursor(int16_t c)
 	}
 
 
-	row = (editor.cury >= editor.nrows) ?
-	      NULL : &editor.row[editor.cury];
+	row = (editor.cury >= editor.nrows) ? NULL : &editor.row[editor.cury];
 	reps = row ? row->size : 0;
 	if (editor.curx > reps) {
 		editor.curx = reps;
@@ -1067,22 +1510,24 @@ move_cursor(int16_t c)
 void
 newline(void)
 {
-	struct erow	*row = NULL;
+    struct erow	*row = NULL;
 
-	if (editor.curx == 0) {
-		erow_insert(editor.cury, "", 0);
-	} else {
-		row = &editor.row[editor.cury];
-		erow_insert(editor.cury + 1, &row->line[editor.curx],
-		    row->size - editor.curx);
-		row = &editor.row[editor.cury];
-		row->size = editor.curx;
-		row->line[row->size] = '\0';
-		erow_update(row);
-	}
+    if (editor.curx == 0) {
+        erow_insert(editor.cury, "", 0);
+    } else {
+        row = &editor.row[editor.cury];
+        erow_insert(editor.cury + 1, &row->line[editor.curx],
+            row->size - editor.curx);
+        row = &editor.row[editor.cury];
+        row->size = editor.curx;
+        row->line[row->size] = '\0';
+        erow_update(row);
+    }
 
-	editor.cury++;
-	editor.curx = 0;
+    editor.cury++;
+    editor.curx = 0;
+    /* Any insertion breaks a delete sequence for killring chaining. */
+    editor.killing = 0;
 }
 
 
@@ -1109,9 +1554,15 @@ process_kcommand(int16_t c)
 		delete_row(editor.cury);
 		break;
 	case 'd':
+		if (editor.curx == 0 && cursor_at_eol()) {
+			delete_row(editor.cury);
+			return;
+		}
+
 		while ((editor.row[editor.cury].size - editor.curx) > 0) {
 			process_normal(DEL_KEY);
 		}
+
 		break;
 	case 'g':
 	case CTRL_KEY('g'):
@@ -1128,6 +1579,11 @@ process_kcommand(int16_t c)
 		abort();
 	case 'e':
 	case CTRL_KEY('e'):
+		if (editor.dirty && editor.dirtyex) {
+			editor_set_status("File not saved - C-k e again to open a new file anyways.");
+			editor.dirtyex = 0;
+			return;
+		}
 		editor_openfile();
 		break;
 	case 'f':
@@ -1140,6 +1596,12 @@ process_kcommand(int16_t c)
 			editor_set_status("make: ok");
 		}
 		break;
+	case 'y':
+		killring_yank();
+		break;
+	default:
+		editor_set_status("unknown kcommand: %04x", c);
+		return;
 	}
 
 	editor.dirtyex = 1;
@@ -1151,6 +1613,8 @@ void
 process_normal(int16_t c)
 {
 	if (is_arrow_key(c)) {
+		/* moving the cursor breaks a delete sequence */
+		editor.killing = 0;
 		move_cursor(c);
 		return;
 	}
@@ -1168,9 +1632,11 @@ process_normal(int16_t c)
 	case DEL_KEY:
 		if (c == DEL_KEY || c == CTRL_KEY('d')) {
 			move_cursor(ARROW_RIGHT);
+			deletech(KILLRING_APPEND);
+		}
+		else {
+			deletech(KILLRING_PREPEND);
 		}
-
-		deletech();
 		break;
 	case CTRL_KEY('l'):
 		display_refresh();
@@ -1182,7 +1648,8 @@ process_normal(int16_t c)
 		editor.mode = MODE_ESCAPE;
 		break;
 	default:
-		if (isprint(c) || c == TAB_KEY) {
+		/* Insert any printable byte: ASCII 0x20–0x7E and all bytes >=0x80. */
+		if ((c == TAB_KEY) || (c >= 0x20 && c != 0x7f)) {
 			insertch(c);
 		}
 		break;
@@ -1195,8 +1662,6 @@ process_normal(int16_t c)
 void
 process_escape(int16_t c)
 {
-	struct erow	*row = &editor.row[editor.cury];
-
 	editor_set_status("hi");
 
 	switch (c) {
@@ -1208,16 +1673,17 @@ process_escape(int16_t c)
 		editor.cury = 0;
 		editor.curx = 0;
 		break;
+	case 'b':
+		find_prev_word();
+		break;
+	case 'd':
+		delete_next_word();
+		break;
+	case 'f':
+		find_next_word();
+		break;
 	case BACKSPACE:
-		if (isalnum(row->line[editor.curx])) {
-			editor_set_status("is alnum");
-			while (editor.curx > 0 && isalnum(row->line[editor.curx])) {
-				process_normal(BACKSPACE);
-			}
-		} else {
-			editor_set_status("not alnum");
-			process_normal(BACKSPACE);
-		}
+		delete_prev_word();
 		break;
 	default:
 		editor_set_status("unknown ESC key: %04x", c);
@@ -1533,6 +1999,9 @@ loop(void)
 int
 main(int argc, char *argv[])
 {
+	// Set locale for proper UTF-8 handling
+	setlocale(LC_ALL, "");
+
 	setup_terminal();
 	init_editor();
 

test.txt.bak

@@ -0,0 +1,15 @@
+hello
+
+
+
+
+
+
+
+
+
+
+
+
+world it is 	me
+	but I am not me