Text: Backspace removes Thai and other marks one at a time (#1220)

Co-Authored-By: GPT-6 Luna <noreply@openai.com>

crates/engine/src/text_behaviour_tests.rs

@@ -60,11 +60,14 @@ fn arrows_and_deletes_step_over_whole_clusters() {
     key(&mut s, "Right", true, false, false).unwrap();
     assert_eq!(caret(&s), (1, 6));
     key(&mut s, "Left", false, false, false).unwrap();
-    // Delete and Backspace remove a whole cluster.
+    // Delete removes a whole cluster; Backspace takes a letter's marks one at a time (#1220) and
+    // an emoji cluster whole.
     key(&mut s, "Delete", false, false, false).unwrap();
     assert_eq!(chars(&s, t), "ab\u{1F1EF}\u{1F1F5}e\u{0301}");
     key(&mut s, "Right", false, false, true).unwrap(); // to the end
     key(&mut s, "Backspace", false, false, false).unwrap();
+    assert_eq!(chars(&s, t), "ab\u{1F1EF}\u{1F1F5}e");
+    key(&mut s, "Backspace", false, false, false).unwrap();
     assert_eq!(chars(&s, t), "ab\u{1F1EF}\u{1F1F5}");
     key(&mut s, "Backspace", false, false, false).unwrap();
     assert_eq!(chars(&s, t), "ab");

crates/engine/src/textedit.rs

@@ -243,7 +243,7 @@ pub fn key(s: &mut Session, key: &str, shift: bool, word: bool, line: bool) -> R
                     } else if word {
                         word_left(&chars, a)
                     } else {
-                        figcraft_text::grapheme_step(&t.characters, a, false)
+                        figcraft_text::backspace_step(&t.characters, a)
                     };
                     (from, a)
                 } else {
@@ -574,6 +574,16 @@ mod tests {
         end(&mut s).unwrap();
     }
 
+    #[test]
+    fn backspace_removes_only_the_final_thai_mark() {
+        let (mut s, id) = setup("น้ำ");
+        begin(&mut s, id, Some(3), false).unwrap();
+
+        key(&mut s, "Backspace", false, false, false).unwrap();
+
+        assert_eq!(data(&s, id).characters, "น้");
+    }
+
     fn ev(kind: PointerKind, x: f64, y: f64, shift: bool) -> PointerEvent {
         let mut e = PointerEvent::at(kind, x, y, 1.0);
         e.shift = shift;

crates/text/src/behaviour_tests.rs

@@ -361,6 +361,18 @@ fn grapheme_clusters() {
     assert!(grapheme_starts("plain ascii").iter().all(|b| *b));
 }
 
+#[test]
+fn backspace_removes_marks_but_keeps_other_graphemes_atomic() {
+    assert_eq!(crate::backspace_step("น้ำ", 3), 2); // Thai spacing mark
+    assert_eq!(crate::backspace_step("e\u{0301}", 2), 1); // Latin combining mark
+    assert_eq!(crate::backspace_step("👨‍👩‍👧", 5), 0); // Emoji ZWJ sequence
+    assert_eq!(crate::backspace_step("🇯🇵", 2), 0); // Flag
+    assert_eq!(crate::backspace_step("x", 1), 0); // Ordinary character
+    assert_eq!(crate::backspace_step("\u{1112}\u{1161}\u{11AB}", 3), 0); // Hangul syllable as conjoining jamo
+    assert_eq!(crate::backspace_step("a\u{200D}", 2), 0); // A letter ending in ZWJ
+    assert_eq!(crate::backspace_step("1\u{FE0F}\u{20E3}", 3), 0); // Keycap
+}
+
 #[test]
 fn a_cluster_is_shaped_with_one_face() {
     if face().is_none() {

crates/text/src/breaks.rs

@@ -89,7 +89,7 @@ pub(crate) fn grapheme_starts(text: &str) -> Vec<bool> {
 }
 
 /// The char index of the grapheme cluster boundary after (`forward`) or before char `index` in
-/// `text`: where an arrow key, Delete or Backspace moves the caret to.
+/// `text`: where an arrow key or Delete moves the caret to.
 pub fn grapheme_step(text: &str, index: usize, forward: bool) -> usize {
     let mut at = 0;
     for g in text.graphemes(true) {
@@ -102,6 +102,29 @@ pub fn grapheme_step(text: &str, index: usize, forward: bool) -> usize {
     at.min(index)
 }
 
+/// The start of the text deleted by Backspace at char `index`.
+///
+/// A letter's trailing mark (Grapheme_Cluster_Break Extend or SpacingMark, such as Thai MAI THO or
+/// SARA AM) is deleted on its own; other grapheme clusters remain atomic, so emoji sequences, flags,
+/// keycaps and Hangul syllables written as conjoining jamo stay intact.
+pub fn backspace_step(text: &str, index: usize) -> usize {
+    let end = text.chars().count().min(index);
+    let start = grapheme_step(text, end, false);
+    if start == end {
+        return end;
+    }
+
+    let cluster = text.chars().skip(start).take(end - start).collect::<Vec<_>>();
+    let letter_cluster = cluster.len() > 1 && cluster.first().is_some_and(|c| c.is_alphabetic());
+    if letter_cluster && cluster.last().is_some_and(|&c| is_trailing_mark(c)) { end - 1 } else { start }
+}
+
+/// Whether `c` is Extend or SpacingMark: it joins a plain letter's cluster. ZWJ also joins one but
+/// is a joiner, not a mark.
+fn is_trailing_mark(c: char) -> bool {
+    c != '\u{200D}' && format!("a{c}").graphemes(true).count() == 1
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

crates/text/src/lib.rs

@@ -13,7 +13,7 @@ pub mod fontdb;
 mod layout;
 mod onpath;
 
-pub use breaks::grapheme_step;
+pub use breaks::{backspace_step, grapheme_step};
 pub use color::{ColorBitmap, ColorInk, ColorLayer};
 pub use fontdb::{
     Axis, CRAFT_FONTS, Coords, Face, FontDb, NamedInstance, Rescan, axis_label, db, face_with_char, family_names, family_styles, find_face,