perf: fast-path grapheme counts and cache terminal layouts

src/editor.rs

@@ -4516,7 +4516,12 @@ fn split_lines(text: &str) -> Vec<String> {
 }
 
 fn grapheme_count(line: &str) -> usize {
-    line.graphemes(true).count()
+    if line.is_ascii() {
+        // CRLF is the only multi-byte ASCII grapheme.
+        line.len() - line.matches("\r\n").count()
+    } else {
+        line.graphemes(true).count()
+    }
 }
 fn grapheme_byte(line: &str, column: usize) -> usize {
     line.grapheme_indices(true)
@@ -5525,6 +5530,35 @@ mod tests {
         assert!(editor.semantic_tokens().is_empty());
     }
 
+    #[test]
+    fn semantic_column_bounds_preserve_ascii_tabs_and_unicode_graphemes() {
+        use crate::plugin::SemanticTokenKind;
+        for text in ["a\tb", "e\u{301}界👩‍🚀"] {
+            let mut editor = editor();
+            editor.apply(vec![Effect::Replace(text.into())]).unwrap();
+            let token = SemanticToken {
+                start: TextPosition { row: 0, column: 0 },
+                end: TextPosition { row: 0, column: 3 },
+                kind: SemanticTokenKind::Variable,
+                modifiers: 0,
+            };
+            editor
+                .replace_semantic_tokens(1, editor.text_revision(), Some(Arc::from([token])))
+                .unwrap();
+            let outside = SemanticToken {
+                end: TextPosition { row: 0, column: 4 },
+                ..token
+            };
+            assert!(
+                editor
+                    .replace_semantic_tokens(1, editor.text_revision(), Some(Arc::from([outside])))
+                    .is_err(),
+                "{text:?}",
+            );
+            assert_eq!(editor.semantic_tokens(), &[token]);
+        }
+    }
+
     #[test]
     fn hover_scroll_and_escape_preserve_carets_and_edits_dismiss_normally() {
         let mut editor = editor();