Fixed grapheme–section mapping during bidi processing

Author: 1ec5

src/symbol/shaping.test.ts

@@ -58,12 +58,12 @@ describe('TaggedString', () => {
         test('splits surrogate pairs', () => {
             const tagged = new TaggedString();
             tagged.text = '𰻞𰻞麵𪚥𪚥';
-            expect(tagged.codeUnitIndex(0)).toBe(0);
-            expect(tagged.codeUnitIndex(1)).toBe(2);
-            expect(tagged.codeUnitIndex(2)).toBe(4);
-            expect(tagged.codeUnitIndex(3)).toBe(5);
-            expect(tagged.codeUnitIndex(4)).toBe(7);
-            expect(tagged.codeUnitIndex(5)).toBe(9);
+            expect(tagged.toCodeUnitIndex(0)).toBe(0);
+            expect(tagged.toCodeUnitIndex(1)).toBe(2);
+            expect(tagged.toCodeUnitIndex(2)).toBe(4);
+            expect(tagged.toCodeUnitIndex(3)).toBe(5);
+            expect(tagged.toCodeUnitIndex(4)).toBe(7);
+            expect(tagged.toCodeUnitIndex(5)).toBe(9);
         });
     });
 });

src/symbol/shaping.ts

@@ -167,7 +167,7 @@ export class TaggedString {
     /**
      * Converts a grapheme cluster index to a UTF-16 code unit (JavaScript character index).
      */
-    codeUnitIndex(unicodeIndex: number): number {
+    toCodeUnitIndex(unicodeIndex: number): number {
         return splitByGraphemeCluster(this.text).slice(0, unicodeIndex).join('').length;
     }
 
@@ -271,7 +271,7 @@ function shapeText(
         // Bidi doesn't have to be style-aware
         lines = [];
         // ICU operates on code units.
-        lineBreaks = lineBreaks.map(index => logicalInput.codeUnitIndex(index));
+        lineBreaks = lineBreaks.map(index => logicalInput.toCodeUnitIndex(index));
         const untaggedLines =
             processBidirectionalText(logicalInput.toString(), lineBreaks);
         for (const line of untaggedLines) {
@@ -289,9 +289,20 @@ function shapeText(
         // with formatting
         lines = [];
         // ICU operates on code units.
-        lineBreaks = lineBreaks.map(index => logicalInput.codeUnitIndex(index));
+        lineBreaks = lineBreaks.map(index => logicalInput.toCodeUnitIndex(index));
+
+        // Convert grapheme cluster–based section index to be based on code units.
+        let i = 0;
+        let elapsedCodeUnits = 0;
+        let sectionIndex = [];
+        for (const grapheme of splitByGraphemeCluster(logicalInput.text)) {
+            sectionIndex.push(...Array(grapheme.length).fill(logicalInput.sectionIndex[i]));
+            i++;
+            elapsedCodeUnits += grapheme.length;
+        }
+
         const processedLines =
-            processStyledBidirectionalText(logicalInput.text, logicalInput.sectionIndex, lineBreaks);
+            processStyledBidirectionalText(logicalInput.text, sectionIndex, lineBreaks);
         for (const line of processedLines) {
             const taggedLine = new TaggedString();
             taggedLine.text = line[0];