From 5209ec8032ceb61b60852dd51b25afa0b9b9934f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Minh=20Nguy=E1=BB=85n?= <mxn@1ec5.org>
Date: Wed, 21 Aug 2024 19:24:23 -0700
Subject: [PATCH] Fixed ligatures in Sinhala

Replace zero-width joiners with temporary strip markers to prevent ICU from stripping them.
---
 src/symbol/shaping.ts | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/src/symbol/shaping.ts b/src/symbol/shaping.ts
index b82875e1ad6..eb802b9dcdd 100644
--- a/src/symbol/shaping.ts
+++ b/src/symbol/shaping.ts
@@ -272,11 +272,14 @@ function shapeText(
         lines = [];
         // ICU operates on code units.
         lineBreaks = lineBreaks.map(index => logicalInput.toCodeUnitIndex(index));
+        // Replace zero-width joiners with temporary strip markers (from the Private Use Area) to prevent ICU from stripping them out.
+        const markedInput = logicalInput.toString().replace(/\u200D/g, '\uF8FF');
         const untaggedLines =
-            processBidirectionalText(logicalInput.toString(), lineBreaks);
+            processBidirectionalText(markedInput, lineBreaks);
         for (const line of untaggedLines) {
             const taggedLine = new TaggedString();
-            taggedLine.text = line;
+            // Restore zero-width joiners from temporary strip markers.
+            taggedLine.text = line.replace(/\uF8FF/g, '\u200D');
             taggedLine.sections = logicalInput.sections;
             // eslint-disable-next-line @typescript-eslint/no-unused-vars
             for (const char of splitByGraphemeCluster(line)) {
@@ -290,20 +293,23 @@ function shapeText(
         lines = [];
         // ICU operates on code units.
         lineBreaks = lineBreaks.map(index => logicalInput.toCodeUnitIndex(index));
+        // Replace zero-width joiners with temporary strip markers (from the Private Use Area) to prevent ICU from stripping them out.
+        const markedInput = logicalInput.toString().replace(/\u200D/g, '\uF8FF');
 
         // Convert grapheme cluster–based section index to be based on code units.
         let i = 0;
         const sectionIndex = [];
-        for (const grapheme of splitByGraphemeCluster(logicalInput.text)) {
+        for (const grapheme of splitByGraphemeCluster(markedInput)) {
             sectionIndex.push(...Array(grapheme.length).fill(logicalInput.sectionIndex[i]));
             i++;
         }
 
         const processedLines =
-            processStyledBidirectionalText(logicalInput.text, sectionIndex, lineBreaks);
+            processStyledBidirectionalText(markedInput, sectionIndex, lineBreaks);
         for (const line of processedLines) {
             const taggedLine = new TaggedString();
-            taggedLine.text = line[0];
+            // Restore zero-width joiners from temporary strip markers.
+            taggedLine.text = line[0].replace(/\uF8FF/g, '\u200D');
             taggedLine.sectionIndex = line[1];
             taggedLine.sections = logicalInput.sections;
             lines.push(taggedLine);