From b4e8ed437142b6821da74a09e5dbeebe6b8703de Mon Sep 17 00:00:00 2001
From: Martynas Bagdonas <git.martynas@gmail.com>
Date: Tue, 18 Jun 2024 08:46:45 +0300
Subject: [PATCH] Fix URL parsing

---
 src/core/module/link/parsed-overlays.js | 3 ++-
 src/core/module/outline-reader.js       | 3 +++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/core/module/link/parsed-overlays.js b/src/core/module/link/parsed-overlays.js
index 8777ad61fc991..0129079d04be0 100644
--- a/src/core/module/link/parsed-overlays.js
+++ b/src/core/module/link/parsed-overlays.js
@@ -44,7 +44,8 @@ export function getParsedOverlays(chars) {
     let text = '';
     for (let j = sequence.from; j <= sequence.to; j++) {
       let char = chars[j];
-      text += char.c;
+      // Use the original char to avoid decomposed ligatures increasing match length
+      text += char.u;
     }
     let match = text.match(urlRegExp);
     if (match) {
diff --git a/src/core/module/outline-reader.js b/src/core/module/outline-reader.js
index 75a7eef4359d1..ceb81c819db9e 100644
--- a/src/core/module/outline-reader.js
+++ b/src/core/module/outline-reader.js
@@ -1,5 +1,8 @@
 import { getPositionFromDestination, getSortIndex } from './util.js';
 
+// TODO:
+// - If root node is single and is a document title, remove it and use its children as top level nodes
+
 async function getSortIndexFromTitle(pdfDocument, structuredCharsProvider, title, dest){
   let position = await getPositionFromDestination(pdfDocument, dest);
   if (!position) {