From 78b47109af9ffbb2f2d8eceb24a1f79107aae2f9 Mon Sep 17 00:00:00 2001
From: Tim Ledbetter <tim.ledbetter@ladybird.org>
Date: Mon, 27 Apr 2026 16:25:50 +0100
Subject: [PATCH] LibWeb: Use the ASCII line segmenter for eligible text

---
 Libraries/LibWeb/DOM/CharacterData.cpp   | 13 +++++++++----
 Libraries/LibWeb/GraphemeEdgeTracker.cpp |  5 ++++-
 Libraries/LibWeb/Layout/TextNode.cpp     |  9 +++++++--
 3 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/Libraries/LibWeb/DOM/CharacterData.cpp b/Libraries/LibWeb/DOM/CharacterData.cpp
index 2ab9ee2228..4f8b1c0ad8 100644
--- a/Libraries/LibWeb/DOM/CharacterData.cpp
+++ b/Libraries/LibWeb/DOM/CharacterData.cpp
@@ -146,8 +146,9 @@ WebIDL::ExceptionOr<void> CharacterData::replace_data(size_t offset, size_t coun
 
     if (m_grapheme_segmenter)
         m_grapheme_segmenter->set_segmented_text(m_data);
-    if (m_line_segmenter)
-        m_line_segmenter->set_segmented_text(m_data);
+    // The line segmenter may be the ASCII fast-path variant, which only accepts a subset of inputs; let the
+    // lazy getter re-pick the implementation against the new data.
+    m_line_segmenter = nullptr;
     if (m_word_segmenter)
         m_word_segmenter->set_segmented_text(m_data);
 
@@ -190,8 +191,12 @@ Unicode::Segmenter& CharacterData::grapheme_segmenter() const
 Unicode::Segmenter& CharacterData::line_segmenter() const
 {
     if (!m_line_segmenter) {
-        m_line_segmenter = document().line_segmenter().clone();
-        m_line_segmenter->set_segmented_text(m_data);
+        if (auto ascii = Unicode::Segmenter::try_create_for_ascii_line(m_data.utf16_view())) {
+            m_line_segmenter = ascii.release_nonnull();
+        } else {
+            m_line_segmenter = document().line_segmenter().clone();
+            m_line_segmenter->set_segmented_text(m_data);
+        }
     }
 
     return *m_line_segmenter;
diff --git a/Libraries/LibWeb/GraphemeEdgeTracker.cpp b/Libraries/LibWeb/GraphemeEdgeTracker.cpp
index e52d71e245..d785869f3a 100644
--- a/Libraries/LibWeb/GraphemeEdgeTracker.cpp
+++ b/Libraries/LibWeb/GraphemeEdgeTracker.cpp
@@ -6,6 +6,7 @@
 
 #include <AK/Utf16View.h>
 #include <LibGfx/TextLayout.h>
+#include <LibWeb/DOM/Document.h>
 #include <LibWeb/DOM/Text.h>
 #include <LibWeb/GraphemeEdgeTracker.h>
 #include <LibWeb/Layout/TextNode.h>
@@ -38,7 +39,9 @@ static float measure_text_width(Layout::TextNode const& text_node, Utf16View con
     auto grapheme_segmenter = text_node.grapheme_segmenter().clone();
     grapheme_segmenter->set_segmented_text(text);
 
-    auto line_segmenter = text_node.line_segmenter().clone();
+    // NB: The text node's cached line segmenter may be the ASCII fast-path variant, which only accepts a subset
+    //     of inputs; route through the document's ICU segmenter so arbitrary subviews are handled.
+    auto line_segmenter = text_node.document().line_segmenter().clone();
     line_segmenter->set_segmented_text(text);
 
     Layout::TextNode::ChunkIterator iterator { text_node, text, *grapheme_segmenter, *line_segmenter, CSS::WordBreak::Normal, false, false };
diff --git a/Libraries/LibWeb/Layout/TextNode.cpp b/Libraries/LibWeb/Layout/TextNode.cpp
index 73345b7969..50cf142b48 100644
--- a/Libraries/LibWeb/Layout/TextNode.cpp
+++ b/Libraries/LibWeb/Layout/TextNode.cpp
@@ -416,8 +416,13 @@ Unicode::Segmenter& TextNode::grapheme_segmenter() const
 Unicode::Segmenter& TextNode::line_segmenter() const
 {
     if (!m_line_segmenter) {
-        m_line_segmenter = document().line_segmenter().clone();
-        m_line_segmenter->set_segmented_text(text_for_rendering());
+        auto const& text = text_for_rendering();
+        if (auto ascii = Unicode::Segmenter::try_create_for_ascii_line(text.utf16_view())) {
+            m_line_segmenter = ascii.release_nonnull();
+        } else {
+            m_line_segmenter = document().line_segmenter().clone();
+            m_line_segmenter->set_segmented_text(text);
+        }
     }
 
     return *m_line_segmenter;