From fcb4d1546abba54102ccfd92382f44ea1d0148b9 Mon Sep 17 00:00:00 2001
From: DragonWork <97512+DragonWork@users.noreply.github.com>
Date: Sat, 12 Sep 2026 15:18:19 +0000
Subject: [PATCH] Preserve non-NBSP UTF-8 bytes during HTML serialization

---
 html5ever/src/serialize/mod.rs | 1 +
 rcdom/tests/html-serializer.rs | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/html5ever/src/serialize/mod.rs b/html5ever/src/serialize/mod.rs
index cce3eae8..7cab15b9 100644
--- a/html5ever/src/serialize/mod.rs
+++ b/html5ever/src/serialize/mod.rs
@@ -139,6 +139,7 @@ impl<Wr: Write> HtmlSerializer<Wr> {
                 },
                 _ => {
                     //  0xC2 not followed by 0xA0 (not NBSP), so keep looking.
+                    self.writer.write_all(&bytes[next_special..search_start])?;
                     continue;
                 },
             };
diff --git a/rcdom/tests/html-serializer.rs b/rcdom/tests/html-serializer.rs
index e3775072..176a417d 100644
--- a/rcdom/tests/html-serializer.rs
+++ b/rcdom/tests/html-serializer.rs
@@ -146,6 +146,10 @@ test!(
     r#"<base foo="&lt;'&gt;">"#
 );
 test!(attr_escape_amp, r#"<base foo="&amp;">"#);
+test!(
+    attr_literal_non_nbsp_c2,
+    r#"<p title="©®&nbsp;«&amp;»"></p>"#
+);
 test!(
     attr_escape_amp_2,
     r#"<base foo=&amp>"#,
@@ -168,6 +172,7 @@ test!(
 );
 
 test!(text_literal, r#"<p>"'"</p>"#);
+test!(text_literal_non_nbsp_c2, "<p>©®&nbsp;«&amp;»</p>");
 test!(text_escape_amp, r#"<p>&amp;</p>"#);
 test!(text_escape_amp_2, r#"<p>&amp</p>"#, r#"<p>&amp;</p>"#);
 test!(text_escape_nbsp, "<p>x\u{a0}y</p>", r#"<p>x&nbsp;y</p>"#);