From 43bad058c20d0ace2a1555c70e330fb1808185e0 Mon Sep 17 00:00:00 2001
From: Mike Auteri <mauteri@gmail.com>
Date: Fri, 11 Sep 2026 23:45:21 -0400
Subject: [PATCH] Leave fragment-only references alone when rewriting URLs

A reference like `#section` points within the document that contains
it, so there is no origin to migrate. Resolving it against the base URL
made it look like a child of the source site, and rewriting it turned an
in-page anchor into `/#section` and a bare `#` into `/`. Blocks whose
saved markup carries such an href then fail validation after an import.

Mirrors WordPress/wordpress-importer#263.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---
 .../DataLiberation/Tests/RewriteUrlsTest.php  | 24 +++++++++++++++++++
 components/DataLiberation/URL/functions.php   | 15 +++++++++++-
 2 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/components/DataLiberation/Tests/RewriteUrlsTest.php b/components/DataLiberation/Tests/RewriteUrlsTest.php
index 17aca6d57..3b947432a 100644
--- a/components/DataLiberation/Tests/RewriteUrlsTest.php
+++ b/components/DataLiberation/Tests/RewriteUrlsTest.php
@@ -213,6 +213,30 @@ public static function provider_test_wp_rewrite_urls() {
 				'https://🚀-science.com/science',
 				'https://science.wordpress.org',
 			),
+			'Fragment-only reference in an HTML attribute is left alone' => array(
+				'<a href="#section">Jump</a>',
+				'<a href="#section">Jump</a>',
+				'https://legacy-blog.com',
+				'https://modern-webstore.org',
+			),
+			'Bare hash in an HTML attribute is left alone' => array(
+				'<a href="#">Placeholder</a>',
+				'<a href="#">Placeholder</a>',
+				'https://legacy-blog.com',
+				'https://modern-webstore.org',
+			),
+			'Fragment-only reference in a block attribute is left alone' => array(
+				'<!-- wp:button {"url":"#login"} -->',
+				'<!-- wp:button {"url":"#login"} -->',
+				'https://legacy-blog.com',
+				'https://modern-webstore.org',
+			),
+			'Absolute URL carrying a fragment is still rewritten and keeps its fragment' => array(
+				'<a href="https://legacy-blog.com/page#section">Deep link</a>',
+				'<a href="https://modern-webstore.org/page#section">Deep link</a>',
+				'https://legacy-blog.com',
+				'https://modern-webstore.org',
+			),
 		);
 	}
 
diff --git a/components/DataLiberation/URL/functions.php b/components/DataLiberation/URL/functions.php
index fe931ce3d..daecc82d2 100644
--- a/components/DataLiberation/URL/functions.php
+++ b/components/DataLiberation/URL/functions.php
@@ -70,7 +70,20 @@ function wp_rewrite_urls( $options ) {
 	while ( $p->next_url() ) {
 		$token_type = $p->get_token_type();
 		$raw_url    = $p->get_raw_url();
-		$cache_key  = $mapping_cache_key . "\0" . $token_type . "\0" . $raw_url;
+
+		/*
+		 * Leave fragment-only references alone. A URL like `#section` points
+		 * within the document that contains it, so there is no origin to
+		 * migrate. Resolving it against the base URL makes it look like a
+		 * child of the site being imported from, and rewriting it then turns
+		 * an in-page anchor into a link somewhere else entirely:
+		 * `#section` becomes `/#section`, and a bare `#` becomes `/`.
+		 */
+		if ( is_string( $raw_url ) && 0 === strpos( $raw_url, '#' ) ) {
+			continue;
+		}
+
+		$cache_key = $mapping_cache_key . "\0" . $token_type . "\0" . $raw_url;
 
 		$cached = $rewrite_cache->get( $cache_key );
 		if ( null !== $cached ) {