Leave fragment-only references alone when rewriting URLs

components/DataLiberation/BlockMarkup/class-blockmarkupurlprocessor.php

@@ -436,6 +436,26 @@ public function is_url_absolute() {
 		return WPURL::can_parse( $this->get_raw_url() );
 	}
 
+	/**
+	 * Returns true if the currently matched URL is a fragment-only reference,
+	 * such as `#section` or a bare `#`.
+	 *
+	 * A fragment-only reference points within the document that contains it,
+	 * so it has no origin of its own to rewrite. URL parsers strip leading C0
+	 * control and space characters before resolving a reference, so a value
+	 * like ` #section` counts as one too.
+	 *
+	 * @return bool Whether the currently matched URL is a fragment-only reference.
+	 */
+	public function is_fragment_only_reference() {
+		$raw_url = $this->get_raw_url();
+		if ( ! is_string( $raw_url ) ) {
+			return false;
+		}
+
+		return 0 === strpos( ltrim( $raw_url, "\x00..\x20" ), '#' );
+	}
+
 	public function get_inspected_attribute_name() {
 		if ( '#tag' !== $this->get_token_type() ) {
 			return false;

components/DataLiberation/Tests/BlockMarkupUrlProcessorTest.php

@@ -467,4 +467,26 @@ public function test_css_urls_with_regular_attributes() {
 		$expected = '<img src="https://new.com/replaced.png" style="border-image: url(&quot;https://new.com/replaced.png&quot;);">';
 		$this->assertEquals( $expected, $p->get_updated_html() );
 	}
+
+	/**
+	 *
+	 * @dataProvider provider_test_is_fragment_only_reference
+	 */
+	public function test_is_fragment_only_reference( $markup, $expected ) {
+		$p = new BlockMarkupUrlProcessor( $markup, 'https://example.com' );
+		$this->assertTrue( $p->next_url(), 'Failed to find the URL in the markup.' );
+		$this->assertSame( $expected, $p->is_fragment_only_reference() );
+	}
+
+	public static function provider_test_is_fragment_only_reference() {
+		return array(
+			'Named fragment'                => array( '<a href="#section">Jump</a>', true ),
+			'Bare hash'                     => array( '<a href="#">Top</a>', true ),
+			'Leading space'                 => array( '<a href=" #section">Jump</a>', true ),
+			'Leading tab and newline'       => array( "<a href=\"\t\n#section\">Jump</a>", true ),
+			'Fragment in a block attribute' => array( '<!-- wp:button {"url":"#login"} -->', true ),
+			'Absolute URL with a fragment'  => array( '<a href="https://example.com/page#section">Deep link</a>', false ),
+			'Relative path with a fragment' => array( '<a href="/page#section">Page</a>', false ),
+		);
+	}
 }

components/DataLiberation/Tests/RewriteUrlsTest.php

@@ -213,6 +213,36 @@ 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 with leading whitespace is left alone' => array(
+				'<a href=" #section">Jump</a>',
+				'<a href=" #section">Jump</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',
+			),
 		);
 	}
 

components/DataLiberation/URL/functions.php

@@ -70,7 +70,18 @@ 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. Resolving one 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 ( $p->is_fragment_only_reference() ) {
+			continue;
+		}
+
+		$cache_key = $mapping_cache_key . "\0" . $token_type . "\0" . $raw_url;
 
 		$cached = $rewrite_cache->get( $cache_key );
 		if ( null !== $cached ) {