HTML API: Preserve decoder match length on named-reference miss.

The HTML decoder could report a match byte length when there was no match. Ensure the pass-by-reference match byte length is unmodified if there is no match.

Developed in WordPress#12379.

Follow-up to [58281].

Props jonsurrell, dmsnell, mukesh27.
See #65372.


git-svn-id: https://develop.svn.wordpress.org/trunk@62626 602fd350-edb4-49c9-b593-d223f7449a82

Author: sirreal

src/wp-includes/html-api/class-wp-html-decoder.php

@@ -361,7 +361,7 @@ public static function read_character_reference( $context, $text, $at = 0, &$mat
 
 		$name_length = 0;
 		$replacement = $html5_named_character_references->read_token( $text, $name_at, $name_length );
-		if ( false === $replacement ) {
+		if ( null === $replacement ) {
 			return null;
 		}
 

tests/phpunit/tests/html-api/wpHtmlDecoder.php

@@ -110,6 +110,39 @@ static function ( int $errno, string $errstr ) use ( &$errors ) {
 		$this->assertSame( "&\x00b", $decoded, 'Should have decoded the text without changing it.' );
 	}
 
+	/**
+	 * Ensures unmatched named character references leave the by-ref match length unchanged.
+	 *
+	 * @ticket 65372
+	 *
+	 * @dataProvider data_unmatched_named_character_references
+	 *
+	 * @param string $context       Decoder context.
+	 * @param string $raw_text_node Raw text containing an unmatched named character reference.
+	 */
+	public function test_unmatched_named_character_reference_does_not_set_match_byte_length( $context, $raw_text_node ): void {
+		$match_byte_length = 'sentinel';
+		$this->assertNull(
+			WP_HTML_Decoder::read_character_reference( $context, $raw_text_node, 0, $match_byte_length ),
+			'Should not have matched an unmatched named character reference.'
+		);
+		$this->assertSame( 'sentinel', $match_byte_length );
+	}
+
+	/**
+	 * Data provider.
+	 *
+	 * @return array<string, array{string, string}>.
+	 */
+	public static function data_unmatched_named_character_references(): array {
+		return array(
+			'text invalid name'                      => array( 'data', '&bogus;' ),
+			'text invalid short-name candidate'      => array( 'data', '&Fv=q' ),
+			'attribute invalid name'                 => array( 'attribute', '&bogus;' ),
+			'attribute invalid short-name candidate' => array( 'attribute', '&Fv=q' ),
+		);
+	}
+
 	/**
 	 * Ensures semicolonless legacy references decode before non-ASCII UTF-8 bytes in attributes.
 	 *