HTML API: Ensure non-string HTML input is safely handled.

sirreal · sirreal · commit 055010bea45d · 2025-10-01T12:57:19.000Z
Prevents an issue where passing `null` to HTML API constructors could result in runtime errors. Developed in #9545. Props kraftbj, jonsurrell, westonruter. Fixes #63854. git-svn-id: https://develop.svn.wordpress.org/trunk@60887 602fd350-edb4-49c9-b593-d223f7449a82
diff --git a/src/wp-includes/html-api/class-wp-html-processor.php b/src/wp-includes/html-api/class-wp-html-processor.php
@@ -297,6 +297,15 @@ public static function create_fragment( $html, $context = '<body>', $encoding =
 			return null;
 		}
 
+		if ( ! is_string( $html ) ) {
+			_doing_it_wrong(
+				__METHOD__,
+				__( 'The HTML parameter must be a string.' ),
+				'6.9.0'
+			);
+			return null;
+		}
+
 		$context_processor = static::create_full_parser( "<!DOCTYPE html>{$context}", $encoding );
 		if ( null === $context_processor ) {
 			return null;
@@ -339,6 +348,14 @@ public static function create_full_parser( $html, $known_definite_encoding = 'UT
 		if ( 'UTF-8' !== $known_definite_encoding ) {
 			return null;
 		}
+		if ( ! is_string( $html ) ) {
+			_doing_it_wrong(
+				__METHOD__,
+				__( 'The HTML parameter must be a string.' ),
+				'6.9.0'
+			);
+			return null;
+		}
 
 		$processor                             = new static( $html, self::CONSTRUCTOR_UNLOCK_CODE );
 		$processor->state->encoding            = $known_definite_encoding;
diff --git a/src/wp-includes/html-api/class-wp-html-tag-processor.php b/src/wp-includes/html-api/class-wp-html-tag-processor.php
@@ -834,6 +834,14 @@ class WP_HTML_Tag_Processor {
 	 * @param string $html HTML to process.
 	 */
 	public function __construct( $html ) {
+		if ( ! is_string( $html ) ) {
+			_doing_it_wrong(
+				__METHOD__,
+				__( 'The HTML parameter must be a string.' ),
+				'6.9.0'
+			);
+			$html = '';
+		}
 		$this->html = $html;
 	}
 
diff --git a/tests/phpunit/tests/html-api/wpHtmlProcessor.php b/tests/phpunit/tests/html-api/wpHtmlProcessor.php
@@ -36,6 +36,28 @@ public function test_warns_that_the_static_creator_methods_should_be_called_inst
 		new WP_HTML_Processor( '<p>Light roast.</p>' );
 	}
 
+	/**
+	 * @ticket 63854
+	 *
+	 * @covers ::create_fragment
+	 * @expectedIncorrectUsage WP_HTML_Processor::create_fragment
+	 */
+	public function test_create_fragment_validates_html_parameter() {
+		$processor = WP_HTML_Processor::create_fragment( null );
+		$this->assertNull( $processor );
+	}
+
+	/**
+	 * @ticket 63854
+	 *
+	 * @covers ::create_full_parser
+	 * @expectedIncorrectUsage WP_HTML_Processor::create_full_parser
+	 */
+	public function test_create_full_parser_validates_html_parameter() {
+		$processor = WP_HTML_Processor::create_full_parser( null );
+		$this->assertNull( $processor );
+	}
+
 	/**
 	 * Once stepping to the end of the document, WP_HTML_Processor::get_tag
 	 * should no longer report a tag. It should report `null` because there
diff --git a/tests/phpunit/tests/html-api/wpHtmlTagProcessor.php b/tests/phpunit/tests/html-api/wpHtmlTagProcessor.php
@@ -72,6 +72,23 @@ public function test_has_self_closing_flag_matches_input_html( $html, $flag_is_s
 		}
 	}
 
+	/**
+	 * @ticket 63854
+	 *
+	 * @covers WP_HTML_Tag_Processor::__construct
+	 * @expectedIncorrectUsage WP_HTML_Tag_Processor::__construct
+	 */
+	public function test_constructor_validates_html_parameter() {
+		// Test that passing null triggers _doing_it_wrong and sets HTML to empty string.
+		$processor = new WP_HTML_Tag_Processor( null );
+
+		// Verify that the HTML was set to an empty string.
+		$this->assertSame( '', $processor->get_updated_html(), 'HTML should be set to empty string when null is passed' );
+
+		// Verify that next_token() works without errors (indicating the processor is in a valid state).
+		$this->assertFalse( $processor->next_token(), 'next_token() should work without errors when HTML is empty string' );
+	}
+
 	/**
 	 * Data provider. HTML tags which might have a self-closing flag, and an indicator if they do.
 	 *
