Benchmark setup HTML API

composer.json

@@ -24,7 +24,8 @@
 		"wp-coding-standards/wpcs": "~3.3.0",
 		"phpcompatibility/phpcompatibility-wp": "~2.1.3",
 		"phpstan/phpstan": "2.1.39",
-		"yoast/phpunit-polyfills": "^1.1.0"
+		"yoast/phpunit-polyfills": "^1.1.0",
+		"phpbench/phpbench": "^1.5"
 	},
 	"config": {
 		"allow-plugins": {

phpbench-bootstrap.php

@@ -0,0 +1,74 @@
+<?php
+/**
+ * Bootstrap file for phpbench benchmarks.
+ *
+ * Loads WordPress for running benchmarks.
+ */
+
+define( 'PHPBENCH_ROOT', __DIR__ );
+
+if ( defined( 'WP_TESTS_CONFIG_FILE_PATH' ) ) {
+	$config_file_path = WP_TESTS_CONFIG_FILE_PATH;
+} else {
+	$config_file_path = __DIR__ . '/wp-tests-config.php';
+}
+
+/*
+ * Globalize some WordPress variables.
+ */
+global $wpdb, $current_site, $current_blog, $wp_rewrite, $shortcode_tags, $wp, $phpmailer, $wp_theme_directories;
+
+if ( ! is_readable( $config_file_path ) ) {
+	exit( 1 );
+}
+
+require_once $config_file_path;
+
+if ( ! is_dir( ABSPATH ) ) {
+	exit( 1 );
+}
+
+function tests_reset__SERVER() { // phpcs:ignore WordPress.NamingConventions.ValidFunctionName.FunctionNameInvalid
+	$_SERVER['HTTP_HOST']       = WP_TESTS_DOMAIN;
+	$_SERVER['REMOTE_ADDR']     = '127.0.0.1';
+	$_SERVER['REQUEST_METHOD']  = 'GET';
+	$_SERVER['REQUEST_URI']     = '';
+	$_SERVER['SERVER_NAME']     = WP_TESTS_DOMAIN;
+	$_SERVER['SERVER_PORT']     = '80';
+	$_SERVER['SERVER_PROTOCOL'] = 'HTTP/1.1';
+
+	unset( $_SERVER['HTTP_REFERER'] );
+	unset( $_SERVER['HTTPS'] );
+}
+tests_reset__SERVER();
+
+define( 'WP_TESTS_TABLE_PREFIX', $table_prefix );
+
+define( 'DIR_BENCHMARKDATA', __DIR__ . '/tests/benchmarks/data' );
+define( 'DIR_TESTDATA', __DIR__ . '/tests/phpunit/data' );
+define( 'WP_LANG_DIR', realpath( DIR_TESTDATA . '/languages' ) );
+
+/*
+ * Cron tries to make an HTTP request to the site, which always fails,
+ * because tests are run in CLI mode only.
+ */
+define( 'DISABLE_WP_CRON', true );
+
+define( 'WP_MEMORY_LIMIT', -1 );
+define( 'WP_MAX_MEMORY_LIMIT', -1 );
+
+$PHP_SELF            = '/index.php';
+$GLOBALS['PHP_SELF'] = '/index.php';
+$_SERVER['PHP_SELF'] = '/index.php';
+
+if ( ! defined( 'WP_DEFAULT_THEME' ) ) {
+	define( 'WP_DEFAULT_THEME', 'default' );
+}
+$wp_theme_directories = array();
+
+if ( file_exists( DIR_TESTDATA . '/themedir1' ) ) {
+	$wp_theme_directories[] = DIR_TESTDATA . '/themedir1';
+}
+
+// Load WordPress.
+require_once ABSPATH . 'wp-settings.php';

phpbench.json

@@ -0,0 +1,5 @@
+{
+	"$schema": "./vendor/phpbench/phpbench/phpbench.schema.json",
+	"runner.bootstrap": "phpbench-bootstrap.php",
+	"runner.path": "tests/benchmarks/benchmarks"
+}

phpcs.xml.dist

@@ -286,6 +286,7 @@
 
 	<rule ref="PEAR.NamingConventions.ValidClassName.Invalid">
 		<exclude-pattern>/tests/phpunit/tests/*</exclude-pattern>
+		<exclude-pattern>/tests/benchmarks/benchmarks/*</exclude-pattern>
 		<!-- Exclude some old classes that cannot be renamed, as it would break back compat. -->
 		<exclude-pattern>/src/wp-admin/includes/class-wp-filesystem-ftpsockets\.php</exclude-pattern>
 		<exclude-pattern>/src/wp-includes/class-wp-oembed\.php</exclude-pattern>
@@ -328,6 +329,7 @@
 	<!-- Exclude the unit tests from the file name rules. -->
 	<rule ref="WordPress.Files.FileName">
 		<exclude-pattern>/tests/phpunit/*</exclude-pattern>
+		<exclude-pattern>/tests/benchmarks/benchmarks/*</exclude-pattern>
 	</rule>
 
 	<!-- WPCS1620: template.php isn't a template tag file. -->

tests/benchmarks/benchmarks/html-api/WpHtmlTagProcessorBench.php

@@ -0,0 +1,177 @@
+<?php
+/**
+ * Benchmarks for WP_HTML_Tag_Processor.
+ *
+ * @package WordPress
+ */
+
+declare(strict_types=1);
+
+use PhpBench\Attributes as Bench;
+
+class WpHtmlTagProcessorBench {
+	/**
+	 * Processor instance for benchmarks.
+	 */
+	private WP_HTML_Tag_Processor|WP_HTML_Processor|null $processor = null;
+
+	public function clean_up_processor(): void {
+		$this->processor = null;
+	}
+
+	/**
+	 * Benchmark normalizing simple Unix paths.
+	 * @param array{0: string} $params
+	 */
+	#[Bench\Warmup( 2 )]
+	#[Bench\Iterations( 50 )]
+	#[Bench\Revs( 10 )]
+	#[Bench\BeforeMethods( 'set_up_script_tag_processor' )]
+	#[Bench\AfterMethods( 'clean_up_processor' )]
+	#[Bench\ParamProviders( 'provide_script_tag_contents' )]
+	public function bench_javascript_custom_escape( array $params ): void {
+		[ $source_text] = $params;
+		assert( $this->processor->set_modifiable_text( $source_text ), 'Failed to set modifiable text.' );
+	}
+
+	public function set_up_script_tag_processor(): void {
+		$this->processor = new WP_HTML_Tag_Processor( '<script></script>' );
+		$this->processor->next_tag();
+	}
+
+	/**
+	 * @return iterable<array{0: string}>
+	 */
+	public static function provide_script_tag_contents(): iterable {
+		yield 'empty' => array( '' );
+
+		yield 'short' => array( 'console.log("Hello, World!");' );
+
+		yield 'many replacements' => array(
+			<<<'JS'
+			/* <!-- and <script> is bad news in JS land! */
+			const templateString = `
+				But can't we talk about <script> and </script> tags without breaking everything?
+				</script>
+				</SCRIPT>
+				</SCRIPT	>
+				</SCRIPT
+				>
+				</SCRIPT/>
+				</script/>
+				</script ignored attributes />
+				</script	/>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+				<script></script>
+			`;
+			/* Good luck! */
+			JS,
+		);
+	}
+
+	/**
+	 * Benchmark HTML parsing.
+	 * @param array{0: string} $params
+	 */
+	#[Bench\Warmup( 2 )]
+	#[Bench\Iterations( 20 )]
+	#[Bench\Revs( 5 )]
+	#[Bench\ParamProviders( 'provide_html' )]
+	#[Bench\BeforeMethods( 'set_up_html_tag_processor' )]
+	#[Bench\AfterMethods( 'clean_up_processor' )]
+	public function bench_tag_processor_html_parsing( array $params ): void {
+		while ( $this->processor->next_token() ) {
+			// No-op.
+		}
+	}
+
+	public function set_up_html_tag_processor( array $params ): void {
+		$this->processor = new WP_HTML_Tag_Processor( file_get_contents( $params[0] ) );
+	}
+
+	/**
+	 * Benchmark HTML parsing.
+	 * @param array{0: string} $params
+	 */
+	#[Bench\Warmup( 2 )]
+	#[Bench\Iterations( 20 )]
+	#[Bench\Revs( 3 )]
+	#[Bench\ParamProviders( 'provide_html' )]
+	#[Bench\BeforeMethods( 'set_up_html_fragment_processor' )]
+	#[Bench\AfterMethods( 'clean_up_processor' )]
+	public function bench_html_fragment_parsing( array $params ): void {
+		while ( $this->processor->next_token() ) {
+			// No-op.
+		}
+	}
+
+	public function set_up_html_fragment_processor( array $params ): void {
+		$this->processor = WP_HTML_Processor::create_fragment( file_get_contents( $params[0] ) );
+	}
+
+	/**
+	 * Benchmark HTML parsing.
+	 * @param array{0: string} $params
+	 */
+	#[Bench\Warmup( 2 )]
+	#[Bench\Iterations( 20 )]
+	#[Bench\Revs( 3 )]
+	#[Bench\ParamProviders( 'provide_html' )]
+	#[Bench\BeforeMethods( 'set_up_html_full_parser' )]
+	#[Bench\AfterMethods( 'clean_up_processor' )]
+	public function bench_html_full_parsing( array $params ): void {
+		while ( $this->processor->next_token() ) {
+			// No-op.
+		}
+	}
+
+	public function set_up_html_full_parser( array $params ): void {
+		$this->processor = WP_HTML_Processor::create_full_parser( file_get_contents( $params[0] ) );
+	}
+
+	public static function provide_html(): iterable {
+		yield 'Empty string' => array( 'data:text/html,' );
+		yield 'Short doc' => array( 'data:text/html,' . rawurlencode( '<h1>Hello, world!</h1>' ) );
+		yield 'HTML Standard' => array( 'file://' . DIR_BENCHMARKDATA . '/html-standard.html' );
+	}
+}