robotstxt-telemetry/includes/class-robotstxt-telemetry.php

<?php
/**
 * Main plugin bootstrap.
 *
 * @package RobotstxtTelemetry
 */

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

if ( ! class_exists( 'Robotstxt_Telemetry' ) ) {
	/**
	 * Class Robotstxt_Telemetry
	 */
	class Robotstxt_Telemetry {

		/**
		 * Initialize the plugin.
		 */
		public static function init() {
			$instance = new self();
			$instance->register_hooks();
		}

		/**
		 * Register hooks.
		 */
		private function register_hooks() {
			add_action( 'plugins_loaded', array( $this, 'load_textdomain' ) );
			add_action( 'init', array( 'Robotstxt_Telemetry_DB', 'maybe_upgrade' ) );
			add_action( 'init', array( $this, 'schedule_cleanup' ) );
			add_action( 'robotstxt_telemetry_cleanup', array( 'Robotstxt_Telemetry_DB', 'cleanup' ) );
			add_filter( 'http_request_args', array( $this, 'filter_useragent' ), 9, 2 );

			$wordpress_api = new Robotstxt_Telemetry_WordPress_Api();
			$wordpress_api->register();

			$plugin_guards = new Robotstxt_Telemetry_Plugin_Guards();
			$plugin_guards->register();

			$endpoints = new Robotstxt_Telemetry_Endpoints();
			$endpoints->register();

			$logger = new Robotstxt_Telemetry_Logger();
			$logger->register();

			if ( is_admin() ) {
				$notice = new Robotstxt_Telemetry_Manager_Notice();
				$notice->register();

				$admin = new Robotstxt_Telemetry_Admin();
				$admin->register();
			}
		}

		/**
		 * Filter the outbound User-Agent.
		 *
		 * The default outbound User-Agent is "WordPress/<version>; <site URL>",
		 * and the WordPress.org update checks send an explicit one with the
		 * same shape. Depending on the settings, the site URL is sent as-is,
		 * replaced with a deterministic SHA-256 hash, or removed, and the
		 * WordPress version can be reduced to its major version ("7.2.n")
		 * or nulled ("0.0.0").
		 *
		 * Runs at priority 9 so the telemetry logger records the value that
		 * is actually sent on the wire.
		 *
		 * @param array  $args HTTP request arguments.
		 * @param string $url  Request URL.
		 * @return array
		 */
		public function filter_useragent( $args, $url ) { // phpcs:ignore Generic.CodeAnalysis.UnusedFunctionParameter -- Second argument required by the http_request_args filter signature.
			$url_mode     = Robotstxt_Telemetry_Network::get_useragent_url_mode();
			$version_mode = Robotstxt_Telemetry_Network::get_wp_version_mode();

			if ( 'url' === $url_mode && 'actual' === $version_mode ) {
				return $args;
			}

			$site_url = get_bloginfo( 'url' );

			if ( isset( $args['user-agent'] ) && is_string( $args['user-agent'] ) ) {
				$args['user-agent'] = $this->mask_user_agent( $args['user-agent'], $url_mode, $version_mode, $site_url );
			}

			foreach ( array( 'User-Agent', 'user-agent' ) as $header ) {
				if ( isset( $args['headers'][ $header ] ) && is_string( $args['headers'][ $header ] ) ) {
					$args['headers'][ $header ] = $this->mask_user_agent( $args['headers'][ $header ], $url_mode, $version_mode, $site_url );
				}
			}

			return $args;
		}

		/**
		 * Apply the URL and version masking to a User-Agent string.
		 *
		 * @param string $user_agent   User-Agent value.
		 * @param string $url_mode     'url', 'hash', or 'none'.
		 * @param string $version_mode 'actual', 'major', or 'nulled'.
		 * @param string $site_url     Site URL.
		 * @return string
		 */
		private function mask_user_agent( $user_agent, $url_mode, $version_mode, $site_url ) {
			// Core sends the site URL with (update checks) and without (HTTP
			// API) a trailing slash; the slashed form must be replaced first
			// so its slash does not leak, then the bare form.
			$bare      = untrailingslashit( $site_url );
			$hash      = substr( hash( 'sha256', $bare ), 0, 16 );
			$url_forms = array( $bare . '/', $bare );
			$url_forms = array_unique( $url_forms );

			foreach ( $url_forms as $url_form ) {
				if ( false === strpos( $user_agent, $url_form ) ) {
					continue;
				}

				if ( 'hash' === $url_mode ) {
					$user_agent = str_replace( $url_form, $hash, $user_agent );
				} elseif ( 'none' === $url_mode ) {
					$user_agent = str_replace( '; ' . $url_form, '', $user_agent );
					$user_agent = str_replace( $url_form, '', $user_agent );
				}
			}

			$user_agent = trim( $user_agent );

			if ( 'actual' !== $version_mode && preg_match( '#^WordPress/(\d+[^\s;]*)#', $user_agent, $matches ) ) {
				$user_agent = preg_replace(
					'#^WordPress/[^\s;]+#',
					'WordPress/' . Robotstxt_Telemetry_Network::mask_wp_version( $matches[1], $version_mode ),
					$user_agent
				);
			}

			return $user_agent;
		}
		/**
		 * Schedule the log cleanup event if it is not scheduled yet.
		 *
		 * In global mode the shared table is cleaned from the main site
		 * only, so subsites do not schedule the event.
		 */
		public function schedule_cleanup() {
			if ( Robotstxt_Telemetry_Network::is_global() && ! is_main_site() ) {
				return;
			}

			if ( ! wp_next_scheduled( 'robotstxt_telemetry_cleanup' ) ) {
				wp_schedule_event( time() + HOUR_IN_SECONDS, 'twicedaily', 'robotstxt_telemetry_cleanup' );
			}
		}

		/**
		 * Deactivation handler: clear the scheduled cleanup event.
		 */
		public static function deactivate() {
			wp_clear_scheduled_hook( 'robotstxt_telemetry_cleanup' );

			if ( is_multisite() ) {
				foreach ( Robotstxt_Telemetry_Network::get_site_ids() as $site_id ) {
					switch_to_blog( $site_id );
					wp_clear_scheduled_hook( 'robotstxt_telemetry_cleanup' );
					restore_current_blog();
				}
			}
		}

		/**
		 * Load plugin translations.
		 */
		public function load_textdomain() {
			load_plugin_textdomain(
				'robotstxt-telemetry',
				false,
				dirname( plugin_basename( ROBOTSTXT_TELEMETRY_PLUGIN_FILE ) ) . '/languages'
			);
		}
	}
}