AIBotTracking.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 5.12.1

matomo / classes / WpMatomo / AIBotTracking.php

matomo / classes / WpMatomo Last commit date

Admin 1 week ago Commands 2 years ago Db 3 weeks ago Ecommerce 1 week ago Report 1 week ago Site 1 week ago TrackingCode 1 week ago Updater 4 years ago User 1 week ago Workarounds 2 years ago WpStatistics 1 week ago views 1 week ago AIBotTracking.php 1 week ago API.php 3 weeks ago Access.php 1 week ago AjaxTracker.php 5 months ago Annotations.php 1 week ago Bootstrap.php 11 months ago Capabilities.php 1 week ago Compatibility.php 1 week ago Email.php 1 week ago ErrorNotice.php 1 week ago Feature.php 3 months ago Installer.php 1 week ago Logger.php 1 year ago MinimumRequirementsNotice.php 1 month ago OptOut.php 2 months ago Paths.php 1 week ago PluginActionLinks.php 3 months ago PluginAdminOverrides.php 1 week ago PluginInit.php 1 week ago PrivacyBadge.php 4 years ago RedirectOnActivation.php 3 months ago Referral.php 3 months ago Roles.php 3 months ago ScheduledTasks.php 1 week ago Settings.php 1 week ago Site.php 3 years ago TrackingCode.php 1 week ago Uninstaller.php 6 months ago Updater.php 1 week ago User.php 3 weeks ago

AIBotTracking.php

232 lines

1	<?php
2	/**
3	* Matomo - free/libre analytics platform
4	*
5	* @link https://matomo.org
6	* @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later
7	* @package matomo
8	*/
9
10	namespace WpMatomo;
11
12	use LiteSpeed\ESI;
13
14	/**
15	* Performs server side tracking for AI bots.
16	*
17	* Normal server side tracking: when no caching plugins or CDNs are being
18	* used, this class will send tracking requests in the wp_footer hook.
19	*
20	* When advanced-cache.php is used: when a caching plugin creates an
21	* advanced-cache.php file that WordPress uses, tracking must be done
22	* through the standalone script in misc/track_ai_bot.php. This script
23	* must be manually added to a user's wp-config.php file, right after
24	* ABSPATH is defined.
25	*
26	* When .htaccess is used: when a caching plugin modifies the .htaccess
27	* to serve cached files directly, AI bot tracking is not possible.
28	*
29	* When a CDN is used: when a CDN is used to serve cached content, AI
30	* bot tracking can be accomplished through the use of ESI (Edge Side Includes).
31	* In this case, this class outputs an `<esi:include>` directive that
32	* loads the misc/track_ai_bot.php script. This technique will only work
33	* for CDNs that support ESI.
34	*
35	* The misc/track_ai_bot.php script uses this class without all of WordPress loaded.
36	* It can also be loaded outside of WordPress. Because of this, it is important
37	* that the script and this class use as few total dependencies as possible. Otherwise,
38	* AI bot tracking reduce the performance of requests to cached content.
39	*/
40	class AIBotTracking {
41
42	private static $ai_bot_tracked = false;
43
44	private static $extensions_to_track = [
45	'', // no extension
46
47	'htm',
48	'html',
49	'php',
50	];
51
52	/**
53	* @var Settings
54	*/
55	private $settings;
56
57	/**
58	* @var AjaxTracker
59	*/
60	private $tracker;
61
62	/**
63	* @param Settings $settings
64	* @param ?AIBotTracking $tracker
65	*/
66	public function __construct( $settings, $tracker = null ) {
67	$this->settings = $settings;
68	$this->tracker = $tracker;
69	}
70
71	public function register_hooks() {
72	add_action( 'litespeed_init', [ $this, 'litespeed_init' ] );
73	add_action( 'wp_footer', [ $this, 'do_ai_bot_tracking' ], 999999 );
74	}
75
76	public function litespeed_init() {
77	if ( class_exists( ESI::class )
78	&& $this->settings->is_tracking_ai_bots_via_esi_includes()
79	) {
80	ESI::set_has_esi();
81	}
82	}
83
84	public function do_ai_bot_tracking( $url = null ) {
85	// track AI bots only once per request
86	if ( self::$ai_bot_tracked ) {
87	return;
88	}
89
90	self::$ai_bot_tracked = true;
91
92	// if using ESI to track, and not within the track_ai_bot.php script, output the appropriate ESI tag
93	$is_using_esi_to_track = $this->settings->is_tracking_ai_bots_via_esi_includes();
94	if (
95	$is_using_esi_to_track
96	&& empty( $GLOBALS['MATOMO_IN_AI_ESI'] )
97	) {
98	$track_script_url = plugins_url( '/misc/track_ai_bot.php', MATOMO_ANALYTICS_FILE ) . '?mtm_esi=1&mtm_url=' . rawurlencode( AjaxTracker::getCurrentUrl() );
99	echo '<esi:include src="' . esc_attr( $track_script_url ) . '" cache-control="no-cache" />';
100	return;
101	}
102
103	if ( ! $this->is_doing_ai_bot_tracking_this_request() ) {
104	return;
105	}
106
107	$response_code = http_response_code();
108	$request_elapsed_ms = null;
109
110	if (
111	empty( $GLOBALS['MATOMO_IN_AI_ESI'] )
112	&& array_key_exists( 'REQUEST_TIME_FLOAT', $_SERVER )
113	) {
114	// phpcs:disable WordPress.Security.ValidatedSanitizedInput.InputNotSanitized
115	// phpcs:disable WordPress.Security.ValidatedSanitizedInput.MissingUnslash
116	$request_elapsed_ms = (int) ( ( microtime( true ) - $_SERVER['REQUEST_TIME_FLOAT'] ) * 1000 );
117	}
118
119	if ( empty( $response_code ) ) {
120	$response_code = 200;
121	}
122
123	// phpcs:ignore WordPress.WP.CapitalPDangit.MisspelledInText
124	$source = 'wordpress';
125
126	if ( empty( $url ) ) {
127	$url = AjaxTracker::getCurrentUrl();
128	}
129
130	$tracker = $this->get_tracker();
131	$tracker->setUrl( $url );
132
133	// cannot count bytes echo'd so no response size tracked
134	$tracker->doTrackPageViewIfAIBot( $response_code, null, $request_elapsed_ms, $source );
135	}
136
137	public function should_track_current_page() {
138	if ( is_admin() ) {
139	return false;
140	}
141
142	if ( defined( 'REST_REQUEST' ) && REST_REQUEST ) {
143	return false;
144	}
145
146	if ( defined( 'DOING_AJAX' ) && DOING_AJAX ) {
147	return false;
148	}
149
150	if (
151	! array_key_exists( 'REQUEST_URI', $_SERVER )
152	\|\| null === $_SERVER['REQUEST_URI']
153	) {
154	return false;
155	}
156
157	// phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized
158	$request_path = (string) wp_parse_url( wp_unslash( $_SERVER['REQUEST_URI'] ), PHP_URL_PATH );
159
160	if ( preg_match( '/matomo\.php$/', $request_path ) ) {
161	return false;
162	}
163
164	if ( $this->is_request_for_file( $request_path ) ) {
165	return false;
166	}
167
168	return true;
169	}
170
171	private function is_request_for_file( $request_path ) {
172	if (
173	! empty( $_SERVER['DOCUMENT_ROOT'] )
174	// phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized
175	&& is_dir( wp_unslash( $_SERVER['DOCUMENT_ROOT'] ) . $request_path )
176	) {
177	return false;
178	}
179
180	$extension = pathinfo( $request_path, PATHINFO_EXTENSION );
181	return ! in_array( $extension, self::$extensions_to_track, true );
182	}
183
184	public static function set_is_ai_bot_tracked( $is_tracked ) {
185	self::$ai_bot_tracked = $is_tracked;
186	}
187
188	public function is_js_execution_detected() {
189	return ! empty( $_COOKIE['matomo_has_js'] )
190	&& '1' === $_COOKIE['matomo_has_js'];
191	}
192
193	private function is_doing_ai_bot_tracking_this_request() {
194	if ( ! empty( $GLOBALS['MATOMO_IN_AI_ESI'] ) ) {
195	return true;
196	}
197
198	if ( ! $this->should_track_current_page() ) {
199	return false;
200	}
201
202	if ( $this->is_js_execution_detected() ) {
203	return false;
204	}
205
206	$tracker = $this->get_tracker();
207
208	// phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
209	if ( ! AjaxTracker::isUserAgentAIBot( $tracker->userAgent ) ) {
210	return false;
211	}
212
213	if (
214	! $this->settings->is_ai_bot_tracking_enabled()
215	\|\| ! $this->settings->is_tracking_enabled()
216	) {
217	return false;
218	}
219
220	return true;
221	}
222
223	private function get_tracker() {
224	if ( empty( $this->tracker ) ) {
225	$this->tracker = new AjaxTracker( $this->settings );
226	$this->tracker->setRequestTimeout( 1 );
227	}
228
229	return $this->tracker;
230	}
231	}
232