SiteContentDetector.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 4.13.5

matomo / app / core / SiteContentDetector.php

matomo / app / core Last commit date

API 3 years ago Access 3 years ago Application 4 years ago Archive 3 years ago ArchiveProcessor 3 years ago Archiver 5 years ago AssetManager 3 years ago Auth 3 years ago Category 5 years ago Changes 4 years ago CliMulti 4 years ago Columns 3 years ago Concurrency 3 years ago Config 4 years ago Container 4 years ago CronArchive 3 years ago DataAccess 3 years ago DataFiles 5 years ago DataTable 3 years ago Db 3 years ago DeviceDetector 3 years ago Email 5 years ago Exception 3 years ago Http 4 years ago Intl 4 years ago Mail 3 years ago Measurable 5 years ago Menu 3 years ago Metrics 4 years ago Notification 4 years ago Period 4 years ago Plugin 3 years ago ProfessionalServices 4 years ago Report 5 years ago ReportRenderer 3 years ago Scheduler 4 years ago Segment 3 years ago Session 4 years ago Settings 3 years ago Tracker 3 years ago Translation 4 years ago UpdateCheck 5 years ago Updater 3 years ago Updates 3 years ago Validators 4 years ago View 4 years ago ViewDataTable 3 years ago Visualization 4 years ago Widget 5 years ago .htaccess 6 years ago Access.php 4 years ago Archive.php 4 years ago ArchiveProcessor.php 4 years ago AssetManager.php 4 years ago Auth.php 5 years ago AuthResult.php 5 years ago BaseFactory.php 5 years ago Cache.php 5 years ago CacheId.php 5 years ago CliMulti.php 4 years ago Common.php 3 years ago Config.php 3 years ago Console.php 3 years ago Context.php 5 years ago Cookie.php 3 years ago CronArchive.php 3 years ago DataArray.php 4 years ago DataTable.php 3 years ago Date.php 4 years ago Db.php 4 years ago DbHelper.php 3 years ago Development.php 5 years ago ErrorHandler.php 5 years ago EventDispatcher.php 3 years ago ExceptionHandler.php 3 years ago FileIntegrity.php 5 years ago Filechecks.php 4 years ago Filesystem.php 3 years ago FrontController.php 3 years ago Http.php 3 years ago IP.php 4 years ago Log.php 4 years ago LogDeleter.php 4 years ago Mail.php 4 years ago Metrics.php 3 years ago NoAccessException.php 5 years ago Nonce.php 3 years ago Notification.php 5 years ago NumberFormatter.php 4 years ago Option.php 4 years ago Period.php 5 years ago Piwik.php 3 years ago Plugin.php 4 years ago Profiler.php 4 years ago ProxyHeaders.php 5 years ago ProxyHttp.php 3 years ago QuickForm2.php 5 years ago RankingQuery.php 3 years ago ReportRenderer.php 4 years ago Segment.php 3 years ago Sequence.php 5 years ago Session.php 3 years ago SettingsPiwik.php 3 years ago SettingsServer.php 5 years ago Singleton.php 5 years ago Site.php 3 years ago SiteContentDetector.php 3 years ago SupportedBrowser.php 3 years ago TCPDF.php 5 years ago Theme.php 5 years ago Timer.php 5 years ago Tracker.php 4 years ago Twig.php 4 years ago Unzip.php 5 years ago UpdateCheck.php 5 years ago Updater.php 4 years ago UpdaterErrorException.php 5 years ago Updates.php 5 years ago Url.php 3 years ago UrlHelper.php 3 years ago Version.php 3 years ago View.php 3 years ago bootstrap.php 3 years ago dispatch.php 5 years ago testMinimumPhpVersion.php 3 years ago

SiteContentDetector.php

434 lines

1	<?php
2
3	/**
4	* Matomo - free/libre analytics platform
5	*
6	* @link https://matomo.org
7	* @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later
8	*
9	*/
10
11	namespace Piwik;
12
13	use Matomo\Cache\Lazy;
14	use Piwik\Config\GeneralConfig;
15
16	/**
17	* This class provides detection functions for specific content on a site. It can be used to easily detect the
18	* presence of known third party code.
19	*
20	* Note: Calling the detect() method will create a HTTP request to the site to retrieve data, only the main site URL
21	* will be checked
22	*
23	* Usage:
24	*
25	* $contentDetector = new SiteContentDetector();
26	* $contentDetector->detectContent([SiteContentDetector::GA3]);
27	* if ($contentDetector->ga3) {
28	* // site is using GA3
29	* }
30	*
31	*/
32	class SiteContentDetector
33	{
34	// Content types
35	const ALL_CONTENT = 1;
36	const CONSENT_MANAGER = 2;
37	const GA3 = 3;
38	const GA4 = 4;
39	const GTM = 5;
40
41	// Detection detail
42	public $consentManagerId; // Id of the detected consent manager, eg. 'osano'
43	public $consentManagerName; // Display name of the detected consent manager, eg. 'Osano'
44	public $consentManagerUrl; // Url for the configuration guide for the detected consent manager
45	public $isConnected = false; // True if the detected consent manager is already connected with Matomo
46	public $ga3; // True if GA3 was detected on the site
47	public $ga4; // True if GA4 was detected on the site
48	public $gtm; // True if GTM was detected on the site
49
50	private $siteData;
51
52	/** @var Lazy */
53	private $cache;
54
55	public function __construct(?Lazy $cache = null)
56	{
57	if ($cache === null) {
58	$this->cache = Cache::getLazyCache();
59	} else {
60	$this->cache = $cache;
61	}
62	}
63
64	/**
65	* Reset the detection properties
66	*
67	* @return void
68	*/
69	private function resetDetectionProperties(): void
70	{
71	$this->consentManagerId = null;
72	$this->consentManagerUrl = null;
73	$this->consentManagerName = null;
74	$this->isConnected = false;
75	$this->ga3 = false;
76	$this->ga4 = false;
77	$this->gtm = false;
78	}
79
80	/**
81	* This will query the site and populate the class properties with
82	* the details of the detected content
83	*
84	* @param array $detectContent Array of content type for which to check, defaults to all, limiting this list
85	* will speed up the detection check
86	* @param ?int $idSite Override the site ID, will use the site from the current request if null
87	* @param string\|null $siteData String containing the site data to search, if blank then data will be retrieved
88	* from the current request site via an http request
89	* @param int $timeOut How long to wait for the site to response, defaults to 5 seconds
90	* @return void
91	*/
92	public function detectContent(array $detectContent = [SiteContentDetector::ALL_CONTENT],
93	?int $idSite = null, ?string $siteData = null, int $timeOut = 5): void
94	{
95	$this->resetDetectionProperties();
96
97	// If site data was passed in, then just run the detection checks against it and return.
98	if ($siteData) {
99	$this->siteData = $siteData;
100	$this->detectionChecks($detectContent);
101	return;
102	}
103
104	// Get the site id from the request object if not explicitly passed
105	if ($idSite === null) {
106	if (!isset($_REQUEST['idSite'])) {
107	return;
108	}
109
110	$idSite = Common::getRequestVar('idSite', null, 'int');
111
112	if (!$idSite) {
113	return;
114	}
115	}
116
117	// Check and load previously cached site content detection data if it exists
118	$cacheKey = 'SiteContentDetector_' . $idSite;
119	$requiredProperties = $this->getRequiredProperties($detectContent);
120	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
121
122	if ($siteContentDetectionCache !== false) {
123	if ($this->checkCacheHasRequiredProperties($requiredProperties, $siteContentDetectionCache)) {
124	$this->loadRequiredPropertiesFromCache($requiredProperties, $siteContentDetectionCache);
125	return;
126	}
127	}
128
129	// No cache hit, no passed data, so make a request for the site content
130	$siteData = $this->requestSiteData($idSite, $timeOut);
131
132	// Abort if still no site data
133	if ($siteData === null \|\| $siteData === '') {
134	return;
135	}
136
137	$this->siteData = $siteData;
138
139	// We now have site data to analyze, so run the detection checks
140	$this->detectionChecks($detectContent);
141
142	// A request was made to get this data and it isn't currently cached, so write it to the cache now
143	$cacheLife = (60 * 60 * 24 * 7);
144	$this->savePropertiesToCache($cacheKey, $requiredProperties, $cacheLife);
145	}
146
147	/**
148	* Returns an array of properties required by the detect content array
149	*
150	* @param array $detectContent
151	*
152	* @return array
153	*/
154	private function getRequiredProperties(array $detectContent): array
155	{
156	$requiredProperties = [];
157
158	if (in_array(SiteContentDetector::CONSENT_MANAGER, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
159	$requiredProperties = array_merge($requiredProperties, ['consentManagerId', 'consentManagerName', 'consentManagerUrl', 'isConnected']);
160	}
161
162	if (in_array(SiteContentDetector::GA3, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
163	$requiredProperties[] = 'ga3';
164	}
165
166	if (in_array(SiteContentDetector::GA4, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
167	$requiredProperties[] = 'ga4';
168	}
169
170	if (in_array(SiteContentDetector::GTM, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
171	$requiredProperties[] = 'gtm';
172	}
173
174	return $requiredProperties;
175	}
176
177	/**
178	* Checks that all required properties are in the cache array
179	*
180	* @param array $properties
181	* @param array $cache
182	*
183	* @return bool
184	*/
185	private function checkCacheHasRequiredProperties(array $properties, array $cache): bool
186	{
187	foreach ($properties as $prop) {
188	if (!array_key_exists($prop, $cache)) {
189	return false;
190	}
191	}
192
193	return true;
194	}
195
196	/**
197	* Load object properties from the cache array
198	*
199	* @param array $properties
200	* @param array $cache
201	*
202	* @return void
203	*/
204	private function loadRequiredPropertiesFromCache(array $properties, array $cache): void
205	{
206	foreach ($properties as $prop) {
207	if (!array_key_exists($prop, $cache)) {
208	continue;
209	}
210
211	$this->{$prop} = $cache[$prop];
212	}
213	}
214
215	/**
216	* Save properties to the cache
217	*
218	* @param string $cacheKey
219	* @param array $properties
220	* @param int $cacheLife
221	*
222	* @return void
223	*/
224	private function savePropertiesToCache(string $cacheKey, array $properties, int $cacheLife): void
225	{
226
227	$cacheData = [];
228
229	// Load any existing cached values
230	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
231
232	if (is_array($siteContentDetectionCache)) {
233	$cacheData = $siteContentDetectionCache;
234	}
235
236	foreach ($properties as $prop) {
237	$cacheData[$prop] = $this->{$prop};
238	}
239
240	$this->cache->save($cacheKey, $cacheData, $cacheLife);
241	}
242
243	/**
244	* Run various detection checks for site content
245	*
246	* @param array $detectContent Array of detection types used to filter the checks that are run
247	*
248	* @return void
249	*/
250	private function detectionChecks($detectContent): void
251	{
252	if (in_array(SiteContentDetector::CONSENT_MANAGER, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
253	$this->detectConsentManager();
254	}
255
256	if (in_array(SiteContentDetector::GA3, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
257	$this->detectGA3();
258	}
259
260	if (in_array(SiteContentDetector::GA4, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
261	$this->detectGA4();
262	}
263
264	if (in_array(SiteContentDetector::GTM, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
265	$this->detectGTM();
266	}
267	}
268
269	/**
270	* Retrieve data from the specified site using an HTTP request
271	*
272	* @param int $idSite
273	* @param int $timeOut
274	*
275	* @return string\|null
276	*/
277	private function requestSiteData(int $idSite, int $timeOut): ?string
278	{
279	// If internet features are disabled, we don't try to fetch any site content
280	if (0 === (int) GeneralConfig::getConfigValue('enable_internet_features')) {
281	return null;
282	}
283
284	$url = Site::getMainUrlFor($idSite);
285
286	if (!$url) {
287	return null;
288	}
289
290	$siteData = null;
291
292	try {
293	$siteData = Http::sendHttpRequestBy(Http::getTransportMethod(), $url, $timeOut, null, null,
294	null, 0, false, true);
295	} catch (\Exception $e) {
296	}
297
298	return $siteData;
299	}
300
301	/**
302	* Detect known consent managers in the site data
303	*
304	* Populate this object's properties with the results
305	*
306	* @return void
307	*/
308	private function detectConsentManager(): void
309	{
310	$defs = SiteContentDetector::getConsentManagerDefinitions();
311
312	if (!$defs) {
313	return;
314	}
315
316	foreach ($defs as $consentManagerId => $consentManagerDef) {
317	foreach ($consentManagerDef['detectStrings'] as $dStr) {
318	if (strpos($this->siteData, $dStr) !== false && array_key_exists($consentManagerId, $defs)) {
319	$this->consentManagerId = $consentManagerId;
320	$this->consentManagerName = $consentManagerDef['name'];
321	$this->consentManagerUrl = $consentManagerDef['url'];
322	break 2;
323	}
324	}
325	}
326
327	if (!isset($defs[$this->consentManagerId]['connectedStrings'])) {
328	return;
329	}
330
331	// If a consent manager was detected then perform an additional check to see if it has been connected to Matomo
332	foreach ($defs[$this->consentManagerId]['connectedStrings'] as $cStr) {
333	if (strpos($this->siteData, $cStr) !== false) {
334	$this->isConnected = true;
335	break;
336	}
337	}
338	}
339
340	/**
341	* Detect GA3 usage from the site data
342	*
343	* @return void
344	*/
345	private function detectGA3(): void
346	{
347	if (strpos($this->siteData, '(i,s,o,g,r,a,m)') !== false) {
348	$this->ga3 = true;
349	}
350	}
351
352	/**
353	* Detect GA4 usage from the site data
354	*
355	* @return void
356	*/
357	private function detectGA4(): void
358	{
359	if (strpos($this->siteData, 'gtag.js') !== false) {
360	$this->ga4 = true;
361	}
362	}
363
364	/**
365	* Detect GTM usage from the site data
366	*
367	* @return void
368	*/
369	private function detectGTM(): void
370	{
371	if (strpos($this->siteData, 'gtm.js') !== false) {
372	$this->gtm = true;
373	}
374	}
375
376	/**
377	* Return an array of consent manager definitions which can be used to detect their presence on the site and show
378	* the associated guide links
379	*
380	* @return array[]
381	*/
382	public static function getConsentManagerDefinitions(): array
383	{
384	return [
385
386	'osano' => [
387	'name' => 'Osano',
388	'detectStrings' => ['osano.com'],
389	'connectedStrings' => ["Osano.cm.addEventListener('osano-cm-consent-changed', (change) => { console.log('cm-change'); consentSet(change); });"],
390	'url' => 'https://matomo.org/faq/how-to/using-osano-consent-manager-with-matomo',
391	],
392
393	'cookiebot' => [
394	'name' => 'Cookiebot',
395	'detectStrings' => ['cookiebot.com'],
396	'connectedStrings' => ["typeof _paq === 'undefined' \|\| typeof Cookiebot === 'undefined'"],
397	'url' => 'https://matomo.org/faq/how-to/using-cookiebot-consent-manager-with-matomo',
398	],
399
400	'cookieyes' => [
401	'name' => 'CookieYes',
402	'detectStrings' => ['cookieyes.com'],
403	'connectedStrings' => ['document.addEventListener("cookieyes_consent_update", function (eventData)'],
404	'url' => 'https://matomo.org/faq/how-to/using-cookieyes-consent-manager-with-matomo',
405	],
406
407	// Note: tarte au citron pro is configured server side so we cannot tell if it has been connected by
408	// crawling the website, however setup of Matomo with the pro version is automatic, so displaying the guide
409	// link for pro isn't necessary. Only the open source version is detected by this definition.
410	'tarteaucitron' => [
411	'name' => 'Tarte au Citron',
412	'detectStrings' => ['tarteaucitron.js'],
413	'connectedStrings' => ['tarteaucitron.user.matomoHost'],
414	'url' => 'https://matomo.org/faq/how-to/using-tarte-au-citron-consent-manager-with-matomo',
415	],
416
417	'klaro' => [
418	'name' => 'Klaro',
419	'detectStrings' => ['klaro.js', 'kiprotect.com'],
420	'connectedStrings' => ['KlaroWatcher()', "title: 'Matomo',"],
421	'url' => 'https://matomo.org/faq/how-to/using-klaro-consent-manager-with-matomo',
422	],
423
424	'complianz' => [
425	'name' => 'Complianz',
426	'detectStrings' => ['complianz-gdpr'],
427	'connectedStrings' => ["if (!cmplz_in_array( 'statistics', consentedCategories )) {
428	_paq.push(['forgetCookieConsentGiven']);"],
429	'url' => 'https://matomo.org/faq/how-to/using-complianz-for-wordpress-consent-manager-with-matomo',
430	],
431	];
432	}
433	}
434