SiteContentDetector.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 4.14.1

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 3 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 3 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 3 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 3 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 3 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

423 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	use Piwik\Plugins\SitesManager\GtmSiteTypeGuesser;
16	use Piwik\Plugins\SitesManager\SitesManager;
17
18	/**
19	* This class provides detection functions for specific content on a site. It can be used to easily detect the
20	* presence of known third party code.
21	*
22	* Note: Calling the detect() method will create a HTTP request to the site to retrieve data, only the main site URL
23	* will be checked
24	*
25	* Usage:
26	*
27	* $contentDetector = new SiteContentDetector();
28	* $contentDetector->detectContent([SiteContentDetector::GA3]);
29	* if ($contentDetector->ga3) {
30	* // site is using GA3
31	* }
32	*
33	*/
34	class SiteContentDetector
35	{
36	// Content types
37	const ALL_CONTENT = 1;
38	const CONSENT_MANAGER = 2;
39	const GA3 = 3;
40	const GA4 = 4;
41	const GTM = 5;
42	const CMS = 6;
43
44	// Detection detail
45	public $consentManagerId; // Id of the detected consent manager, eg. 'osano'
46	public $consentManagerName; // Display name of the detected consent manager, eg. 'Osano'
47	public $consentManagerUrl; // Url for the configuration guide for the detected consent manager
48	public $isConnected = false; // True if the detected consent manager is already connected with Matomo
49	public $ga3; // True if GA3 was detected on the site
50	public $ga4; // True if GA4 was detected on the site
51	public $gtm; // True if GTM was detected on the site
52	public $cms; // The CMS that was detected on the site
53
54	private $siteResponse = [
55	'data' => '',
56	'headers' => []
57	];
58
59	/** @var Lazy */
60	private $cache;
61
62	/**
63	* @var GtmSiteTypeGuesser
64	*/
65	private $siteGuesser;
66
67	public function __construct(?Lazy $cache = null)
68	{
69	if ($cache === null) {
70	$this->cache = Cache::getLazyCache();
71	} else {
72	$this->cache = $cache;
73	}
74	$this->siteGuesser = new GtmSiteTypeGuesser();
75	}
76
77	/**
78	* Reset the detection properties
79	*
80	* @return void
81	*/
82	private function resetDetectionProperties(): void
83	{
84	$this->consentManagerId = null;
85	$this->consentManagerUrl = null;
86	$this->consentManagerName = null;
87	$this->isConnected = false;
88	$this->ga3 = false;
89	$this->ga4 = false;
90	$this->gtm = false;
91	$this->cms = SitesManager::SITE_TYPE_UNKNOWN;
92	}
93
94	/**
95	* This will query the site and populate the class properties with
96	* the details of the detected content
97	*
98	* @param array $detectContent Array of content type for which to check, defaults to all, limiting this list
99	* will speed up the detection check
100	* @param ?int $idSite Override the site ID, will use the site from the current request if null
101	* @param ?array $siteResponse String containing the site data to search, if blank then data will be retrieved
102	* from the current request site via an http request
103	* @param int $timeOut How long to wait for the site to response, defaults to 5 seconds
104	* @return void
105	*/
106	public function detectContent(array $detectContent = [SiteContentDetector::ALL_CONTENT],
107	?int $idSite = null, ?array $siteResponse = null, int $timeOut = 5): void
108	{
109	$this->resetDetectionProperties();
110
111	// If site data was passed in, then just run the detection checks against it and return.
112	if ($siteResponse) {
113	$this->siteResponse = $siteResponse;
114	$this->detectionChecks($detectContent);
115	return;
116	}
117
118	// Get the site id from the request object if not explicitly passed
119	if ($idSite === null) {
120	if (!isset($_REQUEST['idSite'])) {
121	return;
122	}
123
124	$idSite = Common::getRequestVar('idSite', null, 'int');
125
126	if (!$idSite) {
127	return;
128	}
129	}
130
131	$url = Site::getMainUrlFor($idSite);
132
133	// Check and load previously cached site content detection data if it exists
134	$cacheKey = 'SiteContentDetector_' . md5($url);
135	$requiredProperties = $this->getRequiredProperties($detectContent);
136	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
137
138	if ($siteContentDetectionCache !== false) {
139	if ($this->checkCacheHasRequiredProperties($requiredProperties, $siteContentDetectionCache)) {
140	$this->loadRequiredPropertiesFromCache($requiredProperties, $siteContentDetectionCache);
141	return;
142	}
143	}
144
145	// No cache hit, no passed data, so make a request for the site content
146	$siteResponse = $this->requestSiteResponse($url, $timeOut);
147
148	// Abort if still no site data
149	if (empty($siteResponse['data'])) {
150	return;
151	}
152
153	$this->siteResponse = $siteResponse;
154
155	// We now have site data to analyze, so run the detection checks
156	$this->detectionChecks($detectContent);
157
158	// A request was made to get this data and it isn't currently cached, so write it to the cache now
159	$cacheLife = (60 * 60 * 24 * 7);
160	$this->savePropertiesToCache($cacheKey, $requiredProperties, $cacheLife);
161	}
162
163	/**
164	* Returns an array of properties required by the detect content array
165	*
166	* @param array $detectContent
167	*
168	* @return array
169	*/
170	private function getRequiredProperties(array $detectContent): array
171	{
172	$requiredProperties = [];
173
174	if (in_array(SiteContentDetector::CONSENT_MANAGER, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
175	$requiredProperties = array_merge($requiredProperties, ['consentManagerId', 'consentManagerName', 'consentManagerUrl', 'isConnected']);
176	}
177
178	if (in_array(SiteContentDetector::GA3, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
179	$requiredProperties[] = 'ga3';
180	}
181
182	if (in_array(SiteContentDetector::GA4, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
183	$requiredProperties[] = 'ga4';
184	}
185
186	if (in_array(SiteContentDetector::GTM, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
187	$requiredProperties[] = 'gtm';
188	}
189
190	if (in_array(SiteContentDetector::CMS, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
191	$requiredProperties[] = 'cms';
192	}
193
194	return $requiredProperties;
195	}
196
197	/**
198	* Checks that all required properties are in the cache array
199	*
200	* @param array $properties
201	* @param array $cache
202	*
203	* @return bool
204	*/
205	private function checkCacheHasRequiredProperties(array $properties, array $cache): bool
206	{
207	foreach ($properties as $prop) {
208	if (!array_key_exists($prop, $cache)) {
209	return false;
210	}
211	}
212
213	return true;
214	}
215
216	/**
217	* Load object properties from the cache array
218	*
219	* @param array $properties
220	* @param array $cache
221	*
222	* @return void
223	*/
224	private function loadRequiredPropertiesFromCache(array $properties, array $cache): void
225	{
226	foreach ($properties as $prop) {
227	if (!array_key_exists($prop, $cache)) {
228	continue;
229	}
230
231	$this->{$prop} = $cache[$prop];
232	}
233	}
234
235	/**
236	* Save properties to the cache
237	*
238	* @param string $cacheKey
239	* @param array $properties
240	* @param int $cacheLife
241	*
242	* @return void
243	*/
244	private function savePropertiesToCache(string $cacheKey, array $properties, int $cacheLife): void
245	{
246	$cacheData = [];
247
248	// Load any existing cached values
249	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
250
251	if (is_array($siteContentDetectionCache)) {
252	$cacheData = $siteContentDetectionCache;
253	}
254
255	foreach ($properties as $prop) {
256	$cacheData[$prop] = $this->{$prop};
257	}
258
259	$this->cache->save($cacheKey, $cacheData, $cacheLife);
260	}
261
262	/**
263	* Run various detection checks for site content
264	*
265	* @param array $detectContent Array of detection types used to filter the checks that are run
266	*
267	* @return void
268	*/
269	private function detectionChecks($detectContent): void
270	{
271	if (in_array(SiteContentDetector::CONSENT_MANAGER, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
272	$this->detectConsentManager();
273	}
274
275	if (in_array(SiteContentDetector::GA3, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
276	$this->ga3 = $this->siteGuesser->detectGA3FromResponse($this->siteResponse);
277	}
278
279	if (in_array(SiteContentDetector::GA4, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
280	$this->ga4 = $this->siteGuesser->detectGA4FromResponse($this->siteResponse);
281	}
282
283	if (in_array(SiteContentDetector::GTM, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
284	$this->gtm = $this->siteGuesser->guessGtmFromResponse($this->siteResponse);
285	}
286
287	if (in_array(SiteContentDetector::CMS, $detectContent) \|\| in_array(SiteContentDetector::ALL_CONTENT, $detectContent)) {
288	$this->cms = $this->siteGuesser->guessSiteTypeFromResponse($this->siteResponse);
289	}
290	}
291
292	/**
293	* Retrieve data from the specified site using an HTTP request
294	*
295	* @param string $url
296	* @param int $timeOut
297	*
298	* @return array
299	*/
300	private function requestSiteResponse(string $url, int $timeOut): array
301	{
302	if (!$url) {
303	return [];
304	}
305
306	// If internet features are disabled, we don't try to fetch any site content
307	if (0 === (int) GeneralConfig::getConfigValue('enable_internet_features')) {
308	return [];
309	}
310
311	$siteData = [];
312
313	try {
314	$siteData = Http::sendHttpRequestBy(Http::getTransportMethod(), $url, $timeOut, null, null,
315	null, 0, false, true, false, true);
316	} catch (\Exception $e) {
317	}
318
319	return $siteData;
320	}
321
322	/**
323	* Detect known consent managers in the site data
324	*
325	* Populate this object's properties with the results
326	*
327	* @return void
328	*/
329	private function detectConsentManager(): void
330	{
331	$defs = self::getConsentManagerDefinitions();
332
333	if (!$defs) {
334	return;
335	}
336
337	if (empty($this->siteResponse['data'])) {
338	return;
339	}
340
341	foreach ($defs as $consentManagerId => $consentManagerDef) {
342	foreach ($consentManagerDef['detectStrings'] as $dStr) {
343	if (strpos($this->siteResponse['data'], $dStr) !== false && array_key_exists($consentManagerId, $defs)) {
344	$this->consentManagerId = $consentManagerId;
345	$this->consentManagerName = $consentManagerDef['name'];
346	$this->consentManagerUrl = $consentManagerDef['url'];
347	break 2;
348	}
349	}
350	}
351
352	if (!isset($defs[$this->consentManagerId]['connectedStrings'])) {
353	return;
354	}
355
356	// If a consent manager was detected then perform an additional check to see if it has been connected to Matomo
357	foreach ($defs[$this->consentManagerId]['connectedStrings'] as $cStr) {
358	if (strpos($this->siteResponse['data'], $cStr) !== false) {
359	$this->isConnected = true;
360	break;
361	}
362	}
363	}
364
365	/**
366	* Return an array of consent manager definitions which can be used to detect their presence on the site and show
367	* the associated guide links
368	*
369	* @return array[]
370	*/
371	public static function getConsentManagerDefinitions(): array
372	{
373	return [
374
375	'osano' => [
376	'name' => 'Osano',
377	'detectStrings' => ['osano.com'],
378	'connectedStrings' => ["Osano.cm.addEventListener('osano-cm-consent-changed', (change) => { console.log('cm-change'); consentSet(change); });"],
379	'url' => 'https://matomo.org/faq/how-to/using-osano-consent-manager-with-matomo',
380	],
381
382	'cookiebot' => [
383	'name' => 'Cookiebot',
384	'detectStrings' => ['cookiebot.com'],
385	'connectedStrings' => ["typeof _paq === 'undefined' \|\| typeof Cookiebot === 'undefined'"],
386	'url' => 'https://matomo.org/faq/how-to/using-cookiebot-consent-manager-with-matomo',
387	],
388
389	'cookieyes' => [
390	'name' => 'CookieYes',
391	'detectStrings' => ['cookieyes.com'],
392	'connectedStrings' => ['document.addEventListener("cookieyes_consent_update", function (eventData)'],
393	'url' => 'https://matomo.org/faq/how-to/using-cookieyes-consent-manager-with-matomo',
394	],
395
396	// Note: tarte au citron pro is configured server side so we cannot tell if it has been connected by
397	// crawling the website, however setup of Matomo with the pro version is automatic, so displaying the guide
398	// link for pro isn't necessary. Only the open source version is detected by this definition.
399	'tarteaucitron' => [
400	'name' => 'Tarte au Citron',
401	'detectStrings' => ['tarteaucitron.js'],
402	'connectedStrings' => ['tarteaucitron.user.matomoHost'],
403	'url' => 'https://matomo.org/faq/how-to/using-tarte-au-citron-consent-manager-with-matomo',
404	],
405
406	'klaro' => [
407	'name' => 'Klaro',
408	'detectStrings' => ['klaro.js', 'kiprotect.com'],
409	'connectedStrings' => ['KlaroWatcher()', "title: 'Matomo',"],
410	'url' => 'https://matomo.org/faq/how-to/using-klaro-consent-manager-with-matomo',
411	],
412
413	'complianz' => [
414	'name' => 'Complianz',
415	'detectStrings' => ['complianz-gdpr'],
416	'connectedStrings' => ["if (!cmplz_in_array( 'statistics', consentedCategories )) {
417	_paq.push(['forgetCookieConsentGiven']);"],
418	'url' => 'https://matomo.org/faq/how-to/using-complianz-for-wordpress-consent-manager-with-matomo',
419	],
420	];
421	}
422	}
423