SiteContentDetector.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 5.8.2

matomo / app / core / SiteContentDetector.php

matomo / app / core Last commit date

API 3 months ago Access 3 months ago Application 3 months ago Archive 3 months ago ArchiveProcessor 3 months ago Archiver 2 years ago AssetManager 3 months ago Auth 6 months ago Category 6 months ago Changes 3 months ago CliMulti 1 year ago Columns 3 months ago Concurrency 3 months ago Config 3 months ago Container 3 months ago CronArchive 3 months ago DataAccess 3 months ago DataFiles 2 years ago DataTable 3 months ago Db 3 months ago DeviceDetector 1 year ago Email 2 years ago Exception 4 months ago Http 4 months ago Intl 3 months ago Log 2 years ago Mail 1 year ago Measurable 6 months ago Menu 3 months ago Metrics 3 months ago Notification 6 months ago Period 3 months ago Plugin 3 months ago Policy 3 months ago ProfessionalServices 1 year ago Report 1 year ago ReportRenderer 3 months ago Request 3 months ago Scheduler 3 months ago Segment 3 months ago Session 3 months ago Settings 3 months ago Tracker 3 months ago Translation 3 months ago Twig 1 year ago UpdateCheck 3 months ago Updater 4 months ago Updates 3 months ago Validators 1 year ago View 6 months ago ViewDataTable 3 months ago Visualization 1 year ago Widget 3 months ago .htaccess 2 years ago Access.php 3 months ago Archive.php 3 months ago ArchiveProcessor.php 4 months ago AssetManager.php 3 months ago Auth.php 6 months ago AuthResult.php 6 months ago BaseFactory.php 2 years ago Cache.php 2 years ago CacheId.php 4 months ago CliMulti.php 3 months ago Common.php 3 months ago Config.php 3 months ago Console.php 3 months ago Context.php 2 years ago Cookie.php 1 year ago CronArchive.php 3 months ago DI.php 3 months ago DataArray.php 5 months ago DataTable.php 3 months ago Date.php 3 months ago Db.php 3 months ago DbHelper.php 3 months ago Development.php 1 year ago ErrorHandler.php 6 months ago EventDispatcher.php 1 year ago ExceptionHandler.php 4 months ago FileIntegrity.php 3 months ago Filechecks.php 1 year ago Filesystem.php 3 months ago FrontController.php 4 months ago Http.php 4 months ago IP.php 1 year ago Log.php 3 months ago LogDeleter.php 1 year ago Mail.php 1 year ago Metrics.php 3 months ago NoAccessException.php 2 years ago Nonce.php 6 months ago Notification.php 6 months ago NumberFormatter.php 5 months ago Option.php 5 months ago Period.php 3 months ago Piwik.php 3 months ago Plugin.php 3 months ago Process.php 1 year ago Profiler.php 6 months ago ProxyHeaders.php 4 months ago ProxyHttp.php 5 months ago QuickForm2.php 3 months ago RankingQuery.php 5 months ago ReportRenderer.php 3 months ago Request.php 3 months ago Segment.php 3 months ago Sequence.php 6 months ago Session.php 3 months ago SettingsPiwik.php 3 months ago SettingsServer.php 1 year ago Singleton.php 2 years ago Site.php 4 months ago SiteContentDetector.php 3 months ago SupportedBrowser.php 2 years ago TCPDF.php 1 year ago Theme.php 1 year ago Timer.php 2 years ago Tracker.php 3 months ago Twig.php 3 months ago Unzip.php 1 year ago UpdateCheck.php 3 months ago Updater.php 3 months ago UpdaterErrorException.php 2 years ago Updates.php 3 months ago Url.php 3 months ago UrlHelper.php 3 months ago Version.php 3 months ago View.php 3 months ago bootstrap.php 1 year ago dispatch.php 2 years ago testMinimumPhpVersion.php 6 months ago

SiteContentDetector.php

315 lines

1	<?php
2
3	/**
4	* Matomo - free/libre analytics platform
5	*
6	* @link https://matomo.org
7	* @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later
8	*/
9	namespace Piwik;
10
11	use Matomo\Cache\Lazy;
12	use Piwik\Config\GeneralConfig;
13	use Piwik\Container\StaticContainer;
14	use Piwik\Plugins\SitesManager\SiteContentDetection\ConsentManagerDetectionAbstract;
15	use Piwik\Plugins\SitesManager\SiteContentDetection\SiteContentDetectionAbstract;
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 `detectContent()` 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	* $contentDetector->detectContent([GoogleAnalytics3::getId()]);
27	* if ($contentDetector->ga3) {
28	* // site is using GA3
29	* }
30	*
31	* @api
32	*/
33	class SiteContentDetector
34	{
35	/**
36	* @var array<string, array<string, SiteContentDetectionAbstract>>
37	*/
38	public $detectedContent = [SiteContentDetectionAbstract::TYPE_TRACKER => [], SiteContentDetectionAbstract::TYPE_CMS => [], SiteContentDetectionAbstract::TYPE_JS_FRAMEWORK => [], SiteContentDetectionAbstract::TYPE_CONSENT_MANAGER => [], SiteContentDetectionAbstract::TYPE_JS_CRASH_ANALYTICS => [], SiteContentDetectionAbstract::TYPE_OTHER => []];
39	public $connectedConsentManagers = [];
40	private $siteResponse = ['data' => '', 'headers' => []];
41	/** @var Lazy */
42	private $cache;
43	public function __construct(?Lazy $cache = null)
44	{
45	if ($cache === null) {
46	$this->cache = \Piwik\Cache::getLazyCache();
47	} else {
48	$this->cache = $cache;
49	}
50	}
51	/**
52	* @return array<string, SiteContentDetectionAbstract[]>
53	*/
54	public static function getSiteContentDetectionsByType() : array
55	{
56	$instancesByType = [];
57	$classes = self::getAllSiteContentDetectionClasses();
58	foreach ($classes as $className) {
59	$instancesByType[$className::getContentType()][] = StaticContainer::get($className);
60	}
61	return $instancesByType;
62	}
63	/**
64	* Returns the site content detection object with the provided id, or null if it can't be found
65	*
66	*/
67	public function getSiteContentDetectionById(string $id) : ?SiteContentDetectionAbstract
68	{
69	$classes = $this->getAllSiteContentDetectionClasses();
70	foreach ($classes as $className) {
71	if ($className::getId() === $id) {
72	return StaticContainer::get($className);
73	}
74	}
75	return null;
76	}
77	/**
78	* @return string[]
79	*/
80	protected static function getAllSiteContentDetectionClasses() : array
81	{
82	return \Piwik\Plugin\Manager::getInstance()->findMultipleComponents('SiteContentDetection', SiteContentDetectionAbstract::class);
83	}
84	/**
85	* Reset the detections
86	*
87	*/
88	private function resetDetections() : void
89	{
90	$this->detectedContent = [SiteContentDetectionAbstract::TYPE_TRACKER => [], SiteContentDetectionAbstract::TYPE_CMS => [], SiteContentDetectionAbstract::TYPE_JS_FRAMEWORK => [], SiteContentDetectionAbstract::TYPE_CONSENT_MANAGER => [], SiteContentDetectionAbstract::TYPE_JS_CRASH_ANALYTICS => [], SiteContentDetectionAbstract::TYPE_OTHER => []];
91	$this->connectedConsentManagers = [];
92	}
93	/**
94	* This will query the site and populate the class properties with
95	* the details of the detected content
96	*
97	* @param array $detectContent Array of content type for which to check, defaults to all, limiting this list
98	* will speed up the detection check.
99	* Allowed values are:
100	* * empty array - to run all detections
101	* * an array containing ids of detections, e.g. Wordpress::getId() or any of the
102	* type constants, e.g. SiteContentDetectionAbstract::TYPE_TRACKER
103	* @param ?int $idSite Override the site ID, will use the site from the current request if null
104	* @param ?array $siteResponse String containing the site data to search, if blank then data will be retrieved
105	* from the current request site via an http request
106	* @param int $timeOut How long to wait for the site to response, defaults to 5 seconds
107	*/
108	public function detectContent(array $detectContent = [], ?int $idSite = null, ?array $siteResponse = null, int $timeOut = 5) : void
109	{
110	$this->resetDetections();
111	// If site data was passed in, then just run the detection checks against it and return.
112	if ($siteResponse) {
113	$this->siteResponse = $siteResponse;
114	$this->detectionChecks($detectContent);
115	return;
116	}
117	// Get the site id from the request object if not explicitly passed
118	if ($idSite === null) {
119	$idSite = \Piwik\Request::fromRequest()->getIntegerParameter('idSite', 0);
120	if (!$idSite) {
121	return;
122	}
123	}
124	$url = \Piwik\Site::getMainUrlFor($idSite);
125	// Check and load previously cached site content detection data if it exists
126	$cacheKey = 'SiteContentDetection_' . md5($url);
127	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
128	if ($siteContentDetectionCache !== \false) {
129	if ($this->checkCacheHasRequiredProperties($detectContent, $siteContentDetectionCache)) {
130	$this->detectedContent = $siteContentDetectionCache['detectedContent'];
131	$this->connectedConsentManagers = $siteContentDetectionCache['connectedConsentManagers'];
132	return;
133	}
134	}
135	// No cache hit, no passed data, so make a request for the site content
136	$siteResponse = $this->requestSiteResponse($url, $timeOut);
137	// Abort if still no site data
138	if (empty($siteResponse['data'])) {
139	return;
140	}
141	$this->siteResponse = $siteResponse;
142	// We now have site data to analyze, so run the detection checks
143	$this->detectionChecks($detectContent);
144	// A request was made to get this data and it isn't currently cached, so write it to the cache now
145	$cacheLife = 60 * 60 * 24 * 7;
146	$this->saveToCache($cacheKey, $cacheLife);
147	}
148	/**
149	* Returns if the detection with the provided id was detected or not
150	*
151	* Note: self::detectContent needs to be called before.
152	*
153	*/
154	public function wasDetected(string $detectionClassId) : bool
155	{
156	foreach ($this->detectedContent as $type => $detectedClassIds) {
157	if (array_key_exists($detectionClassId, $detectedClassIds)) {
158	return $detectedClassIds[$detectionClassId] ?? \false;
159	}
160	}
161	return \false;
162	}
163	/**
164	* Returns an array containing ids of all detected detections of the given type
165	*
166	* @param int $type One of the SiteContentDetectionAbstract::TYPE_* constants
167	* @return array
168	*/
169	public function getDetectsByType(int $type) : array
170	{
171	$detected = [];
172	foreach ($this->detectedContent[$type] as $objId => $wasDetected) {
173	if (\true === $wasDetected) {
174	$detected[] = $objId;
175	}
176	}
177	return $detected;
178	}
179	/**
180	* Checks that all required detections are in the cache array
181	*
182	* @param array $detectContent
183	* @param array $cache
184	*
185	*/
186	private function checkCacheHasRequiredProperties(array $detectContent, array $cache) : bool
187	{
188	if (empty($detectContent)) {
189	foreach (self::getSiteContentDetectionsByType() as $type => $entries) {
190	foreach ($entries as $entry) {
191	if (!isset($cache['detectedContent'][$type][$entry::getId()])) {
192	return \false;
193	// random detection missing
194	}
195	}
196	}
197	return \true;
198	}
199	foreach ($detectContent as $requestedDetection) {
200	if (is_string($requestedDetection)) {
201	// specific detection
202	$detectionObj = $this->getSiteContentDetectionById($requestedDetection);
203	if (null !== $detectionObj && !isset($cache['detectedContent'][$detectionObj::getContentType()][$detectionObj::getId()])) {
204	return \false;
205	// specific detection was run before
206	}
207	} elseif (is_int($requestedDetection)) {
208	// detection type requested
209	$detectionsByType = self::getSiteContentDetectionsByType();
210	if (isset($detectionsByType[$requestedDetection])) {
211	foreach ($detectionsByType[$requestedDetection] as $detectionObj) {
212	if (!isset($cache['detectedContent'][$requestedDetection][$detectionObj::getId()])) {
213	return \false;
214	// random detection missing
215	}
216	}
217	}
218	}
219	}
220	return \true;
221	}
222	/**
223	* Save data to the cache
224	*
225	*
226	*/
227	private function saveToCache(string $cacheKey, int $cacheLife) : void
228	{
229	$cacheData = ['detectedContent' => [], 'connectedConsentManagers' => []];
230	// Load any existing cached values
231	$siteContentDetectionCache = $this->cache->fetch($cacheKey);
232	if (is_array($siteContentDetectionCache)) {
233	$cacheData = $siteContentDetectionCache;
234	}
235	foreach ($this->detectedContent as $type => $detections) {
236	if (!isset($cacheData['detectedContent'][$type])) {
237	$cacheData['detectedContent'][$type] = [];
238	}
239	foreach ($detections as $detectionId => $wasDetected) {
240	if (null !== $wasDetected) {
241	$cacheData['detectedContent'][$type][$detectionId] = $wasDetected;
242	}
243	}
244	}
245	$cacheData['connectedConsentManagers'] = array_merge($cacheData['connectedConsentManagers'], $this->connectedConsentManagers);
246	$this->cache->save($cacheKey, $cacheData, $cacheLife);
247	}
248	/**
249	* Run various detection checks for site content
250	*
251	* @param array $detectContent Array of detection types used to filter the checks that are run
252	*
253	*/
254	private function detectionChecks(array $detectContent) : void
255	{
256	$detections = $this->getSiteContentDetectionsByType();
257	foreach ($detections as $type => $typeDetections) {
258	foreach ($typeDetections as $typeDetection) {
259	$this->detectedContent[$type][$typeDetection::getId()] = null;
260	if (in_array($type, $detectContent) \|\| in_array($typeDetection::getId(), $detectContent) \|\| empty($detectContent)) {
261	$this->detectedContent[$type][$typeDetection::getId()] = \false;
262	if ($typeDetection->isDetected($this->siteResponse['data'], $this->siteResponse['headers'])) {
263	if ($typeDetection instanceof ConsentManagerDetectionAbstract && $typeDetection->checkIsConnected($this->siteResponse['data'], $this->siteResponse['headers'])) {
264	$this->connectedConsentManagers[] = $typeDetection::getId();
265	}
266	$this->detectedContent[$type][$typeDetection::getId()] = \true;
267	}
268	}
269	}
270	}
271	}
272	/**
273	* Retrieve data from the specified site using an HTTP request
274	*
275	*
276	* @return array
277	*/
278	private function requestSiteResponse(string $url, int $timeOut) : array
279	{
280	if (!$url) {
281	return [];
282	}
283	// If internet features are disabled, we don't try to fetch any site content
284	if (0 === (int) GeneralConfig::getConfigValue('enable_internet_features')) {
285	return [];
286	}
287	$siteData = [];
288	try {
289	$siteData = \Piwik\Http::sendHttpRequestBy(\Piwik\Http::getTransportMethod(), $url, $timeOut, null, null, null, 0, \false, \true, \false, \true);
290	} catch (\Exception $e) {
291	}
292	return $siteData;
293	}
294	/**
295	* Return an array of consent manager definitions which can be used to detect their presence on the site and show
296	* the associated guide links
297	*
298	* Note: This list is also used to display the known / supported consent managers on the "Ask for Consent" page
299	* For adding a new consent manager to this page, it needs to be added here. If a consent manager can't be detected
300	* automatically, simply leave the detections empty.
301	*
302	* @return array[]
303	*/
304	public static function getKnownConsentManagers() : array
305	{
306	$detections = self::getSiteContentDetectionsByType();
307	$cmDetections = $detections[SiteContentDetectionAbstract::TYPE_CONSENT_MANAGER];
308	$consentManagers = [];
309	foreach ($cmDetections as $detection) {
310	$consentManagers[$detection::getId()] = ['name' => $detection::getName(), 'instructionUrl' => $detection::getInstructionUrl()];
311	}
312	return $consentManagers;
313	}
314	}
315