UrlHelper.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 5.0.7

matomo / app / core / UrlHelper.php

matomo / app / core Last commit date

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

UrlHelper.php

315 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	namespace Piwik;
11
12	use Piwik\Container\StaticContainer;
13	use Piwik\Intl\Data\Provider\RegionDataProvider;
14	/**
15	* Contains less commonly needed URL helper methods.
16	*
17	*/
18	class UrlHelper
19	{
20	private static $validLinkProtocols = ['http', 'https', 'tel', 'sms', 'mailto', 'callto'];
21	/**
22	* Checks if a string matches/is equal to one of the patterns/strings.
23	*
24	* @static
25	* @param $test String to test.
26	* @param $patterns Array of strings or regexs.
27	*
28	* @return bool true if $test matches or is equal to one of the regex/string in $patterns, false otherwise.
29	*/
30	protected static function inArrayMatchesRegex($test, $patterns) : bool
31	{
32	foreach ($patterns as $val) {
33	if (@preg_match($val, null) === false) {
34	if (strcasecmp($val, $test) === 0) {
35	return true;
36	}
37	} else {
38	if (preg_match($val, $test) === 1) {
39	return true;
40	}
41	}
42	}
43	return false;
44	}
45	/**
46	* Converts an array of query parameter name/value mappings into a query string.
47	* Parameters that are in `$parametersToExclude` will not appear in the result.
48	*
49	* @static
50	* @param $queryParameters Array of query parameters, eg, `array('site' => '0', 'date' => '2012-01-01')`.
51	* @param $parametersToExclude Array of query parameter names that shouldn't be
52	* in the result query string, eg, `array('date', 'period')`.
53	* @return string A query string, eg, `"?site=0"`.
54	* @api
55	*/
56	public static function getQueryStringWithExcludedParameters($queryParameters, $parametersToExclude)
57	{
58	$validQuery = '';
59	$separator = '&';
60	foreach ($queryParameters as $name => $value) {
61	// decode encoded square brackets
62	$name = str_replace(array('%5B', '%5D'), array('[', ']'), $name);
63	if (!self::inArrayMatchesRegex(strtolower($name), $parametersToExclude)) {
64	if (is_array($value)) {
65	foreach ($value as $param) {
66	if ($param === false) {
67	$validQuery .= $name . '[]' . $separator;
68	} else {
69	$validQuery .= $name . '[]=' . $param . $separator;
70	}
71	}
72	} elseif ($value === false) {
73	$validQuery .= $name . $separator;
74	} else {
75	$validQuery .= $name . '=' . $value . $separator;
76	}
77	}
78	}
79	$validQuery = substr($validQuery, 0, -strlen($separator));
80	return $validQuery;
81	}
82	/**
83	* Reduce URL to more minimal form. 2 letter country codes are
84	* replaced by '{}', while other parts are simply removed.
85	*
86	* Examples:
87	* www.example.com -> example.com
88	* search.example.com -> example.com
89	* m.example.com -> example.com
90	* de.example.com -> {}.example.com
91	* example.de -> example.{}
92	* example.co.uk -> example.{}
93	*
94	* @param string $url
95	* @return string
96	*/
97	public static function getLossyUrl($url)
98	{
99	static $countries;
100	if (!isset($countries)) {
101	/** @var RegionDataProvider $regionDataProvider */
102	$regionDataProvider = StaticContainer::get('Piwik\\Intl\\Data\\Provider\\RegionDataProvider');
103	$countries = implode('\|', array_keys($regionDataProvider->getCountryList(true)));
104	}
105	return preg_replace(array('/^(w+[0-9]*\|search)\\./', '/(^\|\\.)m\\./', '/(\\.(com\|org\|net\|co\|it\|edu))?\\.(' . $countries . ')(\\/\|$)/', '/(^\|\\.)(' . $countries . ')\\./'), array('', '$1', '.{}$4', '$1{}.'), $url);
106	}
107	/**
108	* Returns true if the string passed may be a URL ie. it starts with protocol://.
109	* We don't need a precise test here because the value comes from the website
110	* tracked source code and the URLs may look very strange.
111	*
112	* @api
113	* @param string $url
114	* @return bool
115	*/
116	public static function isLookLikeUrl($url)
117	{
118	return $url && preg_match('~^(([[:alpha:]][[:alnum:]+.-])?:)?//(.)$~D', $url, $matches) !== 0 && strlen($matches[3]) > 0 && !preg_match('/^(javascript:\|vbscript:\|data:)/i', $matches[1]);
119	}
120	public static function isLookLikeSafeUrl($url)
121	{
122	if (preg_match('/[\\x00-\\x1F\\x7F]/', $url)) {
123	return false;
124	}
125	if (strpos($url, ':') === false) {
126	return true;
127	}
128	$protocol = explode(':', $url, 2)[0];
129	return preg_match('/^(' . implode('\|', self::$validLinkProtocols) . ')$/i', $protocol);
130	}
131	/**
132	* Returns a URL created from the result of the [parse_url](http://php.net/manual/en/function.parse-url.php)
133	* function.
134	*
135	* Copied from the PHP comments at [http://php.net/parse_url](http://php.net/parse_url).
136	*
137	* @param array $parsed Result of [parse_url](http://php.net/manual/en/function.parse-url.php).
138	* @return false\|string The URL or `false` if `$parsed` isn't an array.
139	* @api
140	*/
141	public static function getParseUrlReverse($parsed)
142	{
143	if (!is_array($parsed)) {
144	return false;
145	}
146	$uri = !empty($parsed['scheme']) ? $parsed['scheme'] . ':' . (!strcasecmp($parsed['scheme'], 'mailto') ? '' : '//') : '';
147	$uri .= !empty($parsed['user']) ? $parsed['user'] . (!empty($parsed['pass']) ? ':' . $parsed['pass'] : '') . '@' : '';
148	$uri .= !empty($parsed['host']) ? $parsed['host'] : '';
149	$uri .= !empty($parsed['port']) ? ':' . $parsed['port'] : '';
150	if (!empty($parsed['path'])) {
151	$uri .= !strncmp($parsed['path'], '/', 1) ? $parsed['path'] : (!empty($uri) ? '/' : '') . $parsed['path'];
152	}
153	$uri .= !empty($parsed['query']) ? '?' . $parsed['query'] : '';
154	$uri .= !empty($parsed['fragment']) ? '#' . $parsed['fragment'] : '';
155	return $uri;
156	}
157	/**
158	* Returns a URL query string as an array.
159	*
160	* @param string $urlQuery The query string, eg, `'?param1=value1&param2=value2'`.
161	* @return array eg, `array('param1' => 'value1', 'param2' => 'value2')`
162	* @api
163	*/
164	public static function getArrayFromQueryString($urlQuery)
165	{
166	if (empty($urlQuery)) {
167	return array();
168	}
169	// TODO: this method should not use a cache. callers should instead have their own cache, configured through DI.
170	// one undesirable side effect of using a cache here, is that this method can now init the StaticContainer, which makes setting
171	// test environment for RequestCommand more complicated.
172	$cache = \Piwik\Cache::getTransientCache();
173	$cacheKey = 'arrayFromQuery' . $urlQuery;
174	if ($cache->contains($cacheKey)) {
175	return $cache->fetch($cacheKey);
176	}
177	if ($urlQuery[0] == '?') {
178	$urlQuery = substr($urlQuery, 1);
179	}
180	$separator = '&';
181	$urlQuery = $separator . $urlQuery;
182	// $urlQuery = str_replace(array('%20'), ' ', $urlQuery);
183	$referrerQuery = trim($urlQuery);
184	$values = explode($separator, $referrerQuery);
185	$nameToValue = array();
186	foreach ($values as $value) {
187	$pos = strpos($value, '=');
188	if ($pos !== false) {
189	$name = substr($value, 0, $pos);
190	$value = substr($value, $pos + 1);
191	if ($value === false) {
192	$value = '';
193	}
194	} else {
195	$name = $value;
196	$value = false;
197	}
198	if (!empty($name)) {
199	$name = \Piwik\Common::sanitizeInputValue($name);
200	}
201	if (!empty($value)) {
202	$value = \Piwik\Common::sanitizeInputValue($value);
203	}
204	// if array without indexes
205	$count = 0;
206	$tmp = preg_replace('/(\\[\|%5b)(]\|%5d)$/i', '', $name, -1, $count);
207	if (!empty($tmp) && $count) {
208	$name = $tmp;
209	if (isset($nameToValue[$name]) == false \|\| is_array($nameToValue[$name]) == false) {
210	$nameToValue[$name] = array();
211	}
212	array_push($nameToValue[$name], $value);
213	} elseif (!empty($name)) {
214	$nameToValue[$name] = $value;
215	}
216	}
217	$cache->save($cacheKey, $nameToValue);
218	return $nameToValue;
219	}
220	/**
221	* Returns the value of a single query parameter from the supplied query string.
222	*
223	* @param string $urlQuery The query string.
224	* @param string $parameter The query parameter name to return.
225	* @return string\|null Parameter value if found (can be the empty string!), null if not found.
226	* @api
227	*/
228	public static function getParameterFromQueryString($urlQuery, $parameter)
229	{
230	$nameToValue = self::getArrayFromQueryString($urlQuery);
231	if (isset($nameToValue[$parameter])) {
232	return $nameToValue[$parameter];
233	}
234	return null;
235	}
236	/**
237	* Returns the path and query string of a URL.
238	*
239	* @param string $url The URL.
240	* @param array $additionalParamsToAdd If not empty the given parameters will be added to the query.
241	* @param bool $preserveAnchor If true then do not remove any #anchor from the url, default false
242	* @return string eg, `/test/index.php?module=CoreHome` if `$url` is `http://piwik.org/test/index.php?module=CoreHome`.
243	* @api
244	*/
245	public static function getPathAndQueryFromUrl($url, array $additionalParamsToAdd = [], bool $preserveAnchor = false)
246	{
247	$parsedUrl = parse_url($url);
248	// If an anchor is included in the URL parse_url() will not split the anchor and query, so we do that there
249	if (isset($parsedUrl['fragment']) && strpos($parsedUrl['fragment'], '?') !== false) {
250	$parsedUrl['query'] = substr($parsedUrl['fragment'], strpos($parsedUrl['fragment'], '?') + 1);
251	$parsedUrl['fragment'] = substr($parsedUrl['fragment'], 0, strpos($parsedUrl['fragment'], '?'));
252	}
253	$result = '';
254	if (isset($parsedUrl['path'])) {
255	if (substr($parsedUrl['path'], 0, 1) == '/') {
256	$parsedUrl['path'] = substr($parsedUrl['path'], 1);
257	}
258	$result .= $parsedUrl['path'];
259	}
260	if ($preserveAnchor && isset($parsedUrl['fragment'])) {
261	$result .= '#' . $parsedUrl['fragment'];
262	}
263	if (isset($parsedUrl['query']) \|\| count($additionalParamsToAdd)) {
264	$query = isset($parsedUrl['query']) ? $parsedUrl['query'] : '';
265	$query = self::addAdditionalParameters($query, $additionalParamsToAdd);
266	$result .= '?' . $query;
267	}
268	return $result;
269	}
270	/**
271	* Returns the query part from any valid url and adds additional parameters to the query part if needed.
272	*
273	* @param string $url Any url eg `"http://example.com/piwik/?foo=bar"`
274	* @param array $additionalParamsToAdd If not empty the given parameters will be added to the query.
275	*
276	* @return string eg. `"foo=bar&foo2=bar2"`
277	* @api
278	*/
279	public static function getQueryFromUrl($url, array $additionalParamsToAdd = [])
280	{
281	$url = @parse_url($url);
282	$query = '';
283	if (!empty($url['query'])) {
284	$query .= $url['query'];
285	}
286	$query = self::addAdditionalParameters($query, $additionalParamsToAdd);
287	return $query;
288	}
289	/**
290	* Add an array of additional parameters to a query string
291	*
292	* @param string $query
293	* @param array $additionalParamsToAdd
294	*
295	* @return string
296	*/
297	private static function addAdditionalParameters(string $query, array $additionalParamsToAdd) : string
298	{
299	if (!empty($additionalParamsToAdd)) {
300	if (!empty($query)) {
301	$query .= '&';
302	}
303	$query .= \Piwik\Url::getQueryStringFromParameters($additionalParamsToAdd);
304	}
305	return $query;
306	}
307	public static function getHostFromUrl($url)
308	{
309	if (!\Piwik\UrlHelper::isLookLikeUrl($url)) {
310	$url = "http://" . $url;
311	}
312	return parse_url($url, PHP_URL_HOST);
313	}
314	}
315