UrlHelper.php — Matomo Analytics – Powerful, Privacy-First Insights for WordPress 5.1.0

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

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