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

matomo / app / core / UrlHelper.php

matomo / app / core Last commit date

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

UrlHelper.php

327 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	/**
20	* @var string[]
21	*/
22	private static $validLinkProtocols = ['http', 'https', 'tel', 'sms', 'mailto', 'callto'];
23	/**
24	* Checks if a string matches/is equal to one of the patterns/strings.
25	*
26	* @static
27	* @param string $test String to test.
28	* @param string[] $patterns Array of strings or regexs.
29	*
30	* @return bool true if $test matches or is equal to one of the regex/string in $patterns, false otherwise.
31	*/
32	protected static function inArrayMatchesRegex($test, $patterns) : bool
33	{
34	foreach ($patterns as $val) {
35	if (@preg_match($val, '') === \false) {
36	if (strcasecmp($val, $test) === 0) {
37	return \true;
38	}
39	} else {
40	if (preg_match($val, $test) === 1) {
41	return \true;
42	}
43	}
44	}
45	return \false;
46	}
47	/**
48	* Converts an array of query parameter name/value mappings into a query string.
49	* Parameters that are in `$parametersToExclude` will not appear in the result.
50	*
51	* @static
52	* @param array<string, string\|false\|array<string\|false>> $queryParameters Array of query parameters, eg, `array('site' => '0', 'date' => '2012-01-01')`.
53	* @param string[] $parametersToExclude Array of query parameter names that shouldn't be
54	* in the result query string, eg, `array('date', 'period')`.
55	* @return string A query string, eg, `"?site=0"`.
56	* @api
57	*/
58	public static function getQueryStringWithExcludedParameters($queryParameters, $parametersToExclude)
59	{
60	$validQuery = '';
61	$separator = '&';
62	foreach ($queryParameters as $name => $value) {
63	// decode encoded square brackets
64	$name = str_replace(array('%5B', '%5D'), array('[', ']'), $name);
65	if (!self::inArrayMatchesRegex(strtolower($name), $parametersToExclude)) {
66	if (is_array($value)) {
67	foreach ($value as $param) {
68	if ($param === \false) {
69	$validQuery .= $name . '[]' . $separator;
70	} else {
71	$validQuery .= $name . '[]=' . $param . $separator;
72	}
73	}
74	} elseif ($value === \false) {
75	$validQuery .= $name . $separator;
76	} else {
77	$validQuery .= $name . '=' . $value . $separator;
78	}
79	}
80	}
81	$validQuery = substr($validQuery, 0, -strlen($separator));
82	return $validQuery;
83	}
84	/**
85	* Reduce URL to more minimal form. 2 letter country codes are
86	* replaced by '{}', while other parts are simply removed.
87	*
88	* Examples:
89	* www.example.com -> example.com
90	* search.example.com -> example.com
91	* m.example.com -> example.com
92	* de.example.com -> {}.example.com
93	* example.de -> example.{}
94	* example.co.uk -> example.{}
95	*
96	* @param string $url
97	* @return string
98	*/
99	public static function getLossyUrl($url)
100	{
101	static $countries;
102	if (!isset($countries)) {
103	/** @var RegionDataProvider $regionDataProvider */
104	$regionDataProvider = StaticContainer::get('Piwik\\Intl\\Data\\Provider\\RegionDataProvider');
105	$countries = implode('\|', array_keys($regionDataProvider->getCountryList(\true)));
106	}
107	return preg_replace(array('/^(w+[0-9]*\|search)\\./', '/(^\|\\.)m\\./', '/(\\.(com\|org\|net\|co\|it\|edu))?\\.(' . $countries . ')(\\/\|$)/', '/(^\|\\.)(' . $countries . ')\\./'), array('', '$1', '.{}$4', '$1{}.'), $url);
108	}
109	/**
110	* Returns true if the string passed may be a URL ie. it starts with protocol://.
111	* We don't need a precise test here because the value comes from the website
112	* tracked source code and the URLs may look very strange.
113	*
114	* @api
115	* @param string $url
116	* @return bool
117	*/
118	public static function isLookLikeUrl($url)
119	{
120	return $url && preg_match('~^(([[:alpha:]][[:alnum:]+.-])?:)?//(.)$~D', $url, $matches) !== 0 && strlen($matches[3]) > 0 && !preg_match('/^(javascript:\|vbscript:\|data:)/i', $matches[1]);
121	}
122	/**
123	* @param string $url
124	* @return bool
125	*/
126	public static function isLookLikeSafeUrl($url)
127	{
128	if (preg_match('/[\\x00-\\x1F\\x7F]/', $url)) {
129	return \false;
130	}
131	if (strpos($url, ':') === \false) {
132	return \true;
133	}
134	$protocol = explode(':', $url, 2)[0];
135	return (bool) preg_match('/^(' . implode('\|', self::$validLinkProtocols) . ')$/i', $protocol);
136	}
137	/**
138	* Returns a URL created from the result of the [parse_url](https://php.net/manual/en/function.parse-url.php)
139	* function.
140	*
141	* Copied from the PHP comments at [https://php.net/parse_url](https://php.net/parse_url).
142	*
143	* @param array $parsed Result of [parse_url](https://php.net/manual/en/function.parse-url.php).
144	* @return false\|string The URL or `false` if `$parsed` isn't an array.
145	* @api
146	*/
147	public static function getParseUrlReverse($parsed)
148	{
149	if (!is_array($parsed)) {
150	return \false;
151	}
152	// According to RFC 1738, the chars ':', '@' and '/' need to be encoded in username or password part of an url
153	// We also encode '\' as a username or password containing that char, might be handled incorrectly by browsers
154	$escapeSpecialChars = function ($value) {
155	return str_replace([':', '@', '/', '\\'], [urlencode(':'), urlencode('@'), urlencode('/'), urlencode('\\')], $value);
156	};
157	$uri = !empty($parsed['scheme']) ? $parsed['scheme'] . ':' . (!strcasecmp($parsed['scheme'], 'mailto') ? '' : '//') : '';
158	$uri .= !empty($parsed['user']) ? $escapeSpecialChars($parsed['user']) . (!empty($parsed['pass']) ? ':' . $escapeSpecialChars($parsed['pass']) : '') . '@' : '';
159	$uri .= !empty($parsed['host']) ? $parsed['host'] : '';
160	$uri .= !empty($parsed['port']) ? ':' . $parsed['port'] : '';
161	if (!empty($parsed['path'])) {
162	$uri .= !strncmp($parsed['path'], '/', 1) ? $parsed['path'] : (!empty($uri) ? '/' : '') . $parsed['path'];
163	}
164	$uri .= !empty($parsed['query']) ? '?' . $parsed['query'] : '';
165	$uri .= !empty($parsed['fragment']) ? '#' . $parsed['fragment'] : '';
166	return $uri;
167	}
168	/**
169	* Returns a URL query string as an array.
170	*
171	* @param string $urlQuery The query string, eg, `'?param1=value1&param2=value2'`.
172	* @return array eg, `array('param1' => 'value1', 'param2' => 'value2')`
173	* @api
174	*/
175	public static function getArrayFromQueryString($urlQuery)
176	{
177	if (empty($urlQuery)) {
178	return array();
179	}
180	// TODO: this method should not use a cache. callers should instead have their own cache, configured through DI.
181	// one undesirable side effect of using a cache here, is that this method can now init the StaticContainer, which makes setting
182	// test environment for RequestCommand more complicated.
183	$cache = \Piwik\Cache::getTransientCache();
184	$cacheKey = 'arrayFromQuery' . $urlQuery;
185	if ($cache->contains($cacheKey)) {
186	return $cache->fetch($cacheKey);
187	}
188	if ($urlQuery[0] == '?') {
189	$urlQuery = substr($urlQuery, 1);
190	}
191	$separator = '&';
192	$urlQuery = $separator . $urlQuery;
193	// $urlQuery = str_replace(array('%20'), ' ', $urlQuery);
194	$referrerQuery = trim($urlQuery);
195	$values = explode($separator, $referrerQuery);
196	$nameToValue = array();
197	foreach ($values as $value) {
198	$pos = strpos($value, '=');
199	if ($pos !== \false) {
200	$name = substr($value, 0, $pos);
201	$value = substr($value, $pos + 1);
202	if ($value === \false) {
203	$value = '';
204	}
205	} else {
206	$name = $value;
207	$value = \false;
208	}
209	if (!empty($name)) {
210	$name = \Piwik\Common::sanitizeInputValue($name);
211	}
212	if (!empty($value)) {
213	$value = \Piwik\Common::sanitizeInputValue($value);
214	}
215	// if array without indexes
216	$count = 0;
217	$tmp = preg_replace('/(\\[\|%5b)(]\|%5d)$/i', '', $name, -1, $count);
218	if (!empty($tmp) && $count) {
219	$name = $tmp;
220	if (isset($nameToValue[$name]) == \false \|\| is_array($nameToValue[$name]) == \false) {
221	$nameToValue[$name] = array();
222	}
223	array_push($nameToValue[$name], $value);
224	} elseif (!empty($name)) {
225	$nameToValue[$name] = $value;
226	}
227	}
228	$cache->save($cacheKey, $nameToValue);
229	return $nameToValue;
230	}
231	/**
232	* Returns the value of a single query parameter from the supplied query string.
233	*
234	* @param string $urlQuery The query string.
235	* @param string $parameter The query parameter name to return.
236	* @return string\|null Parameter value if found (can be the empty string!), null if not found.
237	* @api
238	*/
239	public static function getParameterFromQueryString($urlQuery, $parameter)
240	{
241	$nameToValue = self::getArrayFromQueryString($urlQuery);
242	if (isset($nameToValue[$parameter])) {
243	return $nameToValue[$parameter];
244	}
245	return null;
246	}
247	/**
248	* Returns the path and query string of a URL.
249	*
250	* @param string $url The URL.
251	* @param array $additionalParamsToAdd If not empty the given parameters will be added to the query.
252	* @param bool $preserveAnchor If true then do not remove any #anchor from the url, default false
253	* @return string eg, `/test/index.php?module=CoreHome` if `$url` is `https://piwik.org/test/index.php?module=CoreHome`.
254	* @api
255	*/
256	public static function getPathAndQueryFromUrl($url, array $additionalParamsToAdd = [], bool $preserveAnchor = \false)
257	{
258	$parsedUrl = parse_url($url);
259	// If an anchor is included in the URL parse_url() will not split the anchor and query, so we do that there
260	if (isset($parsedUrl['fragment']) && strpos($parsedUrl['fragment'], '?') !== \false) {
261	$parsedUrl['query'] = substr($parsedUrl['fragment'], strpos($parsedUrl['fragment'], '?') + 1);
262	$parsedUrl['fragment'] = substr($parsedUrl['fragment'], 0, strpos($parsedUrl['fragment'], '?'));
263	}
264	$result = '';
265	if (isset($parsedUrl['path'])) {
266	if (substr($parsedUrl['path'], 0, 1) == '/') {
267	$parsedUrl['path'] = substr($parsedUrl['path'], 1);
268	}
269	$result .= $parsedUrl['path'];
270	}
271	if ($preserveAnchor && isset($parsedUrl['fragment'])) {
272	$result .= '#' . $parsedUrl['fragment'];
273	}
274	if (isset($parsedUrl['query']) \|\| count($additionalParamsToAdd)) {
275	$query = isset($parsedUrl['query']) ? $parsedUrl['query'] : '';
276	$query = self::addAdditionalParameters($query, $additionalParamsToAdd);
277	$result .= '?' . $query;
278	}
279	return $result;
280	}
281	/**
282	* Returns the query part from any valid url and adds additional parameters to the query part if needed.
283	*
284	* @param string $url Any url eg `"https://example.com/piwik/?foo=bar"`
285	* @param array $additionalParamsToAdd If not empty the given parameters will be added to the query.
286	*
287	* @return string eg. `"foo=bar&foo2=bar2"`
288	* @api
289	*/
290	public static function getQueryFromUrl($url, array $additionalParamsToAdd = [])
291	{
292	$url = @parse_url($url);
293	$query = '';
294	if (!empty($url['query'])) {
295	$query .= $url['query'];
296	}
297	$query = self::addAdditionalParameters($query, $additionalParamsToAdd);
298	return $query;
299	}
300	/**
301	* Add an array of additional parameters to a query string
302	*
303	* @param array $additionalParamsToAdd
304	*/
305	private static function addAdditionalParameters(string $query, array $additionalParamsToAdd) : string
306	{
307	if (!empty($additionalParamsToAdd)) {
308	if (!empty($query)) {
309	$query .= '&';
310	}
311	$query .= \Piwik\Url::getQueryStringFromParameters($additionalParamsToAdd);
312	}
313	return $query;
314	}
315	/**
316	* @param string $url
317	* @return string\|false\|null
318	*/
319	public static function getHostFromUrl($url)
320	{
321	if (!\Piwik\UrlHelper::isLookLikeUrl($url)) {
322	$url = "http://" . $url;
323	}
324	return parse_url($url, \PHP_URL_HOST);
325	}
326	}
327