UriString.php — Independent Analytics – WordPress Analytics Plugin 1.18

independent-analytics / vendor / league / uri / src / UriString.php

independent-analytics / vendor / league / uri / src Last commit date

Exceptions 3 years ago UriTemplate 3 years ago Http.php 3 years ago HttpFactory.php 3 years ago Uri.php 3 years ago UriInfo.php 3 years ago UriResolver.php 3 years ago UriString.php 3 years ago UriTemplate.php 3 years ago

UriString.php

385 lines

1	<?php
2
3	/**
4	* League.Uri (https://uri.thephpleague.com)
5	*
6	* (c) Ignace Nyamagana Butera <nyamsprod@gmail.com>
7	*
8	* For the full copyright and license information, please view the LICENSE
9	* file that was distributed with this source code.
10	*/
11	declare (strict_types=1);
12	namespace IAWP\League\Uri;
13
14	use IAWP\League\Uri\Exceptions\IdnaConversionFailed;
15	use IAWP\League\Uri\Exceptions\IdnSupportMissing;
16	use IAWP\League\Uri\Exceptions\SyntaxError;
17	use IAWP\League\Uri\Idna\Idna;
18	use function array_merge;
19	use function explode;
20	use function filter_var;
21	use function gettype;
22	use function inet_pton;
23	use function is_object;
24	use function is_scalar;
25	use function method_exists;
26	use function preg_match;
27	use function rawurldecode;
28	use function sprintf;
29	use function strpos;
30	use function substr;
31	use const FILTER_FLAG_IPV6;
32	use const FILTER_VALIDATE_IP;
33	/**
34	* A class to parse a URI string according to RFC3986.
35	*
36	* @link https://tools.ietf.org/html/rfc3986
37	* @package League\Uri
38	* @author Ignace Nyamagana Butera <nyamsprod@gmail.com>
39	* @since 6.0.0
40	*/
41	final class UriString
42	{
43	/**
44	* Default URI component values.
45	*/
46	private const URI_COMPONENTS = ['scheme' => null, 'user' => null, 'pass' => null, 'host' => null, 'port' => null, 'path' => '', 'query' => null, 'fragment' => null];
47	/**
48	* Simple URI which do not need any parsing.
49	*/
50	private const URI_SCHORTCUTS = ['' => [], '#' => ['fragment' => ''], '?' => ['query' => ''], '?#' => ['query' => '', 'fragment' => ''], '/' => ['path' => '/'], '//' => ['host' => '']];
51	/**
52	* Range of invalid characters in URI string.
53	*/
54	private const REGEXP_INVALID_URI_CHARS = '/[\\x00-\\x1f\\x7f]/';
55	/**
56	* RFC3986 regular expression URI splitter.
57	*
58	* @link https://tools.ietf.org/html/rfc3986#appendix-B
59	*/
60	private const REGEXP_URI_PARTS = ',^
61	(?<scheme>(?<scontent>[^:/?\\#]+):)? # URI scheme component
62	(?<authority>//(?<acontent>[^/?\\#]*))? # URI authority part
63	(?<path>[^?\\#]*) # URI path component
64	(?<query>\\?(?<qcontent>[^\\#]*))? # URI query component
65	(?<fragment>\\#(?<fcontent>.*))? # URI fragment component
66	,x';
67	/**
68	* URI scheme regular expresssion.
69	*
70	* @link https://tools.ietf.org/html/rfc3986#section-3.1
71	*/
72	private const REGEXP_URI_SCHEME = '/^([a-z][a-z\\d\\+\\.\\-]*)?$/i';
73	/**
74	* IPvFuture regular expression.
75	*
76	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
77	*/
78	private const REGEXP_IP_FUTURE = '/^
79	v(?<version>[A-F0-9])+\\.
80	(?:
81	(?<unreserved>[a-z0-9_~\\-\\.])\|
82	(?<sub_delims>[!$&\'()*+,;=:]) # also include the : character
83	)+
84	$/ix';
85	/**
86	* General registered name regular expression.
87	*
88	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
89	*/
90	private const REGEXP_REGISTERED_NAME = '/(?(DEFINE)
91	(?<unreserved>[a-z0-9_~\\-]) # . is missing as it is used to separate labels
92	(?<sub_delims>[!$&\'()*+,;=])
93	(?<encoded>%[A-F0-9]{2})
94	(?<reg_name>(?:(?&unreserved)\|(?&sub_delims)\|(?&encoded))*)
95	)
96	^(?:(?&reg_name)\\.)*(?&reg_name)\\.?$/ix';
97	/**
98	* Invalid characters in host regular expression.
99	*
100	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
101	*/
102	private const REGEXP_INVALID_HOST_CHARS = '/
103	[:\\/?#\\[\\]@ ] # gen-delims characters as well as the space character
104	/ix';
105	/**
106	* Invalid path for URI without scheme and authority regular expression.
107	*
108	* @link https://tools.ietf.org/html/rfc3986#section-3.3
109	*/
110	private const REGEXP_INVALID_PATH = ',^(([^/]):)(.)?/,';
111	/**
112	* Host and Port splitter regular expression.
113	*/
114	private const REGEXP_HOST_PORT = ',^(?<host>\\[.\\]\|[^:])(:(?<port>.*))?$,';
115	/**
116	* IDN Host detector regular expression.
117	*/
118	private const REGEXP_IDN_PATTERN = '/[^\\x20-\\x7f]/';
119	/**
120	* Only the address block fe80::/10 can have a Zone ID attach to
121	* let's detect the link local significant 10 bits.
122	*/
123	private const ZONE_ID_ADDRESS_BLOCK = "\xfe\x80";
124	/**
125	* Generate an URI string representation from its parsed representation
126	* returned by League\Uri\parse() or PHP's parse_url.
127	*
128	* If you supply your own array, you are responsible for providing
129	* valid components without their URI delimiters.
130	*
131	* @link https://tools.ietf.org/html/rfc3986#section-5.3
132	* @link https://tools.ietf.org/html/rfc3986#section-7.5
133	*
134	* @param array{
135	* scheme:?string,
136	* user:?string,
137	* pass:?string,
138	* host:?string,
139	* port:?int,
140	* path:string,
141	* query:?string,
142	* fragment:?string
143	* } $components
144	*/
145	public static function build(array $components) : string
146	{
147	$result = $components['path'] ?? '';
148	if (isset($components['query'])) {
149	$result .= '?' . $components['query'];
150	}
151	if (isset($components['fragment'])) {
152	$result .= '#' . $components['fragment'];
153	}
154	$scheme = null;
155	if (isset($components['scheme'])) {
156	$scheme = $components['scheme'] . ':';
157	}
158	if (!isset($components['host'])) {
159	return $scheme . $result;
160	}
161	$scheme .= '//';
162	$authority = $components['host'];
163	if (isset($components['port'])) {
164	$authority .= ':' . $components['port'];
165	}
166	if (!isset($components['user'])) {
167	return $scheme . $authority . $result;
168	}
169	$authority = '@' . $authority;
170	if (!isset($components['pass'])) {
171	return $scheme . $components['user'] . $authority . $result;
172	}
173	return $scheme . $components['user'] . ':' . $components['pass'] . $authority . $result;
174	}
175	/**
176	* Parse an URI string into its components.
177	*
178	* This method parses a URI and returns an associative array containing any
179	* of the various components of the URI that are present.
180	*
181	* <code>
182	* $components = (new Parser())->parse('http://foo@test.example.com:42?query#');
183	* var_export($components);
184	* //will display
185	* array(
186	* 'scheme' => 'http', // the URI scheme component
187	* 'user' => 'foo', // the URI user component
188	* 'pass' => null, // the URI pass component
189	* 'host' => 'test.example.com', // the URI host component
190	* 'port' => 42, // the URI port component
191	* 'path' => '', // the URI path component
192	* 'query' => 'query', // the URI query component
193	* 'fragment' => '', // the URI fragment component
194	* );
195	* </code>
196	*
197	* The returned array is similar to PHP's parse_url return value with the following
198	* differences:
199	*
200	* <ul>
201	* <li>All components are always present in the returned array</li>
202	* <li>Empty and undefined component are treated differently. And empty component is
203	* set to the empty string while an undefined component is set to the `null` value.</li>
204	* <li>The path component is never undefined</li>
205	* <li>The method parses the URI following the RFC3986 rules but you are still
206	* required to validate the returned components against its related scheme specific rules.</li>
207	* </ul>
208	*
209	* @link https://tools.ietf.org/html/rfc3986
210	*
211	* @param mixed $uri any scalar or stringable object
212	*
213	* @throws SyntaxError if the URI contains invalid characters
214	* @throws SyntaxError if the URI contains an invalid scheme
215	* @throws SyntaxError if the URI contains an invalid path
216	*
217	* @return array{
218	* scheme:?string,
219	* user:?string,
220	* pass:?string,
221	* host:?string,
222	* port:?int,
223	* path:string,
224	* query:?string,
225	* fragment:?string
226	* }
227	*/
228	public static function parse($uri) : array
229	{
230	if (is_object($uri) && method_exists($uri, '__toString')) {
231	$uri = (string) $uri;
232	}
233	if (!is_scalar($uri)) {
234	throw new \TypeError(sprintf('The uri must be a scalar or a stringable object `%s` given', gettype($uri)));
235	}
236	$uri = (string) $uri;
237	if (isset(self::URI_SCHORTCUTS[$uri])) {
238	/** @var array{scheme:?string, user:?string, pass:?string, host:?string, port:?int, path:string, query:?string, fragment:?string} $components */
239	$components = array_merge(self::URI_COMPONENTS, self::URI_SCHORTCUTS[$uri]);
240	return $components;
241	}
242	if (1 === preg_match(self::REGEXP_INVALID_URI_CHARS, $uri)) {
243	throw new SyntaxError(sprintf('The uri `%s` contains invalid characters', $uri));
244	}
245	//if the first character is a known URI delimiter parsing can be simplified
246	$first_char = $uri[0];
247	//The URI is made of the fragment only
248	if ('#' === $first_char) {
249	[, $fragment] = explode('#', $uri, 2);
250	$components = self::URI_COMPONENTS;
251	$components['fragment'] = $fragment;
252	return $components;
253	}
254	//The URI is made of the query and fragment
255	if ('?' === $first_char) {
256	[, $partial] = explode('?', $uri, 2);
257	[$query, $fragment] = explode('#', $partial, 2) + [1 => null];
258	$components = self::URI_COMPONENTS;
259	$components['query'] = $query;
260	$components['fragment'] = $fragment;
261	return $components;
262	}
263	//use RFC3986 URI regexp to split the URI
264	preg_match(self::REGEXP_URI_PARTS, $uri, $parts);
265	$parts += ['query' => '', 'fragment' => ''];
266	if (':' === $parts['scheme'] \|\| 1 !== preg_match(self::REGEXP_URI_SCHEME, $parts['scontent'])) {
267	throw new SyntaxError(sprintf('The uri `%s` contains an invalid scheme', $uri));
268	}
269	if ('' === $parts['scheme'] . $parts['authority'] && 1 === preg_match(self::REGEXP_INVALID_PATH, $parts['path'])) {
270	throw new SyntaxError(sprintf('The uri `%s` contains an invalid path.', $uri));
271	}
272	/** @var array{scheme:?string, user:?string, pass:?string, host:?string, port:?int, path:string, query:?string, fragment:?string} $components */
273	$components = array_merge(self::URI_COMPONENTS, '' === $parts['authority'] ? [] : self::parseAuthority($parts['acontent']), ['path' => $parts['path'], 'scheme' => '' === $parts['scheme'] ? null : $parts['scontent'], 'query' => '' === $parts['query'] ? null : $parts['qcontent'], 'fragment' => '' === $parts['fragment'] ? null : $parts['fcontent']]);
274	return $components;
275	}
276	/**
277	* Parses the URI authority part.
278	*
279	* @link https://tools.ietf.org/html/rfc3986#section-3.2
280	*
281	* @throws SyntaxError If the port component is invalid
282	*
283	* @return array{user:?string, pass:?string, host:?string, port:?int}
284	*/
285	private static function parseAuthority(string $authority) : array
286	{
287	$components = ['user' => null, 'pass' => null, 'host' => '', 'port' => null];
288	if ('' === $authority) {
289	return $components;
290	}
291	$parts = explode('@', $authority, 2);
292	if (isset($parts[1])) {
293	[$components['user'], $components['pass']] = explode(':', $parts[0], 2) + [1 => null];
294	}
295	preg_match(self::REGEXP_HOST_PORT, $parts[1] ?? $parts[0], $matches);
296	$matches += ['port' => ''];
297	$components['port'] = self::filterPort($matches['port']);
298	$components['host'] = self::filterHost($matches['host']);
299	return $components;
300	}
301	/**
302	* Filter and format the port component.
303	*
304	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
305	*
306	* @throws SyntaxError if the registered name is invalid
307	*/
308	private static function filterPort(string $port) : ?int
309	{
310	if ('' === $port) {
311	return null;
312	}
313	if (1 === preg_match('/^\\d*$/', $port)) {
314	return (int) $port;
315	}
316	throw new SyntaxError(sprintf('The port `%s` is invalid', $port));
317	}
318	/**
319	* Returns whether a hostname is valid.
320	*
321	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
322	*
323	* @throws SyntaxError if the registered name is invalid
324	*/
325	private static function filterHost(string $host) : string
326	{
327	if ('' === $host) {
328	return $host;
329	}
330	if ('[' !== $host[0] \|\| ']' !== substr($host, -1)) {
331	return self::filterRegisteredName($host);
332	}
333	if (!self::isIpHost(substr($host, 1, -1))) {
334	throw new SyntaxError(sprintf('Host `%s` is invalid : the IP host is malformed', $host));
335	}
336	return $host;
337	}
338	/**
339	* Returns whether the host is an IPv4 or a registered named.
340	*
341	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
342	*
343	* @throws SyntaxError if the registered name is invalid
344	* @throws IdnSupportMissing if IDN support or ICU requirement are not available or met.
345	*/
346	private static function filterRegisteredName(string $host) : string
347	{
348	$formatted_host = rawurldecode($host);
349	if (1 === preg_match(self::REGEXP_REGISTERED_NAME, $formatted_host)) {
350	return $host;
351	}
352	//to test IDN host non-ascii characters must be present in the host
353	if (1 !== preg_match(self::REGEXP_IDN_PATTERN, $formatted_host)) {
354	throw new SyntaxError(sprintf('Host `%s` is invalid : the host is not a valid registered name', $host));
355	}
356	$info = Idna::toAscii($host, Idna::IDNA2008_ASCII);
357	if (0 !== $info->errors()) {
358	throw IdnaConversionFailed::dueToIDNAError($host, $info);
359	}
360	return $host;
361	}
362	/**
363	* Validates a IPv6/IPvfuture host.
364	*
365	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
366	* @link https://tools.ietf.org/html/rfc6874#section-2
367	* @link https://tools.ietf.org/html/rfc6874#section-4
368	*/
369	private static function isIpHost(string $ip_host) : bool
370	{
371	if (\false !== filter_var($ip_host, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6)) {
372	return \true;
373	}
374	if (1 === preg_match(self::REGEXP_IP_FUTURE, $ip_host, $matches)) {
375	return !\in_array($matches['version'], ['4', '6'], \true);
376	}
377	$pos = strpos($ip_host, '%');
378	if (\false === $pos \|\| 1 === preg_match(self::REGEXP_INVALID_HOST_CHARS, rawurldecode(substr($ip_host, $pos)))) {
379	return \false;
380	}
381	$ip_host = substr($ip_host, 0, $pos);
382	return \false !== filter_var($ip_host, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6) && 0 === strpos((string) inet_pton($ip_host), self::ZONE_ID_ADDRESS_BLOCK);
383	}
384	}
385