UriString.php — Independent Analytics – WordPress Analytics Plugin 2.2.0

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

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

Exceptions 2 years ago UriTemplate 2 years ago Http.php 2 years ago HttpFactory.php 2 years ago Uri.php 2 years ago UriInfo.php 2 years ago UriResolver.php 2 years ago UriString.php 2 years ago UriTemplate.php 2 years ago

UriString.php

386 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_SCOPED\League\Uri;
13
14	use IAWP_SCOPED\League\Uri\Exceptions\IdnaConversionFailed;
15	use IAWP_SCOPED\League\Uri\Exceptions\IdnSupportMissing;
16	use IAWP_SCOPED\League\Uri\Exceptions\SyntaxError;
17	use IAWP_SCOPED\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	* @internal
41	*/
42	final class UriString
43	{
44	/**
45	* Default URI component values.
46	*/
47	private const URI_COMPONENTS = ['scheme' => null, 'user' => null, 'pass' => null, 'host' => null, 'port' => null, 'path' => '', 'query' => null, 'fragment' => null];
48	/**
49	* Simple URI which do not need any parsing.
50	*/
51	private const URI_SCHORTCUTS = ['' => [], '#' => ['fragment' => ''], '?' => ['query' => ''], '?#' => ['query' => '', 'fragment' => ''], '/' => ['path' => '/'], '//' => ['host' => '']];
52	/**
53	* Range of invalid characters in URI string.
54	*/
55	private const REGEXP_INVALID_URI_CHARS = '/[\\x00-\\x1f\\x7f]/';
56	/**
57	* RFC3986 regular expression URI splitter.
58	*
59	* @link https://tools.ietf.org/html/rfc3986#appendix-B
60	*/
61	private const REGEXP_URI_PARTS = ',^
62	(?<scheme>(?<scontent>[^:/?\\#]+):)? # URI scheme component
63	(?<authority>//(?<acontent>[^/?\\#]*))? # URI authority part
64	(?<path>[^?\\#]*) # URI path component
65	(?<query>\\?(?<qcontent>[^\\#]*))? # URI query component
66	(?<fragment>\\#(?<fcontent>.*))? # URI fragment component
67	,x';
68	/**
69	* URI scheme regular expresssion.
70	*
71	* @link https://tools.ietf.org/html/rfc3986#section-3.1
72	*/
73	private const REGEXP_URI_SCHEME = '/^([a-z][a-z\\d\\+\\.\\-]*)?$/i';
74	/**
75	* IPvFuture regular expression.
76	*
77	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
78	*/
79	private const REGEXP_IP_FUTURE = '/^
80	v(?<version>[A-F0-9])+\\.
81	(?:
82	(?<unreserved>[a-z0-9_~\\-\\.])\|
83	(?<sub_delims>[!$&\'()*+,;=:]) # also include the : character
84	)+
85	$/ix';
86	/**
87	* General registered name regular expression.
88	*
89	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
90	*/
91	private const REGEXP_REGISTERED_NAME = '/(?(DEFINE)
92	(?<unreserved>[a-z0-9_~\\-]) # . is missing as it is used to separate labels
93	(?<sub_delims>[!$&\'()*+,;=])
94	(?<encoded>%[A-F0-9]{2})
95	(?<reg_name>(?:(?&unreserved)\|(?&sub_delims)\|(?&encoded))*)
96	)
97	^(?:(?&reg_name)\\.)*(?&reg_name)\\.?$/ix';
98	/**
99	* Invalid characters in host regular expression.
100	*
101	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
102	*/
103	private const REGEXP_INVALID_HOST_CHARS = '/
104	[:\\/?#\\[\\]@ ] # gen-delims characters as well as the space character
105	/ix';
106	/**
107	* Invalid path for URI without scheme and authority regular expression.
108	*
109	* @link https://tools.ietf.org/html/rfc3986#section-3.3
110	*/
111	private const REGEXP_INVALID_PATH = ',^(([^/]):)(.)?/,';
112	/**
113	* Host and Port splitter regular expression.
114	*/
115	private const REGEXP_HOST_PORT = ',^(?<host>\\[.\\]\|[^:])(:(?<port>.*))?$,';
116	/**
117	* IDN Host detector regular expression.
118	*/
119	private const REGEXP_IDN_PATTERN = '/[^\\x20-\\x7f]/';
120	/**
121	* Only the address block fe80::/10 can have a Zone ID attach to
122	* let's detect the link local significant 10 bits.
123	*/
124	private const ZONE_ID_ADDRESS_BLOCK = "\xfe\x80";
125	/**
126	* Generate an URI string representation from its parsed representation
127	* returned by League\Uri\parse() or PHP's parse_url.
128	*
129	* If you supply your own array, you are responsible for providing
130	* valid components without their URI delimiters.
131	*
132	* @link https://tools.ietf.org/html/rfc3986#section-5.3
133	* @link https://tools.ietf.org/html/rfc3986#section-7.5
134	*
135	* @param array{
136	* scheme:?string,
137	* user:?string,
138	* pass:?string,
139	* host:?string,
140	* port:?int,
141	* path:string,
142	* query:?string,
143	* fragment:?string
144	* } $components
145	*/
146	public static function build(array $components) : string
147	{
148	$result = $components['path'] ?? '';
149	if (isset($components['query'])) {
150	$result .= '?' . $components['query'];
151	}
152	if (isset($components['fragment'])) {
153	$result .= '#' . $components['fragment'];
154	}
155	$scheme = null;
156	if (isset($components['scheme'])) {
157	$scheme = $components['scheme'] . ':';
158	}
159	if (!isset($components['host'])) {
160	return $scheme . $result;
161	}
162	$scheme .= '//';
163	$authority = $components['host'];
164	if (isset($components['port'])) {
165	$authority .= ':' . $components['port'];
166	}
167	if (!isset($components['user'])) {
168	return $scheme . $authority . $result;
169	}
170	$authority = '@' . $authority;
171	if (!isset($components['pass'])) {
172	return $scheme . $components['user'] . $authority . $result;
173	}
174	return $scheme . $components['user'] . ':' . $components['pass'] . $authority . $result;
175	}
176	/**
177	* Parse an URI string into its components.
178	*
179	* This method parses a URI and returns an associative array containing any
180	* of the various components of the URI that are present.
181	*
182	* <code>
183	* $components = (new Parser())->parse('http://foo@test.example.com:42?query#');
184	* var_export($components);
185	* //will display
186	* array(
187	* 'scheme' => 'http', // the URI scheme component
188	* 'user' => 'foo', // the URI user component
189	* 'pass' => null, // the URI pass component
190	* 'host' => 'test.example.com', // the URI host component
191	* 'port' => 42, // the URI port component
192	* 'path' => '', // the URI path component
193	* 'query' => 'query', // the URI query component
194	* 'fragment' => '', // the URI fragment component
195	* );
196	* </code>
197	*
198	* The returned array is similar to PHP's parse_url return value with the following
199	* differences:
200	*
201	* <ul>
202	* <li>All components are always present in the returned array</li>
203	* <li>Empty and undefined component are treated differently. And empty component is
204	* set to the empty string while an undefined component is set to the `null` value.</li>
205	* <li>The path component is never undefined</li>
206	* <li>The method parses the URI following the RFC3986 rules but you are still
207	* required to validate the returned components against its related scheme specific rules.</li>
208	* </ul>
209	*
210	* @link https://tools.ietf.org/html/rfc3986
211	*
212	* @param mixed $uri any scalar or stringable object
213	*
214	* @throws SyntaxError if the URI contains invalid characters
215	* @throws SyntaxError if the URI contains an invalid scheme
216	* @throws SyntaxError if the URI contains an invalid path
217	*
218	* @return array{
219	* scheme:?string,
220	* user:?string,
221	* pass:?string,
222	* host:?string,
223	* port:?int,
224	* path:string,
225	* query:?string,
226	* fragment:?string
227	* }
228	*/
229	public static function parse($uri) : array
230	{
231	if (is_object($uri) && method_exists($uri, '__toString')) {
232	$uri = (string) $uri;
233	}
234	if (!is_scalar($uri)) {
235	throw new \TypeError(sprintf('The uri must be a scalar or a stringable object `%s` given', gettype($uri)));
236	}
237	$uri = (string) $uri;
238	if (isset(self::URI_SCHORTCUTS[$uri])) {
239	/** @var array{scheme:?string, user:?string, pass:?string, host:?string, port:?int, path:string, query:?string, fragment:?string} $components */
240	$components = array_merge(self::URI_COMPONENTS, self::URI_SCHORTCUTS[$uri]);
241	return $components;
242	}
243	if (1 === preg_match(self::REGEXP_INVALID_URI_CHARS, $uri)) {
244	throw new SyntaxError(sprintf('The uri `%s` contains invalid characters', $uri));
245	}
246	//if the first character is a known URI delimiter parsing can be simplified
247	$first_char = $uri[0];
248	//The URI is made of the fragment only
249	if ('#' === $first_char) {
250	[, $fragment] = explode('#', $uri, 2);
251	$components = self::URI_COMPONENTS;
252	$components['fragment'] = $fragment;
253	return $components;
254	}
255	//The URI is made of the query and fragment
256	if ('?' === $first_char) {
257	[, $partial] = explode('?', $uri, 2);
258	[$query, $fragment] = explode('#', $partial, 2) + [1 => null];
259	$components = self::URI_COMPONENTS;
260	$components['query'] = $query;
261	$components['fragment'] = $fragment;
262	return $components;
263	}
264	//use RFC3986 URI regexp to split the URI
265	preg_match(self::REGEXP_URI_PARTS, $uri, $parts);
266	$parts += ['query' => '', 'fragment' => ''];
267	if (':' === $parts['scheme'] \|\| 1 !== preg_match(self::REGEXP_URI_SCHEME, $parts['scontent'])) {
268	throw new SyntaxError(sprintf('The uri `%s` contains an invalid scheme', $uri));
269	}
270	if ('' === $parts['scheme'] . $parts['authority'] && 1 === preg_match(self::REGEXP_INVALID_PATH, $parts['path'])) {
271	throw new SyntaxError(sprintf('The uri `%s` contains an invalid path.', $uri));
272	}
273	/** @var array{scheme:?string, user:?string, pass:?string, host:?string, port:?int, path:string, query:?string, fragment:?string} $components */
274	$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']]);
275	return $components;
276	}
277	/**
278	* Parses the URI authority part.
279	*
280	* @link https://tools.ietf.org/html/rfc3986#section-3.2
281	*
282	* @throws SyntaxError If the port component is invalid
283	*
284	* @return array{user:?string, pass:?string, host:?string, port:?int}
285	*/
286	private static function parseAuthority(string $authority) : array
287	{
288	$components = ['user' => null, 'pass' => null, 'host' => '', 'port' => null];
289	if ('' === $authority) {
290	return $components;
291	}
292	$parts = explode('@', $authority, 2);
293	if (isset($parts[1])) {
294	[$components['user'], $components['pass']] = explode(':', $parts[0], 2) + [1 => null];
295	}
296	preg_match(self::REGEXP_HOST_PORT, $parts[1] ?? $parts[0], $matches);
297	$matches += ['port' => ''];
298	$components['port'] = self::filterPort($matches['port']);
299	$components['host'] = self::filterHost($matches['host']);
300	return $components;
301	}
302	/**
303	* Filter and format the port component.
304	*
305	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
306	*
307	* @throws SyntaxError if the registered name is invalid
308	*/
309	private static function filterPort(string $port) : ?int
310	{
311	if ('' === $port) {
312	return null;
313	}
314	if (1 === preg_match('/^\\d*$/', $port)) {
315	return (int) $port;
316	}
317	throw new SyntaxError(sprintf('The port `%s` is invalid', $port));
318	}
319	/**
320	* Returns whether a hostname is valid.
321	*
322	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
323	*
324	* @throws SyntaxError if the registered name is invalid
325	*/
326	private static function filterHost(string $host) : string
327	{
328	if ('' === $host) {
329	return $host;
330	}
331	if ('[' !== $host[0] \|\| ']' !== substr($host, -1)) {
332	return self::filterRegisteredName($host);
333	}
334	if (!self::isIpHost(substr($host, 1, -1))) {
335	throw new SyntaxError(sprintf('Host `%s` is invalid : the IP host is malformed', $host));
336	}
337	return $host;
338	}
339	/**
340	* Returns whether the host is an IPv4 or a registered named.
341	*
342	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
343	*
344	* @throws SyntaxError if the registered name is invalid
345	* @throws IdnSupportMissing if IDN support or ICU requirement are not available or met.
346	*/
347	private static function filterRegisteredName(string $host) : string
348	{
349	$formatted_host = rawurldecode($host);
350	if (1 === preg_match(self::REGEXP_REGISTERED_NAME, $formatted_host)) {
351	return $host;
352	}
353	//to test IDN host non-ascii characters must be present in the host
354	if (1 !== preg_match(self::REGEXP_IDN_PATTERN, $formatted_host)) {
355	throw new SyntaxError(sprintf('Host `%s` is invalid : the host is not a valid registered name', $host));
356	}
357	$info = Idna::toAscii($host, Idna::IDNA2008_ASCII);
358	if (0 !== $info->errors()) {
359	throw IdnaConversionFailed::dueToIDNAError($host, $info);
360	}
361	return $host;
362	}
363	/**
364	* Validates a IPv6/IPvfuture host.
365	*
366	* @link https://tools.ietf.org/html/rfc3986#section-3.2.2
367	* @link https://tools.ietf.org/html/rfc6874#section-2
368	* @link https://tools.ietf.org/html/rfc6874#section-4
369	*/
370	private static function isIpHost(string $ip_host) : bool
371	{
372	if (\false !== filter_var($ip_host, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6)) {
373	return \true;
374	}
375	if (1 === preg_match(self::REGEXP_IP_FUTURE, $ip_host, $matches)) {
376	return !\in_array($matches['version'], ['4', '6'], \true);
377	}
378	$pos = strpos($ip_host, '%');
379	if (\false === $pos \|\| 1 === preg_match(self::REGEXP_INVALID_HOST_CHARS, rawurldecode(substr($ip_host, $pos)))) {
380	return \false;
381	}
382	$ip_host = substr($ip_host, 0, $pos);
383	return \false !== filter_var($ip_host, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6) && 0 === strpos((string) inet_pton($ip_host), self::ZONE_ID_ADDRESS_BLOCK);
384	}
385	}
386