UnicodeString.php — Independent Analytics – WordPress Analytics Plugin 2.15.0

independent-analytics / vendor / symfony / string / UnicodeString.php

independent-analytics / vendor / symfony / string Last commit date

Exception 2 years ago Inflector 5 days ago Resources 5 days ago Slugger 5 days ago AbstractString.php 5 days ago AbstractUnicodeString.php 5 days ago ByteString.php 5 days ago CodePointString.php 5 days ago LazyString.php 5 days ago UnicodeString.php 5 days ago

UnicodeString.php

285 lines

1	<?php
2
3	/*
4	* This file is part of the Symfony package.
5	*
6	* (c) Fabien Potencier <fabien@symfony.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	namespace IAWPSCOPED\Symfony\Component\String;
12
13	use IAWPSCOPED\Symfony\Component\String\Exception\ExceptionInterface;
14	use IAWPSCOPED\Symfony\Component\String\Exception\InvalidArgumentException;
15	/**
16	* Represents a string of Unicode grapheme clusters encoded as UTF-8.
17	*
18	* A letter followed by combining characters (accents typically) form what Unicode defines
19	* as a grapheme cluster: a character as humans mean it in written texts. This class knows
20	* about the concept and won't split a letter apart from its combining accents. It also
21	* ensures all string comparisons happen on their canonically-composed representation,
22	* ignoring e.g. the order in which accents are listed when a letter has many of them.
23	*
24	* @see https://unicode.org/reports/tr15/
25	*
26	* @author Nicolas Grekas <p@tchwork.com>
27	* @author Hugo Hamon <hugohamon@neuf.fr>
28	*
29	* @throws ExceptionInterface
30	* @internal
31	*/
32	class UnicodeString extends AbstractUnicodeString
33	{
34	public function __construct(string $string = '')
35	{
36	$this->string = \normalizer_is_normalized($string) ? $string : \normalizer_normalize($string);
37	if (\false === $this->string) {
38	throw new InvalidArgumentException('Invalid UTF-8 string.');
39	}
40	}
41	public function append(string ...$suffix) : static
42	{
43	$str = clone $this;
44	$str->string = $this->string . (1 >= \count($suffix) ? $suffix[0] ?? '' : \implode('', $suffix));
45	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
46	if (\false === $str->string) {
47	throw new InvalidArgumentException('Invalid UTF-8 string.');
48	}
49	return $str;
50	}
51	public function chunk(int $length = 1) : array
52	{
53	if (1 > $length) {
54	throw new InvalidArgumentException('The chunk length must be greater than zero.');
55	}
56	if ('' === $this->string) {
57	return [];
58	}
59	$rx = '/(';
60	while (65535 < $length) {
61	$rx .= '\\X{65535}';
62	$length -= 65535;
63	}
64	$rx .= '\\X{' . $length . '})/u';
65	$str = clone $this;
66	$chunks = [];
67	foreach (\preg_split($rx, $this->string, -1, \PREG_SPLIT_DELIM_CAPTURE \| \PREG_SPLIT_NO_EMPTY) as $chunk) {
68	$str->string = $chunk;
69	$chunks[] = clone $str;
70	}
71	return $chunks;
72	}
73	public function endsWith(string\|iterable\|AbstractString $suffix) : bool
74	{
75	if ($suffix instanceof AbstractString) {
76	$suffix = $suffix->string;
77	} elseif (!\is_string($suffix)) {
78	return parent::endsWith($suffix);
79	}
80	$form = null === $this->ignoreCase ? \Normalizer::NFD : \Normalizer::NFC;
81	\normalizer_is_normalized($suffix, $form) ?: ($suffix = \normalizer_normalize($suffix, $form));
82	if ('' === $suffix \|\| \false === $suffix) {
83	return \false;
84	}
85	if ($this->ignoreCase) {
86	return 0 === \mb_stripos(\grapheme_extract($this->string, \strlen($suffix), \GRAPHEME_EXTR_MAXBYTES, \strlen($this->string) - \strlen($suffix)), $suffix, 0, 'UTF-8');
87	}
88	return $suffix === \grapheme_extract($this->string, \strlen($suffix), \GRAPHEME_EXTR_MAXBYTES, \strlen($this->string) - \strlen($suffix));
89	}
90	public function equalsTo(string\|iterable\|AbstractString $string) : bool
91	{
92	if ($string instanceof AbstractString) {
93	$string = $string->string;
94	} elseif (!\is_string($string)) {
95	return parent::equalsTo($string);
96	}
97	$form = null === $this->ignoreCase ? \Normalizer::NFD : \Normalizer::NFC;
98	\normalizer_is_normalized($string, $form) ?: ($string = \normalizer_normalize($string, $form));
99	if ('' !== $string && \false !== $string && $this->ignoreCase) {
100	return \strlen($string) === \strlen($this->string) && 0 === \mb_stripos($this->string, $string, 0, 'UTF-8');
101	}
102	return $string === $this->string;
103	}
104	public function indexOf(string\|iterable\|AbstractString $needle, int $offset = 0) : ?int
105	{
106	if ($needle instanceof AbstractString) {
107	$needle = $needle->string;
108	} elseif (!\is_string($needle)) {
109	return parent::indexOf($needle, $offset);
110	}
111	$form = null === $this->ignoreCase ? \Normalizer::NFD : \Normalizer::NFC;
112	\normalizer_is_normalized($needle, $form) ?: ($needle = \normalizer_normalize($needle, $form));
113	if ('' === $needle \|\| \false === $needle) {
114	return null;
115	}
116	try {
117	$i = $this->ignoreCase ? \grapheme_stripos($this->string, $needle, $offset) : \grapheme_strpos($this->string, $needle, $offset);
118	} catch (\ValueError $e) {
119	return null;
120	}
121	return \false === $i ? null : $i;
122	}
123	public function indexOfLast(string\|iterable\|AbstractString $needle, int $offset = 0) : ?int
124	{
125	if ($needle instanceof AbstractString) {
126	$needle = $needle->string;
127	} elseif (!\is_string($needle)) {
128	return parent::indexOfLast($needle, $offset);
129	}
130	$form = null === $this->ignoreCase ? \Normalizer::NFD : \Normalizer::NFC;
131	\normalizer_is_normalized($needle, $form) ?: ($needle = \normalizer_normalize($needle, $form));
132	if ('' === $needle \|\| \false === $needle) {
133	return null;
134	}
135	$string = $this->string;
136	if (0 > $offset) {
137	// workaround https://bugs.php.net/74264
138	if (0 > ($offset += \grapheme_strlen($needle))) {
139	$string = \grapheme_substr($string, 0, $offset);
140	}
141	$offset = 0;
142	}
143	$i = $this->ignoreCase ? \grapheme_strripos($string, $needle, $offset) : \grapheme_strrpos($string, $needle, $offset);
144	return \false === $i ? null : $i;
145	}
146	public function join(array $strings, string $lastGlue = null) : static
147	{
148	$str = parent::join($strings, $lastGlue);
149	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
150	return $str;
151	}
152	public function length() : int
153	{
154	return \grapheme_strlen($this->string);
155	}
156	public function normalize(int $form = self::NFC) : static
157	{
158	$str = clone $this;
159	if (\in_array($form, [self::NFC, self::NFKC], \true)) {
160	\normalizer_is_normalized($str->string, $form) ?: ($str->string = \normalizer_normalize($str->string, $form));
161	} elseif (!\in_array($form, [self::NFD, self::NFKD], \true)) {
162	throw new InvalidArgumentException('Unsupported normalization form.');
163	} elseif (!\normalizer_is_normalized($str->string, $form)) {
164	$str->string = \normalizer_normalize($str->string, $form);
165	$str->ignoreCase = null;
166	}
167	return $str;
168	}
169	public function prepend(string ...$prefix) : static
170	{
171	$str = clone $this;
172	$str->string = (1 >= \count($prefix) ? $prefix[0] ?? '' : \implode('', $prefix)) . $this->string;
173	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
174	if (\false === $str->string) {
175	throw new InvalidArgumentException('Invalid UTF-8 string.');
176	}
177	return $str;
178	}
179	public function replace(string $from, string $to) : static
180	{
181	$str = clone $this;
182	\normalizer_is_normalized($from) ?: ($from = \normalizer_normalize($from));
183	if ('' !== $from && \false !== $from) {
184	$tail = $str->string;
185	$result = '';
186	$indexOf = $this->ignoreCase ? 'grapheme_stripos' : 'grapheme_strpos';
187	while ('' !== $tail && \false !== ($i = $indexOf($tail, $from))) {
188	$slice = \grapheme_substr($tail, 0, $i);
189	$result .= $slice . $to;
190	$tail = \substr($tail, \strlen($slice) + \strlen($from));
191	}
192	$str->string = $result . $tail;
193	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
194	if (\false === $str->string) {
195	throw new InvalidArgumentException('Invalid UTF-8 string.');
196	}
197	}
198	return $str;
199	}
200	public function replaceMatches(string $fromRegexp, string\|callable $to) : static
201	{
202	$str = parent::replaceMatches($fromRegexp, $to);
203	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
204	return $str;
205	}
206	public function slice(int $start = 0, int $length = null) : static
207	{
208	$str = clone $this;
209	$str->string = (string) \grapheme_substr($this->string, $start, $length ?? 2147483647);
210	return $str;
211	}
212	public function splice(string $replacement, int $start = 0, int $length = null) : static
213	{
214	$str = clone $this;
215	$start = $start ? \strlen(\grapheme_substr($this->string, 0, $start)) : 0;
216	$length = $length ? \strlen(\grapheme_substr($this->string, $start, $length ?? 2147483647)) : $length;
217	$str->string = \substr_replace($this->string, $replacement, $start, $length ?? 2147483647);
218	\normalizer_is_normalized($str->string) ?: ($str->string = \normalizer_normalize($str->string));
219	if (\false === $str->string) {
220	throw new InvalidArgumentException('Invalid UTF-8 string.');
221	}
222	return $str;
223	}
224	public function split(string $delimiter, int $limit = null, int $flags = null) : array
225	{
226	if (1 > ($limit = $limit ?? 2147483647)) {
227	throw new InvalidArgumentException('Split limit must be a positive integer.');
228	}
229	if ('' === $delimiter) {
230	throw new InvalidArgumentException('Split delimiter is empty.');
231	}
232	if (null !== $flags) {
233	return parent::split($delimiter . 'u', $limit, $flags);
234	}
235	\normalizer_is_normalized($delimiter) ?: ($delimiter = \normalizer_normalize($delimiter));
236	if (\false === $delimiter) {
237	throw new InvalidArgumentException('Split delimiter is not a valid UTF-8 string.');
238	}
239	$str = clone $this;
240	$tail = $this->string;
241	$chunks = [];
242	$indexOf = $this->ignoreCase ? 'grapheme_stripos' : 'grapheme_strpos';
243	while (1 < $limit && \false !== ($i = $indexOf($tail, $delimiter))) {
244	$str->string = \grapheme_substr($tail, 0, $i);
245	$chunks[] = clone $str;
246	$tail = \substr($tail, \strlen($str->string) + \strlen($delimiter));
247	--$limit;
248	}
249	$str->string = $tail;
250	$chunks[] = clone $str;
251	return $chunks;
252	}
253	public function startsWith(string\|iterable\|AbstractString $prefix) : bool
254	{
255	if ($prefix instanceof AbstractString) {
256	$prefix = $prefix->string;
257	} elseif (!\is_string($prefix)) {
258	return parent::startsWith($prefix);
259	}
260	$form = null === $this->ignoreCase ? \Normalizer::NFD : \Normalizer::NFC;
261	\normalizer_is_normalized($prefix, $form) ?: ($prefix = \normalizer_normalize($prefix, $form));
262	if ('' === $prefix \|\| \false === $prefix) {
263	return \false;
264	}
265	if ($this->ignoreCase) {
266	return 0 === \mb_stripos(\grapheme_extract($this->string, \strlen($prefix), \GRAPHEME_EXTR_MAXBYTES), $prefix, 0, 'UTF-8');
267	}
268	return $prefix === \grapheme_extract($this->string, \strlen($prefix), \GRAPHEME_EXTR_MAXBYTES);
269	}
270	public function __wakeup()
271	{
272	if (!\is_string($this->string)) {
273	throw new \BadMethodCallException('Cannot unserialize ' . __CLASS__);
274	}
275	\normalizer_is_normalized($this->string) ?: ($this->string = \normalizer_normalize($this->string));
276	}
277	public function __clone()
278	{
279	if (null === $this->ignoreCase) {
280	\normalizer_is_normalized($this->string) ?: ($this->string = \normalizer_normalize($this->string));
281	}
282	$this->ignoreCase = \false;
283	}
284	}
285