Mbstring.php 28 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873
  1. <?php
  2. /*
  3. * This file is part of the Symfony package.
  4. *
  5. * (c) Fabien Potencier <fabien@symfony.com>
  6. *
  7. * For the full copyright and license information, please view the LICENSE
  8. * file that was distributed with this source code.
  9. */
  10. namespace Symfony\Polyfill\Mbstring;
  11. /**
  12. * Partial mbstring implementation in PHP, iconv based, UTF-8 centric.
  13. *
  14. * Implemented:
  15. * - mb_chr - Returns a specific character from its Unicode code point
  16. * - mb_convert_encoding - Convert character encoding
  17. * - mb_convert_variables - Convert character code in variable(s)
  18. * - mb_decode_mimeheader - Decode string in MIME header field
  19. * - mb_encode_mimeheader - Encode string for MIME header XXX NATIVE IMPLEMENTATION IS REALLY BUGGED
  20. * - mb_decode_numericentity - Decode HTML numeric string reference to character
  21. * - mb_encode_numericentity - Encode character to HTML numeric string reference
  22. * - mb_convert_case - Perform case folding on a string
  23. * - mb_detect_encoding - Detect character encoding
  24. * - mb_get_info - Get internal settings of mbstring
  25. * - mb_http_input - Detect HTTP input character encoding
  26. * - mb_http_output - Set/Get HTTP output character encoding
  27. * - mb_internal_encoding - Set/Get internal character encoding
  28. * - mb_list_encodings - Returns an array of all supported encodings
  29. * - mb_ord - Returns the Unicode code point of a character
  30. * - mb_output_handler - Callback function converts character encoding in output buffer
  31. * - mb_scrub - Replaces ill-formed byte sequences with substitute characters
  32. * - mb_strlen - Get string length
  33. * - mb_strpos - Find position of first occurrence of string in a string
  34. * - mb_strrpos - Find position of last occurrence of a string in a string
  35. * - mb_str_split - Convert a string to an array
  36. * - mb_strtolower - Make a string lowercase
  37. * - mb_strtoupper - Make a string uppercase
  38. * - mb_substitute_character - Set/Get substitution character
  39. * - mb_substr - Get part of string
  40. * - mb_stripos - Finds position of first occurrence of a string within another, case insensitive
  41. * - mb_stristr - Finds first occurrence of a string within another, case insensitive
  42. * - mb_strrchr - Finds the last occurrence of a character in a string within another
  43. * - mb_strrichr - Finds the last occurrence of a character in a string within another, case insensitive
  44. * - mb_strripos - Finds position of last occurrence of a string within another, case insensitive
  45. * - mb_strstr - Finds first occurrence of a string within another
  46. * - mb_strwidth - Return width of string
  47. * - mb_substr_count - Count the number of substring occurrences
  48. *
  49. * Not implemented:
  50. * - mb_convert_kana - Convert "kana" one from another ("zen-kaku", "han-kaku" and more)
  51. * - mb_ereg_* - Regular expression with multibyte support
  52. * - mb_parse_str - Parse GET/POST/COOKIE data and set global variable
  53. * - mb_preferred_mime_name - Get MIME charset string
  54. * - mb_regex_encoding - Returns current encoding for multibyte regex as string
  55. * - mb_regex_set_options - Set/Get the default options for mbregex functions
  56. * - mb_send_mail - Send encoded mail
  57. * - mb_split - Split multibyte string using regular expression
  58. * - mb_strcut - Get part of string
  59. * - mb_strimwidth - Get truncated string with specified width
  60. *
  61. * @author Nicolas Grekas <p@tchwork.com>
  62. *
  63. * @internal
  64. */
  65. final class Mbstring
  66. {
  67. public const MB_CASE_FOLD = \PHP_INT_MAX;
  68. private const CASE_FOLD = [
  69. ['µ', 'ſ', "\xCD\x85", 'ς', "\xCF\x90", "\xCF\x91", "\xCF\x95", "\xCF\x96", "\xCF\xB0", "\xCF\xB1", "\xCF\xB5", "\xE1\xBA\x9B", "\xE1\xBE\xBE"],
  70. ['μ', 's', 'ι', 'σ', 'β', 'θ', 'φ', 'π', 'κ', 'ρ', 'ε', "\xE1\xB9\xA1", 'ι'],
  71. ];
  72. private static $encodingList = ['ASCII', 'UTF-8'];
  73. private static $language = 'neutral';
  74. private static $internalEncoding = 'UTF-8';
  75. public static function mb_convert_encoding($s, $toEncoding, $fromEncoding = null)
  76. {
  77. if (\is_array($fromEncoding) || ($fromEncoding !== null && false !== strpos($fromEncoding, ','))) {
  78. $fromEncoding = self::mb_detect_encoding($s, $fromEncoding);
  79. } else {
  80. $fromEncoding = self::getEncoding($fromEncoding);
  81. }
  82. $toEncoding = self::getEncoding($toEncoding);
  83. if ('BASE64' === $fromEncoding) {
  84. $s = base64_decode($s);
  85. $fromEncoding = $toEncoding;
  86. }
  87. if ('BASE64' === $toEncoding) {
  88. return base64_encode($s);
  89. }
  90. if ('HTML-ENTITIES' === $toEncoding || 'HTML' === $toEncoding) {
  91. if ('HTML-ENTITIES' === $fromEncoding || 'HTML' === $fromEncoding) {
  92. $fromEncoding = 'Windows-1252';
  93. }
  94. if ('UTF-8' !== $fromEncoding) {
  95. $s = \iconv($fromEncoding, 'UTF-8//IGNORE', $s);
  96. }
  97. return preg_replace_callback('/[\x80-\xFF]+/', [__CLASS__, 'html_encoding_callback'], $s);
  98. }
  99. if ('HTML-ENTITIES' === $fromEncoding) {
  100. $s = html_entity_decode($s, \ENT_COMPAT, 'UTF-8');
  101. $fromEncoding = 'UTF-8';
  102. }
  103. return \iconv($fromEncoding, $toEncoding.'//IGNORE', $s);
  104. }
  105. public static function mb_convert_variables($toEncoding, $fromEncoding, &...$vars)
  106. {
  107. $ok = true;
  108. array_walk_recursive($vars, function (&$v) use (&$ok, $toEncoding, $fromEncoding) {
  109. if (false === $v = self::mb_convert_encoding($v, $toEncoding, $fromEncoding)) {
  110. $ok = false;
  111. }
  112. });
  113. return $ok ? $fromEncoding : false;
  114. }
  115. public static function mb_decode_mimeheader($s)
  116. {
  117. return \iconv_mime_decode($s, 2, self::$internalEncoding);
  118. }
  119. public static function mb_encode_mimeheader($s, $charset = null, $transferEncoding = null, $linefeed = null, $indent = null)
  120. {
  121. trigger_error('mb_encode_mimeheader() is bugged. Please use iconv_mime_encode() instead', \E_USER_WARNING);
  122. }
  123. public static function mb_decode_numericentity($s, $convmap, $encoding = null)
  124. {
  125. if (null !== $s && !is_scalar($s) && !(\is_object($s) && method_exists($s, '__toString'))) {
  126. trigger_error('mb_decode_numericentity() expects parameter 1 to be string, '.\gettype($s).' given', \E_USER_WARNING);
  127. return null;
  128. }
  129. if (!\is_array($convmap) || (80000 > \PHP_VERSION_ID && !$convmap)) {
  130. return false;
  131. }
  132. if (null !== $encoding && !is_scalar($encoding)) {
  133. trigger_error('mb_decode_numericentity() expects parameter 3 to be string, '.\gettype($s).' given', \E_USER_WARNING);
  134. return ''; // Instead of null (cf. mb_encode_numericentity).
  135. }
  136. $s = (string) $s;
  137. if ('' === $s) {
  138. return '';
  139. }
  140. $encoding = self::getEncoding($encoding);
  141. if ('UTF-8' === $encoding) {
  142. $encoding = null;
  143. if (!preg_match('//u', $s)) {
  144. $s = @\iconv('UTF-8', 'UTF-8//IGNORE', $s);
  145. }
  146. } else {
  147. $s = \iconv($encoding, 'UTF-8//IGNORE', $s);
  148. }
  149. $cnt = floor(\count($convmap) / 4) * 4;
  150. for ($i = 0; $i < $cnt; $i += 4) {
  151. // collector_decode_htmlnumericentity ignores $convmap[$i + 3]
  152. $convmap[$i] += $convmap[$i + 2];
  153. $convmap[$i + 1] += $convmap[$i + 2];
  154. }
  155. $s = preg_replace_callback('/&#(?:0*([0-9]+)|x0*([0-9a-fA-F]+))(?!&);?/', function (array $m) use ($cnt, $convmap) {
  156. $c = isset($m[2]) ? (int) hexdec($m[2]) : $m[1];
  157. for ($i = 0; $i < $cnt; $i += 4) {
  158. if ($c >= $convmap[$i] && $c <= $convmap[$i + 1]) {
  159. return self::mb_chr($c - $convmap[$i + 2]);
  160. }
  161. }
  162. return $m[0];
  163. }, $s);
  164. if (null === $encoding) {
  165. return $s;
  166. }
  167. return \iconv('UTF-8', $encoding.'//IGNORE', $s);
  168. }
  169. public static function mb_encode_numericentity($s, $convmap, $encoding = null, $is_hex = false)
  170. {
  171. if (null !== $s && !is_scalar($s) && !(\is_object($s) && method_exists($s, '__toString'))) {
  172. trigger_error('mb_encode_numericentity() expects parameter 1 to be string, '.\gettype($s).' given', \E_USER_WARNING);
  173. return null;
  174. }
  175. if (!\is_array($convmap) || (80000 > \PHP_VERSION_ID && !$convmap)) {
  176. return false;
  177. }
  178. if (null !== $encoding && !is_scalar($encoding)) {
  179. trigger_error('mb_encode_numericentity() expects parameter 3 to be string, '.\gettype($s).' given', \E_USER_WARNING);
  180. return null; // Instead of '' (cf. mb_decode_numericentity).
  181. }
  182. if (null !== $is_hex && !is_scalar($is_hex)) {
  183. trigger_error('mb_encode_numericentity() expects parameter 4 to be boolean, '.\gettype($s).' given', \E_USER_WARNING);
  184. return null;
  185. }
  186. $s = (string) $s;
  187. if ('' === $s) {
  188. return '';
  189. }
  190. $encoding = self::getEncoding($encoding);
  191. if ('UTF-8' === $encoding) {
  192. $encoding = null;
  193. if (!preg_match('//u', $s)) {
  194. $s = @\iconv('UTF-8', 'UTF-8//IGNORE', $s);
  195. }
  196. } else {
  197. $s = \iconv($encoding, 'UTF-8//IGNORE', $s);
  198. }
  199. static $ulenMask = ["\xC0" => 2, "\xD0" => 2, "\xE0" => 3, "\xF0" => 4];
  200. $cnt = floor(\count($convmap) / 4) * 4;
  201. $i = 0;
  202. $len = \strlen($s);
  203. $result = '';
  204. while ($i < $len) {
  205. $ulen = $s[$i] < "\x80" ? 1 : $ulenMask[$s[$i] & "\xF0"];
  206. $uchr = substr($s, $i, $ulen);
  207. $i += $ulen;
  208. $c = self::mb_ord($uchr);
  209. for ($j = 0; $j < $cnt; $j += 4) {
  210. if ($c >= $convmap[$j] && $c <= $convmap[$j + 1]) {
  211. $cOffset = ($c + $convmap[$j + 2]) & $convmap[$j + 3];
  212. $result .= $is_hex ? sprintf('&#x%X;', $cOffset) : '&#'.$cOffset.';';
  213. continue 2;
  214. }
  215. }
  216. $result .= $uchr;
  217. }
  218. if (null === $encoding) {
  219. return $result;
  220. }
  221. return \iconv('UTF-8', $encoding.'//IGNORE', $result);
  222. }
  223. public static function mb_convert_case($s, $mode, $encoding = null)
  224. {
  225. $s = (string) $s;
  226. if ('' === $s) {
  227. return '';
  228. }
  229. $encoding = self::getEncoding($encoding);
  230. if ('UTF-8' === $encoding) {
  231. $encoding = null;
  232. if (!preg_match('//u', $s)) {
  233. $s = @\iconv('UTF-8', 'UTF-8//IGNORE', $s);
  234. }
  235. } else {
  236. $s = \iconv($encoding, 'UTF-8//IGNORE', $s);
  237. }
  238. if (\MB_CASE_TITLE == $mode) {
  239. static $titleRegexp = null;
  240. if (null === $titleRegexp) {
  241. $titleRegexp = self::getData('titleCaseRegexp');
  242. }
  243. $s = preg_replace_callback($titleRegexp, [__CLASS__, 'title_case'], $s);
  244. } else {
  245. if (\MB_CASE_UPPER == $mode) {
  246. static $upper = null;
  247. if (null === $upper) {
  248. $upper = self::getData('upperCase');
  249. }
  250. $map = $upper;
  251. } else {
  252. if (self::MB_CASE_FOLD === $mode) {
  253. $s = str_replace(self::CASE_FOLD[0], self::CASE_FOLD[1], $s);
  254. }
  255. static $lower = null;
  256. if (null === $lower) {
  257. $lower = self::getData('lowerCase');
  258. }
  259. $map = $lower;
  260. }
  261. static $ulenMask = ["\xC0" => 2, "\xD0" => 2, "\xE0" => 3, "\xF0" => 4];
  262. $i = 0;
  263. $len = \strlen($s);
  264. while ($i < $len) {
  265. $ulen = $s[$i] < "\x80" ? 1 : $ulenMask[$s[$i] & "\xF0"];
  266. $uchr = substr($s, $i, $ulen);
  267. $i += $ulen;
  268. if (isset($map[$uchr])) {
  269. $uchr = $map[$uchr];
  270. $nlen = \strlen($uchr);
  271. if ($nlen == $ulen) {
  272. $nlen = $i;
  273. do {
  274. $s[--$nlen] = $uchr[--$ulen];
  275. } while ($ulen);
  276. } else {
  277. $s = substr_replace($s, $uchr, $i - $ulen, $ulen);
  278. $len += $nlen - $ulen;
  279. $i += $nlen - $ulen;
  280. }
  281. }
  282. }
  283. }
  284. if (null === $encoding) {
  285. return $s;
  286. }
  287. return \iconv('UTF-8', $encoding.'//IGNORE', $s);
  288. }
  289. public static function mb_internal_encoding($encoding = null)
  290. {
  291. if (null === $encoding) {
  292. return self::$internalEncoding;
  293. }
  294. $normalizedEncoding = self::getEncoding($encoding);
  295. if ('UTF-8' === $normalizedEncoding || false !== @\iconv($normalizedEncoding, $normalizedEncoding, ' ')) {
  296. self::$internalEncoding = $normalizedEncoding;
  297. return true;
  298. }
  299. if (80000 > \PHP_VERSION_ID) {
  300. return false;
  301. }
  302. throw new \ValueError(sprintf('Argument #1 ($encoding) must be a valid encoding, "%s" given', $encoding));
  303. }
  304. public static function mb_language($lang = null)
  305. {
  306. if (null === $lang) {
  307. return self::$language;
  308. }
  309. switch ($normalizedLang = strtolower($lang)) {
  310. case 'uni':
  311. case 'neutral':
  312. self::$language = $normalizedLang;
  313. return true;
  314. }
  315. if (80000 > \PHP_VERSION_ID) {
  316. return false;
  317. }
  318. throw new \ValueError(sprintf('Argument #1 ($language) must be a valid language, "%s" given', $lang));
  319. }
  320. public static function mb_list_encodings()
  321. {
  322. return ['UTF-8'];
  323. }
  324. public static function mb_encoding_aliases($encoding)
  325. {
  326. switch (strtoupper($encoding)) {
  327. case 'UTF8':
  328. case 'UTF-8':
  329. return ['utf8'];
  330. }
  331. return false;
  332. }
  333. public static function mb_check_encoding($var = null, $encoding = null)
  334. {
  335. if (null === $encoding) {
  336. if (null === $var) {
  337. return false;
  338. }
  339. $encoding = self::$internalEncoding;
  340. }
  341. return self::mb_detect_encoding($var, [$encoding]) || false !== @\iconv($encoding, $encoding, $var);
  342. }
  343. public static function mb_detect_encoding($str, $encodingList = null, $strict = false)
  344. {
  345. if (null === $encodingList) {
  346. $encodingList = self::$encodingList;
  347. } else {
  348. if (!\is_array($encodingList)) {
  349. $encodingList = array_map('trim', explode(',', $encodingList));
  350. }
  351. $encodingList = array_map('strtoupper', $encodingList);
  352. }
  353. foreach ($encodingList as $enc) {
  354. switch ($enc) {
  355. case 'ASCII':
  356. if (!preg_match('/[\x80-\xFF]/', $str)) {
  357. return $enc;
  358. }
  359. break;
  360. case 'UTF8':
  361. case 'UTF-8':
  362. if (preg_match('//u', $str)) {
  363. return 'UTF-8';
  364. }
  365. break;
  366. default:
  367. if (0 === strncmp($enc, 'ISO-8859-', 9)) {
  368. return $enc;
  369. }
  370. }
  371. }
  372. return false;
  373. }
  374. public static function mb_detect_order($encodingList = null)
  375. {
  376. if (null === $encodingList) {
  377. return self::$encodingList;
  378. }
  379. if (!\is_array($encodingList)) {
  380. $encodingList = array_map('trim', explode(',', $encodingList));
  381. }
  382. $encodingList = array_map('strtoupper', $encodingList);
  383. foreach ($encodingList as $enc) {
  384. switch ($enc) {
  385. default:
  386. if (strncmp($enc, 'ISO-8859-', 9)) {
  387. return false;
  388. }
  389. // no break
  390. case 'ASCII':
  391. case 'UTF8':
  392. case 'UTF-8':
  393. }
  394. }
  395. self::$encodingList = $encodingList;
  396. return true;
  397. }
  398. public static function mb_strlen($s, $encoding = null)
  399. {
  400. $encoding = self::getEncoding($encoding);
  401. if ('CP850' === $encoding || 'ASCII' === $encoding) {
  402. return \strlen($s);
  403. }
  404. return @\iconv_strlen($s, $encoding);
  405. }
  406. public static function mb_strpos($haystack, $needle, $offset = 0, $encoding = null)
  407. {
  408. $encoding = self::getEncoding($encoding);
  409. if ('CP850' === $encoding || 'ASCII' === $encoding) {
  410. return strpos($haystack, $needle, $offset);
  411. }
  412. $needle = (string) $needle;
  413. if ('' === $needle) {
  414. if (80000 > \PHP_VERSION_ID) {
  415. trigger_error(__METHOD__.': Empty delimiter', \E_USER_WARNING);
  416. return false;
  417. }
  418. return 0;
  419. }
  420. return \iconv_strpos($haystack, $needle, $offset, $encoding);
  421. }
  422. public static function mb_strrpos($haystack, $needle, $offset = 0, $encoding = null)
  423. {
  424. $encoding = self::getEncoding($encoding);
  425. if ('CP850' === $encoding || 'ASCII' === $encoding) {
  426. return strrpos($haystack, $needle, $offset);
  427. }
  428. if ($offset != (int) $offset) {
  429. $offset = 0;
  430. } elseif ($offset = (int) $offset) {
  431. if ($offset < 0) {
  432. if (0 > $offset += self::mb_strlen($needle)) {
  433. $haystack = self::mb_substr($haystack, 0, $offset, $encoding);
  434. }
  435. $offset = 0;
  436. } else {
  437. $haystack = self::mb_substr($haystack, $offset, 2147483647, $encoding);
  438. }
  439. }
  440. $pos = '' !== $needle || 80000 > \PHP_VERSION_ID
  441. ? \iconv_strrpos($haystack, $needle, $encoding)
  442. : self::mb_strlen($haystack, $encoding);
  443. return false !== $pos ? $offset + $pos : false;
  444. }
  445. public static function mb_str_split($string, $split_length = 1, $encoding = null)
  446. {
  447. if (null !== $string && !is_scalar($string) && !(\is_object($string) && method_exists($string, '__toString'))) {
  448. trigger_error('mb_str_split() expects parameter 1 to be string, '.\gettype($string).' given', \E_USER_WARNING);
  449. return null;
  450. }
  451. if (1 > $split_length = (int) $split_length) {
  452. if (80000 > \PHP_VERSION_ID) {
  453. trigger_error('The length of each segment must be greater than zero', \E_USER_WARNING);
  454. return false;
  455. }
  456. throw new \ValueError('Argument #2 ($length) must be greater than 0');
  457. }
  458. if (null === $encoding) {
  459. $encoding = mb_internal_encoding();
  460. }
  461. if ('UTF-8' === $encoding = self::getEncoding($encoding)) {
  462. $rx = '/(';
  463. while (65535 < $split_length) {
  464. $rx .= '.{65535}';
  465. $split_length -= 65535;
  466. }
  467. $rx .= '.{'.$split_length.'})/us';
  468. return preg_split($rx, $string, null, \PREG_SPLIT_DELIM_CAPTURE | \PREG_SPLIT_NO_EMPTY);
  469. }
  470. $result = [];
  471. $length = mb_strlen($string, $encoding);
  472. for ($i = 0; $i < $length; $i += $split_length) {
  473. $result[] = mb_substr($string, $i, $split_length, $encoding);
  474. }
  475. return $result;
  476. }
  477. public static function mb_strtolower($s, $encoding = null)
  478. {
  479. return self::mb_convert_case($s, \MB_CASE_LOWER, $encoding);
  480. }
  481. public static function mb_strtoupper($s, $encoding = null)
  482. {
  483. return self::mb_convert_case($s, \MB_CASE_UPPER, $encoding);
  484. }
  485. public static function mb_substitute_character($c = null)
  486. {
  487. if (null === $c) {
  488. return 'none';
  489. }
  490. if (0 === strcasecmp($c, 'none')) {
  491. return true;
  492. }
  493. if (80000 > \PHP_VERSION_ID) {
  494. return false;
  495. }
  496. if (\is_int($c) || 'long' === $c || 'entity' === $c) {
  497. return false;
  498. }
  499. throw new \ValueError('Argument #1 ($substitute_character) must be "none", "long", "entity" or a valid codepoint');
  500. }
  501. public static function mb_substr($s, $start, $length = null, $encoding = null)
  502. {
  503. $encoding = self::getEncoding($encoding);
  504. if ('CP850' === $encoding || 'ASCII' === $encoding) {
  505. return (string) substr($s, $start, null === $length ? 2147483647 : $length);
  506. }
  507. if ($start < 0) {
  508. $start = \iconv_strlen($s, $encoding) + $start;
  509. if ($start < 0) {
  510. $start = 0;
  511. }
  512. }
  513. if (null === $length) {
  514. $length = 2147483647;
  515. } elseif ($length < 0) {
  516. $length = \iconv_strlen($s, $encoding) + $length - $start;
  517. if ($length < 0) {
  518. return '';
  519. }
  520. }
  521. return (string) \iconv_substr($s, $start, $length, $encoding);
  522. }
  523. public static function mb_stripos($haystack, $needle, $offset = 0, $encoding = null)
  524. {
  525. $haystack = self::mb_convert_case($haystack, self::MB_CASE_FOLD, $encoding);
  526. $needle = self::mb_convert_case($needle, self::MB_CASE_FOLD, $encoding);
  527. return self::mb_strpos($haystack, $needle, $offset, $encoding);
  528. }
  529. public static function mb_stristr($haystack, $needle, $part = false, $encoding = null)
  530. {
  531. $pos = self::mb_stripos($haystack, $needle, 0, $encoding);
  532. return self::getSubpart($pos, $part, $haystack, $encoding);
  533. }
  534. public static function mb_strrchr($haystack, $needle, $part = false, $encoding = null)
  535. {
  536. $encoding = self::getEncoding($encoding);
  537. if ('CP850' === $encoding || 'ASCII' === $encoding) {
  538. $pos = strrpos($haystack, $needle);
  539. } else {
  540. $needle = self::mb_substr($needle, 0, 1, $encoding);
  541. $pos = \iconv_strrpos($haystack, $needle, $encoding);
  542. }
  543. return self::getSubpart($pos, $part, $haystack, $encoding);
  544. }
  545. public static function mb_strrichr($haystack, $needle, $part = false, $encoding = null)
  546. {
  547. $needle = self::mb_substr($needle, 0, 1, $encoding);
  548. $pos = self::mb_strripos($haystack, $needle, $encoding);
  549. return self::getSubpart($pos, $part, $haystack, $encoding);
  550. }
  551. public static function mb_strripos($haystack, $needle, $offset = 0, $encoding = null)
  552. {
  553. $haystack = self::mb_convert_case($haystack, self::MB_CASE_FOLD, $encoding);
  554. $needle = self::mb_convert_case($needle, self::MB_CASE_FOLD, $encoding);
  555. return self::mb_strrpos($haystack, $needle, $offset, $encoding);
  556. }
  557. public static function mb_strstr($haystack, $needle, $part = false, $encoding = null)
  558. {
  559. $pos = strpos($haystack, $needle);
  560. if (false === $pos) {
  561. return false;
  562. }
  563. if ($part) {
  564. return substr($haystack, 0, $pos);
  565. }
  566. return substr($haystack, $pos);
  567. }
  568. public static function mb_get_info($type = 'all')
  569. {
  570. $info = [
  571. 'internal_encoding' => self::$internalEncoding,
  572. 'http_output' => 'pass',
  573. 'http_output_conv_mimetypes' => '^(text/|application/xhtml\+xml)',
  574. 'func_overload' => 0,
  575. 'func_overload_list' => 'no overload',
  576. 'mail_charset' => 'UTF-8',
  577. 'mail_header_encoding' => 'BASE64',
  578. 'mail_body_encoding' => 'BASE64',
  579. 'illegal_chars' => 0,
  580. 'encoding_translation' => 'Off',
  581. 'language' => self::$language,
  582. 'detect_order' => self::$encodingList,
  583. 'substitute_character' => 'none',
  584. 'strict_detection' => 'Off',
  585. ];
  586. if ('all' === $type) {
  587. return $info;
  588. }
  589. if (isset($info[$type])) {
  590. return $info[$type];
  591. }
  592. return false;
  593. }
  594. public static function mb_http_input($type = '')
  595. {
  596. return false;
  597. }
  598. public static function mb_http_output($encoding = null)
  599. {
  600. return null !== $encoding ? 'pass' === $encoding : 'pass';
  601. }
  602. public static function mb_strwidth($s, $encoding = null)
  603. {
  604. $encoding = self::getEncoding($encoding);
  605. if ('UTF-8' !== $encoding) {
  606. $s = \iconv($encoding, 'UTF-8//IGNORE', $s);
  607. }
  608. $s = preg_replace('/[\x{1100}-\x{115F}\x{2329}\x{232A}\x{2E80}-\x{303E}\x{3040}-\x{A4CF}\x{AC00}-\x{D7A3}\x{F900}-\x{FAFF}\x{FE10}-\x{FE19}\x{FE30}-\x{FE6F}\x{FF00}-\x{FF60}\x{FFE0}-\x{FFE6}\x{20000}-\x{2FFFD}\x{30000}-\x{3FFFD}]/u', '', $s, -1, $wide);
  609. return ($wide << 1) + \iconv_strlen($s, 'UTF-8');
  610. }
  611. public static function mb_substr_count($haystack, $needle, $encoding = null)
  612. {
  613. return substr_count($haystack, $needle);
  614. }
  615. public static function mb_output_handler($contents, $status)
  616. {
  617. return $contents;
  618. }
  619. public static function mb_chr($code, $encoding = null)
  620. {
  621. if (0x80 > $code %= 0x200000) {
  622. $s = \chr($code);
  623. } elseif (0x800 > $code) {
  624. $s = \chr(0xC0 | $code >> 6).\chr(0x80 | $code & 0x3F);
  625. } elseif (0x10000 > $code) {
  626. $s = \chr(0xE0 | $code >> 12).\chr(0x80 | $code >> 6 & 0x3F).\chr(0x80 | $code & 0x3F);
  627. } else {
  628. $s = \chr(0xF0 | $code >> 18).\chr(0x80 | $code >> 12 & 0x3F).\chr(0x80 | $code >> 6 & 0x3F).\chr(0x80 | $code & 0x3F);
  629. }
  630. if ('UTF-8' !== $encoding = self::getEncoding($encoding)) {
  631. $s = mb_convert_encoding($s, $encoding, 'UTF-8');
  632. }
  633. return $s;
  634. }
  635. public static function mb_ord($s, $encoding = null)
  636. {
  637. if ('UTF-8' !== $encoding = self::getEncoding($encoding)) {
  638. $s = mb_convert_encoding($s, 'UTF-8', $encoding);
  639. }
  640. if (1 === \strlen($s)) {
  641. return \ord($s);
  642. }
  643. $code = ($s = unpack('C*', substr($s, 0, 4))) ? $s[1] : 0;
  644. if (0xF0 <= $code) {
  645. return (($code - 0xF0) << 18) + (($s[2] - 0x80) << 12) + (($s[3] - 0x80) << 6) + $s[4] - 0x80;
  646. }
  647. if (0xE0 <= $code) {
  648. return (($code - 0xE0) << 12) + (($s[2] - 0x80) << 6) + $s[3] - 0x80;
  649. }
  650. if (0xC0 <= $code) {
  651. return (($code - 0xC0) << 6) + $s[2] - 0x80;
  652. }
  653. return $code;
  654. }
  655. private static function getSubpart($pos, $part, $haystack, $encoding)
  656. {
  657. if (false === $pos) {
  658. return false;
  659. }
  660. if ($part) {
  661. return self::mb_substr($haystack, 0, $pos, $encoding);
  662. }
  663. return self::mb_substr($haystack, $pos, null, $encoding);
  664. }
  665. private static function html_encoding_callback(array $m)
  666. {
  667. $i = 1;
  668. $entities = '';
  669. $m = unpack('C*', htmlentities($m[0], \ENT_COMPAT, 'UTF-8'));
  670. while (isset($m[$i])) {
  671. if (0x80 > $m[$i]) {
  672. $entities .= \chr($m[$i++]);
  673. continue;
  674. }
  675. if (0xF0 <= $m[$i]) {
  676. $c = (($m[$i++] - 0xF0) << 18) + (($m[$i++] - 0x80) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++] - 0x80;
  677. } elseif (0xE0 <= $m[$i]) {
  678. $c = (($m[$i++] - 0xE0) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++] - 0x80;
  679. } else {
  680. $c = (($m[$i++] - 0xC0) << 6) + $m[$i++] - 0x80;
  681. }
  682. $entities .= '&#'.$c.';';
  683. }
  684. return $entities;
  685. }
  686. private static function title_case(array $s)
  687. {
  688. return self::mb_convert_case($s[1], \MB_CASE_UPPER, 'UTF-8').self::mb_convert_case($s[2], \MB_CASE_LOWER, 'UTF-8');
  689. }
  690. private static function getData($file)
  691. {
  692. if (file_exists($file = __DIR__.'/Resources/unidata/'.$file.'.php')) {
  693. return require $file;
  694. }
  695. return false;
  696. }
  697. private static function getEncoding($encoding)
  698. {
  699. if (null === $encoding) {
  700. return self::$internalEncoding;
  701. }
  702. if ('UTF-8' === $encoding) {
  703. return 'UTF-8';
  704. }
  705. $encoding = strtoupper($encoding);
  706. if ('8BIT' === $encoding || 'BINARY' === $encoding) {
  707. return 'CP850';
  708. }
  709. if ('UTF8' === $encoding) {
  710. return 'UTF-8';
  711. }
  712. return $encoding;
  713. }
  714. }