lib/textlib.class.php

   1 <?php
   2 // This file is part of Moodle - http://moodle.org/
   3 //
   4 // Moodle is free software: you can redistribute it and/or modify
   5 // it under the terms of the GNU General Public License as published by
   6 // the Free Software Foundation, either version 3 of the License, or
   7 // (at your option) any later version.
   8 //
   9 // Moodle is distributed in the hope that it will be useful,
  10 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 // GNU General Public License for more details.
  13 //
  14 // You should have received a copy of the GNU General Public License
  15 // along with Moodle.  If not, see <http://www.gnu.org/licenses/>.
  16
  17 /**
  18  * Defines string apis
  19  *
  20  * @package    core
  21  * @copyright  (C) 2001-3001 Eloy Lafuente (stronk7) {@link http://contiento.com}
  22  * @license    http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
  23  */
  24
  25 defined('MOODLE_INTERNAL') || die();
  26
  27 /**
  28  * defines string api's for manipulating strings
  29  *
  30  * This class is used to manipulate strings under Moodle 1.6 an later. As
  31  * utf-8 text become mandatory a pool of safe functions under this encoding
  32  * become necessary. The name of the methods is exactly the
  33  * same than their PHP originals.
  34  *
  35  * A big part of this class acts as a wrapper over the Typo3 charset library,
  36  * really a cool group of utilities to handle texts and encoding conversion.
  37  *
  38  * Take a look to its own copyright and license details.
  39  *
  40  * IMPORTANT Note: Typo3 libraries always expect lowercase charsets to use 100%
  41  * its capabilities so, don't forget to make the conversion
  42  * from every wrapper function!
  43  *
  44  * @package   core
  45  * @category  string
  46  * @copyright 1999 onwards Martin Dougiamas  {@link http://moodle.com}
  47  * @license   http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
  48  */
  49 class textlib {
  50
  51     /**
  52      * Return t3lib helper class, which is used for conversion between charsets
  53      *
  54      * @param bool $reset
  55      * @return t3lib_cs
  56      */
  57     protected static function typo3($reset = false) {
  58         static $typo3cs = null;
  59
  60         if ($reset) {
  61             $typo3cs = null;
  62             return null;
  63         }
  64
  65         if (isset($typo3cs)) {
  66             return $typo3cs;
  67         }
  68
  69         global $CFG;
  70
  71         // Required files
  72         require_once($CFG->libdir.'/typo3/class.t3lib_cs.php');
  73         require_once($CFG->libdir.'/typo3/class.t3lib_div.php');
  74         require_once($CFG->libdir.'/typo3/interface.t3lib_singleton.php');
  75         require_once($CFG->libdir.'/typo3/class.t3lib_l10n_locales.php');
  76
  77         // do not use mbstring or recode because it may return invalid results in some corner cases
  78         $GLOBALS['TYPO3_CONF_VARS']['SYS']['t3lib_cs_convMethod'] = 'iconv';
  79         $GLOBALS['TYPO3_CONF_VARS']['SYS']['t3lib_cs_utils'] = 'iconv';
  80
  81         // Tell Typo3 we are curl enabled always (mandatory since 2.0)
  82         $GLOBALS['TYPO3_CONF_VARS']['SYS']['curlUse'] = '1';
  83
  84         // And this directory must exist to allow Typo to cache conversion
  85         // tables when using internal functions
  86         make_temp_directory('typo3temp/cs');
  87
  88         // Make sure typo is using our dir permissions
  89         $GLOBALS['TYPO3_CONF_VARS']['BE']['folderCreateMask'] = decoct($CFG->directorypermissions);
  90
  91         // Default mask for Typo
  92         $GLOBALS['TYPO3_CONF_VARS']['BE']['fileCreateMask'] = $CFG->directorypermissions;
  93
  94         // This full path constants must be defined too, transforming backslashes
  95         // to forward slashed because Typo3 requires it.
  96         if (!defined('PATH_t3lib')) {
  97             define('PATH_t3lib', str_replace('\\','/',$CFG->libdir.'/typo3/'));
  98             define('PATH_typo3', str_replace('\\','/',$CFG->libdir.'/typo3/'));
  99             define('PATH_site', str_replace('\\','/',$CFG->tempdir.'/'));
 100             define('TYPO3_OS', stristr(PHP_OS,'win')&&!stristr(PHP_OS,'darwin')?'WIN':'');
 101         }
 102
 103         $typo3cs = new t3lib_cs();
 104
 105         return $typo3cs;
 106     }
 107
 108     /**
 109      * Reset internal textlib caches.
 110      * @static
 111      */
 112     public static function reset_caches() {
 113         self::typo3(true);
 114     }
 115
 116     /**
 117      * Standardise charset name
 118      *
 119      * Please note it does not mean the returned charset is actually supported.
 120      *
 121      * @static
 122      * @param string $charset raw charset name
 123      * @return string normalised lowercase charset name
 124      */
 125     public static function parse_charset($charset) {
 126         $charset = strtolower($charset);
 127
 128         // shortcuts so that we do not have to load typo3 on every page
 129
 130         if ($charset === 'utf8' or $charset === 'utf-8') {
 131             return 'utf-8';
 132         }
 133
 134         if (preg_match('/^(cp|win|windows)-?(12[0-9]{2})$/', $charset, $matches)) {
 135             return 'windows-'.$matches[2];
 136         }
 137
 138         if (preg_match('/^iso-8859-[0-9]+$/', $charset, $matches)) {
 139             return $charset;
 140         }
 141
 142         if ($charset === 'euc-jp') {
 143             return 'euc-jp';
 144         }
 145         if ($charset === 'iso-2022-jp') {
 146             return 'iso-2022-jp';
 147         }
 148         if ($charset === 'shift-jis' or $charset === 'shift_jis') {
 149             return 'shift_jis';
 150         }
 151         if ($charset === 'gb2312') {
 152             return 'gb2312';
 153         }
 154         if ($charset === 'gb18030') {
 155             return 'gb18030';
 156         }
 157
 158         // fallback to typo3
 159         return self::typo3()->parse_charset($charset);
 160     }
 161
 162     /**
 163      * Converts the text between different encodings. It uses iconv extension with //TRANSLIT parameter,
 164      * falls back to typo3.
 165      * Returns false if fails.
 166      *
 167      * @param string $text
 168      * @param string $fromCS source encoding
 169      * @param string $toCS result encoding
 170      * @return string|bool converted string or false on error
 171      */
 172     public static function convert($text, $fromCS, $toCS='utf-8') {
 173         $fromCS = self::parse_charset($fromCS);
 174         $toCS   = self::parse_charset($toCS);
 175
 176         $text = (string)$text; // we can work only with strings
 177
 178         if ($text === '') {
 179             return '';
 180         }
 181
 182         $result = iconv($fromCS, $toCS.'//TRANSLIT', $text);
 183
 184         if ($result === false or $result === '') {
 185             // note: iconv is prone to return empty string when invalid char encountered, or false if encoding unsupported
 186             $oldlevel = error_reporting(E_PARSE);
 187             $result = self::typo3()->conv((string)$text, $fromCS, $toCS);
 188             error_reporting($oldlevel);
 189         }
 190
 191         return $result;
 192     }
 193
 194     /**
 195      * Multibyte safe substr() function, uses mbstring or iconv for UTF-8, falls back to typo3.
 196      *
 197      * @param string $text string to truncate
 198      * @param int $start negative value means from end
 199      * @param int $len maximum length of characters beginning from start
 200      * @param string $charset encoding of the text
 201      * @return string portion of string specified by the $start and $len
 202      */
 203     public static function substr($text, $start, $len=null, $charset='utf-8') {
 204         $charset = self::parse_charset($charset);
 205
 206         if ($charset === 'utf-8') {
 207             if (function_exists('mb_substr')) {
 208                 // this is much faster than iconv - see MDL-31142
 209                 if ($len === null) {
 210                     $oldcharset = mb_internal_encoding();
 211                     mb_internal_encoding('UTF-8');
 212                     $result = mb_substr($text, $start);
 213                     mb_internal_encoding($oldcharset);
 214                     return $result;
 215                 } else {
 216                     return mb_substr($text, $start, $len, 'UTF-8');
 217                 }
 218
 219             } else {
 220                 if ($len === null) {
 221                     $len = iconv_strlen($text, 'UTF-8');
 222                 }
 223                 return iconv_substr($text, $start, $len, 'UTF-8');
 224             }
 225         }
 226
 227         $oldlevel = error_reporting(E_PARSE);
 228         if ($len === null) {
 229             $result = self::typo3()->substr($charset, (string)$text, $start);
 230         } else {
 231             $result = self::typo3()->substr($charset, (string)$text, $start, $len);
 232         }
 233         error_reporting($oldlevel);
 234
 235         return $result;
 236     }
 237
 238     /**
 239      * Multibyte safe strlen() function, uses mbstring or iconv for UTF-8, falls back to typo3.
 240      *
 241      * @param string $text input string
 242      * @param string $charset encoding of the text
 243      * @return int number of characters
 244      */
 245     public static function strlen($text, $charset='utf-8') {
 246         $charset = self::parse_charset($charset);
 247
 248         if ($charset === 'utf-8') {
 249             if (function_exists('mb_strlen')) {
 250                 return mb_strlen($text, 'UTF-8');
 251             } else {
 252                 return iconv_strlen($text, 'UTF-8');
 253             }
 254         }
 255
 256         $oldlevel = error_reporting(E_PARSE);
 257         $result = self::typo3()->strlen($charset, (string)$text);
 258         error_reporting($oldlevel);
 259
 260         return $result;
 261     }
 262
 263     /**
 264      * Multibyte safe strtolower() function, uses mbstring, falls back to typo3.
 265      *
 266      * @param string $text input string
 267      * @param string $charset encoding of the text (may not work for all encodings)
 268      * @return string lower case text
 269      */
 270     public static function strtolower($text, $charset='utf-8') {
 271         $charset = self::parse_charset($charset);
 272
 273         if ($charset === 'utf-8' and function_exists('mb_strtolower')) {
 274             return mb_strtolower($text, 'UTF-8');
 275         }
 276
 277         $oldlevel = error_reporting(E_PARSE);
 278         $result = self::typo3()->conv_case($charset, (string)$text, 'toLower');
 279         error_reporting($oldlevel);
 280
 281         return $result;
 282     }
 283
 284     /**
 285      * Multibyte safe strtoupper() function, uses mbstring, falls back to typo3.
 286      *
 287      * @param string $text input string
 288      * @param string $charset encoding of the text (may not work for all encodings)
 289      * @return string upper case text
 290      */
 291     public static function strtoupper($text, $charset='utf-8') {
 292         $charset = self::parse_charset($charset);
 293
 294         if ($charset === 'utf-8' and function_exists('mb_strtoupper')) {
 295             return mb_strtoupper($text, 'UTF-8');
 296         }
 297
 298         $oldlevel = error_reporting(E_PARSE);
 299         $result = self::typo3()->conv_case($charset, (string)$text, 'toUpper');
 300         error_reporting($oldlevel);
 301
 302         return $result;
 303     }
 304
 305     /**
 306      * Find the position of the first occurrence of a substring in a string.
 307      * UTF-8 ONLY safe strpos(), uses mbstring, falls back to iconv.
 308      *
 309      * @param string $haystack the string to search in
 310      * @param string $needle one or more charachters to search for
 311      * @param int $offset offset from begining of string
 312      * @return int the numeric position of the first occurrence of needle in haystack.
 313      */
 314     public static function strpos($haystack, $needle, $offset=0) {
 315         if (function_exists('mb_strpos')) {
 316             return mb_strpos($haystack, $needle, $offset, 'UTF-8');
 317         } else {
 318             return iconv_strpos($haystack, $needle, $offset, 'UTF-8');
 319         }
 320     }
 321
 322     /**
 323      * Find the position of the last occurrence of a substring in a string
 324      * UTF-8 ONLY safe strrpos(), uses mbstring, falls back to iconv.
 325      *
 326      * @param string $haystack the string to search in
 327      * @param string $needle one or more charachters to search for
 328      * @return int the numeric position of the last occurrence of needle in haystack
 329      */
 330     public static function strrpos($haystack, $needle) {
 331         if (function_exists('mb_strpos')) {
 332             return mb_strrpos($haystack, $needle, null, 'UTF-8');
 333         } else {
 334             return iconv_strrpos($haystack, $needle, 'UTF-8');
 335         }
 336     }
 337
 338     /**
 339      * Try to convert upper unicode characters to plain ascii,
 340      * the returned string may contain unconverted unicode characters.
 341      *
 342      * @param string $text input string
 343      * @param string $charset encoding of the text
 344      * @return string converted ascii string
 345      */
 346     public static function specialtoascii($text, $charset='utf-8') {
 347         $charset = self::parse_charset($charset);
 348         $oldlevel = error_reporting(E_PARSE);
 349         $result = self::typo3()->specCharsToASCII($charset, (string)$text);
 350         error_reporting($oldlevel);
 351         return $result;
 352     }
 353
 354     /**
 355      * Generate a correct base64 encoded header to be used in MIME mail messages.
 356      * This function seems to be 100% compliant with RFC1342. Credits go to:
 357      * paravoid (http://www.php.net/manual/en/function.mb-encode-mimeheader.php#60283).
 358      *
 359      * @param string $text input string
 360      * @param string $charset encoding of the text
 361      * @return string base64 encoded header
 362      */
 363     public static function encode_mimeheader($text, $charset='utf-8') {
 364         if (empty($text)) {
 365             return (string)$text;
 366         }
 367         // Normalize charset
 368         $charset = self::parse_charset($charset);
 369         // If the text is pure ASCII, we don't need to encode it
 370         if (self::convert($text, $charset, 'ascii') == $text) {
 371             return $text;
 372         }
 373         // Although RFC says that line feed should be \r\n, it seems that
 374         // some mailers double convert \r, so we are going to use \n alone
 375         $linefeed="\n";
 376         // Define start and end of every chunk
 377         $start = "=?$charset?B?";
 378         $end = "?=";
 379         // Accumulate results
 380         $encoded = '';
 381         // Max line length is 75 (including start and end)
 382         $length = 75 - strlen($start) - strlen($end);
 383         // Multi-byte ratio
 384         $multilength = self::strlen($text, $charset);
 385         // Detect if strlen and friends supported
 386         if ($multilength === false) {
 387             if ($charset == 'GB18030' or $charset == 'gb18030') {
 388                 while (strlen($text)) {
 389                     // try to encode first 22 chars - we expect most chars are two bytes long
 390                     if (preg_match('/^(([\x00-\x7f])|([\x81-\xfe][\x40-\x7e])|([\x81-\xfe][\x80-\xfe])|([\x81-\xfe][\x30-\x39]..)){1,22}/m', $text, $matches)) {
 391                         $chunk = $matches[0];
 392                         $encchunk = base64_encode($chunk);
 393                         if (strlen($encchunk) > $length) {
 394                             // find first 11 chars - each char in 4 bytes - worst case scenario
 395                             preg_match('/^(([\x00-\x7f])|([\x81-\xfe][\x40-\x7e])|([\x81-\xfe][\x80-\xfe])|([\x81-\xfe][\x30-\x39]..)){1,11}/m', $text, $matches);
 396                             $chunk = $matches[0];
 397                             $encchunk = base64_encode($chunk);
 398                         }
 399                         $text = substr($text, strlen($chunk));
 400                         $encoded .= ' '.$start.$encchunk.$end.$linefeed;
 401                     } else {
 402                         break;
 403                     }
 404                 }
 405                 $encoded = trim($encoded);
 406                 return $encoded;
 407             } else {
 408                 return false;
 409             }
 410         }
 411         $ratio = $multilength / strlen($text);
 412         // Base64 ratio
 413         $magic = $avglength = floor(3 * $length * $ratio / 4);
 414         // basic infinite loop protection
 415         $maxiterations = strlen($text)*2;
 416         $iteration = 0;
 417         // Iterate over the string in magic chunks
 418         for ($i=0; $i <= $multilength; $i+=$magic) {
 419             if ($iteration++ > $maxiterations) {
 420                 return false; // probably infinite loop
 421             }
 422             $magic = $avglength;
 423             $offset = 0;
 424             // Ensure the chunk fits in length, reducing magic if necessary
 425             do {
 426                 $magic -= $offset;
 427                 $chunk = self::substr($text, $i, $magic, $charset);
 428                 $chunk = base64_encode($chunk);
 429                 $offset++;
 430             } while (strlen($chunk) > $length);
 431             // This chunk doesn't break any multi-byte char. Use it.
 432             if ($chunk)
 433                 $encoded .= ' '.$start.$chunk.$end.$linefeed;
 434         }
 435         // Strip the first space and the last linefeed
 436         $encoded = substr($encoded, 1, -strlen($linefeed));
 437
 438         return $encoded;
 439     }
 440
 441     /**
 442      * Converts all the numeric entities &#nnnn; or &#xnnn; to UTF-8
 443      * Original from laurynas dot butkus at gmail at:
 444      * http://php.net/manual/en/function.html-entity-decode.php#75153
 445      * with some custom mods to provide more functionality
 446      *
 447      * @param string $str input string
 448      * @param boolean $htmlent convert also html entities (defaults to true)
 449      * @return string encoded UTF-8 string
 450      *
 451      * NOTE: we could have used typo3 entities_to_utf8() here
 452      *       but the direct alternative used runs 400% quicker
 453      *       and uses 0.5Mb less memory, so, let's use it
 454      *       (tested against 10^6 conversions)
 455      */
 456     public static function entities_to_utf8($str, $htmlent=true) {
 457         static $trans_tbl; // Going to use static transliteration table
 458
 459         // Replace numeric entities
 460         $result = preg_replace('~&#x([0-9a-f]+);~ei', 'textlib::code2utf8(hexdec("\\1"))', $str);
 461         $result = preg_replace('~&#([0-9]+);~e', 'textlib::code2utf8(\\1)', $result);
 462
 463         // Replace literal entities (if desired)
 464         if ($htmlent) {
 465             // Generate/create $trans_tbl
 466             if (!isset($trans_tbl)) {
 467                 $trans_tbl = array();
 468                 foreach (get_html_translation_table(HTML_ENTITIES) as $val=>$key) {
 469                     $trans_tbl[$key] = utf8_encode($val);
 470                 }
 471             }
 472             $result = strtr($result, $trans_tbl);
 473         }
 474         // Return utf8-ised string
 475         return $result;
 476     }
 477
 478     /**
 479      * Converts all Unicode chars > 127 to numeric entities &#nnnn; or &#xnnn;.
 480      *
 481      * @param string $str input string
 482      * @param boolean $dec output decadic only number entities
 483      * @param boolean $nonnum remove all non-numeric entities
 484      * @return string converted string
 485      */
 486     public static function utf8_to_entities($str, $dec=false, $nonnum=false) {
 487         // Avoid some notices from Typo3 code
 488         $oldlevel = error_reporting(E_PARSE);
 489         if ($nonnum) {
 490             $str = self::typo3()->entities_to_utf8((string)$str, true);
 491         }
 492         $result = self::typo3()->utf8_to_entities((string)$str);
 493         if ($dec) {
 494             $result = preg_replace('/&#x([0-9a-f]+);/ie', "'&#'.hexdec('$1').';'", $result);
 495         }
 496         // Restore original debug level
 497         error_reporting($oldlevel);
 498         return $result;
 499     }
 500
 501     /**
 502      * Removes the BOM from unicode string {@link http://unicode.org/faq/utf_bom.html}
 503      *
 504      * @param string $str input string
 505      * @return string
 506      */
 507     public static function trim_utf8_bom($str) {
 508         $bom = "\xef\xbb\xbf";
 509         if (strpos($str, $bom) === 0) {
 510             return substr($str, strlen($bom));
 511         }
 512         return $str;
 513     }
 514
 515     /**
 516      * Returns encoding options for select boxes, utf-8 and platform encoding first
 517      *
 518      * @return array encodings
 519      */
 520     public static function get_encodings() {
 521         $encodings = array();
 522         $encodings['UTF-8'] = 'UTF-8';
 523         $winenc = strtoupper(get_string('localewincharset', 'langconfig'));
 524         if ($winenc != '') {
 525             $encodings[$winenc] = $winenc;
 526         }
 527         $nixenc = strtoupper(get_string('oldcharset', 'langconfig'));
 528         $encodings[$nixenc] = $nixenc;
 529
 530         foreach (self::typo3()->synonyms as $enc) {
 531             $enc = strtoupper($enc);
 532             $encodings[$enc] = $enc;
 533         }
 534         return $encodings;
 535     }
 536
 537     /**
 538      * Returns the utf8 string corresponding to the unicode value
 539      * (from php.net, courtesy - romans@void.lv)
 540      *
 541      * @param  int    $num one unicode value
 542      * @return string the UTF-8 char corresponding to the unicode value
 543      */
 544     public static function code2utf8($num) {
 545         if ($num < 128) {
 546             return chr($num);
 547         }
 548         if ($num < 2048) {
 549             return chr(($num >> 6) + 192) . chr(($num & 63) + 128);
 550         }
 551         if ($num < 65536) {
 552             return chr(($num >> 12) + 224) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
 553         }
 554         if ($num < 2097152) {
 555             return chr(($num >> 18) + 240) . chr((($num >> 12) & 63) + 128) . chr((($num >> 6) & 63) + 128) . chr(($num & 63) + 128);
 556         }
 557         return '';
 558     }
 559
 560     /**
 561      * Makes first letter of each word capital - words must be separated by spaces.
 562      * Use with care, this function does not work properly in many locales!!!
 563      *
 564      * @param string $text input string
 565      * @return string
 566      */
 567     public static function strtotitle($text) {
 568         if (empty($text)) {
 569             return $text;
 570         }
 571
 572         if (function_exists('mb_convert_case')) {
 573             return mb_convert_case($text, MB_CASE_TITLE, 'UTF-8');
 574         }
 575
 576         $text = self::strtolower($text);
 577         $words = explode(' ', $text);
 578         foreach ($words as $i=>$word) {
 579             $length = self::strlen($word);
 580             if (!$length) {
 581                 continue;
 582
 583             } else if ($length == 1) {
 584                 $words[$i] = self::strtoupper($word);
 585
 586             } else {
 587                 $letter = self::substr($word, 0, 1);
 588                 $letter = self::strtoupper($letter);
 589                 $rest   = self::substr($word, 1);
 590                 $words[$i] = $letter.$rest;
 591             }
 592         }
 593         return implode(' ', $words);
 594     }
 595
 596     /**
 597      * Locale aware sorting, the key associations are kept, values are sorted alphabetically.
 598      *
 599      * @param array $arr array to be sorted (reference)
 600      * @param int $sortflag One of Collator::SORT_REGULAR, Collator::SORT_NUMERIC, Collator::SORT_STRING
 601      * @return void modifies parameter
 602      */
 603     public static function asort(array &$arr, $sortflag = null) {
 604         debugging('textlib::asort has been superseeded by collatorlib::asort please upgrade your code to use that', DEBUG_DEVELOPER);
 605         collatorlib::asort($arr, $sortflag);
 606     }
 607 }
 608
 609
 610 /**
 611  * A collator class with static methods that can be used for sorting.
 612  *
 613  * @package   core
 614  * @copyright 2011 Sam Hemelryk
 615  *            2012 Petr Skoda
 616  * @license   http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
 617  */
 618 class collatorlib {
 619     /** @const compare items using general PHP comparison, equivalent to Collator::SORT_REGULAR, this may bot be locale aware! */
 620     const SORT_REGULAR = 0;
 621
 622     /** @const compare items as strings, equivalent to Collator::SORT_STRING */
 623     const SORT_STRING = 1;
 624
 625     /** @const compare items as numbers, equivalent to Collator::SORT_NUMERIC */
 626     const SORT_NUMERIC = 2;
 627
 628     /** @const compare items like natsort(), equivalent to SORT_NATURAL */
 629     const SORT_NATURAL = 6;
 630
 631     /** @const do not ignore case when sorting, use bitwise "|" with SORT_NATURAL or SORT_STRING, equivalent to Collator::UPPER_FIRST */
 632     const CASE_SENSITIVE = 64;
 633
 634     /** @var Collator|false|null **/
 635     protected static $collator = null;
 636
 637     /** @var string|null The locale that was used in instantiating the current collator **/
 638     protected static $locale = null;
 639
 640     /**
 641      * Prevent class instances, all methods are static.
 642      */
 643     private function __construct() {
 644     }
 645
 646     /**
 647      * Ensures that a collator is available and created
 648      *
 649      * @return bool Returns true if collation is available and ready
 650      */
 651     protected static function ensure_collator_available() {
 652         $locale = get_string('locale', 'langconfig');
 653         if (is_null(self::$collator) || $locale != self::$locale) {
 654             self::$collator = false;
 655             self::$locale = $locale;
 656             if (class_exists('Collator', false)) {
 657                 $collator = new Collator($locale);
 658                 if (!empty($collator) && $collator instanceof Collator) {
 659                     // Check for non fatal error messages. This has to be done immediately
 660                     // after instantiation as any further calls to collation will cause
 661                     // it to reset to 0 again (or another error code if one occurred)
 662                     $errorcode = $collator->getErrorCode();
 663                     $errormessage = $collator->getErrorMessage();
 664                     // Check for an error code, 0 means no error occurred
 665                     if ($errorcode !== 0) {
 666                         // Get the actual locale being used, e.g. en, he, zh
 667                         $localeinuse = $collator->getLocale(Locale::ACTUAL_LOCALE);
 668                         // Check for the common fallback warning error codes. If this occurred
 669                         // there is normally little to worry about:
 670                         // - U_USING_DEFAULT_WARNING (127)  - default fallback locale used (pt => UCA)
 671                         // - U_USING_FALLBACK_WARNING (128) - fallback locale used (de_CH => de)
 672                         // (UCA: Unicode Collation Algorithm http://unicode.org/reports/tr10/)
 673                         if ($errorcode === -127 || $errorcode === -128) {
 674                             // Check if the locale in use is UCA default one ('root') or
 675                             // if it is anything like the locale we asked for
 676                             if ($localeinuse !== 'root' && strpos($locale, $localeinuse) !== 0) {
 677                                 // The locale we asked for is completely different to the locale
 678                                 // we have received, let the user know via debugging
 679                                 debugging('Invalid locale: "' . $locale . '", with warning (not fatal) "' . $errormessage .
 680                                     '", falling back to "' . $collator->getLocale(Locale::VALID_LOCALE) . '"');
 681                             } else {
 682                                 // Nothing to do here, this is expected!
 683                                 // The Moodle locale setting isn't what the collator expected but
 684                                 // it is smart enough to match the first characters of our locale
 685                                 // to find the correct locale or to use UCA collation
 686                             }
 687                         } else {
 688                             // We've received some other sort of non fatal warning - let the
 689                             // user know about it via debugging.
 690                             debugging('Problem with locale: "' . $locale . '", with message "' . $errormessage .
 691                                 '", falling back to "' . $collator->getLocale(Locale::VALID_LOCALE) . '"');
 692                         }
 693                     }
 694                     // Store the collator object now that we can be sure it is in a workable condition
 695                     self::$collator = $collator;
 696                 } else {
 697                     // Fatal error while trying to instantiate the collator... something went wrong
 698                     debugging('Error instantiating collator for locale: "' . $locale . '", with error [' .
 699                         intl_get_error_code() . '] ' . intl_get_error_message($collator));
 700                 }
 701             }
 702         }
 703         return (self::$collator instanceof Collator);
 704     }
 705
 706     /**
 707      * Restore array contents keeping new keys.
 708      * @static
 709      * @param array $arr
 710      * @param array $original
 711      * @return void modifies $arr
 712      */
 713     protected static function restore_array(array &$arr, array &$original) {
 714         foreach ($arr as $key => $ignored) {
 715             $arr[$key] = $original[$key];
 716         }
 717     }
 718
 719     /**
 720      * Normalise numbers in strings for natural sorting comparisons.
 721      * @static
 722      * @param string $string
 723      * @return string string with normalised numbers
 724      */
 725     protected static function naturalise($string) {
 726         return preg_replace_callback('/[0-9]+/', array('collatorlib', 'callback_naturalise'), $string);
 727     }
 728
 729     /**
 730      * @internal
 731      * @static
 732      * @param array $matches
 733      * @return string
 734      */
 735     public static function callback_naturalise($matches) {
 736         return str_pad($matches[0], 20, '0', STR_PAD_LEFT);
 737     }
 738
 739     /**
 740      * Locale aware sorting, the key associations are kept, values are sorted alphabetically.
 741      *
 742      * @param array $arr array to be sorted (reference)
 743      * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
 744      *      optionally "|" collatorlib::CASE_SENSITIVE
 745      * @return bool True on success
 746      */
 747     public static function asort(array &$arr, $sortflag = collatorlib::SORT_STRING) {
 748         if (empty($arr)) {
 749             // nothing to do
 750             return true;
 751         }
 752
 753         $original = null;
 754
 755         $casesensitive = (bool)($sortflag & collatorlib::CASE_SENSITIVE);
 756         $sortflag = ($sortflag & ~collatorlib::CASE_SENSITIVE);
 757         if ($sortflag != collatorlib::SORT_NATURAL and $sortflag != collatorlib::SORT_STRING) {
 758             $casesensitive = false;
 759         }
 760
 761         if (self::ensure_collator_available()) {
 762             if ($sortflag == collatorlib::SORT_NUMERIC) {
 763                 $flag = Collator::SORT_NUMERIC;
 764
 765             } else if ($sortflag == collatorlib::SORT_REGULAR) {
 766                 $flag = Collator::SORT_REGULAR;
 767
 768             } else {
 769                 $flag = Collator::SORT_STRING;
 770             }
 771
 772             if ($sortflag == collatorlib::SORT_NATURAL) {
 773                 $original = $arr;
 774                 if ($sortflag == collatorlib::SORT_NATURAL) {
 775                     foreach ($arr as $key => $value) {
 776                         $arr[$key] = self::naturalise((string)$value);
 777                     }
 778                 }
 779             }
 780             if ($casesensitive) {
 781                 self::$collator->setAttribute(Collator::CASE_FIRST, Collator::UPPER_FIRST);
 782             } else {
 783                 self::$collator->setAttribute(Collator::CASE_FIRST, Collator::OFF);
 784             }
 785             $result = self::$collator->asort($arr, $flag);
 786             if ($original) {
 787                 self::restore_array($arr, $original);
 788             }
 789             return $result;
 790         }
 791
 792         // try some fallback that works at least for English
 793
 794         if ($sortflag == collatorlib::SORT_NUMERIC) {
 795             return asort($arr, SORT_NUMERIC);
 796
 797         } else if ($sortflag == collatorlib::SORT_REGULAR) {
 798             return asort($arr, SORT_REGULAR);
 799         }
 800
 801         if (!$casesensitive) {
 802             $original = $arr;
 803             foreach ($arr as $key => $value) {
 804                 $arr[$key] = textlib::strtolower($value);
 805             }
 806         }
 807
 808         if ($sortflag == collatorlib::SORT_NATURAL) {
 809             $result = natsort($arr);
 810
 811         } else {
 812             $result = asort($arr, SORT_LOCALE_STRING);
 813         }
 814
 815         if ($original) {
 816             self::restore_array($arr, $original);
 817         }
 818
 819         return $result;
 820     }
 821
 822     /**
 823      * Locale aware sort of objects by a property in common to all objects
 824      *
 825      * @param array $objects An array of objects to sort (handled by reference)
 826      * @param string $property The property to use for comparison
 827      * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
 828      *      optionally "|" collatorlib::CASE_SENSITIVE
 829      * @return bool True on success
 830      */
 831     public static function asort_objects_by_property(array &$objects, $property, $sortflag = collatorlib::SORT_STRING) {
 832         $original = $objects;
 833         foreach ($objects as $key => $object) {
 834             $objects[$key] = $object->$property;
 835         }
 836         $result = self::asort($objects, $sortflag);
 837         self::restore_array($objects, $original);
 838         return $result;
 839     }
 840
 841     /**
 842      * Locale aware sort of objects by a method in common to all objects
 843      *
 844      * @param array $objects An array of objects to sort (handled by reference)
 845      * @param string $method The method to call to generate a value for comparison
 846      * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
 847      *      optionally "|" collatorlib::CASE_SENSITIVE
 848      * @return bool True on success
 849      */
 850     public static function asort_objects_by_method(array &$objects, $method, $sortflag = collatorlib::SORT_STRING) {
 851         $original = $objects;
 852         foreach ($objects as $key => $object) {
 853             $objects[$key] = $object->{$method}();
 854         }
 855         $result = self::asort($objects, $sortflag);
 856         self::restore_array($objects, $original);
 857         return $result;
 858     }
 859
 860     /**
 861      * Locale aware sorting, the key associations are kept, keys are sorted alphabetically.
 862      *
 863      * @param array $arr array to be sorted (reference)
 864      * @param int $sortflag One of collatorlib::SORT_NUMERIC, collatorlib::SORT_STRING, collatorlib::SORT_NATURAL, collatorlib::SORT_REGULAR
 865      *      optionally "|" collatorlib::CASE_SENSITIVE
 866      * @return bool True on success
 867      */
 868     public static function ksort(array &$arr, $sortflag = collatorlib::SORT_STRING) {
 869         $keys = array_keys($arr);
 870         if (!self::asort($keys, $sortflag)) {
 871             return false;
 872         }
 873         // This is a bit slow, but we need to keep the references
 874         $original = $arr;
 875         $arr = array(); // Surprisingly this does not break references outside
 876         foreach ($keys as $key) {
 877             $arr[$key] = $original[$key];
 878         }
 879
 880         return true;
 881     }
 882 }
 883