nucleus/libs/i18n.php

   1 <?php\r
   2 /**\r
   3  * i18n class for Nucleus CMS\r
   4  * written by Takashi Sakamoto as of Feb 03, 2012\r
   5  * \r
   6  * This includes wrapper functions of iconv and mbstring\r
   7  * for multibyte processing and includes parameters related to locale.\r
   8  * \r
   9  * This program is free software; you can redistribute it and/or\r
  10  * modify it under the terms of the GNU General Public License\r
  11  * as published by the Free Software Foundation; either version 2\r
  12  * of the License, or (at your option) any later version.\r
  13  * (see nucleus/documentation/index.html#license for more info)\r
  14  *\r
  15  * @license http://nucleuscms.org/license.txt GNU General Public License\r
  16  * @copyright Copyright (C) 2002-2011 The Nucleus Group\r
  17  * @version $Id: i18n.php 1678 2012-02-26 07:31:36Z sakamocchi $\r
  18  */\r
  19 class i18n\r
  20 {\r
  21         static private $mode = FALSE;\r
  22         \r
  23         static private $charset = '';\r
  24         static private $language = '';\r
  25         static private $script = '';\r
  26         static private $region = '';\r
  27         static private $locale_list = array();\r
  28         static private $timezone = 'UTC';\r
  29         \r
  30         /**\r
  31          * i18n::init\r
  32          * Initializing i18n class\r
  33          * \r
  34          * @static\r
  35          * @param       string  $charset        character set\r
  36          * @return      boolean \r
  37          */\r
  38         static public function init($charset, $dir)\r
  39         {\r
  40                 /* i18n is already initialized */\r
  41                 if ( self::$mode )\r
  42                 {\r
  43                         return TRUE;\r
  44                 }\r
  45                 \r
  46                 /* make locale list in this Nucleus CMS */\r
  47                 if ( ($handle = opendir($dir)) === FALSE )\r
  48                 {\r
  49                         return FALSE;\r
  50                 }\r
  51                 while ($filename = readdir($handle))\r
  52                 {\r
  53                         if (preg_match("#^(.+_.+_.+)\.{$charset}\.php$#", $filename, $matches) )\r
  54                         {\r
  55                                 if ( !in_array($matches[1], self::$locale_list) )\r
  56                                 {\r
  57                                         self::$locale_list[] = $matches[1];\r
  58                                 }\r
  59                         }\r
  60                 }\r
  61                 closedir($handle);\r
  62                 \r
  63                 /* set i18n backend and validate character set */\r
  64                 if ( extension_loaded('iconv') )\r
  65                 {\r
  66                         /* this is just for checking the charset. */\r
  67                         if ( iconv_set_encoding('internal_encoding', $charset)\r
  68                          && iconv_set_encoding('output_encoding', $charset)\r
  69                          && iconv_set_encoding('internal_encoding', $charset) )\r
  70                         {\r
  71                                 self::$charset = $charset;\r
  72                                 self::$mode = 'iconv';\r
  73                         }\r
  74                 }\r
  75                 else if ( extension_loaded('mbstring') )\r
  76                 {\r
  77                         /* this is just for checking the charset. */\r
  78                         if ( mb_http_output($charset)\r
  79                          && mb_internal_encoding($charset)\r
  80                          && mb_regex_encoding($charset) )\r
  81                         {\r
  82                                 self::$charset = $charset;\r
  83                                 self::$mode = 'mbstring';\r
  84                         }\r
  85                 }\r
  86                 \r
  87                 return TRUE;\r
  88         }\r
  89         \r
  90         /**\r
  91          * i18n::get_available_locale_list\r
  92          * return available locale list with current charset\r
  93          * \r
  94          * @static\r
  95          * @param       void\r
  96          * @return      array   available locale list\r
  97          */\r
  98         static public function get_available_locale_list()\r
  99         {\r
 100                 return self::$locale_list;\r
 101         }\r
 102         \r
 103         /**\r
 104          * i18n::get_current_charset\r
 105          * return current charset\r
 106          * \r
 107          * @static\r
 108          * @param       void\r
 109          * @return      string  $charset        current character set\r
 110          */\r
 111         static public function get_current_charset()\r
 112         {\r
 113                 return self::$charset;\r
 114         }\r
 115         \r
 116         /**\r
 117          * i18n::set_locale\r
 118          * Set current locale\r
 119          * \r
 120          * NOTE:\r
 121          * naming rule is "$language_$script_$region.$charset.php", refer to RFC 5646.\r
 122          * @link http://www.ietf.org/rfc/rfc5646.txt\r
 123          * @see 2.  The Language Tag\r
 124          * \r
 125          * @static\r
 126          * @param       string  $locale\r
 127          * @return      bool    TRUE/FALSE\r
 128          * \r
 129          */\r
 130         static public function set_current_locale($locale)\r
 131         {\r
 132                 if ( preg_match('#^(.+)_(.+)_(.+)$#', $locale, $match) )\r
 133                 {\r
 134                         self::$language = $match[1];\r
 135                         self::$script   = $match[2];\r
 136                         self::$region   = $match[3];\r
 137                         return TRUE;\r
 138                 }\r
 139                 return FALSE;\r
 140         }\r
 141         \r
 142         /**\r
 143          * i18n::get_locale\r
 144          * Get current locale\r
 145          * \r
 146          * @static\r
 147          * @param       void\r
 148          * @return      $locale\r
 149          */\r
 150         static public function get_current_locale()\r
 151         {\r
 152                 $elements = array(self::$language, self::$script, self::$region);\r
 153                 return implode('_', $elements);\r
 154         }\r
 155         \r
 156         /**\r
 157          * i18n::confirm_default_date_timezone\r
 158          * to avoid E_NOTICE or E_WARNING generated when every calling to a date/time function.\r
 159          * \r
 160          * NOTE:\r
 161          * Some private servers are lack of its timezone setting\r
 162          * http://www.php.net/manual/en/function.date-default-timezone-set.php\r
 163          * \r
 164          * @static\r
 165          * @param       void\r
 166          * @return      void\r
 167          */\r
 168         static public function confirm_default_date_timezone()\r
 169         {\r
 170                 if ( function_exists('date_default_timezone_get') \r
 171                  && FALSE !== ($timezone = @date_default_timezone_get()))\r
 172                 {\r
 173                         self::$timezone = $timezone;\r
 174                 }\r
 175                 if (function_exists('date_default_timezone_set')) {\r
 176                          @date_default_timezone_set(self::$timezone);\r
 177                 }\r
 178                 return;\r
 179         }\r
 180         \r
 181         /**\r
 182          * i18n::get_current_date_timezone()\r
 183          * get current timezone\r
 184          * \r
 185          * @static\r
 186          * @param       void\r
 187          * @return      $timezone\r
 188          */\r
 189         static public function get_date_timezone()\r
 190         {\r
 191                 return self::$timezone;\r
 192         }\r
 193         \r
 194         /**\r
 195          * i18n::convert\r
 196          * character set converter\r
 197          * \r
 198          * @static\r
 199          * @param       string  $string target string binary\r
 200          * @param       string  $from   original character set encoding\r
 201          * @param       string  $to     target character set encoding\r
 202          * @return      string  converted string\r
 203          */\r
 204         static public function convert($string, $from, $to='')\r
 205         {\r
 206                 if ( $to == '' )\r
 207                 {\r
 208                         $to = self::$charset;\r
 209                 }\r
 210                 \r
 211                 if ( self::$mode == 'iconv' )\r
 212                 {\r
 213                         $string = iconv($from, $to.'//TRANSLIT', $string);\r
 214                 }\r
 215                 else if ( self::$mode == 'mbstring' )\r
 216                 {\r
 217                         $string = mb_convert_encoding($string, $to, $from);\r
 218                 }\r
 219                 return (string) $string;\r
 220         }\r
 221         \r
 222         /**\r
 223          * i18n::strlen\r
 224          * strlen wrapper\r
 225          * \r
 226          * @static\r
 227          * @param       string  $string target string\r
 228          * @return      integer the number of letters\r
 229          */\r
 230         static public function strlen($string)\r
 231         {\r
 232                 $length = 0;\r
 233                 if ( self::$mode == 'iconv' )\r
 234                 {\r
 235                         $length = iconv_strlen($string, self::$charset);\r
 236                 }\r
 237                 else if ( self::$mode == 'mbstring' )\r
 238                 {\r
 239                         $length = mb_strlen($string, self::$charset);\r
 240                 }\r
 241                 else\r
 242                 {\r
 243                         $length = strlen($string);\r
 244                 }\r
 245                 return (integer) $length;\r
 246         }\r
 247         \r
 248         /**\r
 249          * i18n::strpos\r
 250          * strpos wrapper\r
 251          * \r
 252          * @static\r
 253          * @param       string  $haystack       string to search\r
 254          * @param       string  $needle string for search\r
 255          * @param       integer $offset the position from which the search should be performed. \r
 256          * @return      integer/FALSE   the numeric position of the first occurrence of needle in haystack\r
 257          */\r
 258         static public function strpos($haystack, $needle, $offset=0)\r
 259         {\r
 260                 $position = 0;\r
 261                 if ( self::$mode == 'iconv' )\r
 262                 {\r
 263                         $position = iconv_strpos($haystack, $needle, $offset, self::$charset);\r
 264                 }\r
 265                 else if ( self::$mode == 'mbstring' )\r
 266                 {\r
 267                         $position = mb_strpos($haystack, $needle, $offset, self::$charset);\r
 268                 }\r
 269                 else\r
 270                 {\r
 271                         $position = strpos($haystack, $needle, $offset);\r
 272                 }\r
 273                 \r
 274                 if ( $position !== FALSE)\r
 275                 {\r
 276                         $position = (integer) $position;\r
 277                 }\r
 278                 return $position;\r
 279         }\r
 280         \r
 281         /**\r
 282          * i18n::strrpos\r
 283          * strrpos wrapper\r
 284          * \r
 285          * @static\r
 286          * @param       string  $haystack       string to search\r
 287          * @param       string  $needle string for search\r
 288          * @return      integer/FALSE   the numeric position of the last occurrence of needle in haystack\r
 289          */\r
 290         static public function strrpos ($haystack, $needle)\r
 291         {\r
 292                 $position = 0;\r
 293                 if ( self::$mode == 'iconv' )\r
 294                 {\r
 295                         $position = iconv_strrpos($haystack, $needle, self::$charset);\r
 296                 }\r
 297                 else if ( self::$mode == 'mbstring' )\r
 298                 {\r
 299                         $position = mb_strrpos($haystack, $needle, 0, self::$charset);\r
 300                 }\r
 301                 else\r
 302                 {\r
 303                         $position = strrpos($haystack, $needle, 0);\r
 304                 }\r
 305                 \r
 306                 if ( $position !== FALSE)\r
 307                 {\r
 308                         $position = (integer) $position;\r
 309                 }\r
 310                 return $position;\r
 311         }\r
 312         \r
 313         /**\r
 314          * i18n::substr\r
 315          * substr wrapper\r
 316          * \r
 317          * @static\r
 318          * @param       string  $string string to be cut\r
 319          * @param       string  $start  the position of starting\r
 320          * @param       integer $length the length to be cut\r
 321          * @return      string  the extracted part of string\r
 322          */\r
 323         static public function substr($string, $start, $length=0)\r
 324         {\r
 325                 $return = '';\r
 326                 if ( self::$mode == 'iconv' )\r
 327                 {\r
 328                         $return = iconv_substr($string, $start, $length, self::$charset);\r
 329                 }\r
 330                 else if ( self::$mode == 'mbstring' )\r
 331                 {\r
 332                         $return = mb_substr($string, $start, $length, self::$charset);\r
 333                 }\r
 334                 else\r
 335                 {\r
 336                         $return = strrpos($string, $start, $length);\r
 337                 }\r
 338                 return (string) $return;\r
 339         }\r
 340         \r
 341         /**\r
 342          * i18n::explode()\r
 343          * explode function based on multibyte processing with non-pcre regular expressions\r
 344          * \r
 345          * NOTE: we SHOULD use preg_split function instead of this,\r
 346          *  and I hope this is obsoleted near future...\r
 347          *  \r
 348          *  preg_split()\r
 349          *  http://www.php.net/manual/en/function.preg-split.php\r
 350          * \r
 351          * @static\r
 352          * @param       string  $delimiter      singlebyte or multibyte delimiter\r
 353          * @param       string  $target target string\r
 354          * @param       integer $limit  the number of index for returned array\r
 355          * @return      array   array splitted by $delimiter\r
 356          */\r
 357         static public function explode($delimiter, $target, $limit=0)\r
 358         {\r
 359                 $array = array();\r
 360                 $preg_delimiter = '#' . preg_quote($delimiter, '#') . '#';\r
 361                 if ( preg_match($preg_delimiter, $target) === 0 )\r
 362                 {\r
 363                         return (array) $target;\r
 364                 }\r
 365                 for ( $count=0; $limit == 0 || $count < $limit; $count++ )\r
 366                 {\r
 367                         $offset = self::strpos($target, $delimiter);\r
 368                         if ( $array != array() && $offset == 0 )\r
 369                         {\r
 370                                 $array[] = $target;\r
 371                                 break;\r
 372                         }\r
 373                         $array[]        = self::substr($target, 0, $offset);\r
 374                         $length = self::strlen($target) - $offset;\r
 375                         $target = self::substr($target, $offset+1, $length);\r
 376                         continue;\r
 377                 }\r
 378                 return (array) $array;\r
 379         }\r
 380         \r
 381         /**\r
 382          * i18n::strftime\r
 383          * strftime function based on multibyte processing\r
 384          * \r
 385          * @static\r
 386          * @param       string  $format format with singlebyte or multibyte\r
 387          * @param       timestamp       $timestamp      UNIX timestamp\r
 388          * @return      string  formatted timestamp\r
 389          */\r
 390         static public function strftime($format, $timestamp='')\r
 391         {\r
 392                 $formatted = '';\r
 393                 \r
 394                 if ( $timestamp == '' )\r
 395                 {\r
 396                         $timestamp = time();\r
 397                 }\r
 398                 \r
 399                 if ( $format == '%%' )\r
 400                 {\r
 401                         return '%';\r
 402                 }\r
 403                 else if ( preg_match('#%[^%]#', $format) === 0 )\r
 404                 {\r
 405                         return $format;\r
 406                 }\r
 407                 \r
 408                 $format = trim(preg_replace('#(%[^%])#', ',$1,', $format), ',');\r
 409                 $elements = preg_split('#,#', $format);\r
 410                 \r
 411                 foreach ( $elements as $element )\r
 412                 {\r
 413                         if ( preg_match('#(%[^%])#', $element) )\r
 414                         {\r
 415                                 $formatted .= strftime($element, $timestamp);\r
 416                         }\r
 417                         else if ( $element == '%%' )\r
 418                         {\r
 419                                 $formatted .= '%';\r
 420                         }\r
 421                         else\r
 422                         {\r
 423                                 $formatted .= $element;\r
 424                         }\r
 425                 }\r
 426                 \r
 427                 return (string) $formatted;\r
 428         }\r
 429         \r
 430         /**\r
 431          * i18n::formatted_datetime()\r
 432          * return formatted datetime string\r
 433          * \r
 434          * Date and Time Formats\r
 435          * @link        http://www.w3.org/TR/NOTE-datetime\r
 436          * \r
 437          * Working with Time Zones\r
 438          * @link        http://www.w3.org/TR/timezone/\r
 439          * \r
 440          * @param       String  $format time expression format\r
 441          * @param       String  $timestamp      UNIX timestamp\r
 442          * @param       Integer $offset timestamp offset\r
 443          * @return      String  formatted datetime\r
 444          */\r
 445         static public function formatted_datetime($format, $timestamp, $offset=0)\r
 446         {\r
 447                 $suffix = '';\r
 448                 $string = '';\r
 449                 \r
 450                 switch ( $format )\r
 451                 {\r
 452                         case 'mysql':\r
 453                                 /*\r
 454                                  * MySQL 5.0 Reference Manual\r
 455                                  *  10.3.1. The DATE, DATETIME, and TIMESTAMP Types\r
 456                                  *   http://dev.mysql.com/doc/refman/5.0/en/datetime.html\r
 457                                  */\r
 458                                 $timestamp += $offset;\r
 459                                 $format = '%Y-%m-%d %H:%M:%S';\r
 460                                 $suffix ='';\r
 461                                 break;\r
 462                         case 'rfc822':\r
 463                                 /*\r
 464                                  * RFC 822: STANDARD FOR THE FORMAT OF ARPA INTERNET TEXT MESSAGES\r
 465                                  *  5.  DATE AND TIME SPECIFICATION\r
 466                                  *   http://www.ietf.org/rfc/rfc0822.txt\r
 467                                  */\r
 468                                 $format = '%a, %d %m %y %H:%M:%S ';\r
 469                                 if ( $offset < 0 )\r
 470                                 {\r
 471                                         $suffix = '-';\r
 472                                         $offset = -$offset;\r
 473                                 }\r
 474                                 else\r
 475                                 {\r
 476                                         $suffix = '+';\r
 477                                 }\r
 478                                 \r
 479                                 $suffix .= sprintf("%02d%02d", floor($offset / 3600), round(($offset % 3600) / 60) );\r
 480                                 break;\r
 481                         case 'rfc822GMT':\r
 482                                 /*\r
 483                                  * RFC 822: STANDARD FOR THE FORMAT OF ARPA INTERNET TEXT MESSAGES\r
 484                                  *  5.  DATE AND TIME SPECIFICATION\r
 485                                  *   http://www.ietf.org/rfc/rfc0822.txt\r
 486                                  */\r
 487                                 $format = '%a, %d %m %y %H:%M:%S ';\r
 488                                 $timestamp -= $offset;\r
 489                                 $suffix = 'GMT';\r
 490                                 break;\r
 491                         case 'iso8601':\r
 492                         case 'rfc3339':\r
 493                                 /*\r
 494                                  * RFC3339: Date and Time on the Internet: Timestamps\r
 495                                  *  5. Date and Time format\r
 496                                  *   http://www.ietf.org/rfc/rfc3339.txt\r
 497                                  */\r
 498                                 $format = '%Y-%m-%dT%H:%M:%S';\r
 499                                 if ( $offset < 0 )\r
 500                                 {\r
 501                                         $suffix = '-';\r
 502                                         $offset = -$offset;\r
 503                                 }\r
 504                                 else\r
 505                                 {\r
 506                                         $suffix = '+';\r
 507                                 }\r
 508                                 $suffix .= sprintf("%02d:%02d", floor($offset / 3600), round(($offset % 3600) / 60) );\r
 509                                 break;\r
 510                         case 'utc':\r
 511                         case 'iso8601UTC':\r
 512                         case 'rfc3339UTC':\r
 513                                 /*\r
 514                                  * RFC3339: Date and Time on the Internet: Timestamps\r
 515                                  *  5. Date and Time format\r
 516                                  *   http://www.ietf.org/rfc/rfc3339.txt\r
 517                                  */\r
 518                                 $timestamp -= $offset;\r
 519                                 $format = '%Y-%m-%dT%H:%M:%SZ';\r
 520                                 $suffix = '';\r
 521                                 break;\r
 522                         case '':\r
 523                                 $format = '%X %x';\r
 524                                 $offset = '';\r
 525                                 break;\r
 526                         default:\r
 527                                 $suffix = '';\r
 528                                 break;\r
 529                 }\r
 530                 return i18n::strftime($format, $timestamp) . $suffix;\r
 531         }\r
 532         \r
 533         /**\r
 534          * i18n::convert_locale_to_old_language_file_name()\r
 535          * NOTE: this should be obsoleted near future.\r
 536          * \r
 537          * @static\r
 538          * @param       string  $target_locale  locale name as language_script_region\r
 539          * @return      string  old translation file name\r
 540          */\r
 541         static public function convert_locale_to_old_language_file_name($target_locale)\r
 542         {\r
 543                 $target_language = '';\r
 544                 foreach ( self::$lang_refs as $language => $locale )\r
 545                 {\r
 546                         if ( preg_match('#-#', $language) )\r
 547                         {\r
 548                                 if ( $target_locale . '.' . self::$charset == $locale )\r
 549                                 {\r
 550                                         $target_language = $language;\r
 551                                         break;\r
 552                                 }\r
 553                         }\r
 554                         else if ( $target_locale == $locale )\r
 555                         {\r
 556                                 $target_language = $language;\r
 557                         }\r
 558                 }\r
 559                 return $target_language;\r
 560         }\r
 561         \r
 562         /**\r
 563          * i18n::convert_old_language_file_name_to_locale()\r
 564          * NOTE: this should be obsoleted near future.\r
 565          * \r
 566          * @static\r
 567          * @param       string  $target_language        old translation file name\r
 568          * @return      string  locale name as language_script_region\r
 569          */\r
 570         static public function convert_old_language_file_name_to_locale($target_language)\r
 571         {\r
 572                 $target_locale = '';\r
 573                 foreach ( self::$lang_refs as $language => $locale )\r
 574                 {\r
 575                         if ( $target_language == $language )\r
 576                         {\r
 577                                 if ( preg_match('#^(.+)\.(.+)$#', $locale, $match) )\r
 578                                 {\r
 579                                         $target_locale = $match[1];\r
 580                                 }\r
 581                                 else\r
 582                                 {\r
 583                                         $target_locale = $locale;\r
 584                                 }\r
 585                                 break;\r
 586                         }\r
 587                 }\r
 588                 return $target_locale;\r
 589         }\r
 590         \r
 591         /**\r
 592          * i18n::$lang_refs\r
 593          * reference table to convert old and new way to name translation files.\r
 594          * NOTE: this should be obsoleted as soon as possible.\r
 595          * \r
 596          * @static\r
 597          */\r
 598         static private $lang_refs = array(\r
 599                 "english"               => "en_Latn_US",\r
 600                 "english-utf8"  => "en_Latn_US.UTF-8",\r
 601                 "bulgarian"     => "bg_Cyrl_BG",\r
 602                 "finnish"               => "fi_Latn_FU",\r
 603                 "catalan"               => "ca_Latn_ES",\r
 604                 "french"                => "fr_Latn_FR",\r
 605                 "russian"               => "ru_Cyrl_RU",\r
 606                 "chinese"               => "zh_Hans_CN",\r
 607                 "simchinese"    => "zh_Hans_CN",\r
 608                 "chineseb5"     => "zh_Hant_TW",\r
 609                 "traditional_chinese"   =>      "zh_Hant_TW",\r
 610                 "galego"                => "gl_Latn_ES",\r
 611                 "german"                => "de_Latn_DE",\r
 612                 "korean-utf"    => "ko_Kore_KR.UTF-8",\r
 613                 "korean-euc-kr" => "ko_Kore_KR.EUC-KR",\r
 614                 "slovak"                => "sk_Latn_SK",\r
 615                 "czech"         => "cs_Latn_CZ",\r
 616                 "hungarian"     => "hu_Latn_HU",\r
 617                 "latvian"               => "lv_Latn_LV",\r
 618                 "nederlands"    => "nl_Latn_NL",\r
 619                 "italiano"              => "it_Latn_IT",\r
 620                 "persian"               => "fa_Arab_IR",\r
 621                 "spanish"               => "es_Latn_ES",\r
 622                 "spanish-utf8"  => "es_Latn_ES.UTF-8",\r
 623                 "japanese-euc"  => "ja_Jpan_JP.EUC-JP",\r
 624                 "japanese-utf8" => "ja_Jpan_JP.UTF-8",\r
 625                 "portuguese_brazil"     => "pt_Latn_BR"\r
 626         );\r
 627 }\r