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