[Summary view] [Print] [Text view]

   1  <?php
   2  /**
   3   * Copyright 2014 Horde LLC (http://www.horde.org/)
   4   *
   5   * See the enclosed file COPYING for license information (LGPL). If you
   6   * did not receive this file, see http://www.horde.org/licenses/lgpl21.
   7   *
   8   * @category  Horde
   9   * @copyright 2014 Horde LLC
  10   * @license   http://www.horde.org/licenses/lgpl21 LGPL 2.1
  11   * @package   Util
  12   */
  13  
  14  /**
  15   * Provides utility methods used to transliterate a string.
  16   *
  17   * @author    Michael Slusarz <slusarz@horde.org>
  18   * @author    Jan Schneider <jan@horde.org>
  19   * @category  Horde
  20   * @copyright 2014 Horde LLC
  21   * @license   http://www.horde.org/licenses/lgpl21 LGPL 2.1
  22   * @package   Util
  23   * @since     2.4.0
  24   */
  25  class Horde_String_Transliterate
  26  {
  27      /**
  28       * Transliterate mapping cache.
  29       *
  30       * @var array
  31       */
  32      static protected $_map;
  33  
  34      /**
  35       * Transliterator instance.
  36       *
  37       * @var Transliterator
  38       */
  39      static protected $_transliterator;
  40  
  41      /**
  42       * Transliterates an UTF-8 string to ASCII, replacing non-English
  43       * characters to their English equivalents.
  44       *
  45       * Note: there is no guarantee that the output string will be ASCII-only,
  46       * since any non-ASCII character not in the transliteration list will
  47       * be ignored.
  48       *
  49       * @param string $str  Input string (UTF-8).
  50       *
  51       * @return string  Transliterated string (UTF-8).
  52       */
  53      static public function toAscii($str)
  54      {
  55          switch (true) {
  56          case class_exists('Transliterator'):
  57              return self::_intlToAscii($str);
  58          case extension_loaded('iconv'):
  59              return self::_iconvToAscii($str);
  60          default:
  61              return self::_fallbackToAscii($str);
  62          }
  63      }
  64  
  65      /**
  66       */
  67      static protected function _intlToAscii($str)
  68      {
  69          if (!isset(self::$_transliterator)) {
  70              self::$_transliterator = Transliterator::create(
  71                  'Any-Latin; Latin-ASCII'
  72              );
  73          }
  74          return self::$_transliterator->transliterate($str);
  75      }
  76  
  77      /**
  78       */
  79      static protected function _iconvToAscii($str)
  80      {
  81          return iconv('UTF-8', 'ASCII//TRANSLIT', $str);
  82      }
  83  
  84      /**
  85       */
  86      static protected function _fallbackToAscii($str)
  87      {
  88          if (!isset(self::$_map)) {
  89              self::$_map = array(
  90                  'À' => 'A',
  91                  'Á' => 'A',
  92                  'Â' => 'A',
  93                  'Ã' => 'A',
  94                  'Ä' => 'A',
  95                  'Å' => 'A',
  96                  'Æ' => 'AE',
  97                  'à' => 'a',
  98                  'á' => 'a',
  99                  'â' => 'a',
 100                  'ã' => 'a',
 101                  'ä' => 'a',
 102                  'å' => 'a',
 103                  'æ' => 'ae',
 104                  'Þ' => 'TH',
 105                  'þ' => 'th',
 106                  'Ç' => 'C',
 107                  'ç' => 'c',
 108                  'Ð' => 'D',
 109                  'ð' => 'd',
 110                  'È' => 'E',
 111                  'É' => 'E',
 112                  'Ê' => 'E',
 113                  'Ë' => 'E',
 114                  'è' => 'e',
 115                  'é' => 'e',
 116                  'ê' => 'e',
 117                  'ë' => 'e',
 118                  'ƒ' => 'f',
 119                  'Ì' => 'I',
 120                  'Í' => 'I',
 121                  'Î' => 'I',
 122                  'Ï' => 'I',
 123                  'ì' => 'i',
 124                  'í' => 'i',
 125                  'î' => 'i',
 126                  'ï' => 'i',
 127                  'Ñ' => 'N',
 128                  'ñ' => 'n',
 129                  'Ò' => 'O',
 130                  'Ó' => 'O',
 131                  'Ô' => 'O',
 132                  'Õ' => 'O',
 133                  'Ö' => 'O',
 134                  'Ø' => 'O',
 135                  'ò' => 'o',
 136                  'ó' => 'o',
 137                  'ô' => 'o',
 138                  'õ' => 'o',
 139                  'ö' => 'o',
 140                  'ø' => 'o',
 141                  'Š' => 'S',
 142                  'ẞ' => 'SS',
 143                  'ß' => 'ss',
 144                  'š' => 's',
 145                  'ś' => 's',
 146                  'Ù' => 'U',
 147                  'Ú' => 'U',
 148                  'Û' => 'U',
 149                  'Ü' => 'U',
 150                  'ù' => 'u',
 151                  'ú' => 'u',
 152                  'û' => 'u',
 153                  'Ý' => 'Y',
 154                  'ý' => 'y',
 155                  'ÿ' => 'y',
 156                  'Ž' => 'Z',
 157                  'ž' => 'z'
 158              );
 159          }
 160  
 161          return strtr($str, self::$_map);
 162      }
 163  }