languages/classes/LanguageJa.php

   1 <?php
   2
   3 /**
   4  * Japanese (日本語)
   5  *
   6  * @ingroup Language
   7  */
   8 class LanguageJa extends Language {
   9
  10         /**
  11          * @param $string string
  12          * @return string
  13          */
  14         function segmentByWord( $string ) {
  15                 // Strip known punctuation ?
  16                 // $s = preg_replace( '/\xe3\x80[\x80-\xbf]/', '', $s ); # U3000-303f
  17
  18                 // Space strings of like hiragana/katakana/kanji
  19                 $hiragana = '(?:\xe3(?:\x81[\x80-\xbf]|\x82[\x80-\x9f]))'; # U3040-309f
  20                 $katakana = '(?:\xe3(?:\x82[\xa0-\xbf]|\x83[\x80-\xbf]))'; # U30a0-30ff
  21                 $kanji = '(?:\xe3[\x88-\xbf][\x80-\xbf]'
  22                         . '|[\xe4-\xe8][\x80-\xbf]{2}'
  23                         . '|\xe9[\x80-\xa5][\x80-\xbf]'
  24                         . '|\xe9\xa6[\x80-\x99])';
  25                         # U3200-9999 = \xe3\x88\x80-\xe9\xa6\x99
  26                 $reg = "/({$hiragana}+|{$katakana}+|{$kanji}+)/";
  27                 $s = self::insertSpace( $string, $reg );
  28                 return $s;
  29         }
  30
  31         /**
  32          * Italic is not appropriate for Japanese script
  33          * Unfortunately most browsers do not recognise this, and render <em> as italic
  34          *
  35          * @param $text string
  36          * @return string
  37          */
  38         function emphasize( $text ) {
  39                 return $text;
  40         }
  41 }