languages/classes/LanguageYue.php

   1 <?php
   2 /**
   3  * Cantonese (粵語)
   4  *
   5  * @ingroup Language
   6  */
   7 class LanguageYue extends Language {
   8         function hasWordBreaks() {
   9                 return false;
  10         }
  11
  12         /**
  13          * Eventually this should be a word segmentation;
  14          * for now just treat each character as a word.
  15          * @todo Fixme: only do this for Han characters...
  16          */
  17         function segmentByWord( $string ) {
  18                 $reg = "/([\\xc0-\\xff][\\x80-\\xbf]*)/";
  19                 $s = self::insertSpace( $string, $reg );
  20                 return $s;
  21         }
  22
  23         function normalizeForSearch( $string ) {
  24                 wfProfileIn( __METHOD__ );
  25
  26                 // Double-width roman characters
  27                 $s = self::convertDoubleWidth( $string );
  28                 $s = trim( $s );
  29                 $s = parent::normalizeForSearch( $s );
  30
  31                 wfProfileOut( __METHOD__ );
  32                 return $s;
  33         }
  34 }