languages/LanguageZh.php

   1 <?php
   2 /**
   3   * @package MediaWiki
   4   * @subpackage Language
   5   */
   6 require_once( dirname(__FILE__).'/LanguageConverter.php' );
   7 require_once( dirname(__FILE__).'/LanguageZh_cn.php' );
   8
   9 class ZhConverter extends LanguageConverter {
  10         function loadDefaultTables() {
  11                 require( "includes/ZhConversion.php" );
  12                 $this->mTables = array();
  13                 $this->mTables['zh-cn'] = $zh2CN;
  14                 $this->mTables['zh-tw'] = $zh2TW;
  15                 $this->mTables['zh-sg'] = array_merge($zh2CN, $zh2SG);
  16                 $this->mTables['zh-hk'] = array_merge($zh2TW, $zh2HK);
  17                 $this->mTables['zh'] = array();
  18         }
  19
  20         function postLoadTables() {
  21                 $this->mTables['zh-sg'] = array_merge($this->mTables['zh-cn'], $this->mTables['zh-sg']);
  22                 $this->mTables['zh-hk'] = array_merge($this->mTables['zh-tw'], $this->mTables['zh-hk']);
  23     }
  24
  25         /* there shouldn't be any latin text in Chinese conversion, so no need
  26            to mark anything
  27     */
  28         function markNoConversion($text) {
  29                 return $text;
  30         }
  31
  32         function convertCategoryKey( $key ) {
  33                 return $this->autoConvert( $key, 'zh-cn' );
  34         }
  35 }
  36
  37
  38 /* class that handles both Traditional and Simplified Chinese
  39    right now it only distinguish zh_cn, zh_tw, zh_sg and zh_hk.
  40 */
  41 class LanguageZh extends LanguageZh_cn {
  42
  43         function __construct() {
  44                 global $wgHooks;
  45                 parent::__construct();
  46                 $this->mConverter = new ZhConverter($this, 'zh',
  47                                             array('zh', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk'),
  48                                                                                         array('zh'=>'zh-cn',
  49                                                                                                   'zh-cn'=>'zh-sg',
  50                                                                                                   'zh-sg'=>'zh-cn',
  51                                                                                                   'zh-tw'=>'zh-hk',
  52                                                                                                   'zh-hk'=>'zh-tw'));
  53                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  54         }
  55
  56
  57         # this should give much better diff info
  58         function segmentForDiff( $text ) {
  59                 return preg_replace(
  60                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  61                         "' ' .\"$1\"", $text);
  62         }
  63
  64         function unsegmentForDiff( $text ) {
  65                 return preg_replace(
  66                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  67                         "\"$1\"", $text);
  68         }
  69
  70         // word segmentation
  71         function stripForSearch( $string ) {
  72                 $fname="LanguageZh::stripForSearch";
  73                 wfProfileIn( $fname );
  74
  75                 // eventually this should be a word segmentation
  76                 // for now just treat each character as a word
  77                 $t = preg_replace(
  78                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  79                                 "' ' .\"$1\"", $string);
  80
  81         //always convert to zh-cn before indexing. it should be
  82                 //better to use zh-cn for search, since conversion from
  83                 //Traditional to Simplified is less ambiguous than the
  84                 //other way around
  85
  86                 $t = $this->mConverter->autoConvert($t, 'zh-cn');
  87                 $t = parent::stripForSearch( $t );
  88                 wfProfileOut( $fname );
  89                 return $t;
  90
  91         }
  92
  93         function convertForSearchResult( $termsArray ) {
  94                 $terms = implode( '|', $termsArray );
  95                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
  96                 $ret = array_unique( explode('|', $terms) );
  97                 return $ret;
  98         }
  99
 100 }
 101 ?>