languages/LanguageZh.php

   1 <?php
   2 /**
   3   * @package MediaWiki
   4   * @subpackage Language
   5   */
   6 require_once( "LanguageConverter.php" );
   7 require_once( "LanguageZh_cn.php");
   8 require_once( "LanguageZh_tw.php");
   9 require_once( "LanguageZh_sg.php");
  10 require_once( "LanguageZh_hk.php");
  11
  12 class ZhConverter extends LanguageConverter {
  13         function loadDefaultTables() {
  14                 require( "includes/ZhConversion.php" );
  15                 $this->mTables = array();
  16                 $this->mTables['zh-cn'] = $zh2CN;
  17                 $this->mTables['zh-tw'] = $zh2TW;
  18                 $this->mTables['zh-sg'] = array_merge($zh2CN, $zh2SG);
  19                 $this->mTables['zh-hk'] = array_merge($zh2TW, $zh2HK);
  20                 $this->mTables['zh'] = array();
  21         }
  22
  23         function postLoadTables() {
  24                 $this->mTables['zh-sg'] = array_merge($this->mTables['zh-cn'], $this->mTables['zh-sg']);
  25                 $this->mTables['zh-hk'] = array_merge($this->mTables['zh-tw'], $this->mTables['zh-hk']);
  26     }
  27
  28         /* there shouldn't be any latin text in Chinese conversion, so no need
  29            to mark anything
  30     */
  31         function markNoConversion($text) {
  32                 return $text;
  33         }
  34
  35         function convertCategoryKey( $key ) {
  36                 return $this->autoConvert( $key, 'zh-cn' );
  37         }
  38 }
  39
  40
  41 /* class that handles both Traditional and Simplified Chinese
  42    right now it only distinguish zh_cn, zh_tw, zh_sg and zh_hk.
  43 */
  44 class LanguageZh extends LanguageZh_cn {
  45
  46         function LanguageZh() {
  47                 global $wgHooks;
  48                 $this->mConverter = new ZhConverter($this, 'zh',
  49                                             array('zh', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk'),
  50                                                                                         array('zh'=>'zh-cn',
  51                                                                                                   'zh-cn'=>'zh-sg',
  52                                                                                                   'zh-sg'=>'zh-cn',
  53                                                                                                   'zh-tw'=>'zh-hk',
  54                                                                                                   'zh-hk'=>'zh-tw'));
  55                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  56         }
  57
  58
  59         # this should give much better diff info
  60         function segmentForDiff( $text ) {
  61                 return preg_replace(
  62                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  63                         "' ' .\"$1\"", $text);
  64         }
  65
  66         function unsegmentForDiff( $text ) {
  67                 return preg_replace(
  68                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  69                         "\"$1\"", $text);
  70         }
  71
  72         // word segmentation
  73         function stripForSearch( $string ) {
  74                 $fname="LanguageZh::stripForSearch";
  75                 wfProfileIn( $fname );
  76
  77                 // eventually this should be a word segmentation
  78                 // for now just treat each character as a word
  79                 $t = preg_replace(
  80                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  81                                 "' ' .\"$1\"", $string);
  82
  83         //always convert to zh-cn before indexing. it should be
  84                 //better to use zh-cn for search, since conversion from
  85                 //Traditional to Simplified is less ambiguous than the
  86                 //other way around
  87
  88                 $t = $this->mConverter->autoConvert($t, 'zh-cn');
  89                 $t = LanguageUtf8::stripForSearch( $t );
  90                 wfProfileOut( $fname );
  91                 return $t;
  92
  93         }
  94
  95         function convertForSearchResult( $termsArray ) {
  96                 $terms = implode( '|', $termsArray );
  97                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
  98                 $ret = array_unique( explode('|', $terms) );
  99                 return $ret;
 100         }
 101
 102 }
 103 ?>