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
  29
  30 /* class that handles both Traditional and Simplified Chinese
  31    right now it only distinguish zh_cn, zh_tw, zh_sg and zh_hk.
  32 */
  33 class LanguageZh extends LanguageZh_cn {
  34
  35         function LanguageZh() {
  36                 global $wgHooks;
  37                 $this->mConverter = new ZhConverter($this, 'zh',
  38                                             array('zh', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk'),
  39                                                                                         array('zh'=>'zh-cn',
  40                                                                                                   'zh-cn'=>'zh-sg',
  41                                                                                                   'zh-sg'=>'zh-cn',
  42                                                                                                   'zh-tw'=>'zh-hk',
  43                                                                                                   'zh-hk'=>'zh-tw'));
  44                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  45         }
  46
  47
  48         # this should give much better diff info
  49         function segmentForDiff( $text ) {
  50                 return preg_replace(
  51                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  52                         "' ' .\"$1\"", $text);
  53         }
  54
  55         function unsegmentForDiff( $text ) {
  56                 return preg_replace(
  57                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  58                         "\"$1\"", $text);
  59         }
  60
  61         // word segmentation
  62         function stripForSearch( $string ) {
  63                 $fname="LanguageZh::stripForSearch";
  64                 wfProfileIn( $fname );
  65
  66                 // eventually this should be a word segmentation
  67                 // for now just treat each character as a word
  68                 $t = preg_replace(
  69                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  70                                 "' ' .\"$1\"", $string);
  71
  72         //always convert to zh-cn before indexing. it should be
  73                 //better to use zh-cn for search, since conversion from
  74                 //Traditional to Simplified is less ambiguous than the
  75                 //other way around
  76
  77                 $t = $this->mConverter->autoConvert($t, 'zh-cn');
  78                 $t = LanguageUtf8::stripForSearch( $t );
  79                 wfProfileOut( $fname );
  80                 return $t;
  81
  82         }
  83
  84         function convertForSearchResult( $termsArray ) {
  85                 $terms = implode( '|', $termsArray );
  86                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
  87                 $ret = array_unique( explode('|', $terms) );
  88                 return $ret;
  89         }
  90
  91 }
  92 ?>