languages/classes/LanguageZh.php

   1 <?php
   2 /**
   3   * @addtogroup Language
   4   */
   5 require_once( dirname(__FILE__).'/../LanguageConverter.php' );
   6 require_once( dirname(__FILE__).'/LanguageZh_hans.php' );
   7
   8 class ZhConverter extends LanguageConverter {
   9         function loadDefaultTables() {
  10                 require( dirname(__FILE__)."/../../includes/ZhConversion.php" );
  11                 $this->mTables = array(
  12                         'zh-cn' => new ReplacementArray( $zh2CN ),
  13                         'zh-tw' => new ReplacementArray( $zh2TW ),
  14                         'zh-sg' => new ReplacementArray( array_merge($zh2CN, $zh2SG) ),
  15                         'zh-hk' => new ReplacementArray( array_merge($zh2TW, $zh2HK) ),
  16                         'zh' => new ReplacementArray
  17                 );
  18         }
  19
  20         function postLoadTables() {
  21                 $this->mTables['zh-sg']->merge( $this->mTables['zh-cn'] );
  22                 $this->mTables['zh-hk']->merge( $this->mTables['zh-tw'] );
  23     }
  24
  25         /* there shouldn't be any latin text in Chinese conversion, so no need
  26            to mark anything.
  27            $noParse is there for compatibility with LanguageConvert::markNoConversion
  28     */
  29         function markNoConversion($text, $noParse = false) {
  30                 return $text;
  31         }
  32
  33         function convertCategoryKey( $key ) {
  34                 return $this->autoConvert( $key, 'zh-cn' );
  35         }
  36 }
  37
  38
  39 /* class that handles both Traditional and Simplified Chinese
  40    right now it only distinguish zh_cn, zh_tw, zh_sg and zh_hk.
  41 */
  42 class LanguageZh extends LanguageZh_hans {
  43
  44         function __construct() {
  45                 global $wgHooks;
  46                 parent::__construct();
  47
  48                 $variants = array('zh', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk');
  49                 $variantfallbacks = array(
  50                         'zh'    => 'zh-cn',
  51                         'zh-cn' => 'zh-sg',
  52                         'zh-sg' => 'zh-cn',
  53                         'zh-tw' => 'zh-hk',
  54                         'zh-hk' => 'zh-tw'
  55                 );
  56
  57                 $this->mConverter = new ZhConverter( $this, 'zh', $variants, $variantfallbacks );
  58
  59                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  60         }
  61
  62
  63         # this should give much better diff info
  64         function segmentForDiff( $text ) {
  65                 return preg_replace(
  66                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  67                         "' ' .\"$1\"", $text);
  68         }
  69
  70         function unsegmentForDiff( $text ) {
  71                 return preg_replace(
  72                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  73                         "\"$1\"", $text);
  74         }
  75
  76         // word segmentation
  77         function stripForSearch( $string ) {
  78                 $fname="LanguageZh::stripForSearch";
  79                 wfProfileIn( $fname );
  80
  81                 // eventually this should be a word segmentation
  82                 // for now just treat each character as a word
  83                 $t = preg_replace(
  84                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  85                                 "' ' .\"$1\"", $string);
  86
  87         //always convert to zh-cn before indexing. it should be
  88                 //better to use zh-cn for search, since conversion from
  89                 //Traditional to Simplified is less ambiguous than the
  90                 //other way around
  91
  92                 $t = $this->mConverter->autoConvert($t, 'zh-cn');
  93                 $t = parent::stripForSearch( $t );
  94                 wfProfileOut( $fname );
  95                 return $t;
  96
  97         }
  98
  99         function convertForSearchResult( $termsArray ) {
 100                 $terms = implode( '|', $termsArray );
 101                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 102                 $ret = array_unique( explode('|', $terms) );
 103                 return $ret;
 104         }
 105
 106 }