languages/classes/LanguageZh.php

   1 <?php
   2
   3 require_once( dirname(__FILE__).'/../LanguageConverter.php' );
   4 require_once( dirname(__FILE__).'/LanguageZh_hans.php' );
   5
   6 /**
   7  * @ingroup Language
   8  */
   9 class ZhConverter extends LanguageConverter {
  10         function loadDefaultTables() {
  11                 require( dirname(__FILE__)."/../../includes/ZhConversion.php" );
  12                 $this->mTables = array(
  13                         'zh-hans' => new ReplacementArray( $zh2Hans ),
  14                         'zh-hant' => new ReplacementArray( $zh2Hant ),
  15                         'zh-cn'   => new ReplacementArray( array_merge($zh2Hans, $zh2CN) ),
  16                         'zh-tw'   => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ),
  17                         'zh-sg'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  18                         'zh-hk'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  19                         'zh'      => new ReplacementArray
  20                 );
  21         }
  22
  23         function postLoadTables() {
  24                 $this->mTables['zh-cn']->merge( $this->mTables['zh-hans'] );
  25                 $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] );
  26                 $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] );
  27                 $this->mTables['zh-hk']->merge( $this->mTables['zh-hant'] );
  28         }
  29
  30         /* there shouldn't be any latin text in Chinese conversion, so no need
  31            to mark anything.
  32            $noParse is there for compatibility with LanguageConvert::markNoConversion
  33     */
  34         function markNoConversion($text, $noParse = false) {
  35                 return $text;
  36         }
  37
  38         function convertCategoryKey( $key ) {
  39                 return $this->autoConvert( $key, 'zh' );
  40         }
  41 }
  42
  43 /**
  44  * class that handles both Traditional and Simplified Chinese
  45  * right now it only distinguish zh_hans, zh_hant, zh_cn, zh_tw, zh_sg and zh_hk.
  46  *
  47  * @ingroup Language
  48  */
  49 class LanguageZh extends LanguageZh_hans {
  50
  51         function __construct() {
  52                 global $wgHooks;
  53                 parent::__construct();
  54
  55                 $variants = array('zh', 'zh-hans', 'zh-hant', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk');
  56                 $variantfallbacks = array(
  57                         'zh'      => 'zh-hans',
  58                         'zh-hans' => 'zh',
  59                         'zh-hant' => 'zh',
  60                         'zh-cn'   => 'zh-hans',
  61                         'zh-sg'   => 'zh-hans',
  62                         'zh-tw'   => 'zh-hant',
  63                         'zh-hk'   => 'zh-hant',
  64                 );
  65
  66                 $this->mConverter = new ZhConverter( $this, 'zh', $variants, $variantfallbacks );
  67
  68                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  69         }
  70
  71         # this should give much better diff info
  72         function segmentForDiff( $text ) {
  73                 return preg_replace(
  74                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  75                         "' ' .\"$1\"", $text);
  76         }
  77
  78         function unsegmentForDiff( $text ) {
  79                 return preg_replace(
  80                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  81                         "\"$1\"", $text);
  82         }
  83
  84         // word segmentation
  85         function stripForSearch( $string ) {
  86                 $fname="LanguageZh::stripForSearch";
  87                 wfProfileIn( $fname );
  88
  89                 // eventually this should be a word segmentation
  90                 // for now just treat each character as a word
  91                 $t = preg_replace(
  92                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  93                                 "' ' .\"$1\"", $string);
  94
  95         //always convert to zh-hans before indexing. it should be
  96                 //better to use zh-hans for search, since conversion from
  97                 //Traditional to Simplified is less ambiguous than the
  98                 //other way around
  99
 100                 $t = $this->mConverter->autoConvert($t, 'zh-hans');
 101                 $t = parent::stripForSearch( $t );
 102                 wfProfileOut( $fname );
 103                 return $t;
 104
 105         }
 106
 107         function convertForSearchResult( $termsArray ) {
 108                 $terms = implode( '|', $termsArray );
 109                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 110                 $ret = array_unique( explode('|', $terms) );
 111                 return $ret;
 112         }
 113 }