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