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
  11         function __construct($langobj, $maincode,
  12                                                                 $variants=array(),
  13                                                                 $variantfallbacks=array(),
  14                                                                 $markup=array(),
  15                                                                 $flags = array(),
  16                                                                 $manualLevel = array() ) {
  17                 parent::__construct($langobj, $maincode,
  18                                                                         $variants,
  19                                                                         $variantfallbacks,
  20                                                                         $markup,
  21                                                                         $flags,
  22                                                                         $manualLevel);
  23                 $names = array(
  24                         'zh'      => '原文',
  25                         'zh-hans' => '简体',
  26                         'zh-hant' => '繁體',
  27                         'zh-cn'   => '大陆',
  28                         'zh-tw'   => '台灣',
  29                         'zh-hk'   => '香港',
  30                         'zh-mo'   => '澳門',
  31                         'zh-sg'   => '新加坡',
  32                         'zh-my'   => '马来西亚',
  33                 );
  34                 $this->mVariantNames = array_merge($this->mVariantNames,$names);
  35         }
  36
  37         function loadDefaultTables() {
  38                 require( dirname(__FILE__)."/../../includes/ZhConversion.php" );
  39                 $this->mTables = array(
  40                         'zh-hans' => new ReplacementArray( $zh2Hans ),
  41                         'zh-hant' => new ReplacementArray( $zh2Hant ),
  42                         'zh-cn'   => new ReplacementArray( array_merge($zh2Hans, $zh2CN) ),
  43                         'zh-hk'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  44                         'zh-mo'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  45                         'zh-my'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  46                         'zh-sg'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  47                         'zh-tw'   => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ),
  48                         'zh'      => new ReplacementArray
  49                 );
  50         }
  51
  52         function postLoadTables() {
  53                 $this->mTables['zh-cn']->merge( $this->mTables['zh-hans'] );
  54                 $this->mTables['zh-hk']->merge( $this->mTables['zh-hant'] );
  55                 $this->mTables['zh-mo']->merge( $this->mTables['zh-hant'] );
  56                 $this->mTables['zh-my']->merge( $this->mTables['zh-hans'] );
  57                 $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] );
  58                 $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] );
  59         }
  60
  61         /* there shouldn't be any latin text in Chinese conversion, so no need
  62            to mark anything.
  63            $noParse is there for compatibility with LanguageConvert::markNoConversion
  64          */
  65         function markNoConversion($text, $noParse = false) {
  66                 return $text;
  67         }
  68
  69         /* description of convert code in chinese language*/
  70         function getRulesDesc($bidtable,$unidtable){
  71                 $text=parent::getRulesDesc($bidtable,$unidtable);
  72                 $text=str_replace(':','\81F',$text);
  73                 $text=str_replace(';','\81G',$text);
  74                 return $text;
  75         }
  76
  77         function convertCategoryKey( $key ) {
  78                 return $this->autoConvert( $key, 'zh' );
  79         }
  80 }
  81
  82 /**
  83  * class that handles both Traditional and Simplified Chinese
  84  * right now it only distinguish zh_hans, zh_hant, zh_cn, zh_tw, zh_sg and zh_hk.
  85  *
  86  * @ingroup Language
  87  */
  88 class LanguageZh extends LanguageZh_hans {
  89
  90         function __construct() {
  91                 global $wgHooks;
  92                 parent::__construct();
  93
  94                 $variants = array('zh','zh-hans','zh-hant','zh-cn','zh-hk','zh-mo','zh-my','zh-sg','zh-tw');
  95                 $variantfallbacks = array(
  96                         'zh'      => array('zh-hans','zh-hant','zh-cn','zh-tw','zh-hk','zh-sg','zh-mo','zh-my'),
  97                         'zh-hans' => array('zh-cn','zh-sg','zh-my'),
  98                         'zh-hant' => array('zh-tw','zh-hk','zh-mo'),
  99                         'zh-cn'   => array('zh-hans','zh-sg','zh-my'),
 100                         'zh-sg'   => array('zh-hans','zh-cn','zh-my'),
 101                         'zh-my'   => array('zh-hant','zh-sg','zh-cn'),
 102                         'zh-tw'   => array('zh-hant','zh-hk','zh-mo'),
 103                         'zh-hk'   => array('zh-hant','zh-mo','zh-tw'),
 104                         'zh-mo'   => array('zh-hant','zh-hk','zh-tw'),
 105                 );
 106                 $ml=array(
 107                         'zh'      => 'disable',
 108                         'zh-hans' => 'unidirectional',
 109                         'zh-hant' => 'unidirectional',
 110                 );
 111
 112                 $this->mConverter = new ZhConverter( $this, 'zh',
 113                                                                 $variants, $variantfallbacks,
 114                                                                 array(),array(),
 115                                                                 $ml);
 116
 117                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
 118         }
 119
 120         # this should give much better diff info
 121         function segmentForDiff( $text ) {
 122                 return preg_replace(
 123                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 124                         "' ' .\"$1\"", $text);
 125         }
 126
 127         function unsegmentForDiff( $text ) {
 128                 return preg_replace(
 129                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 130                         "\"$1\"", $text);
 131         }
 132
 133         // word segmentation
 134         function stripForSearch( $string ) {
 135                 $fname="LanguageZh::stripForSearch";
 136                 wfProfileIn( $fname );
 137
 138                 // eventually this should be a word segmentation
 139                 // for now just treat each character as a word
 140                 $t = preg_replace(
 141                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 142                                 "' ' .\"$1\"", $string);
 143
 144         //always convert to zh-hans before indexing. it should be
 145                 //better to use zh-hans for search, since conversion from
 146                 //Traditional to Simplified is less ambiguous than the
 147                 //other way around
 148
 149                 $t = $this->mConverter->autoConvert($t, 'zh-hans');
 150                 $t = parent::stripForSearch( $t );
 151                 wfProfileOut( $fname );
 152                 return $t;
 153
 154         }
 155
 156         function convertForSearchResult( $termsArray ) {
 157                 $terms = implode( '|', $termsArray );
 158                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 159                 $ret = array_unique( explode('|', $terms) );
 160                 return $ret;
 161         }
 162 }
 163