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