languages/classes/LanguageZh.php

   1 <?php
   2
   3 $dir = dirname(__FILE__) . '/';
   4 require_once( $dir . '../LanguageConverter.php' );
   5 require_once( $dir . 'LanguageZh_hans.php' );
   6
   7 /**
   8  * @ingroup Language
   9  */
  10 class ZhConverter extends LanguageConverter {
  11
  12         function __construct($langobj, $maincode,
  13                                                                 $variants=array(),
  14                                                                 $variantfallbacks=array(),
  15                                                                 $markup=array(),
  16                                                                 $flags = array(),
  17                                                                 $manualLevel = array() ) {
  18                 $this->mDescCodeSep = '：';
  19                 $this->mDescVarSep = '；';
  20                 parent::__construct($langobj, $maincode,
  21                                                                         $variants,
  22                                                                         $variantfallbacks,
  23                                                                         $markup,
  24                                                                         $flags,
  25                                                                         $manualLevel);
  26                 $names = array(
  27                         'zh'      => '原文',
  28                         'zh-hans' => '简体',
  29                         'zh-hant' => '繁體',
  30                         'zh-cn'   => '大陆',
  31                         'zh-tw'   => '台灣',
  32                         'zh-hk'   => '香港',
  33                         'zh-mo'   => '澳門',
  34                         'zh-sg'   => '新加坡',
  35                         'zh-my'   => '大马',
  36                 );
  37                 $this->mVariantNames = array_merge($this->mVariantNames,$names);
  38         }
  39
  40         function loadDefaultTables() {
  41                 require( dirname(__FILE__)."/../../includes/ZhConversion.php" );
  42                 $this->mTables = array(
  43                         'zh-hans' => new ReplacementArray( $zh2Hans ),
  44                         'zh-hant' => new ReplacementArray( $zh2Hant ),
  45                         'zh-cn'   => new ReplacementArray( array_merge($zh2Hans, $zh2CN) ),
  46                         'zh-hk'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  47                         'zh-mo'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  48                         'zh-my'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  49                         'zh-sg'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  50                         'zh-tw'   => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ),
  51                         'zh'      => new ReplacementArray
  52                 );
  53         }
  54
  55         function postLoadTables() {
  56                 $this->mTables['zh-cn']->merge( $this->mTables['zh-hans'] );
  57                 $this->mTables['zh-hk']->merge( $this->mTables['zh-hant'] );
  58                 $this->mTables['zh-mo']->merge( $this->mTables['zh-hant'] );
  59                 $this->mTables['zh-my']->merge( $this->mTables['zh-hans'] );
  60                 $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] );
  61                 $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] );
  62         }
  63
  64         /* there shouldn't be any latin text in Chinese conversion, so no need
  65            to mark anything.
  66            $noParse is there for compatibility with LanguageConvert::markNoConversion
  67          */
  68         function markNoConversion($text, $noParse = false) {
  69                 return $text;
  70         }
  71
  72         function convertCategoryKey( $key ) {
  73                 return $this->autoConvert( $key, 'zh' );
  74         }
  75 }
  76
  77 /**
  78  * class that handles both Traditional and Simplified Chinese
  79  * right now it only distinguish zh_hans, zh_hant, zh_cn, zh_tw, zh_sg and zh_hk.
  80  *
  81  * @ingroup Language
  82  */
  83 class LanguageZh extends LanguageZh_hans {
  84
  85         function __construct() {
  86                 global $wgHooks;
  87                 parent::__construct();
  88
  89                 $variants = array('zh','zh-hans','zh-hant','zh-cn','zh-hk','zh-mo','zh-my','zh-sg','zh-tw');
  90                 $variantfallbacks = array(
  91                         'zh'      => array('zh-hans','zh-hant','zh-cn','zh-tw','zh-hk','zh-sg','zh-mo','zh-my'),
  92                         'zh-hans' => array('zh-cn','zh-sg','zh-my'),
  93                         'zh-hant' => array('zh-tw','zh-hk','zh-mo'),
  94                         'zh-cn'   => array('zh-hans','zh-sg','zh-my'),
  95                         'zh-sg'   => array('zh-hans','zh-cn','zh-my'),
  96                         'zh-my'   => array('zh-hans','zh-sg','zh-cn'),
  97                         'zh-tw'   => array('zh-hant','zh-hk','zh-mo'),
  98                         'zh-hk'   => array('zh-hant','zh-mo','zh-tw'),
  99                         'zh-mo'   => array('zh-hant','zh-hk','zh-tw'),
 100                 );
 101                 $ml=array(
 102                         'zh'      => 'disable',
 103                         'zh-hans' => 'unidirectional',
 104                         'zh-hant' => 'unidirectional',
 105                 );
 106
 107                 $this->mConverter = new ZhConverter( $this, 'zh',
 108                                                                 $variants, $variantfallbacks,
 109                                                                 array(),array(),
 110                                                                 $ml);
 111
 112                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
 113         }
 114
 115         # this should give much better diff info
 116         function segmentForDiff( $text ) {
 117                 return preg_replace(
 118                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 119                         "' ' .\"$1\"", $text);
 120         }
 121
 122         function unsegmentForDiff( $text ) {
 123                 return preg_replace(
 124                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 125                         "\"$1\"", $text);
 126         }
 127
 128         // word segmentation
 129         function stripForSearch( $string ) {
 130                 wfProfileIn( __METHOD__ );
 131
 132                 // eventually this should be a word segmentation
 133                 // for now just treat each character as a word
 134                 // @fixme only do this for Han characters...
 135                 $t = preg_replace(
 136                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 137                                 " $1", $string);
 138
 139         //always convert to zh-hans before indexing. it should be
 140                 //better to use zh-hans for search, since conversion from
 141                 //Traditional to Simplified is less ambiguous than the
 142                 //other way around
 143
 144                 $t = $this->mConverter->autoConvert($t, 'zh-hans');
 145                 $t = parent::stripForSearch( $t );
 146                 wfProfileOut( __METHOD__ );
 147                 return $t;
 148
 149         }
 150
 151         function convertForSearchResult( $termsArray ) {
 152                 $terms = implode( '|', $termsArray );
 153                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 154                 $ret = array_unique( explode('|', $terms) );
 155                 return $ret;
 156         }
 157 }
 158