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                 $this->loadNamespaceTables();
  38         }
  39
  40         function loadNamespaceTables() {
  41                 global $wgMetaNamespace;
  42                 $nsproject     = $wgMetaNamespace;
  43                 $projecttable  = array(
  44                         'Wikipedia'       => '维基百科',
  45                         'Wikisource'      => '维基文库',
  46                         'Wikinews'        => '维基新闻',
  47                         'Wiktionary'      => '维基词典',
  48                         'Wikibooks'       => '维基教科书',
  49                         'Wikiquote'       => '维基语录',
  50                 );
  51                 $this->mNamespaceTables['zh-hans'] = array(
  52                         'Media'          => '媒体',
  53                         'Special'        => '特殊',
  54                         'Talk'           => '讨论',
  55                         'User'           => '用户',
  56                         'User talk'      => '用户讨论',
  57                         $nsproject
  58                                         => isset($projecttable[$nsproject]) ?
  59                                                 $projecttable[$nsproject] : $nsproject,
  60                         $nsproject . ' talk'
  61                                         => isset($projecttable[$nsproject]) ?
  62                                                 $projecttable[$nsproject] . '讨论' : $nsproject . '讨论',
  63                         'File'           => '文件',
  64                         'File talk'      => '文件讨论',
  65                         'MediaWiki'      => 'MediaWiki',
  66                         'MediaWiki talk' => 'MediaWiki讨论',
  67                         'Template'       => '模板',
  68                         'Template talk'  => '模板讨论',
  69                         'Help'           => '帮助',
  70                         'Help talk'      => '帮助讨论',
  71                         'Category'       => '分类',
  72                         'Category talk'  => '分类讨论',
  73                 );
  74                 $this->mNamespaceTables['zh-hant'] = array_merge($this->mNamespaceTables['zh-hans']);
  75                 $this->mNamespaceTables['zh-hant']['File'] = '檔案';
  76                 $this->mNamespaceTables['zh-hant']['File talk'] = '檔案討論';
  77                 $this->mNamespaceTables['zh'] = array_merge($this->mNamespaceTables['zh-hans']);
  78                 $this->mNamespaceTables['zh-cn'] = array_merge($this->mNamespaceTables['zh-hans']);
  79                 $this->mNamespaceTables['zh-hk'] = array_merge($this->mNamespaceTables['zh-hant']);
  80                 $this->mNamespaceTables['zh-mo'] = array_merge($this->mNamespaceTables['zh-hant']);
  81                 $this->mNamespaceTables['zh-my'] = array_merge($this->mNamespaceTables['zh-hans']);
  82                 $this->mNamespaceTables['zh-sg'] = array_merge($this->mNamespaceTables['zh-hans']);
  83                 $this->mNamespaceTables['zh-tw'] = array_merge($this->mNamespaceTables['zh-hant']);
  84         }
  85
  86         function loadDefaultTables() {
  87                 require( dirname(__FILE__)."/../../includes/ZhConversion.php" );
  88                 $this->mTables = array(
  89                         'zh-hans' => new ReplacementArray( $zh2Hans ),
  90                         'zh-hant' => new ReplacementArray( $zh2Hant ),
  91                         'zh-cn'   => new ReplacementArray( array_merge($zh2Hans, $zh2CN) ),
  92                         'zh-hk'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  93                         'zh-mo'   => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ),
  94                         'zh-my'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  95                         'zh-sg'   => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ),
  96                         'zh-tw'   => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ),
  97                         'zh'      => new ReplacementArray
  98                 );
  99         }
 100
 101         function postLoadTables() {
 102                 $this->mTables['zh-cn']->merge( $this->mTables['zh-hans'] );
 103                 $this->mTables['zh-hk']->merge( $this->mTables['zh-hant'] );
 104                 $this->mTables['zh-mo']->merge( $this->mTables['zh-hant'] );
 105                 $this->mTables['zh-my']->merge( $this->mTables['zh-hans'] );
 106                 $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] );
 107                 $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] );
 108         }
 109
 110         /* there shouldn't be any latin text in Chinese conversion, so no need
 111            to mark anything.
 112            $noParse is there for compatibility with LanguageConvert::markNoConversion
 113          */
 114         function markNoConversion($text, $noParse = false) {
 115                 return $text;
 116         }
 117
 118         function convertCategoryKey( $key ) {
 119                 return $this->autoConvert( $key, 'zh' );
 120         }
 121 }
 122
 123 /**
 124  * class that handles both Traditional and Simplified Chinese
 125  * right now it only distinguish zh_hans, zh_hant, zh_cn, zh_tw, zh_sg and zh_hk.
 126  *
 127  * @ingroup Language
 128  */
 129 class LanguageZh extends LanguageZh_hans {
 130
 131         function __construct() {
 132                 global $wgHooks;
 133                 parent::__construct();
 134
 135                 $variants = array('zh','zh-hans','zh-hant','zh-cn','zh-hk','zh-mo','zh-my','zh-sg','zh-tw');
 136                 $variantfallbacks = array(
 137                         'zh'      => array('zh-hans','zh-hant','zh-cn','zh-tw','zh-hk','zh-sg','zh-mo','zh-my'),
 138                         'zh-hans' => array('zh-cn','zh-sg','zh-my'),
 139                         'zh-hant' => array('zh-tw','zh-hk','zh-mo'),
 140                         'zh-cn'   => array('zh-hans','zh-sg','zh-my'),
 141                         'zh-sg'   => array('zh-hans','zh-cn','zh-my'),
 142                         'zh-my'   => array('zh-hans','zh-sg','zh-cn'),
 143                         'zh-tw'   => array('zh-hant','zh-hk','zh-mo'),
 144                         'zh-hk'   => array('zh-hant','zh-mo','zh-tw'),
 145                         'zh-mo'   => array('zh-hant','zh-hk','zh-tw'),
 146                 );
 147                 $ml=array(
 148                         'zh'      => 'disable',
 149                         'zh-hans' => 'unidirectional',
 150                         'zh-hant' => 'unidirectional',
 151                 );
 152
 153                 $this->mConverter = new ZhConverter( $this, 'zh',
 154                                                                 $variants, $variantfallbacks,
 155                                                                 array(),array(),
 156                                                                 $ml);
 157
 158                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
 159         }
 160
 161         # this should give much better diff info
 162         function segmentForDiff( $text ) {
 163                 return preg_replace(
 164                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 165                         "' ' .\"$1\"", $text);
 166         }
 167
 168         function unsegmentForDiff( $text ) {
 169                 return preg_replace(
 170                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
 171                         "\"$1\"", $text);
 172         }
 173
 174         // word segmentation
 175         function stripForSearch( $string ) {
 176                 wfProfileIn( __METHOD__ );
 177
 178                 // eventually this should be a word segmentation
 179                 // for now just treat each character as a word
 180                 // @fixme only do this for Han characters...
 181                 $t = preg_replace(
 182                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/",
 183                                 " $1", $string);
 184
 185         //always convert to zh-hans before indexing. it should be
 186                 //better to use zh-hans for search, since conversion from
 187                 //Traditional to Simplified is less ambiguous than the
 188                 //other way around
 189
 190                 $t = $this->mConverter->autoConvert($t, 'zh-hans');
 191                 $t = parent::stripForSearch( $t );
 192                 wfProfileOut( __METHOD__ );
 193                 return $t;
 194
 195         }
 196
 197         function convertForSearchResult( $termsArray ) {
 198                 $terms = implode( '|', $termsArray );
 199                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 200                 $ret = array_unique( explode('|', $terms) );
 201                 return $ret;
 202         }
 203 }
 204