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 /* class that handles both Traditional and Simplified Chinese
  43    right now it only distinguish zh_hans, zh_hant, zh_cn, zh_tw, zh_sg and zh_hk.
  44 */
  45 class LanguageZh extends LanguageZh_hans {
  46
  47         function __construct() {
  48                 global $wgHooks;
  49                 parent::__construct();
  50
  51                 $variants = array('zh', 'zh-hans', 'zh-hant', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk');
  52                 $variantfallbacks = array(
  53                         'zh'      => 'zh-hans',
  54                         'zh-hans' => 'zh',
  55                         'zh-hant' => 'zh',
  56                         'zh-cn'   => 'zh-hans',
  57                         'zh-sg'   => 'zh-hans',
  58                         'zh-tw'   => 'zh-hant',
  59                         'zh-hk'   => 'zh-hant',
  60                 );
  61
  62                 $this->mConverter = new ZhConverter( $this, 'zh', $variants, $variantfallbacks );
  63
  64                 $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
  65         }
  66
  67         # this should give much better diff info
  68         function segmentForDiff( $text ) {
  69                 return preg_replace(
  70                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  71                         "' ' .\"$1\"", $text);
  72         }
  73
  74         function unsegmentForDiff( $text ) {
  75                 return preg_replace(
  76                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  77                         "\"$1\"", $text);
  78         }
  79
  80         // word segmentation
  81         function stripForSearch( $string ) {
  82                 $fname="LanguageZh::stripForSearch";
  83                 wfProfileIn( $fname );
  84
  85                 // eventually this should be a word segmentation
  86                 // for now just treat each character as a word
  87                 $t = preg_replace(
  88                                 "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  89                                 "' ' .\"$1\"", $string);
  90
  91         //always convert to zh-hans before indexing. it should be
  92                 //better to use zh-hans for search, since conversion from
  93                 //Traditional to Simplified is less ambiguous than the
  94                 //other way around
  95
  96                 $t = $this->mConverter->autoConvert($t, 'zh-hans');
  97                 $t = parent::stripForSearch( $t );
  98                 wfProfileOut( $fname );
  99                 return $t;
 100
 101         }
 102
 103         function convertForSearchResult( $termsArray ) {
 104                 $terms = implode( '|', $termsArray );
 105                 $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
 106                 $ret = array_unique( explode('|', $terms) );
 107                 return $ret;
 108         }
 109 }