languages/LanguageZh.php

   1 <?php
   2 require_once( "includes/ZhClient.php" );
   3 require_once( "LanguageZh_cn.php");
   4 require_once( "LanguageZh_tw.php");
   5 require_once( "LanguageZh_sg.php");
   6 require_once( "LanguageZh_hk.php");
   7
   8 /* class that handles both Traditional and Simplified Chinese
   9    right now it only distinguish zh_cn and zh_tw (actuall, zh_cn and
  10    non-zh_cn), will add support for zh_sg, zh_hk, etc, later.
  11 */
  12 class LanguageZh extends LanguageZh_cn {
  13
  14         var $mZhLanguageCode=false;
  15         var $mZhClient=false;
  16         function LanguageZh() {
  17                 global $wgUseZhdaemon, $wgZhdaemonHost, $wgZhdaemonPort;
  18                 global $wgDisableLangConversion;
  19
  20                 $this->mZhLanguageCode = $this->getPreferredVariant();
  21                 if($wgUseZhdaemon) {
  22                         $this->mZhClient=new ZhClient($wgZhdaemonHost, $wgZhdaemonPort);
  23                         if(!$this->mZhClient->isconnected())
  24                                 $this->mZhClient = false;
  25                 }
  26                 // fallback to fake client
  27                 if($this->mZhClient == false)
  28                         $this->mZhClient=new ZhClientFake();
  29         }
  30
  31         /*
  32                 get preferred language variants. eventually this will check the
  33                 user's preference setting as well, once the language option in
  34                 the setting pages is finalized.
  35         */
  36         function getPreferredVariant() {
  37                 global $wgUser;
  38
  39                 if($this->mZhLanguageCode)
  40                         return $this->mZhLanguageCode;
  41
  42                 // get language variant preference for logged in users
  43                 if($wgUser->getID()!=0) {
  44                         $this->mZhLanguageCode = $wgUser->getOption('variant');
  45                 }
  46                 else {
  47                         // see if some zh- variant is set in the http header,
  48                         $this->mZhLanguageCode="zh-cn";
  49                         $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
  50                         $zh = strstr($header, 'zh-');
  51                         if($zh) {
  52                                 $this->mZhLanguageCode = substr($zh,0,5);
  53                         }
  54                 }
  55                 return $this->mZhLanguageCode;
  56         }
  57
  58         # this should give much better diff info
  59         function segmentForDiff( $text ) {
  60                 return preg_replace(
  61                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  62                         "' ' .\"$1\"", $text);
  63         }
  64
  65         function unsegmentForDiff( $text ) {
  66                 return preg_replace(
  67                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  68                         "\"$1\"", $text);
  69         }
  70
  71
  72
  73         function autoConvert($text, $toVariant=false) {
  74                 if(!$toVariant)
  75                         $toVariant = $this->getPreferredVariant();
  76                 $fname="zhautoConvert";
  77                 wfProfileIn( $fname );
  78                 $t = $this->mZhClient->convert($text, $toVariant);
  79                 wfProfileOut( $fname );
  80                 return $t;
  81         }
  82
  83         function autoConvertToAllVariants($text) {
  84                 $fname="zhautoConvertToAll";
  85                 wfProfileIn( $fname );
  86                 $ret = $this->mZhClient->convertToAllVariants($text);
  87                 if($ret == false) {//fall back...
  88                         $ret = Language::autoConvertToAllVariants($text);
  89                 }
  90                 wfProfileOut( $fname );
  91                 return $ret;
  92         }
  93
  94         # only convert titles having more than one character
  95         function convertTitle($text) {
  96                 $len=0;
  97                 if( function_exists( 'mb_strlen' ) )
  98                         $len = mb_strlen($text);
  99                 else
 100                         $len = strlen($text)/3;
 101                 if($len>1)
 102                         return $this->autoConvert( $text);
 103                 return $text;
 104         }
 105
 106         function getVariants() {
 107                 return array("zh-cn", "zh-tw", "zh-sg", "zh-hk");
 108         }
 109
 110         function getVariantFallback($v) {
 111                 switch ($v) {
 112                 case 'zh-cn': return 'zh-sg'; break;
 113                 case 'zh-sg': return 'zh-cn'; break;
 114                 case 'zh-tw': return 'zh-hk'; break;
 115                 case 'zh-hk': return 'zh-tw'; break;
 116                 }
 117                 return false;
 118         }
 119
 120         // word segmentation through ZhClient
 121         function stripForSearch( $string ) {
 122                 $fname="zhsegment";
 123                 wfProfileIn( $fname );
 124         //always convert to zh-cn before indexing. it should be
 125                 //better to use zh-cn for search, since conversion from
 126                 //Traditional to Simplified is less ambiguous than the
 127                 //other way around
 128                 $t = $this->mZhClient->segment($string);
 129         $t = $this->autoConvert($t, 'zh-cn');
 130                 $t = LanguageUtf8::stripForSearch( $t );
 131                 wfProfileOut( $fname );
 132                 return $t;
 133
 134         }
 135
 136         function convertForSearchResult( $termsArray ) {
 137                 $terms = implode( '|', $termsArray );
 138                 $terms = implode( '|', $this->autoConvertToAllVariants( $terms ) );
 139                 $ret = array_unique( explode('|', $terms) );
 140                 return $ret;
 141         }
 142 }
 143 ?>