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, $wgUser;
  19
  20                 if( $wgUser->getID()!=0 ) {
  21                         /* allow user to diable conversion */
  22                         if( $wgDisableLangConversion == false &&
  23                                 $wgUser->getOption('nolangconversion') == 1)
  24                                 $wgDisableLangConversion = true;
  25                 }
  26
  27                 $this->mZhLanguageCode = $this->getPreferredVariant();
  28                 if($wgUseZhdaemon) {
  29                         $this->mZhClient=new ZhClient($wgZhdaemonHost, $wgZhdaemonPort);
  30                         if(!$this->mZhClient->isconnected())
  31                                 $this->mZhClient = false;
  32                 }
  33                 // fallback to fake client
  34                 if($this->mZhClient == false)
  35                         $this->mZhClient=new ZhClientFake();
  36         }
  37
  38         /*
  39                 get preferred language variants. eventually this will check the
  40                 user's preference setting as well, once the language option in
  41                 the setting pages is finalized.
  42         */
  43         function getPreferredVariant() {
  44                 global $wgUser;
  45
  46                 if($this->mZhLanguageCode)
  47                         return $this->mZhLanguageCode;
  48
  49                 // get language variant preference for logged in users
  50                 if($wgUser->getID()!=0) {
  51                         $this->mZhLanguageCode = $wgUser->getOption('variant');
  52                 }
  53                 else {
  54                         // see if some zh- variant is set in the http header,
  55                         $this->mZhLanguageCode="zh-cn";
  56                         $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
  57                         $zh = strstr($header, 'zh-');
  58                         if($zh) {
  59                                 $this->mZhLanguageCode = substr($zh,0,5);
  60                         }
  61                 }
  62                 return $this->mZhLanguageCode;
  63         }
  64
  65         # this should give much better diff info
  66         function segmentForDiff( $text ) {
  67                 return preg_replace(
  68                         "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  69                         "' ' .\"$1\"", $text);
  70         }
  71
  72         function unsegmentForDiff( $text ) {
  73                 return preg_replace(
  74                         "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
  75                         "\"$1\"", $text);
  76         }
  77
  78
  79
  80         function autoConvert($text, $toVariant=false) {
  81                 if(!$toVariant)
  82                         $toVariant = $this->getPreferredVariant();
  83                 $fname="zhautoConvert";
  84                 wfProfileIn( $fname );
  85                 $t = $this->mZhClient->convert($text, $toVariant);
  86                 wfProfileOut( $fname );
  87                 return $t;
  88         }
  89
  90         function autoConvertToAllVariants($text) {
  91                 $fname="zhautoConvertToAll";
  92                 wfProfileIn( $fname );
  93                 $ret = $this->mZhClient->convertToAllVariants($text);
  94                 if($ret == false) {//fall back...
  95                         $ret = Language::autoConvertToAllVariants($text);
  96                 }
  97                 wfProfileOut( $fname );
  98                 return $ret;
  99         }
 100
 101         # only convert titles having more than one character
 102         function convertTitle($text) {
 103                 $len=0;
 104                 if( function_exists( 'mb_strlen' ) )
 105                         $len = mb_strlen($text);
 106                 else
 107                         $len = strlen($text)/3;
 108                 if($len>1)
 109                         return $this->autoConvert( $text);
 110                 return $text;
 111         }
 112
 113         function getVariants() {
 114                 return array("zh-cn", "zh-tw", "zh-sg", "zh-hk");
 115         }
 116
 117         function getVariantFallback($v) {
 118                 switch ($v) {
 119                 case 'zh-cn': return 'zh-sg'; break;
 120                 case 'zh-sg': return 'zh-cn'; break;
 121                 case 'zh-tw': return 'zh-hk'; break;
 122                 case 'zh-hk': return 'zh-tw'; break;
 123                 }
 124                 return false;
 125         }
 126
 127         // word segmentation through ZhClient
 128         function stripForSearch( $string ) {
 129                 $fname="zhsegment";
 130                 wfProfileIn( $fname );
 131         //always convert to zh-cn before indexing. it should be
 132                 //better to use zh-cn for search, since conversion from
 133                 //Traditional to Simplified is less ambiguous than the
 134                 //other way around
 135                 $t = $this->mZhClient->segment($string);
 136         $t = $this->autoConvert($t, 'zh-cn');
 137                 $t = LanguageUtf8::stripForSearch( $t );
 138                 wfProfileOut( $fname );
 139                 return $t;
 140
 141         }
 142
 143         function convertForSearchResult( $termsArray ) {
 144                 $terms = implode( '|', $termsArray );
 145                 $terms = implode( '|', $this->autoConvertToAllVariants( $terms ) );
 146                 $ret = array_unique( explode('|', $terms) );
 147                 return $ret;
 148         }
 149
 150         function getExtraHashOptions() {
 151                 return array('variant', 'nolangconversion');
 152         }
 153 }
 154 ?>