Handle automatic reassignment of edits from duplicate accounts to the main

[lhc/web/wiklou.git] / languages / LanguageZh.php

diff --git a/languages/LanguageZh.php b/languages/LanguageZh.php
index 2034e1f..f7f3255 100644 (file)
--- a/languages/LanguageZh.php
+++ b/languages/LanguageZh.php
@@ -1,78 +1,103 @@
 <?php
+/**
+  * @package MediaWiki
+  * @subpackage Language
+  */
+require_once( "LanguageConverter.php" );
 require_once( "LanguageZh_cn.php");
+require_once( "LanguageZh_tw.php");
+require_once( "LanguageZh_sg.php");
+require_once( "LanguageZh_hk.php");
 
-/* caching the conversion tables */
-$zhSimp2Trad = $wgMemc->get($key1 = "$wgDBname:zhConvert:s2t");
-$zhTrad2Simp = $wgMemc->get($key2 = "$wgDBname:zhConvert:t2s");
-if(empty($zhSimp2Trad) || empty($zhTrad2Simp)) {
-    require_once("includes/ZhConversion.php");
-    $wgMemc->set($key1, $zhSimp2Trad);
-    $wgMemc->set($key2, $zhTrad2Simp);
+class ZhConverter extends LanguageConverter {
+       function loadDefaultTables() {
+               require( "includes/ZhConversion.php" );
+               $this->mTables = array();
+               $this->mTables['zh-cn'] = $zh2CN;
+               $this->mTables['zh-tw'] = $zh2TW;
+               $this->mTables['zh-sg'] = array_merge($zh2CN, $zh2SG);
+               $this->mTables['zh-hk'] = array_merge($zh2TW, $zh2HK);
+               $this->mTables['zh'] = array();
+       }
+
+       function postLoadTables() {
+               $this->mTables['zh-sg'] = array_merge($this->mTables['zh-cn'], $this->mTables['zh-sg']);
+               $this->mTables['zh-hk'] = array_merge($this->mTables['zh-tw'], $this->mTables['zh-hk']);
+    }
+
+       /* there shouldn't be any latin text in Chinese conversion, so no need 
+          to mark anything
+    */
+       function markNoConversion($text) {
+               return $text;
+       }
+
+       function convertCategoryKey( $key ) {
+               return $this->autoConvert( $key, 'zh-cn' ); 
+       }
 }
 
+
 /* class that handles both Traditional and Simplified Chinese
-   right now it only distinguish zh_cn and zh_tw (actuall, zh_cn and
-   non-zh_cn), will add support for zh_sg, zh_hk, etc, later.
+   right now it only distinguish zh_cn, zh_tw, zh_sg and zh_hk.
 */
 class LanguageZh extends LanguageZh_cn {
-    
-    var $mZhLanguageCode=false;
+       
+       function LanguageZh() {
+               global $wgHooks;
+               $this->mConverter = new ZhConverter($this, 'zh', 
+                                            array('zh', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk'),
+                                                                                       array('zh'=>'zh-cn',
+                                                                                                 'zh-cn'=>'zh-sg',
+                                                                                                 'zh-sg'=>'zh-cn',
+                                                                                                 'zh-tw'=>'zh-hk',
+                                                                                                 'zh-hk'=>'zh-tw'));      
+               $wgHooks['ArticleSaveComplete'][] = $this->mConverter;
+       }
 
-    function LanguageZh() {
-        $this->mZhLanguageCode = $this->getPreferredVariant();
-    }
+       
+       # this should give much better diff info
+       function segmentForDiff( $text ) {
+               return preg_replace(
+                       "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
+                       "' ' .\"$1\"", $text);
+       }
 
-    /* 
-       get preferred language variants. eventually this will check the
-       user's preference setting as well, once the language option in
-       the setting pages is finalized.
-    */
-    function getPreferredVariant() {
-        global $wgUser;
+       function unsegmentForDiff( $text ) {
+               return preg_replace(
+                       "/ ([\\xc0-\\xff][\\x80-\\xbf]*)/e",
+                       "\"$1\"", $text);
+       }
 
-        if($this->mZhLanguageCode)
-            return $this->mZhLanguageCode;
+       // word segmentation
+       function stripForSearch( $string ) {
+               $fname="LanguageZh::stripForSearch";
+               wfProfileIn( $fname );
 
-        // get language variant preference for logged in users 
-        if($wgUser->getID()!=0) {
-            $this->mZhLanguageCode = $wgUser->getOption('variant');
-        }
-        else {
-            // see if some zh- variant is set in the http header,
-            $this->mZhLanguageCode="zh-cn";
-            $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
-            $zh = strstr($header, 'zh-');
-            if($zh) {
-                $this->mZhLanguageCode = substr($zh,0,5);
-            }
-        }
-        return $this->mZhLanguageCode;
-    }
-    
-    
-  /* the Simplified/Traditional conversion stuff */
+               // eventually this should be a word segmentation
+               // for now just treat each character as a word
+               $t = preg_replace(
+                               "/([\\xc0-\\xff][\\x80-\\xbf]*)/e",
+                               "' ' .\"$1\"", $string);
+
+        //always convert to zh-cn before indexing. it should be
+               //better to use zh-cn for search, since conversion from 
+               //Traditional to Simplified is less ambiguous than the
+               //other way around
+
+               $t = $this->mConverter->autoConvert($t, 'zh-cn');
+               $t = LanguageUtf8::stripForSearch( $t );
+               wfProfileOut( $fname );
+               return $t;
 
-       function simp2trad($text) {
-               global $zhSimp2Trad;
-               return strtr($text, $zhSimp2Trad);
        }
 
-       function trad2simp($text) {
-               global $zhTrad2Simp;
-               return strtr($text, $zhTrad2Simp);
+       function convertForSearchResult( $termsArray ) {
+               $terms = implode( '|', $termsArray );
+               $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
+               $ret = array_unique( explode('|', $terms) );
+               return $ret;
        }
-       
-       function autoConvert($text) {
-        if($this->getPreferredVariant() == "zh-cn") {
-            return $this->trad2simp($text);
-        }
-        else {
-            return $this->simp2trad($text);
-        }
-    }
 
-    function getVariants() {
-        return array("zh-cn", "zh-tw");
-    }
 }
 ?>