<?php
-require_once( "includes/ZhClient.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");
+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;
- var $mZhClient=false;
- function LanguageZh() {
- global $wgUseZhdaemon, $wgZhdaemonHost, $wgZhdaemonPort;
- global $wgDisableLangConversion, $wgUser;
- if($wgUseZhdaemon) {
- $this->mZhClient=new ZhClient($wgZhdaemonHost, $wgZhdaemonPort);
- if(!$this->mZhClient->isconnected())
- $this->mZhClient = false;
- }
- // fallback to fake client
- if($this->mZhClient == false)
- $this->mZhClient=new ZhClientFake();
+ 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;
}
-
- /*
- get preferred language variants.
- */
- function getPreferredVariant() {
- global $wgUser, $wgRequest;
-
- if($this->mZhLanguageCode)
- return $this->mZhLanguageCode;
-
- // see if the preference is set in the request
- $zhreq = $wgRequest->getText( 'variant' );
- if( in_array( $zhreq, $this->getVariants() ) ) {
- $this->mZhLanguageCode = $zhreq;
- return $zhreq;
- }
- // get language variant preference from logged in users
- if($wgUser->getID()!=0) {
- $this->mZhLanguageCode = $wgUser->getOption('variant');
- }
- if( !$this->mZhLanguageCode ) {
- // see if some zh- variant is set in the http header,
- $this->mZhLanguageCode="zh";
- $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
- $zh = strstr($header, 'zh-');
- if($zh) {
- $this->mZhLanguageCode = substr($zh,0,5);
- }
- }
- return $this->mZhLanguageCode;
- }
-
# this should give much better diff info
function segmentForDiff( $text ) {
return preg_replace(
"\"$1\"", $text);
}
- function autoConvert($text, $toVariant=false) {
- if(!$toVariant)
- $toVariant = $this->getPreferredVariant();
- if($toVariant == 'zh')
- return $text;
- $fname="zhautoConvert";
- wfProfileIn( $fname );
- $t = $this->mZhClient->convert($text, $toVariant);
- wfProfileOut( $fname );
- return $t;
- }
-
- function autoConvertToAllVariants($text) {
- $fname="zhautoConvertToAll";
+ // word segmentation
+ function stripForSearch( $string ) {
+ $fname="LanguageZh::stripForSearch";
wfProfileIn( $fname );
- $ret = $this->mZhClient->convertToAllVariants($text);
- if($ret == false) {//fall back...
- $ret = ZhClientFake::autoConvertToAllVariants($text);
- }
- wfProfileOut( $fname );
- return $ret;
- }
-
- # convert text to different variants of a language. the automatic
- # conversion is done in autoConvert(). here we parse the text
- # marked with -{}-, which specifies special conversions of the
- # text that can not be accomplished in autoConvert()
- #
- # syntax of the markup:
- # -{code1:text1;code2:text2;...}- or
- # -{text}- in which case no conversion should take place for text
- function convert( $text , $isTitle=false) {
- global $wgDisableLangConversion;
-
- if($wgDisableLangConversion)
- return $text;
-
- // no conversion if redirecting
- if(strtolower( substr( $text,0,9 ) ) == "#redirect") {
- return $text;
- }
-
- if( $isTitle ) {
- global $wgRequest;
- $isredir = $wgRequest->getText( 'redirect', 'yes' );
- $action = $wgRequest->getText( 'action' );
- if ( $isredir == 'no' || $action == 'edit' ) {
- return $text;
- }
- else {
- $text = $this->convertTitle($text);
- return $text;
- }
- }
-
- $plang = $this->getPreferredVariant();
- $fallback = $this->getVariantFallback($plang);
-
- $tarray = explode("-{", $text);
- $tfirst = array_shift($tarray);
- $text = $this->autoConvert($tfirst);
- foreach($tarray as $txt) {
- $marked = explode("}-", $txt);
-
- $choice = explode(";", $marked{0});
- if(!array_key_exists(1, $choice)) {
- /* a single choice */
- $text .= $choice{0};
- } else {
- $choice1=false;
- $choice2=false;
- foreach($choice as $c) {
- $v = explode(":", $c);
- if(!array_key_exists(1, $v)) {
- //syntax error in the markup, give up
- break;
- }
- $code = trim($v{0});
- $content = trim($v{1});
- if($code == $plang) {
- $choice1 = $content;
- break;
- }
- if($code == $fallback)
- $choice2 = $content;
- }
- if ( $choice1 )
- $text .= $choice1;
- elseif ( $choice2 )
- $text .= $choice2;
- else
- $text .= $marked{0};
- }
- if(array_key_exists(1, $marked))
- $text .= $this->autoConvert($marked{1});
- }
-
- return $text;
- }
-
-
- # only convert titles having more than one character
- function convertTitle($text) {
- $len=0;
- if( function_exists( 'mb_strlen' ) )
- $len = mb_strlen($text);
- else
- $len = strlen($text)/3;
- if($len>1)
- return $this->autoConvert( $text);
- return $text;
- }
- function getVariants() {
- return array("zh", "zh-cn", "zh-tw", "zh-sg", "zh-hk");
- }
+ // 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\"", $string);
- function getVariantFallback($v) {
- switch ($v) {
- case 'zh': return 'zh-cn'; break;
- case 'zh-cn': return 'zh-sg'; break;
- case 'zh-sg': return 'zh-cn'; break;
- case 'zh-tw': return 'zh-hk'; break;
- case 'zh-hk': return 'zh-tw'; break;
- }
- return false;
- }
-
- // word segmentation through ZhClient
- function stripForSearch( $string ) {
- $fname="zhsegment";
- wfProfileIn( $fname );
//always convert to zh-cn before indexing. it should be
- //better to use zh-cn for search, since conversion from
+ //better to use zh-cn for search, since conversion from
//Traditional to Simplified is less ambiguous than the
//other way around
- $t = $this->mZhClient->segment($string);
- $t = $this->autoConvert($t, 'zh-cn');
+
+ $t = $this->mConverter->autoConvert($t, 'zh-cn');
$t = LanguageUtf8::stripForSearch( $t );
wfProfileOut( $fname );
return $t;
function convertForSearchResult( $termsArray ) {
$terms = implode( '|', $termsArray );
- $terms = implode( '|', $this->autoConvertToAllVariants( $terms ) );
+ $terms = implode( '|', $this->mConverter->autoConvertToAllVariants( $terms ) );
$ret = array_unique( explode('|', $terms) );
return $ret;
}
- function findVariantLink( &$link, &$nt ) {
- static $count=0; //used to limit this operation
- static $cache=array();
- global $wgDisableLangConversion;
- $pref = $this->getPreferredVariant();
- if( $wgDisableLangConversion || $pref == 'zh' || $count > 50)
- return;
- $count++;
- $variants = $this->autoConvertToAllVariants($link);
- if($variants == false) //give up
- return;
- foreach( $variants as $v ) {
- if(isset($cache[$v]))
- continue;
- $cache[$v] = 1;
- $varnt = Title::newFromText( $v );
- if( $varnt && $varnt->getArticleID() > 0 ) {
- $nt = $varnt;
- $link = $v;
- break;
- }
- }
- }
-
- function getExtraHashOptions() {
- global $wgUser;
- $variant = $this->getPreferredVariant();
- return '!' . $variant ;
- }
}
?>