<?php
/**
- * @ingroup Language
- *
- * @author Zhengzhu Feng <zhengzhu@gmail.com>
- * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
- * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>
- */
+ * Contains the LanguageConverter class and ConverterRule class
+ * @ingroup Language
+ *
+ * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
+ * @file
+ */
+/**
+ * base class for language convert
+ * @ingroup Language
+ *
+ * @author Zhengzhu Feng <zhengzhu@gmail.com>
+ * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
+ */
class LanguageConverter {
var $mPreferredVariant='';
var $mMainLanguageCode;
var $mVariants, $mVariantFallbacks, $mVariantNames;
var $mTablesLoaded = false;
var $mTables;
- var $mTitleDisplay='';
+ var $mNamespaceTables;
var $mDoTitleConvert=true, $mDoContentConvert=true;
var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
- var $mManualCodeError='<span style="color: red;">code error!</span>';
var $mTitleFromFlag = false;
var $mCacheKey;
var $mLangObj;
var $mMarkup;
var $mFlags;
+ var $mDescCodeSep = ':',$mDescVarSep = ';';
var $mUcfirst = false;
+ var $mTitleOriginal = '';
+ var $mTitleDisplay = '';
const CACHE_VERSION_KEY = 'VERSION 6';
* @param array $variantfallback the fallback language of each variant
* @param array $markup array defining the markup used for manual conversion
* @param array $flags array defining the custom strings that maps to the flags
- * @access public
+ * @param array $manualLevel limit for supported variants
+ * @public
*/
- function __construct($langobj, $maincode,
+ function __construct( $langobj, $maincode,
$variants=array(),
$variantfallbacks=array(),
$markup=array(),
$manualLevel = array() ) {
$this->mLangObj = $langobj;
$this->mMainLanguageCode = $maincode;
- $this->mVariants = $variants;
+
+ global $wgDisabledVariants;
+ $this->mVariants = array();
+ foreach( $variants as $variant ) {
+ if( !in_array( $variant, $wgDisabledVariants ) )
+ $this->mVariants[] = $variant;
+ }
$this->mVariantFallbacks = $variantfallbacks;
global $wgLanguageNames;
$this->mVariantNames = $wgLanguageNames;
'N'=>'N' // current variant name
);
$this->mFlags = array_merge($f, $flags);
- foreach( $this->mVariants as $v)
+ foreach( $this->mVariants as $v) {
$this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
?$manualLevel[$v]
:'bidirectional';
+ $this->mNamespaceTables[$v] = array();
+ $this->mFlags[$v] = $v;
+ }
}
/**
- * @access public
+ * @public
*/
function getVariants() {
return $this->mVariants;
*
* @param string $v the language code of the variant
* @return string array the code of the fallback language or false if there is no fallback
- * @private
+ * @public
*/
function getVariantFallbacks($v) {
if( isset( $this->mVariantFallbacks[$v] ) ) {
return $this->mMainLanguageCode;
}
- /**
- * check if variants array in convert array
- *
- * @param string $variant Variant language code
- * @param array $carray convert array
- * @param string $text Text to convert
- * @return string Translated text
- * @private
- */
- function getTextInCArray($variants,$carray){
- if(is_string($variants)){ $variants=array($variants); }
- if(!is_array($variants)) return false;
- foreach ($variants as $variant){
- if(array_key_exists($variant, $carray)){
- return $carray[$variant];
- }
- }
- return false;
- }
-
/**
* get preferred language variants.
* @param boolean $fromUser Get it from $wgUser's preferences
+ * @param boolean $fromHeader Get it from Accept-Language
* @return string the preferred language code
- * @access public
+ * @public
*/
- function getPreferredVariant( $fromUser = true ) {
- global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
+ function getPreferredVariant( $fromUser = true, $fromHeader = false ) {
+ global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant, $wgOut;
- if($this->mPreferredVariant)
+ // bug 21974, don't return $this->mPreferredVariant if $fromUser = false
+ if( $fromUser && $this->mPreferredVariant )
return $this->mPreferredVariant;
+ // figure out user lang without constructing wgLang to avoid infinite recursion
+ if( $fromUser )
+ $defaultUserLang = $wgUser->getOption( 'language' );
+ else
+ $defaultUserLang = $this->mMainLanguageCode;
+ $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
+ // see if interface language is same as content, if not, prevent conversion
+ if( ! in_array( $userLang, $this->mVariants ) ){
+ $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
+ return $this->mPreferredVariant;
+ }
+
// see if the preference is set in the request
$req = $wgRequest->getText( 'variant' );
if( in_array( $req, $this->mVariants ) ) {
}
// see if default variant is globaly set
- if($wgDefaultLanguageVariant != false && in_array( $wgDefaultLanguageVariant, $this->mVariants )){
+ if($wgDefaultLanguageVariant != false && in_array( $wgDefaultLanguageVariant, $this->mVariants )){
$this->mPreferredVariant = $wgDefaultLanguageVariant;
return $this->mPreferredVariant;
}
- # FIXME rewrite code for parsing http header. The current code
- # is written specific for detecting zh- variants
if( !$this->mPreferredVariant ) {
// see if some supported language variant is set in the
// http header, but we don't set the mPreferredVariant
// variable in case this is called before the user's
// preference is loaded
- $pv=$this->mMainLanguageCode;
- if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
- $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
- $zh = strstr($header, $pv.'-');
- if($zh) {
- $pv = substr($zh,0,5);
+ if( $fromHeader && array_key_exists( 'HTTP_ACCEPT_LANGUAGE', $_SERVER ) ) {
+ $acceptLanguage = strtolower( $_SERVER['HTTP_ACCEPT_LANGUAGE'] );
+ // explode by comma
+ $result = explode(',', $acceptLanguage);
+
+ $languages = array();
+
+ foreach( $result as $elem ) {
+ // if $elem likes 'zh-cn;q=0.9'
+ if(($posi = strpos( $elem, ';' )) !== false ) {
+ // get the real language code likes 'zh-cn'
+ $languages[] = substr( $elem, 0, $posi );
+ }
+ else {
+ $languages[] = $elem;
+ }
+ }
+
+ $fallback_languages = array();
+ foreach( $languages as $language ) {
+ // strip whitespace
+ $language = trim( $language );
+ if( in_array( $language, $this->mVariants ) ) {
+ return $language;
+ }
+ else {
+ // To see if there are fallbacks of current language.
+ // We record these fallback variants, and process
+ // them later.
+ $fallbacks = $this->getVariantFallbacks( $language );
+ if( is_string( $fallbacks ) )
+ $fallback_languages[] = $fallbacks;
+ elseif( is_array( $fallbacks ) )
+ $fallback_languages = array_merge( $fallback_languages, $fallbacks );
+ }
+ }
+
+ // process fallback languages now
+ $fallback_languages = array_unique( $fallback_languages );
+ foreach( $fallback_languages as $language ) {
+ if( in_array( $language, $this->mVariants ) ) {
+ return $language;
+ }
}
}
- // don't try to return bad variant
- if(in_array( $pv, $this->mVariants ))
- return $pv;
}
-
return $this->mMainLanguageCode;
-
}
+ /**
+ * caption convert, base on preg_replace_callback
+ *
+ * to convert text in "title" or "alt", like '<img alt="text" ... '
+ * or '<span title="text" ... '
+ *
+ * @return string like ' alt="yyyy"' or ' title="yyyy"'
+ * @private
+ */
+ function captionConvert( $matches ) {
+ $toVariant = $this->getPreferredVariant();
+ $title = $matches[1];
+ $text = $matches[2];
+ // we convert captions except URL
+ if( !strpos( $text, '://' ) )
+ $text = $this->translate($text, $toVariant);
+ return " $title=\"$text\"";
+ }
+
/**
* dictionary-based conversion
*
$ret = $this->translate($m[0], $toVariant);
$mstart = $m[1]+strlen($m[0]);
+
+ // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
+ $captionpattern = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
+
+ $trtext = '';
+ $trtextmark = "\0";
+ $notrtext = array();
foreach($matches as $m) {
- $ret .= substr($text, $mstart, $m[1]-$mstart);
- $ret .= $this->translate($m[0], $toVariant);
+ $mark = substr($text, $mstart, $m[1]-$mstart);
+ $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
+ // Let's convert the trtext only once,
+ // it would give us more performance improvement
+ $notrtext[] = $mark;
+ $trtext .= $m[0] . $trtextmark;
$mstart = $m[1] + strlen($m[0]);
}
+ $notrtext[] = '';
+ $trtext = $this->translate( $trtext, $toVariant );
+ $trtext = StringUtils::explode( $trtextmark, $trtext );
+ foreach( $trtext as $t ) {
+ $ret .= array_shift($notrtext);
+ $ret .= $t;
+ }
wfProfileOut( $fname );
return $ret;
}
*/
function translate( $text, $variant ) {
wfProfileIn( __METHOD__ );
- if( !$this->mTablesLoaded )
- $this->loadTables();
- $text = $this->mTables[$variant]->replace( $text );
+ // If $text is empty or only includes spaces, do nothing
+ // Otherwise translate it
+ if( trim($text) ) {
+ if( !$this->mTablesLoaded )
+ $this->loadTables();
+ $text = $this->mTables[$variant]->replace( $text );
+ }
wfProfileOut( __METHOD__ );
return $text;
}
return $ret;
}
+
+ /**
+ * prepare manual conversion table
+ * @private
+ */
+ function applyManualConv( $convRule ){
+ // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
+ $title = $convRule->getTitle();
+ if( $title ){
+ $this->mTitleFromFlag = true;
+ $this->mTitleDisplay = $title;
+ }
+ //apply manual conversion table to global table
+ $convTable = $convRule->getConvTable();
+ $action = $convRule->getRulesAction();
+ foreach( $convTable as $variant => $pair ) {
+ if( !in_array( $variant, $this->mVariants ) )continue;
+ if( $action == 'add' ) {
+ foreach( $pair as $from => $to ) {
+ // to ensure that $from and $to not be left blank
+ // so $this->translate() could always return a string
+ if ( $from || $to )
+ // more efficient than array_merge(), about 2.5 times.
+ $this->mTables[$variant]->setPair( $from, $to );
+ }
+ }
+ elseif ( $action == 'remove' ) {
+ $this->mTables[$variant]->removeArray( $pair );
+ }
+ }
+ }
/**
* Convert text using a parser object for context
return $text;
}
- if($wgDisableLangConversion)
+ if ( $wgDisableLangConversion )
return $text;
$text = $this->convert( $text );
- $parser->mOutput->setTitleText( $this->mTitleDisplay );
- return $text;
- }
- /**
- * Parse flags with syntax -{FLAG| ... }-
- * @private
- */
- function parseFlags($marked){
- $flags = array();
-
- // for multi-FLAGs
- if(strlen($marked) < 2 )
- return array($marked,array('R'));
-
- $tt = explode($this->mMarkup['flagsep'], $marked, 2);
-
- if(count($tt) == 2) {
- $f = explode($this->mMarkup['varsep'], $tt[0]);
- foreach($f as $ff) {
- $ff = trim($ff);
- if(array_key_exists($ff, $this->mFlags) &&
- !in_array($this->mFlags[$ff], $flags))
- $flags[] = $this->mFlags[$ff];
- }
- $rules = $tt[1];
- } else {
- $rules = $marked;
- }
-
- if( !in_array('R',$flags) ){
- //FIXME: may cause trouble here...
- //strip since it interferes with the parsing, plus,
- //all spaces should be stripped in this tag anyway.
- $rules = str_replace(' ', '', $rules);
- $rules = str_replace('=>','=>',$rules);
- }
-
- //check flags
- if( in_array('R',$flags) ){
- $flags = array('R');// remove other flags
- } elseif ( in_array('N',$flags) ){
- $flags = array('N');// remove other flags
- } elseif ( in_array('-',$flags) ){
- $flags = array('-');// remove other flags
- } elseif (count($flags)==1 && $flags[0]=='T'){
- $flags[]='H';
- } elseif ( in_array('H',$flags) ){
- // replace A flag, and remove other flags except T
- $temp=array('+','H');
- if(in_array('T',$flags)) $temp[] = 'T';
- if(in_array('D',$flags)) $temp[] = 'D';
- $flags = $temp;
- } else {
- if ( in_array('A',$flags)) {
- $flags[]='+';
- $flags[]='S';
- }
- if ( in_array('D',$flags) )
- $flags=array_diff($flags,array('S'));
- }
- if ( count($flags)==0 )
- $flags = array('S');
- return array($rules,$flags);
- }
+ $this->convertTitle();
+ $parser->mOutput->setTitleText( $this->mTitleDisplay );
- /**
- * @private
- */
- function getRulesDesc($bidtable,$unidtable){
- $text='';
- foreach($bidtable as $k => $v)
- $text .= $this->mVariantNames[$k].':'.$v.';';
- foreach($unidtable as $k => $a)
- foreach($a as $from=>$to)
- $text.=$from.'⇒'.$this->mVariantNames[$k].':'.$to.';';
return $text;
}
-
+
/**
- * parse the manually marked conversion rule
- * @param string $rule the text of the rule
- * @return array of the translation in each variant
+ * convert namespace
+ * @param string $title the title included namespace
+ * @return array of string
* @private
*/
- function getConvTableFromRules($rules,$flags=array()) {
- $bidtable = array();
- $unidtable = array();
- $choice = explode($this->mMarkup['varsep'], $rules );
- foreach($choice as $c) {
- $v = explode($this->mMarkup['codesep'], $c);
- if(count($v) != 2)
- continue;// syntax error, skip
- $to=trim($v[1]);
- $v=trim($v[0]);
- $u = explode($this->mMarkup['unidsep'], $v);
- if(count($u) == 1) {
- $bidtable[$v] = $to;
- } else if(count($u) == 2){
- $from=trim($u[0]);$v=trim($u[1]);
- if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
- $unidtable[$v]=array($from=>$to);
- else
- $unidtable[$v][$from]=$to;
- }
- // syntax error, pass
- }
- return array($bidtable,$unidtable);
+ function convertNamespace( $title, $variant ) {
+ $splittitle = explode( ':', $title );
+ if (count($splittitle) < 2)
+ return $title;
+ if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
+ $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
+ $ret = implode(':', $splittitle );
+ return $ret;
}
/**
- * get display text on markup -{...}-
- * @param string $rules the original code
- * @param array $flags FLAGs
- * @param array $bidtable bidirectional convert table
- * @param string $unidtable unidirectional convert table
- * @param string $variant the current variant
- * @param bool $$doConvert if do convert
+ * Pre convert title. Store the original title $this->mTitleOrginal;
+ * store the default converted title to $this->mTitleDisplay.
* @private
*/
- function getRulesDisplay($rules,$flags,
- $bidtable,$unidtable,
- $variant=false,$doConvert=true){
- if(!$variant) $variant = $this->getPreferredVariant();
- $is_mc_disable = $this->mManualLevel[$variant]=='disable';
-
- if( in_array('R',$flags) ) {
- // if we don't do content convert, still strip the -{}- tags
- $disp = $rules;
- } elseif ( in_array('N',$flags) ){
- // proces N flag: output current variant name
- $disp = $this->mVariantNames[trim($rules)];
- } elseif ( in_array('D',$flags) ){
- // proces D flag: output rules description
- $disp = $this->getRulesDesc($bidtable,$unidtable);
- } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
- // proces H,- flag or T only: output nothing
- $disp = '';
- } elseif ( in_array('S',$flags) ){
- if( count($bidtable) + count($unidtable) == 0 ){
- $disp = $rules;
- } elseif ($doConvert){// the text converted
- // display current variant in bidirectional array
- $disp = $this->getTextInCArray($variant,$bidtable);
- // or display current variant in fallbacks
- if(!$disp)
- $disp = $this->getTextInCArray($this->getVariantFallbacks($variant),$bidtable);
- // or display current variant in unidirectional array
- if(!$disp && array_key_exists($variant,$unidtable)){
- $disp = array_values($unidtable[$variant]);
- $disp = $disp[0];
- }
- // or display frist text under disable manual convert
- if(!$disp && $is_mc_disable) {
- if(count($bidtable)>0){
- $disp = array_values($bidtable);
- $disp = $disp[0];
- } else {
- $disp = array_values($unidtable);
- $disp = array_values($disp[0]);
- $disp = $disp[0];
- }
- }
- } else {// no convert
- $disp = $rules;
- }
- } elseif ( in_array('T',$flags) ) {
- // proces T flag : output nothing
- $disp = '';
- }
- else
- $disp= $this->mManualCodeError;
-
- return $disp;
- }
-
- /**
- * @access private
- */
- function applyManualFlag($rules,$flags,$bidtable,$unidtable,$variant=false){
- if(!$variant) $variant = $this->getPreferredVariant();
-
- $is_title_flag = in_array('T', $flags);
- // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
- if($is_title_flag){
- $this->mTitleFromFlag = true;
- $this->mTitleDisplay = $this->getRulesDisplay($rules,array('S'),
- $bidtable,$unidtable,
- $variant,
- $this->mDoTitleConvert);
- }
-
- if($this->mManualLevel[$variant]=='disable') return;
-
- $is_remove_flag = !$is_title_flag && in_array('-', $flags);
- $is_add_flag = !$is_remove_flag && in_array('+', $flags);
- $is_bidMC = $this->mManualLevel[$variant]=='bidirectional';
- $is_unidMC = $this->mManualLevel[$variant]=='unidirectional';
- $vmarked=array();
-
- foreach($this->mVariants as $v) {
- /* for bidirectional array
- fill in the missing variants, if any,
- with fallbacks */
- if($is_bidMC && !array_key_exists($v, $bidtable)) {
- $vf = $this->getTextInCArray($this->getVariantFallbacks($v),$bidtable);
- if($vf) $bidtable[$v] = $vf;
- }
- if($is_bidMC && array_key_exists($v,$bidtable)){
- foreach($vmarked as $vo){
- // use syntax:
- // -{A|zh:WordZh;zh-tw:WordTw}- or -{+|zh:WordZh;zh-tw:WordTw}-
- // to introduce a custom mapping between
- // words WordZh and WordTw in the whole text
- if($is_add_flag){
- $this->mTables[$v]->setPair($bidtable[$vo], $bidtable[$v]);
- $this->mTables[$vo]->setPair($bidtable[$v], $bidtable[$vo]);
- }
- // use syntax -{-|zh:WordZh;zh-tw:WordTw}- to remove a conversion
- // words WordZh and WordTw in the whole text
- if($is_remove_flag){
- $this->mTables[$v]->removePair($bidtable[$vo]);
- $this->mTables[$vo]->removePair($bidtable[$v]);
- }
- }
- $vmarked[]=$v;
- }
- /*for unidirectional array
- fill to convert tables */
- if($is_unidMC && array_key_exists($v,$unidtable)){
- if($is_add_flag)$this->mTables[$v]->mergeArray($unidtable[$v]);
- if($is_remove_flag)$this->mTables[$v]->removeArray($unidtable[$v]);
- }
- }
+ function preConvertTitle( $text, $variant ){
+ $this->mTitleOriginal = $text;
+
+ $text = $this->convertNamespace( $text, $variant );
+ $this->mTitleDisplay = $this->convert( $text );
}
/**
- * Parse rules and flags
+ * convert title
* @private
*/
- function parseRules($rules,$flags,$variant=false){
- if(!$variant) $variant = $this->getPreferredVariant();
-
- list($bidtable,$unidtable) = $this->getConvTableFromRules($rules, $flags);
- if(count($bidtable)==0 && count($unidtable)==0
- && !in_array('N',$flags) && !in_array('T',$flags) )
- $flags = array('R');
- $disp = $this->getRulesDisplay($rules,$flags,
- $bidtable,$unidtable,
- $variant,
- $this->mDoContentConvert);
- $this->applyManualFlag($rules,$flags,$bidtable,$unidtable);
-
- return $disp;
- }
-
- /**
- * convert title
- * @private
- */
- function convertTitle($text){
- // check for __NOTC__ tag
- if( !$this->mDoTitleConvert ) {
- $this->mTitleDisplay = $text;
- return $text;
- }
-
- // use the title from the T flag if any
- if($this->mTitleFromFlag){
- $this->mTitleFromFlag = false;
- return $this->mTitleDisplay;
- }
-
- global $wgRequest;
+ function convertTitle(){
+ global $wgDisableTitleConversion, $wgUser, $wgRequest;
$isredir = $wgRequest->getText( 'redirect', 'yes' );
$action = $wgRequest->getText( 'action' );
- if ( $isredir == 'no' || $action == 'edit' ) {
- return $text;
- } else {
- $this->mTitleDisplay = $this->convert($text);
- return $this->mTitleDisplay;
+ $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
+
+ // check for the global variable, __NOTC__ magic word, and user setting
+ if( $wgDisableTitleConversion || !$this->mDoTitleConvert ||
+ $wgUser->getOption('noconvertlink') == 1 ) {
+ $this->mTitleDisplay = $this->mTitleOriginal;
+ }
+
+ // check for GET params
+ elseif ( $isredir == 'no' || $action == 'edit' || $linkconvert == 'no' ) {
+ $this->mTitleDisplay = $this->mTitleOriginal;
}
}
* @param string $text text to be converted
* @param bool $isTitle whether this conversion is for the article title
* @return string converted text
- * @access public
+ * @public
*/
- function convert( $text , $isTitle=false) {
+ function convert( $text, $isTitle = false ) {
$mw =& MagicWord::get( 'notitleconvert' );
if( $mw->matchAndRemove( $text ) )
// no conversion if redirecting
$mw =& MagicWord::get( 'redirect' );
- if( $mw->matchStart( $text ))
+ if( $mw->matchStart( $text ) )
return $text;
+ $plang = $this->getPreferredVariant();
+
// for title convertion
- if ($isTitle) return $this->convertTitle($text);
+ if ( $isTitle ) {
+ $this->preConvertTitle( $text, $plang );
+ return $text;
+ }
- $plang = $this->getPreferredVariant();
+ $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
+ $text = '';
- $tarray = explode($this->mMarkup['begin'], $text);
- $tfirst = array_shift($tarray);
- if($this->mDoContentConvert)
- $text = $this->autoConvert($tfirst,$plang);
- else
- $text = $tfirst;
- foreach($tarray as $txt) {
- $marked = explode($this->mMarkup['end'], $txt, 2);
+ foreach ( $tarray as $txt ) {
- // strip the flags from syntax like -{T| ... }-
- list($rules,$flags) = $this->parseFlags($marked[0]);
+ $marked = explode( $this->mMarkup['begin'], $txt, 2 );
- $text .= $this->parseRules($rules,$flags,$plang);
+ if( $this->mDoContentConvert )
+ // Bug 19620: should convert a string immediately after a new rule added.
+ $text .= $this->autoConvert( $marked[0], $plang );
+ else
+ $text .= $marked[0];
- if(array_key_exists(1, $marked)){
- if( $this->mDoContentConvert )
- $text .= $this->autoConvert($marked[1],$plang);
- else
- $text .= $marked[1];
+ if ( array_key_exists( 1, $marked ) ) {
+ $crule = new ConverterRule($marked[1], $this);
+ $crule->parse( $plang );
+ $text .= $crule->getDisplay();
+ $this->applyManualConv( $crule );
}
+ else
+ $text .= $this->mMarkup['end'];
+
}
+ // Remove the last delimiter (wasn't real)
+ $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
return $text;
}
*
* @param string $link the name of the link
* @param mixed $nt the title object of the link
+ * @param boolean $ignoreOtherCond: to disable other conditions when
+ * we need to transclude a template or update a category's link
* @return null the input parameters may be modified upon return
- * @access public
+ * @public
*/
- function findVariantLink( &$link, &$nt ) {
- global $wgDisableLangConversion;
+ function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
+ # If the article has already existed, there is no need to
+ # check it again, otherwise it may cause a fault.
+ if ( is_object( $nt ) && $nt->exists() )
+ return;
+
+ global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
+ $isredir = $wgRequest->getText( 'redirect', 'yes' );
+ $action = $wgRequest->getText( 'action' );
+ $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
+ $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
$linkBatch = new LinkBatch();
$ns=NS_MAIN;
- if(is_object($nt))
+ if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
+ || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
+ return;
+
+ if ( is_object( $nt ) )
$ns = $nt->getNamespace();
$variants = $this->autoConvertToAllVariants($link);
foreach( $titles as $varnt ) {
if( $varnt->getArticleID() > 0 ) {
$nt = $varnt;
- if( !$wgDisableLangConversion )
- $link = $v;
+ $link = $varnt->getText();
break;
}
}
/**
* returns language specific hash options
*
- * @access public
+ * @public
*/
function getExtraHashOptions() {
$variant = $this->getPreferredVariant();
/**
* get title text as defined in the body of the article text
*
- * @access public
+ * @public
*/
function getParsedTitle() {
return $this->mTitleDisplay;
}
- /**
- * a write lock to the cache
- *
- * @private
- */
- function lockCache() {
- global $wgMemc;
- $success = false;
- for($i=0; $i<30; $i++) {
- if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
- break;
- sleep(1);
- }
- return $success;
- }
-
- /**
- * unlock cache
- *
- * @private
- */
- function unlockCache() {
- global $wgMemc;
- $wgMemc->delete($this->mCacheKey . "lock");
- }
-
-
/**
* Load default conversion tables
* This method must be implemented in derived class
$this->postLoadTables();
$this->mTables[self::CACHE_VERSION_KEY] = true;
-
- if($this->lockCache()) {
- $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
- $this->unlockCache();
- }
+
+ $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
wfProfileOut( __METHOD__.'-recache' );
}
wfProfileOut( __METHOD__ );
* @public
*/
function armourMath($text){
+ // we need to convert '-{' and '}-' to '-{' and '}-'
+ // to avoid a unwanted '}-' appeared after the math-image.
+ $text = strtr( $text, array('-{' => '-{', '}-' => '}-') );
$ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
return $ret;
}
}
+
+/**
+ * parser for rules of language conversion , parse rules in -{ }- tag
+ * @ingroup Language
+ * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
+ */
+class ConverterRule {
+ var $mText; // original text in -{text}-
+ var $mConverter; // LanguageConverter object
+ var $mManualCodeError = '<strong class="error">code error!</strong>';
+ var $mRuleDisplay = '';
+ var $mRuleTitle = false;
+ var $mRules = '';// string : the text of the rules
+ var $mRulesAction = 'none';
+ var $mFlags = array();
+ var $mConvTable = array();
+ var $mBidtable = array();// array of the translation in each variant
+ var $mUnidtable = array();// array of the translation in each variant
+
+ /**
+ * Constructor
+ *
+ * @param string $text the text between -{ and }-
+ * @param object $converter a LanguageConverter object
+ * @access public
+ */
+ function __construct( $text, $converter ){
+ $this->mText = $text;
+ $this->mConverter = $converter;
+ foreach( $converter->mVariants as $v ){
+ $this->mConvTable[$v] = array();
+ }
+ }
+
+ /**
+ * check if variants array in convert array
+ *
+ * @param string $variant Variant language code
+ * @return string Translated text
+ * @public
+ */
+ function getTextInBidtable( $variants ){
+ if( is_string( $variants ) ){ $variants = array( $variants ); }
+ if( !is_array( $variants ) ) return false;
+ foreach( $variants as $variant ){
+ if( array_key_exists( $variant, $this->mBidtable ) ){
+ return $this->mBidtable[$variant];
+ }
+ }
+ return false;
+ }
+
+ /**
+ * Parse flags with syntax -{FLAG| ... }-
+ * @private
+ */
+ function parseFlags(){
+ $text = $this->mText;
+ if(strlen($text) < 2 ) {
+ $this->mFlags = array( 'R' );
+ $this->mRules = $text;
+ return;
+ }
+
+ $flags = array();
+ $markup = $this->mConverter->mMarkup;
+ $validFlags = $this->mConverter->mFlags;
+ $variants = $this->mConverter->mVariants;
+
+ $tt = explode($markup['flagsep'], $text, 2);
+ if(count($tt) == 2) {
+ $f = explode($markup['varsep'], $tt[0]);
+ foreach($f as $ff) {
+ $ff = trim($ff);
+ if(array_key_exists($ff, $validFlags) &&
+ !in_array($validFlags[$ff], $flags))
+ $flags[] = $validFlags[$ff];
+ }
+ $rules = $tt[1];
+ } else {
+ $rules = $text;
+ }
+
+ //check flags
+ if( in_array('R',$flags) ){
+ $flags = array('R');// remove other flags
+ } elseif ( in_array('N',$flags) ){
+ $flags = array('N');// remove other flags
+ } elseif ( in_array('-',$flags) ){
+ $flags = array('-');// remove other flags
+ } elseif (count($flags)==1 && $flags[0]=='T'){
+ $flags[]='H';
+ } elseif ( in_array('H',$flags) ){
+ // replace A flag, and remove other flags except T
+ $temp=array('+','H');
+ if(in_array('T',$flags)) $temp[] = 'T';
+ if(in_array('D',$flags)) $temp[] = 'D';
+ $flags = $temp;
+ } else {
+ if ( in_array('A',$flags) ) {
+ $flags[]='+';
+ $flags[]='S';
+ }
+ if ( in_array('D',$flags) )
+ $flags=array_diff($flags,array('S'));
+ $flags_temp = array();
+ foreach ($variants as $variant) {
+ // try to find flags like "zh-hans", "zh-hant"
+ // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
+ if ( in_array($variant, $flags) )
+ $flags_temp[] = $variant;
+ }
+ if ( count($flags_temp) !== 0 )
+ $flags = $flags_temp;
+ }
+ if ( count($flags) == 0 )
+ $flags = array('S');
+ $this->mRules=$rules;
+ $this->mFlags=$flags;
+ }
+
+ /**
+ * generate conversion table
+ * @private
+ */
+ function parseRules() {
+ $rules = $this->mRules;
+ $flags = $this->mFlags;
+ $bidtable = array();
+ $unidtable = array();
+ $markup = $this->mConverter->mMarkup;
+ $variants = $this->mConverter->mVariants;
+
+ // varsep_pattern for preg_split:
+ // text should be splited by ";" only if a valid variant
+ // name exist after the markup, for example:
+ // -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
+ // we should split it as:
+ // array(
+ // [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
+ // [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
+ // [2] => ''
+ // )
+ $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
+ foreach( $variants as $variant ) {
+ $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
+ $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
+ . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
+ }
+ $varsep_pattern .= '\s*$)/';
+
+ $choice = preg_split($varsep_pattern, $rules);
+
+ foreach( $choice as $c ) {
+ $v = explode($markup['codesep'], $c, 2);
+ if( count($v) != 2 )
+ continue;// syntax error, skip
+ $to = trim($v[1]);
+ $v = trim($v[0]);
+ $u = explode($markup['unidsep'], $v, 2);
+ // if $to is empty, strtr() could return a wrong result
+ if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
+ $bidtable[$v] = $to;
+ } else if(count($u) == 2){
+ $from = trim($u[0]);
+ $v = trim($u[1]);
+ if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
+ && $to && in_array( $v, $variants ) )
+ $unidtable[$v] = array( $from=>$to );
+ elseif ( $to && in_array( $v, $variants ) )
+ $unidtable[$v][$from] = $to;
+ }
+ // syntax error, pass
+ if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
+ $bidtable = array();
+ $unidtable = array();
+ break;
+ }
+ }
+ $this->mBidtable = $bidtable;
+ $this->mUnidtable = $unidtable;
+ }
+
+ /**
+ * @private
+ */
+ function getRulesDesc(){
+ $codesep = $this->mConverter->mDescCodeSep;
+ $varsep = $this->mConverter->mDescVarSep;
+ $text='';
+ foreach($this->mBidtable as $k => $v)
+ $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
+ foreach($this->mUnidtable as $k => $a)
+ foreach($a as $from=>$to)
+ $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
+ return $text;
+ }
+
+ /**
+ * Parse rules conversion
+ * @private
+ */
+ function getRuleConvertedStr( $variant, $doConvert ){
+ $bidtable = $this->mBidtable;
+ $unidtable = $this->mUnidtable;
+
+ if( count($bidtable) + count($unidtable) == 0 ){
+ return $this->mRules;
+ } elseif ( $doConvert ){// the text converted
+ // display current variant in bidirectional array
+ $disp = $this->getTextInBidtable($variant);
+ // or display current variant in fallbacks
+ if(!$disp)
+ $disp = $this->getTextInBidtable(
+ $this->mConverter->getVariantFallbacks($variant));
+ // or display current variant in unidirectional array
+ if(!$disp && array_key_exists($variant,$unidtable)){
+ $disp = array_values($unidtable[$variant]);
+ $disp = $disp[0];
+ }
+ // or display frist text under disable manual convert
+ if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
+ if(count($bidtable)>0){
+ $disp = array_values($bidtable);
+ $disp = $disp[0];
+ } else {
+ $disp = array_values($unidtable);
+ $disp = array_values($disp[0]);
+ $disp = $disp[0];
+ }
+ }
+ return $disp;
+ } else {// no convert
+ return $this->mRules;
+ }
+ }
+
+ /**
+ * generate conversion table for all text
+ * @private
+ */
+ function generateConvTable(){
+ $flags = $this->mFlags;
+ $bidtable = $this->mBidtable;
+ $unidtable = $this->mUnidtable;
+ $manLevel = $this->mConverter->mManualLevel;
+
+ $vmarked=array();
+ foreach($this->mConverter->mVariants as $v) {
+ /* for bidirectional array
+ fill in the missing variants, if any,
+ with fallbacks */
+ if(!array_key_exists($v, $bidtable)) {
+ $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
+ $vf = $this->getTextInBidtable($variantFallbacks);
+ if($vf) $bidtable[$v] = $vf;
+ }
+
+ if(array_key_exists($v,$bidtable)){
+ foreach($vmarked as $vo){
+ // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
+ // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
+ // to introduce a custom mapping between
+ // words WordZh and WordTw in the whole text
+ if($manLevel[$v]=='bidirectional'){
+ $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
+ }
+ if($manLevel[$vo]=='bidirectional'){
+ $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
+ }
+ }
+ $vmarked[]=$v;
+ }
+ /*for unidirectional array
+ fill to convert tables */
+ $allow_unid = $manLevel[$v]=='bidirectional'
+ || $manLevel[$v]=='unidirectional';
+ if( $allow_unid && array_key_exists( $v, $unidtable ) ){
+ $ct = $this->mConvTable[$v];
+ $this->mConvTable[$v] = array_merge($ct, $unidtable[$v]);
+ }
+ }
+ }
+
+ /**
+ * Parse rules and flags
+ * @public
+ */
+ function parse($variant){
+ if(!$variant)
+ $variant = $this->mConverter->getPreferredVariant();
+
+ $variants = $this->mConverter->mVariants;
+ $this->parseFlags();
+ $flags = $this->mFlags;
+
+ // convert to specified variant
+ // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
+ if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
+ if ( in_array( $variant, $flags ) ) // check if current variant in flags
+ // then convert <text to convert> to current language
+ $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
+ else { // if current variant no in flags,
+ // then we check its fallback variants.
+ $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
+ foreach ( $variantFallbacks as $variantFallback ) {
+ // if current variant's fallback exist in flags
+ if ( in_array( $variantFallback, $flags ) ) {
+ // then convert <text to convert> to fallback language
+ $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
+ break;
+ }
+ }
+ }
+ $this->mFlags = $flags = array('R');
+ }
+
+ if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
+ // decode => HTML entities modified by Sanitizer::removeHTMLtags
+ $this->mRules = str_replace('=>','=>',$this->mRules);
+
+ $this->parseRules();
+ }
+ $rules = $this->mRules;
+
+ if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
+ if(in_array('+',$flags) || in_array('-',$flags))
+ // fill all variants if text in -{A/H/-|text} without rules
+ foreach($this->mConverter->mVariants as $v)
+ $this->mBidtable[$v] = $rules;
+ elseif (!in_array('N',$flags) && !in_array('T',$flags) )
+ $this->mFlags = $flags = array('R');
+ }
+
+ if( in_array('R',$flags) ) {
+ // if we don't do content convert, still strip the -{}- tags
+ $this->mRuleDisplay = $rules;
+ } elseif ( in_array('N',$flags) ){
+ // proces N flag: output current variant name
+ $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
+ } elseif ( in_array('D',$flags) ){
+ // proces D flag: output rules description
+ $this->mRuleDisplay = $this->getRulesDesc();
+ } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
+ // proces H,- flag or T only: output nothing
+ $this->mRuleDisplay = '';
+ } elseif ( in_array('S',$flags) ){
+ $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
+ $this->mConverter->mDoContentConvert);
+ } else {
+ $this->mRuleDisplay= $this->mManualCodeError;
+ }
+ // proces T flag
+ if ( in_array('T',$flags) ) {
+ $this->mRuleTitle = $this->getRuleConvertedStr($variant,
+ $this->mConverter->mDoTitleConvert);
+ }
+
+ if (in_array('-', $flags))
+ $this->mRulesAction='remove';
+ if (in_array('+', $flags))
+ $this->mRulesAction='add';
+
+ $this->generateConvTable();
+ }
+
+ /**
+ * @public
+ */
+ function hasRules(){
+ // TODO:
+ }
+
+ /**
+ * get display text on markup -{...}-
+ * @public
+ */
+ function getDisplay(){
+ return $this->mRuleDisplay;
+ }
+ /**
+ * get converted title
+ * @public
+ */
+ function getTitle(){
+ return $this->mRuleTitle;
+ }
+
+ /**
+ * return how deal with conversion rules
+ * @public
+ */
+ function getRulesAction(){
+ return $this->mRulesAction;
+ }
+
+ /**
+ * get conversion table ( bidirectional and unidirectional conversion table )
+ * @public
+ */
+ function getConvTable(){
+ return $this->mConvTable;
+ }
+
+ /**
+ * get conversion rules string
+ * @public
+ */
+ function getRules(){
+ return $this->mRules;
+ }
+
+ /**
+ * get conversion flags
+ * @public
+ */
+ function getFlags(){
+ return $this->mFlags;
+ }
+}