Localization update for he.
[lhc/web/wiklou.git] / languages / LanguageConverter.php
index 86b9da9..b54fcbc 100644 (file)
@@ -21,10 +21,7 @@ class LanguageConverter {
        var $mVariants, $mVariantFallbacks, $mVariantNames;
        var $mTablesLoaded = false;
        var $mTables;
-       var $mManualAddTables;
-       var $mManualRemoveTables;
        var $mNamespaceTables;
-       var $mTitleDisplay='';
        var $mDoTitleConvert=true, $mDoContentConvert=true;
        var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
        var $mTitleFromFlag = false;
@@ -34,6 +31,8 @@ class LanguageConverter {
        var $mFlags;
        var $mDescCodeSep = ':',$mDescVarSep = ';';
        var $mUcfirst = false;
+       var $mTitleOriginal = '';
+       var $mTitleDisplay = '';
 
        const CACHE_VERSION_KEY = 'VERSION 6';
 
@@ -48,7 +47,7 @@ class LanguageConverter {
         * @param array $manualLevel limit for supported variants
         * @public
         */
-       function __construct($langobj, $maincode,
+       function __construct( $langobj, $maincode,
                                                                $variants=array(),
                                                                $variantfallbacks=array(),
                                                                $markup=array(),
@@ -56,7 +55,13 @@ class LanguageConverter {
                                                                $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;
@@ -88,8 +93,6 @@ class LanguageConverter {
                        $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
                                                                ?$manualLevel[$v]
                                                                :'bidirectional';
-                       $this->mManualAddTables[$v] = array();
-                       $this->mManualRemoveTables[$v] = array();
                        $this->mNamespaceTables[$v] = array();
                        $this->mFlags[$v] = $v;
                }
@@ -123,13 +126,15 @@ class LanguageConverter {
        /**
         * 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
         * @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
@@ -171,34 +176,63 @@ class LanguageConverter {
                }
 
                // 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) {
-                                       $ary = split("[,;]",$zh);
-                                       $pv = $ary[0];
+                       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;
-
        }
        
        /**
@@ -273,13 +307,26 @@ class LanguageConverter {
                
                // 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) {
                        $mark = substr($text, $mstart, $m[1]-$mstart);
                        $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
-                       $ret .= $mark;
-                       $ret .= $this->translate($m[0], $toVariant);
+                       // 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;
        }
@@ -295,9 +342,13 @@ class LanguageConverter {
         */
        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;
        }
@@ -360,7 +411,7 @@ class LanguageConverter {
         * prepare manual conversion table
         * @private
         */
-       function prepareManualConv( $convRule ){
+       function applyManualConv( $convRule ){
                // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
                $title = $convRule->getTitle();
                if( $title ){
@@ -371,38 +422,20 @@ class LanguageConverter {
                //apply manual conversion table to global table
                $convTable = $convRule->getConvTable();
                $action = $convRule->getRulesAction();
-               foreach( $convTable as $v => $t ) {
-                       if( !in_array( $v, $this->mVariants ) )continue;
-                       if( $action=="add" ) {
-                               foreach( $t as $from => $to ) {
+               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.
-                                               $this->mManualAddTables[$v][$from] = $to;
+                                               $this->mTables[$variant]->setPair( $from, $to );
                                }
                        }
-                       elseif ( $action == "remove" ) {
-                               foreach ( $t as $from=>$to ) {
-                                       if ( $from || $to )
-                                               $this->mManualRemoveTables[$v][$from] = $to;
-                               }
-                       }
-               }
-       }
-
-       /**
-        * apply manual conversion from $this->mManualAddTables and $this->mManualRemoveTables
-        * @private
-        */
-       function applyManualConv(){
-               //apply manual conversion table to global table
-               foreach($this->mVariants as $v) {
-                       if (count($this->mManualAddTables[$v]) > 0) {
-                               $this->mTables[$v]->mergeArray($this->mManualAddTables[$v]);
+                       elseif ( $action == 'remove' ) {
+                               $this->mTables[$variant]->removeArray( $pair );
                        }
-                       if (count($this->mManualRemoveTables[$v]) > 0)
-                               $this->mTables[$v]->removeArray($this->mManualRemoveTables[$v]);
                }
        }
 
@@ -424,8 +457,9 @@ class LanguageConverter {
 
                $text = $this->convert( $text );
 
-               if ( $this->mTitleFromFlag )
-                       $parser->mOutput->setTitleText( $this->mTitleDisplay );
+               $this->convertTitle();
+               $parser->mOutput->setTitleText( $this->mTitleDisplay );
+
                return $text;
        }
        
@@ -446,34 +480,36 @@ class LanguageConverter {
        }
 
        /**
-        * convert title
+        * Pre convert title. Store the original title $this->mTitleOrginal;
+        * store the default converted title to $this->mTitleDisplay.
         * @private
         */
-       function convertTitle( $text, $variant ){
-               global $wgDisableTitleConversion, $wgUser;
-
-               // check for global param and __NOTC__ tag
-               if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
-                       $this->mTitleDisplay = $text;
-                       return $text;
-               }
-
-               // use the title from the T flag if any
-               if( $this->mTitleFromFlag ){
-                       $this->mTitleFromFlag = false;
-                       return $this->mTitleDisplay;
-               }
+       function preConvertTitle( $text, $variant ){
+               $this->mTitleOriginal = $text;
+               
+               $text = $this->convertNamespace( $text, $variant );
+               $this->mTitleDisplay = $this->convert( $text );
+       }
 
-               global $wgRequest;
+       /**
+        * convert title
+        * @private
+        */
+       function convertTitle(){
+               global $wgDisableTitleConversion, $wgUser, $wgRequest;
                $isredir = $wgRequest->getText( 'redirect', 'yes' );
                $action = $wgRequest->getText( 'action' );
                $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
-               if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
-                       return $text;
-               } else {
-                       $text = $this->convertNamespace( $text, $variant );
-                       $this->mTitleDisplay = $this->convert( $text );
-                       return $this->mTitleDisplay;
+
+               // 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;
                }
        }
 
@@ -511,36 +547,37 @@ class LanguageConverter {
                $plang = $this->getPreferredVariant();
 
                // for title convertion
-               if ( $isTitle ) return $this->convertTitle( $text, $plang );
+               if ( $isTitle ) {
+                       $this->preConvertTitle( $text, $plang );
+                       return $text;
+               }
 
                $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
                $text = '';
 
-               $marks = array();
                foreach ( $tarray as $txt ) {
+
                        $marked = explode( $this->mMarkup['begin'], $txt, 2 );
+
+                       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 ) ) {
                                $crule = new ConverterRule($marked[1], $this);
                                $crule->parse( $plang );
-                               $marked[1] = $crule->getDisplay();
-                               $this->prepareManualConv( $crule );
+                               $text .= $crule->getDisplay();
+                               $this->applyManualConv( $crule );
                        }
                        else
-                               $marked[0] .= $this->mMarkup['end'];
-                       array_push( $marks, $marked );
-               }
-               $this->applyManualConv();
-               foreach ( $marks as $marked ) {
-                       if( $this->mDoContentConvert )
-                               $text .= $this->autoConvert( $marked[0], $plang );
-                       else
-                               $text .= $marked[0];
-                       if( array_key_exists( 1, $marked ) )
-                               $text .= $marked[1];
+                               $text .= $this->mMarkup['end'];
+
                }
+
                // Remove the last delimiter (wasn't real)
                $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
-
                return $text;
        }
 
@@ -558,6 +595,11 @@ class LanguageConverter {
         * @public
         */
        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' );
@@ -571,7 +613,7 @@ class LanguageConverter {
                        || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
                        return;
 
-               if(is_object($nt))
+               if ( is_object( $nt ) )
                        $ns = $nt->getNamespace();
 
                $variants = $this->autoConvertToAllVariants($link);
@@ -621,33 +663,6 @@ class LanguageConverter {
                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
@@ -688,11 +703,8 @@ class LanguageConverter {
 
                        $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, $this->mTables, 43200);
                        wfProfileOut( __METHOD__.'-recache' );
                }
                wfProfileOut( __METHOD__ );
@@ -866,6 +878,9 @@ class LanguageConverter {
         * @public
         */
        function armourMath($text){ 
+               // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
+               // to avoid a unwanted '}-' appeared after the math-image.
+               $text = strtr( $text, array('-{' => '-&#123;', '}-' => '&#125;-') );
                $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
                return $ret;
        }
@@ -879,8 +894,9 @@ class LanguageConverter {
 class ConverterRule {
        var $mText; // original text in -{text}-
        var $mConverter; // LanguageConverter object 
-       var $mManualCodeError='<strong class="error">code error!</strong>';
-       var $mRuleDisplay = '',$mRuleTitle=false;
+       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();
@@ -895,11 +911,11 @@ class ConverterRule {
         * @param object $converter a  LanguageConverter object 
         * @access public
         */
-       function __construct($text,$converter){
+       function __construct( $text, $converter ){
                $this->mText = $text;
-               $this->mConverter=$converter;
-               foreach($converter->mVariants as $v){
-                       $this->mConvTable[$v]=array();
+               $this->mConverter = $converter;
+               foreach( $converter->mVariants as $v ){
+                       $this->mConvTable[$v] = array();
                }
        }
 
@@ -910,11 +926,11 @@ class ConverterRule {
         * @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)){
+       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];
                        }
                }
@@ -976,10 +992,12 @@ class ConverterRule {
                                $flags=array_diff($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 )
+                       if ( count($flags_temp) !== 0 )
                                $flags = $flags_temp;
                }
                if ( count($flags) == 0 )
@@ -1000,19 +1018,34 @@ class ConverterRule {
                $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'] . '|';
+               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);
+                       $u  = explode($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){
@@ -1054,13 +1087,13 @@ class ConverterRule {
         *  Parse rules conversion
         * @private
         */
-       function getRuleConvertedStr($variant,$doConvert){
+       function getRuleConvertedStr( $variant, $doConvert ){
                $bidtable = $this->mBidtable;
                $unidtable = $this->mUnidtable;
 
                if( count($bidtable) + count($unidtable) == 0 ){
                        return $this->mRules;
-               } elseif ($doConvert){// the text converted 
+               } elseif ( $doConvert ){// the text converted 
                        // display current variant in bidirectional array
                        $disp = $this->getTextInBidtable($variant);
                        // or display current variant in fallbacks
@@ -1129,9 +1162,9 @@ class ConverterRule {
                                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]);
+                       if( $allow_unid && array_key_exists( $v, $unidtable ) ){
+                               $ct = $this->mConvTable[$v];
+                               $this->mConvTable[$v] = array_merge($ct, $unidtable[$v]);
                        }
                }
        }
@@ -1148,14 +1181,19 @@ class ConverterRule {
                $this->parseFlags();
                $flags = $this->mFlags;
 
-               //convert to specified variant
+               // 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 ) )
+                       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 {
+                       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;
                                        }