Revert r51128, and its follow-up revisions r51129, r51130, r51390.
[lhc/web/wiklou.git] / languages / LanguageConverter.php
index 207742f..57af7bc 100644 (file)
@@ -176,25 +176,39 @@ class LanguageConverter {
                        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( 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;
+                                       }
+                               }
+
+                               foreach( $languages as $language ) {
+                                       // strip whitespace
+                                       $language = trim( $language );
+                                       if( in_array( $language, $this->mVariants ) ) {
+                                               return $language;
+                                               break;
+                                       }
                                }
                        }
-                       // don't try to return bad variant
-                       if(in_array( $pv, $this->mVariants ))
-                               return $pv;
                }
 
                return $this->mMainLanguageCode;
@@ -866,6 +880,9 @@ class LanguageConverter {
         * @public
         */
        function armourMath($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;
        }
@@ -976,10 +993,12 @@ 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 )
@@ -998,23 +1017,45 @@ class ConverterRule {
                $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);
 
-               $choice = explode($markup['varsep'], $rules);
-               foreach($choice as $c) {
+               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'], $v);
-                       if( count($u) == 1 ) {
+                       // 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] ) )
+                               if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
+                                       && $to && in_array( $v, $variants ) )
                                        $unidtable[$v] = array( $from=>$to );
-                               else
+                               elseif ( $to && in_array( $v, $variants ) )
                                        $unidtable[$v][$from] = $to;
                        }
                        // syntax error, pass
@@ -1141,14 +1182,19 @@ 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;
                                        }
@@ -1158,10 +1204,6 @@ class ConverterRule {
                }
 
                if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
-                       //FIXME: may cause trouble here...
-                       //strip &nbsp; since it interferes with the parsing, plus,
-                       //all spaces should be stripped in this tag anyway.
-                       $this->mRules = str_replace('&nbsp;', '', $this->mRules);
                        // decode => HTML entities modified by Sanitizer::removeHTMLtags 
                        $this->mRules = str_replace('=&gt;','=>',$this->mRules);