var $mTables;
var $mManualAddTables;
var $mManualRemoveTables;
+ var $mNamespaceTables;
var $mTitleDisplay='';
var $mDoTitleConvert=true, $mDoContentConvert=true;
var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
var $mFlags;
var $mDescCodeSep = ':',$mDescVarSep = ';';
var $mUcfirst = false;
+ var $mGroups = array();
const CACHE_VERSION_KEY = 'VERSION 6';
:'bidirectional';
$this->mManualAddTables[$v] = array();
$this->mManualRemoveTables[$v] = array();
+ $this->mNamespaceTables[$v] = array();
+ $this->mFlags[$v] = $v;
}
}
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;
* to convert text in "title" or "alt", like '<img alt="text" ... '
* or '<span title="text" ... '
*
- * @param string $title the "title" or "alt" text
- * @param string $text the text to be converted
- * @param string $toVariant the target language code
* @return string like ' alt="yyyy"' or ' title="yyyy"'
+ * @private
*/
- protected function captionConvert( $title, $text, $toVariant ) {
+ 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);
$mstart = $m[1]+strlen($m[0]);
// enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
- $captionpattern = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/e';
- $replacement = "\$this->captionConvert('\\1', '\\2', \$toVariant)";
+ $captionpattern = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
foreach($matches as $m) {
$mark = substr($text, $mstart, $m[1]-$mstart);
- $mark = preg_replace($captionpattern, $replacement, $mark);
+ $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
$ret .= $mark;
$ret .= $this->translate($m[0], $toVariant);
$mstart = $m[1] + strlen($m[0]);
* prepare manual conversion table
* @private
*/
- function prepareManualConv($convRule){
+ function prepareManualConv( $convRule ){
// use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
$title = $convRule->getTitle();
- if($title){
+ if( $title ){
$this->mTitleFromFlag = true;
$this->mTitleDisplay = $title;
}
//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;
+ foreach( $convTable as $v => $t ) {
+ if( !in_array( $v, $this->mVariants ) )continue;
if( $action=="add" ) {
- foreach($t as $from=>$to) {
- // more efficient than array_merge(), about 2.5 times.
- $this->mManualAddTables[$v][$from] = $to;
+ foreach( $t 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;
+ }
+ }
+ elseif ( $action == "remove" ) {
+ foreach ( $t as $from=>$to ) {
+ if ( $from || $to )
+ $this->mManualRemoveTables[$v][$from] = $to;
}
}
- elseif ( $action=="remove" )
- $this->mManualRemoveTables[$v] = array_merge($this->mManualRemoveTables[$v], $t);
}
}
* @public
*/
function parserConvert( $text, &$parser ) {
- global $wgDisableLangConversion;
+ global $wgUser, $wgEnableVariants;
+
+ /* disable entire conversion engine if the user had been disabled it */
+ $variantConv = $wgUser->getOption( 'variantconversion' );
+ if ( ( $wgEnableVariants == true ) && ( $variantConv == 0 ) ) {
+ $wgEnableVariants = false;
+ }
+
/* don't do anything if this is the conversion table */
if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
- strpos($parser->mTitle->getText(), "Conversiontable") !== false )
+ strpos($parser->mTitle->getText(), 'onversiontable') !== false )
{
return $text;
}
- if($wgDisableLangConversion)
+ if ( !$wgEnableVariants )
return $text;
$text = $this->convert( $text );
- $parser->mOutput->setTitleText( $this->mTitleDisplay );
+
+ if ( $this->mTitleFromFlag )
+ $parser->mOutput->setTitleText( $this->mTitleDisplay );
return $text;
}
+
+ /**
+ * convert namespace
+ * @param string $title the title included namespace
+ * @return array of string
+ * @private
+ */
+ 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 $ret;
+ }
/**
- * convert title
+ * convert title
* @private
*/
- function convertTitle($text){
+ function convertTitle( $text, $variant ){
global $wgDisableTitleConversion, $wgUser;
// check for global param and __NOTC__ tag
}
// use the title from the T flag if any
- if($this->mTitleFromFlag){
+ if( $this->mTitleFromFlag ){
$this->mTitleFromFlag = false;
return $this->mTitleDisplay;
}
if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
return $text;
} else {
- $this->mTitleDisplay = $this->convert($text);
+ $text = $this->convertNamespace( $text, $variant );
+ $this->mTitleDisplay = $this->convert( $text );
return $this->mTitleDisplay;
}
}
* @return string converted text
* @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 ) return $this->convertTitle( $text, $plang );
- $plang = $this->getPreferredVariant();
- $tarray = StringUtils::explode($this->mMarkup['end'], $text);
+ $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
$text = '';
$marks = array();
- foreach($tarray as $txt) {
- $marked = explode($this->mMarkup['begin'], $txt, 2);
- if (array_key_exists(1, $marked)) {
+ foreach ( $tarray as $txt ) {
+ $marked = explode( $this->mMarkup['begin'], $txt, 2 );
+ if ( array_key_exists( 1, $marked ) ) {
$crule = new ConverterRule($marked[1], $this);
- $crule->parse($plang);
+ $crule->parse( $plang );
$marked[1] = $crule->getDisplay();
- $this->prepareManualConv($crule);
+ $this->prepareManualConv( $crule );
}
else
$marked[0] .= $this->mMarkup['end'];
- array_push($marks, $marked);
+ array_push( $marks, $marked );
}
$this->applyManualConv();
- foreach ($marks as $marked) {
+ foreach ( $marks as $marked ) {
if( $this->mDoContentConvert )
- $text .= $this->autoConvert($marked[0],$plang);
+ $text .= $this->autoConvert( $marked[0], $plang );
else
$text .= $marked[0];
- if( array_key_exists(1, $marked) )
+ if( array_key_exists( 1, $marked ) )
$text .= $marked[1];
}
// Remove the last delimiter (wasn't real)
*
* @param string $link the name of the link
* @param mixed $nt the title object of the link
+ * @param boolean $ignoreOtherCond: to disable other conditions when
+ * we need to transclude a template or update a category's link
* @return null the input parameters may be modified upon return
* @public
*/
- function findVariantLink( &$link, &$nt, $forTemplate = false ) {
- global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
+ function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
+ global $wgEnableVariants, $wgDisableTitleConversion, $wgRequest, $wgUser;
$isredir = $wgRequest->getText( 'redirect', 'yes' );
$action = $wgRequest->getText( 'action' );
$linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
- $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
+ $disableLinkConversion = !$wgEnableVariants || $wgDisableTitleConversion;
$linkBatch = new LinkBatch();
$ns=NS_MAIN;
- if ( $disableLinkConversion || ( !$forTemplate && ( $isredir == 'no' || $action == 'edit'
- || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) ) {
+ if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
+ || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
return;
- }
if(is_object($nt))
$ns = $nt->getNamespace();
foreach( $titles as $varnt ) {
if( $varnt->getArticleID() > 0 ) {
$nt = $varnt;
- $link = $v;
+ $link = $varnt->getText();
break;
}
}
* load conversion tables either from the cache or the disk
* @private
*/
- function loadTables($fromcache=true) {
+ function loadTables( $fromcache = true ) {
global $wgMemc;
if( $this->mTablesLoaded )
return;
// we will first load the default tables
// then update them using things in MediaWiki:Zhconversiontable/*
$this->loadDefaultTables();
- foreach($this->mVariants as $var) {
- $cached = $this->parseCachedTable($var);
- $this->mTables[$var]->mergeArray($cached);
+ foreach( $this->mVariants as $var ) {
+ $cached = $this->parseCachedTable( $var );
+ // load group convert table, e.g. [[MediaWiki:Groupconversiontable-StarTrek]].
+ foreach( $this->mGroups as $group ) {
+ $cachedgroup = $this->parseCachedTable( $var, '', true, "Groupconversiontable-$group" );
+ $cached = array_merge( $cached, $cachedgroup );
+ }
+ $this->mTables[$var]->mergeArray( $cached );
}
$this->postLoadTables();
* and will be parsed recursively if $recursive=true
*
*/
- function parseCachedTable($code, $subpage='', $recursive=true) {
+ function parseCachedTable($code, $subpage='', $recursive=true, $table='Conversiontable') {
global $wgMessageCache;
static $parsed = array();
if(!is_object($wgMessageCache))
return array();
- $key = 'Conversiontable/'.$code;
+ $key = "$table/".$code;
if($subpage)
$key .= '/' . $subpage;
if(array_key_exists($key, $parsed))
return array();
-
+
if ( strpos( $code, '/' ) === false ) {
- $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
+ $txt = $wgMessageCache->get( $table, true, $code );
} else {
- $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
+ $title = Title::makeTitleSafe( NS_MEDIAWIKI, "$table/$code" );
if ( $title && $title->exists() ) {
$article = new Article( $title );
$txt = $article->getContents();
$txt = '';
}
}
-
// get all subpage links of the form
// [[MediaWiki:conversiontable/zh-xx/...|...]]
- $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
+ $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ":$table";
$subs = explode('[[', $txt);
$sublinks = array();
foreach( $subs as $sub ) {
}
}
-
// parse the mappings in this page
$blocks = explode($this->mMarkup['begin'], $txt);
array_shift($blocks);
* @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;
}
+
+ /**
+ * Callback function for magicword 'groupconvert'
+ *
+ * @param string $group: the group name called for
+ * @return blank string
+ */
+ function groupConvert( $group ) {
+ $this->mGroups[] = $group;
+ return '';
+ }
}
/**
* parser for rules of language conversion , parse rules in -{ }- tag
* @ingroup Language
- * @author fdcn <fdcn64@gmail.com>
+ * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
*/
class ConverterRule {
var $mText; // original text in -{text}-
$flags = array();
$markup = $this->mConverter->mMarkup;
$validFlags = $this->mConverter->mFlags;
+ $variants = $this->mConverter->mVariants;
$tt = explode($markup['flagsep'], $text, 2);
if(count($tt) == 2) {
if(in_array('D',$flags)) $temp[] = 'D';
$flags = $temp;
} else {
- if ( in_array('A',$flags)) {
+ 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 )
+ if ( count($flags) == 0 )
$flags = array('S');
$this->mRules=$rules;
$this->mFlags=$flags;
$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) {
- $v = explode($markup['codesep'], $c);
- if(count($v) != 2)
+ 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) {
+ $to = trim($v[1]);
+ $v = trim($v[0]);
+ $u = explode($markup['unidsep'], $v);
+ // 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]) )
- $unidtable[$v]=array($from=>$to);
- else
- $unidtable[$v][$from]=$to;
+ $from = trim($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)){
+ if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
$bidtable = array();
$unidtable = array();
break;
* @public
*/
function parse($variant){
- if(!$variant) $variant = $this->mConverter->getPreferredVariant();
+ if(!$variant)
+ $variant = $this->mConverter->getPreferredVariant();
+ $variants = $this->mConverter->mVariants;
$this->parseFlags();
$flags = $this->mFlags;
- if( !in_array('R',$flags) || !in_array('N',$flags) ){
- //FIXME: may cause trouble here...
- //strip since it interferes with the parsing, plus,
- //all spaces should be stripped in this tag anyway.
- $this->mRules = str_replace(' ', '', $this->mRules);
+ // 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);
}
$rules = $this->mRules;
- if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
+ 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)
dépôts / lhc / web / wiklou.git / blobdiff