From 69dbeb97f15b65dd5773852d76c3bc93caefc862 Mon Sep 17 00:00:00 2001 From: Shinjiman Date: Thu, 26 Jun 2008 03:00:34 +0000 Subject: [PATCH] * (bug 14604) Introduced the following features for the LanguageConverter: Multi-tag support, single conversion flag, remove conversion flag on a single page, description flag, variant name, multi-variant fallbacks. patch by fdcn * Added zh-mo and zh-my variants for the zh language --- RELEASE-NOTES | 5 +- includes/StringUtils.php | 11 + languages/LanguageConverter.php | 674 ++++++++++++++++---------- languages/Names.php | 6 +- languages/classes/LanguageZh.php | 78 ++- languages/messages/MessagesZh.php | 2 + languages/messages/MessagesZh_mo.php | 10 + languages/messages/MessagesZh_my.php | 10 + maintenance/language/messageTypes.inc | 2 + maintenance/language/messages.inc | 2 + 10 files changed, 530 insertions(+), 270 deletions(-) create mode 100644 languages/messages/MessagesZh_mo.php create mode 100644 languages/messages/MessagesZh_my.php diff --git a/RELEASE-NOTES b/RELEASE-NOTES index 73889b15f9..7fec797712 100644 --- a/RELEASE-NOTES +++ b/RELEASE-NOTES @@ -164,6 +164,8 @@ it from source control: http://www.mediawiki.org/wiki/Download_from_SVN * Add support for Hijri (Islamic) calendar * Add a new hook LinkerMakeExternalImage to allow extensions to modify the output of external (hotlinked) images. +* (bug 14604) Introduced the following features for the LanguageConverter: Multi-tag support, single conversion flag, remove conversion flag on a single page, description flag, variant name, multi-variant fallbacks. +* Add zh-mo and zh-my variants for the zh language === Bug fixes in 1.13 === @@ -484,7 +486,8 @@ regularly. Below only new and removed languages are listed. * Silesian (szl) (new) * Tajiki (Cyrllic script) (tg-cyrl) (new) * Tajiki (Latin script) (tg-latn) (new) - +* Chinese (Macau) (zh-mo) (new) +* Chinese (Malaysia) (zh-my) (new) == Compatibility == diff --git a/includes/StringUtils.php b/includes/StringUtils.php index 95c755d437..70d0bff1d3 100644 --- a/includes/StringUtils.php +++ b/includes/StringUtils.php @@ -283,6 +283,17 @@ class ReplacementArray { $this->fss = false; } + function removePair( $from ) { + unset($this->data[$from]); + $this->fss = false; + } + + function removeArray( $data ) { + foreach( $data as $from => $to ) + $this->removePair( $from ); + $this->fss = false; + } + function replace( $subject ) { if ( function_exists( 'fss_prep_replace' ) ) { wfProfileIn( __METHOD__.'-fss' ); diff --git a/languages/LanguageConverter.php b/languages/LanguageConverter.php index 57b5f0311c..5e47ded72b 100644 --- a/languages/LanguageConverter.php +++ b/languages/LanguageConverter.php @@ -1,19 +1,24 @@ * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License + * @maintainers fdcn , shinjiman */ class LanguageConverter { var $mPreferredVariant=''; var $mMainLanguageCode; - var $mVariants, $mVariantFallbacks; + var $mVariants, $mVariantFallbacks, $mVariantNames; var $mTablesLoaded = false; var $mTables; var $mTitleDisplay=''; var $mDoTitleConvert=true, $mDoContentConvert=true; + var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants + var $mManualCodeError='code error!'; var $mTitleFromFlag = false; var $mCacheKey; var $mLangObj; @@ -21,38 +26,63 @@ class LanguageConverter { var $mFlags; var $mUcfirst = false; - const CACHE_VERSION_KEY = 'VERSION 5'; + const CACHE_VERSION_KEY = 'VERSION 6'; /** - * Constructor + * Constructor * - * @param string $maincode the main language code of this language - * @param array $variants the supported variants of this language - * @param array $variantfallback the fallback language of each variant - * @param array $markup array defining the markup used for manual conversion + * @param string $maincode the main language code of this language + * @param array $variants the supported variants of this language + * @param array $variantfallback the fallback language of each variant + * @param array $markup array defining the markup used for manual conversion * @param array $flags array defining the custom strings that maps to the flags - * @access public - */ + * @access public + */ function __construct($langobj, $maincode, $variants=array(), $variantfallbacks=array(), $markup=array(), - $flags = array()) { + $flags = array(), + $manualLevel = array() ) { $this->mLangObj = $langobj; $this->mMainLanguageCode = $maincode; $this->mVariants = $variants; $this->mVariantFallbacks = $variantfallbacks; + global $wgLanguageNames; + $this->mVariantNames = $wgLanguageNames; $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode ); - $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':', - 'varsep'=>';', 'end'=>'}-'); + $m = array( + 'begin'=>'-{', + 'flagsep'=>'|', + 'unidsep'=>'=>', //for unidirectional conversion + 'codesep'=>':', + 'varsep'=>';', + 'end'=>'}-' + ); $this->mMarkup = array_merge($m, $markup); - $f = array('A'=>'A', 'T'=>'T', 'R' => 'R'); + $f = array( + // 'S' show converted text + // '+' add rules for alltext + // 'E' the gave flags is error + // these flags above are reserved for program + 'A'=>'A', // add rule for convert code (all text convert) + 'T'=>'T', // title convert + 'R'=>'R', // raw content + 'D'=>'D', // convert description (subclass implement) + '-'=>'-', // remove convert (not implement) + 'H'=>'H', // add rule for convert code (but no display in placed code ) + 'N'=>'N' // current variant name + ); $this->mFlags = array_merge($f, $flags); + foreach( $this->mVariants as $v) + $this->mManualLevel[$v]=array_key_exists($v,$manualLevel) + ?$manualLevel[$v] + :'bidirectional'; } /** - * @access public - */ + * @access public + */ function getVariants() { return $this->mVariants; } @@ -60,25 +90,47 @@ class LanguageConverter { /** * in case some variant is not defined in the markup, we need * to have some fallback. for example, in zh, normally people - * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk. - * when zh-sg is preferred but not defined, we will pick zh-cn + * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk. + * when zh-sg is preferred but not defined, we will pick zh-hans * in this case. right now this is only used by zh. * * @param string $v the language code of the variant - * @return string the code of the fallback language or false if there is no fallback - * @private - */ - function getVariantFallback($v) { - return $this->mVariantFallbacks[$v]; + * @return string array the code of the fallback language or false if there is no fallback + * @private + */ + function getVariantFallbacks($v) { + if( isset( $this->mVariantFallbacks[$v] ) ) { + return $this->mVariantFallbacks[$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 - * @return string the preferred language code - * @access public - */ + * @return string the preferred language code + * @access public + */ function getPreferredVariant( $fromUser = true ) { global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant; @@ -140,15 +192,15 @@ class LanguageConverter { return $this->mMainLanguageCode; } - + /** - * dictionary-based conversion - * - * @param string $text the text to be converted - * @param string $toVariant the target language code - * @return string the converted text - * @private - */ + * dictionary-based conversion + * + * @param string $text the text to be converted + * @param string $toVariant the target language code + * @return string the converted text + * @private + */ function autoConvert($text, $toVariant=false) { $fname="LanguageConverter::autoConvert"; @@ -182,7 +234,7 @@ class LanguageConverter { $scriptfix = '.*?<\/script>|'; $reg = '/'.$codefix . $scriptfix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s'; - + $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE); $m = array_shift($matches); @@ -216,12 +268,12 @@ class LanguageConverter { } /** - * convert text to all supported variants - * - * @param string $text the text to be converted - * @return array of string - * @public - */ + * convert text to all supported variants + * + * @param string $text the text to be converted + * @return array of string + * @public + */ function autoConvertToAllVariants($text) { $fname="LanguageConverter::autoConvertToAllVariants"; wfProfileIn( $fname ); @@ -238,12 +290,12 @@ class LanguageConverter { } /** - * convert link text to all supported variants - * - * @param string $text the text to be converted - * @return array of string - * @public - */ + * convert link text to all supported variants + * + * @param string $text the text to be converted + * @return array of string + * @public + */ function convertLinkToAllVariants($text) { if( !$this->mTablesLoaded ) $this->loadTables(); @@ -295,35 +347,269 @@ class LanguageConverter { * */ function parseFlags($marked){ - $flags = array(); + $flags = array(); + + // for multi-FLAGs + if(strlen($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 ) + $flags = array('S'); - // process flag only if the flag is valid - if(strlen($marked) < 2 || !(in_array($marked[0],$this->mFlags) && $marked[1]=='|' ) ) - return array($marked,array()); + return array($rules,$flags); + } + + 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.'Ë'.$this->mVariantNames[$k].':'.$to.';'; + return $text; + } - $tt = explode($this->mMarkup['flagsep'], $marked, 2); + /** + * parse the manually marked conversion rule + * @param string $rule the text of the rule + * @return array of the translation in each variant + * @private + */ + function getConvTableFromRules($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); + } - if(sizeof($tt) == 2) { - $f = explode($this->mMarkup['varsep'], $tt[0]); - foreach($f as $ff) { - $ff = trim($ff); - if(array_key_exists($ff, $this->mFlags) && - !array_key_exists($this->mFlags[$ff], $flags)) - $flags[] = $this->mFlags[$ff]; + /** + * 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 + * @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) ){ + // the text converted + if($doConvert){ + // 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]; } - $rules = $tt[1]; + // or display frist text under disable manual convert + if(!$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; } - else - $rules = $marked; + } elseif ( in_array('T',$flags) ) { + // proces T flag : output nothing + $disp = ''; + } + else + $disp= $this->mManualCodeError; - if( !in_array('R',$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); + return $disp; + } + + function applyManualFlag($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[$v]); + } + } + } - return array($rules,$flags); + /** + * Parse rules and flags + * @private + */ + function parseRules($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); + $this->applyManualFlag($flags,$bidtable,$unidtable); + + return $disp; + } + + function convertTitle($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; + $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; + } } /** @@ -334,134 +620,51 @@ class LanguageConverter { * * syntax of the markup: * -{code1:text1;code2:text2;...}- or + * -{flags|code1:text1;code2:text2;...}- or * -{text}- in which case no conversion should take place for text - * - * @param string $text text to be converted - * @param bool $isTitle whether this conversion is for the article title - * @return string converted text - * @access public - */ + * + * @param string $text text to be converted + * @param bool $isTitle whether this conversion is for the article title + * @return string converted text + * @access public + */ function convert( $text , $isTitle=false) { - $mw =& MagicWord::get( 'notitleconvert' ); + + $mw =& MagicWord::get( 'notitleconvert' ); if( $mw->matchAndRemove( $text ) ) $this->mDoTitleConvert = false; - - $mw =& MagicWord::get( 'nocontentconvert' ); + $mw =& MagicWord::get( 'nocontentconvert' ); if( $mw->matchAndRemove( $text ) ) { $this->mDoContentConvert = false; } // no conversion if redirecting - $mw =& MagicWord::get( 'redirect' ); + $mw =& MagicWord::get( 'redirect' ); if( $mw->matchStart( $text )) return $text; - if( $isTitle ) { - - // use the title from the T flag if any - if($this->mTitleFromFlag){ - $this->mTitleFromFlag = false; - return $this->mTitleDisplay; - } - - // check for __NOTC__ tag - if( !$this->mDoTitleConvert ) { - $this->mTitleDisplay = $text; - return $text; - } - - global $wgRequest; - $isredir = $wgRequest->getText( 'redirect', 'yes' ); - $action = $wgRequest->getText( 'action' ); - if ( $isredir == 'no' || $action == 'edit' ) { - return $text; - } - else { - $this->mTitleDisplay = $this->convert($text); - return $this->mTitleDisplay; - } - } + // for title convertion + if ($isTitle) return $this->convertTitle($text); $plang = $this->getPreferredVariant(); - if( isset( $this->mVariantFallbacks[$plang] ) ) { - $fallback = $this->mVariantFallbacks[$plang]; - } else { - $fallback = $this->mMainLanguageCode; - } $tarray = explode($this->mMarkup['begin'], $text); $tfirst = array_shift($tarray); if($this->mDoContentConvert) - $text = $this->autoConvert($tfirst); + $text = $this->autoConvert($tfirst,$plang); else $text = $tfirst; - foreach($tarray as $txt) { + foreach($tarray as $txt) { $marked = explode($this->mMarkup['end'], $txt, 2); // strip the flags from syntax like -{T| ... }- list($rules,$flags) = $this->parseFlags($marked[0]); - // proces R flag: output raw content of -{ ... }- - if( in_array('R',$flags) ){ - $disp = $rules; - } else if( $this->mDoContentConvert){ - // parse the contents -{ ... }- - $carray = $this->parseManualRule($rules, $flags); - - $disp = ''; - if(array_key_exists($plang, $carray)) { - $disp = $carray[$plang]; - } else if(array_key_exists($fallback, $carray)) { - $disp = $carray[$fallback]; - } - } else{ - // if we don't do content convert, still strip the -{}- tags - $disp = $rules; - $flags = array(); - } + $text .= $this->parseRules($rules,$flags,$plang); - if($disp) { - // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title - if(in_array('T', $flags)){ - $this->mTitleFromFlag = true; - $this->mTitleDisplay = $disp; - } - else - $text .= $disp; - - // use syntax -{A|zh:WordZh;zh-tw:WordTw}- to introduce a custom mapping between - // words WordZh and WordTw in the whole text - if(in_array('A', $flags)) { - - /* fill in the missing variants, if any, - with fallbacks */ - foreach($this->mVariants as $v) { - if(!array_key_exists($v, $carray)) { - $vf = $this->getVariantFallback($v); - if(array_key_exists($vf, $carray)) - $carray[$v] = $carray[$vf]; - } - } - - foreach($this->mVariants as $vfrom) { - if(!array_key_exists($vfrom, $carray)) - continue; - foreach($this->mVariants as $vto) { - if($vfrom == $vto) - continue; - if(!array_key_exists($vto, $carray)) - continue; - $this->mTables[$vto]->setPair($carray[$vfrom], $carray[$vto]); - } - } - } - } - else { - $text .= $marked[0]; - } if(array_key_exists(1, $marked)){ if( $this->mDoContentConvert ) - $text .= $this->autoConvert($marked[1]); + $text .= $this->autoConvert($marked[1],$plang); else $text .= $marked[1]; } @@ -470,32 +673,6 @@ class LanguageConverter { return $text; } - /** - * parse the manually marked conversion rule - * @param string $rule the text of the rule - * @return array of the translation in each variant - * @private - */ - function parseManualRule($rules, $flags=array()) { - - $choice = explode($this->mMarkup['varsep'], $rules); - $carray = array(); - if(sizeof($choice) == 1) { - /* a single choice */ - foreach($this->mVariants as $v) - $carray[$v] = $choice[0]; - } - else { - foreach($choice as $c) { - $v = explode($this->mMarkup['codesep'], $c); - if(sizeof($v) != 2) // syntax error, skip - continue; - $carray[trim($v[0])] = trim($v[1]); - } - } - return $carray; - } - /** * if a language supports multiple variants, it is * possible that non-existing link in one variant @@ -505,7 +682,7 @@ class LanguageConverter { * @param string $link the name of the link * @param mixed $nt the title object of the link * @return null the input parameters may be modified upon return - * @access public + * @access public */ function findVariantLink( &$link, &$nt ) { global $wgDisableLangConversion; @@ -546,29 +723,29 @@ class LanguageConverter { } /** - * returns language specific hash options - * - * @access public - */ + * returns language specific hash options + * + * @access public + */ function getExtraHashOptions() { $variant = $this->getPreferredVariant(); return '!' . $variant ; } /** - * get title text as defined in the body of the article text - * - * @access public - */ + * get title text as defined in the body of the article text + * + * @access public + */ function getParsedTitle() { return $this->mTitleDisplay; } /** - * a write lock to the cache - * - * @private - */ + * a write lock to the cache + * + * @private + */ function lockCache() { global $wgMemc; $success = false; @@ -581,10 +758,10 @@ class LanguageConverter { } /** - * unlock cache - * - * @private - */ + * unlock cache + * + * @private + */ function unlockCache() { global $wgMemc; $wgMemc->delete($this->mCacheKey . "lock"); @@ -592,20 +769,20 @@ class LanguageConverter { /** - * Load default conversion tables - * This method must be implemented in derived class - * - * @private - */ + * Load default conversion tables + * This method must be implemented in derived class + * + * @private + */ function loadDefaultTables() { $name = get_class($this); wfDie("Must implement loadDefaultTables() method in class $name"); } /** - * load conversion tables either from the cache or the disk - * @private - */ + * load conversion tables either from the cache or the disk + * @private + */ function loadTables($fromcache=true) { global $wgMemc; if( $this->mTablesLoaded ) @@ -642,16 +819,16 @@ class LanguageConverter { } /** - * Hook for post processig after conversion tables are loaded - * - */ + * Hook for post processig after conversion tables are loaded + * + */ function postLoadTables() {} /** - * Reload the conversion tables - * - * @private - */ + * Reload the conversion tables + * + * @private + */ function reloadTables() { if($this->mTables) unset($this->mTables); @@ -661,20 +838,18 @@ class LanguageConverter { /** - * parse the conversion table stored in the cache - * - * the tables should be in blocks of the following form: - - * -{ - * word => word ; - * word => word ; - * ... - * }- - * - * to make the tables more manageable, subpages are allowed - * and will be parsed recursively if $recursive=true - * - * @private + * parse the conversion table stored in the cache + * + * the tables should be in blocks of the following form: + * -{ + * word => word ; + * word => word ; + * ... + * }- + * + * to make the tables more manageable, subpages are allowed + * and will be parsed recursively if $recursive=true + * */ function parseCachedTable($code, $subpage='', $recursive=true) { global $wgMessageCache; @@ -757,37 +932,32 @@ class LanguageConverter { * * @param string $text text to be tagged for no conversion * @return string the tagged text - */ + */ function markNoConversion($text, $noParse=false) { # don't mark if already marked if(strpos($text, $this->mMarkup['begin']) || strpos($text, $this->mMarkup['end'])) return $text; - $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end']; + $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end']; return $ret; } /** * convert the sorting key for category links. this should make different * keys that are variants of each other map to the same key - */ + */ function convertCategoryKey( $key ) { return $key; } /** - * hook to refresh the cache of conversion tables when - * MediaWiki:conversiontable* is updated - * @private - */ + * hook to refresh the cache of conversion tables when + * MediaWiki:conversiontable* is updated + * @private + */ function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) { $titleobj = $article->getTitle(); if($titleobj->getNamespace() == NS_MEDIAWIKI) { - /* - global $wgContLang; // should be an LanguageZh. - if(get_class($wgContLang) != 'languagezh') - return true; - */ $title = $titleobj->getDBkey(); $t = explode('/', $title, 3); $c = count($t); @@ -808,6 +978,4 @@ class LanguageConverter { $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end']; return $ret; } - - } diff --git a/languages/Names.php b/languages/Names.php index e7a5dfb4f6..a1a3faf599 100644 --- a/languages/Names.php +++ b/languages/Names.php @@ -340,10 +340,12 @@ 'zh-cn' => "\xE2\x80\xAA中文(中国大陆)\xE2\x80\xAC", # Chinese (PRC) 'zh-hans' => "\xE2\x80\xAA中文(简体)\xE2\x80\xAC", # Chinese written using the Simplified Chinese script 'zh-hant' => "\xE2\x80\xAA中文(繁體)\xE2\x80\xAC", # Chinese written using the Traditional Chinese script - 'zh-hk' => "\xE2\x80\xAA中文(香港)\xE2\x80\xAC", # Chinese (Hong Kong) + 'zh-hk' => "\xE2\x80\xAA中文(香港)\xE2\x80\xAC", # Chinese (Hong Kong) 'zh-min-nan' => 'Bân-lâm-gú', # Min-nan -- (see bug 8217) + 'zh-mo' => "\xE2\x80\xAA中文(澳門)\xE2\x80\xAC", # Chinese (Macau) + 'zh-my' => "\xE2\x80\xAA中文(马来西亚)\xE2\x80\xAC", # Chinese (Malaysia) 'zh-sg' => "\xE2\x80\xAA中文(新加坡)\xE2\x80\xAC", # Chinese (Singapore) - 'zh-tw' => "\xE2\x80\xAA中文(台灣)\xE2\x80\xAC", # Chinese (Taiwan) + 'zh-tw' => "\xE2\x80\xAA中文(台灣)\xE2\x80\xAC", # Chinese (Taiwan) 'zh-yue' => '粵語', # Cantonese -- (see bug 8217) 'zu' => 'isiZulu' # Zulu ); diff --git a/languages/classes/LanguageZh.php b/languages/classes/LanguageZh.php index ff36798850..ec427b7baf 100644 --- a/languages/classes/LanguageZh.php +++ b/languages/classes/LanguageZh.php @@ -7,33 +7,72 @@ require_once( dirname(__FILE__).'/LanguageZh_hans.php' ); * @ingroup Language */ class ZhConverter extends LanguageConverter { + + function __construct($langobj, $maincode, + $variants=array(), + $variantfallbacks=array(), + $markup=array(), + $flags = array(), + $manualLevel = array() ) { + parent::__construct($langobj, $maincode, + $variants, + $variantfallbacks, + $markup, + $flags, + $manualLevel); + $names = array( + 'zh' => '原文', + 'zh-hans' => '简体', + 'zh-hant' => '繁體', + 'zh-cn' => '大陆', + 'zh-tw' => '台灣', + 'zh-hk' => '香港', + 'zh-mo' => '澳門', + 'zh-sg' => '新加坡', + 'zh-my' => '马来西亚', + ); + $this->mVariantNames = array_merge($this->mVariantNames,$names); + } + function loadDefaultTables() { require( dirname(__FILE__)."/../../includes/ZhConversion.php" ); $this->mTables = array( 'zh-hans' => new ReplacementArray( $zh2Hans ), 'zh-hant' => new ReplacementArray( $zh2Hant ), 'zh-cn' => new ReplacementArray( array_merge($zh2Hans, $zh2CN) ), - 'zh-tw' => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ), - 'zh-sg' => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ), 'zh-hk' => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ), + 'zh-mo' => new ReplacementArray( array_merge($zh2Hant, $zh2HK) ), + 'zh-my' => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ), + 'zh-sg' => new ReplacementArray( array_merge($zh2Hans, $zh2SG) ), + 'zh-tw' => new ReplacementArray( array_merge($zh2Hant, $zh2TW) ), 'zh' => new ReplacementArray ); } function postLoadTables() { $this->mTables['zh-cn']->merge( $this->mTables['zh-hans'] ); - $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] ); - $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] ); $this->mTables['zh-hk']->merge( $this->mTables['zh-hant'] ); + $this->mTables['zh-mo']->merge( $this->mTables['zh-hant'] ); + $this->mTables['zh-my']->merge( $this->mTables['zh-hans'] ); + $this->mTables['zh-sg']->merge( $this->mTables['zh-hans'] ); + $this->mTables['zh-tw']->merge( $this->mTables['zh-hant'] ); } /* there shouldn't be any latin text in Chinese conversion, so no need to mark anything. $noParse is there for compatibility with LanguageConvert::markNoConversion - */ + */ function markNoConversion($text, $noParse = false) { return $text; } + + /* description of convert code in chinese language*/ + function getRulesDesc($bidtable,$unidtable){ + $text=parent::getRulesDesc($bidtable,$unidtable); + $text=str_replace(':','F',$text); + $text=str_replace(';','G',$text); + return $text; + } function convertCategoryKey( $key ) { return $this->autoConvert( $key, 'zh' ); @@ -52,18 +91,28 @@ class LanguageZh extends LanguageZh_hans { global $wgHooks; parent::__construct(); - $variants = array('zh', 'zh-hans', 'zh-hant', 'zh-cn', 'zh-tw', 'zh-sg', 'zh-hk'); + $variants = array('zh','zh-hans','zh-hant','zh-cn','zh-hk','zh-mo','zh-my','zh-sg','zh-tw'); $variantfallbacks = array( - 'zh' => 'zh-hans', - 'zh-hans' => 'zh', - 'zh-hant' => 'zh', - 'zh-cn' => 'zh-hans', - 'zh-sg' => 'zh-hans', - 'zh-tw' => 'zh-hant', - 'zh-hk' => 'zh-hant', + 'zh' => array('zh-hans','zh-hant','zh-cn','zh-tw','zh-hk','zh-sg','zh-mo','zh-my'), + 'zh-hans' => array('zh-cn','zh-sg','zh-my'), + 'zh-hant' => array('zh-tw','zh-hk','zh-mo'), + 'zh-cn' => array('zh-hans','zh-sg','zh-my'), + 'zh-sg' => array('zh-hans','zh-cn','zh-my'), + 'zh-my' => array('zh-hant','zh-sg','zh-cn'), + 'zh-tw' => array('zh-hant','zh-hk','zh-mo'), + 'zh-hk' => array('zh-hant','zh-mo','zh-tw'), + 'zh-mo' => array('zh-hant','zh-hk','zh-tw'), + ); + $ml=array( + 'zh' => 'disable', + 'zh-hans' => 'unidirectional', + 'zh-hant' => 'unidirectional', ); - $this->mConverter = new ZhConverter( $this, 'zh', $variants, $variantfallbacks ); + $this->mConverter = new ZhConverter( $this, 'zh', + $variants, $variantfallbacks, + array(),array(), + $ml); $wgHooks['ArticleSaveComplete'][] = $this->mConverter; } @@ -111,3 +160,4 @@ class LanguageZh extends LanguageZh_hans { return $ret; } } + diff --git a/languages/messages/MessagesZh.php b/languages/messages/MessagesZh.php index 4a901f9e70..4337ce7978 100644 --- a/languages/messages/MessagesZh.php +++ b/languages/messages/MessagesZh.php @@ -22,7 +22,9 @@ Variants for Chinese language 'variantname-zh-cn' => '大陆简体', 'variantname-zh-tw' => '台灣正體', 'variantname-zh-hk' => '香港繁體', +'variantname-zh-mo' => '澳門繁體', 'variantname-zh-sg' => '新加坡简体', +'variantname-zh-my' => '大马简体', 'variantname-zh' => '不转换／不轉換', ); diff --git a/languages/messages/MessagesZh_mo.php b/languages/messages/MessagesZh_mo.php new file mode 100644 index 0000000000..fac8ae0396 --- /dev/null +++ b/languages/messages/MessagesZh_mo.php @@ -0,0 +1,10 @@ + array( -- 2.20.1