X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=languages%2FLanguageConverter.php;h=08171c0a6f5226e9707904646b7379911b9f5787;hb=4b12409823cb5eab8d3f150a021e0e3152ab8a0b;hp=017edd7c1dd69fa2a6fee1b75587201d05269d1f;hpb=0587e5a674332d9104dbf6ef09e65bdd4fdcac50;p=lhc%2Fweb%2Fwiklou.git diff --git a/languages/LanguageConverter.php b/languages/LanguageConverter.php index 017edd7c1d..08171c0a6f 100644 --- a/languages/LanguageConverter.php +++ b/languages/LanguageConverter.php @@ -1,7 +1,6 @@ * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License @@ -15,10 +14,15 @@ class LanguageConverter { var $mTables; var $mTitleDisplay=''; var $mDoTitleConvert=true, $mDoContentConvert=true; + var $mTitleFromFlag = false; var $mCacheKey; var $mLangObj; var $mMarkup; var $mFlags; + var $mUcfirst = false; + + const CACHE_VERSION_KEY = 'VERSION 5'; + /** * Constructor * @@ -29,28 +33,27 @@ class LanguageConverter { * @param array $flags array defining the custom strings that maps to the flags * @access public */ - function LanguageConverter($langobj, $maincode, - $variants=array(), - $variantfallbacks=array(), + function __construct($langobj, $maincode, + $variants=array(), + $variantfallbacks=array(), $markup=array(), $flags = array()) { - global $wgDBname; $this->mLangObj = $langobj; $this->mMainLanguageCode = $maincode; - $this->mVariants = $variants; + $this->mVariants = $variants; $this->mVariantFallbacks = $variantfallbacks; - $this->mCacheKey = $wgDBname . ":conversiontables"; + $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode ); $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':', 'varsep'=>';', 'end'=>'}-'); $this->mMarkup = array_merge($m, $markup); - $f = array('A'=>'A', 'T'=>'T'); + $f = array('A'=>'A', 'T'=>'T', 'R' => 'R'); $this->mFlags = array_merge($f, $flags); } /** * @access public */ - function getVariants() { + function getVariants() { return $this->mVariants; } @@ -60,24 +63,25 @@ class LanguageConverter { * 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 * in this case. right now this is only used by zh. - * + * * @param string $v the language code of the variant * @return string the code of the fallback language or false if there is no fallback - * @access private + * @private */ function getVariantFallback($v) { return $this->mVariantFallbacks[$v]; } - - /** - * get preferred language variants. + + /** + * get preferred language variants. + * @param boolean $fromUser Get it from $wgUser's preferences * @return string the preferred language code * @access public */ - function getPreferredVariant() { - global $wgUser, $wgRequest; - + function getPreferredVariant( $fromUser = true ) { + global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant; + if($this->mPreferredVariant) return $this->mPreferredVariant; @@ -88,26 +92,53 @@ class LanguageConverter { return $req; } - // get language variant preference from logged in users - if(is_object($wgUser) && $wgUser->isLoggedIn() ) { + // check the syntax /code/ArticleTitle + if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){ + // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI + // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations + $scriptBase = basename( $_SERVER['SCRIPT_NAME'] ); + if(in_array($scriptBase,$this->mVariants)){ + $this->mPreferredVariant = $scriptBase; + return $this->mPreferredVariant; + } + } + + // get language variant preference from logged in users + // Don't call this on stub objects because that causes infinite + // recursion during initialisation + if( $fromUser && $wgUser->isLoggedIn() ) { $this->mPreferredVariant = $wgUser->getOption('variant'); + return $this->mPreferredVariant; + } + + // see if default variant is globaly set + if($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 zh- variant is set in the http header, - $this->mPreferredVariant=$this->mMainLanguageCode; + // 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, 'zh-'); + $zh = strstr($header, $pv.'-'); if($zh) { - $this->mPreferredVariant = substr($zh,0,5); + $pv = substr($zh,0,5); } } + // don't try to return bad variant + if(in_array( $pv, $this->mVariants )) + return $pv; } - return $this->mPreferredVariant; + return $this->mMainLanguageCode; + } /** @@ -116,7 +147,7 @@ class LanguageConverter { * @param string $text the text to be converted * @param string $toVariant the target language code * @return string the converted text - * @access private + * @private */ function autoConvert($text, $toVariant=false) { $fname="LanguageConverter::autoConvert"; @@ -126,34 +157,70 @@ class LanguageConverter { if(!$this->mTablesLoaded) $this->loadTables(); - if(!$toVariant) + if(!$toVariant) $toVariant = $this->getPreferredVariant(); if(!in_array($toVariant, $this->mVariants)) return $text; - - $reg = '/<[^>]+>|&[a-z#][a-z0-9]+;|'.UNIQ_PREFIX.'-[a-zA-Z0-9]+/'; + /* we convert everything except: + 1. html markups (anything between < and >) + 2. html entities + 3. place holders created by the parser + */ + global $wgParser; + if (isset($wgParser)) + $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+'; + else + $marker = ""; + + // this one is needed when the text is inside an html markup + $htmlfix = '|<[^>]+$|^[^<>]*>'; + + // disable convert to variants between tags + $codefix = '.+?<\/code>|'; + // disable convertsion of + $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); - $ret = strtr($m[0], $this->mTables[$toVariant]); + + $ret = $this->translate($m[0], $toVariant); $mstart = $m[1]+strlen($m[0]); foreach($matches as $m) { $ret .= substr($text, $mstart, $m[1]-$mstart); - $ret .= strtr($m[0], $this->mTables[$toVariant]); + $ret .= $this->translate($m[0], $toVariant); $mstart = $m[1] + strlen($m[0]); - } + } wfProfileOut( $fname ); return $ret; } - + + /** + * Translate a string to a variant + * Doesn't process markup or do any of that other stuff, for that use convert() + * + * @param string $text Text to convert + * @param string $variant Variant language code + * @return string Translated text + */ + function translate( $text, $variant ) { + wfProfileIn( __METHOD__ ); + if( !$this->mTablesLoaded ) + $this->loadTables(); + $text = $this->mTables[$variant]->replace( $text ); + wfProfileOut( __METHOD__ ); + return $text; + } + /** * convert text to all supported variants * * @param string $text the text to be converted * @return array of string - * @access private + * @public */ function autoConvertToAllVariants($text) { $fname="LanguageConverter::autoConvertToAllVariants"; @@ -163,16 +230,106 @@ class LanguageConverter { $ret = array(); foreach($this->mVariants as $variant) { - $ret[$variant] = strtr($text, $this->mTables[$variant]); + $ret[$variant] = $this->translate($text, $variant); } + wfProfileOut( $fname ); return $ret; } + /** + * 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(); + + $ret = array(); + $tarray = explode($this->mMarkup['begin'], $text); + $tfirst = array_shift($tarray); + + foreach($this->mVariants as $variant) + $ret[$variant] = $this->translate($tfirst,$variant); + + foreach($tarray as $txt) { + $marked = explode($this->mMarkup['end'], $txt, 2); + + foreach($this->mVariants as $variant){ + $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end']; + if(array_key_exists(1, $marked)) + $ret[$variant] .= $this->translate($marked[1],$variant); + } + + } + + return $ret; + } + + + /** + * Convert text using a parser object for context + */ + function parserConvert( $text, &$parser ) { + global $wgDisableLangConversion; + /* don't do anything if this is the conversion table */ + if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI && + strpos($parser->mTitle->getText(), "Conversiontable") !== false ) + { + return $text; + } + + if($wgDisableLangConversion) + return $text; + + $text = $this->convert( $text ); + $parser->mOutput->setTitleText( $this->mTitleDisplay ); + return $text; + } + + /** + * Parse flags with syntax -{FLAG| ... }- + * + */ + function parseFlags($marked){ + $flags = array(); + + // process flag only if the flag is valid + if(strlen($marked) < 2 || !(in_array($marked[0],$this->mFlags) && $marked[1]=='|' ) ) + return array($marked,array()); + + $tt = explode($this->mMarkup['flagsep'], $marked, 2); + + if(sizeof($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]; + } + $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(' ', '', $rules); + } + + return array($rules,$flags); + } + /** * convert text to different variants of a language. the automatic - * conversion is done in autoConvert(). here we parse the text - * marked with -{}-, which specifies special conversions of the + * conversion is done in autoConvert(). here we parse the text + * marked with -{}-, which specifies special conversions of the * text that can not be accomplished in autoConvert() * * syntax of the markup: @@ -180,43 +337,38 @@ class LanguageConverter { * -{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 + * @param bool $isTitle whether this conversion is for the article title * @return string converted text * @access public */ function convert( $text , $isTitle=false) { - global $wgDisableLangConversion; - global $wgTitle; - - /* don't do anything if this is the conversion table */ - if($wgTitle->getNamespace() == NS_MEDIAWIKI && - strpos($wgTitle->getText(), "Conversiontable")!==false) - return $text; - - if($wgDisableLangConversion) - return $text; - - $mw =& MagicWord::get( MAG_NOTITLECONVERT ); + $mw =& MagicWord::get( 'notitleconvert' ); if( $mw->matchAndRemove( $text ) ) $this->mDoTitleConvert = false; - $mw =& MagicWord::get( MAG_NOCONTENTCONVERT ); + $mw =& MagicWord::get( 'nocontentconvert' ); if( $mw->matchAndRemove( $text ) ) { $this->mDoContentConvert = false; } // no conversion if redirecting - $mw =& MagicWord::get( MAG_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 $text; } - if( !empty($this->mTitleDisplay)) - return $this->mTitleDisplay; global $wgRequest; $isredir = $wgRequest->getText( 'redirect', 'yes' ); @@ -225,60 +377,64 @@ class LanguageConverter { return $text; } else { - $this->mTitleDisplay = $this->autoConvert($text); + $this->mTitleDisplay = $this->convert($text); return $this->mTitleDisplay; } } - if( !$this->mDoContentConvert ) - return $text; - $plang = $this->getPreferredVariant(); - $fallback = $this->mVariantFallbacks[$plang]; + if( isset( $this->mVariantFallbacks[$plang] ) ) { + $fallback = $this->mVariantFallbacks[$plang]; + } else { + $fallback = $this->mMainLanguageCode; + } $tarray = explode($this->mMarkup['begin'], $text); $tfirst = array_shift($tarray); - $text = $this->autoConvert($tfirst); - foreach($tarray as $txt) { - $marked = explode($this->mMarkup['end'], $txt); - $flags = array(); - $tt = explode($this->mMarkup['flagsep'], $marked[0], 2); - - if(sizeof($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]; + if($this->mDoContentConvert) + $text = $this->autoConvert($tfirst); + else + $text = $tfirst; + foreach($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]; } - $rules = $tt[1]; + } else{ + // if we don't do content convert, still strip the -{}- tags + $disp = $rules; + $flags = array(); } - else - $rules = $marked[0]; - -#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); - - $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]; + if($disp) { - if(in_array('T', $flags)) + // 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', $flags)) { - /* modify the conversion table for this session*/ /* fill in the missing variants, if any, - with fallbacks */ + with fallbacks */ foreach($this->mVariants as $v) { if(!array_key_exists($v, $carray)) { $vf = $this->getVariantFallback($v); @@ -295,8 +451,7 @@ class LanguageConverter { continue; if(!array_key_exists($vto, $carray)) continue; - $this->mTables[$vto][$carray[$vfrom]] = $carray[$vto]; - + $this->mTables[$vto]->setPair($carray[$vfrom], $carray[$vto]); } } } @@ -304,10 +459,14 @@ class LanguageConverter { else { $text .= $marked[0]; } - if(array_key_exists(1, $marked)) - $text .= $this->autoConvert($marked[1]); + if(array_key_exists(1, $marked)){ + if( $this->mDoContentConvert ) + $text .= $this->autoConvert($marked[1]); + else + $text .= $marked[1]; + } } - + return $text; } @@ -315,7 +474,7 @@ class LanguageConverter { * parse the manually marked conversion rule * @param string $rule the text of the rule * @return array of the translation in each variant - * @access private + * @private */ function parseManualRule($rules, $flags=array()) { @@ -340,7 +499,7 @@ class LanguageConverter { /** * if a language supports multiple variants, it is * possible that non-existing link in one variant - * actually exists in another variant. this function + * actually exists in another variant. this function * tries to find it. See e.g. LanguageZh.php * * @param string $link the name of the link @@ -349,25 +508,35 @@ class LanguageConverter { * @access public */ function findVariantLink( &$link, &$nt ) { - static $count=0; //used to limit this operation - static $cache=array(); global $wgDisableLangConversion; - $pref = $this->getPreferredVariant(); - $ns=0; + $linkBatch = new LinkBatch(); + + $ns=NS_MAIN; + if(is_object($nt)) $ns = $nt->getNamespace(); - if( $count > 50 && $ns != NS_CATEGORY ) - return; - $count++; + $variants = $this->autoConvertToAllVariants($link); if($variants == false) //give up return; + + $titles = array(); + foreach( $variants as $v ) { - if(isset($cache[$v])) - continue; - $cache[$v] = 1; - $varnt = Title::newFromText( $v ); - if( $varnt && $varnt->getArticleID() > 0 ) { + if($v != $link){ + $varnt = Title::newFromText( $v, $ns ); + if(!is_null($varnt)){ + $linkBatch->addObj($varnt); + $titles[]=$varnt; + } + } + } + + // fetch all variants in single query + $linkBatch->execute(); + + foreach( $titles as $varnt ) { + if( $varnt->getArticleID() > 0 ) { $nt = $varnt; if( !$wgDisableLangConversion ) $link = $v; @@ -378,7 +547,7 @@ class LanguageConverter { /** * returns language specific hash options - * + * * @access public */ function getExtraHashOptions() { @@ -398,7 +567,7 @@ class LanguageConverter { /** * a write lock to the cache * - * @access private + * @private */ function lockCache() { global $wgMemc; @@ -408,13 +577,13 @@ class LanguageConverter { break; sleep(1); } - return $success; + return $success; } /** * unlock cache * - * @access private + * @private */ function unlockCache() { global $wgMemc; @@ -425,46 +594,53 @@ class LanguageConverter { /** * Load default conversion tables * This method must be implemented in derived class - * - * @access private + * + * @private */ function loadDefaultTables() { $name = get_class($this); - die("Must implement loadDefaultTables() method in class $name"); + wfDie("Must implement loadDefaultTables() method in class $name"); } /** * load conversion tables either from the cache or the disk - * @access private + * @private */ function loadTables($fromcache=true) { global $wgMemc; if( $this->mTablesLoaded ) return; + wfProfileIn( __METHOD__ ); $this->mTablesLoaded = true; + $this->mTables = false; if($fromcache) { + wfProfileIn( __METHOD__.'-cache' ); $this->mTables = $wgMemc->get( $this->mCacheKey ); - if( !empty( $this->mTables ) ) //all done - return; - } - // not in cache, or we need a fresh reload. - // we will first load the default tables - // then update them using things in MediaWiki:Zhconversiontable/* - global $wgMessageCache; - $this->loadDefaultTables(); - foreach($this->mVariants as $var) { - $cached = $this->parseCachedTable($var); - $this->mTables[$var] = array_merge($this->mTables[$var], $cached); + wfProfileOut( __METHOD__.'-cache' ); } + if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) { + wfProfileIn( __METHOD__.'-recache' ); + // not in cache, or we need a fresh reload. + // 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); + } - $this->postLoadTables(); - - if($this->lockCache()) { - $wgMemc->set($this->mCacheKey, $this->mTables, 43200); - $this->unlockCache(); + $this->postLoadTables(); + $this->mTables[self::CACHE_VERSION_KEY] = true; + + if($this->lockCache()) { + $wgMemc->set($this->mCacheKey, $this->mTables, 43200); + $this->unlockCache(); + } + wfProfileOut( __METHOD__.'-recache' ); } + wfProfileOut( __METHOD__ ); } - + /** * Hook for post processig after conversion tables are loaded * @@ -473,8 +649,8 @@ class LanguageConverter { /** * Reload the conversion tables - * - * @access private + * + * @private */ function reloadTables() { if($this->mTables) @@ -485,7 +661,7 @@ class LanguageConverter { /** - * parse the conversion table stored in the cache + * parse the conversion table stored in the cache * * the tables should be in blocks of the following form: @@ -494,11 +670,11 @@ class LanguageConverter { * word => word ; * ... * }- - * + * * to make the tables more manageable, subpages are allowed * and will be parsed recursively if $recursive=true * - * @access private + * @private */ function parseCachedTable($code, $subpage='', $recursive=true) { global $wgMessageCache; @@ -513,7 +689,7 @@ class LanguageConverter { if(array_key_exists($key, $parsed)) return array(); - + $txt = $wgMessageCache->get( $key, true, true, true ); @@ -530,7 +706,7 @@ class LanguageConverter { $b = explode('/', trim($b[0]), 3); if(count($b)==3) $sublink = $b[2]; - else + else $sublink = ''; if($b[0] == $linkhead && $b[1] == $code) { @@ -542,7 +718,7 @@ class LanguageConverter { // parse the mappings in this page $blocks = explode($this->mMarkup['begin'], $txt); array_shift($blocks); - $ret = array(); + $ret = array(); foreach($blocks as $block) { $mappings = explode($this->mMarkup['end'], $block, 2); $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]); @@ -566,17 +742,23 @@ class LanguageConverter { $ret = array_merge($ret, $s); } } + + if ($this->mUcfirst) { + foreach ($ret as $k => $v) { + $ret[Language::ucfirst($k)] = Language::ucfirst($v); + } + } return $ret; } /** * Enclose a string with the "no conversion" tag. This is used by * various functions in the Parser - * + * * @param string $text text to be tagged for no conversion * @return string the tagged text */ - function markNoConversion($text) { + function markNoConversion($text, $noParse=false) { # don't mark if already marked if(strpos($text, $this->mMarkup['begin']) || strpos($text, $this->mMarkup['end'])) @@ -587,23 +769,23 @@ class LanguageConverter { } /** - * convert the sorting key for category links. this should make different + * 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 $key; } /** - * hook to refresh the cache of conversion tables when + * hook to refresh the cache of conversion tables when * MediaWiki:conversiontable* is updated - * @access private + * @private */ - function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section) { + function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) { $titleobj = $article->getTitle(); - if($titleobj->getNamespace() == NS_MEDIAWIKI) { + if($titleobj->getNamespace() == NS_MEDIAWIKI) { /* global $wgContLang; // should be an LanguageZh. - if(get_class($wgContLang) != 'languagezh') + if(get_class($wgContLang) != 'languagezh') return true; */ $title = $titleobj->getDBkey(); @@ -617,6 +799,15 @@ class LanguageConverter { } return true; } -} -?> + /** + * Armour rendered math against conversion + * Wrap math into rawoutput -{R| math }- syntax + */ + function armourMath($text){ + $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end']; + return $ret; + } + + +}