languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4   * @ingroup Language
   5   *
   6   * @author Zhengzhu Feng <zhengzhu@gmail.com>
   7   * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8   * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>
   9   */
  10
  11 class LanguageConverter {
  12         var $mPreferredVariant='';
  13         var $mMainLanguageCode;
  14         var $mVariants, $mVariantFallbacks, $mVariantNames;
  15         var $mTablesLoaded = false;
  16         var $mTables;
  17         var $mTitleDisplay='';
  18         var $mDoTitleConvert=true, $mDoContentConvert=true;
  19         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  20         var $mTitleFromFlag = false;
  21         var $mCacheKey;
  22         var $mLangObj;
  23         var $mMarkup;
  24         var $mFlags;
  25         var $mUcfirst = false;
  26
  27         const CACHE_VERSION_KEY = 'VERSION 6';
  28
  29         /**
  30          * Constructor
  31          *
  32          * @param string $maincode the main language code of this language
  33          * @param array $variants the supported variants of this language
  34          * @param array $variantfallback the fallback language of each variant
  35          * @param array $markup array defining the markup used for manual conversion
  36          * @param array $flags array defining the custom strings that maps to the flags
  37          * @access public
  38          */
  39         function __construct($langobj, $maincode,
  40                                                                 $variants=array(),
  41                                                                 $variantfallbacks=array(),
  42                                                                 $markup=array(),
  43                                                                 $flags = array(),
  44                                                                 $manualLevel = array() ) {
  45                 $this->mLangObj = $langobj;
  46                 $this->mMainLanguageCode = $maincode;
  47                 $this->mVariants = $variants;
  48                 $this->mVariantFallbacks = $variantfallbacks;
  49                 global $wgLanguageNames;
  50                 $this->mVariantNames = $wgLanguageNames;
  51                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  52                 $m = array(
  53                         'begin'=>'-{',
  54                         'flagsep'=>'|',
  55                         'unidsep'=>'=>', //for unidirectional conversion
  56                         'codesep'=>':',
  57                         'varsep'=>';',
  58                         'end'=>'}-'
  59                 );
  60                 $this->mMarkup = array_merge($m, $markup);
  61                 $f = array(
  62                         // 'S' show converted text
  63                         // '+' add rules for alltext
  64                         // 'E' the gave flags is error
  65                         // these flags above are reserved for program
  66                         'A'=>'A',       // add rule for convert code (all text convert)
  67                         'T'=>'T',       // title convert
  68                         'R'=>'R',       // raw content
  69                         'D'=>'D',       // convert description (subclass implement)
  70                         '-'=>'-',       // remove convert (not implement)
  71                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  72                         'N'=>'N'        // current variant name
  73                 );
  74                 $this->mFlags = array_merge($f, $flags);
  75                 foreach( $this->mVariants as $v)
  76                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  77                                                                 ?$manualLevel[$v]
  78                                                                 :'bidirectional';
  79         }
  80
  81         /**
  82          * @public
  83          */
  84         function getVariants() {
  85                 return $this->mVariants;
  86         }
  87
  88         /**
  89          * in case some variant is not defined in the markup, we need
  90          * to have some fallback. for example, in zh, normally people
  91          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
  92          * when zh-sg is preferred but not defined, we will pick zh-hans
  93          * in this case. right now this is only used by zh.
  94          *
  95          * @param string $v the language code of the variant
  96          * @return string array the code of the fallback language or false if there is no fallback
  97          * @private
  98          */
  99         function getVariantFallbacks($v) {
 100                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 101                         return $this->mVariantFallbacks[$v];
 102                 }
 103                 return $this->mMainLanguageCode;
 104         }
 105
 106         /**
 107          * get preferred language variants.
 108          * @param boolean $fromUser Get it from $wgUser's preferences
 109          * @return string the preferred language code
 110          * @public
 111          */
 112         function getPreferredVariant( $fromUser = true ) {
 113                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 114
 115                 if($this->mPreferredVariant)
 116                         return $this->mPreferredVariant;
 117
 118                 // see if the preference is set in the request
 119                 $req = $wgRequest->getText( 'variant' );
 120                 if( in_array( $req, $this->mVariants ) ) {
 121                         $this->mPreferredVariant = $req;
 122                         return $req;
 123                 }
 124
 125                 // check the syntax /code/ArticleTitle
 126                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 127                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 128                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 129                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 130                         if(in_array($scriptBase,$this->mVariants)){
 131                                 $this->mPreferredVariant = $scriptBase;
 132                                 return $this->mPreferredVariant;
 133                         }
 134                 }
 135
 136                 // get language variant preference from logged in users
 137                 // Don't call this on stub objects because that causes infinite
 138                 // recursion during initialisation
 139                 if( $fromUser && $wgUser->isLoggedIn() )  {
 140                         $this->mPreferredVariant = $wgUser->getOption('variant');
 141                         return $this->mPreferredVariant;
 142                 }
 143
 144                 // see if default variant is globaly set
 145                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 146                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 147                         return $this->mPreferredVariant;
 148                 }
 149
 150                 # FIXME rewrite code for parsing http header. The current code
 151                 # is written specific for detecting zh- variants
 152                 if( !$this->mPreferredVariant ) {
 153                         // see if some supported language variant is set in the
 154                         // http header, but we don't set the mPreferredVariant
 155                         // variable in case this is called before the user's
 156                         // preference is loaded
 157                         $pv=$this->mMainLanguageCode;
 158                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 159                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 160                                 $zh = strstr($header, $pv.'-');
 161                                 if($zh) {
 162                                         $pv = substr($zh,0,5);
 163                                 }
 164                         }
 165                         // don't try to return bad variant
 166                         if(in_array( $pv, $this->mVariants ))
 167                                 return $pv;
 168                 }
 169
 170                 return $this->mMainLanguageCode;
 171
 172         }
 173
 174         /**
 175          * dictionary-based conversion
 176          *
 177          * @param string $text the text to be converted
 178          * @param string $toVariant the target language code
 179          * @return string the converted text
 180          * @private
 181          */
 182         function autoConvert($text, $toVariant=false) {
 183                 $fname="LanguageConverter::autoConvert";
 184
 185                 wfProfileIn( $fname );
 186
 187                 if(!$this->mTablesLoaded)
 188                         $this->loadTables();
 189
 190                 if(!$toVariant)
 191                         $toVariant = $this->getPreferredVariant();
 192                 if(!in_array($toVariant, $this->mVariants))
 193                         return $text;
 194
 195                 /* we convert everything except:
 196                    1. html markups (anything between < and >)
 197                    2. html entities
 198                    3. place holders created by the parser
 199                 */
 200                 global $wgParser;
 201                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 202                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 203                 } else
 204                         $marker = "";
 205
 206                 // this one is needed when the text is inside an html markup
 207                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 208
 209                 // disable convert to variants between <code></code> tags
 210                 $codefix = '<code>.+?<\/code>|';
 211                 // disable convertsion of <script type="text/javascript"> ... </script>
 212                 $scriptfix = '<script.*?>.*?<\/script>|';
 213                 // disable conversion of <pre xxxx> ... </pre>
 214                 $prefix = '<pre.*?>.*?<\/pre>|';
 215
 216                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 217
 218                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 219
 220                 $m = array_shift($matches);
 221
 222                 $ret = $this->translate($m[0], $toVariant);
 223                 $mstart = $m[1]+strlen($m[0]);
 224                 foreach($matches as $m) {
 225                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 226                         $ret .= $this->translate($m[0], $toVariant);
 227                         $mstart = $m[1] + strlen($m[0]);
 228                 }
 229                 wfProfileOut( $fname );
 230                 return $ret;
 231         }
 232
 233         /**
 234          * Translate a string to a variant
 235          * Doesn't process markup or do any of that other stuff, for that use convert()
 236          *
 237          * @param string $text Text to convert
 238          * @param string $variant Variant language code
 239          * @return string Translated text
 240          * @private
 241          */
 242         function translate( $text, $variant ) {
 243                 wfProfileIn( __METHOD__ );
 244                 if( !$this->mTablesLoaded )
 245                         $this->loadTables();
 246                 $text = $this->mTables[$variant]->replace( $text );
 247                 wfProfileOut( __METHOD__ );
 248                 return $text;
 249         }
 250
 251         /**
 252          * convert text to all supported variants
 253          *
 254          * @param string $text the text to be converted
 255          * @return array of string
 256          * @public
 257          */
 258         function autoConvertToAllVariants($text) {
 259                 $fname="LanguageConverter::autoConvertToAllVariants";
 260                 wfProfileIn( $fname );
 261                 if( !$this->mTablesLoaded )
 262                         $this->loadTables();
 263
 264                 $ret = array();
 265                 foreach($this->mVariants as $variant) {
 266                         $ret[$variant] = $this->translate($text, $variant);
 267                 }
 268
 269                 wfProfileOut( $fname );
 270                 return $ret;
 271         }
 272
 273         /**
 274          * convert link text to all supported variants
 275          *
 276          * @param string $text the text to be converted
 277          * @return array of string
 278          * @public
 279          */
 280         function convertLinkToAllVariants($text) {
 281                 if( !$this->mTablesLoaded )
 282                         $this->loadTables();
 283
 284                 $ret = array();
 285                 $tarray = explode($this->mMarkup['begin'], $text);
 286                 $tfirst = array_shift($tarray);
 287
 288                 foreach($this->mVariants as $variant)
 289                         $ret[$variant] = $this->translate($tfirst,$variant);
 290
 291                 foreach($tarray as $txt) {
 292                         $marked = explode($this->mMarkup['end'], $txt, 2);
 293
 294                         foreach($this->mVariants as $variant){
 295                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 296                                 if(array_key_exists(1, $marked))
 297                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 298                         }
 299
 300                 }
 301
 302                 return $ret;
 303         }
 304
 305
 306         /**
 307          * Convert text using a parser object for context
 308          * @public
 309          */
 310         function parserConvert( $text, &$parser ) {
 311                 global $wgDisableLangConversion;
 312                 /* don't do anything if this is the conversion table */
 313                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 314                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 315                 {
 316                         return $text;
 317                 }
 318
 319                 if($wgDisableLangConversion)
 320                         return $text;
 321
 322                 $text = $this->convert( $text );
 323                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 324                 return $text;
 325         }
 326
 327         /**
 328          * @private
 329          */
 330         function applyManualConv($convRule,$variant){
 331                 if(!$variant) $variant = $this->getPreferredVariant();
 332                 $rules = $convRule->getRules();
 333                 $flags = $convRule->getFlags();
 334                 list($bidtable, $unidtable) = $convRule->getConvTable();
 335
 336                 $is_title_flag = in_array('T',  $flags);
 337                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 338                 if($is_title_flag){
 339                         $this->mTitleFromFlag = true;
 340                         $this->mTitleDisplay =  $convRule->getTitle();
 341                 }
 342
 343                 if($this->mManualLevel[$variant]=='disable') return;
 344
 345                 $is_remove_flag = !$is_title_flag && in_array('-', $flags);
 346                 $is_add_flag = !$is_remove_flag && in_array('+', $flags);
 347                 $is_bidMC = $this->mManualLevel[$variant]=='bidirectional';
 348                 $is_unidMC = $this->mManualLevel[$variant]=='unidirectional';
 349                 $vmarked=array();
 350
 351                 foreach($this->mVariants as $v) {
 352                         /* for bidirectional array
 353                                 fill in the missing variants, if any,
 354                                 with fallbacks */
 355                         if($is_bidMC && !array_key_exists($v, $bidtable)) {
 356                                 $vf = $convRule->getTextInBidtable($this->getVariantFallbacks($v) );
 357                                 if($vf) $bidtable[$v] = $vf;
 358                         }
 359                         if($is_bidMC && array_key_exists($v,$bidtable)){
 360                                 foreach($vmarked as $vo){
 361                                         // use syntax:
 362                                         //  -{A|zh:WordZh;zh-tw:WordTw}- or -{+|zh:WordZh;zh-tw:WordTw}-
 363                                         // to introduce a custom mapping between
 364                                         // words WordZh and WordTw in the whole text
 365                                         if($is_add_flag){
 366                                                 $this->mTables[$v]->setPair($bidtable[$vo], $bidtable[$v]);
 367                                                 $this->mTables[$vo]->setPair($bidtable[$v], $bidtable[$vo]);
 368                                         }
 369                                         // use syntax -{-|zh:WordZh;zh-tw:WordTw}- to remove a conversion
 370                                         // words WordZh and WordTw in the whole text
 371                                         if($is_remove_flag){
 372                                                 $this->mTables[$v]->removePair($bidtable[$vo]);
 373                                                 $this->mTables[$vo]->removePair($bidtable[$v]);
 374                                         }
 375                                 }
 376                                 $vmarked[]=$v;
 377                         }
 378                         /*for unidirectional array
 379                                 fill to convert tables */
 380                         if($is_unidMC && array_key_exists($v,$unidtable)){
 381                                 if($is_add_flag)$this->mTables[$v]->mergeArray($unidtable[$v]);
 382                                 if($is_remove_flag)$this->mTables[$v]->removeArray($unidtable[$v]);
 383                         }
 384                 }
 385         }
 386
 387         /**
 388          *  convert title
 389          * @private
 390          */
 391         function convertTitle($text){
 392                 // check for __NOTC__ tag
 393                 if( !$this->mDoTitleConvert ) {
 394                         $this->mTitleDisplay = $text;
 395                         return $text;
 396                 }
 397
 398                 // use the title from the T flag if any
 399                 if($this->mTitleFromFlag){
 400                         $this->mTitleFromFlag = false;
 401                         return $this->mTitleDisplay;
 402                 }
 403
 404                 global $wgRequest;
 405                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 406                 $action = $wgRequest->getText( 'action' );
 407                 if ( $isredir == 'no' || $action == 'edit' ) {
 408                         return $text;
 409                 } else {
 410                         $this->mTitleDisplay = $this->convert($text);
 411                         return $this->mTitleDisplay;
 412                 }
 413         }
 414
 415         /**
 416          * convert text to different variants of a language. the automatic
 417          * conversion is done in autoConvert(). here we parse the text
 418          * marked with -{}-, which specifies special conversions of the
 419          * text that can not be accomplished in autoConvert()
 420          *
 421          * syntax of the markup:
 422          * -{code1:text1;code2:text2;...}-  or
 423          * -{flags|code1:text1;code2:text2;...}-  or
 424          * -{text}- in which case no conversion should take place for text
 425          *
 426          * @param string $text text to be converted
 427          * @param bool $isTitle whether this conversion is for the article title
 428          * @return string converted text
 429          * @public
 430          */
 431         function convert( $text , $isTitle=false) {
 432
 433                 $mw =& MagicWord::get( 'notitleconvert' );
 434                 if( $mw->matchAndRemove( $text ) )
 435                         $this->mDoTitleConvert = false;
 436                 $mw =& MagicWord::get( 'nocontentconvert' );
 437                 if( $mw->matchAndRemove( $text ) ) {
 438                         $this->mDoContentConvert = false;
 439                 }
 440
 441                 // no conversion if redirecting
 442                 $mw =& MagicWord::get( 'redirect' );
 443                 if( $mw->matchStart( $text ))
 444                         return $text;
 445
 446                 // for title convertion
 447                 if ($isTitle) return $this->convertTitle($text);
 448
 449                 $plang = $this->getPreferredVariant();
 450
 451                 $tarray = explode($this->mMarkup['begin'], $text);
 452                 $tfirst = array_shift($tarray);
 453                 if($this->mDoContentConvert)
 454                         $text = $this->autoConvert($tfirst,$plang);
 455                 else
 456                         $text = $tfirst;
 457                 foreach($tarray as $txt) {
 458                         $marked = explode($this->mMarkup['end'], $txt, 2);
 459
 460                         // strip the flags from syntax like -{T| ... }-
 461                         $crule = new ConverterRule($marked[0],
 462                                                         $this->mMarkup, $this->mFlags,
 463                                                         $this->mVariantNames,
 464                                                         $this->mDoContentConvert,
 465                                                         $this->mDoTitleConvert);
 466                         $crule->parse($plang,
 467                                                         $this->getVariantFallbacks($plang),
 468                                                         $this->mManualLevel[$plang]=='disable');
 469                         $text .= $crule->getDisplay();
 470                         $this->applyManualConv($crule,$plang);
 471
 472                         if(array_key_exists(1, $marked)){
 473                                 if( $this->mDoContentConvert )
 474                                         $text .= $this->autoConvert($marked[1],$plang);
 475                                 else
 476                                         $text .= $marked[1];
 477                         }
 478                 }
 479
 480                 return $text;
 481         }
 482
 483         /**
 484          * if a language supports multiple variants, it is
 485          * possible that non-existing link in one variant
 486          * actually exists in another variant. this function
 487          * tries to find it. See e.g. LanguageZh.php
 488          *
 489          * @param string $link the name of the link
 490          * @param mixed $nt the title object of the link
 491          * @return null the input parameters may be modified upon return
 492          * @public
 493          */
 494         function findVariantLink( &$link, &$nt ) {
 495                 global $wgDisableLangConversion;
 496                 $linkBatch = new LinkBatch();
 497
 498                 $ns=NS_MAIN;
 499
 500                 if(is_object($nt))
 501                         $ns = $nt->getNamespace();
 502
 503                 $variants = $this->autoConvertToAllVariants($link);
 504                 if($variants == false) //give up
 505                         return;
 506
 507                 $titles = array();
 508
 509                 foreach( $variants as $v ) {
 510                         if($v != $link){
 511                                 $varnt = Title::newFromText( $v, $ns );
 512                                 if(!is_null($varnt)){
 513                                         $linkBatch->addObj($varnt);
 514                                         $titles[]=$varnt;
 515                                 }
 516                         }
 517                 }
 518
 519                 // fetch all variants in single query
 520                 $linkBatch->execute();
 521
 522                 foreach( $titles as $varnt ) {
 523                         if( $varnt->getArticleID() > 0 ) {
 524                                 $nt = $varnt;
 525                                 if( !$wgDisableLangConversion )
 526                                         $link = $v;
 527                                 break;
 528                         }
 529                 }
 530         }
 531
 532     /**
 533          * returns language specific hash options
 534          *
 535          * @public
 536          */
 537         function getExtraHashOptions() {
 538                 $variant = $this->getPreferredVariant();
 539                 return '!' . $variant ;
 540         }
 541
 542     /**
 543          * get title text as defined in the body of the article text
 544          *
 545          * @public
 546          */
 547         function getParsedTitle() {
 548                 return $this->mTitleDisplay;
 549         }
 550
 551         /**
 552          * a write lock to the cache
 553          *
 554          * @private
 555          */
 556         function lockCache() {
 557                 global $wgMemc;
 558                 $success = false;
 559                 for($i=0; $i<30; $i++) {
 560                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 561                                 break;
 562                         sleep(1);
 563                 }
 564                 return $success;
 565         }
 566
 567         /**
 568          * unlock cache
 569          *
 570          * @private
 571          */
 572         function unlockCache() {
 573                 global $wgMemc;
 574                 $wgMemc->delete($this->mCacheKey . "lock");
 575         }
 576
 577
 578         /**
 579          * Load default conversion tables
 580          * This method must be implemented in derived class
 581          *
 582          * @private
 583          */
 584         function loadDefaultTables() {
 585                 $name = get_class($this);
 586                 wfDie("Must implement loadDefaultTables() method in class $name");
 587         }
 588
 589         /**
 590          * load conversion tables either from the cache or the disk
 591          * @private
 592          */
 593         function loadTables($fromcache=true) {
 594                 global $wgMemc;
 595                 if( $this->mTablesLoaded )
 596                         return;
 597                 wfProfileIn( __METHOD__ );
 598                 $this->mTablesLoaded = true;
 599                 $this->mTables = false;
 600                 if($fromcache) {
 601                         wfProfileIn( __METHOD__.'-cache' );
 602                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 603                         wfProfileOut( __METHOD__.'-cache' );
 604                 }
 605                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 606                         wfProfileIn( __METHOD__.'-recache' );
 607                         // not in cache, or we need a fresh reload.
 608                         // we will first load the default tables
 609                         // then update them using things in MediaWiki:Zhconversiontable/*
 610                         $this->loadDefaultTables();
 611                         foreach($this->mVariants as $var) {
 612                                 $cached = $this->parseCachedTable($var);
 613                                 $this->mTables[$var]->mergeArray($cached);
 614                         }
 615
 616                         $this->postLoadTables();
 617                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 618
 619                         if($this->lockCache()) {
 620                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 621                                 $this->unlockCache();
 622                         }
 623                         wfProfileOut( __METHOD__.'-recache' );
 624                 }
 625                 wfProfileOut( __METHOD__ );
 626         }
 627
 628     /**
 629          * Hook for post processig after conversion tables are loaded
 630          *
 631          */
 632         function postLoadTables() {}
 633
 634     /**
 635          * Reload the conversion tables
 636          *
 637          * @private
 638          */
 639         function reloadTables() {
 640                 if($this->mTables)
 641                         unset($this->mTables);
 642                 $this->mTablesLoaded = false;
 643                 $this->loadTables(false);
 644         }
 645
 646
 647         /**
 648          * parse the conversion table stored in the cache
 649          *
 650          * the tables should be in blocks of the following form:
 651          *              -{
 652          *                      word => word ;
 653          *                      word => word ;
 654          *                      ...
 655          *              }-
 656          *
 657          *      to make the tables more manageable, subpages are allowed
 658          *      and will be parsed recursively if $recursive=true
 659          *
 660          */
 661         function parseCachedTable($code, $subpage='', $recursive=true) {
 662                 global $wgMessageCache;
 663                 static $parsed = array();
 664
 665                 if(!is_object($wgMessageCache))
 666                         return array();
 667
 668                 $key = 'Conversiontable/'.$code;
 669                 if($subpage)
 670                         $key .= '/' . $subpage;
 671
 672                 if(array_key_exists($key, $parsed))
 673                         return array();
 674
 675                 if ( strpos( $code, '/' ) === false ) {
 676                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 677                 } else {
 678                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 679                         if ( $title && $title->exists() ) {
 680                                 $article = new Article( $title );
 681                                 $txt = $article->getContents();
 682                         } else {
 683                                 $txt = '';
 684                         }
 685                 }
 686
 687                 // get all subpage links of the form
 688                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 689                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 690                 $subs = explode('[[', $txt);
 691                 $sublinks = array();
 692                 foreach( $subs as $sub ) {
 693                         $link = explode(']]', $sub, 2);
 694                         if(count($link) != 2)
 695                                 continue;
 696                         $b = explode('|', $link[0]);
 697                         $b = explode('/', trim($b[0]), 3);
 698                         if(count($b)==3)
 699                                 $sublink = $b[2];
 700                         else
 701                                 $sublink = '';
 702
 703                         if($b[0] == $linkhead && $b[1] == $code) {
 704                                 $sublinks[] = $sublink;
 705                         }
 706                 }
 707
 708
 709                 // parse the mappings in this page
 710                 $blocks = explode($this->mMarkup['begin'], $txt);
 711                 array_shift($blocks);
 712                 $ret = array();
 713                 foreach($blocks as $block) {
 714                         $mappings = explode($this->mMarkup['end'], $block, 2);
 715                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 716                         $table = explode( ';', $stripped );
 717                         foreach( $table as $t ) {
 718                                 $m = explode( '=>', $t );
 719                                 if( count( $m ) != 2)
 720                                         continue;
 721                                 // trim any trailling comments starting with '//'
 722                                 $tt = explode('//', $m[1], 2);
 723                                 $ret[trim($m[0])] = trim($tt[0]);
 724                         }
 725                 }
 726                 $parsed[$key] = true;
 727
 728
 729                 // recursively parse the subpages
 730                 if($recursive) {
 731                         foreach($sublinks as $link) {
 732                                 $s = $this->parseCachedTable($code, $link, $recursive);
 733                                 $ret = array_merge($ret, $s);
 734                         }
 735                 }
 736
 737                 if ($this->mUcfirst) {
 738                         foreach ($ret as $k => $v) {
 739                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 740                         }
 741                 }
 742                 return $ret;
 743         }
 744
 745         /**
 746          * Enclose a string with the "no conversion" tag. This is used by
 747          * various functions in the Parser
 748          *
 749          * @param string $text text to be tagged for no conversion
 750          * @return string the tagged text
 751          * @public
 752          */
 753         function markNoConversion($text, $noParse=false) {
 754                 # don't mark if already marked
 755                 if(strpos($text, $this->mMarkup['begin']) ||
 756                    strpos($text, $this->mMarkup['end']))
 757                         return $text;
 758
 759                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 760                 return $ret;
 761         }
 762
 763         /**
 764          * convert the sorting key for category links. this should make different
 765          * keys that are variants of each other map to the same key
 766          */
 767         function convertCategoryKey( $key ) {
 768                 return $key;
 769         }
 770         /**
 771          * hook to refresh the cache of conversion tables when
 772          * MediaWiki:conversiontable* is updated
 773          * @private
 774          */
 775         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 776                 $titleobj = $article->getTitle();
 777                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 778                         $title = $titleobj->getDBkey();
 779                         $t = explode('/', $title, 3);
 780                         $c = count($t);
 781                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 782                                 if(in_array($t[1], $this->mVariants)) {
 783                                         $this->reloadTables();
 784                                 }
 785                         }
 786                 }
 787                 return true;
 788         }
 789
 790         /**
 791          * Armour rendered math against conversion
 792          * Wrap math into rawoutput -{R| math }- syntax
 793          * @public
 794          */
 795         function armourMath($text){
 796                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 797                 return $ret;
 798         }
 799 }
 800
 801 /**
 802  * parser for rules of language conversion , parse rules in -{ }- tag
 803  * @ingroup Language
 804  * @author  fdcn <fdcn64@gmail.com>
 805  */
 806 class ConverterRule {
 807         var $mText; // original text in -{text}-
 808         var $mVariantNames;
 809         var $mMarkup;
 810         var $mRules;
 811         var $mFlags;
 812         var $mManualCodeError='<span style="color: red;">code error!</span>';
 813         var $mDoTitleConvert=true, $mDoContentConvert=true;
 814         var $ruleDisplay = '',$ruleTitle=false;
 815         var $rules = '';// string $rule the text of the rule
 816         var $flags = array();
 817         var $bidtable = array();// array of the translation in each variant
 818         var $unidtable = array();// array of the translation in each variant
 819
 820         /**
 821          * Constructor
 822          *
 823          * @param string $text the text between -{ and }-
 824          * @param array $markup the supported markup for conversion
 825          * @param array $flags the supported flags for conversion
 826          * @param array $variantNames the names list of supported language
 827          * @param bool $doTitleConvert if do title convert
 828          * @param bool $doContentConvert if do content convert
 829          * @access public
 830          */
 831         function __construct($text,$markup,$flags,$variantNames,
 832                                         $doTitleConvert=true, $doContentConvert=true){
 833                 $this->mText = $text;
 834                 $this->mMarkup = $markup;
 835                 $this->mFlags = $flags;
 836                 $this->mVariantNames = $variantNames;
 837                 $this->mDoTitleConvert=$doTitleConvert;
 838                 $this->mDoContentConvert=$doContentConvert;
 839         }
 840
 841         /**
 842          * check if variants array in convert array
 843          *
 844          * @param string $variant Variant language code
 845          * @return string Translated text
 846          * @public
 847          */
 848         function getTextInBidtable($variants){
 849                 if(is_string($variants)){ $variants=array($variants); }
 850                 if(!is_array($variants)) return false;
 851                 foreach ($variants as $variant){
 852                         if(array_key_exists($variant, $this->bidtable)){
 853                                 return $this->bidtable[$variant];
 854                         }
 855                 }
 856                 return false;
 857         }
 858
 859         /**
 860          * Parse flags with syntax -{FLAG| ... }-
 861          * @private
 862          */
 863         function parseFlags(){
 864                 $text = $this->mText;
 865                 if(strlen($text) < 2 ) {
 866                         $this->flags = array( 'R' );
 867                         $this->rules = $text;
 868                         return;
 869                 }
 870
 871                 $flags = array();
 872                 $tt = explode($this->mMarkup['flagsep'], $text, 2);
 873
 874                 if(count($tt) == 2) {
 875                         $f = explode($this->mMarkup['varsep'], $tt[0]);
 876                         foreach($f as $ff) {
 877                                 $ff = trim($ff);
 878                                 if(array_key_exists($ff, $this->mFlags) &&
 879                                                         !in_array($this->mFlags[$ff], $flags))
 880                                         $flags[] = $this->mFlags[$ff];
 881                         }
 882                         $rules = $tt[1];
 883                 } else {
 884                         $rules = $text;
 885                 }
 886
 887                 if( !in_array('R',$flags) ){
 888                         //FIXME: may cause trouble here...
 889                         //strip &nbsp; since it interferes with the parsing, plus,
 890                         //all spaces should be stripped in this tag anyway.
 891                         $rules = str_replace('&nbsp;', '', $rules);
 892                         $rules = str_replace('=&gt;','=>',$rules);
 893                 }
 894
 895                 //check flags
 896                 if( in_array('R',$flags) ){
 897                         $flags = array('R');// remove other flags
 898                 } elseif ( in_array('N',$flags) ){
 899                         $flags = array('N');// remove other flags
 900                 } elseif ( in_array('-',$flags) ){
 901                         $flags = array('-');// remove other flags
 902                 } elseif (count($flags)==1 && $flags[0]=='T'){
 903                         $flags[]='H';
 904                 } elseif ( in_array('H',$flags) ){
 905                         // replace A flag, and remove other flags except T
 906                         $temp=array('+','H');
 907                         if(in_array('T',$flags)) $temp[] = 'T';
 908                         if(in_array('D',$flags)) $temp[] = 'D';
 909                         $flags = $temp;
 910                 } else {
 911                         if ( in_array('A',$flags)) {
 912                                 $flags[]='+';
 913                                 $flags[]='S';
 914                         }
 915                         if ( in_array('D',$flags) )
 916                                 $flags=array_diff($flags,array('S'));
 917                 }
 918                 if ( count($flags)==0 )
 919                         $flags = array('S');
 920                 $this->rules=$rules;
 921                 $this->flags=$flags;
 922         }
 923
 924         /**
 925          * generate conversion table
 926          * @private
 927          */
 928         function generateConvTable() {
 929                 $rules = $this->rules;
 930                 $flags = $this->flags;
 931                 $bidtable = array();
 932                 $unidtable = array();
 933                 $choice = explode($this->mMarkup['varsep'], $rules );
 934                 foreach($choice as $c) {
 935                         $v = explode($this->mMarkup['codesep'], $c);
 936                         if(count($v) != 2)
 937                                 continue;// syntax error, skip
 938                         $to=trim($v[1]);
 939                         $v=trim($v[0]);
 940                         $u = explode($this->mMarkup['unidsep'], $v);
 941                         if(count($u) == 1) {
 942                                 $bidtable[$v] = $to;
 943                         } else if(count($u) == 2){
 944                                 $from=trim($u[0]);$v=trim($u[1]);
 945                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 946                                         $unidtable[$v]=array($from=>$to);
 947                                 else
 948                                         $unidtable[$v][$from]=$to;
 949                         }
 950                         // syntax error, pass
 951                         if (!array_key_exists($v,$this->mVariantNames))
 952                                 return;
 953                 }
 954                 $this->bidtable = $bidtable;
 955                 $this->unidtable = $unidtable;
 956         }
 957
 958         /**
 959          * @private
 960          */
 961         function getRulesDesc(){
 962                 $text='';
 963                 foreach($this->bidtable as $k => $v)
 964                         $text .= $this->mVariantNames[$k].':'.$v.';';
 965                 foreach($this->unidtable as $k => $a)
 966                         foreach($a as $from=>$to)
 967                                 $text.=$from.'⇒'.$this->mVariantNames[$k].':'.$to.';';
 968                 return $text;
 969         }
 970
 971         /**
 972          *  Parse rules conversion
 973          * @private
 974          */
 975         function getRuleConvertedStr($variant,$variantFallbacks,
 976                                                 $doConvert,$isMCDisable=false){
 977                 $bidtable = $this->bidtable;
 978                 $unidtable = $this->unidtable;
 979
 980                 if( count($bidtable) + count($unidtable) == 0 ){
 981                         return $this->rules;
 982                 } elseif ($doConvert){// the text converted
 983                         // display current variant in bidirectional array
 984                         $disp = $this->getTextInBidtable($variant);
 985                         // or display current variant in fallbacks
 986                         if(!$disp)
 987                                 $disp = $this->getTextInBidtable($variantFallbacks);
 988                         // or display current variant in unidirectional array
 989                         if(!$disp && array_key_exists($variant,$unidtable)){
 990                                 $disp = array_values($unidtable[$variant]);
 991                                 $disp = $disp[0];
 992                         }
 993                         // or display frist text under disable manual convert
 994                         if(!$disp && $isMCDisable) {
 995                                 if(count($bidtable)>0){
 996                                         $disp = array_values($bidtable);
 997                                         $disp = $disp[0];
 998                                 } else {
 999                                         $disp = array_values($unidtable);
1000                                         $disp = array_values($disp[0]);
1001                                         $disp = $disp[0];
1002                                 }
1003                         }
1004                         return $disp;
1005                 } else {// no convert
1006                         return $this->rules;
1007                 }
1008         }
1009
1010         /**
1011          * Parse rules and flags
1012          * @public
1013          */
1014         function parse($variant,$variantFallbacks,$isMCDisable=false){
1015                 $this->parseFlags();
1016                 $this->generateConvTable();
1017
1018                 $rules = $this->rules;
1019                 $flags = $this->flags;
1020
1021                 if(count($this->bidtable)==0 && count($this->unidtable)==0
1022                         && !in_array('N',$flags) && !in_array('T',$flags) )
1023                 {
1024                         $this->flags = $flags = array('R');
1025                 }
1026
1027                 if( in_array('R',$flags) ) {
1028                         // if we don't do content convert, still strip the -{}- tags
1029                         $this->ruleDisplay = $rules;
1030                 } elseif ( in_array('N',$flags) ){
1031                         // proces N flag: output current variant name
1032                         $this->ruleDisplay = $this->mVariantNames[trim($rules)];
1033                 } elseif ( in_array('D',$flags) ){
1034                         // proces D flag: output rules description
1035                         $this->ruleDisplay = $this->getRulesDesc();
1036                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1037                         // proces H,- flag or T only: output nothing
1038                         $this->ruleDisplay = '';
1039                 } elseif ( in_array('S',$flags) ){
1040                         $this->ruleDisplay = $this->getRuleConvertedStr(
1041                                                         $variant,$variantFallbacks,
1042                                                         $this->mDoContentConvert,$isMCDisable);
1043                 } else {
1044                         $this->ruleDisplay= $this->mManualCodeError;
1045                 }
1046                 // proces T flag : output nothing
1047                 if ( in_array('T',$flags) ) {
1048                         $this->ruleTitle = $this->getRuleConvertedStr(
1049                                                         $variant,$variantFallbacks,
1050                                                         $this->mDoTitleConvert,$isMCDisable);
1051                 }
1052         }
1053
1054         /**
1055          * get display text on markup -{...}-
1056          * @param string $variant the current variant
1057          * @public
1058          */
1059         function getDisplay(){
1060                 return $this->ruleDisplay;
1061         }
1062         /**
1063          * get converted title
1064          * @param string $variant the current variant
1065          * @public
1066          */
1067         function getTitle(){
1068                 return $this->ruleTitle;
1069         }
1070
1071         /**
1072          * get conversion table ( bidirectional and unidirectional conversion table )
1073          * @public
1074          */
1075         function getConvTable(){
1076                 return array($this->bidtable, $this->unidtable);
1077         }
1078
1079         /**
1080          * get conversion rules string
1081          * @public
1082          */
1083         function getRules(){
1084                 return $this->rules;
1085         }
1086
1087         /**
1088          * get conversion flags
1089          * @public
1090          */
1091         function getFlags(){
1092                 return $this->flags;
1093         }
1094 }