languages/LanguageConverter.php

   1 <?php
   2 /**
   3   * @addtogroup Language
   4   *
   5   * @author Zhengzhu Feng <zhengzhu@gmail.com>
   6   * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   7   */
   8
   9 class LanguageConverter {
  10         var $mPreferredVariant='';
  11         var $mMainLanguageCode;
  12         var $mVariants, $mVariantFallbacks;
  13         var $mTablesLoaded = false;
  14         var $mTables;
  15         var $mTitleDisplay='';
  16         var $mDoTitleConvert=true, $mDoContentConvert=true;
  17         var $mTitleFromFlag = false;
  18         var $mCacheKey;
  19         var $mLangObj;
  20         var $mMarkup;
  21         var $mFlags;
  22         var $mUcfirst = false;
  23
  24         const CACHE_VERSION_KEY = 'VERSION 5';
  25
  26         /**
  27      * Constructor
  28          *
  29      * @param string $maincode the main language code of this language
  30      * @param array $variants the supported variants of this language
  31      * @param array $variantfallback the fallback language of each variant
  32      * @param array $markup array defining the markup used for manual conversion
  33          * @param array $flags array defining the custom strings that maps to the flags
  34      * @access public
  35      */
  36         function __construct($langobj, $maincode,
  37                                                                 $variants=array(),
  38                                                                 $variantfallbacks=array(),
  39                                                                 $markup=array(),
  40                                                                 $flags = array()) {
  41                 $this->mLangObj = $langobj;
  42                 $this->mMainLanguageCode = $maincode;
  43                 $this->mVariants = $variants;
  44                 $this->mVariantFallbacks = $variantfallbacks;
  45                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  46                 $m = array('begin'=>'-{', 'flagsep'=>'|', 'codesep'=>':',
  47                                    'varsep'=>';', 'end'=>'}-');
  48                 $this->mMarkup = array_merge($m, $markup);
  49                 $f = array('A'=>'A', 'T'=>'T', 'R' => 'R');
  50                 $this->mFlags = array_merge($f, $flags);
  51         }
  52
  53         /**
  54      * @access public
  55      */
  56         function getVariants() {
  57                 return $this->mVariants;
  58         }
  59
  60         /**
  61          * in case some variant is not defined in the markup, we need
  62          * to have some fallback. for example, in zh, normally people
  63          * will define zh-cn and zh-tw, but less so for zh-sg or zh-hk.
  64          * when zh-sg is preferred but not defined, we will pick zh-cn
  65          * in this case. right now this is only used by zh.
  66          *
  67          * @param string $v the language code of the variant
  68          * @return string the code of the fallback language or false if there is no fallback
  69      * @private
  70         */
  71         function getVariantFallback($v) {
  72                 return $this->mVariantFallbacks[$v];
  73         }
  74
  75
  76         /**
  77          * get preferred language variants.
  78          * @param boolean $fromUser Get it from $wgUser's preferences
  79      * @return string the preferred language code
  80      * @access public
  81         */
  82         function getPreferredVariant( $fromUser = true ) {
  83                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
  84
  85                 if($this->mPreferredVariant)
  86                         return $this->mPreferredVariant;
  87
  88                 // see if the preference is set in the request
  89                 $req = $wgRequest->getText( 'variant' );
  90                 if( in_array( $req, $this->mVariants ) ) {
  91                         $this->mPreferredVariant = $req;
  92                         return $req;
  93                 }
  94
  95                 // check the syntax /code/ArticleTitle
  96                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
  97                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
  98                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
  99                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 100                         if(in_array($scriptBase,$this->mVariants)){
 101                                 $this->mPreferredVariant = $scriptBase;
 102                                 return $this->mPreferredVariant;
 103                         }
 104                 }
 105
 106                 // get language variant preference from logged in users
 107                 // Don't call this on stub objects because that causes infinite
 108                 // recursion during initialisation
 109                 if( $fromUser && $wgUser->isLoggedIn() )  {
 110                         $this->mPreferredVariant = $wgUser->getOption('variant');
 111                         return $this->mPreferredVariant;
 112                 }
 113
 114                 // see if default variant is globaly set
 115                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 116                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 117                         return $this->mPreferredVariant;
 118                 }
 119
 120                 # FIXME rewrite code for parsing http header. The current code
 121                 # is written specific for detecting zh- variants
 122                 if( !$this->mPreferredVariant ) {
 123                         // see if some supported language variant is set in the
 124                         // http header, but we don't set the mPreferredVariant
 125                         // variable in case this is called before the user's
 126                         // preference is loaded
 127                         $pv=$this->mMainLanguageCode;
 128                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 129                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 130                                 $zh = strstr($header, $pv.'-');
 131                                 if($zh) {
 132                                         $pv = substr($zh,0,5);
 133                                 }
 134                         }
 135                         // don't try to return bad variant
 136                         if(in_array( $pv, $this->mVariants ))
 137                                 return $pv;
 138                 }
 139
 140                 return $this->mMainLanguageCode;
 141
 142         }
 143
 144         /**
 145      * dictionary-based conversion
 146      *
 147      * @param string $text the text to be converted
 148      * @param string $toVariant the target language code
 149      * @return string the converted text
 150      * @private
 151      */
 152         function autoConvert($text, $toVariant=false) {
 153                 $fname="LanguageConverter::autoConvert";
 154
 155                 wfProfileIn( $fname );
 156
 157                 if(!$this->mTablesLoaded)
 158                         $this->loadTables();
 159
 160                 if(!$toVariant)
 161                         $toVariant = $this->getPreferredVariant();
 162                 if(!in_array($toVariant, $this->mVariants))
 163                         return $text;
 164
 165                 /* we convert everything except:
 166                    1. html markups (anything between < and >)
 167                    2. html entities
 168                    3. place holders created by the parser
 169                 */
 170                 global $wgParser;
 171                 if (isset($wgParser))
 172                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 173                 else
 174                         $marker = "";
 175
 176                 // this one is needed when the text is inside an html markup
 177                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 178
 179                 // disable convert to variants between <code></code> tags
 180                 $codefix = '<code>.+?<\/code>|';
 181                 // disable convertsion of <script type="text/javascript"> ... </script>
 182                 $scriptfix = '<script.*?>.*?<\/script>|';
 183
 184                 $reg = '/'.$codefix . $scriptfix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 185
 186                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 187
 188                 $m = array_shift($matches);
 189
 190                 $ret = $this->translate($m[0], $toVariant);
 191                 $mstart = $m[1]+strlen($m[0]);
 192                 foreach($matches as $m) {
 193                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 194                         $ret .= $this->translate($m[0], $toVariant);
 195                         $mstart = $m[1] + strlen($m[0]);
 196                 }
 197                 wfProfileOut( $fname );
 198                 return $ret;
 199         }
 200
 201         /**
 202          * Translate a string to a variant
 203          * Doesn't process markup or do any of that other stuff, for that use convert()
 204          *
 205          * @param string $text Text to convert
 206          * @param string $variant Variant language code
 207          * @return string Translated text
 208          */
 209         function translate( $text, $variant ) {
 210                 wfProfileIn( __METHOD__ );
 211                 if( !$this->mTablesLoaded )
 212                         $this->loadTables();
 213                 $text = $this->mTables[$variant]->replace( $text );
 214                 wfProfileOut( __METHOD__ );
 215                 return $text;
 216         }
 217
 218         /**
 219      * convert text to all supported variants
 220      *
 221      * @param string $text the text to be converted
 222      * @return array of string
 223      * @public
 224      */
 225         function autoConvertToAllVariants($text) {
 226                 $fname="LanguageConverter::autoConvertToAllVariants";
 227                 wfProfileIn( $fname );
 228                 if( !$this->mTablesLoaded )
 229                         $this->loadTables();
 230
 231                 $ret = array();
 232                 foreach($this->mVariants as $variant) {
 233                         $ret[$variant] = $this->translate($text, $variant);
 234                 }
 235
 236                 wfProfileOut( $fname );
 237                 return $ret;
 238         }
 239
 240         /**
 241      * convert link text to all supported variants
 242      *
 243      * @param string $text the text to be converted
 244      * @return array of string
 245      * @public
 246      */
 247         function convertLinkToAllVariants($text) {
 248                 if( !$this->mTablesLoaded )
 249                         $this->loadTables();
 250
 251                 $ret = array();
 252                 $tarray = explode($this->mMarkup['begin'], $text);
 253                 $tfirst = array_shift($tarray);
 254
 255                 foreach($this->mVariants as $variant)
 256                         $ret[$variant] = $this->translate($tfirst,$variant);
 257
 258                 foreach($tarray as $txt) {
 259                         $marked = explode($this->mMarkup['end'], $txt, 2);
 260
 261                         foreach($this->mVariants as $variant){
 262                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 263                                 if(array_key_exists(1, $marked))
 264                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 265                         }
 266
 267                 }
 268
 269                 return $ret;
 270         }
 271
 272
 273         /**
 274          * Convert text using a parser object for context
 275          */
 276         function parserConvert( $text, &$parser ) {
 277                 global $wgDisableLangConversion;
 278                 /* don't do anything if this is the conversion table */
 279                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 280                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 281                 {
 282                         return $text;
 283                 }
 284
 285                 if($wgDisableLangConversion)
 286                         return $text;
 287
 288                 $text = $this->convert( $text );
 289                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 290                 return $text;
 291         }
 292
 293         /**
 294          *  Parse flags with syntax -{FLAG| ... }-
 295          *
 296          */
 297         function parseFlags($marked){
 298                         $flags = array();
 299
 300                         // process flag only if the flag is valid
 301                         if(strlen($marked) < 2 || !(in_array($marked[0],$this->mFlags) && $marked[1]=='|' ) )
 302                                 return array($marked,array());
 303
 304                         $tt = explode($this->mMarkup['flagsep'], $marked, 2);
 305
 306                         if(sizeof($tt) == 2) {
 307                                 $f = explode($this->mMarkup['varsep'], $tt[0]);
 308                                 foreach($f as $ff) {
 309                                         $ff = trim($ff);
 310                                         if(array_key_exists($ff, $this->mFlags) &&
 311                                                 !array_key_exists($this->mFlags[$ff], $flags))
 312                                                 $flags[] = $this->mFlags[$ff];
 313                                 }
 314                                 $rules = $tt[1];
 315                         }
 316                         else
 317                                 $rules = $marked;
 318
 319                         if( !in_array('R',$flags) ){
 320                                 //FIXME: may cause trouble here...
 321                                 //strip &nbsp; since it interferes with the parsing, plus,
 322                                 //all spaces should be stripped in this tag anyway.
 323                                 $rules = str_replace('&nbsp;', '', $rules);
 324                         }
 325
 326                         return array($rules,$flags);
 327         }
 328
 329         /**
 330          * convert text to different variants of a language. the automatic
 331          * conversion is done in autoConvert(). here we parse the text
 332          * marked with -{}-, which specifies special conversions of the
 333          * text that can not be accomplished in autoConvert()
 334          *
 335          * syntax of the markup:
 336          * -{code1:text1;code2:text2;...}-  or
 337          * -{text}- in which case no conversion should take place for text
 338      *
 339      * @param string $text text to be converted
 340      * @param bool $isTitle whether this conversion is for the article title
 341      * @return string converted text
 342      * @access public
 343      */
 344         function convert( $text , $isTitle=false) {
 345                 $mw =& MagicWord::get( 'notitleconvert'   );
 346                 if( $mw->matchAndRemove( $text ) )
 347                         $this->mDoTitleConvert = false;
 348
 349                 $mw =& MagicWord::get( 'nocontentconvert'   );
 350                 if( $mw->matchAndRemove( $text ) ) {
 351                         $this->mDoContentConvert = false;
 352                 }
 353
 354                 // no conversion if redirecting
 355                 $mw =& MagicWord::get( 'redirect'   );
 356                 if( $mw->matchStart( $text ))
 357                         return $text;
 358
 359                 if( $isTitle ) {
 360
 361                         // use the title from the T flag if any
 362                         if($this->mTitleFromFlag){
 363                                 $this->mTitleFromFlag = false;
 364                                 return $this->mTitleDisplay;
 365                         }
 366
 367                         // check for __NOTC__ tag
 368                         if( !$this->mDoTitleConvert ) {
 369                                 $this->mTitleDisplay = $text;
 370                                 return $text;
 371                         }
 372
 373                         global $wgRequest;
 374                         $isredir = $wgRequest->getText( 'redirect', 'yes' );
 375                         $action = $wgRequest->getText( 'action' );
 376                         if ( $isredir == 'no' || $action == 'edit' ) {
 377                                 return $text;
 378                         }
 379                         else {
 380                                 $this->mTitleDisplay = $this->convert($text);
 381                                 return $this->mTitleDisplay;
 382                         }
 383                 }
 384
 385                 $plang = $this->getPreferredVariant();
 386                 if( isset( $this->mVariantFallbacks[$plang] ) ) {
 387                         $fallback = $this->mVariantFallbacks[$plang];
 388                 } else {
 389                         $fallback = $this->mMainLanguageCode;
 390                 }
 391
 392                 $tarray = explode($this->mMarkup['begin'], $text);
 393                 $tfirst = array_shift($tarray);
 394                 if($this->mDoContentConvert)
 395                         $text = $this->autoConvert($tfirst);
 396                 else
 397                         $text = $tfirst;
 398                 foreach($tarray as $txt) {
 399                         $marked = explode($this->mMarkup['end'], $txt, 2);
 400
 401                         // strip the flags from syntax like -{T| ... }-
 402                         list($rules,$flags) = $this->parseFlags($marked[0]);
 403
 404                         // proces R flag: output raw content of -{ ... }-
 405                         if( in_array('R',$flags) ){
 406                                 $disp = $rules;
 407                         } else if( $this->mDoContentConvert){
 408                                 // parse the contents -{ ... }-
 409                                 $carray = $this->parseManualRule($rules, $flags);
 410
 411                                 $disp = '';
 412                                 if(array_key_exists($plang, $carray)) {
 413                                         $disp = $carray[$plang];
 414                                 } else if(array_key_exists($fallback, $carray)) {
 415                                         $disp = $carray[$fallback];
 416                                 }
 417                         } else{
 418                                 // if we don't do content convert, still strip the -{}- tags
 419                                 $disp = $rules;
 420                                 $flags = array();
 421                         }
 422
 423                         if($disp) {
 424                                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 425                                 if(in_array('T',  $flags)){
 426                                         $this->mTitleFromFlag = true;
 427                                         $this->mTitleDisplay = $disp;
 428                                 }
 429                                 else
 430                                         $text .= $disp;
 431
 432                                 // use syntax -{A|zh:WordZh;zh-tw:WordTw}- to introduce a custom mapping between
 433                                 // words WordZh and WordTw in the whole text
 434                                 if(in_array('A', $flags)) {
 435
 436                                         /* fill in the missing variants, if any,
 437                                             with fallbacks */
 438                                         foreach($this->mVariants as $v) {
 439                                                 if(!array_key_exists($v, $carray)) {
 440                                                         $vf = $this->getVariantFallback($v);
 441                                                         if(array_key_exists($vf, $carray))
 442                                                                 $carray[$v] = $carray[$vf];
 443                                                 }
 444                                         }
 445
 446                                         foreach($this->mVariants as $vfrom) {
 447                                                 if(!array_key_exists($vfrom, $carray))
 448                                                         continue;
 449                                                 foreach($this->mVariants as $vto) {
 450                                                         if($vfrom == $vto)
 451                                                                 continue;
 452                                                         if(!array_key_exists($vto, $carray))
 453                                                                 continue;
 454                                                         $this->mTables[$vto]->setPair($carray[$vfrom], $carray[$vto]);
 455                                                 }
 456                                         }
 457                                 }
 458                         }
 459                         else {
 460                                 $text .= $marked[0];
 461                         }
 462                         if(array_key_exists(1, $marked)){
 463                                 if( $this->mDoContentConvert )
 464                                         $text .= $this->autoConvert($marked[1]);
 465                                 else
 466                                         $text .= $marked[1];
 467                         }
 468                 }
 469
 470                 return $text;
 471         }
 472
 473         /**
 474          * parse the manually marked conversion rule
 475          * @param string $rule the text of the rule
 476          * @return array of the translation in each variant
 477          * @private
 478          */
 479         function parseManualRule($rules, $flags=array()) {
 480
 481                 $choice = explode($this->mMarkup['varsep'], $rules);
 482                 $carray = array();
 483                 if(sizeof($choice) == 1) {
 484                         /* a single choice */
 485                         foreach($this->mVariants as $v)
 486                                 $carray[$v] = $choice[0];
 487                 }
 488                 else {
 489                         foreach($choice as $c) {
 490                                 $v = explode($this->mMarkup['codesep'], $c);
 491                                 if(sizeof($v) != 2) // syntax error, skip
 492                                         continue;
 493                                 $carray[trim($v[0])] = trim($v[1]);
 494                         }
 495                 }
 496                 return $carray;
 497         }
 498
 499         /**
 500          * if a language supports multiple variants, it is
 501          * possible that non-existing link in one variant
 502          * actually exists in another variant. this function
 503          * tries to find it. See e.g. LanguageZh.php
 504          *
 505          * @param string $link the name of the link
 506          * @param mixed $nt the title object of the link
 507          * @return null the input parameters may be modified upon return
 508      * @access public
 509          */
 510         function findVariantLink( &$link, &$nt ) {
 511                 global $wgDisableLangConversion;
 512                 $linkBatch = new LinkBatch();
 513
 514                 $ns=NS_MAIN;
 515
 516                 if(is_object($nt))
 517                         $ns = $nt->getNamespace();
 518
 519                 $variants = $this->autoConvertToAllVariants($link);
 520                 if($variants == false) //give up
 521                         return;
 522
 523                 $titles = array();
 524
 525                 foreach( $variants as $v ) {
 526                         if($v != $link){
 527                                 $varnt = Title::newFromText( $v, $ns );
 528                                 if(!is_null($varnt)){
 529                                         $linkBatch->addObj($varnt);
 530                                         $titles[]=$varnt;
 531                                 }
 532                         }
 533                 }
 534
 535                 // fetch all variants in single query
 536                 $linkBatch->execute();
 537
 538                 foreach( $titles as $varnt ) {
 539                         if( $varnt->getArticleID() > 0 ) {
 540                                 $nt = $varnt;
 541                                 if( !$wgDisableLangConversion )
 542                                         $link = $v;
 543                                 break;
 544                         }
 545                 }
 546         }
 547
 548     /**
 549      * returns language specific hash options
 550      *
 551      * @access public
 552      */
 553         function getExtraHashOptions() {
 554                 $variant = $this->getPreferredVariant();
 555                 return '!' . $variant ;
 556         }
 557
 558     /**
 559      * get title text as defined in the body of the article text
 560      *
 561      * @access public
 562      */
 563         function getParsedTitle() {
 564                 return $this->mTitleDisplay;
 565         }
 566
 567         /**
 568      * a write lock to the cache
 569      *
 570      * @private
 571      */
 572         function lockCache() {
 573                 global $wgMemc;
 574                 $success = false;
 575                 for($i=0; $i<30; $i++) {
 576                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 577                                 break;
 578                         sleep(1);
 579                 }
 580                 return $success;
 581         }
 582
 583         /**
 584      * unlock cache
 585      *
 586      * @private
 587      */
 588         function unlockCache() {
 589                 global $wgMemc;
 590                 $wgMemc->delete($this->mCacheKey . "lock");
 591         }
 592
 593
 594         /**
 595      * Load default conversion tables
 596      * This method must be implemented in derived class
 597      *
 598      * @private
 599      */
 600         function loadDefaultTables() {
 601                 $name = get_class($this);
 602                 wfDie("Must implement loadDefaultTables() method in class $name");
 603         }
 604
 605         /**
 606      * load conversion tables either from the cache or the disk
 607      * @private
 608      */
 609         function loadTables($fromcache=true) {
 610                 global $wgMemc;
 611                 if( $this->mTablesLoaded )
 612                         return;
 613                 wfProfileIn( __METHOD__ );
 614                 $this->mTablesLoaded = true;
 615                 $this->mTables = false;
 616                 if($fromcache) {
 617                         wfProfileIn( __METHOD__.'-cache' );
 618                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 619                         wfProfileOut( __METHOD__.'-cache' );
 620                 }
 621                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 622                         wfProfileIn( __METHOD__.'-recache' );
 623                         // not in cache, or we need a fresh reload.
 624                         // we will first load the default tables
 625                         // then update them using things in MediaWiki:Zhconversiontable/*
 626                         $this->loadDefaultTables();
 627                         foreach($this->mVariants as $var) {
 628                                 $cached = $this->parseCachedTable($var);
 629                                 $this->mTables[$var]->mergeArray($cached);
 630                         }
 631
 632                         $this->postLoadTables();
 633                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 634
 635                         if($this->lockCache()) {
 636                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 637                                 $this->unlockCache();
 638                         }
 639                         wfProfileOut( __METHOD__.'-recache' );
 640                 }
 641                 wfProfileOut( __METHOD__ );
 642         }
 643
 644     /**
 645      * Hook for post processig after conversion tables are loaded
 646      *
 647      */
 648         function postLoadTables() {}
 649
 650     /**
 651      * Reload the conversion tables
 652      *
 653      * @private
 654      */
 655         function reloadTables() {
 656                 if($this->mTables)
 657                         unset($this->mTables);
 658                 $this->mTablesLoaded = false;
 659                 $this->loadTables(false);
 660         }
 661
 662
 663         /**
 664      * parse the conversion table stored in the cache
 665      *
 666      * the tables should be in blocks of the following form:
 667
 668      *          -{
 669      *                  word => word ;
 670      *                  word => word ;
 671      *                  ...
 672      *          }-
 673      *
 674      *  to make the tables more manageable, subpages are allowed
 675      *  and will be parsed recursively if $recursive=true
 676      *
 677      * @private
 678          */
 679         function parseCachedTable($code, $subpage='', $recursive=true) {
 680                 global $wgMessageCache;
 681                 static $parsed = array();
 682
 683                 if(!is_object($wgMessageCache))
 684                         return array();
 685
 686                 $key = 'Conversiontable/'.$code;
 687                 if($subpage)
 688                         $key .= '/' . $subpage;
 689
 690                 if(array_key_exists($key, $parsed))
 691                         return array();
 692
 693
 694                 $txt = $wgMessageCache->get( $key, true, true, true );
 695
 696                 // get all subpage links of the form
 697                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 698                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 699                 $subs = explode('[[', $txt);
 700                 $sublinks = array();
 701                 foreach( $subs as $sub ) {
 702                         $link = explode(']]', $sub, 2);
 703                         if(count($link) != 2)
 704                                 continue;
 705                         $b = explode('|', $link[0]);
 706                         $b = explode('/', trim($b[0]), 3);
 707                         if(count($b)==3)
 708                                 $sublink = $b[2];
 709                         else
 710                                 $sublink = '';
 711
 712                         if($b[0] == $linkhead && $b[1] == $code) {
 713                                 $sublinks[] = $sublink;
 714                         }
 715                 }
 716
 717
 718                 // parse the mappings in this page
 719                 $blocks = explode($this->mMarkup['begin'], $txt);
 720                 array_shift($blocks);
 721                 $ret = array();
 722                 foreach($blocks as $block) {
 723                         $mappings = explode($this->mMarkup['end'], $block, 2);
 724                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 725                         $table = explode( ';', $stripped );
 726                         foreach( $table as $t ) {
 727                                 $m = explode( '=>', $t );
 728                                 if( count( $m ) != 2)
 729                                         continue;
 730                                 // trim any trailling comments starting with '//'
 731                                 $tt = explode('//', $m[1], 2);
 732                                 $ret[trim($m[0])] = trim($tt[0]);
 733                         }
 734                 }
 735                 $parsed[$key] = true;
 736
 737
 738                 // recursively parse the subpages
 739                 if($recursive) {
 740                         foreach($sublinks as $link) {
 741                                 $s = $this->parseCachedTable($code, $link, $recursive);
 742                                 $ret = array_merge($ret, $s);
 743                         }
 744                 }
 745
 746                 if ($this->mUcfirst) {
 747                         foreach ($ret as $k => $v) {
 748                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 749                         }
 750                 }
 751                 return $ret;
 752         }
 753
 754         /**
 755          * Enclose a string with the "no conversion" tag. This is used by
 756          * various functions in the Parser
 757          *
 758          * @param string $text text to be tagged for no conversion
 759          * @return string the tagged text
 760         */
 761         function markNoConversion($text, $noParse=false) {
 762                 # don't mark if already marked
 763                 if(strpos($text, $this->mMarkup['begin']) ||
 764                    strpos($text, $this->mMarkup['end']))
 765                         return $text;
 766
 767                 $ret = $this->mMarkup['begin'] . $text . $this->mMarkup['end'];
 768                 return $ret;
 769         }
 770
 771         /**
 772          * convert the sorting key for category links. this should make different
 773          * keys that are variants of each other map to the same key
 774         */
 775         function convertCategoryKey( $key ) {
 776                 return $key;
 777         }
 778         /**
 779      * hook to refresh the cache of conversion tables when
 780      * MediaWiki:conversiontable* is updated
 781      * @private
 782         */
 783         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 784                 $titleobj = $article->getTitle();
 785                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 786             /*
 787                         global $wgContLang; // should be an LanguageZh.
 788                         if(get_class($wgContLang) != 'languagezh')
 789                                 return true;
 790             */
 791                         $title = $titleobj->getDBkey();
 792                         $t = explode('/', $title, 3);
 793                         $c = count($t);
 794                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 795                                 if(in_array($t[1], $this->mVariants)) {
 796                                         $this->reloadTables();
 797                                 }
 798                         }
 799                 }
 800                 return true;
 801         }
 802
 803         /**
 804          * Armour rendered math against conversion
 805          * Wrap math into rawoutput -{R| math }- syntax
 806          */
 807         function armourMath($text){
 808                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 809                 return $ret;
 810         }
 811
 812
 813 }