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