languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * base class for language convert
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mLanguageNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mManualAddTables;
  25         var $mManualRemoveTables;
  26         var $mTitleDisplay='';
  27         var $mDoTitleConvert=true, $mDoContentConvert=true;
  28         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  29         var $mTitleFromFlag = false;
  30         var $mCacheKey;
  31         var $mLangObj;
  32         var $mMarkup;
  33         var $mFlags;
  34         var $mDescCodeSep = ':',$mDescVarSep = ';';
  35         var $mUcfirst = false;
  36
  37         const CACHE_VERSION_KEY = 'VERSION 6';
  38
  39         /**
  40          * Constructor
  41          *
  42          * @param string $maincode the main language code of this language
  43          * @param array $variants the supported variants of this language
  44          * @param array $variantfallback the fallback language of each variant
  45          * @param array $markup array defining the markup used for manual conversion
  46          * @param array $flags array defining the custom strings that maps to the flags
  47          * @param array $manualLevel limit for supported variants
  48          * @public
  49          */
  50         function __construct($langobj, $maincode,
  51                                                                 $variants=array(),
  52                                                                 $variantfallbacks=array(),
  53                                                                 $markup=array(),
  54                                                                 $flags = array(),
  55                                                                 $manualLevel = array() ) {
  56                 $this->mLangObj = $langobj;
  57                 $this->mMainLanguageCode = $maincode;
  58                 $this->mVariants = $variants;
  59                 $this->mVariantFallbacks = $variantfallbacks;
  60                 global $wgLanguageNames;
  61                 $this->mLanguageNames = $wgLanguageNames;
  62                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  63                 $m = array(
  64                         'begin'=>'-{',
  65                         'flagsep'=>'|',
  66                         'unidsep'=>'=>', //for unidirectional conversion
  67                         'codesep'=>':',
  68                         'varsep'=>';',
  69                         'end'=>'}-'
  70                 );
  71                 $this->mMarkup = array_merge($m, $markup);
  72                 $f = array(
  73                         // 'S' show converted text
  74                         // '+' add rules for alltext
  75                         // 'E' the gave flags is error
  76                         // these flags above are reserved for program
  77                         'A'=>'A',       // add rule for convert code (all text convert)
  78                         'T'=>'T',       // title convert
  79                         'R'=>'R',       // raw content
  80                         'D'=>'D',       // convert description (subclass implement)
  81                         '-'=>'-',       // remove convert (not implement)
  82                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  83                         'N'=>'N'        // current variant name
  84                 );
  85                 $this->mFlags = array_merge($f, $flags);
  86                 foreach( $this->mVariants as $v) {
  87                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  88                                                                 ?$manualLevel[$v]
  89                                                                 :'bidirectional';
  90                         $this->mManualAddTables[$v] = array();
  91                         $this->mManualRemoveTables[$v] = array();
  92                 }
  93         }
  94
  95         /**
  96          * @public
  97          */
  98         function getVariants() {
  99                 return $this->mVariants;
 100         }
 101
 102         /**
 103          * in case some variant is not defined in the markup, we need
 104          * to have some fallback. for example, in zh, normally people
 105          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 106          * when zh-sg is preferred but not defined, we will pick zh-hans
 107          * in this case. right now this is only used by zh.
 108          *
 109          * @param string $v the language code of the variant
 110          * @return string array the code of the fallback language or false if there is no fallback
 111          * @public
 112          */
 113         function getVariantFallbacks($v) {
 114                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 115                         return $this->mVariantFallbacks[$v];
 116                 }
 117                 return $this->mMainLanguageCode;
 118         }
 119
 120         // this method body in Language class
 121         function getPreferredVariant( $fromUser = true ) {
 122                 return $this->mLangObj->getPreferredVariant( $fromUser );
 123         }
 124
 125         /**
 126          * caption convert, base on preg_replace_callback
 127          *
 128          * to convert text in "title" or "alt", like '<img alt="text" ... '
 129          * or '<span title="text" ... '
 130          *
 131          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 132          * @private
 133          */
 134         function captionConvert( $matches ) {
 135                 $toVariant = $this->getPreferredVariant();
 136                 $title = $matches[1];
 137                 $text  = $matches[2];
 138                 // we convert captions except URL
 139                 if( !strpos( $text, '://' ) )
 140                         $text = $this->translate($text, $toVariant);
 141                 return " $title=\"$text\"";
 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) && $wgParser->UniqPrefix()!=''){
 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                 // disable conversion of <pre xxxx> ... </pre>
 184                 $prefix = '<pre.*?>.*?<\/pre>|';
 185
 186                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 187
 188                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 189
 190                 $m = array_shift($matches);
 191
 192                 $ret = $this->translate($m[0], $toVariant);
 193                 $mstart = $m[1]+strlen($m[0]);
 194
 195                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 196                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 197                 foreach($matches as $m) {
 198                         $mark = substr($text, $mstart, $m[1]-$mstart);
 199                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 200                         $ret .= $mark;
 201                         $ret .= $this->translate($m[0], $toVariant);
 202                         $mstart = $m[1] + strlen($m[0]);
 203                 }
 204                 wfProfileOut( $fname );
 205                 return $ret;
 206         }
 207
 208         /**
 209          * Translate a string to a variant
 210          * Doesn't process markup or do any of that other stuff, for that use convert()
 211          *
 212          * @param string $text Text to convert
 213          * @param string $variant Variant language code
 214          * @return string Translated text
 215          * @private
 216          */
 217         function translate( $text, $variant ) {
 218                 wfProfileIn( __METHOD__ );
 219                 if( !$this->mTablesLoaded )
 220                         $this->loadTables();
 221                 $text = $this->mTables[$variant]->replace( $text );
 222                 wfProfileOut( __METHOD__ );
 223                 return $text;
 224         }
 225
 226         /**
 227          * convert text to all supported variants
 228          *
 229          * @param string $text the text to be converted
 230          * @return array of string
 231          * @public
 232          */
 233         function autoConvertToAllVariants($text) {
 234                 $fname="LanguageConverter::autoConvertToAllVariants";
 235                 wfProfileIn( $fname );
 236                 if( !$this->mTablesLoaded )
 237                         $this->loadTables();
 238
 239                 $ret = array();
 240                 foreach($this->mVariants as $variant) {
 241                         $ret[$variant] = $this->translate($text, $variant);
 242                 }
 243
 244                 wfProfileOut( $fname );
 245                 return $ret;
 246         }
 247
 248         /**
 249          * convert link text to all supported variants
 250          *
 251          * @param string $text the text to be converted
 252          * @return array of string
 253          * @public
 254          */
 255         function convertLinkToAllVariants($text) {
 256                 if( !$this->mTablesLoaded )
 257                         $this->loadTables();
 258
 259                 $ret = array();
 260                 $tarray = explode($this->mMarkup['begin'], $text);
 261                 $tfirst = array_shift($tarray);
 262
 263                 foreach($this->mVariants as $variant)
 264                         $ret[$variant] = $this->translate($tfirst,$variant);
 265
 266                 foreach($tarray as $txt) {
 267                         $marked = explode($this->mMarkup['end'], $txt, 2);
 268
 269                         foreach($this->mVariants as $variant){
 270                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 271                                 if(array_key_exists(1, $marked))
 272                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 273                         }
 274
 275                 }
 276
 277                 return $ret;
 278         }
 279
 280         /**
 281          * prepare manual conversion table
 282          * @private
 283          */
 284         function prepareManualConv($convRule){
 285                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 286                 $title = $convRule->getTitle();
 287                 if($title){
 288                         $this->mTitleFromFlag = true;
 289                         $this->mTitleDisplay =  $title;
 290                 }
 291
 292                 //apply manual conversion table to global table
 293                 $convTable = $convRule->getConvTable();
 294                 $action = $convRule->getRulesAction();
 295                 foreach($convTable as $v=>$t) {
 296                         if( !in_array($v,$this->mVariants) )continue;
 297                         if( $action=="add" ) {
 298                                 foreach($t as $from=>$to) {
 299                                         // more efficient than array_merge(), about 2.5 times.
 300                                         $this->mManualAddTables[$v][$from] = $to;
 301                                 }
 302                         }
 303                         elseif ( $action=="remove" )
 304                                 $this->mManualRemoveTables[$v] = array_merge($this->mManualRemoveTables[$v], $t);
 305                 }
 306         }
 307
 308         /**
 309          * apply manual conversion from $this->mManualAddTables and $this->mManualRemoveTables
 310          * @private
 311          */
 312         function applyManualConv(){
 313                 //apply manual conversion table to global table
 314                 foreach($this->mVariants as $v) {
 315                         if (count($this->mManualAddTables[$v]) > 0) {
 316                                 $this->mTables[$v]->mergeArray($this->mManualAddTables[$v]);
 317                         }
 318                         if (count($this->mManualRemoveTables[$v]) > 0)
 319                                 $this->mTables[$v]->removeArray($this->mManualRemoveTables[$v]);
 320                 }
 321         }
 322
 323         /**
 324          * Convert text using a parser object for context
 325          * @public
 326          */
 327         function parserConvert( $text, &$parser ) {
 328                 global $wgDisableLangConversion;
 329                 /* don't do anything if this is the conversion table */
 330                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 331                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 332                 {
 333                         return $text;
 334                 }
 335
 336                 if($wgDisableLangConversion)
 337                         return $text;
 338
 339                 $text = $this->convert( $text );
 340                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 341                 return $text;
 342         }
 343
 344         /**
 345          *  convert title
 346          * @private
 347          */
 348         function convertTitle($text){
 349                 global $wgDisableTitleConversion, $wgUser;
 350
 351                 // check for global param and __NOTC__ tag
 352                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 353                         $this->mTitleDisplay = $text;
 354                         return $text;
 355                 }
 356
 357                 // use the title from the T flag if any
 358                 if($this->mTitleFromFlag){
 359                         $this->mTitleFromFlag = false;
 360                         return $this->mTitleDisplay;
 361                 }
 362
 363                 global $wgRequest;
 364                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 365                 $action = $wgRequest->getText( 'action' );
 366                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 367                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 368                         return $text;
 369                 } else {
 370                         $this->mTitleDisplay = $this->convert($text);
 371                         return $this->mTitleDisplay;
 372                 }
 373         }
 374
 375         /**
 376          * convert text to different variants of a language. the automatic
 377          * conversion is done in autoConvert(). here we parse the text
 378          * marked with -{}-, which specifies special conversions of the
 379          * text that can not be accomplished in autoConvert()
 380          *
 381          * syntax of the markup:
 382          * -{code1:text1;code2:text2;...}-  or
 383          * -{flags|code1:text1;code2:text2;...}-  or
 384          * -{text}- in which case no conversion should take place for text
 385          *
 386          * @param string $text text to be converted
 387          * @param bool $isTitle whether this conversion is for the article title
 388          * @return string converted text
 389          * @public
 390          */
 391         function convert( $text , $isTitle=false) {
 392
 393                 $mw =& MagicWord::get( 'notitleconvert' );
 394                 if( $mw->matchAndRemove( $text ) )
 395                         $this->mDoTitleConvert = false;
 396                 $mw =& MagicWord::get( 'nocontentconvert' );
 397                 if( $mw->matchAndRemove( $text ) ) {
 398                         $this->mDoContentConvert = false;
 399                 }
 400
 401                 // no conversion if redirecting
 402                 $mw =& MagicWord::get( 'redirect' );
 403                 if( $mw->matchStart( $text ))
 404                         return $text;
 405
 406                 // for title convertion
 407                 if ($isTitle) return $this->convertTitle($text);
 408
 409                 $plang = $this->getPreferredVariant();
 410                 $tarray = StringUtils::explode($this->mMarkup['end'], $text);
 411                 $text = '';
 412
 413                 $marks = array();
 414                 foreach($tarray as $txt) {
 415                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 416                         if (array_key_exists(1, $marked)) {
 417                                 $crule = new ConverterRule($marked[1], $this);
 418                                 $crule->parse($plang);
 419                                 $marked[1] = $crule->getDisplay();
 420                                 $this->prepareManualConv($crule);
 421                         }
 422                         else
 423                                 $marked[0] .= $this->mMarkup['end'];
 424                         array_push($marks, $marked);
 425                 }
 426                 $this->applyManualConv();
 427                 foreach ($marks as $marked) {
 428                         if( $this->mDoContentConvert )
 429                                 $text .= $this->autoConvert($marked[0],$plang);
 430                         else
 431                                 $text .= $marked[0];
 432                         if( array_key_exists(1, $marked) )
 433                                 $text .= $marked[1];
 434                 }
 435                 // Remove the last delimiter (wasn't real)
 436                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 437
 438                 return $text;
 439         }
 440
 441         /**
 442          * if a language supports multiple variants, it is
 443          * possible that non-existing link in one variant
 444          * actually exists in another variant. this function
 445          * tries to find it. See e.g. LanguageZh.php
 446          *
 447          * @param string $link the name of the link
 448          * @param mixed $nt the title object of the link
 449          * @return null the input parameters may be modified upon return
 450          * @public
 451          */
 452         function findVariantLink( &$link, &$nt, $forTemplate = false ) {
 453                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 454                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 455                 $action = $wgRequest->getText( 'action' );
 456                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 457                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 458                 $linkBatch = new LinkBatch();
 459
 460                 $ns=NS_MAIN;
 461
 462                 if ( $disableLinkConversion || ( !$forTemplate && ( $isredir == 'no' || $action == 'edit'
 463                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) ) {
 464                         return;
 465                 }
 466
 467                 if(is_object($nt))
 468                         $ns = $nt->getNamespace();
 469
 470                 $variants = $this->autoConvertToAllVariants($link);
 471                 if($variants == false) //give up
 472                         return;
 473
 474                 $titles = array();
 475
 476                 foreach( $variants as $v ) {
 477                         if($v != $link){
 478                                 $varnt = Title::newFromText( $v, $ns );
 479                                 if(!is_null($varnt)){
 480                                         $linkBatch->addObj($varnt);
 481                                         $titles[]=$varnt;
 482                                 }
 483                         }
 484                 }
 485
 486                 // fetch all variants in single query
 487                 $linkBatch->execute();
 488
 489                 foreach( $titles as $varnt ) {
 490                         if( $varnt->getArticleID() > 0 ) {
 491                                 $nt = $varnt;
 492                                 $link = $v;
 493                                 break;
 494                         }
 495                 }
 496         }
 497
 498     /**
 499          * returns language specific hash options
 500          *
 501          * @public
 502          */
 503         function getExtraHashOptions() {
 504                 $variant = $this->getPreferredVariant();
 505                 return '!' . $variant ;
 506         }
 507
 508     /**
 509          * get title text as defined in the body of the article text
 510          *
 511          * @public
 512          */
 513         function getParsedTitle() {
 514                 return $this->mTitleDisplay;
 515         }
 516
 517         /**
 518          * a write lock to the cache
 519          *
 520          * @private
 521          */
 522         function lockCache() {
 523                 global $wgMemc;
 524                 $success = false;
 525                 for($i=0; $i<30; $i++) {
 526                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 527                                 break;
 528                         sleep(1);
 529                 }
 530                 return $success;
 531         }
 532
 533         /**
 534          * unlock cache
 535          *
 536          * @private
 537          */
 538         function unlockCache() {
 539                 global $wgMemc;
 540                 $wgMemc->delete($this->mCacheKey . "lock");
 541         }
 542
 543
 544         /**
 545          * Load default conversion tables
 546          * This method must be implemented in derived class
 547          *
 548          * @private
 549          */
 550         function loadDefaultTables() {
 551                 $name = get_class($this);
 552                 wfDie("Must implement loadDefaultTables() method in class $name");
 553         }
 554
 555         /**
 556          * load conversion tables either from the cache or the disk
 557          * @private
 558          */
 559         function loadTables($fromcache=true) {
 560                 global $wgMemc;
 561                 if( $this->mTablesLoaded )
 562                         return;
 563                 wfProfileIn( __METHOD__ );
 564                 $this->mTablesLoaded = true;
 565                 $this->mTables = false;
 566                 if($fromcache) {
 567                         wfProfileIn( __METHOD__.'-cache' );
 568                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 569                         wfProfileOut( __METHOD__.'-cache' );
 570                 }
 571                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 572                         wfProfileIn( __METHOD__.'-recache' );
 573                         // not in cache, or we need a fresh reload.
 574                         // we will first load the default tables
 575                         // then update them using things in MediaWiki:Zhconversiontable/*
 576                         $this->loadDefaultTables();
 577                         foreach($this->mVariants as $var) {
 578                                 $cached = $this->parseCachedTable($var);
 579                                 $this->mTables[$var]->mergeArray($cached);
 580                         }
 581
 582                         $this->postLoadTables();
 583                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 584
 585                         if($this->lockCache()) {
 586                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 587                                 $this->unlockCache();
 588                         }
 589                         wfProfileOut( __METHOD__.'-recache' );
 590                 }
 591                 wfProfileOut( __METHOD__ );
 592         }
 593
 594     /**
 595          * Hook for post processig after conversion tables are loaded
 596          *
 597          */
 598         function postLoadTables() {}
 599
 600     /**
 601          * Reload the conversion tables
 602          *
 603          * @private
 604          */
 605         function reloadTables() {
 606                 if($this->mTables)
 607                         unset($this->mTables);
 608                 $this->mTablesLoaded = false;
 609                 $this->loadTables(false);
 610         }
 611
 612
 613         /**
 614          * parse the conversion table stored in the cache
 615          *
 616          * the tables should be in blocks of the following form:
 617          *              -{
 618          *                      word => word ;
 619          *                      word => word ;
 620          *                      ...
 621          *              }-
 622          *
 623          *      to make the tables more manageable, subpages are allowed
 624          *      and will be parsed recursively if $recursive=true
 625          *
 626          */
 627         function parseCachedTable($code, $subpage='', $recursive=true) {
 628                 global $wgMessageCache;
 629                 static $parsed = array();
 630
 631                 if(!is_object($wgMessageCache))
 632                         return array();
 633
 634                 $key = 'Conversiontable/'.$code;
 635                 if($subpage)
 636                         $key .= '/' . $subpage;
 637
 638                 if(array_key_exists($key, $parsed))
 639                         return array();
 640
 641                 if ( strpos( $code, '/' ) === false ) {
 642                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 643                 } else {
 644                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 645                         if ( $title && $title->exists() ) {
 646                                 $article = new Article( $title );
 647                                 $txt = $article->getContents();
 648                         } else {
 649                                 $txt = '';
 650                         }
 651                 }
 652
 653                 // get all subpage links of the form
 654                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 655                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 656                 $subs = explode('[[', $txt);
 657                 $sublinks = array();
 658                 foreach( $subs as $sub ) {
 659                         $link = explode(']]', $sub, 2);
 660                         if(count($link) != 2)
 661                                 continue;
 662                         $b = explode('|', $link[0]);
 663                         $b = explode('/', trim($b[0]), 3);
 664                         if(count($b)==3)
 665                                 $sublink = $b[2];
 666                         else
 667                                 $sublink = '';
 668
 669                         if($b[0] == $linkhead && $b[1] == $code) {
 670                                 $sublinks[] = $sublink;
 671                         }
 672                 }
 673
 674
 675                 // parse the mappings in this page
 676                 $blocks = explode($this->mMarkup['begin'], $txt);
 677                 array_shift($blocks);
 678                 $ret = array();
 679                 foreach($blocks as $block) {
 680                         $mappings = explode($this->mMarkup['end'], $block, 2);
 681                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 682                         $table = explode( ';', $stripped );
 683                         foreach( $table as $t ) {
 684                                 $m = explode( '=>', $t );
 685                                 if( count( $m ) != 2)
 686                                         continue;
 687                                 // trim any trailling comments starting with '//'
 688                                 $tt = explode('//', $m[1], 2);
 689                                 $ret[trim($m[0])] = trim($tt[0]);
 690                         }
 691                 }
 692                 $parsed[$key] = true;
 693
 694
 695                 // recursively parse the subpages
 696                 if($recursive) {
 697                         foreach($sublinks as $link) {
 698                                 $s = $this->parseCachedTable($code, $link, $recursive);
 699                                 $ret = array_merge($ret, $s);
 700                         }
 701                 }
 702
 703                 if ($this->mUcfirst) {
 704                         foreach ($ret as $k => $v) {
 705                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 706                         }
 707                 }
 708                 return $ret;
 709         }
 710
 711         /**
 712          * Enclose a string with the "no conversion" tag. This is used by
 713          * various functions in the Parser
 714          *
 715          * @param string $text text to be tagged for no conversion
 716          * @return string the tagged text
 717          * @public
 718          */
 719         function markNoConversion($text, $noParse=false) {
 720                 # don't mark if already marked
 721                 if(strpos($text, $this->mMarkup['begin']) ||
 722                    strpos($text, $this->mMarkup['end']))
 723                         return $text;
 724
 725                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 726                 return $ret;
 727         }
 728
 729         /**
 730          * convert the sorting key for category links. this should make different
 731          * keys that are variants of each other map to the same key
 732          */
 733         function convertCategoryKey( $key ) {
 734                 return $key;
 735         }
 736         /**
 737          * hook to refresh the cache of conversion tables when
 738          * MediaWiki:conversiontable* is updated
 739          * @private
 740          */
 741         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 742                 $titleobj = $article->getTitle();
 743                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 744                         $title = $titleobj->getDBkey();
 745                         $t = explode('/', $title, 3);
 746                         $c = count($t);
 747                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 748                                 if(in_array($t[1], $this->mVariants)) {
 749                                         $this->reloadTables();
 750                                 }
 751                         }
 752                 }
 753                 return true;
 754         }
 755
 756         /**
 757          * Armour rendered math against conversion
 758          * Wrap math into rawoutput -{R| math }- syntax
 759          * @public
 760          */
 761         function armourMath($text){
 762                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 763                 return $ret;
 764         }
 765 }
 766
 767 /**
 768  * parser for rules of language conversion , parse rules in -{ }- tag
 769  * @ingroup Language
 770  * @author  fdcn <fdcn64@gmail.com>
 771  */
 772 class ConverterRule {
 773         var $mText; // original text in -{text}-
 774         var $mConverter; // LanguageConverter object
 775         var $mManualCodeError='<strong class="error">code error!</strong>';
 776         var $mRuleDisplay = '',$mRuleTitle=false;
 777         var $mRules = '';// string : the text of the rules
 778         var $mRulesAction = 'none';
 779         var $mFlags = array();
 780         var $mConvTable = array();
 781         var $mBidtable = array();// array of the translation in each variant
 782         var $mUnidtable = array();// array of the translation in each variant
 783
 784         /**
 785          * Constructor
 786          *
 787          * @param string $text the text between -{ and }-
 788          * @param object $converter a  LanguageConverter object
 789          * @access public
 790          */
 791         function __construct($text,$converter){
 792                 $this->mText = $text;
 793                 $this->mConverter=$converter;
 794                 foreach($converter->mVariants as $v){
 795                         $this->mConvTable[$v]=array();
 796                 }
 797         }
 798
 799         /**
 800          * check if variants array in convert array
 801          *
 802          * @param string $variant Variant language code
 803          * @return string Translated text
 804          * @public
 805          */
 806         function getTextInBidtable($variants){
 807                 if(is_string($variants)){ $variants=array($variants); }
 808                 if(!is_array($variants)) return false;
 809                 foreach ($variants as $variant){
 810                         if(array_key_exists($variant, $this->mBidtable)){
 811                                 return $this->mBidtable[$variant];
 812                         }
 813                 }
 814                 return false;
 815         }
 816
 817         /**
 818          * Parse flags with syntax -{FLAG| ... }-
 819          * @private
 820          */
 821         function parseFlags(){
 822                 $text = $this->mText;
 823                 if(strlen($text) < 2 ) {
 824                         $this->mFlags = array( 'R' );
 825                         $this->mRules = $text;
 826                         return;
 827                 }
 828
 829                 $flags = array();
 830                 $markup = $this->mConverter->mMarkup;
 831                 $validFlags = $this->mConverter->mFlags;
 832
 833                 $tt = explode($markup['flagsep'], $text, 2);
 834                 if(count($tt) == 2) {
 835                         $f = explode($markup['varsep'], $tt[0]);
 836                         foreach($f as $ff) {
 837                                 $ff = trim($ff);
 838                                 if(array_key_exists($ff, $validFlags) &&
 839                                                         !in_array($validFlags[$ff], $flags))
 840                                         $flags[] = $validFlags[$ff];
 841                         }
 842                         $rules = $tt[1];
 843                 } else {
 844                         $rules = $text;
 845                 }
 846
 847                 //check flags
 848                 if( in_array('R',$flags) ){
 849                         $flags = array('R');// remove other flags
 850                 } elseif ( in_array('N',$flags) ){
 851                         $flags = array('N');// remove other flags
 852                 } elseif ( in_array('-',$flags) ){
 853                         $flags = array('-');// remove other flags
 854                 } elseif (count($flags)==1 && $flags[0]=='T'){
 855                         $flags[]='H';
 856                 } elseif ( in_array('H',$flags) ){
 857                         // replace A flag, and remove other flags except T
 858                         $temp=array('+','H');
 859                         if(in_array('T',$flags)) $temp[] = 'T';
 860                         if(in_array('D',$flags)) $temp[] = 'D';
 861                         $flags = $temp;
 862                 } else {
 863                         if ( in_array('A',$flags)) {
 864                                 $flags[]='+';
 865                                 $flags[]='S';
 866                         }
 867                         if ( in_array('D',$flags) )
 868                                 $flags=array_diff($flags,array('S'));
 869                 }
 870                 if ( count($flags)==0 )
 871                         $flags = array('S');
 872                 $this->mRules=$rules;
 873                 $this->mFlags=$flags;
 874         }
 875
 876         /**
 877          * generate conversion table
 878          * @private
 879          */
 880         function parseRules() {
 881                 $rules = $this->mRules;
 882                 $flags = $this->mFlags;
 883                 $bidtable = array();
 884                 $unidtable = array();
 885                 $markup = $this->mConverter->mMarkup;
 886
 887                 $choice = explode($markup['varsep'], $rules );
 888                 foreach($choice as $c) {
 889                         $v = explode($markup['codesep'], $c);
 890                         if(count($v) != 2)
 891                                 continue;// syntax error, skip
 892                         $to=trim($v[1]);
 893                         $v=trim($v[0]);
 894                         $u = explode($markup['unidsep'], $v);
 895                         if(count($u) == 1) {
 896                                 $bidtable[$v] = $to;
 897                         } else if(count($u) == 2){
 898                                 $from=trim($u[0]);$v=trim($u[1]);
 899                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 900                                         $unidtable[$v]=array($from=>$to);
 901                                 else
 902                                         $unidtable[$v][$from]=$to;
 903                         }
 904                         // syntax error, pass
 905                         if (!array_key_exists($v,$this->mConverter->mLanguageNames)){
 906                                 $bidtable = array();
 907                                 $unidtable = array();
 908                                 break;
 909                         }
 910                 }
 911                 $this->mBidtable = $bidtable;
 912                 $this->mUnidtable = $unidtable;
 913         }
 914
 915         /**
 916          * @private
 917          */
 918         function getRulesDesc(){
 919                 $codesep = $this->mConverter->mDescCodeSep;
 920                 $varsep = $this->mConverter->mDescVarSep;
 921                 $text='';
 922                 foreach($this->mBidtable as $k => $v)
 923                         $text .= $this->mConverter->mLanguageNames[$k]."$codesep$v$varsep";
 924                 foreach($this->mUnidtable as $k => $a)
 925                         foreach($a as $from=>$to)
 926                                 $text.="$from&#8658;".$this->mConverter->mLanguageNames[$k]."$codesep$to$varsep";
 927                 return $text;
 928         }
 929
 930         /**
 931          *  Parse rules conversion
 932          * @private
 933          */
 934         function getRuleConvertedStr($variant,$doConvert){
 935                 $bidtable = $this->mBidtable;
 936                 $unidtable = $this->mUnidtable;
 937
 938                 if( count($bidtable) + count($unidtable) == 0 ){
 939                         return $this->mRules;
 940                 } elseif ($doConvert){// the text converted
 941                         // display current variant in bidirectional array
 942                         $disp = $this->getTextInBidtable($variant);
 943                         // or display current variant in fallbacks
 944                         if(!$disp)
 945                                 $disp = $this->getTextInBidtable(
 946                                                 $this->mConverter->getVariantFallbacks($variant));
 947                         // or display current variant in unidirectional array
 948                         if(!$disp && array_key_exists($variant,$unidtable)){
 949                                 $disp = array_values($unidtable[$variant]);
 950                                 $disp = $disp[0];
 951                         }
 952                         // or display frist text under disable manual convert
 953                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
 954                                 if(count($bidtable)>0){
 955                                         $disp = array_values($bidtable);
 956                                         $disp = $disp[0];
 957                                 } else {
 958                                         $disp = array_values($unidtable);
 959                                         $disp = array_values($disp[0]);
 960                                         $disp = $disp[0];
 961                                 }
 962                         }
 963                         return $disp;
 964                 } else {// no convert
 965                         return $this->mRules;
 966                 }
 967         }
 968
 969         /**
 970          * generate conversion table for all text
 971          * @private
 972          */
 973         function generateConvTable(){
 974                 $flags = $this->mFlags;
 975                 $bidtable = $this->mBidtable;
 976                 $unidtable = $this->mUnidtable;
 977                 $manLevel = $this->mConverter->mManualLevel;
 978
 979                 $vmarked=array();
 980                 foreach($this->mConverter->mVariants as $v) {
 981                         /* for bidirectional array
 982                                 fill in the missing variants, if any,
 983                                 with fallbacks */
 984                         if(!array_key_exists($v, $bidtable)) {
 985                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
 986                                 $vf = $this->getTextInBidtable($variantFallbacks);
 987                                 if($vf) $bidtable[$v] = $vf;
 988                         }
 989
 990                         if(array_key_exists($v,$bidtable)){
 991                                 foreach($vmarked as $vo){
 992                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
 993                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
 994                                         // to introduce a custom mapping between
 995                                         // words WordZh and WordTw in the whole text
 996                                         if($manLevel[$v]=='bidirectional'){
 997                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
 998                                         }
 999                                         if($manLevel[$vo]=='bidirectional'){
1000                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1001                                         }
1002                                 }
1003                                 $vmarked[]=$v;
1004                         }
1005                         /*for unidirectional array
1006                                 fill to convert tables */
1007                         $allow_unid = $manLevel[$v]=='bidirectional'
1008                                         || $manLevel[$v]=='unidirectional';
1009                         if($allow_unid && array_key_exists($v,$unidtable)){
1010                                 $ct=$this->mConvTable[$v];
1011                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1012                         }
1013                 }
1014         }
1015
1016         /**
1017          * Parse rules and flags
1018          * @public
1019          */
1020         function parse($variant){
1021                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1022
1023                 $this->parseFlags();
1024                 $flags = $this->mFlags;
1025
1026                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1027                         //FIXME: may cause trouble here...
1028                         //strip &nbsp; since it interferes with the parsing, plus,
1029                         //all spaces should be stripped in this tag anyway.
1030                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1031                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1032                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1033
1034                         $this->parseRules();
1035                 }
1036                 $rules = $this->mRules;
1037
1038                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1039                         if(in_array('+',$flags) || in_array('-',$flags))
1040                                 // fill all variants if text in -{A/H/-|text} without rules
1041                                 foreach($this->mConverter->mVariants as $v)
1042                                         $this->mBidtable[$v] = $rules;
1043                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1044                                 $this->mFlags = $flags = array('R');
1045                 }
1046
1047                 if( in_array('R',$flags) ) {
1048                         // if we don't do content convert, still strip the -{}- tags
1049                         $this->mRuleDisplay = $rules;
1050                 } elseif ( in_array('N',$flags) ){
1051                         // proces N flag: output current variant name
1052                         $this->mRuleDisplay = $this->mConverter->mLanguageNames[trim($rules)];
1053                 } elseif ( in_array('D',$flags) ){
1054                         // proces D flag: output rules description
1055                         $this->mRuleDisplay = $this->getRulesDesc();
1056                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1057                         // proces H,- flag or T only: output nothing
1058                         $this->mRuleDisplay = '';
1059                 } elseif ( in_array('S',$flags) ){
1060                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1061                                                         $this->mConverter->mDoContentConvert);
1062                 } else {
1063                         $this->mRuleDisplay= $this->mManualCodeError;
1064                 }
1065                 // proces T flag
1066                 if ( in_array('T',$flags) ) {
1067                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1068                                                         $this->mConverter->mDoTitleConvert);
1069                 }
1070
1071                 if (in_array('-', $flags))
1072                         $this->mRulesAction='remove';
1073                 if (in_array('+', $flags))
1074                         $this->mRulesAction='add';
1075
1076                 $this->generateConvTable();
1077         }
1078
1079         /**
1080          * @public
1081          */
1082         function hasRules(){
1083                 // TODO:
1084         }
1085
1086         /**
1087          * get display text on markup -{...}-
1088          * @public
1089          */
1090         function getDisplay(){
1091                 return $this->mRuleDisplay;
1092         }
1093         /**
1094          * get converted title
1095          * @public
1096          */
1097         function getTitle(){
1098                 return $this->mRuleTitle;
1099         }
1100
1101         /**
1102          * return how deal with conversion rules
1103          * @public
1104          */
1105         function getRulesAction(){
1106                 return $this->mRulesAction;
1107         }
1108
1109         /**
1110          * get conversion table ( bidirectional and unidirectional conversion table )
1111          * @public
1112          */
1113         function getConvTable(){
1114                 return $this->mConvTable;
1115         }
1116
1117         /**
1118          * get conversion rules string
1119          * @public
1120          */
1121         function getRules(){
1122                 return $this->mRules;
1123         }
1124
1125         /**
1126          * get conversion flags
1127          * @public
1128          */
1129         function getFlags(){
1130                 return $this->mFlags;
1131         }
1132 }