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>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mVariantNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mTitleDisplay='';
  25         var $mDoTitleConvert=true, $mDoContentConvert=true;
  26         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  27         var $mTitleFromFlag = false;
  28         var $mCacheKey;
  29         var $mLangObj;
  30         var $mMarkup;
  31         var $mFlags;
  32         var $mDescCodeSep = ':',$mDescVarSep = ';';
  33         var $mUcfirst = false;
  34
  35         const CACHE_VERSION_KEY = 'VERSION 6';
  36
  37         /**
  38          * Constructor
  39          *
  40          * @param string $maincode the main language code of this language
  41          * @param array $variants the supported variants of this language
  42          * @param array $variantfallback the fallback language of each variant
  43          * @param array $markup array defining the markup used for manual conversion
  44          * @param array $flags array defining the custom strings that maps to the flags
  45          * @param array $manualLevel limit for supported variants
  46          * @public
  47          */
  48         function __construct($langobj, $maincode,
  49                                                                 $variants=array(),
  50                                                                 $variantfallbacks=array(),
  51                                                                 $markup=array(),
  52                                                                 $flags = array(),
  53                                                                 $manualLevel = array() ) {
  54                 $this->mLangObj = $langobj;
  55                 $this->mMainLanguageCode = $maincode;
  56                 $this->mVariants = $variants;
  57                 $this->mVariantFallbacks = $variantfallbacks;
  58                 global $wgLanguageNames;
  59                 $this->mVariantNames = $wgLanguageNames;
  60                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  61                 $m = array(
  62                         'begin'=>'-{',
  63                         'flagsep'=>'|',
  64                         'unidsep'=>'=>', //for unidirectional conversion
  65                         'codesep'=>':',
  66                         'varsep'=>';',
  67                         'end'=>'}-'
  68                 );
  69                 $this->mMarkup = array_merge($m, $markup);
  70                 $f = array(
  71                         // 'S' show converted text
  72                         // '+' add rules for alltext
  73                         // 'E' the gave flags is error
  74                         // these flags above are reserved for program
  75                         'A'=>'A',       // add rule for convert code (all text convert)
  76                         'T'=>'T',       // title convert
  77                         'R'=>'R',       // raw content
  78                         'D'=>'D',       // convert description (subclass implement)
  79                         '-'=>'-',       // remove convert (not implement)
  80                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  81                         'N'=>'N'        // current variant name
  82                 );
  83                 $this->mFlags = array_merge($f, $flags);
  84                 foreach( $this->mVariants as $v)
  85                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  86                                                                 ?$manualLevel[$v]
  87                                                                 :'bidirectional';
  88         }
  89
  90         /**
  91          * @public
  92          */
  93         function getVariants() {
  94                 return $this->mVariants;
  95         }
  96
  97         /**
  98          * in case some variant is not defined in the markup, we need
  99          * to have some fallback. for example, in zh, normally people
 100          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 101          * when zh-sg is preferred but not defined, we will pick zh-hans
 102          * in this case. right now this is only used by zh.
 103          *
 104          * @param string $v the language code of the variant
 105          * @return string array the code of the fallback language or false if there is no fallback
 106          * @public
 107          */
 108         function getVariantFallbacks($v) {
 109                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 110                         return $this->mVariantFallbacks[$v];
 111                 }
 112                 return $this->mMainLanguageCode;
 113         }
 114
 115         /**
 116          * get preferred language variants.
 117          * @param boolean $fromUser Get it from $wgUser's preferences
 118          * @return string the preferred language code
 119          * @public
 120          */
 121         function getPreferredVariant( $fromUser = true ) {
 122                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 123
 124                 if($this->mPreferredVariant)
 125                         return $this->mPreferredVariant;
 126
 127                 // figure out user lang without constructing wgLang to avoid infinite recursion
 128                 if( $fromUser )
 129                         $defaultUserLang = $wgUser->getOption( 'language' );
 130                 else
 131                         $defaultUserLang = $this->mMainLanguageCode;
 132                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 133                 // see if interface language is same as content, if not, prevent conversion
 134                 if( ! in_array( $userLang, $this->mVariants ) ){
 135                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 136                         return $this->mPreferredVariant;
 137                 }
 138
 139                 // see if the preference is set in the request
 140                 $req = $wgRequest->getText( 'variant' );
 141                 if( in_array( $req, $this->mVariants ) ) {
 142                         $this->mPreferredVariant = $req;
 143                         return $req;
 144                 }
 145
 146                 // check the syntax /code/ArticleTitle
 147                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 148                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 149                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 150                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 151                         if(in_array($scriptBase,$this->mVariants)){
 152                                 $this->mPreferredVariant = $scriptBase;
 153                                 return $this->mPreferredVariant;
 154                         }
 155                 }
 156
 157                 // get language variant preference from logged in users
 158                 // Don't call this on stub objects because that causes infinite
 159                 // recursion during initialisation
 160                 if( $fromUser && $wgUser->isLoggedIn() )  {
 161                         $this->mPreferredVariant = $wgUser->getOption('variant');
 162                         return $this->mPreferredVariant;
 163                 }
 164
 165                 // see if default variant is globaly set
 166                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 167                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 168                         return $this->mPreferredVariant;
 169                 }
 170
 171                 # FIXME rewrite code for parsing http header. The current code
 172                 # is written specific for detecting zh- variants
 173                 if( !$this->mPreferredVariant ) {
 174                         // see if some supported language variant is set in the
 175                         // http header, but we don't set the mPreferredVariant
 176                         // variable in case this is called before the user's
 177                         // preference is loaded
 178                         $pv=$this->mMainLanguageCode;
 179                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 180                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 181                                 $zh = strstr($header, $pv.'-');
 182                                 if($zh) {
 183                                         $pv = substr($zh,0,5);
 184                                 }
 185                         }
 186                         // don't try to return bad variant
 187                         if(in_array( $pv, $this->mVariants ))
 188                                 return $pv;
 189                 }
 190
 191                 return $this->mMainLanguageCode;
 192
 193         }
 194
 195         /**
 196          * dictionary-based conversion
 197          *
 198          * @param string $text the text to be converted
 199          * @param string $toVariant the target language code
 200          * @return string the converted text
 201          * @private
 202          */
 203         function autoConvert($text, $toVariant=false) {
 204                 $fname="LanguageConverter::autoConvert";
 205
 206                 wfProfileIn( $fname );
 207
 208                 if(!$this->mTablesLoaded)
 209                         $this->loadTables();
 210
 211                 if(!$toVariant)
 212                         $toVariant = $this->getPreferredVariant();
 213                 if(!in_array($toVariant, $this->mVariants))
 214                         return $text;
 215
 216                 /* we convert everything except:
 217                    1. html markups (anything between < and >)
 218                    2. html entities
 219                    3. place holders created by the parser
 220                 */
 221                 global $wgParser;
 222                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 223                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 224                 } else
 225                         $marker = "";
 226
 227                 // this one is needed when the text is inside an html markup
 228                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 229
 230                 // disable convert to variants between <code></code> tags
 231                 $codefix = '<code>.+?<\/code>|';
 232                 // disable convertsion of <script type="text/javascript"> ... </script>
 233                 $scriptfix = '<script.*?>.*?<\/script>|';
 234                 // disable conversion of <pre xxxx> ... </pre>
 235                 $prefix = '<pre.*?>.*?<\/pre>|';
 236
 237                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 238
 239                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 240
 241                 $m = array_shift($matches);
 242
 243                 $ret = $this->translate($m[0], $toVariant);
 244                 $mstart = $m[1]+strlen($m[0]);
 245                 foreach($matches as $m) {
 246                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 247                         $ret .= $this->translate($m[0], $toVariant);
 248                         $mstart = $m[1] + strlen($m[0]);
 249                 }
 250                 wfProfileOut( $fname );
 251                 return $ret;
 252         }
 253
 254         /**
 255          * Translate a string to a variant
 256          * Doesn't process markup or do any of that other stuff, for that use convert()
 257          *
 258          * @param string $text Text to convert
 259          * @param string $variant Variant language code
 260          * @return string Translated text
 261          * @private
 262          */
 263         function translate( $text, $variant ) {
 264                 wfProfileIn( __METHOD__ );
 265                 if( !$this->mTablesLoaded )
 266                         $this->loadTables();
 267                 $text = $this->mTables[$variant]->replace( $text );
 268                 wfProfileOut( __METHOD__ );
 269                 return $text;
 270         }
 271
 272         /**
 273          * convert text to all supported variants
 274          *
 275          * @param string $text the text to be converted
 276          * @return array of string
 277          * @public
 278          */
 279         function autoConvertToAllVariants($text) {
 280                 $fname="LanguageConverter::autoConvertToAllVariants";
 281                 wfProfileIn( $fname );
 282                 if( !$this->mTablesLoaded )
 283                         $this->loadTables();
 284
 285                 $ret = array();
 286                 foreach($this->mVariants as $variant) {
 287                         $ret[$variant] = $this->translate($text, $variant);
 288                 }
 289
 290                 wfProfileOut( $fname );
 291                 return $ret;
 292         }
 293
 294         /**
 295          * convert link text to all supported variants
 296          *
 297          * @param string $text the text to be converted
 298          * @return array of string
 299          * @public
 300          */
 301         function convertLinkToAllVariants($text) {
 302                 if( !$this->mTablesLoaded )
 303                         $this->loadTables();
 304
 305                 $ret = array();
 306                 $tarray = explode($this->mMarkup['begin'], $text);
 307                 $tfirst = array_shift($tarray);
 308
 309                 foreach($this->mVariants as $variant)
 310                         $ret[$variant] = $this->translate($tfirst,$variant);
 311
 312                 foreach($tarray as $txt) {
 313                         $marked = explode($this->mMarkup['end'], $txt, 2);
 314
 315                         foreach($this->mVariants as $variant){
 316                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 317                                 if(array_key_exists(1, $marked))
 318                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 319                         }
 320
 321                 }
 322
 323                 return $ret;
 324         }
 325
 326
 327         /**
 328          * apply manual conversion
 329          * @private
 330          */
 331         function applyManualConv($convRule){
 332                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 333                 $title = $convRule->getTitle();
 334                 if($title){
 335                         $this->mTitleFromFlag = true;
 336                         $this->mTitleDisplay =  $title;
 337                 }
 338
 339                 //apply manual conversion table to global table
 340                 $convTable = $convRule->getConvTable();
 341                 $action = $convRule->getRulesAction();
 342                 foreach($convTable as $v=>$t) {
 343                         if( !in_array($v,$this->mVariants) )continue;
 344                         if( $action=="add" )
 345                                 $this->mTables[$v]->mergeArray($t);
 346                         elseif ( $action=="remove" )
 347                                 $this->mTables[$v]->removeArray($t);
 348                 }
 349         }
 350
 351         /**
 352          * Convert text using a parser object for context
 353          * @public
 354          */
 355         function parserConvert( $text, &$parser ) {
 356                 global $wgDisableLangConversion;
 357                 /* don't do anything if this is the conversion table */
 358                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 359                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 360                 {
 361                         return $text;
 362                 }
 363
 364                 if($wgDisableLangConversion)
 365                         return $text;
 366
 367                 $text = $this->convert( $text );
 368                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 369                 return $text;
 370         }
 371
 372         /**
 373          *  convert title
 374          * @private
 375          */
 376         function convertTitle($text){
 377                 // check for __NOTC__ tag
 378                 if( !$this->mDoTitleConvert ) {
 379                         $this->mTitleDisplay = $text;
 380                         return $text;
 381                 }
 382
 383                 // use the title from the T flag if any
 384                 if($this->mTitleFromFlag){
 385                         $this->mTitleFromFlag = false;
 386                         return $this->mTitleDisplay;
 387                 }
 388
 389                 global $wgRequest;
 390                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 391                 $action = $wgRequest->getText( 'action' );
 392                 if ( $isredir == 'no' || $action == 'edit' ) {
 393                         return $text;
 394                 } else {
 395                         $this->mTitleDisplay = $this->convert($text);
 396                         return $this->mTitleDisplay;
 397                 }
 398         }
 399
 400         /**
 401          * convert text to different variants of a language. the automatic
 402          * conversion is done in autoConvert(). here we parse the text
 403          * marked with -{}-, which specifies special conversions of the
 404          * text that can not be accomplished in autoConvert()
 405          *
 406          * syntax of the markup:
 407          * -{code1:text1;code2:text2;...}-  or
 408          * -{flags|code1:text1;code2:text2;...}-  or
 409          * -{text}- in which case no conversion should take place for text
 410          *
 411          * @param string $text text to be converted
 412          * @param bool $isTitle whether this conversion is for the article title
 413          * @return string converted text
 414          * @public
 415          */
 416         function convert( $text , $isTitle=false) {
 417
 418                 $mw =& MagicWord::get( 'notitleconvert' );
 419                 if( $mw->matchAndRemove( $text ) )
 420                         $this->mDoTitleConvert = false;
 421                 $mw =& MagicWord::get( 'nocontentconvert' );
 422                 if( $mw->matchAndRemove( $text ) ) {
 423                         $this->mDoContentConvert = false;
 424                 }
 425
 426                 // no conversion if redirecting
 427                 $mw =& MagicWord::get( 'redirect' );
 428                 if( $mw->matchStart( $text ))
 429                         return $text;
 430
 431                 // for title convertion
 432                 if ($isTitle) return $this->convertTitle($text);
 433
 434                 $plang = $this->getPreferredVariant();
 435                 $tarray = explode($this->mMarkup['end'], $text);
 436                 $text = '';
 437                 foreach($tarray as $txt) {
 438                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 439
 440                         if( $this->mDoContentConvert )
 441                                 $text .= $this->autoConvert($marked[0],$plang);
 442                         else
 443                                 $text .= $marked[0];
 444
 445                         if(array_key_exists(1, $marked)){
 446                                 // strip the flags from syntax like -{T| ... }-
 447                                 $crule = new ConverterRule($marked[1], $this);
 448                                 $crule->parse($plang);
 449
 450                                 $text .= $crule->getDisplay();
 451                                 $this->applyManualConv($crule);
 452                         }
 453                 }
 454
 455                 return $text;
 456         }
 457
 458         /**
 459          * if a language supports multiple variants, it is
 460          * possible that non-existing link in one variant
 461          * actually exists in another variant. this function
 462          * tries to find it. See e.g. LanguageZh.php
 463          *
 464          * @param string $link the name of the link
 465          * @param mixed $nt the title object of the link
 466          * @return null the input parameters may be modified upon return
 467          * @public
 468          */
 469         function findVariantLink( &$link, &$nt ) {
 470                 global $wgDisableLangConversion;
 471                 $linkBatch = new LinkBatch();
 472
 473                 $ns=NS_MAIN;
 474
 475                 if(is_object($nt))
 476                         $ns = $nt->getNamespace();
 477
 478                 $variants = $this->autoConvertToAllVariants($link);
 479                 if($variants == false) //give up
 480                         return;
 481
 482                 $titles = array();
 483
 484                 foreach( $variants as $v ) {
 485                         if($v != $link){
 486                                 $varnt = Title::newFromText( $v, $ns );
 487                                 if(!is_null($varnt)){
 488                                         $linkBatch->addObj($varnt);
 489                                         $titles[]=$varnt;
 490                                 }
 491                         }
 492                 }
 493
 494                 // fetch all variants in single query
 495                 $linkBatch->execute();
 496
 497                 foreach( $titles as $varnt ) {
 498                         if( $varnt->getArticleID() > 0 ) {
 499                                 $nt = $varnt;
 500                                 if( !$wgDisableLangConversion )
 501                                         $link = $v;
 502                                 break;
 503                         }
 504                 }
 505         }
 506
 507     /**
 508          * returns language specific hash options
 509          *
 510          * @public
 511          */
 512         function getExtraHashOptions() {
 513                 $variant = $this->getPreferredVariant();
 514                 return '!' . $variant ;
 515         }
 516
 517     /**
 518          * get title text as defined in the body of the article text
 519          *
 520          * @public
 521          */
 522         function getParsedTitle() {
 523                 return $this->mTitleDisplay;
 524         }
 525
 526         /**
 527          * a write lock to the cache
 528          *
 529          * @private
 530          */
 531         function lockCache() {
 532                 global $wgMemc;
 533                 $success = false;
 534                 for($i=0; $i<30; $i++) {
 535                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 536                                 break;
 537                         sleep(1);
 538                 }
 539                 return $success;
 540         }
 541
 542         /**
 543          * unlock cache
 544          *
 545          * @private
 546          */
 547         function unlockCache() {
 548                 global $wgMemc;
 549                 $wgMemc->delete($this->mCacheKey . "lock");
 550         }
 551
 552
 553         /**
 554          * Load default conversion tables
 555          * This method must be implemented in derived class
 556          *
 557          * @private
 558          */
 559         function loadDefaultTables() {
 560                 $name = get_class($this);
 561                 wfDie("Must implement loadDefaultTables() method in class $name");
 562         }
 563
 564         /**
 565          * load conversion tables either from the cache or the disk
 566          * @private
 567          */
 568         function loadTables($fromcache=true) {
 569                 global $wgMemc;
 570                 if( $this->mTablesLoaded )
 571                         return;
 572                 wfProfileIn( __METHOD__ );
 573                 $this->mTablesLoaded = true;
 574                 $this->mTables = false;
 575                 if($fromcache) {
 576                         wfProfileIn( __METHOD__.'-cache' );
 577                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 578                         wfProfileOut( __METHOD__.'-cache' );
 579                 }
 580                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 581                         wfProfileIn( __METHOD__.'-recache' );
 582                         // not in cache, or we need a fresh reload.
 583                         // we will first load the default tables
 584                         // then update them using things in MediaWiki:Zhconversiontable/*
 585                         $this->loadDefaultTables();
 586                         foreach($this->mVariants as $var) {
 587                                 $cached = $this->parseCachedTable($var);
 588                                 $this->mTables[$var]->mergeArray($cached);
 589                         }
 590
 591                         $this->postLoadTables();
 592                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 593
 594                         if($this->lockCache()) {
 595                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 596                                 $this->unlockCache();
 597                         }
 598                         wfProfileOut( __METHOD__.'-recache' );
 599                 }
 600                 wfProfileOut( __METHOD__ );
 601         }
 602
 603     /**
 604          * Hook for post processig after conversion tables are loaded
 605          *
 606          */
 607         function postLoadTables() {}
 608
 609     /**
 610          * Reload the conversion tables
 611          *
 612          * @private
 613          */
 614         function reloadTables() {
 615                 if($this->mTables)
 616                         unset($this->mTables);
 617                 $this->mTablesLoaded = false;
 618                 $this->loadTables(false);
 619         }
 620
 621
 622         /**
 623          * parse the conversion table stored in the cache
 624          *
 625          * the tables should be in blocks of the following form:
 626          *              -{
 627          *                      word => word ;
 628          *                      word => word ;
 629          *                      ...
 630          *              }-
 631          *
 632          *      to make the tables more manageable, subpages are allowed
 633          *      and will be parsed recursively if $recursive=true
 634          *
 635          */
 636         function parseCachedTable($code, $subpage='', $recursive=true) {
 637                 global $wgMessageCache;
 638                 static $parsed = array();
 639
 640                 if(!is_object($wgMessageCache))
 641                         return array();
 642
 643                 $key = 'Conversiontable/'.$code;
 644                 if($subpage)
 645                         $key .= '/' . $subpage;
 646
 647                 if(array_key_exists($key, $parsed))
 648                         return array();
 649
 650                 if ( strpos( $code, '/' ) === false ) {
 651                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 652                 } else {
 653                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 654                         if ( $title && $title->exists() ) {
 655                                 $article = new Article( $title );
 656                                 $txt = $article->getContents();
 657                         } else {
 658                                 $txt = '';
 659                         }
 660                 }
 661
 662                 // get all subpage links of the form
 663                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 664                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 665                 $subs = explode('[[', $txt);
 666                 $sublinks = array();
 667                 foreach( $subs as $sub ) {
 668                         $link = explode(']]', $sub, 2);
 669                         if(count($link) != 2)
 670                                 continue;
 671                         $b = explode('|', $link[0]);
 672                         $b = explode('/', trim($b[0]), 3);
 673                         if(count($b)==3)
 674                                 $sublink = $b[2];
 675                         else
 676                                 $sublink = '';
 677
 678                         if($b[0] == $linkhead && $b[1] == $code) {
 679                                 $sublinks[] = $sublink;
 680                         }
 681                 }
 682
 683
 684                 // parse the mappings in this page
 685                 $blocks = explode($this->mMarkup['begin'], $txt);
 686                 array_shift($blocks);
 687                 $ret = array();
 688                 foreach($blocks as $block) {
 689                         $mappings = explode($this->mMarkup['end'], $block, 2);
 690                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 691                         $table = explode( ';', $stripped );
 692                         foreach( $table as $t ) {
 693                                 $m = explode( '=>', $t );
 694                                 if( count( $m ) != 2)
 695                                         continue;
 696                                 // trim any trailling comments starting with '//'
 697                                 $tt = explode('//', $m[1], 2);
 698                                 $ret[trim($m[0])] = trim($tt[0]);
 699                         }
 700                 }
 701                 $parsed[$key] = true;
 702
 703
 704                 // recursively parse the subpages
 705                 if($recursive) {
 706                         foreach($sublinks as $link) {
 707                                 $s = $this->parseCachedTable($code, $link, $recursive);
 708                                 $ret = array_merge($ret, $s);
 709                         }
 710                 }
 711
 712                 if ($this->mUcfirst) {
 713                         foreach ($ret as $k => $v) {
 714                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 715                         }
 716                 }
 717                 return $ret;
 718         }
 719
 720         /**
 721          * Enclose a string with the "no conversion" tag. This is used by
 722          * various functions in the Parser
 723          *
 724          * @param string $text text to be tagged for no conversion
 725          * @return string the tagged text
 726          * @public
 727          */
 728         function markNoConversion($text, $noParse=false) {
 729                 # don't mark if already marked
 730                 if(strpos($text, $this->mMarkup['begin']) ||
 731                    strpos($text, $this->mMarkup['end']))
 732                         return $text;
 733
 734                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 735                 return $ret;
 736         }
 737
 738         /**
 739          * convert the sorting key for category links. this should make different
 740          * keys that are variants of each other map to the same key
 741          */
 742         function convertCategoryKey( $key ) {
 743                 return $key;
 744         }
 745         /**
 746          * hook to refresh the cache of conversion tables when
 747          * MediaWiki:conversiontable* is updated
 748          * @private
 749          */
 750         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 751                 $titleobj = $article->getTitle();
 752                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 753                         $title = $titleobj->getDBkey();
 754                         $t = explode('/', $title, 3);
 755                         $c = count($t);
 756                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 757                                 if(in_array($t[1], $this->mVariants)) {
 758                                         $this->reloadTables();
 759                                 }
 760                         }
 761                 }
 762                 return true;
 763         }
 764
 765         /**
 766          * Armour rendered math against conversion
 767          * Wrap math into rawoutput -{R| math }- syntax
 768          * @public
 769          */
 770         function armourMath($text){
 771                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 772                 return $ret;
 773         }
 774 }
 775
 776 /**
 777  * parser for rules of language conversion , parse rules in -{ }- tag
 778  * @ingroup Language
 779  * @author  fdcn <fdcn64@gmail.com>
 780  */
 781 class ConverterRule {
 782         var $mText; // original text in -{text}-
 783         var $mConverter; // LanguageConverter object
 784         var $mManualCodeError='<strong class="error">code error!</strong>';
 785         var $mRuleDisplay = '',$mRuleTitle=false;
 786         var $mRules = '';// string : the text of the rules
 787         var $mRulesAction = 'none';
 788         var $mFlags = array();
 789         var $mConvTable = array();
 790         var $mBidtable = array();// array of the translation in each variant
 791         var $mUnidtable = array();// array of the translation in each variant
 792
 793         /**
 794          * Constructor
 795          *
 796          * @param string $text the text between -{ and }-
 797          * @param object $converter a  LanguageConverter object
 798          * @access public
 799          */
 800         function __construct($text,$converter){
 801                 $this->mText = $text;
 802                 $this->mConverter=$converter;
 803                 foreach($converter->mVariants as $v){
 804                         $this->mConvTable[$v]=array();
 805                 }
 806         }
 807
 808         /**
 809          * check if variants array in convert array
 810          *
 811          * @param string $variant Variant language code
 812          * @return string Translated text
 813          * @public
 814          */
 815         function getTextInBidtable($variants){
 816                 if(is_string($variants)){ $variants=array($variants); }
 817                 if(!is_array($variants)) return false;
 818                 foreach ($variants as $variant){
 819                         if(array_key_exists($variant, $this->mBidtable)){
 820                                 return $this->mBidtable[$variant];
 821                         }
 822                 }
 823                 return false;
 824         }
 825
 826         /**
 827          * Parse flags with syntax -{FLAG| ... }-
 828          * @private
 829          */
 830         function parseFlags(){
 831                 $text = $this->mText;
 832                 if(strlen($text) < 2 ) {
 833                         $this->mFlags = array( 'R' );
 834                         $this->mRules = $text;
 835                         return;
 836                 }
 837
 838                 $flags = array();
 839                 $markup = $this->mConverter->mMarkup;
 840                 $validFlags = $this->mConverter->mFlags;
 841
 842                 $tt = explode($markup['flagsep'], $text, 2);
 843                 if(count($tt) == 2) {
 844                         $f = explode($markup['varsep'], $tt[0]);
 845                         foreach($f as $ff) {
 846                                 $ff = trim($ff);
 847                                 if(array_key_exists($ff, $validFlags) &&
 848                                                         !in_array($validFlags[$ff], $flags))
 849                                         $flags[] = $validFlags[$ff];
 850                         }
 851                         $rules = $tt[1];
 852                 } else {
 853                         $rules = $text;
 854                 }
 855
 856                 //check flags
 857                 if( in_array('R',$flags) ){
 858                         $flags = array('R');// remove other flags
 859                 } elseif ( in_array('N',$flags) ){
 860                         $flags = array('N');// remove other flags
 861                 } elseif ( in_array('-',$flags) ){
 862                         $flags = array('-');// remove other flags
 863                 } elseif (count($flags)==1 && $flags[0]=='T'){
 864                         $flags[]='H';
 865                 } elseif ( in_array('H',$flags) ){
 866                         // replace A flag, and remove other flags except T
 867                         $temp=array('+','H');
 868                         if(in_array('T',$flags)) $temp[] = 'T';
 869                         if(in_array('D',$flags)) $temp[] = 'D';
 870                         $flags = $temp;
 871                 } else {
 872                         if ( in_array('A',$flags)) {
 873                                 $flags[]='+';
 874                                 $flags[]='S';
 875                         }
 876                         if ( in_array('D',$flags) )
 877                                 $flags=array_diff($flags,array('S'));
 878                 }
 879                 if ( count($flags)==0 )
 880                         $flags = array('S');
 881                 $this->mRules=$rules;
 882                 $this->mFlags=$flags;
 883         }
 884
 885         /**
 886          * generate conversion table
 887          * @private
 888          */
 889         function parseRules() {
 890                 $rules = $this->mRules;
 891                 $flags = $this->mFlags;
 892                 $bidtable = array();
 893                 $unidtable = array();
 894                 $markup = $this->mConverter->mMarkup;
 895
 896                 $choice = explode($markup['varsep'], $rules );
 897                 foreach($choice as $c) {
 898                         $v = explode($markup['codesep'], $c);
 899                         if(count($v) != 2)
 900                                 continue;// syntax error, skip
 901                         $to=trim($v[1]);
 902                         $v=trim($v[0]);
 903                         $u = explode($markup['unidsep'], $v);
 904                         if(count($u) == 1) {
 905                                 $bidtable[$v] = $to;
 906                         } else if(count($u) == 2){
 907                                 $from=trim($u[0]);$v=trim($u[1]);
 908                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 909                                         $unidtable[$v]=array($from=>$to);
 910                                 else
 911                                         $unidtable[$v][$from]=$to;
 912                         }
 913                         // syntax error, pass
 914                         if (!array_key_exists($v,$this->mConverter->mVariantNames)){
 915                                 $bidtable = array();
 916                                 $unidtable = array();
 917                                 break;
 918                         }
 919                 }
 920                 $this->mBidtable = $bidtable;
 921                 $this->mUnidtable = $unidtable;
 922         }
 923
 924         /**
 925          * @private
 926          */
 927         function getRulesDesc(){
 928                 $codesep = $this->mConverter->mDescCodeSep;
 929                 $varsep = $this->mConverter->mDescVarSep;
 930                 $text='';
 931                 foreach($this->mBidtable as $k => $v)
 932                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
 933                 foreach($this->mUnidtable as $k => $a)
 934                         foreach($a as $from=>$to)
 935                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
 936                 return $text;
 937         }
 938
 939         /**
 940          *  Parse rules conversion
 941          * @private
 942          */
 943         function getRuleConvertedStr($variant,$doConvert){
 944                 $bidtable = $this->mBidtable;
 945                 $unidtable = $this->mUnidtable;
 946
 947                 if( count($bidtable) + count($unidtable) == 0 ){
 948                         return $this->mRules;
 949                 } elseif ($doConvert){// the text converted
 950                         // display current variant in bidirectional array
 951                         $disp = $this->getTextInBidtable($variant);
 952                         // or display current variant in fallbacks
 953                         if(!$disp)
 954                                 $disp = $this->getTextInBidtable(
 955                                                 $this->mConverter->getVariantFallbacks($variant));
 956                         // or display current variant in unidirectional array
 957                         if(!$disp && array_key_exists($variant,$unidtable)){
 958                                 $disp = array_values($unidtable[$variant]);
 959                                 $disp = $disp[0];
 960                         }
 961                         // or display frist text under disable manual convert
 962                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
 963                                 if(count($bidtable)>0){
 964                                         $disp = array_values($bidtable);
 965                                         $disp = $disp[0];
 966                                 } else {
 967                                         $disp = array_values($unidtable);
 968                                         $disp = array_values($disp[0]);
 969                                         $disp = $disp[0];
 970                                 }
 971                         }
 972                         return $disp;
 973                 } else {// no convert
 974                         return $this->mRules;
 975                 }
 976         }
 977
 978         /**
 979          * generate conversion table for all text
 980          * @private
 981          */
 982         function generateConvTable(){
 983                 $flags = $this->mFlags;
 984                 $bidtable = $this->mBidtable;
 985                 $unidtable = $this->mUnidtable;
 986                 $manLevel = $this->mConverter->mManualLevel;
 987
 988                 $vmarked=array();
 989                 foreach($this->mConverter->mVariants as $v) {
 990                         /* for bidirectional array
 991                                 fill in the missing variants, if any,
 992                                 with fallbacks */
 993                         if(!array_key_exists($v, $bidtable)) {
 994                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
 995                                 $vf = $this->getTextInBidtable($variantFallbacks);
 996                                 if($vf) $bidtable[$v] = $vf;
 997                         }
 998
 999                         if(array_key_exists($v,$bidtable)){
1000                                 foreach($vmarked as $vo){
1001                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1002                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1003                                         // to introduce a custom mapping between
1004                                         // words WordZh and WordTw in the whole text
1005                                         if($manLevel[$v]=='bidirectional'){
1006                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1007                                         }
1008                                         if($manLevel[$vo]=='bidirectional'){
1009                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1010                                         }
1011                                 }
1012                                 $vmarked[]=$v;
1013                         }
1014                         /*for unidirectional array
1015                                 fill to convert tables */
1016                         $allow_unid = $manLevel[$v]=='bidirectional'
1017                                         || $manLevel[$v]=='unidirectional';
1018                         if($allow_unid && array_key_exists($v,$unidtable)){
1019                                 $ct=$this->mConvTable[$v];
1020                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1021                         }
1022                 }
1023         }
1024
1025         /**
1026          * Parse rules and flags
1027          * @public
1028          */
1029         function parse($variant){
1030                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1031
1032                 $this->parseFlags();
1033                 $flags = $this->mFlags;
1034
1035                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1036                         //FIXME: may cause trouble here...
1037                         //strip &nbsp; since it interferes with the parsing, plus,
1038                         //all spaces should be stripped in this tag anyway.
1039                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1040                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1041                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1042
1043                         $this->parseRules();
1044                 }
1045                 $rules = $this->mRules;
1046
1047                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1048                         if(in_array('+',$flags) || in_array('-',$flags))
1049                                 // fill all variants if text in -{A/H/-|text} without rules
1050                                 foreach($this->mConverter->mVariants as $v)
1051                                         $this->mBidtable[$v] = $rules;
1052                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1053                                 $this->mFlags = $flags = array('R');
1054                 }
1055
1056                 if( in_array('R',$flags) ) {
1057                         // if we don't do content convert, still strip the -{}- tags
1058                         $this->mRuleDisplay = $rules;
1059                 } elseif ( in_array('N',$flags) ){
1060                         // proces N flag: output current variant name
1061                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1062                 } elseif ( in_array('D',$flags) ){
1063                         // proces D flag: output rules description
1064                         $this->mRuleDisplay = $this->getRulesDesc();
1065                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1066                         // proces H,- flag or T only: output nothing
1067                         $this->mRuleDisplay = '';
1068                 } elseif ( in_array('S',$flags) ){
1069                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1070                                                         $this->mConverter->mDoContentConvert);
1071                 } else {
1072                         $this->mRuleDisplay= $this->mManualCodeError;
1073                 }
1074                 // proces T flag
1075                 if ( in_array('T',$flags) ) {
1076                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1077                                                         $this->mConverter->mDoTitleConvert);
1078                 }
1079
1080                 if (in_array('-', $flags))
1081                         $this->mRulesAction='remove';
1082                 if (in_array('+', $flags))
1083                         $this->mRulesAction='add';
1084
1085                 $this->generateConvTable();
1086         }
1087
1088         /**
1089          * @public
1090          */
1091         function hasRules(){
1092                 // TODO:
1093         }
1094
1095         /**
1096          * get display text on markup -{...}-
1097          * @public
1098          */
1099         function getDisplay(){
1100                 return $this->mRuleDisplay;
1101         }
1102         /**
1103          * get converted title
1104          * @public
1105          */
1106         function getTitle(){
1107                 return $this->mRuleTitle;
1108         }
1109
1110         /**
1111          * return how deal with conversion rules
1112          * @public
1113          */
1114         function getRulesAction(){
1115                 return $this->mRulesAction;
1116         }
1117
1118         /**
1119          * get conversion table ( bidirectional and unidirectional conversion table )
1120          * @public
1121          */
1122         function getConvTable(){
1123                 return $this->mConvTable;
1124         }
1125
1126         /**
1127          * get conversion rules string
1128          * @public
1129          */
1130         function getRules(){
1131                 return $this->mRules;
1132         }
1133
1134         /**
1135          * get conversion flags
1136          * @public
1137          */
1138         function getFlags(){
1139                 return $this->mFlags;
1140         }
1141 }