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                                         $ary = split("[,;]",$zh);
 184                                         $pv = $ary[0];
 185                                 }
 186                         }
 187                         // don't try to return bad variant
 188                         if(in_array( $pv, $this->mVariants ))
 189                                 return $pv;
 190                 }
 191
 192                 return $this->mMainLanguageCode;
 193
 194         }
 195
 196         /**
 197          * dictionary-based conversion
 198          *
 199          * @param string $text the text to be converted
 200          * @param string $toVariant the target language code
 201          * @return string the converted text
 202          * @private
 203          */
 204         function autoConvert($text, $toVariant=false) {
 205                 $fname="LanguageConverter::autoConvert";
 206
 207                 wfProfileIn( $fname );
 208
 209                 if(!$this->mTablesLoaded)
 210                         $this->loadTables();
 211
 212                 if(!$toVariant)
 213                         $toVariant = $this->getPreferredVariant();
 214                 if(!in_array($toVariant, $this->mVariants))
 215                         return $text;
 216
 217                 /* we convert everything except:
 218                    1. html markups (anything between < and >)
 219                    2. html entities
 220                    3. place holders created by the parser
 221                 */
 222                 global $wgParser;
 223                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 224                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 225                 } else
 226                         $marker = "";
 227
 228                 // this one is needed when the text is inside an html markup
 229                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 230
 231                 // disable convert to variants between <code></code> tags
 232                 $codefix = '<code>.+?<\/code>|';
 233                 // disable convertsion of <script type="text/javascript"> ... </script>
 234                 $scriptfix = '<script.*?>.*?<\/script>|';
 235                 // disable conversion of <pre xxxx> ... </pre>
 236                 $prefix = '<pre.*?>.*?<\/pre>|';
 237
 238                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 239
 240                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 241
 242                 $m = array_shift($matches);
 243
 244                 $ret = $this->translate($m[0], $toVariant);
 245                 $mstart = $m[1]+strlen($m[0]);
 246                 foreach($matches as $m) {
 247                         $ret .= substr($text, $mstart, $m[1]-$mstart);
 248                         $ret .= $this->translate($m[0], $toVariant);
 249                         $mstart = $m[1] + strlen($m[0]);
 250                 }
 251                 wfProfileOut( $fname );
 252                 return $ret;
 253         }
 254
 255         /**
 256          * Translate a string to a variant
 257          * Doesn't process markup or do any of that other stuff, for that use convert()
 258          *
 259          * @param string $text Text to convert
 260          * @param string $variant Variant language code
 261          * @return string Translated text
 262          * @private
 263          */
 264         function translate( $text, $variant ) {
 265                 wfProfileIn( __METHOD__ );
 266                 if( !$this->mTablesLoaded )
 267                         $this->loadTables();
 268                 $text = $this->mTables[$variant]->replace( $text );
 269                 wfProfileOut( __METHOD__ );
 270                 return $text;
 271         }
 272
 273         /**
 274          * convert text to all supported variants
 275          *
 276          * @param string $text the text to be converted
 277          * @return array of string
 278          * @public
 279          */
 280         function autoConvertToAllVariants($text) {
 281                 $fname="LanguageConverter::autoConvertToAllVariants";
 282                 wfProfileIn( $fname );
 283                 if( !$this->mTablesLoaded )
 284                         $this->loadTables();
 285
 286                 $ret = array();
 287                 foreach($this->mVariants as $variant) {
 288                         $ret[$variant] = $this->translate($text, $variant);
 289                 }
 290
 291                 wfProfileOut( $fname );
 292                 return $ret;
 293         }
 294
 295         /**
 296          * convert link text to all supported variants
 297          *
 298          * @param string $text the text to be converted
 299          * @return array of string
 300          * @public
 301          */
 302         function convertLinkToAllVariants($text) {
 303                 if( !$this->mTablesLoaded )
 304                         $this->loadTables();
 305
 306                 $ret = array();
 307                 $tarray = explode($this->mMarkup['begin'], $text);
 308                 $tfirst = array_shift($tarray);
 309
 310                 foreach($this->mVariants as $variant)
 311                         $ret[$variant] = $this->translate($tfirst,$variant);
 312
 313                 foreach($tarray as $txt) {
 314                         $marked = explode($this->mMarkup['end'], $txt, 2);
 315
 316                         foreach($this->mVariants as $variant){
 317                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 318                                 if(array_key_exists(1, $marked))
 319                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 320                         }
 321
 322                 }
 323
 324                 return $ret;
 325         }
 326
 327
 328         /**
 329          * apply manual conversion
 330          * @private
 331          */
 332         function applyManualConv($convRule){
 333                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 334                 $title = $convRule->getTitle();
 335                 if($title){
 336                         $this->mTitleFromFlag = true;
 337                         $this->mTitleDisplay =  $title;
 338                 }
 339
 340                 //apply manual conversion table to global table
 341                 $convTable = $convRule->getConvTable();
 342                 $action = $convRule->getRulesAction();
 343                 foreach($convTable as $v=>$t) {
 344                         if( !in_array($v,$this->mVariants) )continue;
 345                         if( $action=="add" )
 346                                 $this->mTables[$v]->mergeArray($t);
 347                         elseif ( $action=="remove" )
 348                                 $this->mTables[$v]->removeArray($t);
 349                 }
 350         }
 351
 352         /**
 353          * Convert text using a parser object for context
 354          * @public
 355          */
 356         function parserConvert( $text, &$parser ) {
 357                 global $wgDisableLangConversion;
 358                 /* don't do anything if this is the conversion table */
 359                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 360                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 361                 {
 362                         return $text;
 363                 }
 364
 365                 if($wgDisableLangConversion)
 366                         return $text;
 367
 368                 $text = $this->convert( $text );
 369                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 370                 return $text;
 371         }
 372
 373         /**
 374          *  convert title
 375          * @private
 376          */
 377         function convertTitle($text){
 378                 global $wgDisableTitleConversion, $wgUser;
 379
 380                 // check for global param and __NOTC__ tag
 381                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 382                         $this->mTitleDisplay = $text;
 383                         return $text;
 384                 }
 385
 386                 // use the title from the T flag if any
 387                 if($this->mTitleFromFlag){
 388                         $this->mTitleFromFlag = false;
 389                         return $this->mTitleDisplay;
 390                 }
 391
 392                 global $wgRequest;
 393                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 394                 $action = $wgRequest->getText( 'action' );
 395                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 396                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 397                         return $text;
 398                 } else {
 399                         $this->mTitleDisplay = $this->convert($text);
 400                         return $this->mTitleDisplay;
 401                 }
 402         }
 403
 404         /**
 405          * convert text to different variants of a language. the automatic
 406          * conversion is done in autoConvert(). here we parse the text
 407          * marked with -{}-, which specifies special conversions of the
 408          * text that can not be accomplished in autoConvert()
 409          *
 410          * syntax of the markup:
 411          * -{code1:text1;code2:text2;...}-  or
 412          * -{flags|code1:text1;code2:text2;...}-  or
 413          * -{text}- in which case no conversion should take place for text
 414          *
 415          * @param string $text text to be converted
 416          * @param bool $isTitle whether this conversion is for the article title
 417          * @return string converted text
 418          * @public
 419          */
 420         function convert( $text , $isTitle=false) {
 421
 422                 $mw =& MagicWord::get( 'notitleconvert' );
 423                 if( $mw->matchAndRemove( $text ) )
 424                         $this->mDoTitleConvert = false;
 425                 $mw =& MagicWord::get( 'nocontentconvert' );
 426                 if( $mw->matchAndRemove( $text ) ) {
 427                         $this->mDoContentConvert = false;
 428                 }
 429
 430                 // no conversion if redirecting
 431                 $mw =& MagicWord::get( 'redirect' );
 432                 if( $mw->matchStart( $text ))
 433                         return $text;
 434
 435                 // for title convertion
 436                 if ($isTitle) return $this->convertTitle($text);
 437
 438                 $plang = $this->getPreferredVariant();
 439                 $tarray = StringUtils::explode($this->mMarkup['end'], $text);
 440                 $text = '';
 441                 $lastDelim = false;
 442                 foreach($tarray as $txt) {
 443                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 444
 445                         if( $this->mDoContentConvert )
 446                                 $text .= $this->autoConvert($marked[0],$plang);
 447                         else
 448                                 $text .= $marked[0];
 449
 450                         if(array_key_exists(1, $marked)){
 451                                 // strip the flags from syntax like -{T| ... }-
 452                                 $crule = new ConverterRule($marked[1], $this);
 453                                 $crule->parse($plang);
 454
 455                                 $text .= $crule->getDisplay();
 456                                 $this->applyManualConv($crule);
 457                                 $lastDelim = false;
 458                         } else {
 459                                 // Reinsert the }- which wasn't part of anything
 460                                 $text .= $this->mMarkup['end'];
 461                                 $lastDelim = true;
 462                         }
 463                 }
 464                 if ( $lastDelim ) {
 465                         // Remove the last delimiter (wasn't real)
 466                         $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 467                 }
 468
 469                 return $text;
 470         }
 471
 472         /**
 473          * if a language supports multiple variants, it is
 474          * possible that non-existing link in one variant
 475          * actually exists in another variant. this function
 476          * tries to find it. See e.g. LanguageZh.php
 477          *
 478          * @param string $link the name of the link
 479          * @param mixed $nt the title object of the link
 480          * @return null the input parameters may be modified upon return
 481          * @public
 482          */
 483         function findVariantLink( &$link, &$nt, $forTemplate = false ) {
 484                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 485                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 486                 $action = $wgRequest->getText( 'action' );
 487                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 488                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 489                 $linkBatch = new LinkBatch();
 490
 491                 $ns=NS_MAIN;
 492
 493                 if ( $disableLinkConversion || ( !$forTemplate && ( $isredir == 'no' || $action == 'edit'
 494                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) ) {
 495                         return;
 496                 }
 497
 498                 if(is_object($nt))
 499                         $ns = $nt->getNamespace();
 500
 501                 $variants = $this->autoConvertToAllVariants($link);
 502                 if($variants == false) //give up
 503                         return;
 504
 505                 $titles = array();
 506
 507                 foreach( $variants as $v ) {
 508                         if($v != $link){
 509                                 $varnt = Title::newFromText( $v, $ns );
 510                                 if(!is_null($varnt)){
 511                                         $linkBatch->addObj($varnt);
 512                                         $titles[]=$varnt;
 513                                 }
 514                         }
 515                 }
 516
 517                 // fetch all variants in single query
 518                 $linkBatch->execute();
 519
 520                 foreach( $titles as $varnt ) {
 521                         if( $varnt->getArticleID() > 0 ) {
 522                                 $nt = $varnt;
 523                                 $link = $v;
 524                                 break;
 525                         }
 526                 }
 527         }
 528
 529     /**
 530          * returns language specific hash options
 531          *
 532          * @public
 533          */
 534         function getExtraHashOptions() {
 535                 $variant = $this->getPreferredVariant();
 536                 return '!' . $variant ;
 537         }
 538
 539     /**
 540          * get title text as defined in the body of the article text
 541          *
 542          * @public
 543          */
 544         function getParsedTitle() {
 545                 return $this->mTitleDisplay;
 546         }
 547
 548         /**
 549          * a write lock to the cache
 550          *
 551          * @private
 552          */
 553         function lockCache() {
 554                 global $wgMemc;
 555                 $success = false;
 556                 for($i=0; $i<30; $i++) {
 557                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 558                                 break;
 559                         sleep(1);
 560                 }
 561                 return $success;
 562         }
 563
 564         /**
 565          * unlock cache
 566          *
 567          * @private
 568          */
 569         function unlockCache() {
 570                 global $wgMemc;
 571                 $wgMemc->delete($this->mCacheKey . "lock");
 572         }
 573
 574
 575         /**
 576          * Load default conversion tables
 577          * This method must be implemented in derived class
 578          *
 579          * @private
 580          */
 581         function loadDefaultTables() {
 582                 $name = get_class($this);
 583                 wfDie("Must implement loadDefaultTables() method in class $name");
 584         }
 585
 586         /**
 587          * load conversion tables either from the cache or the disk
 588          * @private
 589          */
 590         function loadTables($fromcache=true) {
 591                 global $wgMemc;
 592                 if( $this->mTablesLoaded )
 593                         return;
 594                 wfProfileIn( __METHOD__ );
 595                 $this->mTablesLoaded = true;
 596                 $this->mTables = false;
 597                 if($fromcache) {
 598                         wfProfileIn( __METHOD__.'-cache' );
 599                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 600                         wfProfileOut( __METHOD__.'-cache' );
 601                 }
 602                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 603                         wfProfileIn( __METHOD__.'-recache' );
 604                         // not in cache, or we need a fresh reload.
 605                         // we will first load the default tables
 606                         // then update them using things in MediaWiki:Zhconversiontable/*
 607                         $this->loadDefaultTables();
 608                         foreach($this->mVariants as $var) {
 609                                 $cached = $this->parseCachedTable($var);
 610                                 $this->mTables[$var]->mergeArray($cached);
 611                         }
 612
 613                         $this->postLoadTables();
 614                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 615
 616                         if($this->lockCache()) {
 617                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 618                                 $this->unlockCache();
 619                         }
 620                         wfProfileOut( __METHOD__.'-recache' );
 621                 }
 622                 wfProfileOut( __METHOD__ );
 623         }
 624
 625     /**
 626          * Hook for post processig after conversion tables are loaded
 627          *
 628          */
 629         function postLoadTables() {}
 630
 631     /**
 632          * Reload the conversion tables
 633          *
 634          * @private
 635          */
 636         function reloadTables() {
 637                 if($this->mTables)
 638                         unset($this->mTables);
 639                 $this->mTablesLoaded = false;
 640                 $this->loadTables(false);
 641         }
 642
 643
 644         /**
 645          * parse the conversion table stored in the cache
 646          *
 647          * the tables should be in blocks of the following form:
 648          *              -{
 649          *                      word => word ;
 650          *                      word => word ;
 651          *                      ...
 652          *              }-
 653          *
 654          *      to make the tables more manageable, subpages are allowed
 655          *      and will be parsed recursively if $recursive=true
 656          *
 657          */
 658         function parseCachedTable($code, $subpage='', $recursive=true) {
 659                 global $wgMessageCache;
 660                 static $parsed = array();
 661
 662                 if(!is_object($wgMessageCache))
 663                         return array();
 664
 665                 $key = 'Conversiontable/'.$code;
 666                 if($subpage)
 667                         $key .= '/' . $subpage;
 668
 669                 if(array_key_exists($key, $parsed))
 670                         return array();
 671
 672                 if ( strpos( $code, '/' ) === false ) {
 673                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 674                 } else {
 675                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 676                         if ( $title && $title->exists() ) {
 677                                 $article = new Article( $title );
 678                                 $txt = $article->getContents();
 679                         } else {
 680                                 $txt = '';
 681                         }
 682                 }
 683
 684                 // get all subpage links of the form
 685                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 686                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 687                 $subs = explode('[[', $txt);
 688                 $sublinks = array();
 689                 foreach( $subs as $sub ) {
 690                         $link = explode(']]', $sub, 2);
 691                         if(count($link) != 2)
 692                                 continue;
 693                         $b = explode('|', $link[0]);
 694                         $b = explode('/', trim($b[0]), 3);
 695                         if(count($b)==3)
 696                                 $sublink = $b[2];
 697                         else
 698                                 $sublink = '';
 699
 700                         if($b[0] == $linkhead && $b[1] == $code) {
 701                                 $sublinks[] = $sublink;
 702                         }
 703                 }
 704
 705
 706                 // parse the mappings in this page
 707                 $blocks = explode($this->mMarkup['begin'], $txt);
 708                 array_shift($blocks);
 709                 $ret = array();
 710                 foreach($blocks as $block) {
 711                         $mappings = explode($this->mMarkup['end'], $block, 2);
 712                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 713                         $table = explode( ';', $stripped );
 714                         foreach( $table as $t ) {
 715                                 $m = explode( '=>', $t );
 716                                 if( count( $m ) != 2)
 717                                         continue;
 718                                 // trim any trailling comments starting with '//'
 719                                 $tt = explode('//', $m[1], 2);
 720                                 $ret[trim($m[0])] = trim($tt[0]);
 721                         }
 722                 }
 723                 $parsed[$key] = true;
 724
 725
 726                 // recursively parse the subpages
 727                 if($recursive) {
 728                         foreach($sublinks as $link) {
 729                                 $s = $this->parseCachedTable($code, $link, $recursive);
 730                                 $ret = array_merge($ret, $s);
 731                         }
 732                 }
 733
 734                 if ($this->mUcfirst) {
 735                         foreach ($ret as $k => $v) {
 736                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 737                         }
 738                 }
 739                 return $ret;
 740         }
 741
 742         /**
 743          * Enclose a string with the "no conversion" tag. This is used by
 744          * various functions in the Parser
 745          *
 746          * @param string $text text to be tagged for no conversion
 747          * @return string the tagged text
 748          * @public
 749          */
 750         function markNoConversion($text, $noParse=false) {
 751                 # don't mark if already marked
 752                 if(strpos($text, $this->mMarkup['begin']) ||
 753                    strpos($text, $this->mMarkup['end']))
 754                         return $text;
 755
 756                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 757                 return $ret;
 758         }
 759
 760         /**
 761          * convert the sorting key for category links. this should make different
 762          * keys that are variants of each other map to the same key
 763          */
 764         function convertCategoryKey( $key ) {
 765                 return $key;
 766         }
 767         /**
 768          * hook to refresh the cache of conversion tables when
 769          * MediaWiki:conversiontable* is updated
 770          * @private
 771          */
 772         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 773                 $titleobj = $article->getTitle();
 774                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 775                         $title = $titleobj->getDBkey();
 776                         $t = explode('/', $title, 3);
 777                         $c = count($t);
 778                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 779                                 if(in_array($t[1], $this->mVariants)) {
 780                                         $this->reloadTables();
 781                                 }
 782                         }
 783                 }
 784                 return true;
 785         }
 786
 787         /**
 788          * Armour rendered math against conversion
 789          * Wrap math into rawoutput -{R| math }- syntax
 790          * @public
 791          */
 792         function armourMath($text){
 793                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 794                 return $ret;
 795         }
 796 }
 797
 798 /**
 799  * parser for rules of language conversion , parse rules in -{ }- tag
 800  * @ingroup Language
 801  * @author  fdcn <fdcn64@gmail.com>
 802  */
 803 class ConverterRule {
 804         var $mText; // original text in -{text}-
 805         var $mConverter; // LanguageConverter object
 806         var $mManualCodeError='<strong class="error">code error!</strong>';
 807         var $mRuleDisplay = '',$mRuleTitle=false;
 808         var $mRules = '';// string : the text of the rules
 809         var $mRulesAction = 'none';
 810         var $mFlags = array();
 811         var $mConvTable = array();
 812         var $mBidtable = array();// array of the translation in each variant
 813         var $mUnidtable = array();// array of the translation in each variant
 814
 815         /**
 816          * Constructor
 817          *
 818          * @param string $text the text between -{ and }-
 819          * @param object $converter a  LanguageConverter object
 820          * @access public
 821          */
 822         function __construct($text,$converter){
 823                 $this->mText = $text;
 824                 $this->mConverter=$converter;
 825                 foreach($converter->mVariants as $v){
 826                         $this->mConvTable[$v]=array();
 827                 }
 828         }
 829
 830         /**
 831          * check if variants array in convert array
 832          *
 833          * @param string $variant Variant language code
 834          * @return string Translated text
 835          * @public
 836          */
 837         function getTextInBidtable($variants){
 838                 if(is_string($variants)){ $variants=array($variants); }
 839                 if(!is_array($variants)) return false;
 840                 foreach ($variants as $variant){
 841                         if(array_key_exists($variant, $this->mBidtable)){
 842                                 return $this->mBidtable[$variant];
 843                         }
 844                 }
 845                 return false;
 846         }
 847
 848         /**
 849          * Parse flags with syntax -{FLAG| ... }-
 850          * @private
 851          */
 852         function parseFlags(){
 853                 $text = $this->mText;
 854                 if(strlen($text) < 2 ) {
 855                         $this->mFlags = array( 'R' );
 856                         $this->mRules = $text;
 857                         return;
 858                 }
 859
 860                 $flags = array();
 861                 $markup = $this->mConverter->mMarkup;
 862                 $validFlags = $this->mConverter->mFlags;
 863
 864                 $tt = explode($markup['flagsep'], $text, 2);
 865                 if(count($tt) == 2) {
 866                         $f = explode($markup['varsep'], $tt[0]);
 867                         foreach($f as $ff) {
 868                                 $ff = trim($ff);
 869                                 if(array_key_exists($ff, $validFlags) &&
 870                                                         !in_array($validFlags[$ff], $flags))
 871                                         $flags[] = $validFlags[$ff];
 872                         }
 873                         $rules = $tt[1];
 874                 } else {
 875                         $rules = $text;
 876                 }
 877
 878                 //check flags
 879                 if( in_array('R',$flags) ){
 880                         $flags = array('R');// remove other flags
 881                 } elseif ( in_array('N',$flags) ){
 882                         $flags = array('N');// remove other flags
 883                 } elseif ( in_array('-',$flags) ){
 884                         $flags = array('-');// remove other flags
 885                 } elseif (count($flags)==1 && $flags[0]=='T'){
 886                         $flags[]='H';
 887                 } elseif ( in_array('H',$flags) ){
 888                         // replace A flag, and remove other flags except T
 889                         $temp=array('+','H');
 890                         if(in_array('T',$flags)) $temp[] = 'T';
 891                         if(in_array('D',$flags)) $temp[] = 'D';
 892                         $flags = $temp;
 893                 } else {
 894                         if ( in_array('A',$flags)) {
 895                                 $flags[]='+';
 896                                 $flags[]='S';
 897                         }
 898                         if ( in_array('D',$flags) )
 899                                 $flags=array_diff($flags,array('S'));
 900                 }
 901                 if ( count($flags)==0 )
 902                         $flags = array('S');
 903                 $this->mRules=$rules;
 904                 $this->mFlags=$flags;
 905         }
 906
 907         /**
 908          * generate conversion table
 909          * @private
 910          */
 911         function parseRules() {
 912                 $rules = $this->mRules;
 913                 $flags = $this->mFlags;
 914                 $bidtable = array();
 915                 $unidtable = array();
 916                 $markup = $this->mConverter->mMarkup;
 917
 918                 $choice = explode($markup['varsep'], $rules );
 919                 foreach($choice as $c) {
 920                         $v = explode($markup['codesep'], $c);
 921                         if(count($v) != 2)
 922                                 continue;// syntax error, skip
 923                         $to=trim($v[1]);
 924                         $v=trim($v[0]);
 925                         $u = explode($markup['unidsep'], $v);
 926                         if(count($u) == 1) {
 927                                 $bidtable[$v] = $to;
 928                         } else if(count($u) == 2){
 929                                 $from=trim($u[0]);$v=trim($u[1]);
 930                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 931                                         $unidtable[$v]=array($from=>$to);
 932                                 else
 933                                         $unidtable[$v][$from]=$to;
 934                         }
 935                         // syntax error, pass
 936                         if (!array_key_exists($v,$this->mConverter->mVariantNames)){
 937                                 $bidtable = array();
 938                                 $unidtable = array();
 939                                 break;
 940                         }
 941                 }
 942                 $this->mBidtable = $bidtable;
 943                 $this->mUnidtable = $unidtable;
 944         }
 945
 946         /**
 947          * @private
 948          */
 949         function getRulesDesc(){
 950                 $codesep = $this->mConverter->mDescCodeSep;
 951                 $varsep = $this->mConverter->mDescVarSep;
 952                 $text='';
 953                 foreach($this->mBidtable as $k => $v)
 954                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
 955                 foreach($this->mUnidtable as $k => $a)
 956                         foreach($a as $from=>$to)
 957                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
 958                 return $text;
 959         }
 960
 961         /**
 962          *  Parse rules conversion
 963          * @private
 964          */
 965         function getRuleConvertedStr($variant,$doConvert){
 966                 $bidtable = $this->mBidtable;
 967                 $unidtable = $this->mUnidtable;
 968
 969                 if( count($bidtable) + count($unidtable) == 0 ){
 970                         return $this->mRules;
 971                 } elseif ($doConvert){// the text converted
 972                         // display current variant in bidirectional array
 973                         $disp = $this->getTextInBidtable($variant);
 974                         // or display current variant in fallbacks
 975                         if(!$disp)
 976                                 $disp = $this->getTextInBidtable(
 977                                                 $this->mConverter->getVariantFallbacks($variant));
 978                         // or display current variant in unidirectional array
 979                         if(!$disp && array_key_exists($variant,$unidtable)){
 980                                 $disp = array_values($unidtable[$variant]);
 981                                 $disp = $disp[0];
 982                         }
 983                         // or display frist text under disable manual convert
 984                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
 985                                 if(count($bidtable)>0){
 986                                         $disp = array_values($bidtable);
 987                                         $disp = $disp[0];
 988                                 } else {
 989                                         $disp = array_values($unidtable);
 990                                         $disp = array_values($disp[0]);
 991                                         $disp = $disp[0];
 992                                 }
 993                         }
 994                         return $disp;
 995                 } else {// no convert
 996                         return $this->mRules;
 997                 }
 998         }
 999
1000         /**
1001          * generate conversion table for all text
1002          * @private
1003          */
1004         function generateConvTable(){
1005                 $flags = $this->mFlags;
1006                 $bidtable = $this->mBidtable;
1007                 $unidtable = $this->mUnidtable;
1008                 $manLevel = $this->mConverter->mManualLevel;
1009
1010                 $vmarked=array();
1011                 foreach($this->mConverter->mVariants as $v) {
1012                         /* for bidirectional array
1013                                 fill in the missing variants, if any,
1014                                 with fallbacks */
1015                         if(!array_key_exists($v, $bidtable)) {
1016                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1017                                 $vf = $this->getTextInBidtable($variantFallbacks);
1018                                 if($vf) $bidtable[$v] = $vf;
1019                         }
1020
1021                         if(array_key_exists($v,$bidtable)){
1022                                 foreach($vmarked as $vo){
1023                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1024                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1025                                         // to introduce a custom mapping between
1026                                         // words WordZh and WordTw in the whole text
1027                                         if($manLevel[$v]=='bidirectional'){
1028                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1029                                         }
1030                                         if($manLevel[$vo]=='bidirectional'){
1031                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1032                                         }
1033                                 }
1034                                 $vmarked[]=$v;
1035                         }
1036                         /*for unidirectional array
1037                                 fill to convert tables */
1038                         $allow_unid = $manLevel[$v]=='bidirectional'
1039                                         || $manLevel[$v]=='unidirectional';
1040                         if($allow_unid && array_key_exists($v,$unidtable)){
1041                                 $ct=$this->mConvTable[$v];
1042                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1043                         }
1044                 }
1045         }
1046
1047         /**
1048          * Parse rules and flags
1049          * @public
1050          */
1051         function parse($variant){
1052                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1053
1054                 $this->parseFlags();
1055                 $flags = $this->mFlags;
1056
1057                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1058                         //FIXME: may cause trouble here...
1059                         //strip &nbsp; since it interferes with the parsing, plus,
1060                         //all spaces should be stripped in this tag anyway.
1061                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1062                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1063                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1064
1065                         $this->parseRules();
1066                 }
1067                 $rules = $this->mRules;
1068
1069                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1070                         if(in_array('+',$flags) || in_array('-',$flags))
1071                                 // fill all variants if text in -{A/H/-|text} without rules
1072                                 foreach($this->mConverter->mVariants as $v)
1073                                         $this->mBidtable[$v] = $rules;
1074                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1075                                 $this->mFlags = $flags = array('R');
1076                 }
1077
1078                 if( in_array('R',$flags) ) {
1079                         // if we don't do content convert, still strip the -{}- tags
1080                         $this->mRuleDisplay = $rules;
1081                 } elseif ( in_array('N',$flags) ){
1082                         // proces N flag: output current variant name
1083                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1084                 } elseif ( in_array('D',$flags) ){
1085                         // proces D flag: output rules description
1086                         $this->mRuleDisplay = $this->getRulesDesc();
1087                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1088                         // proces H,- flag or T only: output nothing
1089                         $this->mRuleDisplay = '';
1090                 } elseif ( in_array('S',$flags) ){
1091                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1092                                                         $this->mConverter->mDoContentConvert);
1093                 } else {
1094                         $this->mRuleDisplay= $this->mManualCodeError;
1095                 }
1096                 // proces T flag
1097                 if ( in_array('T',$flags) ) {
1098                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1099                                                         $this->mConverter->mDoTitleConvert);
1100                 }
1101
1102                 if (in_array('-', $flags))
1103                         $this->mRulesAction='remove';
1104                 if (in_array('+', $flags))
1105                         $this->mRulesAction='add';
1106
1107                 $this->generateConvTable();
1108         }
1109
1110         /**
1111          * @public
1112          */
1113         function hasRules(){
1114                 // TODO:
1115         }
1116
1117         /**
1118          * get display text on markup -{...}-
1119          * @public
1120          */
1121         function getDisplay(){
1122                 return $this->mRuleDisplay;
1123         }
1124         /**
1125          * get converted title
1126          * @public
1127          */
1128         function getTitle(){
1129                 return $this->mRuleTitle;
1130         }
1131
1132         /**
1133          * return how deal with conversion rules
1134          * @public
1135          */
1136         function getRulesAction(){
1137                 return $this->mRulesAction;
1138         }
1139
1140         /**
1141          * get conversion table ( bidirectional and unidirectional conversion table )
1142          * @public
1143          */
1144         function getConvTable(){
1145                 return $this->mConvTable;
1146         }
1147
1148         /**
1149          * get conversion rules string
1150          * @public
1151          */
1152         function getRules(){
1153                 return $this->mRules;
1154         }
1155
1156         /**
1157          * get conversion flags
1158          * @public
1159          */
1160         function getFlags(){
1161                 return $this->mFlags;
1162         }
1163 }