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