languages/LanguageConverter.php

   1 <?php
   2
   3 /**
   4  * Contains the LanguageConverter class and ConverterRule class
   5  * @ingroup Language
   6  *
   7  * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
   8  * @file
   9  */
  10
  11 /**
  12  * base class for language convert
  13  * @ingroup Language
  14  *
  15  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  16  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  17  */
  18 class LanguageConverter {
  19         var $mPreferredVariant='';
  20         var $mMainLanguageCode;
  21         var $mVariants, $mVariantFallbacks, $mLanguageNames;
  22         var $mTablesLoaded = false;
  23         var $mTables;
  24         var $mManualAddTables;
  25         var $mManualRemoveTables;
  26         var $mTitleDisplay='';
  27         var $mDoTitleConvert=true, $mDoContentConvert=true;
  28         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  29         var $mTitleFromFlag = false;
  30         var $mCacheKey;
  31         var $mLangObj;
  32         var $mMarkup;
  33         var $mFlags;
  34         var $mDescCodeSep = ':',$mDescVarSep = ';';
  35         var $mUcfirst = false;
  36
  37         const CACHE_VERSION_KEY = 'VERSION 6';
  38
  39         /**
  40          * Constructor
  41          *
  42          * @param string $maincode the main language code of this language
  43          * @param array $variants the supported variants of this language
  44          * @param array $variantfallback the fallback language of each variant
  45          * @param array $markup array defining the markup used for manual conversion
  46          * @param array $flags array defining the custom strings that maps to the flags
  47          * @param array $manualLevel limit for supported variants
  48          * @public
  49          */
  50         function __construct($langobj, $maincode,
  51                                                                 $variants=array(),
  52                                                                 $variantfallbacks=array(),
  53                                                                 $markup=array(),
  54                                                                 $flags = array(),
  55                                                                 $manualLevel = array() ) {
  56                 $this->mLangObj = $langobj;
  57                 $this->mMainLanguageCode = $maincode;
  58                 $this->mVariants = $variants;
  59                 $this->mVariantFallbacks = $variantfallbacks;
  60                 global $wgLanguageNames;
  61                 $this->mLanguageNames = $wgLanguageNames;
  62                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  63                 $m = array(
  64                         'begin'=>'-{',
  65                         'flagsep'=>'|',
  66                         'unidsep'=>'=>', //for unidirectional conversion
  67                         'codesep'=>':',
  68                         'varsep'=>';',
  69                         'end'=>'}-'
  70                 );
  71                 $this->mMarkup = array_merge($m, $markup);
  72                 $f = array(
  73                         // 'S' show converted text
  74                         // '+' add rules for alltext
  75                         // 'E' the gave flags is error
  76                         // these flags above are reserved for program
  77                         'A'=>'A',       // add rule for convert code (all text convert)
  78                         'T'=>'T',       // title convert
  79                         'R'=>'R',       // raw content
  80                         'D'=>'D',       // convert description (subclass implement)
  81                         '-'=>'-',       // remove convert (not implement)
  82                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  83                         'N'=>'N'        // current variant name
  84                 );
  85                 $this->mFlags = array_merge($f, $flags);
  86                 foreach( $this->mVariants as $v) {
  87                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  88                                                                 ?$manualLevel[$v]
  89                                                                 :'bidirectional';
  90                         $this->mManualAddTables[$v] = array();
  91                         $this->mManualRemoveTables[$v] = array();
  92                 }
  93         }
  94
  95         /**
  96          * @public
  97          */
  98         function getVariants() {
  99                 return $this->mVariants;
 100         }
 101
 102         /**
 103          * in case some variant is not defined in the markup, we need
 104          * to have some fallback. for example, in zh, normally people
 105          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 106          * when zh-sg is preferred but not defined, we will pick zh-hans
 107          * in this case. right now this is only used by zh.
 108          *
 109          * @param string $v the language code of the variant
 110          * @return string array the code of the fallback language or false if there is no fallback
 111          * @public
 112          */
 113         function getVariantFallbacks($v) {
 114                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 115                         return $this->mVariantFallbacks[$v];
 116                 }
 117                 return $this->mMainLanguageCode;
 118         }
 119
 120         // this method body in Language class
 121         function getPreferredVariant( $fromUser = true ) {
 122                 return $this->mLangObj->getPreferredVariant( $fromUser );
 123         }
 124
 125         /**
 126          * caption convert, base on preg_replace_callback
 127          *
 128          * to convert text in "title" or "alt", like '<img alt="text" ... '
 129          * or '<span title="text" ... '
 130          *
 131          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 132          * @private
 133          */
 134         function captionConvert( $matches ) {
 135                 $toVariant = $this->getPreferredVariant();
 136                 $title = $matches[1];
 137                 $text  = $matches[2];
 138                 // we convert captions except URL
 139                 if( !strpos( $text, '://' ) )
 140                         $text = $this->translate($text, $toVariant);
 141                 return " $title=\"$text\"";
 142         }
 143
 144         /**
 145          * dictionary-based conversion
 146          *
 147          * @param string $text the text to be converted
 148          * @param string $toVariant the target language code
 149          * @return string the converted text
 150          * @private
 151          */
 152         function autoConvert($text, $toVariant=false) {
 153                 $fname="LanguageConverter::autoConvert";
 154
 155                 wfProfileIn( $fname );
 156
 157                 if(!$this->mTablesLoaded)
 158                         $this->loadTables();
 159
 160                 if(!$toVariant)
 161                         $toVariant = $this->getPreferredVariant();
 162                 if(!in_array($toVariant, $this->mVariants))
 163                         return $text;
 164
 165                 /* we convert everything except:
 166                    1. html markups (anything between < and >)
 167                    2. html entities
 168                    3. place holders created by the parser
 169                 */
 170                 global $wgParser;
 171                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 172                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 173                 } else
 174                         $marker = "";
 175
 176                 // this one is needed when the text is inside an html markup
 177                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 178
 179                 // disable convert to variants between <code></code> tags
 180                 $codefix = '<code>.+?<\/code>|';
 181                 // disable convertsion of <script type="text/javascript"> ... </script>
 182                 $scriptfix = '<script.*?>.*?<\/script>|';
 183                 // disable conversion of <pre xxxx> ... </pre>
 184                 $prefix = '<pre.*?>.*?<\/pre>|';
 185
 186                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 187
 188                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 189
 190                 $m = array_shift($matches);
 191
 192                 $ret = $this->translate($m[0], $toVariant);
 193                 $mstart = $m[1]+strlen($m[0]);
 194
 195                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 196                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 197                 foreach($matches as $m) {
 198                         $mark = substr($text, $mstart, $m[1]-$mstart);
 199                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 200                         $ret .= $mark;
 201                         $ret .= $this->translate($m[0], $toVariant);
 202                         $mstart = $m[1] + strlen($m[0]);
 203                 }
 204                 wfProfileOut( $fname );
 205                 return $ret;
 206         }
 207
 208         /**
 209          * Translate a string to a variant
 210          * Doesn't process markup or do any of that other stuff, for that use convert()
 211          *
 212          * @param string $text Text to convert
 213          * @param string $variant Variant language code
 214          * @return string Translated text
 215          * @private
 216          */
 217         function translate( $text, $variant ) {
 218                 wfProfileIn( __METHOD__ );
 219                 if( !$this->mTablesLoaded )
 220                         $this->loadTables();
 221                 $text = $this->mTables[$variant]->replace( $text );
 222                 wfProfileOut( __METHOD__ );
 223                 return $text;
 224         }
 225
 226         /**
 227          * convert text to all supported variants
 228          *
 229          * @param string $text the text to be converted
 230          * @return array of string
 231          * @public
 232          */
 233         function autoConvertToAllVariants($text) {
 234                 $fname="LanguageConverter::autoConvertToAllVariants";
 235                 wfProfileIn( $fname );
 236                 if( !$this->mTablesLoaded )
 237                         $this->loadTables();
 238
 239                 $ret = array();
 240                 foreach($this->mVariants as $variant) {
 241                         $ret[$variant] = $this->translate($text, $variant);
 242                 }
 243
 244                 wfProfileOut( $fname );
 245                 return $ret;
 246         }
 247
 248         /**
 249          * convert link text to all supported variants
 250          *
 251          * @param string $text the text to be converted
 252          * @return array of string
 253          * @public
 254          */
 255         function convertLinkToAllVariants($text) {
 256                 if( !$this->mTablesLoaded )
 257                         $this->loadTables();
 258
 259                 $ret = array();
 260                 $tarray = explode($this->mMarkup['begin'], $text);
 261                 $tfirst = array_shift($tarray);
 262
 263                 foreach($this->mVariants as $variant)
 264                         $ret[$variant] = $this->translate($tfirst,$variant);
 265
 266                 foreach($tarray as $txt) {
 267                         $marked = explode($this->mMarkup['end'], $txt, 2);
 268
 269                         foreach($this->mVariants as $variant){
 270                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 271                                 if(array_key_exists(1, $marked))
 272                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 273                         }
 274
 275                 }
 276
 277                 return $ret;
 278         }
 279
 280         /**
 281          * prepare manual conversion table
 282          * @private
 283          */
 284         function prepareManualConv($convRule){
 285                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 286                 $title = $convRule->getTitle();
 287                 if($title){
 288                         $this->mTitleFromFlag = true;
 289                         $this->mTitleDisplay =  $title;
 290                 }
 291
 292                 //apply manual conversion table to global table
 293                 $convTable = $convRule->getConvTable();
 294                 $action = $convRule->getRulesAction();
 295                 foreach($convTable as $v=>$t) {
 296                         if( !in_array($v,$this->mVariants) )continue;
 297                         if( $action=="add" ) {
 298                                 foreach($t as $from=>$to) {
 299                                         // to ensure that $from and $to not be left blank
 300                                         // so $this->translate() could always return a string
 301                                         if ($from || $to)
 302                                                 // more efficient than array_merge(), about 2.5 times.
 303                                                 $this->mManualAddTables[$v][$from] = $to;
 304                                 }
 305                         }
 306                         elseif ( $action=="remove" ) {
 307                                 foreach($t as $from=>$to) {
 308                                         if ($from || $to)
 309                                                 $this->mManualRemoveTables[$v][$from] = $to;
 310                                 }
 311                         }
 312                 }
 313         }
 314
 315         /**
 316          * apply manual conversion from $this->mManualAddTables and $this->mManualRemoveTables
 317          * @private
 318          */
 319         function applyManualConv(){
 320                 //apply manual conversion table to global table
 321                 foreach($this->mVariants as $v) {
 322                         if (count($this->mManualAddTables[$v]) > 0) {
 323                                 $this->mTables[$v]->mergeArray($this->mManualAddTables[$v]);
 324                         }
 325                         if (count($this->mManualRemoveTables[$v]) > 0)
 326                                 $this->mTables[$v]->removeArray($this->mManualRemoveTables[$v]);
 327                 }
 328         }
 329
 330         /**
 331          * Convert text using a parser object for context
 332          * @public
 333          */
 334         function parserConvert( $text, &$parser ) {
 335                 global $wgDisableLangConversion;
 336                 /* don't do anything if this is the conversion table */
 337                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 338                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 339                 {
 340                         return $text;
 341                 }
 342
 343                 if($wgDisableLangConversion)
 344                         return $text;
 345
 346                 $text = $this->convert( $text );
 347                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 348                 return $text;
 349         }
 350
 351         /**
 352          *  convert title
 353          * @private
 354          */
 355         function convertTitle($text){
 356                 global $wgDisableTitleConversion, $wgUser;
 357
 358                 // check for global param and __NOTC__ tag
 359                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 360                         $this->mTitleDisplay = $text;
 361                         return $text;
 362                 }
 363
 364                 // use the title from the T flag if any
 365                 if($this->mTitleFromFlag){
 366                         $this->mTitleFromFlag = false;
 367                         return $this->mTitleDisplay;
 368                 }
 369
 370                 global $wgRequest;
 371                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 372                 $action = $wgRequest->getText( 'action' );
 373                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 374                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 375                         return $text;
 376                 } else {
 377                         $this->mTitleDisplay = $this->convert($text);
 378                         return $this->mTitleDisplay;
 379                 }
 380         }
 381
 382         /**
 383          * convert text to different variants of a language. the automatic
 384          * conversion is done in autoConvert(). here we parse the text
 385          * marked with -{}-, which specifies special conversions of the
 386          * text that can not be accomplished in autoConvert()
 387          *
 388          * syntax of the markup:
 389          * -{code1:text1;code2:text2;...}-  or
 390          * -{flags|code1:text1;code2:text2;...}-  or
 391          * -{text}- in which case no conversion should take place for text
 392          *
 393          * @param string $text text to be converted
 394          * @param bool $isTitle whether this conversion is for the article title
 395          * @return string converted text
 396          * @public
 397          */
 398         function convert( $text , $isTitle=false) {
 399
 400                 $mw =& MagicWord::get( 'notitleconvert' );
 401                 if( $mw->matchAndRemove( $text ) )
 402                         $this->mDoTitleConvert = false;
 403                 $mw =& MagicWord::get( 'nocontentconvert' );
 404                 if( $mw->matchAndRemove( $text ) ) {
 405                         $this->mDoContentConvert = false;
 406                 }
 407
 408                 // no conversion if redirecting
 409                 $mw =& MagicWord::get( 'redirect' );
 410                 if( $mw->matchStart( $text ))
 411                         return $text;
 412
 413                 // for title convertion
 414                 if ($isTitle) return $this->convertTitle($text);
 415
 416                 $plang = $this->getPreferredVariant();
 417                 $tarray = StringUtils::explode($this->mMarkup['end'], $text);
 418                 $text = '';
 419
 420                 $marks = array();
 421                 foreach($tarray as $txt) {
 422                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 423                         if (array_key_exists(1, $marked)) {
 424                                 $crule = new ConverterRule($marked[1], $this);
 425                                 $crule->parse($plang);
 426                                 $marked[1] = $crule->getDisplay();
 427                                 $this->prepareManualConv($crule);
 428                         }
 429                         else
 430                                 $marked[0] .= $this->mMarkup['end'];
 431                         array_push($marks, $marked);
 432                 }
 433                 $this->applyManualConv();
 434                 foreach ($marks as $marked) {
 435                         if( $this->mDoContentConvert )
 436                                 $text .= $this->autoConvert($marked[0],$plang);
 437                         else
 438                                 $text .= $marked[0];
 439                         if( array_key_exists(1, $marked) )
 440                                 $text .= $marked[1];
 441                 }
 442                 // Remove the last delimiter (wasn't real)
 443                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 444
 445                 return $text;
 446         }
 447
 448         /**
 449          * if a language supports multiple variants, it is
 450          * possible that non-existing link in one variant
 451          * actually exists in another variant. this function
 452          * tries to find it. See e.g. LanguageZh.php
 453          *
 454          * @param string $link the name of the link
 455          * @param mixed $nt the title object of the link
 456          * @return null the input parameters may be modified upon return
 457          * @public
 458          */
 459         function findVariantLink( &$link, &$nt, $forTemplate = false ) {
 460                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 461                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 462                 $action = $wgRequest->getText( 'action' );
 463                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 464                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 465                 $linkBatch = new LinkBatch();
 466
 467                 $ns=NS_MAIN;
 468
 469                 if ( $disableLinkConversion || ( !$forTemplate && ( $isredir == 'no' || $action == 'edit'
 470                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) ) {
 471                         return;
 472                 }
 473
 474                 if(is_object($nt))
 475                         $ns = $nt->getNamespace();
 476
 477                 $variants = $this->autoConvertToAllVariants($link);
 478                 if($variants == false) //give up
 479                         return;
 480
 481                 $titles = array();
 482
 483                 foreach( $variants as $v ) {
 484                         if($v != $link){
 485                                 $varnt = Title::newFromText( $v, $ns );
 486                                 if(!is_null($varnt)){
 487                                         $linkBatch->addObj($varnt);
 488                                         $titles[]=$varnt;
 489                                 }
 490                         }
 491                 }
 492
 493                 // fetch all variants in single query
 494                 $linkBatch->execute();
 495
 496                 foreach( $titles as $varnt ) {
 497                         if( $varnt->getArticleID() > 0 ) {
 498                                 $nt = $varnt;
 499                                 $link = $v;
 500                                 break;
 501                         }
 502                 }
 503         }
 504
 505     /**
 506          * returns language specific hash options
 507          *
 508          * @public
 509          */
 510         function getExtraHashOptions() {
 511                 $variant = $this->getPreferredVariant();
 512                 return '!' . $variant ;
 513         }
 514
 515     /**
 516          * get title text as defined in the body of the article text
 517          *
 518          * @public
 519          */
 520         function getParsedTitle() {
 521                 return $this->mTitleDisplay;
 522         }
 523
 524         /**
 525          * a write lock to the cache
 526          *
 527          * @private
 528          */
 529         function lockCache() {
 530                 global $wgMemc;
 531                 $success = false;
 532                 for($i=0; $i<30; $i++) {
 533                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 534                                 break;
 535                         sleep(1);
 536                 }
 537                 return $success;
 538         }
 539
 540         /**
 541          * unlock cache
 542          *
 543          * @private
 544          */
 545         function unlockCache() {
 546                 global $wgMemc;
 547                 $wgMemc->delete($this->mCacheKey . "lock");
 548         }
 549
 550
 551         /**
 552          * Load default conversion tables
 553          * This method must be implemented in derived class
 554          *
 555          * @private
 556          */
 557         function loadDefaultTables() {
 558                 $name = get_class($this);
 559                 wfDie("Must implement loadDefaultTables() method in class $name");
 560         }
 561
 562         /**
 563          * load conversion tables either from the cache or the disk
 564          * @private
 565          */
 566         function loadTables($fromcache=true) {
 567                 global $wgMemc;
 568                 if( $this->mTablesLoaded )
 569                         return;
 570                 wfProfileIn( __METHOD__ );
 571                 $this->mTablesLoaded = true;
 572                 $this->mTables = false;
 573                 if($fromcache) {
 574                         wfProfileIn( __METHOD__.'-cache' );
 575                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 576                         wfProfileOut( __METHOD__.'-cache' );
 577                 }
 578                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 579                         wfProfileIn( __METHOD__.'-recache' );
 580                         // not in cache, or we need a fresh reload.
 581                         // we will first load the default tables
 582                         // then update them using things in MediaWiki:Zhconversiontable/*
 583                         $this->loadDefaultTables();
 584                         foreach($this->mVariants as $var) {
 585                                 $cached = $this->parseCachedTable($var);
 586                                 $this->mTables[$var]->mergeArray($cached);
 587                         }
 588
 589                         $this->postLoadTables();
 590                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 591
 592                         if($this->lockCache()) {
 593                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 594                                 $this->unlockCache();
 595                         }
 596                         wfProfileOut( __METHOD__.'-recache' );
 597                 }
 598                 wfProfileOut( __METHOD__ );
 599         }
 600
 601     /**
 602          * Hook for post processig after conversion tables are loaded
 603          *
 604          */
 605         function postLoadTables() {}
 606
 607     /**
 608          * Reload the conversion tables
 609          *
 610          * @private
 611          */
 612         function reloadTables() {
 613                 if($this->mTables)
 614                         unset($this->mTables);
 615                 $this->mTablesLoaded = false;
 616                 $this->loadTables(false);
 617         }
 618
 619
 620         /**
 621          * parse the conversion table stored in the cache
 622          *
 623          * the tables should be in blocks of the following form:
 624          *              -{
 625          *                      word => word ;
 626          *                      word => word ;
 627          *                      ...
 628          *              }-
 629          *
 630          *      to make the tables more manageable, subpages are allowed
 631          *      and will be parsed recursively if $recursive=true
 632          *
 633          */
 634         function parseCachedTable($code, $subpage='', $recursive=true) {
 635                 global $wgMessageCache;
 636                 static $parsed = array();
 637
 638                 if(!is_object($wgMessageCache))
 639                         return array();
 640
 641                 $key = 'Conversiontable/'.$code;
 642                 if($subpage)
 643                         $key .= '/' . $subpage;
 644
 645                 if(array_key_exists($key, $parsed))
 646                         return array();
 647
 648                 if ( strpos( $code, '/' ) === false ) {
 649                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 650                 } else {
 651                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 652                         if ( $title && $title->exists() ) {
 653                                 $article = new Article( $title );
 654                                 $txt = $article->getContents();
 655                         } else {
 656                                 $txt = '';
 657                         }
 658                 }
 659
 660                 // get all subpage links of the form
 661                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 662                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 663                 $subs = explode('[[', $txt);
 664                 $sublinks = array();
 665                 foreach( $subs as $sub ) {
 666                         $link = explode(']]', $sub, 2);
 667                         if(count($link) != 2)
 668                                 continue;
 669                         $b = explode('|', $link[0]);
 670                         $b = explode('/', trim($b[0]), 3);
 671                         if(count($b)==3)
 672                                 $sublink = $b[2];
 673                         else
 674                                 $sublink = '';
 675
 676                         if($b[0] == $linkhead && $b[1] == $code) {
 677                                 $sublinks[] = $sublink;
 678                         }
 679                 }
 680
 681
 682                 // parse the mappings in this page
 683                 $blocks = explode($this->mMarkup['begin'], $txt);
 684                 array_shift($blocks);
 685                 $ret = array();
 686                 foreach($blocks as $block) {
 687                         $mappings = explode($this->mMarkup['end'], $block, 2);
 688                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 689                         $table = explode( ';', $stripped );
 690                         foreach( $table as $t ) {
 691                                 $m = explode( '=>', $t );
 692                                 if( count( $m ) != 2)
 693                                         continue;
 694                                 // trim any trailling comments starting with '//'
 695                                 $tt = explode('//', $m[1], 2);
 696                                 $ret[trim($m[0])] = trim($tt[0]);
 697                         }
 698                 }
 699                 $parsed[$key] = true;
 700
 701
 702                 // recursively parse the subpages
 703                 if($recursive) {
 704                         foreach($sublinks as $link) {
 705                                 $s = $this->parseCachedTable($code, $link, $recursive);
 706                                 $ret = array_merge($ret, $s);
 707                         }
 708                 }
 709
 710                 if ($this->mUcfirst) {
 711                         foreach ($ret as $k => $v) {
 712                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 713                         }
 714                 }
 715                 return $ret;
 716         }
 717
 718         /**
 719          * Enclose a string with the "no conversion" tag. This is used by
 720          * various functions in the Parser
 721          *
 722          * @param string $text text to be tagged for no conversion
 723          * @return string the tagged text
 724          * @public
 725          */
 726         function markNoConversion($text, $noParse=false) {
 727                 # don't mark if already marked
 728                 if(strpos($text, $this->mMarkup['begin']) ||
 729                    strpos($text, $this->mMarkup['end']))
 730                         return $text;
 731
 732                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 733                 return $ret;
 734         }
 735
 736         /**
 737          * convert the sorting key for category links. this should make different
 738          * keys that are variants of each other map to the same key
 739          */
 740         function convertCategoryKey( $key ) {
 741                 return $key;
 742         }
 743         /**
 744          * hook to refresh the cache of conversion tables when
 745          * MediaWiki:conversiontable* is updated
 746          * @private
 747          */
 748         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 749                 $titleobj = $article->getTitle();
 750                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 751                         $title = $titleobj->getDBkey();
 752                         $t = explode('/', $title, 3);
 753                         $c = count($t);
 754                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 755                                 if(in_array($t[1], $this->mVariants)) {
 756                                         $this->reloadTables();
 757                                 }
 758                         }
 759                 }
 760                 return true;
 761         }
 762
 763         /**
 764          * Armour rendered math against conversion
 765          * Wrap math into rawoutput -{R| math }- syntax
 766          * @public
 767          */
 768         function armourMath($text){
 769                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 770                 return $ret;
 771         }
 772 }
 773
 774 /**
 775  * parser for rules of language conversion , parse rules in -{ }- tag
 776  * @ingroup Language
 777  * @author  fdcn <fdcn64@gmail.com>
 778  */
 779 class ConverterRule {
 780         var $mText; // original text in -{text}-
 781         var $mConverter; // LanguageConverter object
 782         var $mManualCodeError='<strong class="error">code error!</strong>';
 783         var $mRuleDisplay = '',$mRuleTitle=false;
 784         var $mRules = '';// string : the text of the rules
 785         var $mRulesAction = 'none';
 786         var $mFlags = array();
 787         var $mConvTable = array();
 788         var $mBidtable = array();// array of the translation in each variant
 789         var $mUnidtable = array();// array of the translation in each variant
 790
 791         /**
 792          * Constructor
 793          *
 794          * @param string $text the text between -{ and }-
 795          * @param object $converter a  LanguageConverter object
 796          * @access public
 797          */
 798         function __construct($text,$converter){
 799                 $this->mText = $text;
 800                 $this->mConverter=$converter;
 801                 foreach($converter->mVariants as $v){
 802                         $this->mConvTable[$v]=array();
 803                 }
 804         }
 805
 806         /**
 807          * check if variants array in convert array
 808          *
 809          * @param string $variant Variant language code
 810          * @return string Translated text
 811          * @public
 812          */
 813         function getTextInBidtable($variants){
 814                 if(is_string($variants)){ $variants=array($variants); }
 815                 if(!is_array($variants)) return false;
 816                 foreach ($variants as $variant){
 817                         if(array_key_exists($variant, $this->mBidtable)){
 818                                 return $this->mBidtable[$variant];
 819                         }
 820                 }
 821                 return false;
 822         }
 823
 824         /**
 825          * Parse flags with syntax -{FLAG| ... }-
 826          * @private
 827          */
 828         function parseFlags(){
 829                 $text = $this->mText;
 830                 if(strlen($text) < 2 ) {
 831                         $this->mFlags = array( 'R' );
 832                         $this->mRules = $text;
 833                         return;
 834                 }
 835
 836                 $flags = array();
 837                 $markup = $this->mConverter->mMarkup;
 838                 $validFlags = $this->mConverter->mFlags;
 839
 840                 $tt = explode($markup['flagsep'], $text, 2);
 841                 if(count($tt) == 2) {
 842                         $f = explode($markup['varsep'], $tt[0]);
 843                         foreach($f as $ff) {
 844                                 $ff = trim($ff);
 845                                 if(array_key_exists($ff, $validFlags) &&
 846                                                         !in_array($validFlags[$ff], $flags))
 847                                         $flags[] = $validFlags[$ff];
 848                         }
 849                         $rules = $tt[1];
 850                 } else {
 851                         $rules = $text;
 852                 }
 853
 854                 //check flags
 855                 if( in_array('R',$flags) ){
 856                         $flags = array('R');// remove other flags
 857                 } elseif ( in_array('N',$flags) ){
 858                         $flags = array('N');// remove other flags
 859                 } elseif ( in_array('-',$flags) ){
 860                         $flags = array('-');// remove other flags
 861                 } elseif (count($flags)==1 && $flags[0]=='T'){
 862                         $flags[]='H';
 863                 } elseif ( in_array('H',$flags) ){
 864                         // replace A flag, and remove other flags except T
 865                         $temp=array('+','H');
 866                         if(in_array('T',$flags)) $temp[] = 'T';
 867                         if(in_array('D',$flags)) $temp[] = 'D';
 868                         $flags = $temp;
 869                 } else {
 870                         if ( in_array('A',$flags)) {
 871                                 $flags[]='+';
 872                                 $flags[]='S';
 873                         }
 874                         if ( in_array('D',$flags) )
 875                                 $flags=array_diff($flags,array('S'));
 876                 }
 877                 if ( count($flags)==0 )
 878                         $flags = array('S');
 879                 $this->mRules=$rules;
 880                 $this->mFlags=$flags;
 881         }
 882
 883         /**
 884          * generate conversion table
 885          * @private
 886          */
 887         function parseRules() {
 888                 $rules = $this->mRules;
 889                 $flags = $this->mFlags;
 890                 $bidtable = array();
 891                 $unidtable = array();
 892                 $markup = $this->mConverter->mMarkup;
 893
 894                 $choice = explode($markup['varsep'], $rules );
 895                 foreach($choice as $c) {
 896                         $v = explode($markup['codesep'], $c);
 897                         if(count($v) != 2)
 898                                 continue;// syntax error, skip
 899                         $to=trim($v[1]);
 900                         $v=trim($v[0]);
 901                         $u = explode($markup['unidsep'], $v);
 902                         if(count($u) == 1) {
 903                                 $bidtable[$v] = $to;
 904                         } else if(count($u) == 2){
 905                                 $from=trim($u[0]);$v=trim($u[1]);
 906                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 907                                         $unidtable[$v]=array($from=>$to);
 908                                 else
 909                                         $unidtable[$v][$from]=$to;
 910                         }
 911                         // syntax error, pass
 912                         if (!array_key_exists($v,$this->mConverter->mLanguageNames)){
 913                                 $bidtable = array();
 914                                 $unidtable = array();
 915                                 break;
 916                         }
 917                 }
 918                 $this->mBidtable = $bidtable;
 919                 $this->mUnidtable = $unidtable;
 920         }
 921
 922         /**
 923          * @private
 924          */
 925         function getRulesDesc(){
 926                 $codesep = $this->mConverter->mDescCodeSep;
 927                 $varsep = $this->mConverter->mDescVarSep;
 928                 $text='';
 929                 foreach($this->mBidtable as $k => $v)
 930                         $text .= $this->mConverter->mLanguageNames[$k]."$codesep$v$varsep";
 931                 foreach($this->mUnidtable as $k => $a)
 932                         foreach($a as $from=>$to)
 933                                 $text.="$from&#8658;".$this->mConverter->mLanguageNames[$k]."$codesep$to$varsep";
 934                 return $text;
 935         }
 936
 937         /**
 938          *  Parse rules conversion
 939          * @private
 940          */
 941         function getRuleConvertedStr($variant,$doConvert){
 942                 $bidtable = $this->mBidtable;
 943                 $unidtable = $this->mUnidtable;
 944
 945                 if( count($bidtable) + count($unidtable) == 0 ){
 946                         return $this->mRules;
 947                 } elseif ($doConvert){// the text converted
 948                         // display current variant in bidirectional array
 949                         $disp = $this->getTextInBidtable($variant);
 950                         // or display current variant in fallbacks
 951                         if(!$disp)
 952                                 $disp = $this->getTextInBidtable(
 953                                                 $this->mConverter->getVariantFallbacks($variant));
 954                         // or display current variant in unidirectional array
 955                         if(!$disp && array_key_exists($variant,$unidtable)){
 956                                 $disp = array_values($unidtable[$variant]);
 957                                 $disp = $disp[0];
 958                         }
 959                         // or display frist text under disable manual convert
 960                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
 961                                 if(count($bidtable)>0){
 962                                         $disp = array_values($bidtable);
 963                                         $disp = $disp[0];
 964                                 } else {
 965                                         $disp = array_values($unidtable);
 966                                         $disp = array_values($disp[0]);
 967                                         $disp = $disp[0];
 968                                 }
 969                         }
 970                         return $disp;
 971                 } else {// no convert
 972                         return $this->mRules;
 973                 }
 974         }
 975
 976         /**
 977          * generate conversion table for all text
 978          * @private
 979          */
 980         function generateConvTable(){
 981                 $flags = $this->mFlags;
 982                 $bidtable = $this->mBidtable;
 983                 $unidtable = $this->mUnidtable;
 984                 $manLevel = $this->mConverter->mManualLevel;
 985
 986                 $vmarked=array();
 987                 foreach($this->mConverter->mVariants as $v) {
 988                         /* for bidirectional array
 989                                 fill in the missing variants, if any,
 990                                 with fallbacks */
 991                         if(!array_key_exists($v, $bidtable)) {
 992                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
 993                                 $vf = $this->getTextInBidtable($variantFallbacks);
 994                                 if($vf) $bidtable[$v] = $vf;
 995                         }
 996
 997                         if(array_key_exists($v,$bidtable)){
 998                                 foreach($vmarked as $vo){
 999                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1000                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1001                                         // to introduce a custom mapping between
1002                                         // words WordZh and WordTw in the whole text
1003                                         if($manLevel[$v]=='bidirectional'){
1004                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1005                                         }
1006                                         if($manLevel[$vo]=='bidirectional'){
1007                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1008                                         }
1009                                 }
1010                                 $vmarked[]=$v;
1011                         }
1012                         /*for unidirectional array
1013                                 fill to convert tables */
1014                         $allow_unid = $manLevel[$v]=='bidirectional'
1015                                         || $manLevel[$v]=='unidirectional';
1016                         if($allow_unid && array_key_exists($v,$unidtable)){
1017                                 $ct=$this->mConvTable[$v];
1018                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1019                         }
1020                 }
1021         }
1022
1023         /**
1024          * Parse rules and flags
1025          * @public
1026          */
1027         function parse($variant){
1028                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1029
1030                 $this->parseFlags();
1031                 $flags = $this->mFlags;
1032
1033                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1034                         //FIXME: may cause trouble here...
1035                         //strip &nbsp; since it interferes with the parsing, plus,
1036                         //all spaces should be stripped in this tag anyway.
1037                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1038                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1039                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1040
1041                         $this->parseRules();
1042                 }
1043                 $rules = $this->mRules;
1044
1045                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1046                         if(in_array('+',$flags) || in_array('-',$flags))
1047                                 // fill all variants if text in -{A/H/-|text} without rules
1048                                 foreach($this->mConverter->mVariants as $v)
1049                                         $this->mBidtable[$v] = $rules;
1050                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1051                                 $this->mFlags = $flags = array('R');
1052                 }
1053
1054                 if( in_array('R',$flags) ) {
1055                         // if we don't do content convert, still strip the -{}- tags
1056                         $this->mRuleDisplay = $rules;
1057                 } elseif ( in_array('N',$flags) ){
1058                         // proces N flag: output current variant name
1059                         $this->mRuleDisplay = $this->mConverter->mLanguageNames[trim($rules)];
1060                 } elseif ( in_array('D',$flags) ){
1061                         // proces D flag: output rules description
1062                         $this->mRuleDisplay = $this->getRulesDesc();
1063                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1064                         // proces H,- flag or T only: output nothing
1065                         $this->mRuleDisplay = '';
1066                 } elseif ( in_array('S',$flags) ){
1067                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1068                                                         $this->mConverter->mDoContentConvert);
1069                 } else {
1070                         $this->mRuleDisplay= $this->mManualCodeError;
1071                 }
1072                 // proces T flag
1073                 if ( in_array('T',$flags) ) {
1074                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1075                                                         $this->mConverter->mDoTitleConvert);
1076                 }
1077
1078                 if (in_array('-', $flags))
1079                         $this->mRulesAction='remove';
1080                 if (in_array('+', $flags))
1081                         $this->mRulesAction='add';
1082
1083                 $this->generateConvTable();
1084         }
1085
1086         /**
1087          * @public
1088          */
1089         function hasRules(){
1090                 // TODO:
1091         }
1092
1093         /**
1094          * get display text on markup -{...}-
1095          * @public
1096          */
1097         function getDisplay(){
1098                 return $this->mRuleDisplay;
1099         }
1100         /**
1101          * get converted title
1102          * @public
1103          */
1104         function getTitle(){
1105                 return $this->mRuleTitle;
1106         }
1107
1108         /**
1109          * return how deal with conversion rules
1110          * @public
1111          */
1112         function getRulesAction(){
1113                 return $this->mRulesAction;
1114         }
1115
1116         /**
1117          * get conversion table ( bidirectional and unidirectional conversion table )
1118          * @public
1119          */
1120         function getConvTable(){
1121                 return $this->mConvTable;
1122         }
1123
1124         /**
1125          * get conversion rules string
1126          * @public
1127          */
1128         function getRules(){
1129                 return $this->mRules;
1130         }
1131
1132         /**
1133          * get conversion flags
1134          * @public
1135          */
1136         function getFlags(){
1137                 return $this->mFlags;
1138         }
1139 }