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, $mVariantNames;
  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->mVariantNames = $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         /**
 121          * get preferred language variants.
 122          * @param boolean $fromUser Get it from $wgUser's preferences
 123          * @return string the preferred language code
 124          * @public
 125          */
 126         function getPreferredVariant( $fromUser = true ) {
 127                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 128
 129                 if($this->mPreferredVariant)
 130                         return $this->mPreferredVariant;
 131
 132                 // figure out user lang without constructing wgLang to avoid infinite recursion
 133                 if( $fromUser )
 134                         $defaultUserLang = $wgUser->getOption( 'language' );
 135                 else
 136                         $defaultUserLang = $this->mMainLanguageCode;
 137                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 138                 // see if interface language is same as content, if not, prevent conversion
 139                 if( ! in_array( $userLang, $this->mVariants ) ){
 140                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 141                         return $this->mPreferredVariant;
 142                 }
 143
 144                 // see if the preference is set in the request
 145                 $req = $wgRequest->getText( 'variant' );
 146                 if( in_array( $req, $this->mVariants ) ) {
 147                         $this->mPreferredVariant = $req;
 148                         return $req;
 149                 }
 150
 151                 // check the syntax /code/ArticleTitle
 152                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 153                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 154                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 155                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 156                         if(in_array($scriptBase,$this->mVariants)){
 157                                 $this->mPreferredVariant = $scriptBase;
 158                                 return $this->mPreferredVariant;
 159                         }
 160                 }
 161
 162                 // get language variant preference from logged in users
 163                 // Don't call this on stub objects because that causes infinite
 164                 // recursion during initialisation
 165                 if( $fromUser && $wgUser->isLoggedIn() )  {
 166                         $this->mPreferredVariant = $wgUser->getOption('variant');
 167                         return $this->mPreferredVariant;
 168                 }
 169
 170                 // see if default variant is globaly set
 171                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 172                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 173                         return $this->mPreferredVariant;
 174                 }
 175
 176                 # FIXME rewrite code for parsing http header. The current code
 177                 # is written specific for detecting zh- variants
 178                 if( !$this->mPreferredVariant ) {
 179                         // see if some supported language variant is set in the
 180                         // http header, but we don't set the mPreferredVariant
 181                         // variable in case this is called before the user's
 182                         // preference is loaded
 183                         $pv=$this->mMainLanguageCode;
 184                         if(array_key_exists('HTTP_ACCEPT_LANGUAGE', $_SERVER)) {
 185                                 $header = str_replace( '_', '-', strtolower($_SERVER["HTTP_ACCEPT_LANGUAGE"]));
 186                                 $zh = strstr($header, $pv.'-');
 187                                 if($zh) {
 188                                         $ary = split("[,;]",$zh);
 189                                         $pv = $ary[0];
 190                                 }
 191                         }
 192                         // don't try to return bad variant
 193                         if(in_array( $pv, $this->mVariants ))
 194                                 return $pv;
 195                 }
 196
 197                 return $this->mMainLanguageCode;
 198
 199         }
 200
 201         /**
 202          * caption convert, base on preg_replace_callback
 203          *
 204          * to convert text in "title" or "alt", like '<img alt="text" ... '
 205          * or '<span title="text" ... '
 206          *
 207          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 208          * @private
 209          */
 210         function captionConvert( $matches ) {
 211                 // we convert captions except URL
 212                 $toVariant = $this->getPreferredVariant();
 213                 $title = $matches[1];
 214                 $text  = $matches[2];
 215                 if( !strpos( $text, '://' ) )
 216                         $text = $this->translate($text, $toVariant);
 217                 return " $title=\"$text\"";
 218         }
 219
 220         /**
 221          * dictionary-based conversion
 222          *
 223          * @param string $text the text to be converted
 224          * @param string $toVariant the target language code
 225          * @return string the converted text
 226          * @private
 227          */
 228         function autoConvert($text, $toVariant=false) {
 229                 $fname="LanguageConverter::autoConvert";
 230
 231                 wfProfileIn( $fname );
 232
 233                 if(!$this->mTablesLoaded)
 234                         $this->loadTables();
 235
 236                 if(!$toVariant)
 237                         $toVariant = $this->getPreferredVariant();
 238                 if(!in_array($toVariant, $this->mVariants))
 239                         return $text;
 240
 241                 /* we convert everything except:
 242                    1. html markups (anything between < and >)
 243                    2. html entities
 244                    3. place holders created by the parser
 245                 */
 246                 global $wgParser;
 247                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 248                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 249                 } else
 250                         $marker = "";
 251
 252                 // this one is needed when the text is inside an html markup
 253                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 254
 255                 // disable convert to variants between <code></code> tags
 256                 $codefix = '<code>.+?<\/code>|';
 257                 // disable convertsion of <script type="text/javascript"> ... </script>
 258                 $scriptfix = '<script.*?>.*?<\/script>|';
 259                 // disable conversion of <pre xxxx> ... </pre>
 260                 $prefix = '<pre.*?>.*?<\/pre>|';
 261
 262                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 263
 264                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 265
 266                 $m = array_shift($matches);
 267
 268                 $ret = $this->translate($m[0], $toVariant);
 269                 $mstart = $m[1]+strlen($m[0]);
 270
 271                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 272                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 273                 foreach($matches as $m) {
 274                         $mark = substr($text, $mstart, $m[1]-$mstart);
 275                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 276                         $ret .= $mark;
 277                         $ret .= $this->translate($m[0], $toVariant);
 278                         $mstart = $m[1] + strlen($m[0]);
 279                 }
 280                 wfProfileOut( $fname );
 281                 return $ret;
 282         }
 283
 284         /**
 285          * Translate a string to a variant
 286          * Doesn't process markup or do any of that other stuff, for that use convert()
 287          *
 288          * @param string $text Text to convert
 289          * @param string $variant Variant language code
 290          * @return string Translated text
 291          * @private
 292          */
 293         function translate( $text, $variant ) {
 294                 wfProfileIn( __METHOD__ );
 295                 if( !$this->mTablesLoaded )
 296                         $this->loadTables();
 297                 $text = $this->mTables[$variant]->replace( $text );
 298                 wfProfileOut( __METHOD__ );
 299                 return $text;
 300         }
 301
 302         /**
 303          * convert text to all supported variants
 304          *
 305          * @param string $text the text to be converted
 306          * @return array of string
 307          * @public
 308          */
 309         function autoConvertToAllVariants($text) {
 310                 $fname="LanguageConverter::autoConvertToAllVariants";
 311                 wfProfileIn( $fname );
 312                 if( !$this->mTablesLoaded )
 313                         $this->loadTables();
 314
 315                 $ret = array();
 316                 foreach($this->mVariants as $variant) {
 317                         $ret[$variant] = $this->translate($text, $variant);
 318                 }
 319
 320                 wfProfileOut( $fname );
 321                 return $ret;
 322         }
 323
 324         /**
 325          * convert link text to all supported variants
 326          *
 327          * @param string $text the text to be converted
 328          * @return array of string
 329          * @public
 330          */
 331         function convertLinkToAllVariants($text) {
 332                 if( !$this->mTablesLoaded )
 333                         $this->loadTables();
 334
 335                 $ret = array();
 336                 $tarray = explode($this->mMarkup['begin'], $text);
 337                 $tfirst = array_shift($tarray);
 338
 339                 foreach($this->mVariants as $variant)
 340                         $ret[$variant] = $this->translate($tfirst,$variant);
 341
 342                 foreach($tarray as $txt) {
 343                         $marked = explode($this->mMarkup['end'], $txt, 2);
 344
 345                         foreach($this->mVariants as $variant){
 346                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 347                                 if(array_key_exists(1, $marked))
 348                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 349                         }
 350
 351                 }
 352
 353                 return $ret;
 354         }
 355
 356         /**
 357          * prepare manual conversion table
 358          * @private
 359          */
 360         function prepareManualConv($convRule){
 361                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 362                 $title = $convRule->getTitle();
 363                 if($title){
 364                         $this->mTitleFromFlag = true;
 365                         $this->mTitleDisplay =  $title;
 366                 }
 367
 368                 //apply manual conversion table to global table
 369                 $convTable = $convRule->getConvTable();
 370                 $action = $convRule->getRulesAction();
 371                 foreach($convTable as $v=>$t) {
 372                         if( !in_array($v,$this->mVariants) )continue;
 373                         if( $action=="add" ) {
 374                                 foreach($t as $from=>$to) {
 375                                         // more efficient than array_merge(), about 2.5 times.
 376                                         $this->mManualAddTables[$v][$from] = $to;
 377                                 }
 378                         }
 379                         elseif ( $action=="remove" )
 380                                 $this->mManualRemoveTables[$v] = array_merge($this->mManualRemoveTables[$v], $t);
 381                 }
 382         }
 383
 384         /**
 385          * apply manual conversion from $this->mManualAddTables and $this->mManualRemoveTables
 386          * @private
 387          */
 388         function applyManualConv(){
 389                 //apply manual conversion table to global table
 390                 foreach($this->mVariants as $v) {
 391                         if (count($this->mManualAddTables[$v]) > 0) {
 392                                 $this->mTables[$v]->mergeArray($this->mManualAddTables[$v]);
 393                         }
 394                         if (count($this->mManualRemoveTables[$v]) > 0)
 395                                 $this->mTables[$v]->removeArray($this->mManualRemoveTables[$v]);
 396                 }
 397         }
 398
 399         /**
 400          * Convert text using a parser object for context
 401          * @public
 402          */
 403         function parserConvert( $text, &$parser ) {
 404                 global $wgDisableLangConversion;
 405                 /* don't do anything if this is the conversion table */
 406                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 407                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 408                 {
 409                         return $text;
 410                 }
 411
 412                 if($wgDisableLangConversion)
 413                         return $text;
 414
 415                 $text = $this->convert( $text );
 416                 $parser->mOutput->setTitleText( $this->mTitleDisplay );
 417                 return $text;
 418         }
 419
 420         /**
 421          *  convert title
 422          * @private
 423          */
 424         function convertTitle($text){
 425                 global $wgDisableTitleConversion, $wgUser;
 426
 427                 // check for global param and __NOTC__ tag
 428                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 429                         $this->mTitleDisplay = $text;
 430                         return $text;
 431                 }
 432
 433                 // use the title from the T flag if any
 434                 if($this->mTitleFromFlag){
 435                         $this->mTitleFromFlag = false;
 436                         return $this->mTitleDisplay;
 437                 }
 438
 439                 global $wgRequest;
 440                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 441                 $action = $wgRequest->getText( 'action' );
 442                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 443                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 444                         return $text;
 445                 } else {
 446                         $this->mTitleDisplay = $this->convert($text);
 447                         return $this->mTitleDisplay;
 448                 }
 449         }
 450
 451         /**
 452          * convert text to different variants of a language. the automatic
 453          * conversion is done in autoConvert(). here we parse the text
 454          * marked with -{}-, which specifies special conversions of the
 455          * text that can not be accomplished in autoConvert()
 456          *
 457          * syntax of the markup:
 458          * -{code1:text1;code2:text2;...}-  or
 459          * -{flags|code1:text1;code2:text2;...}-  or
 460          * -{text}- in which case no conversion should take place for text
 461          *
 462          * @param string $text text to be converted
 463          * @param bool $isTitle whether this conversion is for the article title
 464          * @return string converted text
 465          * @public
 466          */
 467         function convert( $text , $isTitle=false) {
 468
 469                 $mw =& MagicWord::get( 'notitleconvert' );
 470                 if( $mw->matchAndRemove( $text ) )
 471                         $this->mDoTitleConvert = false;
 472                 $mw =& MagicWord::get( 'nocontentconvert' );
 473                 if( $mw->matchAndRemove( $text ) ) {
 474                         $this->mDoContentConvert = false;
 475                 }
 476
 477                 // no conversion if redirecting
 478                 $mw =& MagicWord::get( 'redirect' );
 479                 if( $mw->matchStart( $text ))
 480                         return $text;
 481
 482                 // for title convertion
 483                 if ($isTitle) return $this->convertTitle($text);
 484
 485                 $plang = $this->getPreferredVariant();
 486                 $tarray = StringUtils::explode($this->mMarkup['end'], $text);
 487                 $text = '';
 488
 489                 $marks = array();
 490                 foreach($tarray as $txt) {
 491                         $marked = explode($this->mMarkup['begin'], $txt, 2);
 492                         if (array_key_exists(1, $marked)) {
 493                                 $crule = new ConverterRule($marked[1], $this);
 494                                 $crule->parse($plang);
 495                                 $marked[1] = $crule->getDisplay();
 496                                 $this->prepareManualConv($crule);
 497                         }
 498                         else
 499                                 $marked[0] .= $this->mMarkup['end'];
 500                         array_push($marks, $marked);
 501                 }
 502                 $this->applyManualConv();
 503                 foreach ($marks as $marked) {
 504                         if( $this->mDoContentConvert )
 505                                 $text .= $this->autoConvert($marked[0],$plang);
 506                         else
 507                                 $text .= $marked[0];
 508                         if( array_key_exists(1, $marked) )
 509                                 $text .= $marked[1];
 510                 }
 511                 // Remove the last delimiter (wasn't real)
 512                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 513
 514                 return $text;
 515         }
 516
 517         /**
 518          * if a language supports multiple variants, it is
 519          * possible that non-existing link in one variant
 520          * actually exists in another variant. this function
 521          * tries to find it. See e.g. LanguageZh.php
 522          *
 523          * @param string $link the name of the link
 524          * @param mixed $nt the title object of the link
 525          * @return null the input parameters may be modified upon return
 526          * @public
 527          */
 528         function findVariantLink( &$link, &$nt, $forTemplate = false ) {
 529                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 530                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 531                 $action = $wgRequest->getText( 'action' );
 532                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 533                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 534                 $linkBatch = new LinkBatch();
 535
 536                 $ns=NS_MAIN;
 537
 538                 if ( $disableLinkConversion || ( !$forTemplate && ( $isredir == 'no' || $action == 'edit'
 539                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) ) {
 540                         return;
 541                 }
 542
 543                 if(is_object($nt))
 544                         $ns = $nt->getNamespace();
 545
 546                 $variants = $this->autoConvertToAllVariants($link);
 547                 if($variants == false) //give up
 548                         return;
 549
 550                 $titles = array();
 551
 552                 foreach( $variants as $v ) {
 553                         if($v != $link){
 554                                 $varnt = Title::newFromText( $v, $ns );
 555                                 if(!is_null($varnt)){
 556                                         $linkBatch->addObj($varnt);
 557                                         $titles[]=$varnt;
 558                                 }
 559                         }
 560                 }
 561
 562                 // fetch all variants in single query
 563                 $linkBatch->execute();
 564
 565                 foreach( $titles as $varnt ) {
 566                         if( $varnt->getArticleID() > 0 ) {
 567                                 $nt = $varnt;
 568                                 $link = $v;
 569                                 break;
 570                         }
 571                 }
 572         }
 573
 574     /**
 575          * returns language specific hash options
 576          *
 577          * @public
 578          */
 579         function getExtraHashOptions() {
 580                 $variant = $this->getPreferredVariant();
 581                 return '!' . $variant ;
 582         }
 583
 584     /**
 585          * get title text as defined in the body of the article text
 586          *
 587          * @public
 588          */
 589         function getParsedTitle() {
 590                 return $this->mTitleDisplay;
 591         }
 592
 593         /**
 594          * a write lock to the cache
 595          *
 596          * @private
 597          */
 598         function lockCache() {
 599                 global $wgMemc;
 600                 $success = false;
 601                 for($i=0; $i<30; $i++) {
 602                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 603                                 break;
 604                         sleep(1);
 605                 }
 606                 return $success;
 607         }
 608
 609         /**
 610          * unlock cache
 611          *
 612          * @private
 613          */
 614         function unlockCache() {
 615                 global $wgMemc;
 616                 $wgMemc->delete($this->mCacheKey . "lock");
 617         }
 618
 619
 620         /**
 621          * Load default conversion tables
 622          * This method must be implemented in derived class
 623          *
 624          * @private
 625          */
 626         function loadDefaultTables() {
 627                 $name = get_class($this);
 628                 wfDie("Must implement loadDefaultTables() method in class $name");
 629         }
 630
 631         /**
 632          * load conversion tables either from the cache or the disk
 633          * @private
 634          */
 635         function loadTables($fromcache=true) {
 636                 global $wgMemc;
 637                 if( $this->mTablesLoaded )
 638                         return;
 639                 wfProfileIn( __METHOD__ );
 640                 $this->mTablesLoaded = true;
 641                 $this->mTables = false;
 642                 if($fromcache) {
 643                         wfProfileIn( __METHOD__.'-cache' );
 644                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 645                         wfProfileOut( __METHOD__.'-cache' );
 646                 }
 647                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 648                         wfProfileIn( __METHOD__.'-recache' );
 649                         // not in cache, or we need a fresh reload.
 650                         // we will first load the default tables
 651                         // then update them using things in MediaWiki:Zhconversiontable/*
 652                         $this->loadDefaultTables();
 653                         foreach($this->mVariants as $var) {
 654                                 $cached = $this->parseCachedTable($var);
 655                                 $this->mTables[$var]->mergeArray($cached);
 656                         }
 657
 658                         $this->postLoadTables();
 659                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 660
 661                         if($this->lockCache()) {
 662                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 663                                 $this->unlockCache();
 664                         }
 665                         wfProfileOut( __METHOD__.'-recache' );
 666                 }
 667                 wfProfileOut( __METHOD__ );
 668         }
 669
 670     /**
 671          * Hook for post processig after conversion tables are loaded
 672          *
 673          */
 674         function postLoadTables() {}
 675
 676     /**
 677          * Reload the conversion tables
 678          *
 679          * @private
 680          */
 681         function reloadTables() {
 682                 if($this->mTables)
 683                         unset($this->mTables);
 684                 $this->mTablesLoaded = false;
 685                 $this->loadTables(false);
 686         }
 687
 688
 689         /**
 690          * parse the conversion table stored in the cache
 691          *
 692          * the tables should be in blocks of the following form:
 693          *              -{
 694          *                      word => word ;
 695          *                      word => word ;
 696          *                      ...
 697          *              }-
 698          *
 699          *      to make the tables more manageable, subpages are allowed
 700          *      and will be parsed recursively if $recursive=true
 701          *
 702          */
 703         function parseCachedTable($code, $subpage='', $recursive=true) {
 704                 global $wgMessageCache;
 705                 static $parsed = array();
 706
 707                 if(!is_object($wgMessageCache))
 708                         return array();
 709
 710                 $key = 'Conversiontable/'.$code;
 711                 if($subpage)
 712                         $key .= '/' . $subpage;
 713
 714                 if(array_key_exists($key, $parsed))
 715                         return array();
 716
 717                 if ( strpos( $code, '/' ) === false ) {
 718                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 719                 } else {
 720                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 721                         if ( $title && $title->exists() ) {
 722                                 $article = new Article( $title );
 723                                 $txt = $article->getContents();
 724                         } else {
 725                                 $txt = '';
 726                         }
 727                 }
 728
 729                 // get all subpage links of the form
 730                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 731                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 732                 $subs = explode('[[', $txt);
 733                 $sublinks = array();
 734                 foreach( $subs as $sub ) {
 735                         $link = explode(']]', $sub, 2);
 736                         if(count($link) != 2)
 737                                 continue;
 738                         $b = explode('|', $link[0]);
 739                         $b = explode('/', trim($b[0]), 3);
 740                         if(count($b)==3)
 741                                 $sublink = $b[2];
 742                         else
 743                                 $sublink = '';
 744
 745                         if($b[0] == $linkhead && $b[1] == $code) {
 746                                 $sublinks[] = $sublink;
 747                         }
 748                 }
 749
 750
 751                 // parse the mappings in this page
 752                 $blocks = explode($this->mMarkup['begin'], $txt);
 753                 array_shift($blocks);
 754                 $ret = array();
 755                 foreach($blocks as $block) {
 756                         $mappings = explode($this->mMarkup['end'], $block, 2);
 757                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 758                         $table = explode( ';', $stripped );
 759                         foreach( $table as $t ) {
 760                                 $m = explode( '=>', $t );
 761                                 if( count( $m ) != 2)
 762                                         continue;
 763                                 // trim any trailling comments starting with '//'
 764                                 $tt = explode('//', $m[1], 2);
 765                                 $ret[trim($m[0])] = trim($tt[0]);
 766                         }
 767                 }
 768                 $parsed[$key] = true;
 769
 770
 771                 // recursively parse the subpages
 772                 if($recursive) {
 773                         foreach($sublinks as $link) {
 774                                 $s = $this->parseCachedTable($code, $link, $recursive);
 775                                 $ret = array_merge($ret, $s);
 776                         }
 777                 }
 778
 779                 if ($this->mUcfirst) {
 780                         foreach ($ret as $k => $v) {
 781                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 782                         }
 783                 }
 784                 return $ret;
 785         }
 786
 787         /**
 788          * Enclose a string with the "no conversion" tag. This is used by
 789          * various functions in the Parser
 790          *
 791          * @param string $text text to be tagged for no conversion
 792          * @return string the tagged text
 793          * @public
 794          */
 795         function markNoConversion($text, $noParse=false) {
 796                 # don't mark if already marked
 797                 if(strpos($text, $this->mMarkup['begin']) ||
 798                    strpos($text, $this->mMarkup['end']))
 799                         return $text;
 800
 801                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 802                 return $ret;
 803         }
 804
 805         /**
 806          * convert the sorting key for category links. this should make different
 807          * keys that are variants of each other map to the same key
 808          */
 809         function convertCategoryKey( $key ) {
 810                 return $key;
 811         }
 812         /**
 813          * hook to refresh the cache of conversion tables when
 814          * MediaWiki:conversiontable* is updated
 815          * @private
 816          */
 817         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 818                 $titleobj = $article->getTitle();
 819                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 820                         $title = $titleobj->getDBkey();
 821                         $t = explode('/', $title, 3);
 822                         $c = count($t);
 823                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 824                                 if(in_array($t[1], $this->mVariants)) {
 825                                         $this->reloadTables();
 826                                 }
 827                         }
 828                 }
 829                 return true;
 830         }
 831
 832         /**
 833          * Armour rendered math against conversion
 834          * Wrap math into rawoutput -{R| math }- syntax
 835          * @public
 836          */
 837         function armourMath($text){
 838                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 839                 return $ret;
 840         }
 841 }
 842
 843 /**
 844  * parser for rules of language conversion , parse rules in -{ }- tag
 845  * @ingroup Language
 846  * @author  fdcn <fdcn64@gmail.com>
 847  */
 848 class ConverterRule {
 849         var $mText; // original text in -{text}-
 850         var $mConverter; // LanguageConverter object
 851         var $mManualCodeError='<strong class="error">code error!</strong>';
 852         var $mRuleDisplay = '',$mRuleTitle=false;
 853         var $mRules = '';// string : the text of the rules
 854         var $mRulesAction = 'none';
 855         var $mFlags = array();
 856         var $mConvTable = array();
 857         var $mBidtable = array();// array of the translation in each variant
 858         var $mUnidtable = array();// array of the translation in each variant
 859
 860         /**
 861          * Constructor
 862          *
 863          * @param string $text the text between -{ and }-
 864          * @param object $converter a  LanguageConverter object
 865          * @access public
 866          */
 867         function __construct($text,$converter){
 868                 $this->mText = $text;
 869                 $this->mConverter=$converter;
 870                 foreach($converter->mVariants as $v){
 871                         $this->mConvTable[$v]=array();
 872                 }
 873         }
 874
 875         /**
 876          * check if variants array in convert array
 877          *
 878          * @param string $variant Variant language code
 879          * @return string Translated text
 880          * @public
 881          */
 882         function getTextInBidtable($variants){
 883                 if(is_string($variants)){ $variants=array($variants); }
 884                 if(!is_array($variants)) return false;
 885                 foreach ($variants as $variant){
 886                         if(array_key_exists($variant, $this->mBidtable)){
 887                                 return $this->mBidtable[$variant];
 888                         }
 889                 }
 890                 return false;
 891         }
 892
 893         /**
 894          * Parse flags with syntax -{FLAG| ... }-
 895          * @private
 896          */
 897         function parseFlags(){
 898                 $text = $this->mText;
 899                 if(strlen($text) < 2 ) {
 900                         $this->mFlags = array( 'R' );
 901                         $this->mRules = $text;
 902                         return;
 903                 }
 904
 905                 $flags = array();
 906                 $markup = $this->mConverter->mMarkup;
 907                 $validFlags = $this->mConverter->mFlags;
 908
 909                 $tt = explode($markup['flagsep'], $text, 2);
 910                 if(count($tt) == 2) {
 911                         $f = explode($markup['varsep'], $tt[0]);
 912                         foreach($f as $ff) {
 913                                 $ff = trim($ff);
 914                                 if(array_key_exists($ff, $validFlags) &&
 915                                                         !in_array($validFlags[$ff], $flags))
 916                                         $flags[] = $validFlags[$ff];
 917                         }
 918                         $rules = $tt[1];
 919                 } else {
 920                         $rules = $text;
 921                 }
 922
 923                 //check flags
 924                 if( in_array('R',$flags) ){
 925                         $flags = array('R');// remove other flags
 926                 } elseif ( in_array('N',$flags) ){
 927                         $flags = array('N');// remove other flags
 928                 } elseif ( in_array('-',$flags) ){
 929                         $flags = array('-');// remove other flags
 930                 } elseif (count($flags)==1 && $flags[0]=='T'){
 931                         $flags[]='H';
 932                 } elseif ( in_array('H',$flags) ){
 933                         // replace A flag, and remove other flags except T
 934                         $temp=array('+','H');
 935                         if(in_array('T',$flags)) $temp[] = 'T';
 936                         if(in_array('D',$flags)) $temp[] = 'D';
 937                         $flags = $temp;
 938                 } else {
 939                         if ( in_array('A',$flags)) {
 940                                 $flags[]='+';
 941                                 $flags[]='S';
 942                         }
 943                         if ( in_array('D',$flags) )
 944                                 $flags=array_diff($flags,array('S'));
 945                 }
 946                 if ( count($flags)==0 )
 947                         $flags = array('S');
 948                 $this->mRules=$rules;
 949                 $this->mFlags=$flags;
 950         }
 951
 952         /**
 953          * generate conversion table
 954          * @private
 955          */
 956         function parseRules() {
 957                 $rules = $this->mRules;
 958                 $flags = $this->mFlags;
 959                 $bidtable = array();
 960                 $unidtable = array();
 961                 $markup = $this->mConverter->mMarkup;
 962
 963                 $choice = explode($markup['varsep'], $rules );
 964                 foreach($choice as $c) {
 965                         $v = explode($markup['codesep'], $c);
 966                         if(count($v) != 2)
 967                                 continue;// syntax error, skip
 968                         $to=trim($v[1]);
 969                         $v=trim($v[0]);
 970                         $u = explode($markup['unidsep'], $v);
 971                         if(count($u) == 1) {
 972                                 $bidtable[$v] = $to;
 973                         } else if(count($u) == 2){
 974                                 $from=trim($u[0]);$v=trim($u[1]);
 975                                 if( array_key_exists($v,$unidtable) && !is_array($unidtable[$v]) )
 976                                         $unidtable[$v]=array($from=>$to);
 977                                 else
 978                                         $unidtable[$v][$from]=$to;
 979                         }
 980                         // syntax error, pass
 981                         if (!array_key_exists($v,$this->mConverter->mVariantNames)){
 982                                 $bidtable = array();
 983                                 $unidtable = array();
 984                                 break;
 985                         }
 986                 }
 987                 $this->mBidtable = $bidtable;
 988                 $this->mUnidtable = $unidtable;
 989         }
 990
 991         /**
 992          * @private
 993          */
 994         function getRulesDesc(){
 995                 $codesep = $this->mConverter->mDescCodeSep;
 996                 $varsep = $this->mConverter->mDescVarSep;
 997                 $text='';
 998                 foreach($this->mBidtable as $k => $v)
 999                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1000                 foreach($this->mUnidtable as $k => $a)
1001                         foreach($a as $from=>$to)
1002                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1003                 return $text;
1004         }
1005
1006         /**
1007          *  Parse rules conversion
1008          * @private
1009          */
1010         function getRuleConvertedStr($variant,$doConvert){
1011                 $bidtable = $this->mBidtable;
1012                 $unidtable = $this->mUnidtable;
1013
1014                 if( count($bidtable) + count($unidtable) == 0 ){
1015                         return $this->mRules;
1016                 } elseif ($doConvert){// the text converted
1017                         // display current variant in bidirectional array
1018                         $disp = $this->getTextInBidtable($variant);
1019                         // or display current variant in fallbacks
1020                         if(!$disp)
1021                                 $disp = $this->getTextInBidtable(
1022                                                 $this->mConverter->getVariantFallbacks($variant));
1023                         // or display current variant in unidirectional array
1024                         if(!$disp && array_key_exists($variant,$unidtable)){
1025                                 $disp = array_values($unidtable[$variant]);
1026                                 $disp = $disp[0];
1027                         }
1028                         // or display frist text under disable manual convert
1029                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1030                                 if(count($bidtable)>0){
1031                                         $disp = array_values($bidtable);
1032                                         $disp = $disp[0];
1033                                 } else {
1034                                         $disp = array_values($unidtable);
1035                                         $disp = array_values($disp[0]);
1036                                         $disp = $disp[0];
1037                                 }
1038                         }
1039                         return $disp;
1040                 } else {// no convert
1041                         return $this->mRules;
1042                 }
1043         }
1044
1045         /**
1046          * generate conversion table for all text
1047          * @private
1048          */
1049         function generateConvTable(){
1050                 $flags = $this->mFlags;
1051                 $bidtable = $this->mBidtable;
1052                 $unidtable = $this->mUnidtable;
1053                 $manLevel = $this->mConverter->mManualLevel;
1054
1055                 $vmarked=array();
1056                 foreach($this->mConverter->mVariants as $v) {
1057                         /* for bidirectional array
1058                                 fill in the missing variants, if any,
1059                                 with fallbacks */
1060                         if(!array_key_exists($v, $bidtable)) {
1061                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1062                                 $vf = $this->getTextInBidtable($variantFallbacks);
1063                                 if($vf) $bidtable[$v] = $vf;
1064                         }
1065
1066                         if(array_key_exists($v,$bidtable)){
1067                                 foreach($vmarked as $vo){
1068                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1069                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1070                                         // to introduce a custom mapping between
1071                                         // words WordZh and WordTw in the whole text
1072                                         if($manLevel[$v]=='bidirectional'){
1073                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1074                                         }
1075                                         if($manLevel[$vo]=='bidirectional'){
1076                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1077                                         }
1078                                 }
1079                                 $vmarked[]=$v;
1080                         }
1081                         /*for unidirectional array
1082                                 fill to convert tables */
1083                         $allow_unid = $manLevel[$v]=='bidirectional'
1084                                         || $manLevel[$v]=='unidirectional';
1085                         if($allow_unid && array_key_exists($v,$unidtable)){
1086                                 $ct=$this->mConvTable[$v];
1087                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1088                         }
1089                 }
1090         }
1091
1092         /**
1093          * Parse rules and flags
1094          * @public
1095          */
1096         function parse($variant){
1097                 if(!$variant) $variant = $this->mConverter->getPreferredVariant();
1098
1099                 $this->parseFlags();
1100                 $flags = $this->mFlags;
1101
1102                 if( !in_array('R',$flags) || !in_array('N',$flags) ){
1103                         //FIXME: may cause trouble here...
1104                         //strip &nbsp; since it interferes with the parsing, plus,
1105                         //all spaces should be stripped in this tag anyway.
1106                         $this->mRules = str_replace('&nbsp;', '', $this->mRules);
1107                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1108                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1109
1110                         $this->parseRules();
1111                 }
1112                 $rules = $this->mRules;
1113
1114                 if(count($this->mBidtable)==0 && count($this->mUnidtable)==0){
1115                         if(in_array('+',$flags) || in_array('-',$flags))
1116                                 // fill all variants if text in -{A/H/-|text} without rules
1117                                 foreach($this->mConverter->mVariants as $v)
1118                                         $this->mBidtable[$v] = $rules;
1119                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1120                                 $this->mFlags = $flags = array('R');
1121                 }
1122
1123                 if( in_array('R',$flags) ) {
1124                         // if we don't do content convert, still strip the -{}- tags
1125                         $this->mRuleDisplay = $rules;
1126                 } elseif ( in_array('N',$flags) ){
1127                         // proces N flag: output current variant name
1128                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1129                 } elseif ( in_array('D',$flags) ){
1130                         // proces D flag: output rules description
1131                         $this->mRuleDisplay = $this->getRulesDesc();
1132                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1133                         // proces H,- flag or T only: output nothing
1134                         $this->mRuleDisplay = '';
1135                 } elseif ( in_array('S',$flags) ){
1136                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1137                                                         $this->mConverter->mDoContentConvert);
1138                 } else {
1139                         $this->mRuleDisplay= $this->mManualCodeError;
1140                 }
1141                 // proces T flag
1142                 if ( in_array('T',$flags) ) {
1143                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1144                                                         $this->mConverter->mDoTitleConvert);
1145                 }
1146
1147                 if (in_array('-', $flags))
1148                         $this->mRulesAction='remove';
1149                 if (in_array('+', $flags))
1150                         $this->mRulesAction='add';
1151
1152                 $this->generateConvTable();
1153         }
1154
1155         /**
1156          * @public
1157          */
1158         function hasRules(){
1159                 // TODO:
1160         }
1161
1162         /**
1163          * get display text on markup -{...}-
1164          * @public
1165          */
1166         function getDisplay(){
1167                 return $this->mRuleDisplay;
1168         }
1169         /**
1170          * get converted title
1171          * @public
1172          */
1173         function getTitle(){
1174                 return $this->mRuleTitle;
1175         }
1176
1177         /**
1178          * return how deal with conversion rules
1179          * @public
1180          */
1181         function getRulesAction(){
1182                 return $this->mRulesAction;
1183         }
1184
1185         /**
1186          * get conversion table ( bidirectional and unidirectional conversion table )
1187          * @public
1188          */
1189         function getConvTable(){
1190                 return $this->mConvTable;
1191         }
1192
1193         /**
1194          * get conversion rules string
1195          * @public
1196          */
1197         function getRules(){
1198                 return $this->mRules;
1199         }
1200
1201         /**
1202          * get conversion flags
1203          * @public
1204          */
1205         function getFlags(){
1206                 return $this->mFlags;
1207         }
1208 }