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 $mNamespaceTables;
  25         var $mTitleDisplay='';
  26         var $mDoTitleConvert=true, $mDoContentConvert=true;
  27         var $mManualLevel; // 'bidirectional' 'unidirectional' 'disable' for each variants
  28         var $mTitleFromFlag = false;
  29         var $mCacheKey;
  30         var $mLangObj;
  31         var $mMarkup;
  32         var $mFlags;
  33         var $mDescCodeSep = ':',$mDescVarSep = ';';
  34         var $mUcfirst = false;
  35
  36         const CACHE_VERSION_KEY = 'VERSION 6';
  37
  38         /**
  39          * Constructor
  40          *
  41          * @param string $maincode the main language code of this language
  42          * @param array $variants the supported variants of this language
  43          * @param array $variantfallback the fallback language of each variant
  44          * @param array $markup array defining the markup used for manual conversion
  45          * @param array $flags array defining the custom strings that maps to the flags
  46          * @param array $manualLevel limit for supported variants
  47          * @public
  48          */
  49         function __construct($langobj, $maincode,
  50                                                                 $variants=array(),
  51                                                                 $variantfallbacks=array(),
  52                                                                 $markup=array(),
  53                                                                 $flags = array(),
  54                                                                 $manualLevel = array() ) {
  55                 $this->mLangObj = $langobj;
  56                 $this->mMainLanguageCode = $maincode;
  57                 $this->mVariants = $variants;
  58                 $this->mVariantFallbacks = $variantfallbacks;
  59                 global $wgLanguageNames;
  60                 $this->mVariantNames = $wgLanguageNames;
  61                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  62                 $m = array(
  63                         'begin'=>'-{',
  64                         'flagsep'=>'|',
  65                         'unidsep'=>'=>', //for unidirectional conversion
  66                         'codesep'=>':',
  67                         'varsep'=>';',
  68                         'end'=>'}-'
  69                 );
  70                 $this->mMarkup = array_merge($m, $markup);
  71                 $f = array(
  72                         // 'S' show converted text
  73                         // '+' add rules for alltext
  74                         // 'E' the gave flags is error
  75                         // these flags above are reserved for program
  76                         'A'=>'A',       // add rule for convert code (all text convert)
  77                         'T'=>'T',       // title convert
  78                         'R'=>'R',       // raw content
  79                         'D'=>'D',       // convert description (subclass implement)
  80                         '-'=>'-',       // remove convert (not implement)
  81                         'H'=>'H',       // add rule for convert code (but no display in placed code )
  82                         'N'=>'N'        // current variant name
  83                 );
  84                 $this->mFlags = array_merge($f, $flags);
  85                 foreach( $this->mVariants as $v) {
  86                         $this->mManualLevel[$v]=array_key_exists($v,$manualLevel)
  87                                                                 ?$manualLevel[$v]
  88                                                                 :'bidirectional';
  89                         $this->mNamespaceTables[$v] = array();
  90                         $this->mFlags[$v] = $v;
  91                 }
  92         }
  93
  94         /**
  95          * @public
  96          */
  97         function getVariants() {
  98                 return $this->mVariants;
  99         }
 100
 101         /**
 102          * in case some variant is not defined in the markup, we need
 103          * to have some fallback. for example, in zh, normally people
 104          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 105          * when zh-sg is preferred but not defined, we will pick zh-hans
 106          * in this case. right now this is only used by zh.
 107          *
 108          * @param string $v the language code of the variant
 109          * @return string array the code of the fallback language or false if there is no fallback
 110          * @public
 111          */
 112         function getVariantFallbacks($v) {
 113                 if( isset( $this->mVariantFallbacks[$v] ) ) {
 114                         return $this->mVariantFallbacks[$v];
 115                 }
 116                 return $this->mMainLanguageCode;
 117         }
 118
 119         /**
 120          * get preferred language variants.
 121          * @param boolean $fromUser Get it from $wgUser's preferences
 122          * @return string the preferred language code
 123          * @public
 124          */
 125         function getPreferredVariant( $fromUser = true ) {
 126                 global $wgUser, $wgRequest, $wgVariantArticlePath, $wgDefaultLanguageVariant;
 127
 128                 if($this->mPreferredVariant)
 129                         return $this->mPreferredVariant;
 130
 131                 // figure out user lang without constructing wgLang to avoid infinite recursion
 132                 if( $fromUser )
 133                         $defaultUserLang = $wgUser->getOption( 'language' );
 134                 else
 135                         $defaultUserLang = $this->mMainLanguageCode;
 136                 $userLang = $wgRequest->getVal( 'uselang', $defaultUserLang );
 137                 // see if interface language is same as content, if not, prevent conversion
 138                 if( ! in_array( $userLang, $this->mVariants ) ){
 139                         $this->mPreferredVariant = $this->mMainLanguageCode; // no conversion
 140                         return $this->mPreferredVariant;
 141                 }
 142
 143                 // see if the preference is set in the request
 144                 $req = $wgRequest->getText( 'variant' );
 145                 if( in_array( $req, $this->mVariants ) ) {
 146                         $this->mPreferredVariant = $req;
 147                         return $req;
 148                 }
 149
 150                 // check the syntax /code/ArticleTitle
 151                 if($wgVariantArticlePath!=false && isset($_SERVER['SCRIPT_NAME'])){
 152                         // Note: SCRIPT_NAME probably won't hold the correct value if PHP is run as CGI
 153                         // (it will hold path to php.cgi binary), and might not exist on some very old PHP installations
 154                         $scriptBase = basename( $_SERVER['SCRIPT_NAME'] );
 155                         if(in_array($scriptBase,$this->mVariants)){
 156                                 $this->mPreferredVariant = $scriptBase;
 157                                 return $this->mPreferredVariant;
 158                         }
 159                 }
 160
 161                 // get language variant preference from logged in users
 162                 // Don't call this on stub objects because that causes infinite
 163                 // recursion during initialisation
 164                 if( $fromUser && $wgUser->isLoggedIn() )  {
 165                         $this->mPreferredVariant = $wgUser->getOption('variant');
 166                         return $this->mPreferredVariant;
 167                 }
 168
 169                 // see if default variant is globaly set
 170                 if($wgDefaultLanguageVariant != false  &&  in_array( $wgDefaultLanguageVariant, $this->mVariants )){
 171                         $this->mPreferredVariant = $wgDefaultLanguageVariant;
 172                         return $this->mPreferredVariant;
 173                 }
 174
 175                 if( !$this->mPreferredVariant ) {
 176                         // see if some supported language variant is set in the
 177                         // http header, but we don't set the mPreferredVariant
 178                         // variable in case this is called before the user's
 179                         // preference is loaded
 180                         if( array_key_exists( 'HTTP_ACCEPT_LANGUAGE', $_SERVER ) ) {
 181                                 $acceptLanguage = strtolower( $_SERVER['HTTP_ACCEPT_LANGUAGE'] );
 182
 183                                 // explode by comma
 184                                 $result = explode(',', $acceptLanguage);
 185
 186                                 $languages  = array();
 187
 188                                 foreach( $result as $elem ) {
 189                                         // if $elem likes 'zh-cn;q=0.9'
 190                                         if(($posi = strpos( $elem, ';' )) !== false ) {
 191                                                 // get the real language code likes 'zh-cn'
 192                                                 $languages[] = substr( $elem, 0, $posi );
 193                                         }
 194                                         else {
 195                                                 $languages[] = $elem;
 196                                         }
 197                                 }
 198
 199                                 foreach( $languages as $language ) {
 200                                         // strip whitespace
 201                                         $language = trim( $language );
 202                                         if( in_array( $language, $this->mVariants ) ) {
 203                                                 return $language;
 204                                                 break;
 205                                         }
 206                                 }
 207                         }
 208                 }
 209
 210                 return $this->mMainLanguageCode;
 211
 212         }
 213
 214         /**
 215          * caption convert, base on preg_replace_callback
 216          *
 217          * to convert text in "title" or "alt", like '<img alt="text" ... '
 218          * or '<span title="text" ... '
 219          *
 220          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 221          * @private
 222          */
 223         function captionConvert( $matches ) {
 224                 $toVariant = $this->getPreferredVariant();
 225                 $title = $matches[1];
 226                 $text  = $matches[2];
 227                 // we convert captions except URL
 228                 if( !strpos( $text, '://' ) )
 229                         $text = $this->translate($text, $toVariant);
 230                 return " $title=\"$text\"";
 231         }
 232
 233         /**
 234          * dictionary-based conversion
 235          *
 236          * @param string $text the text to be converted
 237          * @param string $toVariant the target language code
 238          * @return string the converted text
 239          * @private
 240          */
 241         function autoConvert($text, $toVariant=false) {
 242                 $fname="LanguageConverter::autoConvert";
 243
 244                 wfProfileIn( $fname );
 245
 246                 if(!$this->mTablesLoaded)
 247                         $this->loadTables();
 248
 249                 if(!$toVariant)
 250                         $toVariant = $this->getPreferredVariant();
 251                 if(!in_array($toVariant, $this->mVariants))
 252                         return $text;
 253
 254                 /* we convert everything except:
 255                    1. html markups (anything between < and >)
 256                    2. html entities
 257                    3. place holders created by the parser
 258                 */
 259                 global $wgParser;
 260                 if (isset($wgParser) && $wgParser->UniqPrefix()!=''){
 261                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 262                 } else
 263                         $marker = "";
 264
 265                 // this one is needed when the text is inside an html markup
 266                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 267
 268                 // disable convert to variants between <code></code> tags
 269                 $codefix = '<code>.+?<\/code>|';
 270                 // disable convertsion of <script type="text/javascript"> ... </script>
 271                 $scriptfix = '<script.*?>.*?<\/script>|';
 272                 // disable conversion of <pre xxxx> ... </pre>
 273                 $prefix = '<pre.*?>.*?<\/pre>|';
 274
 275                 $reg = '/'.$codefix . $scriptfix . $prefix . '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 276
 277                 $matches = preg_split($reg, $text, -1, PREG_SPLIT_OFFSET_CAPTURE);
 278
 279                 $m = array_shift($matches);
 280
 281                 $ret = $this->translate($m[0], $toVariant);
 282                 $mstart = $m[1]+strlen($m[0]);
 283
 284                 // enable convertsion of '<img alt="xxxx" ... ' or '<span title="xxxx" ... '
 285                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 286                 foreach($matches as $m) {
 287                         $mark = substr($text, $mstart, $m[1]-$mstart);
 288                         $mark = preg_replace_callback($captionpattern, array(&$this, 'captionConvert'), $mark);
 289                         $ret .= $mark;
 290                         $ret .= $this->translate($m[0], $toVariant);
 291                         $mstart = $m[1] + strlen($m[0]);
 292                 }
 293                 wfProfileOut( $fname );
 294                 return $ret;
 295         }
 296
 297         /**
 298          * Translate a string to a variant
 299          * Doesn't process markup or do any of that other stuff, for that use convert()
 300          *
 301          * @param string $text Text to convert
 302          * @param string $variant Variant language code
 303          * @return string Translated text
 304          * @private
 305          */
 306         function translate( $text, $variant ) {
 307                 wfProfileIn( __METHOD__ );
 308                 if( !$this->mTablesLoaded )
 309                         $this->loadTables();
 310                 $text = $this->mTables[$variant]->replace( $text );
 311                 wfProfileOut( __METHOD__ );
 312                 return $text;
 313         }
 314
 315         /**
 316          * convert text to all supported variants
 317          *
 318          * @param string $text the text to be converted
 319          * @return array of string
 320          * @public
 321          */
 322         function autoConvertToAllVariants($text) {
 323                 $fname="LanguageConverter::autoConvertToAllVariants";
 324                 wfProfileIn( $fname );
 325                 if( !$this->mTablesLoaded )
 326                         $this->loadTables();
 327
 328                 $ret = array();
 329                 foreach($this->mVariants as $variant) {
 330                         $ret[$variant] = $this->translate($text, $variant);
 331                 }
 332
 333                 wfProfileOut( $fname );
 334                 return $ret;
 335         }
 336
 337         /**
 338          * convert link text to all supported variants
 339          *
 340          * @param string $text the text to be converted
 341          * @return array of string
 342          * @public
 343          */
 344         function convertLinkToAllVariants($text) {
 345                 if( !$this->mTablesLoaded )
 346                         $this->loadTables();
 347
 348                 $ret = array();
 349                 $tarray = explode($this->mMarkup['begin'], $text);
 350                 $tfirst = array_shift($tarray);
 351
 352                 foreach($this->mVariants as $variant)
 353                         $ret[$variant] = $this->translate($tfirst,$variant);
 354
 355                 foreach($tarray as $txt) {
 356                         $marked = explode($this->mMarkup['end'], $txt, 2);
 357
 358                         foreach($this->mVariants as $variant){
 359                                 $ret[$variant] .= $this->mMarkup['begin'].$marked[0].$this->mMarkup['end'];
 360                                 if(array_key_exists(1, $marked))
 361                                         $ret[$variant] .= $this->translate($marked[1],$variant);
 362                         }
 363
 364                 }
 365
 366                 return $ret;
 367         }
 368
 369         /**
 370          * prepare manual conversion table
 371          * @private
 372          */
 373         function applyManualConv( $convRule ){
 374                 // use syntax -{T|zh:TitleZh;zh-tw:TitleTw}- for custom conversion in title
 375                 $title = $convRule->getTitle();
 376                 if( $title ){
 377                         $this->mTitleFromFlag = true;
 378                         $this->mTitleDisplay =  $title;
 379                 }
 380
 381                 //apply manual conversion table to global table
 382                 $convTable = $convRule->getConvTable();
 383                 $action = $convRule->getRulesAction();
 384                 foreach( $convTable as $variant => $pair ) {
 385                         if( !in_array( $variant, $this->mVariants ) )continue;
 386                         if( $action=="add" ) {
 387                                 foreach( $pair as $from => $to ) {
 388                                         // to ensure that $from and $to not be left blank
 389                                         // so $this->translate() could always return a string
 390                                         if ( $from || $to )
 391                                                 // more efficient than array_merge(), about 2.5 times.
 392                                                 $this->mTables[$variant]->setPair( $from, $to );
 393                                 }
 394                         }
 395                         elseif ( $action == "remove" ) {
 396                                 $this->mTables[$variant]->removeArray( $pair );
 397                         }
 398                 }
 399         }
 400
 401         /**
 402          * Convert text using a parser object for context
 403          * @public
 404          */
 405         function parserConvert( $text, &$parser ) {
 406                 global $wgDisableLangConversion;
 407                 /* don't do anything if this is the conversion table */
 408                 if ( $parser->getTitle()->getNamespace() == NS_MEDIAWIKI &&
 409                                  strpos($parser->mTitle->getText(), "Conversiontable") !== false )
 410                 {
 411                         return $text;
 412                 }
 413
 414                 if ( $wgDisableLangConversion )
 415                         return $text;
 416
 417                 $text = $this->convert( $text );
 418
 419                 if ( $this->mTitleFromFlag )
 420                         $parser->mOutput->setTitleText( $this->mTitleDisplay );
 421                 return $text;
 422         }
 423
 424         /**
 425          * convert namespace
 426          * @param string $title the title included namespace
 427          * @return array of string
 428          * @private
 429          */
 430         function convertNamespace( $title, $variant ) {
 431                 $splittitle = explode( ':', $title );
 432                 if (count($splittitle) < 2)
 433                         return $title;
 434                 if ( isset( $this->mNamespaceTables[$variant][$splittitle[0]] ) )
 435                         $splittitle[0] = $this->mNamespaceTables[$variant][$splittitle[0]];
 436                 $ret = implode(':', $splittitle );
 437                 return $ret;
 438         }
 439
 440         /**
 441          * convert title
 442          * @private
 443          */
 444         function convertTitle( $text, $variant ){
 445                 global $wgDisableTitleConversion, $wgUser;
 446
 447                 // check for global param and __NOTC__ tag
 448                 if( $wgDisableTitleConversion || !$this->mDoTitleConvert || $wgUser->getOption('noconvertlink') == 1 ) {
 449                         $this->mTitleDisplay = $text;
 450                         return $text;
 451                 }
 452
 453                 // use the title from the T flag if any
 454                 if( $this->mTitleFromFlag ){
 455                         $this->mTitleFromFlag = false;
 456                         return $this->mTitleDisplay;
 457                 }
 458
 459                 global $wgRequest;
 460                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 461                 $action = $wgRequest->getText( 'action' );
 462                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 463                 if ( $isredir == 'no' || $action == 'edit' || $action == 'submit' || $linkconvert == 'no' ) {
 464                         return $text;
 465                 } else {
 466                         $text = $this->convertNamespace( $text, $variant );
 467                         $this->mTitleDisplay = $this->convert( $text );
 468                         return $this->mTitleDisplay;
 469                 }
 470         }
 471
 472         /**
 473          * convert text to different variants of a language. the automatic
 474          * conversion is done in autoConvert(). here we parse the text
 475          * marked with -{}-, which specifies special conversions of the
 476          * text that can not be accomplished in autoConvert()
 477          *
 478          * syntax of the markup:
 479          * -{code1:text1;code2:text2;...}-  or
 480          * -{flags|code1:text1;code2:text2;...}-  or
 481          * -{text}- in which case no conversion should take place for text
 482          *
 483          * @param string $text text to be converted
 484          * @param bool $isTitle whether this conversion is for the article title
 485          * @return string converted text
 486          * @public
 487          */
 488         function convert( $text, $isTitle = false ) {
 489
 490                 $mw =& MagicWord::get( 'notitleconvert' );
 491                 if( $mw->matchAndRemove( $text ) )
 492                         $this->mDoTitleConvert = false;
 493                 $mw =& MagicWord::get( 'nocontentconvert' );
 494                 if( $mw->matchAndRemove( $text ) ) {
 495                         $this->mDoContentConvert = false;
 496                 }
 497
 498                 // no conversion if redirecting
 499                 $mw =& MagicWord::get( 'redirect' );
 500                 if( $mw->matchStart( $text ) )
 501                         return $text;
 502
 503                 $plang = $this->getPreferredVariant();
 504
 505                 // for title convertion
 506                 if ( $isTitle ) return $this->convertTitle( $text, $plang );
 507
 508                 $tarray = StringUtils::explode( $this->mMarkup['end'], $text );
 509                 $text = '';
 510
 511                 foreach ( $tarray as $txt ) {
 512
 513                         $marked = explode( $this->mMarkup['begin'], $txt, 2 );
 514
 515                         if( $this->mDoContentConvert )
 516                                 // Bug 19620: should convert a string immediately after a new rule added.
 517                                 $text .= $this->autoConvert( $marked[0], $plang );
 518
 519                         if ( array_key_exists( 1, $marked ) ) {
 520                                 $crule = new ConverterRule($marked[1], $this);
 521                                 $crule->parse( $plang );
 522                                 $text .= $crule->getDisplay();
 523                                 $this->applyManualConv( $crule );
 524                         }
 525                         else
 526                                 $text .= $this->mMarkup['end'];
 527
 528                 }
 529
 530                 // Remove the last delimiter (wasn't real)
 531                 $text = substr( $text, 0, -strlen( $this->mMarkup['end'] ) );
 532
 533                 return $text;
 534         }
 535
 536         /**
 537          * if a language supports multiple variants, it is
 538          * possible that non-existing link in one variant
 539          * actually exists in another variant. this function
 540          * tries to find it. See e.g. LanguageZh.php
 541          *
 542          * @param string $link the name of the link
 543          * @param mixed $nt the title object of the link
 544          * @param boolean $ignoreOtherCond: to disable other conditions when
 545          *      we need to transclude a template or update a category's link
 546          * @return null the input parameters may be modified upon return
 547          * @public
 548          */
 549         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 550                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest, $wgUser;
 551                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 552                 $action = $wgRequest->getText( 'action' );
 553                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 554                 $disableLinkConversion = $wgDisableLangConversion || $wgDisableTitleConversion;
 555                 $linkBatch = new LinkBatch();
 556
 557                 $ns=NS_MAIN;
 558
 559                 if ( $disableLinkConversion || ( !$ignoreOtherCond && ( $isredir == 'no' || $action == 'edit'
 560                         || $action == 'submit' || $linkconvert == 'no' || $wgUser->getOption('noconvertlink') == 1 ) ) )
 561                         return;
 562
 563                 if(is_object($nt))
 564                         $ns = $nt->getNamespace();
 565
 566                 $variants = $this->autoConvertToAllVariants($link);
 567                 if($variants == false) //give up
 568                         return;
 569
 570                 $titles = array();
 571
 572                 foreach( $variants as $v ) {
 573                         if($v != $link){
 574                                 $varnt = Title::newFromText( $v, $ns );
 575                                 if(!is_null($varnt)){
 576                                         $linkBatch->addObj($varnt);
 577                                         $titles[]=$varnt;
 578                                 }
 579                         }
 580                 }
 581
 582                 // fetch all variants in single query
 583                 $linkBatch->execute();
 584
 585                 foreach( $titles as $varnt ) {
 586                         if( $varnt->getArticleID() > 0 ) {
 587                                 $nt = $varnt;
 588                                 $link = $varnt->getText();
 589                                 break;
 590                         }
 591                 }
 592         }
 593
 594     /**
 595          * returns language specific hash options
 596          *
 597          * @public
 598          */
 599         function getExtraHashOptions() {
 600                 $variant = $this->getPreferredVariant();
 601                 return '!' . $variant ;
 602         }
 603
 604     /**
 605          * get title text as defined in the body of the article text
 606          *
 607          * @public
 608          */
 609         function getParsedTitle() {
 610                 return $this->mTitleDisplay;
 611         }
 612
 613         /**
 614          * a write lock to the cache
 615          *
 616          * @private
 617          */
 618         function lockCache() {
 619                 global $wgMemc;
 620                 $success = false;
 621                 for($i=0; $i<30; $i++) {
 622                         if($success = $wgMemc->add($this->mCacheKey . "lock", 1, 10))
 623                                 break;
 624                         sleep(1);
 625                 }
 626                 return $success;
 627         }
 628
 629         /**
 630          * unlock cache
 631          *
 632          * @private
 633          */
 634         function unlockCache() {
 635                 global $wgMemc;
 636                 $wgMemc->delete($this->mCacheKey . "lock");
 637         }
 638
 639
 640         /**
 641          * Load default conversion tables
 642          * This method must be implemented in derived class
 643          *
 644          * @private
 645          */
 646         function loadDefaultTables() {
 647                 $name = get_class($this);
 648                 wfDie("Must implement loadDefaultTables() method in class $name");
 649         }
 650
 651         /**
 652          * load conversion tables either from the cache or the disk
 653          * @private
 654          */
 655         function loadTables($fromcache=true) {
 656                 global $wgMemc;
 657                 if( $this->mTablesLoaded )
 658                         return;
 659                 wfProfileIn( __METHOD__ );
 660                 $this->mTablesLoaded = true;
 661                 $this->mTables = false;
 662                 if($fromcache) {
 663                         wfProfileIn( __METHOD__.'-cache' );
 664                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 665                         wfProfileOut( __METHOD__.'-cache' );
 666                 }
 667                 if ( !$this->mTables || !isset( $this->mTables[self::CACHE_VERSION_KEY] ) ) {
 668                         wfProfileIn( __METHOD__.'-recache' );
 669                         // not in cache, or we need a fresh reload.
 670                         // we will first load the default tables
 671                         // then update them using things in MediaWiki:Zhconversiontable/*
 672                         $this->loadDefaultTables();
 673                         foreach($this->mVariants as $var) {
 674                                 $cached = $this->parseCachedTable($var);
 675                                 $this->mTables[$var]->mergeArray($cached);
 676                         }
 677
 678                         $this->postLoadTables();
 679                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 680
 681                         if($this->lockCache()) {
 682                                 $wgMemc->set($this->mCacheKey, $this->mTables, 43200);
 683                                 $this->unlockCache();
 684                         }
 685                         wfProfileOut( __METHOD__.'-recache' );
 686                 }
 687                 wfProfileOut( __METHOD__ );
 688         }
 689
 690     /**
 691          * Hook for post processig after conversion tables are loaded
 692          *
 693          */
 694         function postLoadTables() {}
 695
 696     /**
 697          * Reload the conversion tables
 698          *
 699          * @private
 700          */
 701         function reloadTables() {
 702                 if($this->mTables)
 703                         unset($this->mTables);
 704                 $this->mTablesLoaded = false;
 705                 $this->loadTables(false);
 706         }
 707
 708
 709         /**
 710          * parse the conversion table stored in the cache
 711          *
 712          * the tables should be in blocks of the following form:
 713          *              -{
 714          *                      word => word ;
 715          *                      word => word ;
 716          *                      ...
 717          *              }-
 718          *
 719          *      to make the tables more manageable, subpages are allowed
 720          *      and will be parsed recursively if $recursive=true
 721          *
 722          */
 723         function parseCachedTable($code, $subpage='', $recursive=true) {
 724                 global $wgMessageCache;
 725                 static $parsed = array();
 726
 727                 if(!is_object($wgMessageCache))
 728                         return array();
 729
 730                 $key = 'Conversiontable/'.$code;
 731                 if($subpage)
 732                         $key .= '/' . $subpage;
 733
 734                 if(array_key_exists($key, $parsed))
 735                         return array();
 736
 737                 if ( strpos( $code, '/' ) === false ) {
 738                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 739                 } else {
 740                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, "Conversiontable/$code" );
 741                         if ( $title && $title->exists() ) {
 742                                 $article = new Article( $title );
 743                                 $txt = $article->getContents();
 744                         } else {
 745                                 $txt = '';
 746                         }
 747                 }
 748
 749                 // get all subpage links of the form
 750                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 751                 $linkhead = $this->mLangObj->getNsText(NS_MEDIAWIKI) . ':Conversiontable';
 752                 $subs = explode('[[', $txt);
 753                 $sublinks = array();
 754                 foreach( $subs as $sub ) {
 755                         $link = explode(']]', $sub, 2);
 756                         if(count($link) != 2)
 757                                 continue;
 758                         $b = explode('|', $link[0]);
 759                         $b = explode('/', trim($b[0]), 3);
 760                         if(count($b)==3)
 761                                 $sublink = $b[2];
 762                         else
 763                                 $sublink = '';
 764
 765                         if($b[0] == $linkhead && $b[1] == $code) {
 766                                 $sublinks[] = $sublink;
 767                         }
 768                 }
 769
 770
 771                 // parse the mappings in this page
 772                 $blocks = explode($this->mMarkup['begin'], $txt);
 773                 array_shift($blocks);
 774                 $ret = array();
 775                 foreach($blocks as $block) {
 776                         $mappings = explode($this->mMarkup['end'], $block, 2);
 777                         $stripped = str_replace(array("'", '"', '*','#'), '', $mappings[0]);
 778                         $table = explode( ';', $stripped );
 779                         foreach( $table as $t ) {
 780                                 $m = explode( '=>', $t );
 781                                 if( count( $m ) != 2)
 782                                         continue;
 783                                 // trim any trailling comments starting with '//'
 784                                 $tt = explode('//', $m[1], 2);
 785                                 $ret[trim($m[0])] = trim($tt[0]);
 786                         }
 787                 }
 788                 $parsed[$key] = true;
 789
 790
 791                 // recursively parse the subpages
 792                 if($recursive) {
 793                         foreach($sublinks as $link) {
 794                                 $s = $this->parseCachedTable($code, $link, $recursive);
 795                                 $ret = array_merge($ret, $s);
 796                         }
 797                 }
 798
 799                 if ($this->mUcfirst) {
 800                         foreach ($ret as $k => $v) {
 801                                 $ret[Language::ucfirst($k)] = Language::ucfirst($v);
 802                         }
 803                 }
 804                 return $ret;
 805         }
 806
 807         /**
 808          * Enclose a string with the "no conversion" tag. This is used by
 809          * various functions in the Parser
 810          *
 811          * @param string $text text to be tagged for no conversion
 812          * @return string the tagged text
 813          * @public
 814          */
 815         function markNoConversion($text, $noParse=false) {
 816                 # don't mark if already marked
 817                 if(strpos($text, $this->mMarkup['begin']) ||
 818                    strpos($text, $this->mMarkup['end']))
 819                         return $text;
 820
 821                 $ret = $this->mMarkup['begin'] .'R|'. $text . $this->mMarkup['end'];
 822                 return $ret;
 823         }
 824
 825         /**
 826          * convert the sorting key for category links. this should make different
 827          * keys that are variants of each other map to the same key
 828          */
 829         function convertCategoryKey( $key ) {
 830                 return $key;
 831         }
 832         /**
 833          * hook to refresh the cache of conversion tables when
 834          * MediaWiki:conversiontable* is updated
 835          * @private
 836          */
 837         function OnArticleSaveComplete($article, $user, $text, $summary, $isminor, $iswatch, $section, $flags, $revision) {
 838                 $titleobj = $article->getTitle();
 839                 if($titleobj->getNamespace() == NS_MEDIAWIKI) {
 840                         $title = $titleobj->getDBkey();
 841                         $t = explode('/', $title, 3);
 842                         $c = count($t);
 843                         if( $c > 1 && $t[0] == 'Conversiontable' ) {
 844                                 if(in_array($t[1], $this->mVariants)) {
 845                                         $this->reloadTables();
 846                                 }
 847                         }
 848                 }
 849                 return true;
 850         }
 851
 852         /**
 853          * Armour rendered math against conversion
 854          * Wrap math into rawoutput -{R| math }- syntax
 855          * @public
 856          */
 857         function armourMath($text){
 858                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 859                 // to avoid a unwanted '}-' appeared after the math-image.
 860                 $text = strtr( $text, array('-{' => '-&#123;', '}-' => '&#125;-') );
 861                 $ret = $this->mMarkup['begin'] . 'R|' . $text . $this->mMarkup['end'];
 862                 return $ret;
 863         }
 864 }
 865
 866 /**
 867  * parser for rules of language conversion , parse rules in -{ }- tag
 868  * @ingroup Language
 869  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
 870  */
 871 class ConverterRule {
 872         var $mText; // original text in -{text}-
 873         var $mConverter; // LanguageConverter object
 874         var $mManualCodeError='<strong class="error">code error!</strong>';
 875         var $mRuleDisplay = '',$mRuleTitle=false;
 876         var $mRules = '';// string : the text of the rules
 877         var $mRulesAction = 'none';
 878         var $mFlags = array();
 879         var $mConvTable = array();
 880         var $mBidtable = array();// array of the translation in each variant
 881         var $mUnidtable = array();// array of the translation in each variant
 882
 883         /**
 884          * Constructor
 885          *
 886          * @param string $text the text between -{ and }-
 887          * @param object $converter a  LanguageConverter object
 888          * @access public
 889          */
 890         function __construct($text,$converter){
 891                 $this->mText = $text;
 892                 $this->mConverter=$converter;
 893                 foreach($converter->mVariants as $v){
 894                         $this->mConvTable[$v]=array();
 895                 }
 896         }
 897
 898         /**
 899          * check if variants array in convert array
 900          *
 901          * @param string $variant Variant language code
 902          * @return string Translated text
 903          * @public
 904          */
 905         function getTextInBidtable($variants){
 906                 if(is_string($variants)){ $variants=array($variants); }
 907                 if(!is_array($variants)) return false;
 908                 foreach ($variants as $variant){
 909                         if(array_key_exists($variant, $this->mBidtable)){
 910                                 return $this->mBidtable[$variant];
 911                         }
 912                 }
 913                 return false;
 914         }
 915
 916         /**
 917          * Parse flags with syntax -{FLAG| ... }-
 918          * @private
 919          */
 920         function parseFlags(){
 921                 $text = $this->mText;
 922                 if(strlen($text) < 2 ) {
 923                         $this->mFlags = array( 'R' );
 924                         $this->mRules = $text;
 925                         return;
 926                 }
 927
 928                 $flags = array();
 929                 $markup = $this->mConverter->mMarkup;
 930                 $validFlags = $this->mConverter->mFlags;
 931                 $variants = $this->mConverter->mVariants;
 932
 933                 $tt = explode($markup['flagsep'], $text, 2);
 934                 if(count($tt) == 2) {
 935                         $f = explode($markup['varsep'], $tt[0]);
 936                         foreach($f as $ff) {
 937                                 $ff = trim($ff);
 938                                 if(array_key_exists($ff, $validFlags) &&
 939                                                         !in_array($validFlags[$ff], $flags))
 940                                         $flags[] = $validFlags[$ff];
 941                         }
 942                         $rules = $tt[1];
 943                 } else {
 944                         $rules = $text;
 945                 }
 946
 947                 //check flags
 948                 if( in_array('R',$flags) ){
 949                         $flags = array('R');// remove other flags
 950                 } elseif ( in_array('N',$flags) ){
 951                         $flags = array('N');// remove other flags
 952                 } elseif ( in_array('-',$flags) ){
 953                         $flags = array('-');// remove other flags
 954                 } elseif (count($flags)==1 && $flags[0]=='T'){
 955                         $flags[]='H';
 956                 } elseif ( in_array('H',$flags) ){
 957                         // replace A flag, and remove other flags except T
 958                         $temp=array('+','H');
 959                         if(in_array('T',$flags)) $temp[] = 'T';
 960                         if(in_array('D',$flags)) $temp[] = 'D';
 961                         $flags = $temp;
 962                 } else {
 963                         if ( in_array('A',$flags) ) {
 964                                 $flags[]='+';
 965                                 $flags[]='S';
 966                         }
 967                         if ( in_array('D',$flags) )
 968                                 $flags=array_diff($flags,array('S'));
 969                         $flags_temp = array();
 970                         foreach ($variants as $variant) {
 971                                 // try to find flags like "zh-hans", "zh-hant"
 972                                 // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
 973                                 if ( in_array($variant, $flags) )
 974                                         $flags_temp[] = $variant;
 975                         }
 976                         if ( count($flags_temp) !== 0 )
 977                                 $flags = $flags_temp;
 978                 }
 979                 if ( count($flags) == 0 )
 980                         $flags = array('S');
 981                 $this->mRules=$rules;
 982                 $this->mFlags=$flags;
 983         }
 984
 985         /**
 986          * generate conversion table
 987          * @private
 988          */
 989         function parseRules() {
 990                 $rules = $this->mRules;
 991                 $flags = $this->mFlags;
 992                 $bidtable = array();
 993                 $unidtable = array();
 994                 $markup = $this->mConverter->mMarkup;
 995                 $variants = $this->mConverter->mVariants;
 996
 997                 // varsep_pattern for preg_split:
 998                 // text should be splited by ";" only if a valid variant
 999                 // name exist after the markup, for example:
1000                 //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:<span style="font-size:120%;">yyy</span>;}-
1001                 // we should split it as:
1002                 //  array(
1003                 //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1004                 //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1005                 //        [2] => ''
1006                 //       )
1007                 $varsep_pattern = '/' . $markup['varsep'] . '\s*' . '(?=';
1008                 foreach( $variants as $variant ) {
1009                         $varsep_pattern .= $variant . '\s*' . $markup['codesep'] . '|'; // zh-hans:xxx;zh-hant:yyy
1010                         $varsep_pattern .= '[^;]*?' . $markup['unidsep'] . '\s*' . $variant
1011                                                         . '\s*' . $markup['codesep'] . '|'; // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1012                 }
1013                 $varsep_pattern .= '\s*$)/';
1014
1015                 $choice = preg_split($varsep_pattern, $rules);
1016
1017                 foreach( $choice as $c ) {
1018                         $v  = explode($markup['codesep'], $c, 2);
1019                         if( count($v) != 2 )
1020                                 continue;// syntax error, skip
1021                         $to = trim($v[1]);
1022                         $v  = trim($v[0]);
1023                         $u  = explode($markup['unidsep'], $v);
1024                         // if $to is empty, strtr() could return a wrong result
1025                         if( count($u) == 1 && $to && in_array( $v, $variants ) ) {
1026                                 $bidtable[$v] = $to;
1027                         } else if(count($u) == 2){
1028                                 $from = trim($u[0]);
1029                                 $v    = trim($u[1]);
1030                                 if( array_key_exists( $v, $unidtable ) && !is_array( $unidtable[$v] )
1031                                         && $to && in_array( $v, $variants ) )
1032                                         $unidtable[$v] = array( $from=>$to );
1033                                 elseif ( $to && in_array( $v, $variants ) )
1034                                         $unidtable[$v][$from] = $to;
1035                         }
1036                         // syntax error, pass
1037                         if ( !array_key_exists( $v, $this->mConverter->mVariantNames ) ){
1038                                 $bidtable = array();
1039                                 $unidtable = array();
1040                                 break;
1041                         }
1042                 }
1043                 $this->mBidtable = $bidtable;
1044                 $this->mUnidtable = $unidtable;
1045         }
1046
1047         /**
1048          * @private
1049          */
1050         function getRulesDesc(){
1051                 $codesep = $this->mConverter->mDescCodeSep;
1052                 $varsep = $this->mConverter->mDescVarSep;
1053                 $text='';
1054                 foreach($this->mBidtable as $k => $v)
1055                         $text .= $this->mConverter->mVariantNames[$k]."$codesep$v$varsep";
1056                 foreach($this->mUnidtable as $k => $a)
1057                         foreach($a as $from=>$to)
1058                                 $text.=$from.'⇒'.$this->mConverter->mVariantNames[$k]."$codesep$to$varsep";
1059                 return $text;
1060         }
1061
1062         /**
1063          *  Parse rules conversion
1064          * @private
1065          */
1066         function getRuleConvertedStr($variant,$doConvert){
1067                 $bidtable = $this->mBidtable;
1068                 $unidtable = $this->mUnidtable;
1069
1070                 if( count($bidtable) + count($unidtable) == 0 ){
1071                         return $this->mRules;
1072                 } elseif ($doConvert){// the text converted
1073                         // display current variant in bidirectional array
1074                         $disp = $this->getTextInBidtable($variant);
1075                         // or display current variant in fallbacks
1076                         if(!$disp)
1077                                 $disp = $this->getTextInBidtable(
1078                                                 $this->mConverter->getVariantFallbacks($variant));
1079                         // or display current variant in unidirectional array
1080                         if(!$disp && array_key_exists($variant,$unidtable)){
1081                                 $disp = array_values($unidtable[$variant]);
1082                                 $disp = $disp[0];
1083                         }
1084                         // or display frist text under disable manual convert
1085                         if(!$disp && $this->mConverter->mManualLevel[$variant]=='disable') {
1086                                 if(count($bidtable)>0){
1087                                         $disp = array_values($bidtable);
1088                                         $disp = $disp[0];
1089                                 } else {
1090                                         $disp = array_values($unidtable);
1091                                         $disp = array_values($disp[0]);
1092                                         $disp = $disp[0];
1093                                 }
1094                         }
1095                         return $disp;
1096                 } else {// no convert
1097                         return $this->mRules;
1098                 }
1099         }
1100
1101         /**
1102          * generate conversion table for all text
1103          * @private
1104          */
1105         function generateConvTable(){
1106                 $flags = $this->mFlags;
1107                 $bidtable = $this->mBidtable;
1108                 $unidtable = $this->mUnidtable;
1109                 $manLevel = $this->mConverter->mManualLevel;
1110
1111                 $vmarked=array();
1112                 foreach($this->mConverter->mVariants as $v) {
1113                         /* for bidirectional array
1114                                 fill in the missing variants, if any,
1115                                 with fallbacks */
1116                         if(!array_key_exists($v, $bidtable)) {
1117                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($v);
1118                                 $vf = $this->getTextInBidtable($variantFallbacks);
1119                                 if($vf) $bidtable[$v] = $vf;
1120                         }
1121
1122                         if(array_key_exists($v,$bidtable)){
1123                                 foreach($vmarked as $vo){
1124                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1125                                         // or -{H|zh:WordZh;zh-tw:WordTw}- or -{-|zh:WordZh;zh-tw:WordTw}-
1126                                         // to introduce a custom mapping between
1127                                         // words WordZh and WordTw in the whole text
1128                                         if($manLevel[$v]=='bidirectional'){
1129                                                 $this->mConvTable[$v][$bidtable[$vo]]=$bidtable[$v];
1130                                         }
1131                                         if($manLevel[$vo]=='bidirectional'){
1132                                                 $this->mConvTable[$vo][$bidtable[$v]]=$bidtable[$vo];
1133                                         }
1134                                 }
1135                                 $vmarked[]=$v;
1136                         }
1137                         /*for unidirectional array
1138                                 fill to convert tables */
1139                         $allow_unid = $manLevel[$v]=='bidirectional'
1140                                         || $manLevel[$v]=='unidirectional';
1141                         if($allow_unid && array_key_exists($v,$unidtable)){
1142                                 $ct=$this->mConvTable[$v];
1143                                 $this->mConvTable[$v] = array_merge($ct,$unidtable[$v]);
1144                         }
1145                 }
1146         }
1147
1148         /**
1149          * Parse rules and flags
1150          * @public
1151          */
1152         function parse($variant){
1153                 if(!$variant)
1154                         $variant = $this->mConverter->getPreferredVariant();
1155
1156                 $variants = $this->mConverter->mVariants;
1157                 $this->parseFlags();
1158                 $flags = $this->mFlags;
1159
1160                 // convert to specified variant
1161                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1162                 if( count( array_diff( $flags, $variants ) ) == 0 and count( $flags ) != 0 ) {
1163                         if ( in_array( $variant, $flags ) ) // check if current variant in flags
1164                                 // then convert <text to convert> to current language
1165                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variant );
1166                         else { // if current variant no in flags,
1167                                    // then we check its fallback variants.
1168                                 $variantFallbacks = $this->mConverter->getVariantFallbacks($variant);
1169                                 foreach ( $variantFallbacks as $variantFallback ) {
1170                                         // if current variant's fallback exist in flags
1171                                         if ( in_array( $variantFallback, $flags ) ) {
1172                                                 // then convert <text to convert> to fallback language
1173                                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules, $variantFallback );
1174                                                 break;
1175                                         }
1176                                 }
1177                         }
1178                         $this->mFlags = $flags = array('R');
1179                 }
1180
1181                 if( !in_array( 'R', $flags ) || !in_array( 'N', $flags ) ) {
1182                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1183                         $this->mRules = str_replace('=&gt;','=>',$this->mRules);
1184
1185                         $this->parseRules();
1186                 }
1187                 $rules = $this->mRules;
1188
1189                 if( count( $this->mBidtable ) == 0 && count( $this->mUnidtable ) == 0 ){
1190                         if(in_array('+',$flags) || in_array('-',$flags))
1191                                 // fill all variants if text in -{A/H/-|text} without rules
1192                                 foreach($this->mConverter->mVariants as $v)
1193                                         $this->mBidtable[$v] = $rules;
1194                         elseif (!in_array('N',$flags) && !in_array('T',$flags) )
1195                                 $this->mFlags = $flags = array('R');
1196                 }
1197
1198                 if( in_array('R',$flags) ) {
1199                         // if we don't do content convert, still strip the -{}- tags
1200                         $this->mRuleDisplay = $rules;
1201                 } elseif ( in_array('N',$flags) ){
1202                         // proces N flag: output current variant name
1203                         $this->mRuleDisplay = $this->mConverter->mVariantNames[trim($rules)];
1204                 } elseif ( in_array('D',$flags) ){
1205                         // proces D flag: output rules description
1206                         $this->mRuleDisplay = $this->getRulesDesc();
1207                 } elseif ( in_array('H',$flags) || in_array('-',$flags) ) {
1208                         // proces H,- flag or T only: output nothing
1209                         $this->mRuleDisplay = '';
1210                 } elseif ( in_array('S',$flags) ){
1211                         $this->mRuleDisplay = $this->getRuleConvertedStr($variant,
1212                                                         $this->mConverter->mDoContentConvert);
1213                 } else {
1214                         $this->mRuleDisplay= $this->mManualCodeError;
1215                 }
1216                 // proces T flag
1217                 if ( in_array('T',$flags) ) {
1218                         $this->mRuleTitle = $this->getRuleConvertedStr($variant,
1219                                                         $this->mConverter->mDoTitleConvert);
1220                 }
1221
1222                 if (in_array('-', $flags))
1223                         $this->mRulesAction='remove';
1224                 if (in_array('+', $flags))
1225                         $this->mRulesAction='add';
1226
1227                 $this->generateConvTable();
1228         }
1229
1230         /**
1231          * @public
1232          */
1233         function hasRules(){
1234                 // TODO:
1235         }
1236
1237         /**
1238          * get display text on markup -{...}-
1239          * @public
1240          */
1241         function getDisplay(){
1242                 return $this->mRuleDisplay;
1243         }
1244         /**
1245          * get converted title
1246          * @public
1247          */
1248         function getTitle(){
1249                 return $this->mRuleTitle;
1250         }
1251
1252         /**
1253          * return how deal with conversion rules
1254          * @public
1255          */
1256         function getRulesAction(){
1257                 return $this->mRulesAction;
1258         }
1259
1260         /**
1261          * get conversion table ( bidirectional and unidirectional conversion table )
1262          * @public
1263          */
1264         function getConvTable(){
1265                 return $this->mConvTable;
1266         }
1267
1268         /**
1269          * get conversion rules string
1270          * @public
1271          */
1272         function getRules(){
1273                 return $this->mRules;
1274         }
1275
1276         /**
1277          * get conversion flags
1278          * @public
1279          */
1280         function getFlags(){
1281                 return $this->mFlags;
1282         }
1283 }