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 conversion.
  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 $mMainLanguageCode;
  20         var $mVariants, $mVariantFallbacks, $mVariantNames;
  21         var $mTablesLoaded = false;
  22         var $mTables;
  23         // 'bidirectional' 'unidirectional' 'disable' for each variant
  24         var $mManualLevel;
  25         var $mCacheKey;
  26         var $mLangObj;
  27         var $mFlags;
  28         var $mDescCodeSep = ':', $mDescVarSep = ';';
  29         var $mUcfirst = false;
  30         var $mConvRuleTitle = false;
  31         var $mURLVariant;
  32         var $mUserVariant;
  33         var $mHeaderVariant;
  34         var $mMaxDepth = 10;
  35         var $mVarSeparatorPattern;
  36
  37         const CACHE_VERSION_KEY = 'VERSION 6';
  38
  39         /**
  40          * Constructor
  41          *
  42          * @param $langobj The Language Object
  43          * @param $maincode String: the main language code of this language
  44          * @param $variants Array: the supported variants of this language
  45          * @param $variantfallbacks Array: the fallback language of each variant
  46          * @param $flags Array: defining the custom strings that maps to the flags
  47          * @param $manualLevel Array: limit for supported variants
  48          */
  49         public function __construct( $langobj, $maincode,
  50                                                                 $variants = array(),
  51                                                                 $variantfallbacks = array(),
  52                                                                 $flags = array(),
  53                                                                 $manualLevel = array() ) {
  54                 $this->mLangObj = $langobj;
  55                 $this->mMainLanguageCode = $maincode;
  56
  57                 global $wgDisabledVariants;
  58                 $this->mVariants = array();
  59                 foreach ( $variants as $variant ) {
  60                         if ( !in_array( $variant, $wgDisabledVariants ) ) {
  61                                 $this->mVariants[] = $variant;
  62                         }
  63                 }
  64                 $this->mVariantFallbacks = $variantfallbacks;
  65                 global $wgLanguageNames;
  66                 $this->mVariantNames = $wgLanguageNames;
  67                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  68                 $f = array(
  69                         // 'S' show converted text
  70                         // '+' add rules for alltext
  71                         // 'E' the gave flags is error
  72                         // these flags above are reserved for program
  73                         'A' => 'A',       // add rule for convert code (all text convert)
  74                         'T' => 'T',       // title convert
  75                         'R' => 'R',       // raw content
  76                         'D' => 'D',       // convert description (subclass implement)
  77                         '-' => '-',       // remove convert (not implement)
  78                         'H' => 'H',       // add rule for convert code
  79                                       // (but no display in placed code )
  80                         'N' => 'N'        // current variant name
  81                 );
  82                 $this->mFlags = array_merge( $f, $flags );
  83                 foreach ( $this->mVariants as $v ) {
  84                         if ( array_key_exists( $v, $manualLevel ) ) {
  85                                 $this->mManualLevel[$v] = $manualLevel[$v];
  86                         } else {
  87                                 $this->mManualLevel[$v] = 'bidirectional';
  88                         }
  89                         $this->mFlags[$v] = $v;
  90                 }
  91         }
  92
  93         /**
  94          * @public
  95          */
  96         function getVariants() {
  97                 return $this->mVariants;
  98         }
  99
 100         /**
 101          * In case some variant is not defined in the markup, we need
 102          * to have some fallback. For example, in zh, normally people
 103          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 104          * when zh-sg is preferred but not defined, we will pick zh-hans
 105          * in this case. Right now this is only used by zh.
 106          *
 107          * @param string $v The language code of the variant
 108          * @return string array The code of the fallback language or false if there
 109          *                      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 the title produced by the conversion rule.
 121          * @returns string
 122          */
 123         function getConvRuleTitle() {
 124                 return $this->mConvRuleTitle;
 125         }
 126
 127         /**
 128          * Get preferred language variants.
 129          * @param boolean $fromUser Get it from $wgUser's preferences
 130          * @param boolean $fromHeader Get it from Accept-Language
 131          * @return string the preferred language code
 132          * @public
 133          */
 134         function getPreferredVariant( $fromUser = true, $fromHeader = false ) {
 135                 global $wgDefaultLanguageVariant;
 136
 137                 $req = $this->getURLVariant();
 138
 139                 if ( $fromUser && !$req ) {
 140                         $req = $this->getUserVariant();
 141                 }
 142
 143                 if ( $fromHeader && !$req ) {
 144                         $req = $this->getHeaderVariant();
 145                 }
 146
 147                 if ( $wgDefaultLanguageVariant && !$req ) {
 148                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 149                 }
 150
 151                 // This function, unlike the other get*Variant functions, is
 152                 // not memoized (i.e. there return value is not cached) since
 153                 // new information might appear during processing after this
 154                 // is first called.
 155                 if ( $req ) {
 156                         return $req;
 157                 }
 158                 return $this->mMainLanguageCode;
 159         }
 160
 161         /**
 162          * Validate the variant
 163          * @param string $v the variant to validate
 164          * @returns mixed returns the variant if it is valid, null otherwise
 165          */
 166         function validateVariant( $v = null ) {
 167                 if ( $v !== null && in_array( $v, $this->mVariants ) ) {
 168                         return $v;
 169                 }
 170                 return null;
 171         }
 172
 173         /**
 174          * Get the variant specified in the URL
 175          *
 176          * @returns mixed variant if one found, false otherwise.
 177          */
 178         function getURLVariant() {
 179                 global $wgRequest;
 180                 $ret = null;
 181
 182                 if ( $this->mURLVariant ) {
 183                         return $this->mURLVariant;
 184                 }
 185
 186                 // see if the preference is set in the request
 187                 $ret = $wgRequest->getText( 'variant' );
 188
 189                 if ( !$ret ) {
 190                         $ret = $wgRequest->getVal( 'uselang' );
 191                 }
 192
 193                 return $this->mURLVariant = $this->validateVariant( $ret );
 194         }
 195
 196         /**
 197          * Determine if the user has a variant set.
 198          *
 199          * @returns mixed variant if one found, false otherwise.
 200          */
 201         function getUserVariant() {
 202                 global $wgUser;
 203                 $ret = null;
 204
 205                 // memoizing this function wreaks havoc on parserTest.php
 206                 /* if ( $this->mUserVariant ) { */
 207                 /*      return $this->mUserVariant; */
 208                 /* } */
 209
 210                 // get language variant preference from logged in users
 211                 // Don't call this on stub objects because that causes infinite
 212                 // recursion during initialisation
 213                 if ( $wgUser->isLoggedIn() )  {
 214                         $ret = $wgUser->getOption( 'variant' );
 215                 }
 216                 else {
 217                         // figure out user lang without constructing wgLang to avoid
 218                         // infinite recursion
 219                         $ret = $wgUser->getOption( 'language' );
 220                 }
 221
 222                 return $this->mUserVariant = $this->validateVariant( $ret );
 223         }
 224
 225
 226         /**
 227          * Determine the language variant from the Accept-Language header.
 228          *
 229          * @returns mixed variant if one found, false otherwise.
 230          */
 231         function getHeaderVariant() {
 232                 global $wgRequest;
 233                 $ret = null;
 234
 235                 if ( $this->mHeaderVariant ) {
 236                         return $this->mHeaderVariant;
 237                 }
 238
 239                 // see if some supported language variant is set in the
 240                 // http header.
 241
 242                 $acceptLanguage = $wgRequest->getHeader( 'Accept-Language' );
 243                 if ( !$acceptLanguage ) {
 244                         return null;
 245                 }
 246
 247                 // explode by comma
 248                 $result = StringUtils::explode( ',', strtolower( $acceptLanguage ) );
 249                 $languages = array();
 250
 251                 foreach ( $result as $elem ) {
 252                         // if $elem likes 'zh-cn;q=0.9'
 253                         if ( ( $posi = strpos( $elem, ';' ) ) !== false ) {
 254                                 // get the real language code likes 'zh-cn'
 255                                 $languages[] = substr( $elem, 0, $posi );
 256                         } else {
 257                                 $languages[] = $elem;
 258                         }
 259                 }
 260
 261                 $fallback_languages = array();
 262                 foreach ( $languages as $language ) {
 263                         // strip whitespace
 264                         $language = trim( $language );
 265                         $this->mHeaderVariant = $this->validateVariant( $language );
 266                         if ( $this->mHeaderVariant ) {
 267                                 break;
 268                         }
 269
 270                         // To see if there are fallbacks of current language.
 271                         // We record these fallback variants, and process
 272                         // them later.
 273                         $fallbacks = $this->getVariantFallbacks( $language );
 274                         if ( is_string( $fallbacks ) ) {
 275                                 $fallback_languages[] = $fallbacks;
 276                         } elseif ( is_array( $fallbacks ) ) {
 277                                 $fallback_languages =
 278                                         array_merge( $fallback_languages,
 279                                                                  $fallbacks );
 280                         }
 281                 }
 282
 283                 if ( !$this->mHeaderVariant ) {
 284                         // process fallback languages now
 285                         $fallback_languages = array_unique( $fallback_languages );
 286                         foreach ( $fallback_languages as $language ) {
 287                                 $this->mHeaderVariant = $this->validateVariant( $language );
 288                                 if ( $this->mHeaderVariant ) {
 289                                         break;
 290                                 }
 291                         }
 292                 }
 293
 294                 return $this->mHeaderVariant;
 295         }
 296
 297         /**
 298          * Caption convert, base on preg_replace_callback.
 299          *
 300          * To convert text in "title" or "alt", like '<img alt="text" ... '
 301          * or '<span title="text" ... '
 302          *
 303          * @return string like ' alt="yyyy"' or ' title="yyyy"'
 304          * @private
 305          */
 306         function captionConvert( $matches ) {
 307                 $toVariant = $this->getPreferredVariant();
 308                 $title = $matches[1];
 309                 $text  = $matches[2];
 310                 // we convert captions except URL
 311                 if ( !strpos( $text, '://' ) ) {
 312                         $text = $this->translate( $text, $toVariant );
 313                 }
 314                 return " $title=\"$text\"";
 315         }
 316
 317         /**
 318          * Dictionary-based conversion.
 319          *
 320          * @param string $text the text to be converted
 321          * @param string $toVariant the target language code
 322          * @return string the converted text
 323          * @private
 324          */
 325         function autoConvert( $text, $toVariant = false ) {
 326                 $fname = 'LanguageConverter::autoConvert';
 327
 328                 wfProfileIn( $fname );
 329
 330                 if ( !$this->mTablesLoaded ) {
 331                         $this->loadTables();
 332                 }
 333
 334                 if ( !$toVariant ) {
 335                         $toVariant = $this->getPreferredVariant();
 336                         if ( !$toVariant ) {
 337                                 return $text;
 338                         }
 339                 }
 340
 341                 /* we convert everything except:
 342                    1. html markups (anything between < and >)
 343                    2. html entities
 344                    3. place holders created by the parser
 345                 */
 346                 global $wgParser;
 347                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 348                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 349                 } else {
 350                         $marker = '';
 351                 }
 352
 353                 // this one is needed when the text is inside an html markup
 354                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 355
 356                 // disable convert to variants between <code></code> tags
 357                 $codefix = '<code>.+?<\/code>|';
 358                 // disable convertsion of <script type="text/javascript"> ... </script>
 359                 $scriptfix = '<script.*?>.*?<\/script>|';
 360                 // disable conversion of <pre xxxx> ... </pre>
 361                 $prefix = '<pre.*?>.*?<\/pre>|';
 362
 363                 $reg = '/' . $codefix . $scriptfix . $prefix .
 364                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 365
 366                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 367
 368                 $m = array_shift( $matches );
 369
 370                 $ret = $this->translate( $m[0], $toVariant );
 371                 $mstart = $m[1] + strlen( $m[0] );
 372
 373                 // enable convertsion of '<img alt="xxxx" ... '
 374                 // or '<span title="xxxx" ... '
 375                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 376
 377                 $trtext = '';
 378                 $trtextmark = "\0";
 379                 $notrtext = array();
 380                 foreach ( $matches as $m ) {
 381                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 382                         $mark = preg_replace_callback( $captionpattern,
 383                                                                                    array( &$this, 'captionConvert' ),
 384                                                                                    $mark );
 385                         // Let's convert the trtext only once,
 386                         // it would give us more performance improvement
 387                         $notrtext[] = $mark;
 388                         $trtext .= $m[0] . $trtextmark;
 389                         $mstart = $m[1] + strlen( $m[0] );
 390                 }
 391                 $notrtext[] = '';
 392                 $trtext = $this->translate( $trtext, $toVariant );
 393                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 394                 foreach ( $trtext as $t ) {
 395                         $ret .= array_shift( $notrtext );
 396                         $ret .= $t;
 397                 }
 398                 wfProfileOut( $fname );
 399                 return $ret;
 400         }
 401
 402         /**
 403          * Translate a string to a variant.
 404          * Doesn't process markup or do any of that other stuff, for that use
 405          * convert().
 406          *
 407          * @param string $text Text to convert
 408          * @param string $variant Variant language code
 409          * @return string Translated text
 410          * @private
 411          */
 412         function translate( $text, $variant ) {
 413                 wfProfileIn( __METHOD__ );
 414                 // If $text is empty or only includes spaces, do nothing
 415                 // Otherwise translate it
 416                 if ( trim( $text ) ) {
 417                         if ( !$this->mTablesLoaded ) {
 418                                 $this->loadTables();
 419                         }
 420                         $text = $this->mTables[$variant]->replace( $text );
 421                 }
 422                 wfProfileOut( __METHOD__ );
 423                 return $text;
 424         }
 425
 426         /**
 427          * Convert text to all supported variants.
 428          *
 429          * @param string $text the text to be converted
 430          * @return array of string
 431          * @public
 432          */
 433         function autoConvertToAllVariants( $text ) {
 434                 $fname = 'LanguageConverter::autoConvertToAllVariants';
 435                 wfProfileIn( $fname );
 436                 if ( !$this->mTablesLoaded ) {
 437                         $this->loadTables();
 438                 }
 439
 440                 $ret = array();
 441                 foreach ( $this->mVariants as $variant ) {
 442                         $ret[$variant] = $this->translate( $text, $variant );
 443                 }
 444
 445                 wfProfileOut( $fname );
 446                 return $ret;
 447         }
 448
 449         /**
 450          * Convert link text to all supported variants.
 451          *
 452          * @param string $text the text to be converted
 453          * @return array of string
 454          * @public
 455          */
 456         function convertLinkToAllVariants( $text ) {
 457                 if ( !$this->mTablesLoaded ) {
 458                         $this->loadTables();
 459                 }
 460
 461                 $ret = array();
 462                 $tarray = StringUtils::explode( '-{', $text );
 463                 $first = true;
 464
 465                 foreach ( $tarray as $txt ) {
 466                         if ( $first ) {
 467                                 $first = false;
 468                                 foreach ( $this->mVariants as $variant ) {
 469                                         $ret[$variant] = $this->translate( $txt, $variant );
 470                                 }
 471                                 continue;
 472                         }
 473
 474                         $marked = explode( '}-', $txt, 2 );
 475
 476                         foreach ( $this->mVariants as $variant ) {
 477                                 $ret[$variant] .= '-{' . $marked[0] . '}-';
 478                                 if ( array_key_exists( 1, $marked ) ) {
 479                                         $ret[$variant] .= $this->translate( $marked[1], $variant );
 480                                 }
 481                         }
 482
 483                 }
 484
 485                 return $ret;
 486         }
 487
 488         /**
 489          * Prepare manual conversion table.
 490          * @private
 491          */
 492         function applyManualConv( $convRule ) {
 493                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 494                 // title conversion.
 495                 // Bug 24072: mConvRuleTitle won't work if the title conversion
 496                 // rule was followed by other manual conversion rule(s).
 497                 $newConvRuleTitle = $convRule->getTitle();
 498                 if( $newConvRuleTitle ) {
 499                         $this->mConvRuleTitle = $newConvRuleTitle;
 500                 }
 501
 502                 // apply manual conversion table to global table
 503                 $convTable = $convRule->getConvTable();
 504                 $action = $convRule->getRulesAction();
 505                 foreach ( $convTable as $variant => $pair ) {
 506                         if ( !$this->validateVariant( $variant ) ) {
 507                                 continue;
 508                         }
 509
 510                         if ( $action == 'add' ) {
 511                                 foreach ( $pair as $from => $to ) {
 512                                         // to ensure that $from and $to not be left blank
 513                                         // so $this->translate() could always return a string
 514                                         if ( $from || $to ) {
 515                                                 // more efficient than array_merge(), about 2.5 times.
 516                                                 $this->mTables[$variant]->setPair( $from, $to );
 517                                         }
 518                                 }
 519                         } elseif ( $action == 'remove' ) {
 520                                 $this->mTables[$variant]->removeArray( $pair );
 521                         }
 522                 }
 523         }
 524
 525         /**
 526          * Convert text to different variants of a language. The automatic
 527          * conversion is done in autoConvert(). Here we parse the text
 528          * marked with -{}-, which specifies special conversions of the
 529          * text that can not be accomplished in autoConvert().
 530          *
 531          * Syntax of the markup:
 532          * -{code1:text1;code2:text2;...}-  or
 533          * -{flags|code1:text1;code2:text2;...}-  or
 534          * -{text}- in which case no conversion should take place for text
 535          *
 536          * @param $text String: text to be converted
 537          * @return String: converted text
 538          */
 539         public function convert( $text ) {
 540                 global $wgDisableLangConversion;
 541                 if ( $wgDisableLangConversion ) return $text;
 542
 543                 $variant = $this->getPreferredVariant();
 544
 545                 return $this->recursiveConvertTopLevel( $text, $variant );
 546         }
 547
 548         /**
 549          * Convert a Title object to a readable string in the preferred variant
 550          */
 551         public function convertTitle( $title ) {
 552                 $variant = $this->getPreferredVariant();
 553                 $index = $title->getNamespace();
 554                 if ( $index === NS_MAIN || $index === NS_SPECIAL ) {
 555                         $text = '';
 556                 } else {
 557                         // first let's check if a message has given us a converted name
 558                         $nsConvKey = 'conversion-ns' . $index;
 559                         $nsLocalText = wfMsgForContentNoTrans( $nsConvKey );
 560                         if ( !wfEmptyMsg( $nsConvKey, $nsLocalText ) ) {
 561                                 $text = $nsLocalText;
 562                         } else {
 563                                 // the message does not exist, try retrieve it from the current
 564                                 // variant's namespace names.
 565                                 $langObj = $this->mLangObj->factory( $variant );
 566                                 $text = $langObj->getFormattedNsText( $index );
 567                         }
 568                         $text .= ':';
 569                 }
 570                 $text .= $title->getText();
 571                 $text = $this->autoConvert( $text, $variant );
 572                 return $text;
 573         }
 574
 575         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 576                 $startPos = 0;
 577                 $out = '';
 578                 $length = strlen( $text );
 579                 while ( $startPos < $length ) {
 580                         $m = false;
 581                         $pos = strpos( $text, '-{', $startPos );
 582
 583                         if ( $pos === false ) {
 584                                 // No more markup, append final segment
 585                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 586                                 $startPos = $length;
 587                                 return $out;
 588                         }
 589
 590                         // Markup found
 591                         // Append initial segment
 592                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 593
 594                         // Advance position
 595                         $startPos = $pos;
 596
 597                         // Do recursive conversion
 598                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 599                 }
 600
 601                 return $out;
 602         }
 603
 604         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 605                 // Quick sanity check (no function calls)
 606                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 607                         throw new MWException( __METHOD__.': invalid input string' );
 608                 }
 609
 610                 $startPos += 2;
 611                 $inner = '';
 612                 $warningDone = false;
 613                 $length = strlen( $text );
 614
 615                 while ( $startPos < $length ) {
 616                         $m = false;
 617                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 618                         if ( !$m ) {
 619                                 // Unclosed rule
 620                                 break;
 621                         }
 622
 623                         $token = $m[0][0];
 624                         $pos = $m[0][1];
 625
 626                         // Markup found
 627                         // Append initial segment
 628                         $inner .= substr( $text, $startPos, $pos - $startPos );
 629
 630                         // Advance position
 631                         $startPos = $pos;
 632
 633                         switch ( $token ) {
 634                                 case '-{':
 635                                         // Check max depth
 636                                         if ( $depth >= $this->mMaxDepth ) {
 637                                                 $inner .= '-{';
 638                                                 if ( !$warningDone ) {
 639                                                         $inner .= '<span class="error">' .
 640                                                                 wfMsgForContent( 'language-converter-depth-warning',
 641                                                                         $this->mMaxDepth ) .
 642                                                                 '</span>';
 643                                                         $warningDone = true;
 644                                                 }
 645                                                 $startPos += 2;
 646                                                 continue;
 647                                         }
 648                                         // Recursively parse another rule
 649                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 650                                         break;
 651                                 case '}-':
 652                                         // Apply the rule
 653                                         $startPos += 2;
 654                                         $rule = new ConverterRule( $inner, $this );
 655                                         $rule->parse( $variant );
 656                                         $this->applyManualConv( $rule );
 657                                         return $rule->getDisplay();
 658                                 default:
 659                                         throw new MWException( __METHOD__.': invalid regex match' );
 660                         }
 661                 }
 662
 663                 // Unclosed rule
 664                 if ( $startPos < $length ) {
 665                         $inner .= substr( $text, $startPos );
 666                 }
 667                 $startPos = $length;
 668                 return '-{' . $this->autoConvert( $inner, $variant );
 669         }
 670
 671         /**
 672          * If a language supports multiple variants, it is
 673          * possible that non-existing link in one variant
 674          * actually exists in another variant. This function
 675          * tries to find it. See e.g. LanguageZh.php
 676          *
 677          * @param string $link the name of the link
 678          * @param mixed $nt the title object of the link
 679          * @param boolean $ignoreOtherCond: to disable other conditions when
 680          *      we need to transclude a template or update a category's link
 681          * @return null the input parameters may be modified upon return
 682          * @public
 683          */
 684         function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 685                 # If the article has already existed, there is no need to
 686                 # check it again, otherwise it may cause a fault.
 687                 if ( is_object( $nt ) && $nt->exists() ) {
 688                         return;
 689                 }
 690
 691                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 692                         $wgUser;
 693                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 694                 $action = $wgRequest->getText( 'action' );
 695                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 696                 $disableLinkConversion = $wgDisableLangConversion
 697                         || $wgDisableTitleConversion;
 698                 $linkBatch = new LinkBatch();
 699
 700                 $ns = NS_MAIN;
 701
 702                 if ( $disableLinkConversion ||
 703                          ( !$ignoreOtherCond &&
 704                            ( $isredir == 'no'
 705                                  || $action == 'edit'
 706                                  || $action == 'submit'
 707                                  || $linkconvert == 'no'
 708                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 709                         return;
 710                 }
 711
 712                 if ( is_object( $nt ) ) {
 713                         $ns = $nt->getNamespace();
 714                 }
 715
 716                 $variants = $this->autoConvertToAllVariants( $link );
 717                 if ( $variants == false ) { // give up
 718                         return;
 719                 }
 720
 721                 $titles = array();
 722
 723                 foreach ( $variants as $v ) {
 724                         if ( $v != $link ) {
 725                                 $varnt = Title::newFromText( $v, $ns );
 726                                 if ( !is_null( $varnt ) ) {
 727                                         $linkBatch->addObj( $varnt );
 728                                         $titles[] = $varnt;
 729                                 }
 730                         }
 731                 }
 732
 733                 // fetch all variants in single query
 734                 $linkBatch->execute();
 735
 736                 foreach ( $titles as $varnt ) {
 737                         if ( $varnt->getArticleID() > 0 ) {
 738                                 $nt = $varnt;
 739                                 $link = $varnt->getText();
 740                                 break;
 741                         }
 742                 }
 743         }
 744
 745     /**
 746          * Returns language specific hash options.
 747          *
 748          * @public
 749          */
 750         function getExtraHashOptions() {
 751                 $variant = $this->getPreferredVariant();
 752                 return '!' . $variant ;
 753         }
 754
 755         /**
 756          * Load default conversion tables.
 757          * This method must be implemented in derived class.
 758          *
 759          * @private
 760          */
 761         function loadDefaultTables() {
 762                 $name = get_class( $this );
 763                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 764         }
 765
 766         /**
 767          * Load conversion tables either from the cache or the disk.
 768          * @private
 769          */
 770         function loadTables( $fromcache = true ) {
 771                 global $wgMemc;
 772                 if ( $this->mTablesLoaded ) {
 773                         return;
 774                 }
 775                 wfProfileIn( __METHOD__ );
 776                 $this->mTablesLoaded = true;
 777                 $this->mTables = false;
 778                 if ( $fromcache ) {
 779                         wfProfileIn( __METHOD__ . '-cache' );
 780                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 781                         wfProfileOut( __METHOD__ . '-cache' );
 782                 }
 783                 if ( !$this->mTables
 784                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 785                         wfProfileIn( __METHOD__ . '-recache' );
 786                         // not in cache, or we need a fresh reload.
 787                         // we will first load the default tables
 788                         // then update them using things in MediaWiki:Zhconversiontable/*
 789                         $this->loadDefaultTables();
 790                         foreach ( $this->mVariants as $var ) {
 791                                 $cached = $this->parseCachedTable( $var );
 792                                 $this->mTables[$var]->mergeArray( $cached );
 793                         }
 794
 795                         $this->postLoadTables();
 796                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 797
 798                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 799                         wfProfileOut( __METHOD__ . '-recache' );
 800                 }
 801                 wfProfileOut( __METHOD__ );
 802         }
 803
 804     /**
 805          * Hook for post processig after conversion tables are loaded.
 806          *
 807          */
 808         function postLoadTables() { }
 809
 810     /**
 811          * Reload the conversion tables.
 812          *
 813          * @private
 814          */
 815         function reloadTables() {
 816                 if ( $this->mTables ) {
 817                         unset( $this->mTables );
 818                 }
 819                 $this->mTablesLoaded = false;
 820                 $this->loadTables( false );
 821         }
 822
 823
 824         /**
 825          * Parse the conversion table stored in the cache.
 826          *
 827          * The tables should be in blocks of the following form:
 828          *              -{
 829          *                      word => word ;
 830          *                      word => word ;
 831          *                      ...
 832          *              }-
 833          *
 834          *      To make the tables more manageable, subpages are allowed
 835          *      and will be parsed recursively if $recursive == true.
 836          *
 837          */
 838         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 839                 global $wgMessageCache;
 840                 static $parsed = array();
 841
 842                 if ( !is_object( $wgMessageCache ) ) {
 843                         return array();
 844                 }
 845
 846                 $key = 'Conversiontable/' . $code;
 847                 if ( $subpage ) {
 848                         $key .= '/' . $subpage;
 849                 }
 850                 if ( array_key_exists( $key, $parsed ) ) {
 851                         return array();
 852                 }
 853
 854                 if ( strpos( $code, '/' ) === false ) {
 855                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 856                 } else {
 857                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 858                                                                                    "Conversiontable/$code" );
 859                         if ( $title && $title->exists() ) {
 860                                 $article = new Article( $title );
 861                                 $txt = $article->getContents();
 862                         } else {
 863                                 $txt = '';
 864                         }
 865                 }
 866
 867                 // get all subpage links of the form
 868                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 869                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 870                         ':Conversiontable';
 871                 $subs = StringUtils::explode( '[[', $txt );
 872                 $sublinks = array();
 873                 foreach ( $subs as $sub ) {
 874                         $link = explode( ']]', $sub, 2 );
 875                         if ( count( $link ) != 2 ) {
 876                                 continue;
 877                         }
 878                         $b = explode( '|', $link[0], 2 );
 879                         $b = explode( '/', trim( $b[0] ), 3 );
 880                         if ( count( $b ) == 3 ) {
 881                                 $sublink = $b[2];
 882                         } else {
 883                                 $sublink = '';
 884                         }
 885
 886                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 887                                 $sublinks[] = $sublink;
 888                         }
 889                 }
 890
 891
 892                 // parse the mappings in this page
 893                 $blocks = StringUtils::explode( '-{', $txt );
 894                 $ret = array();
 895                 $first = true;
 896                 foreach ( $blocks as $block ) {
 897                         if ( $first ) {
 898                                 // Skip the part before the first -{
 899                                 $first = false;
 900                                 continue;
 901                         }
 902                         $mappings = explode( '}-', $block, 2 );
 903                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 904                                                                          $mappings[0] );
 905                         $table = StringUtils::explode( ';', $stripped );
 906                         foreach ( $table as $t ) {
 907                                 $m = explode( '=>', $t, 3 );
 908                                 if ( count( $m ) != 2 )
 909                                         continue;
 910                                 // trim any trailling comments starting with '//'
 911                                 $tt = explode( '//', $m[1], 2 );
 912                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 913                         }
 914                 }
 915                 $parsed[$key] = true;
 916
 917
 918                 // recursively parse the subpages
 919                 if ( $recursive ) {
 920                         foreach ( $sublinks as $link ) {
 921                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 922                                 $ret = array_merge( $ret, $s );
 923                         }
 924                 }
 925
 926                 if ( $this->mUcfirst ) {
 927                         foreach ( $ret as $k => $v ) {
 928                                 $ret[Language::ucfirst( $k )] = Language::ucfirst( $v );
 929                         }
 930                 }
 931                 return $ret;
 932         }
 933
 934         /**
 935          * Enclose a string with the "no conversion" tag. This is used by
 936          * various functions in the Parser.
 937          *
 938          * @param string $text text to be tagged for no conversion
 939          * @return string the tagged text
 940          * @public
 941          */
 942         function markNoConversion( $text, $noParse = false ) {
 943                 # don't mark if already marked
 944                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 945                         return $text;
 946                 }
 947
 948                 $ret = "-{R|$text}-";
 949                 return $ret;
 950         }
 951
 952         /**
 953          * Convert the sorting key for category links. This should make different
 954          * keys that are variants of each other map to the same key.
 955          */
 956         function convertCategoryKey( $key ) {
 957                 return $key;
 958         }
 959
 960         /**
 961          * Hook to refresh the cache of conversion tables when
 962          * MediaWiki:conversiontable* is updated.
 963          * @private
 964          */
 965         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 966                         $iswatch, $section, $flags, $revision ) {
 967                 $titleobj = $article->getTitle();
 968                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 969                         $title = $titleobj->getDBkey();
 970                         $t = explode( '/', $title, 3 );
 971                         $c = count( $t );
 972                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 973                                 if ( $this->validateVariant( $t[1] ) ) {
 974                                         $this->reloadTables();
 975                                 }
 976                         }
 977                 }
 978                 return true;
 979         }
 980
 981         /**
 982          * Armour rendered math against conversion.
 983          * Wrap math into rawoutput -{R| math }- syntax.
 984          * @public
 985          */
 986         function armourMath( $text ) {
 987                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 988                 // to avoid a unwanted '}-' appeared after the math-image.
 989                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 990                 $ret = "-{R|$text}-";
 991                 return $ret;
 992         }
 993
 994         /**
 995          * Get the cached separator pattern for ConverterRule::parseRules()
 996          */
 997         function getVarSeparatorPattern() {
 998                 if ( is_null( $this->mVarSeparatorPattern ) ) {
 999                         // varsep_pattern for preg_split:
1000                         // text should be splited by ";" only if a valid variant
1001                         // name exist after the markup, for example:
1002                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1003                         //    <span style="font-size:120%;">yyy</span>;}-
1004                         // we should split it as:
1005                         //  array(
1006                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1007                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1008                         //        [2] => ''
1009                         //       )
1010                         $pat = '/;\s*(?=';
1011                         foreach ( $this->mVariants as $variant ) {
1012                                 // zh-hans:xxx;zh-hant:yyy
1013                                 $pat .= $variant . '\s*:|';
1014                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1015                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1016                         }
1017                         $pat .= '\s*$)/';
1018                         $this->mVarSeparatorPattern = $pat;
1019                 }
1020                 return $this->mVarSeparatorPattern;
1021         }
1022 }
1023
1024 /**
1025  * Parser for rules of language conversion , parse rules in -{ }- tag.
1026  * @ingroup Language
1027  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1028  */
1029 class ConverterRule {
1030         var $mText; // original text in -{text}-
1031         var $mConverter; // LanguageConverter object
1032         var $mManualCodeError = '<strong class="error">code error!</strong>';
1033         var $mRuleDisplay = '';
1034         var $mRuleTitle = false;
1035         var $mRules = '';// string : the text of the rules
1036         var $mRulesAction = 'none';
1037         var $mFlags = array();
1038         var $mVariantFlags = array();
1039         var $mConvTable = array();
1040         var $mBidtable = array();// array of the translation in each variant
1041         var $mUnidtable = array();// array of the translation in each variant
1042
1043         /**
1044          * Constructor
1045          *
1046          * @param $text String: the text between -{ and }-
1047          * @param $converter LanguageConverter object
1048          */
1049         public function __construct( $text, $converter ) {
1050                 $this->mText = $text;
1051                 $this->mConverter = $converter;
1052         }
1053
1054         /**
1055          * Check if variants array in convert array.
1056          *
1057          * @param $variants Array or string: variant language code
1058          * @return String: translated text
1059          */
1060         public function getTextInBidtable( $variants ) {
1061                 $variants = (array)$variants;
1062                 if ( !$variants ) {
1063                         return false;
1064                 }
1065                 foreach ( $variants as $variant ) {
1066                         if ( isset( $this->mBidtable[$variant] ) ) {
1067                                 return $this->mBidtable[$variant];
1068                         }
1069                 }
1070                 return false;
1071         }
1072
1073         /**
1074          * Parse flags with syntax -{FLAG| ... }-
1075          * @private
1076          */
1077         function parseFlags() {
1078                 $text = $this->mText;
1079                 $flags = array();
1080                 $variantFlags = array();
1081
1082                 $sepPos = strpos( $text, '|' );
1083                 if ( $sepPos !== false ) {
1084                         $validFlags = $this->mConverter->mFlags;
1085                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1086                         foreach ( $f as $ff ) {
1087                                 $ff = trim( $ff );
1088                                 if ( isset( $validFlags[$ff] ) ) {
1089                                         $flags[$validFlags[$ff]] = true;
1090                                 }
1091                         }
1092                         $text = strval( substr( $text, $sepPos + 1 ) );
1093                 }
1094
1095                 if ( !$flags ) {
1096                         $flags['S'] = true;
1097                 } elseif ( isset( $flags['R'] ) ) {
1098                         $flags = array( 'R' => true );// remove other flags
1099                 } elseif ( isset( $flags['N'] ) ) {
1100                         $flags = array( 'N' => true );// remove other flags
1101                 } elseif ( isset( $flags['-'] ) ) {
1102                         $flags = array( '-' => true );// remove other flags
1103                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1104                         $flags['H'] = true;
1105                 } elseif ( isset( $flags['H'] ) ) {
1106                         // replace A flag, and remove other flags except T
1107                         $temp = array( '+' => true, 'H' => true );
1108                         if ( isset( $flags['T'] ) ) {
1109                                 $temp['T'] = true;
1110                         }
1111                         if ( isset( $flags['D'] ) ) {
1112                                 $temp['D'] = true;
1113                         }
1114                         $flags = $temp;
1115                 } else {
1116                         if ( isset( $flags['A'] ) ) {
1117                                 $flags['+'] = true;
1118                                 $flags['S'] = true;
1119                         }
1120                         if ( isset( $flags['D'] ) ) {
1121                                 unset( $flags['S'] );
1122                         }
1123                         // try to find flags like "zh-hans", "zh-hant"
1124                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1125                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1126                         if ( $variantFlags ) {
1127                                 $variantFlags = array_flip( $variantFlags );
1128                                 $flags = array();
1129                         }
1130                 }
1131                 $this->mVariantFlags = $variantFlags;
1132                 $this->mRules = $text;
1133                 $this->mFlags = $flags;
1134         }
1135
1136         /**
1137          * Generate conversion table.
1138          * @private
1139          */
1140         function parseRules() {
1141                 $rules = $this->mRules;
1142                 $flags = $this->mFlags;
1143                 $bidtable = array();
1144                 $unidtable = array();
1145                 $variants = $this->mConverter->mVariants;
1146                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1147
1148                 $choice = preg_split( $varsep_pattern, $rules );
1149
1150                 foreach ( $choice as $c ) {
1151                         $v  = explode( ':', $c, 2 );
1152                         if ( count( $v ) != 2 ) {
1153                                 // syntax error, skip
1154                                 continue;
1155                         }
1156                         $to = trim( $v[1] );
1157                         $v  = trim( $v[0] );
1158                         $u  = explode( '=>', $v, 2 );
1159                         // if $to is empty, strtr() could return a wrong result
1160                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1161                                 $bidtable[$v] = $to;
1162                         } elseif ( count( $u ) == 2 ) {
1163                                 $from = trim( $u[0] );
1164                                 $v    = trim( $u[1] );
1165                                 if ( array_key_exists( $v, $unidtable )
1166                                          && !is_array( $unidtable[$v] )
1167                                          && $to
1168                                          && in_array( $v, $variants ) ) {
1169                                         $unidtable[$v] = array( $from => $to );
1170                                 } elseif ( $to && in_array( $v, $variants ) ) {
1171                                         $unidtable[$v][$from] = $to;
1172                                 }
1173                         }
1174                         // syntax error, pass
1175                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1176                                 $bidtable = array();
1177                                 $unidtable = array();
1178                                 break;
1179                         }
1180                 }
1181                 $this->mBidtable = $bidtable;
1182                 $this->mUnidtable = $unidtable;
1183         }
1184
1185         /**
1186          * @private
1187          */
1188         function getRulesDesc() {
1189                 $codesep = $this->mConverter->mDescCodeSep;
1190                 $varsep = $this->mConverter->mDescVarSep;
1191                 $text = '';
1192                 foreach ( $this->mBidtable as $k => $v ) {
1193                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1194                 }
1195                 foreach ( $this->mUnidtable as $k => $a ) {
1196                         foreach ( $a as $from => $to ) {
1197                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1198                                         "$codesep$to$varsep";
1199                         }
1200                 }
1201                 return $text;
1202         }
1203
1204         /**
1205          * Parse rules conversion.
1206          * @private
1207          */
1208         function getRuleConvertedStr( $variant ) {
1209                 $bidtable = $this->mBidtable;
1210                 $unidtable = $this->mUnidtable;
1211
1212                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1213                         return $this->mRules;
1214                 } else {
1215                         // display current variant in bidirectional array
1216                         $disp = $this->getTextInBidtable( $variant );
1217                         // or display current variant in fallbacks
1218                         if ( !$disp ) {
1219                                 $disp = $this->getTextInBidtable(
1220                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1221                         }
1222                         // or display current variant in unidirectional array
1223                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1224                                 $disp = array_values( $unidtable[$variant] );
1225                                 $disp = $disp[0];
1226                         }
1227                         // or display frist text under disable manual convert
1228                         if ( !$disp
1229                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1230                                 if ( count( $bidtable ) > 0 ) {
1231                                         $disp = array_values( $bidtable );
1232                                         $disp = $disp[0];
1233                                 } else {
1234                                         $disp = array_values( $unidtable );
1235                                         $disp = array_values( $disp[0] );
1236                                         $disp = $disp[0];
1237                                 }
1238                         }
1239                         return $disp;
1240                 }
1241         }
1242
1243         /**
1244          * Generate conversion table for all text.
1245          * @private
1246          */
1247         function generateConvTable() {
1248                 // Special case optimisation
1249                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1250                         $this->mConvTable = array();
1251                         return;
1252                 }
1253
1254                 $bidtable = $this->mBidtable;
1255                 $unidtable = $this->mUnidtable;
1256                 $manLevel = $this->mConverter->mManualLevel;
1257
1258                 $vmarked = array();
1259                 foreach ( $this->mConverter->mVariants as $v ) {
1260                         /* for bidirectional array
1261                                 fill in the missing variants, if any,
1262                                 with fallbacks */
1263                         if ( !isset( $bidtable[$v] ) ) {
1264                                 $variantFallbacks =
1265                                         $this->mConverter->getVariantFallbacks( $v );
1266                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1267                                 if ( $vf ) {
1268                                         $bidtable[$v] = $vf;
1269                                 }
1270                         }
1271
1272                         if ( isset( $bidtable[$v] ) ) {
1273                                 foreach ( $vmarked as $vo ) {
1274                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1275                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1276                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1277                                         // to introduce a custom mapping between
1278                                         // words WordZh and WordTw in the whole text
1279                                         if ( $manLevel[$v] == 'bidirectional' ) {
1280                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1281                                         }
1282                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1283                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1284                                         }
1285                                 }
1286                                 $vmarked[] = $v;
1287                         }
1288                         /*for unidirectional array fill to convert tables */
1289                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1290                                 && isset( $unidtable[$v] ) )
1291                         {
1292                                 if ( isset( $this->mConvTable[$v] ) ) {
1293                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1294                                 } else {
1295                                         $this->mConvTable[$v] = $unidtable[$v];
1296                                 }
1297                         }
1298                 }
1299         }
1300
1301         /**
1302          * Parse rules and flags.
1303          * @public
1304          */
1305         function parse( $variant = NULL ) {
1306                 if ( !$variant ) {
1307                         $variant = $this->mConverter->getPreferredVariant();
1308                 }
1309
1310                 $variants = $this->mConverter->mVariants;
1311                 $this->parseFlags();
1312                 $flags = $this->mFlags;
1313
1314                 // convert to specified variant
1315                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1316                 if ( $this->mVariantFlags ) {
1317                         // check if current variant in flags
1318                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1319                                 // then convert <text to convert> to current language
1320                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1321                                                                                                                                 $variant );
1322                         } else { // if current variant no in flags,
1323                                    // then we check its fallback variants.
1324                                 $variantFallbacks =
1325                                         $this->mConverter->getVariantFallbacks( $variant );
1326                                 foreach ( $variantFallbacks as $variantFallback ) {
1327                                         // if current variant's fallback exist in flags
1328                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1329                                                 // then convert <text to convert> to fallback language
1330                                                 $this->mRules =
1331                                                         $this->mConverter->autoConvert( $this->mRules,
1332                                                                                                                         $variantFallback );
1333                                                 break;
1334                                         }
1335                                 }
1336                         }
1337                         $this->mFlags = $flags = array( 'R' => true );
1338                 }
1339
1340                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1341                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1342                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1343                         $this->parseRules();
1344                 }
1345                 $rules = $this->mRules;
1346
1347                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1348                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1349                                 // fill all variants if text in -{A/H/-|text} without rules
1350                                 foreach ( $this->mConverter->mVariants as $v ) {
1351                                         $this->mBidtable[$v] = $rules;
1352                                 }
1353                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1354                                 $this->mFlags = $flags = array( 'R' => true );
1355                         }
1356                 }
1357
1358                 $this->mRuleDisplay = false;
1359                 foreach ( $flags as $flag => $unused ) {
1360                         switch ( $flag ) {
1361                                 case 'R':
1362                                         // if we don't do content convert, still strip the -{}- tags
1363                                         $this->mRuleDisplay = $rules;
1364                                         break;
1365                                 case 'N':
1366                                         // process N flag: output current variant name
1367                                         $ruleVar = trim( $rules );
1368                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1369                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1370                                         } else {
1371                                                 $this->mRuleDisplay = '';
1372                                         }
1373                                         break;
1374                                 case 'D':
1375                                         // process D flag: output rules description
1376                                         $this->mRuleDisplay = $this->getRulesDesc();
1377                                         break;
1378                                 case 'H':
1379                                         // process H,- flag or T only: output nothing
1380                                         $this->mRuleDisplay = '';
1381                                         break;
1382                                 case '-':
1383                                         $this->mRulesAction = 'remove';
1384                                         $this->mRuleDisplay = '';
1385                                         break;
1386                                 case '+':
1387                                         $this->mRulesAction = 'add';
1388                                         $this->mRuleDisplay = '';
1389                                         break;
1390                                 case 'S':
1391                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1392                                         break;
1393                                 case 'T':
1394                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1395                                         $this->mRuleDisplay = '';
1396                                         break;
1397                                 default:
1398                                         // ignore unknown flags (but see error case below)
1399                         }
1400                 }
1401                 if ( $this->mRuleDisplay === false ) {
1402                         $this->mRuleDisplay = $this->mManualCodeError;
1403                 }
1404
1405                 $this->generateConvTable();
1406         }
1407
1408         /**
1409          * @public
1410          */
1411         function hasRules() {
1412                 // TODO:
1413         }
1414
1415         /**
1416          * Get display text on markup -{...}-
1417          * @public
1418          */
1419         function getDisplay() {
1420                 return $this->mRuleDisplay;
1421         }
1422
1423         /**
1424          * Get converted title.
1425          * @public
1426          */
1427         function getTitle() {
1428                 return $this->mRuleTitle;
1429         }
1430
1431         /**
1432          * Return how deal with conversion rules.
1433          * @public
1434          */
1435         function getRulesAction() {
1436                 return $this->mRulesAction;
1437         }
1438
1439         /**
1440          * Get conversion table. ( bidirectional and unidirectional
1441          * conversion table )
1442          * @public
1443          */
1444         function getConvTable() {
1445                 return $this->mConvTable;
1446         }
1447
1448         /**
1449          * Get conversion rules string.
1450          * @public
1451          */
1452         function getRules() {
1453                 return $this->mRules;
1454         }
1455
1456         /**
1457          * Get conversion flags.
1458          * @public
1459          */
1460         function getFlags() {
1461                 return $this->mFlags;
1462         }
1463 }