languages/LanguageConverter.php

   1 <?php
   2 /**
   3  * Contains the LanguageConverter class and ConverterRule class
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Language
  22  */
  23
  24 /**
  25  * Base class for language conversion.
  26  * @ingroup Language
  27  *
  28  * @author Zhengzhu Feng <zhengzhu@gmail.com>
  29  * @maintainers fdcn <fdcn64@gmail.com>, shinjiman <shinjiman@gmail.com>, PhiLiP <philip.npc@gmail.com>
  30  */
  31 class LanguageConverter {
  32         var $mMainLanguageCode;
  33         var $mVariants, $mVariantFallbacks, $mVariantNames;
  34         var $mTablesLoaded = false;
  35         var $mTables;
  36         // 'bidirectional' 'unidirectional' 'disable' for each variant
  37         var $mManualLevel;
  38         var $mCacheKey;
  39         var $mLangObj;
  40         var $mFlags;
  41         var $mDescCodeSep = ':', $mDescVarSep = ';';
  42         var $mUcfirst = false;
  43         var $mConvRuleTitle = false;
  44         var $mURLVariant;
  45         var $mUserVariant;
  46         var $mHeaderVariant;
  47         var $mMaxDepth = 10;
  48         var $mVarSeparatorPattern;
  49
  50         const CACHE_VERSION_KEY = 'VERSION 6';
  51
  52         /**
  53          * Constructor
  54          *
  55          * @param $langobj The Language Object
  56          * @param $maincode String: the main language code of this language
  57          * @param $variants Array: the supported variants of this language
  58          * @param $variantfallbacks Array: the fallback language of each variant
  59          * @param $flags Array: defining the custom strings that maps to the flags
  60          * @param $manualLevel Array: limit for supported variants
  61          */
  62         public function __construct( $langobj, $maincode,
  63                                                                 $variants = array(),
  64                                                                 $variantfallbacks = array(),
  65                                                                 $flags = array(),
  66                                                                 $manualLevel = array() ) {
  67                 global $wgDisabledVariants, $wgLanguageNames;
  68                 $this->mLangObj = $langobj;
  69                 $this->mMainLanguageCode = $maincode;
  70                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  71                 $this->mVariantFallbacks = $variantfallbacks;
  72                 $this->mVariantNames = $wgLanguageNames;
  73                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  74                 $defaultflags = array(
  75                         // 'S' show converted text
  76                         // '+' add rules for alltext
  77                         // 'E' the gave flags is error
  78                         // these flags above are reserved for program
  79                         'A' => 'A',       // add rule for convert code (all text convert)
  80                         'T' => 'T',       // title convert
  81                         'R' => 'R',       // raw content
  82                         'D' => 'D',       // convert description (subclass implement)
  83                         '-' => '-',       // remove convert (not implement)
  84                         'H' => 'H',       // add rule for convert code
  85                                                   // (but no display in placed code )
  86                         'N' => 'N'        // current variant name
  87                 );
  88                 $this->mFlags = array_merge( $defaultflags, $flags );
  89                 foreach ( $this->mVariants as $v ) {
  90                         if ( array_key_exists( $v, $manualLevel ) ) {
  91                                 $this->mManualLevel[$v] = $manualLevel[$v];
  92                         } else {
  93                                 $this->mManualLevel[$v] = 'bidirectional';
  94                         }
  95                         $this->mFlags[$v] = $v;
  96                 }
  97         }
  98
  99         /**
 100          * Get all valid variants.
 101          * Call this instead of using $this->mVariants directly.
 102          *
 103          * @return Array: contains all valid variants
 104          */
 105         public function getVariants() {
 106                 return $this->mVariants;
 107         }
 108
 109         /**
 110          * In case some variant is not defined in the markup, we need
 111          * to have some fallback. For example, in zh, normally people
 112          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 113          * when zh-sg is preferred but not defined, we will pick zh-hans
 114          * in this case. Right now this is only used by zh.
 115          *
 116          * @param $variant String: the language code of the variant
 117          * @return String: The code of the fallback language or the
 118          *                               main code if there is no fallback
 119          */
 120         public function getVariantFallbacks( $variant ) {
 121                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 122                         return $this->mVariantFallbacks[$variant];
 123                 }
 124                 return $this->mMainLanguageCode;
 125         }
 126
 127         /**
 128          * Get the title produced by the conversion rule.
 129          * @return String: The converted title text
 130          */
 131         public function getConvRuleTitle() {
 132                 return $this->mConvRuleTitle;
 133         }
 134
 135         /**
 136          * Get preferred language variant.
 137          * @return String: the preferred language code
 138          */
 139         public function getPreferredVariant() {
 140                 global $wgDefaultLanguageVariant, $wgUser;
 141
 142                 $req = $this->getURLVariant();
 143
 144                 if ( $wgUser->isLoggedIn() && !$req ) {
 145                         $req = $this->getUserVariant();
 146                 }
 147
 148                 elseif ( !$req ) {
 149                         $req = $this->getHeaderVariant();
 150                 }
 151
 152                 if ( $wgDefaultLanguageVariant && !$req ) {
 153                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 154                 }
 155
 156                 // This function, unlike the other get*Variant functions, is
 157                 // not memoized (i.e. there return value is not cached) since
 158                 // new information might appear during processing after this
 159                 // is first called.
 160                 if ( $req ) {
 161                         return $req;
 162                 }
 163                 return $this->mMainLanguageCode;
 164         }
 165
 166         /**
 167          * Get default variant.
 168          * This function would not be affected by user's settings or headers
 169          * @return String: the default variant code
 170          */
 171         public function getDefaultVariant() {
 172                 global $wgDefaultLanguageVariant;
 173
 174                 $req = $this->getURLVariant();
 175
 176                 if ( $wgDefaultLanguageVariant && !$req ) {
 177                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 178                 }
 179
 180                 if ( $req ) {
 181                         return $req;
 182                 }
 183                 return $this->mMainLanguageCode;
 184         }
 185
 186         /**
 187          * Validate the variant
 188          * @param $variant String: the variant to validate
 189          * @return Mixed: returns the variant if it is valid, null otherwise
 190          */
 191         protected function validateVariant( $variant = null ) {
 192                 if ( $variant !== null &&
 193                          in_array( $variant, $this->mVariants ) ) {
 194                         return $variant;
 195                 }
 196                 return null;
 197         }
 198
 199         /**
 200          * Get the variant specified in the URL
 201          *
 202          * @return Mixed: variant if one found, false otherwise.
 203          */
 204         public function getURLVariant() {
 205                 global $wgRequest;
 206
 207                 if ( $this->mURLVariant ) {
 208                         return $this->mURLVariant;
 209                 }
 210
 211                 // see if the preference is set in the request
 212                 $ret = $wgRequest->getText( 'variant' );
 213
 214                 if ( !$ret ) {
 215                         $ret = $wgRequest->getVal( 'uselang' );
 216                 }
 217
 218                 return $this->mURLVariant = $this->validateVariant( $ret );
 219         }
 220
 221         /**
 222          * Determine if the user has a variant set.
 223          *
 224          * @return Mixed: variant if one found, false otherwise.
 225          */
 226         protected function getUserVariant() {
 227                 global $wgUser;
 228
 229                 // memoizing this function wreaks havoc on parserTest.php
 230                 /* if ( $this->mUserVariant ) { */
 231                 /*      return $this->mUserVariant; */
 232                 /* } */
 233
 234                 // get language variant preference from logged in users
 235                 // Don't call this on stub objects because that causes infinite
 236                 // recursion during initialisation
 237                 if ( $wgUser->isLoggedIn() )  {
 238                         $ret = $wgUser->getOption( 'variant' );
 239                 }
 240                 else {
 241                         // figure out user lang without constructing wgLang to avoid
 242                         // infinite recursion
 243                         $ret = $wgUser->getOption( 'language' );
 244                 }
 245
 246                 return $this->mUserVariant = $this->validateVariant( $ret );
 247         }
 248
 249         /**
 250          * Determine the language variant from the Accept-Language header.
 251          *
 252          * @return Mixed: variant if one found, false otherwise.
 253          */
 254         protected function getHeaderVariant() {
 255                 global $wgRequest;
 256
 257                 if ( $this->mHeaderVariant ) {
 258                         return $this->mHeaderVariant;
 259                 }
 260
 261                 // see if some supported language variant is set in the
 262                 // http header.
 263                 $languages = array_keys( $wgRequest->getAcceptLang() );
 264                 if ( empty( $languages ) ) {
 265                         return null;
 266                 }
 267
 268                 $fallback_languages = array();
 269                 foreach ( $languages as $language ) {
 270                         $this->mHeaderVariant = $this->validateVariant( $language );
 271                         if ( $this->mHeaderVariant ) {
 272                                 break;
 273                         }
 274
 275                         // To see if there are fallbacks of current language.
 276                         // We record these fallback variants, and process
 277                         // them later.
 278                         $fallbacks = $this->getVariantFallbacks( $language );
 279                         if ( is_string( $fallbacks ) ) {
 280                                 $fallback_languages[] = $fallbacks;
 281                         } elseif ( is_array( $fallbacks ) ) {
 282                                 $fallback_languages =
 283                                         array_merge( $fallback_languages,
 284                                                                  $fallbacks );
 285                         }
 286                 }
 287
 288                 if ( !$this->mHeaderVariant ) {
 289                         // process fallback languages now
 290                         $fallback_languages = array_unique( $fallback_languages );
 291                         foreach ( $fallback_languages as $language ) {
 292                                 $this->mHeaderVariant = $this->validateVariant( $language );
 293                                 if ( $this->mHeaderVariant ) {
 294                                         break;
 295                                 }
 296                         }
 297                 }
 298
 299                 return $this->mHeaderVariant;
 300         }
 301
 302         /**
 303          * Caption convert, base on preg_replace_callback.
 304          *
 305          * To convert text in "title" or "alt", like '<img alt="text" ... '
 306          * or '<span title="text" ... '
 307          *
 308          * @return String like ' alt="yyyy"' or ' title="yyyy"'
 309          */
 310         protected function captionConvert( $matches ) {
 311                 // TODO: cache the preferred variant in every autoConvert() process,
 312                 // this helps improve performance in a way.
 313                 $toVariant = $this->getPreferredVariant();
 314                 $title = $matches[1];
 315                 $text = $matches[2];
 316
 317                 // we convert captions except URL
 318                 if ( !strpos( $text, '://' ) ) {
 319                         $text = $this->translate( $text, $toVariant );
 320                 }
 321
 322                 // remove HTML tags to prevent disrupting the layout
 323                 $text = preg_replace( '/<[^>]+>/', '', $text );
 324                 // escape HTML special chars to prevent disrupting the layout
 325                 $text = htmlspecialchars( $text );
 326
 327                 return " {$title}=\"{$text}\"";
 328         }
 329
 330         /**
 331          * Dictionary-based conversion.
 332          * This function would not parse the conversion rules.
 333          * If you want to parse rules, try to use convert() or
 334          * convertTo().
 335          *
 336          * @param $text String: the text to be converted
 337          * @param $toVariant String: the target language code
 338          * @return String: the converted text
 339          */
 340         public function autoConvert( $text, $toVariant = false ) {
 341                 wfProfileIn( __METHOD__ );
 342
 343                 $this->loadTables();
 344
 345                 if ( !$toVariant ) {
 346                         $toVariant = $this->getPreferredVariant();
 347                         if ( !$toVariant ) {
 348                                 wfProfileOut( __METHOD__ );
 349                                 return $text;
 350                         }
 351                 }
 352
 353                 /* we convert everything except:
 354                    1. html markups (anything between < and >)
 355                    2. html entities
 356                    3. place holders created by the parser
 357                 */
 358                 global $wgParser;
 359                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 360                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 361                 } else {
 362                         $marker = '';
 363                 }
 364
 365                 // this one is needed when the text is inside an html markup
 366                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 367
 368                 // disable convert to variants between <code></code> tags
 369                 $codefix = '<code>.+?<\/code>|';
 370                 // disable convertsion of <script type="text/javascript"> ... </script>
 371                 $scriptfix = '<script.*?>.*?<\/script>|';
 372                 // disable conversion of <pre xxxx> ... </pre>
 373                 $prefix = '<pre.*?>.*?<\/pre>|';
 374
 375                 $reg = '/' . $codefix . $scriptfix . $prefix .
 376                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 377
 378                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 379
 380                 $m = array_shift( $matches );
 381
 382                 $ret = $this->translate( $m[0], $toVariant );
 383                 $mstart = $m[1] + strlen( $m[0] );
 384
 385                 // enable convertsion of '<img alt="xxxx" ... '
 386                 // or '<span title="xxxx" ... '
 387                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 388
 389                 $trtext = '';
 390                 $trtextmark = "\0";
 391                 $notrtext = array();
 392                 foreach ( $matches as $m ) {
 393                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 394                         $mark = preg_replace_callback( $captionpattern,
 395                                                                                    array( &$this, 'captionConvert' ),
 396                                                                                    $mark );
 397                         // Let's convert the trtext only once,
 398                         // it would give us more performance improvement
 399                         $notrtext[] = $mark;
 400                         $trtext .= $m[0] . $trtextmark;
 401                         $mstart = $m[1] + strlen( $m[0] );
 402                 }
 403                 $notrtext[] = '';
 404                 $trtext = $this->translate( $trtext, $toVariant );
 405                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 406                 foreach ( $trtext as $t ) {
 407                         $ret .= array_shift( $notrtext );
 408                         $ret .= $t;
 409                 }
 410                 wfProfileOut( __METHOD__ );
 411                 return $ret;
 412         }
 413
 414         /**
 415          * Translate a string to a variant.
 416          * Doesn't parse rules or do any of that other stuff, for that use
 417          * convert() or convertTo().
 418          *
 419          * @param $text String: text to convert
 420          * @param $variant String: variant language code
 421          * @return String: translated text
 422          */
 423         protected function translate( $text, $variant ) {
 424                 wfProfileIn( __METHOD__ );
 425                 // If $text is empty or only includes spaces, do nothing
 426                 // Otherwise translate it
 427                 if ( trim( $text ) ) {
 428                         $this->loadTables();
 429                         $text = $this->mTables[$variant]->replace( $text );
 430                 }
 431                 wfProfileOut( __METHOD__ );
 432                 return $text;
 433         }
 434
 435         /**
 436          * Call translate() to convert text to all valid variants.
 437          *
 438          * @param $text String: the text to be converted
 439          * @return Array: variant => converted text
 440          */
 441         public function autoConvertToAllVariants( $text ) {
 442                 wfProfileIn( __METHOD__ );
 443                 $this->loadTables();
 444
 445                 $ret = array();
 446                 foreach ( $this->mVariants as $variant ) {
 447                         $ret[$variant] = $this->translate( $text, $variant );
 448                 }
 449
 450                 wfProfileOut( __METHOD__ );
 451                 return $ret;
 452         }
 453
 454         /**
 455          * Convert link text to all valid variants.
 456          * In the first, this function only convert text outside the
 457          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 458          * titles, the text will get all converted always.
 459          * So I removed this feature and deprecated the function.
 460          *
 461          * @param $text String: the text to be converted
 462          * @return Array: variant => converted text
 463          * @deprecated Use autoConvertToAllVariants() instead
 464          */
 465         public function convertLinkToAllVariants( $text ) {
 466                 return $this->autoConvertToAllVariants( $text );
 467         }
 468
 469         /**
 470          * Apply manual conversion rules.
 471          *
 472          * @param $convRule Object: Object of ConverterRule
 473          */
 474         protected function applyManualConv( $convRule ) {
 475                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 476                 // title conversion.
 477                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 478                 // rule(s) not for title, this breaks the title conversion.
 479                 $newConvRuleTitle = $convRule->getTitle();
 480                 if ( $newConvRuleTitle ) {
 481                         // So I add an empty check for getTitle()
 482                         $this->mConvRuleTitle = $newConvRuleTitle;
 483                 }
 484
 485                 // merge/remove manual conversion rules to/from global table
 486                 $convTable = $convRule->getConvTable();
 487                 $action = $convRule->getRulesAction();
 488                 foreach ( $convTable as $variant => $pair ) {
 489                         if ( !$this->validateVariant( $variant ) ) {
 490                                 continue;
 491                         }
 492
 493                         if ( $action == 'add' ) {
 494                                 foreach ( $pair as $from => $to ) {
 495                                         // to ensure that $from and $to not be left blank
 496                                         // so $this->translate() could always return a string
 497                                         if ( $from || $to ) {
 498                                                 // more efficient than array_merge(), about 2.5 times.
 499                                                 $this->mTables[$variant]->setPair( $from, $to );
 500                                         }
 501                                 }
 502                         } elseif ( $action == 'remove' ) {
 503                                 $this->mTables[$variant]->removeArray( $pair );
 504                         }
 505                 }
 506         }
 507
 508         /**
 509          * Auto convert a Title object to a readable string in the
 510          * preferred variant.
 511          *
 512          *@param $title Object: a object of Title
 513          *@return String: converted title text
 514          */
 515         public function convertTitle( $title ) {
 516                 $variant = $this->getPreferredVariant();
 517                 $index = $title->getNamespace();
 518                 if ( $index === NS_MAIN ) {
 519                         $text = '';
 520                 } else {
 521                         // first let's check if a message has given us a converted name
 522                         $nsConvKey = 'conversion-ns' . $index;
 523                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 524                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 525                         } else {
 526                                 // the message does not exist, try retrieve it from the current
 527                                 // variant's namespace names.
 528                                 $langObj = $this->mLangObj->factory( $variant );
 529                                 $text = $langObj->getFormattedNsText( $index );
 530                         }
 531                         $text .= ':';
 532                 }
 533                 $text .= $title->getText();
 534                 $text = $this->translate( $text, $variant );
 535                 return $text;
 536         }
 537
 538         /**
 539          * Convert text to different variants of a language. The automatic
 540          * conversion is done in autoConvert(). Here we parse the text
 541          * marked with -{}-, which specifies special conversions of the
 542          * text that can not be accomplished in autoConvert().
 543          *
 544          * Syntax of the markup:
 545          * -{code1:text1;code2:text2;...}-  or
 546          * -{flags|code1:text1;code2:text2;...}-  or
 547          * -{text}- in which case no conversion should take place for text
 548          *
 549          * @param $text String: text to be converted
 550          * @return String: converted text
 551          */
 552         public function convert( $text ) {
 553                 $variant = $this->getPreferredVariant();
 554                 return $this->convertTo( $text, $variant );
 555         }
 556
 557         /**
 558          * Same as convert() except a extra parameter to custom variant.
 559          *
 560          * @param $text String: text to be converted
 561          * @param $variant String: the target variant code
 562          * @return String: converted text
 563          */
 564         public function convertTo( $text, $variant ) {
 565                 global $wgDisableLangConversion;
 566                 if ( $wgDisableLangConversion ) return $text;
 567                 return $this->recursiveConvertTopLevel( $text, $variant );
 568         }
 569
 570         /**
 571          * Recursively convert text on the outside. Allow to use nested
 572          * markups to custom rules.
 573          *
 574          * @param $text String: text to be converted
 575          * @param $variant String: the target variant code
 576          * @param $depth Integer: depth of recursion
 577          * @return String: converted text
 578          */
 579         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 580                 $startPos = 0;
 581                 $out = '';
 582                 $length = strlen( $text );
 583                 while ( $startPos < $length ) {
 584                         $pos = strpos( $text, '-{', $startPos );
 585
 586                         if ( $pos === false ) {
 587                                 // No more markup, append final segment
 588                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 589                                 return $out;
 590                         }
 591
 592                         // Markup found
 593                         // Append initial segment
 594                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 595
 596                         // Advance position
 597                         $startPos = $pos;
 598
 599                         // Do recursive conversion
 600                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 601                 }
 602
 603                 return $out;
 604         }
 605
 606         /**
 607          * Recursively convert text on the inside.
 608          *
 609          * @param $text String: text to be converted
 610          * @param $variant String: the target variant code
 611          * @param $depth Integer: depth of recursion
 612          * @return String: converted text
 613          */
 614         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 615                 // Quick sanity check (no function calls)
 616                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 617                         throw new MWException( __METHOD__ . ': invalid input string' );
 618                 }
 619
 620                 $startPos += 2;
 621                 $inner = '';
 622                 $warningDone = false;
 623                 $length = strlen( $text );
 624
 625                 while ( $startPos < $length ) {
 626                         $m = false;
 627                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 628                         if ( !$m ) {
 629                                 // Unclosed rule
 630                                 break;
 631                         }
 632
 633                         $token = $m[0][0];
 634                         $pos = $m[0][1];
 635
 636                         // Markup found
 637                         // Append initial segment
 638                         $inner .= substr( $text, $startPos, $pos - $startPos );
 639
 640                         // Advance position
 641                         $startPos = $pos;
 642
 643                         switch ( $token ) {
 644                                 case '-{':
 645                                         // Check max depth
 646                                         if ( $depth >= $this->mMaxDepth ) {
 647                                                 $inner .= '-{';
 648                                                 if ( !$warningDone ) {
 649                                                         $inner .= '<span class="error">' .
 650                                                                 wfMsgForContent( 'language-converter-depth-warning',
 651                                                                         $this->mMaxDepth ) .
 652                                                                 '</span>';
 653                                                         $warningDone = true;
 654                                                 }
 655                                                 $startPos += 2;
 656                                                 continue;
 657                                         }
 658                                         // Recursively parse another rule
 659                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 660                                         break;
 661                                 case '}-':
 662                                         // Apply the rule
 663                                         $startPos += 2;
 664                                         $rule = new ConverterRule( $inner, $this );
 665                                         $rule->parse( $variant );
 666                                         $this->applyManualConv( $rule );
 667                                         return $rule->getDisplay();
 668                                 default:
 669                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 670                         }
 671                 }
 672
 673                 // Unclosed rule
 674                 if ( $startPos < $length ) {
 675                         $inner .= substr( $text, $startPos );
 676                 }
 677                 $startPos = $length;
 678                 return '-{' . $this->autoConvert( $inner, $variant );
 679         }
 680
 681         /**
 682          * If a language supports multiple variants, it is
 683          * possible that non-existing link in one variant
 684          * actually exists in another variant. This function
 685          * tries to find it. See e.g. LanguageZh.php
 686          *
 687          * @param $link String: the name of the link
 688          * @param $nt Mixed: the title object of the link
 689          * @param $ignoreOtherCond Boolean: to disable other conditions when
 690          *              we need to transclude a template or update a category's link
 691          * @return Null, the input parameters may be modified upon return
 692          */
 693         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 694                 # If the article has already existed, there is no need to
 695                 # check it again, otherwise it may cause a fault.
 696                 if ( is_object( $nt ) && $nt->exists() ) {
 697                         return;
 698                 }
 699
 700                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 701                         $wgUser;
 702                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 703                 $action = $wgRequest->getText( 'action' );
 704                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 705                 $disableLinkConversion = $wgDisableLangConversion
 706                         || $wgDisableTitleConversion;
 707                 $linkBatch = new LinkBatch();
 708
 709                 $ns = NS_MAIN;
 710
 711                 if ( $disableLinkConversion ||
 712                          ( !$ignoreOtherCond &&
 713                            ( $isredir == 'no'
 714                                  || $action == 'edit'
 715                                  || $action == 'submit'
 716                                  || $linkconvert == 'no'
 717                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 718                         return;
 719                 }
 720
 721                 if ( is_object( $nt ) ) {
 722                         $ns = $nt->getNamespace();
 723                 }
 724
 725                 $variants = $this->autoConvertToAllVariants( $link );
 726                 if ( !$variants ) { // give up
 727                         return;
 728                 }
 729
 730                 $titles = array();
 731
 732                 foreach ( $variants as $v ) {
 733                         if ( $v != $link ) {
 734                                 $varnt = Title::newFromText( $v, $ns );
 735                                 if ( !is_null( $varnt ) ) {
 736                                         $linkBatch->addObj( $varnt );
 737                                         $titles[] = $varnt;
 738                                 }
 739                         }
 740                 }
 741
 742                 // fetch all variants in single query
 743                 $linkBatch->execute();
 744
 745                 foreach ( $titles as $varnt ) {
 746                         if ( $varnt->getArticleID() > 0 ) {
 747                                 $nt = $varnt;
 748                                 $link = $varnt->getText();
 749                                 break;
 750                         }
 751                 }
 752         }
 753
 754         /**
 755          * Returns language specific hash options.
 756          */
 757         public function getExtraHashOptions() {
 758                 $variant = $this->getPreferredVariant();
 759                 return '!' . $variant ;
 760         }
 761
 762         /**
 763          * Load default conversion tables.
 764          * This method must be implemented in derived class.
 765          *
 766          * @private
 767          */
 768         function loadDefaultTables() {
 769                 $name = get_class( $this );
 770                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 771         }
 772
 773         /**
 774          * Load conversion tables either from the cache or the disk.
 775          * @private
 776          */
 777         function loadTables( $fromcache = true ) {
 778                 if ( $this->mTablesLoaded ) {
 779                         return;
 780                 }
 781                 global $wgMemc;
 782                 wfProfileIn( __METHOD__ );
 783                 $this->mTablesLoaded = true;
 784                 $this->mTables = false;
 785                 if ( $fromcache ) {
 786                         wfProfileIn( __METHOD__ . '-cache' );
 787                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 788                         wfProfileOut( __METHOD__ . '-cache' );
 789                 }
 790                 if ( !$this->mTables
 791                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 792                         wfProfileIn( __METHOD__ . '-recache' );
 793                         // not in cache, or we need a fresh reload.
 794                         // we will first load the default tables
 795                         // then update them using things in MediaWiki:Zhconversiontable/*
 796                         $this->loadDefaultTables();
 797                         foreach ( $this->mVariants as $var ) {
 798                                 $cached = $this->parseCachedTable( $var );
 799                                 $this->mTables[$var]->mergeArray( $cached );
 800                         }
 801
 802                         $this->postLoadTables();
 803                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 804
 805                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 806                         wfProfileOut( __METHOD__ . '-recache' );
 807                 }
 808                 wfProfileOut( __METHOD__ );
 809         }
 810
 811         /**
 812          * Hook for post processig after conversion tables are loaded.
 813          *
 814          */
 815         function postLoadTables() { }
 816
 817         /**
 818          * Reload the conversion tables.
 819          *
 820          * @private
 821          */
 822         function reloadTables() {
 823                 if ( $this->mTables ) {
 824                         unset( $this->mTables );
 825                 }
 826                 $this->mTablesLoaded = false;
 827                 $this->loadTables( false );
 828         }
 829
 830
 831         /**
 832          * Parse the conversion table stored in the cache.
 833          *
 834          * The tables should be in blocks of the following form:
 835          *              -{
 836          *                      word => word ;
 837          *                      word => word ;
 838          *                      ...
 839          *              }-
 840          *
 841          *      To make the tables more manageable, subpages are allowed
 842          *      and will be parsed recursively if $recursive == true.
 843          *
 844          */
 845         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 846                 static $parsed = array();
 847
 848                 $key = 'Conversiontable/' . $code;
 849                 if ( $subpage ) {
 850                         $key .= '/' . $subpage;
 851                 }
 852                 if ( array_key_exists( $key, $parsed ) ) {
 853                         return array();
 854                 }
 855
 856                 if ( strpos( $code, '/' ) === false ) {
 857                         $txt = MessageCache::singleton()->get( 'Conversiontable', true, $code );
 858                         if ( $txt === false ) {
 859                                 # FIXME: this method doesn't seem to be expecting
 860                                 # this possible outcome...
 861                                 $txt = '&lt;Conversiontable&gt;';
 862                         }
 863                 } else {
 864                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 865                                                                                    "Conversiontable/$code" );
 866                         if ( $title && $title->exists() ) {
 867                                 $article = new Article( $title );
 868                                 $txt = $article->getContents();
 869                         } else {
 870                                 $txt = '';
 871                         }
 872                 }
 873
 874                 // get all subpage links of the form
 875                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 876                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 877                         ':Conversiontable';
 878                 $subs = StringUtils::explode( '[[', $txt );
 879                 $sublinks = array();
 880                 foreach ( $subs as $sub ) {
 881                         $link = explode( ']]', $sub, 2 );
 882                         if ( count( $link ) != 2 ) {
 883                                 continue;
 884                         }
 885                         $b = explode( '|', $link[0], 2 );
 886                         $b = explode( '/', trim( $b[0] ), 3 );
 887                         if ( count( $b ) == 3 ) {
 888                                 $sublink = $b[2];
 889                         } else {
 890                                 $sublink = '';
 891                         }
 892
 893                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 894                                 $sublinks[] = $sublink;
 895                         }
 896                 }
 897
 898                 // parse the mappings in this page
 899                 $blocks = StringUtils::explode( '-{', $txt );
 900                 $ret = array();
 901                 $first = true;
 902                 foreach ( $blocks as $block ) {
 903                         if ( $first ) {
 904                                 // Skip the part before the first -{
 905                                 $first = false;
 906                                 continue;
 907                         }
 908                         $mappings = explode( '}-', $block, 2 );
 909                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 910                                                                          $mappings[0] );
 911                         $table = StringUtils::explode( ';', $stripped );
 912                         foreach ( $table as $t ) {
 913                                 $m = explode( '=>', $t, 3 );
 914                                 if ( count( $m ) != 2 )
 915                                         continue;
 916                                 // trim any trailling comments starting with '//'
 917                                 $tt = explode( '//', $m[1], 2 );
 918                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 919                         }
 920                 }
 921                 $parsed[$key] = true;
 922
 923                 // recursively parse the subpages
 924                 if ( $recursive ) {
 925                         foreach ( $sublinks as $link ) {
 926                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 927                                 $ret = array_merge( $ret, $s );
 928                         }
 929                 }
 930
 931                 if ( $this->mUcfirst ) {
 932                         foreach ( $ret as $k => $v ) {
 933                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 934                         }
 935                 }
 936                 return $ret;
 937         }
 938
 939         /**
 940          * Enclose a string with the "no conversion" tag. This is used by
 941          * various functions in the Parser.
 942          *
 943          * @param $text String: text to be tagged for no conversion
 944          * @param $noParse Unused (?)
 945          * @return String: the tagged text
 946          */
 947         public function markNoConversion( $text, $noParse = false ) {
 948                 # don't mark if already marked
 949                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 950                         return $text;
 951                 }
 952
 953                 $ret = "-{R|$text}-";
 954                 return $ret;
 955         }
 956
 957         /**
 958          * Convert the sorting key for category links. This should make different
 959          * keys that are variants of each other map to the same key.
 960          */
 961         function convertCategoryKey( $key ) {
 962                 return $key;
 963         }
 964
 965         /**
 966          * Hook to refresh the cache of conversion tables when
 967          * MediaWiki:conversiontable* is updated.
 968          * @private
 969          */
 970         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 971                         $iswatch, $section, $flags, $revision ) {
 972                 $titleobj = $article->getTitle();
 973                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 974                         $title = $titleobj->getDBkey();
 975                         $t = explode( '/', $title, 3 );
 976                         $c = count( $t );
 977                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 978                                 if ( $this->validateVariant( $t[1] ) ) {
 979                                         $this->reloadTables();
 980                                 }
 981                         }
 982                 }
 983                 return true;
 984         }
 985
 986         /**
 987          * Armour rendered math against conversion.
 988          * Escape special chars in parsed math text.(in most cases are img elements)
 989          */
 990         public function armourMath( $text ) {
 991                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
 992                 // any unwanted markup appearing in the math image tag.
 993                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 994                 return $text;
 995         }
 996
 997         /**
 998          * Get the cached separator pattern for ConverterRule::parseRules()
 999          */
1000         function getVarSeparatorPattern() {
1001                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1002                         // varsep_pattern for preg_split:
1003                         // text should be splited by ";" only if a valid variant
1004                         // name exist after the markup, for example:
1005                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1006                         //      <span style="font-size:120%;">yyy</span>;}-
1007                         // we should split it as:
1008                         //  array(
1009                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1010                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1011                         //        [2] => ''
1012                         //       )
1013                         $pat = '/;\s*(?=';
1014                         foreach ( $this->mVariants as $variant ) {
1015                                 // zh-hans:xxx;zh-hant:yyy
1016                                 $pat .= $variant . '\s*:|';
1017                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1018                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1019                         }
1020                         $pat .= '\s*$)/';
1021                         $this->mVarSeparatorPattern = $pat;
1022                 }
1023                 return $this->mVarSeparatorPattern;
1024         }
1025 }
1026
1027 /**
1028  * Parser for rules of language conversion , parse rules in -{ }- tag.
1029  * @ingroup Language
1030  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1031  */
1032 class ConverterRule {
1033         var $mText; // original text in -{text}-
1034         var $mConverter; // LanguageConverter object
1035         var $mManualCodeError = '<strong class="error">code error!</strong>';
1036         var $mRuleDisplay = '';
1037         var $mRuleTitle = false;
1038         var $mRules = '';// string : the text of the rules
1039         var $mRulesAction = 'none';
1040         var $mFlags = array();
1041         var $mVariantFlags = array();
1042         var $mConvTable = array();
1043         var $mBidtable = array();// array of the translation in each variant
1044         var $mUnidtable = array();// array of the translation in each variant
1045
1046         /**
1047          * Constructor
1048          *
1049          * @param $text String: the text between -{ and }-
1050          * @param $converter LanguageConverter object
1051          */
1052         public function __construct( $text, $converter ) {
1053                 $this->mText = $text;
1054                 $this->mConverter = $converter;
1055         }
1056
1057         /**
1058          * Check if variants array in convert array.
1059          *
1060          * @param $variants Array or string: variant language code
1061          * @return String: translated text
1062          */
1063         public function getTextInBidtable( $variants ) {
1064                 $variants = (array)$variants;
1065                 if ( !$variants ) {
1066                         return false;
1067                 }
1068                 foreach ( $variants as $variant ) {
1069                         if ( isset( $this->mBidtable[$variant] ) ) {
1070                                 return $this->mBidtable[$variant];
1071                         }
1072                 }
1073                 return false;
1074         }
1075
1076         /**
1077          * Parse flags with syntax -{FLAG| ... }-
1078          * @private
1079          */
1080         function parseFlags() {
1081                 $text = $this->mText;
1082                 $flags = array();
1083                 $variantFlags = array();
1084
1085                 $sepPos = strpos( $text, '|' );
1086                 if ( $sepPos !== false ) {
1087                         $validFlags = $this->mConverter->mFlags;
1088                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1089                         foreach ( $f as $ff ) {
1090                                 $ff = trim( $ff );
1091                                 if ( isset( $validFlags[$ff] ) ) {
1092                                         $flags[$validFlags[$ff]] = true;
1093                                 }
1094                         }
1095                         $text = strval( substr( $text, $sepPos + 1 ) );
1096                 }
1097
1098                 if ( !$flags ) {
1099                         $flags['S'] = true;
1100                 } elseif ( isset( $flags['R'] ) ) {
1101                         $flags = array( 'R' => true );// remove other flags
1102                 } elseif ( isset( $flags['N'] ) ) {
1103                         $flags = array( 'N' => true );// remove other flags
1104                 } elseif ( isset( $flags['-'] ) ) {
1105                         $flags = array( '-' => true );// remove other flags
1106                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1107                         $flags['H'] = true;
1108                 } elseif ( isset( $flags['H'] ) ) {
1109                         // replace A flag, and remove other flags except T
1110                         $temp = array( '+' => true, 'H' => true );
1111                         if ( isset( $flags['T'] ) ) {
1112                                 $temp['T'] = true;
1113                         }
1114                         if ( isset( $flags['D'] ) ) {
1115                                 $temp['D'] = true;
1116                         }
1117                         $flags = $temp;
1118                 } else {
1119                         if ( isset( $flags['A'] ) ) {
1120                                 $flags['+'] = true;
1121                                 $flags['S'] = true;
1122                         }
1123                         if ( isset( $flags['D'] ) ) {
1124                                 unset( $flags['S'] );
1125                         }
1126                         // try to find flags like "zh-hans", "zh-hant"
1127                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1128                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1129                         if ( $variantFlags ) {
1130                                 $variantFlags = array_flip( $variantFlags );
1131                                 $flags = array();
1132                         }
1133                 }
1134                 $this->mVariantFlags = $variantFlags;
1135                 $this->mRules = $text;
1136                 $this->mFlags = $flags;
1137         }
1138
1139         /**
1140          * Generate conversion table.
1141          * @private
1142          */
1143         function parseRules() {
1144                 $rules = $this->mRules;
1145                 $bidtable = array();
1146                 $unidtable = array();
1147                 $variants = $this->mConverter->mVariants;
1148                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1149
1150                 $choice = preg_split( $varsep_pattern, $rules );
1151
1152                 foreach ( $choice as $c ) {
1153                         $v  = explode( ':', $c, 2 );
1154                         if ( count( $v ) != 2 ) {
1155                                 // syntax error, skip
1156                                 continue;
1157                         }
1158                         $to = trim( $v[1] );
1159                         $v  = trim( $v[0] );
1160                         $u  = explode( '=>', $v, 2 );
1161                         // if $to is empty, strtr() could return a wrong result
1162                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1163                                 $bidtable[$v] = $to;
1164                         } elseif ( count( $u ) == 2 ) {
1165                                 $from = trim( $u[0] );
1166                                 $v      = trim( $u[1] );
1167                                 if ( array_key_exists( $v, $unidtable )
1168                                          && !is_array( $unidtable[$v] )
1169                                          && $to
1170                                          && in_array( $v, $variants ) ) {
1171                                         $unidtable[$v] = array( $from => $to );
1172                                 } elseif ( $to && in_array( $v, $variants ) ) {
1173                                         $unidtable[$v][$from] = $to;
1174                                 }
1175                         }
1176                         // syntax error, pass
1177                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1178                                 $bidtable = array();
1179                                 $unidtable = array();
1180                                 break;
1181                         }
1182                 }
1183                 $this->mBidtable = $bidtable;
1184                 $this->mUnidtable = $unidtable;
1185         }
1186
1187         /**
1188          * @private
1189          */
1190         function getRulesDesc() {
1191                 $codesep = $this->mConverter->mDescCodeSep;
1192                 $varsep = $this->mConverter->mDescVarSep;
1193                 $text = '';
1194                 foreach ( $this->mBidtable as $k => $v ) {
1195                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1196                 }
1197                 foreach ( $this->mUnidtable as $k => $a ) {
1198                         foreach ( $a as $from => $to ) {
1199                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1200                                         "$codesep$to$varsep";
1201                         }
1202                 }
1203                 return $text;
1204         }
1205
1206         /**
1207          * Parse rules conversion.
1208          * @private
1209          */
1210         function getRuleConvertedStr( $variant ) {
1211                 $bidtable = $this->mBidtable;
1212                 $unidtable = $this->mUnidtable;
1213
1214                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1215                         return $this->mRules;
1216                 } else {
1217                         // display current variant in bidirectional array
1218                         $disp = $this->getTextInBidtable( $variant );
1219                         // or display current variant in fallbacks
1220                         if ( !$disp ) {
1221                                 $disp = $this->getTextInBidtable(
1222                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1223                         }
1224                         // or display current variant in unidirectional array
1225                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1226                                 $disp = array_values( $unidtable[$variant] );
1227                                 $disp = $disp[0];
1228                         }
1229                         // or display frist text under disable manual convert
1230                         if ( !$disp
1231                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1232                                 if ( count( $bidtable ) > 0 ) {
1233                                         $disp = array_values( $bidtable );
1234                                         $disp = $disp[0];
1235                                 } else {
1236                                         $disp = array_values( $unidtable );
1237                                         $disp = array_values( $disp[0] );
1238                                         $disp = $disp[0];
1239                                 }
1240                         }
1241                         return $disp;
1242                 }
1243         }
1244
1245         /**
1246          * Generate conversion table for all text.
1247          * @private
1248          */
1249         function generateConvTable() {
1250                 // Special case optimisation
1251                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1252                         $this->mConvTable = array();
1253                         return;
1254                 }
1255
1256                 $bidtable = $this->mBidtable;
1257                 $unidtable = $this->mUnidtable;
1258                 $manLevel = $this->mConverter->mManualLevel;
1259
1260                 $vmarked = array();
1261                 foreach ( $this->mConverter->mVariants as $v ) {
1262                         /* for bidirectional array
1263                                 fill in the missing variants, if any,
1264                                 with fallbacks */
1265                         if ( !isset( $bidtable[$v] ) ) {
1266                                 $variantFallbacks =
1267                                         $this->mConverter->getVariantFallbacks( $v );
1268                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1269                                 if ( $vf ) {
1270                                         $bidtable[$v] = $vf;
1271                                 }
1272                         }
1273
1274                         if ( isset( $bidtable[$v] ) ) {
1275                                 foreach ( $vmarked as $vo ) {
1276                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1277                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1278                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1279                                         // to introduce a custom mapping between
1280                                         // words WordZh and WordTw in the whole text
1281                                         if ( $manLevel[$v] == 'bidirectional' ) {
1282                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1283                                         }
1284                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1285                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1286                                         }
1287                                 }
1288                                 $vmarked[] = $v;
1289                         }
1290                         /*for unidirectional array fill to convert tables */
1291                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1292                                 && isset( $unidtable[$v] ) )
1293                         {
1294                                 if ( isset( $this->mConvTable[$v] ) ) {
1295                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1296                                 } else {
1297                                         $this->mConvTable[$v] = $unidtable[$v];
1298                                 }
1299                         }
1300                 }
1301         }
1302
1303         /**
1304          * Parse rules and flags.
1305          * @public
1306          */
1307         function parse( $variant = NULL ) {
1308                 if ( !$variant ) {
1309                         $variant = $this->mConverter->getPreferredVariant();
1310                 }
1311
1312                 $this->parseFlags();
1313                 $flags = $this->mFlags;
1314
1315                 // convert to specified variant
1316                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1317                 if ( $this->mVariantFlags ) {
1318                         // check if current variant in flags
1319                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1320                                 // then convert <text to convert> to current language
1321                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1322                                                                                                                                 $variant );
1323                         } else { // if current variant no in flags,
1324                                    // then we check its fallback variants.
1325                                 $variantFallbacks =
1326                                         $this->mConverter->getVariantFallbacks( $variant );
1327                                 foreach ( $variantFallbacks as $variantFallback ) {
1328                                         // if current variant's fallback exist in flags
1329                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1330                                                 // then convert <text to convert> to fallback language
1331                                                 $this->mRules =
1332                                                         $this->mConverter->autoConvert( $this->mRules,
1333                                                                                                                         $variantFallback );
1334                                                 break;
1335                                         }
1336                                 }
1337                         }
1338                         $this->mFlags = $flags = array( 'R' => true );
1339                 }
1340
1341                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1342                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1343                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1344                         $this->parseRules();
1345                 }
1346                 $rules = $this->mRules;
1347
1348                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1349                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1350                                 // fill all variants if text in -{A/H/-|text} without rules
1351                                 foreach ( $this->mConverter->mVariants as $v ) {
1352                                         $this->mBidtable[$v] = $rules;
1353                                 }
1354                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1355                                 $this->mFlags = $flags = array( 'R' => true );
1356                         }
1357                 }
1358
1359                 $this->mRuleDisplay = false;
1360                 foreach ( $flags as $flag => $unused ) {
1361                         switch ( $flag ) {
1362                                 case 'R':
1363                                         // if we don't do content convert, still strip the -{}- tags
1364                                         $this->mRuleDisplay = $rules;
1365                                         break;
1366                                 case 'N':
1367                                         // process N flag: output current variant name
1368                                         $ruleVar = trim( $rules );
1369                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1370                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1371                                         } else {
1372                                                 $this->mRuleDisplay = '';
1373                                         }
1374                                         break;
1375                                 case 'D':
1376                                         // process D flag: output rules description
1377                                         $this->mRuleDisplay = $this->getRulesDesc();
1378                                         break;
1379                                 case 'H':
1380                                         // process H,- flag or T only: output nothing
1381                                         $this->mRuleDisplay = '';
1382                                         break;
1383                                 case '-':
1384                                         $this->mRulesAction = 'remove';
1385                                         $this->mRuleDisplay = '';
1386                                         break;
1387                                 case '+':
1388                                         $this->mRulesAction = 'add';
1389                                         $this->mRuleDisplay = '';
1390                                         break;
1391                                 case 'S':
1392                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1393                                         break;
1394                                 case 'T':
1395                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1396                                         $this->mRuleDisplay = '';
1397                                         break;
1398                                 default:
1399                                         // ignore unknown flags (but see error case below)
1400                         }
1401                 }
1402                 if ( $this->mRuleDisplay === false ) {
1403                         $this->mRuleDisplay = $this->mManualCodeError;
1404                 }
1405
1406                 $this->generateConvTable();
1407         }
1408
1409         /**
1410          * @public
1411          */
1412         function hasRules() {
1413                 // TODO:
1414         }
1415
1416         /**
1417          * Get display text on markup -{...}-
1418          * @public
1419          */
1420         function getDisplay() {
1421                 return $this->mRuleDisplay;
1422         }
1423
1424         /**
1425          * Get converted title.
1426          * @public
1427          */
1428         function getTitle() {
1429                 return $this->mRuleTitle;
1430         }
1431
1432         /**
1433          * Return how deal with conversion rules.
1434          * @public
1435          */
1436         function getRulesAction() {
1437                 return $this->mRulesAction;
1438         }
1439
1440         /**
1441          * Get conversion table. ( bidirectional and unidirectional
1442          * conversion table )
1443          * @public
1444          */
1445         function getConvTable() {
1446                 return $this->mConvTable;
1447         }
1448
1449         /**
1450          * Get conversion rules string.
1451          * @public
1452          */
1453         function getRules() {
1454                 return $this->mRules;
1455         }
1456
1457         /**
1458          * Get conversion flags.
1459          * @public
1460          */
1461         function getFlags() {
1462                 return $this->mFlags;
1463         }
1464 }