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                 $toVariant = $this->getPreferredVariant();
 312                 $title = $matches[1];
 313                 $text  = $matches[2];
 314                 // we convert captions except URL
 315                 if ( !strpos( $text, '://' ) ) {
 316                         $text = $this->translate( $text, $toVariant );
 317                 }
 318                 return " $title=\"$text\"";
 319         }
 320
 321         /**
 322          * Dictionary-based conversion.
 323          * This function would not parse the conversion rules.
 324          * If you want to parse rules, try to use convert() or
 325          * convertTo().
 326          *
 327          * @param $text String: the text to be converted
 328          * @param $toVariant String: the target language code
 329          * @return String: the converted text
 330          */
 331         public function autoConvert( $text, $toVariant = false ) {
 332                 wfProfileIn( __METHOD__ );
 333
 334                 $this->loadTables();
 335
 336                 if ( !$toVariant ) {
 337                         $toVariant = $this->getPreferredVariant();
 338                         if ( !$toVariant ) {
 339                                 return $text;
 340                         }
 341                 }
 342
 343                 /* we convert everything except:
 344                    1. html markups (anything between < and >)
 345                    2. html entities
 346                    3. place holders created by the parser
 347                 */
 348                 global $wgParser;
 349                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 350                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 351                 } else {
 352                         $marker = '';
 353                 }
 354
 355                 // this one is needed when the text is inside an html markup
 356                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 357
 358                 // disable convert to variants between <code></code> tags
 359                 $codefix = '<code>.+?<\/code>|';
 360                 // disable convertsion of <script type="text/javascript"> ... </script>
 361                 $scriptfix = '<script.*?>.*?<\/script>|';
 362                 // disable conversion of <pre xxxx> ... </pre>
 363                 $prefix = '<pre.*?>.*?<\/pre>|';
 364
 365                 $reg = '/' . $codefix . $scriptfix . $prefix .
 366                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 367
 368                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 369
 370                 $m = array_shift( $matches );
 371
 372                 $ret = $this->translate( $m[0], $toVariant );
 373                 $mstart = $m[1] + strlen( $m[0] );
 374
 375                 // enable convertsion of '<img alt="xxxx" ... '
 376                 // or '<span title="xxxx" ... '
 377                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 378
 379                 $trtext = '';
 380                 $trtextmark = "\0";
 381                 $notrtext = array();
 382                 foreach ( $matches as $m ) {
 383                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 384                         $mark = preg_replace_callback( $captionpattern,
 385                                                                                    array( &$this, 'captionConvert' ),
 386                                                                                    $mark );
 387                         // Let's convert the trtext only once,
 388                         // it would give us more performance improvement
 389                         $notrtext[] = $mark;
 390                         $trtext .= $m[0] . $trtextmark;
 391                         $mstart = $m[1] + strlen( $m[0] );
 392                 }
 393                 $notrtext[] = '';
 394                 $trtext = $this->translate( $trtext, $toVariant );
 395                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 396                 foreach ( $trtext as $t ) {
 397                         $ret .= array_shift( $notrtext );
 398                         $ret .= $t;
 399                 }
 400                 wfProfileOut( __METHOD__ );
 401                 return $ret;
 402         }
 403
 404         /**
 405          * Translate a string to a variant.
 406          * Doesn't parse rules or do any of that other stuff, for that use
 407          * convert() or convertTo().
 408          *
 409          * @param $text String: text to convert
 410          * @param $variant String: variant language code
 411          * @return String: translated text
 412          */
 413         protected function translate( $text, $variant ) {
 414                 wfProfileIn( __METHOD__ );
 415                 // If $text is empty or only includes spaces, do nothing
 416                 // Otherwise translate it
 417                 if ( trim( $text ) ) {
 418                         $this->loadTables();
 419                         $text = $this->mTables[$variant]->replace( $text );
 420                 }
 421                 wfProfileOut( __METHOD__ );
 422                 return $text;
 423         }
 424
 425         /**
 426          * Call translate() to convert text to all valid variants.
 427          *
 428          * @param $text String: the text to be converted
 429          * @return Array: variant => converted text
 430          */
 431         public function autoConvertToAllVariants( $text ) {
 432                 wfProfileIn( __METHOD__ );
 433                 $this->loadTables();
 434
 435                 $ret = array();
 436                 foreach ( $this->mVariants as $variant ) {
 437                         $ret[$variant] = $this->translate( $text, $variant );
 438                 }
 439
 440                 wfProfileOut( __METHOD__ );
 441                 return $ret;
 442         }
 443
 444         /**
 445          * Convert link text to all valid variants.
 446          * In the first, this function only convert text outside the
 447          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 448          * titles, the text will get all converted always.
 449          * So I removed this feature and deprecated the function.
 450          *
 451          * @param $text String: the text to be converted
 452          * @return Array: variant => converted text
 453          * @deprecated Use autoConvertToAllVariants() instead
 454          */
 455         public function convertLinkToAllVariants( $text ) {
 456                 return $this->autoConvertToAllVariants( $text );
 457         }
 458
 459         /**
 460          * Apply manual conversion rules.
 461          *
 462          * @param $convRule Object: Object of ConverterRule
 463          */
 464         protected function applyManualConv( $convRule ) {
 465                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 466                 // title conversion.
 467                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 468                 // rule(s) not for title, this breaks the title conversion.
 469                 $newConvRuleTitle = $convRule->getTitle();
 470                 if ( $newConvRuleTitle ) {
 471                         // So I add an empty check for getTitle()
 472                         $this->mConvRuleTitle = $newConvRuleTitle;
 473                 }
 474
 475                 // merge/remove manual conversion rules to/from global table
 476                 $convTable = $convRule->getConvTable();
 477                 $action = $convRule->getRulesAction();
 478                 foreach ( $convTable as $variant => $pair ) {
 479                         if ( !$this->validateVariant( $variant ) ) {
 480                                 continue;
 481                         }
 482
 483                         if ( $action == 'add' ) {
 484                                 foreach ( $pair as $from => $to ) {
 485                                         // to ensure that $from and $to not be left blank
 486                                         // so $this->translate() could always return a string
 487                                         if ( $from || $to ) {
 488                                                 // more efficient than array_merge(), about 2.5 times.
 489                                                 $this->mTables[$variant]->setPair( $from, $to );
 490                                         }
 491                                 }
 492                         } elseif ( $action == 'remove' ) {
 493                                 $this->mTables[$variant]->removeArray( $pair );
 494                         }
 495                 }
 496         }
 497
 498         /**
 499          * Auto convert a Title object to a readable string in the
 500          * preferred variant.
 501          *
 502          *@param $title Object: a object of Title
 503          *@return String: converted title text
 504          */
 505         public function convertTitle( $title ) {
 506                 $variant = $this->getPreferredVariant();
 507                 $index = $title->getNamespace();
 508                 if ( $index === NS_MAIN ) {
 509                         $text = '';
 510                 } else {
 511                         // first let's check if a message has given us a converted name
 512                         $nsConvKey = 'conversion-ns' . $index;
 513                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 514                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 515                         } else {
 516                                 // the message does not exist, try retrieve it from the current
 517                                 // variant's namespace names.
 518                                 $langObj = $this->mLangObj->factory( $variant );
 519                                 $text = $langObj->getFormattedNsText( $index );
 520                         }
 521                         $text .= ':';
 522                 }
 523                 $text .= $title->getText();
 524                 $text = $this->translate( $text, $variant );
 525                 return $text;
 526         }
 527
 528         /**
 529          * Convert text to different variants of a language. The automatic
 530          * conversion is done in autoConvert(). Here we parse the text
 531          * marked with -{}-, which specifies special conversions of the
 532          * text that can not be accomplished in autoConvert().
 533          *
 534          * Syntax of the markup:
 535          * -{code1:text1;code2:text2;...}-  or
 536          * -{flags|code1:text1;code2:text2;...}-  or
 537          * -{text}- in which case no conversion should take place for text
 538          *
 539          * @param $text String: text to be converted
 540          * @return String: converted text
 541          */
 542         public function convert( $text ) {
 543                 $variant = $this->getPreferredVariant();
 544                 return $this->convertTo( $text, $variant );
 545         }
 546
 547         /**
 548          * Same as convert() except a extra parameter to custom variant.
 549          *
 550          * @param $text String: text to be converted
 551          * @param $variant String: the target variant code
 552          * @return String: converted text
 553          */
 554         public function convertTo( $text, $variant ) {
 555                 global $wgDisableLangConversion;
 556                 if ( $wgDisableLangConversion ) return $text;
 557                 return $this->recursiveConvertTopLevel( $text, $variant );
 558         }
 559
 560         /**
 561          * Recursively convert text on the outside. Allow to use nested
 562          * markups to custom rules.
 563          *
 564          * @param $text String: text to be converted
 565          * @param $variant String: the target variant code
 566          * @param $depth Integer: depth of recursion
 567          * @return String: converted text
 568          */
 569         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 570                 $startPos = 0;
 571                 $out = '';
 572                 $length = strlen( $text );
 573                 while ( $startPos < $length ) {
 574                         $pos = strpos( $text, '-{', $startPos );
 575
 576                         if ( $pos === false ) {
 577                                 // No more markup, append final segment
 578                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 579                                 return $out;
 580                         }
 581
 582                         // Markup found
 583                         // Append initial segment
 584                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 585
 586                         // Advance position
 587                         $startPos = $pos;
 588
 589                         // Do recursive conversion
 590                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 591                 }
 592
 593                 return $out;
 594         }
 595
 596         /**
 597          * Recursively convert text on the inside.
 598          *
 599          * @param $text String: text to be converted
 600          * @param $variant String: the target variant code
 601          * @param $depth Integer: depth of recursion
 602          * @return String: converted text
 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 $link String: the name of the link
 678          * @param $nt Mixed: the title object of the link
 679          * @param $ignoreOtherCond Boolean: to disable other conditions when
 680          *              we need to transclude a template or update a category's link
 681          * @return Null, the input parameters may be modified upon return
 682          */
 683         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 684                 # If the article has already existed, there is no need to
 685                 # check it again, otherwise it may cause a fault.
 686                 if ( is_object( $nt ) && $nt->exists() ) {
 687                         return;
 688                 }
 689
 690                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 691                         $wgUser;
 692                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 693                 $action = $wgRequest->getText( 'action' );
 694                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 695                 $disableLinkConversion = $wgDisableLangConversion
 696                         || $wgDisableTitleConversion;
 697                 $linkBatch = new LinkBatch();
 698
 699                 $ns = NS_MAIN;
 700
 701                 if ( $disableLinkConversion ||
 702                          ( !$ignoreOtherCond &&
 703                            ( $isredir == 'no'
 704                                  || $action == 'edit'
 705                                  || $action == 'submit'
 706                                  || $linkconvert == 'no'
 707                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 708                         return;
 709                 }
 710
 711                 if ( is_object( $nt ) ) {
 712                         $ns = $nt->getNamespace();
 713                 }
 714
 715                 $variants = $this->autoConvertToAllVariants( $link );
 716                 if ( !$variants ) { // give up
 717                         return;
 718                 }
 719
 720                 $titles = array();
 721
 722                 foreach ( $variants as $v ) {
 723                         if ( $v != $link ) {
 724                                 $varnt = Title::newFromText( $v, $ns );
 725                                 if ( !is_null( $varnt ) ) {
 726                                         $linkBatch->addObj( $varnt );
 727                                         $titles[] = $varnt;
 728                                 }
 729                         }
 730                 }
 731
 732                 // fetch all variants in single query
 733                 $linkBatch->execute();
 734
 735                 foreach ( $titles as $varnt ) {
 736                         if ( $varnt->getArticleID() > 0 ) {
 737                                 $nt = $varnt;
 738                                 $link = $varnt->getText();
 739                                 break;
 740                         }
 741                 }
 742         }
 743
 744         /**
 745          * Returns language specific hash options.
 746          */
 747         public function getExtraHashOptions() {
 748                 $variant = $this->getPreferredVariant();
 749                 return '!' . $variant ;
 750         }
 751
 752         /**
 753          * Load default conversion tables.
 754          * This method must be implemented in derived class.
 755          *
 756          * @private
 757          */
 758         function loadDefaultTables() {
 759                 $name = get_class( $this );
 760                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 761         }
 762
 763         /**
 764          * Load conversion tables either from the cache or the disk.
 765          * @private
 766          */
 767         function loadTables( $fromcache = true ) {
 768                 if ( $this->mTablesLoaded ) {
 769                         return;
 770                 }
 771                 global $wgMemc;
 772                 wfProfileIn( __METHOD__ );
 773                 $this->mTablesLoaded = true;
 774                 $this->mTables = false;
 775                 if ( $fromcache ) {
 776                         wfProfileIn( __METHOD__ . '-cache' );
 777                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 778                         wfProfileOut( __METHOD__ . '-cache' );
 779                 }
 780                 if ( !$this->mTables
 781                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 782                         wfProfileIn( __METHOD__ . '-recache' );
 783                         // not in cache, or we need a fresh reload.
 784                         // we will first load the default tables
 785                         // then update them using things in MediaWiki:Zhconversiontable/*
 786                         $this->loadDefaultTables();
 787                         foreach ( $this->mVariants as $var ) {
 788                                 $cached = $this->parseCachedTable( $var );
 789                                 $this->mTables[$var]->mergeArray( $cached );
 790                         }
 791
 792                         $this->postLoadTables();
 793                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 794
 795                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 796                         wfProfileOut( __METHOD__ . '-recache' );
 797                 }
 798                 wfProfileOut( __METHOD__ );
 799         }
 800
 801         /**
 802          * Hook for post processig after conversion tables are loaded.
 803          *
 804          */
 805         function postLoadTables() { }
 806
 807         /**
 808          * Reload the conversion tables.
 809          *
 810          * @private
 811          */
 812         function reloadTables() {
 813                 if ( $this->mTables ) {
 814                         unset( $this->mTables );
 815                 }
 816                 $this->mTablesLoaded = false;
 817                 $this->loadTables( false );
 818         }
 819
 820
 821         /**
 822          * Parse the conversion table stored in the cache.
 823          *
 824          * The tables should be in blocks of the following form:
 825          *              -{
 826          *                      word => word ;
 827          *                      word => word ;
 828          *                      ...
 829          *              }-
 830          *
 831          *      To make the tables more manageable, subpages are allowed
 832          *      and will be parsed recursively if $recursive == true.
 833          *
 834          */
 835         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 836                 global $wgMessageCache;
 837                 static $parsed = array();
 838
 839                 if ( !is_object( $wgMessageCache ) ) {
 840                         return array();
 841                 }
 842
 843                 $key = 'Conversiontable/' . $code;
 844                 if ( $subpage ) {
 845                         $key .= '/' . $subpage;
 846                 }
 847                 if ( array_key_exists( $key, $parsed ) ) {
 848                         return array();
 849                 }
 850
 851                 if ( strpos( $code, '/' ) === false ) {
 852                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 853                         if ( $txt === false ) {
 854                                 # FIXME: this method doesn't seem to be expecting
 855                                 # this possible outcome...
 856                                 $txt = '&lt;Conversiontable&gt;';
 857                         }
 858                 } else {
 859                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 860                                                                                    "Conversiontable/$code" );
 861                         if ( $title && $title->exists() ) {
 862                                 $article = new Article( $title );
 863                                 $txt = $article->getContents();
 864                         } else {
 865                                 $txt = '';
 866                         }
 867                 }
 868
 869                 // get all subpage links of the form
 870                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 871                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 872                         ':Conversiontable';
 873                 $subs = StringUtils::explode( '[[', $txt );
 874                 $sublinks = array();
 875                 foreach ( $subs as $sub ) {
 876                         $link = explode( ']]', $sub, 2 );
 877                         if ( count( $link ) != 2 ) {
 878                                 continue;
 879                         }
 880                         $b = explode( '|', $link[0], 2 );
 881                         $b = explode( '/', trim( $b[0] ), 3 );
 882                         if ( count( $b ) == 3 ) {
 883                                 $sublink = $b[2];
 884                         } else {
 885                                 $sublink = '';
 886                         }
 887
 888                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 889                                 $sublinks[] = $sublink;
 890                         }
 891                 }
 892
 893                 // parse the mappings in this page
 894                 $blocks = StringUtils::explode( '-{', $txt );
 895                 $ret = array();
 896                 $first = true;
 897                 foreach ( $blocks as $block ) {
 898                         if ( $first ) {
 899                                 // Skip the part before the first -{
 900                                 $first = false;
 901                                 continue;
 902                         }
 903                         $mappings = explode( '}-', $block, 2 );
 904                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 905                                                                          $mappings[0] );
 906                         $table = StringUtils::explode( ';', $stripped );
 907                         foreach ( $table as $t ) {
 908                                 $m = explode( '=>', $t, 3 );
 909                                 if ( count( $m ) != 2 )
 910                                         continue;
 911                                 // trim any trailling comments starting with '//'
 912                                 $tt = explode( '//', $m[1], 2 );
 913                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 914                         }
 915                 }
 916                 $parsed[$key] = true;
 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                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->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 $text String: text to be tagged for no conversion
 939          * @param $noParse Unused (?)
 940          * @return String: the tagged text
 941          */
 942         public 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          */
 985         public function armourMath( $text ) {
 986                 // we need to convert '-{' and '}-' to '-&#123;' and '&#125;-'
 987                 // to avoid a unwanted '}-' appeared after the math-image.
 988                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 989                 $ret = "-{R|$text}-";
 990                 return $ret;
 991         }
 992
 993         /**
 994          * Get the cached separator pattern for ConverterRule::parseRules()
 995          */
 996         function getVarSeparatorPattern() {
 997                 if ( is_null( $this->mVarSeparatorPattern ) ) {
 998                         // varsep_pattern for preg_split:
 999                         // text should be splited by ";" only if a valid variant
1000                         // name exist after the markup, for example:
1001                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1002                         //      <span style="font-size:120%;">yyy</span>;}-
1003                         // we should split it as:
1004                         //  array(
1005                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1006                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1007                         //        [2] => ''
1008                         //       )
1009                         $pat = '/;\s*(?=';
1010                         foreach ( $this->mVariants as $variant ) {
1011                                 // zh-hans:xxx;zh-hant:yyy
1012                                 $pat .= $variant . '\s*:|';
1013                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1014                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1015                         }
1016                         $pat .= '\s*$)/';
1017                         $this->mVarSeparatorPattern = $pat;
1018                 }
1019                 return $this->mVarSeparatorPattern;
1020         }
1021 }
1022
1023 /**
1024  * Parser for rules of language conversion , parse rules in -{ }- tag.
1025  * @ingroup Language
1026  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1027  */
1028 class ConverterRule {
1029         var $mText; // original text in -{text}-
1030         var $mConverter; // LanguageConverter object
1031         var $mManualCodeError = '<strong class="error">code error!</strong>';
1032         var $mRuleDisplay = '';
1033         var $mRuleTitle = false;
1034         var $mRules = '';// string : the text of the rules
1035         var $mRulesAction = 'none';
1036         var $mFlags = array();
1037         var $mVariantFlags = array();
1038         var $mConvTable = array();
1039         var $mBidtable = array();// array of the translation in each variant
1040         var $mUnidtable = array();// array of the translation in each variant
1041
1042         /**
1043          * Constructor
1044          *
1045          * @param $text String: the text between -{ and }-
1046          * @param $converter LanguageConverter object
1047          */
1048         public function __construct( $text, $converter ) {
1049                 $this->mText = $text;
1050                 $this->mConverter = $converter;
1051         }
1052
1053         /**
1054          * Check if variants array in convert array.
1055          *
1056          * @param $variants Array or string: variant language code
1057          * @return String: translated text
1058          */
1059         public function getTextInBidtable( $variants ) {
1060                 $variants = (array)$variants;
1061                 if ( !$variants ) {
1062                         return false;
1063                 }
1064                 foreach ( $variants as $variant ) {
1065                         if ( isset( $this->mBidtable[$variant] ) ) {
1066                                 return $this->mBidtable[$variant];
1067                         }
1068                 }
1069                 return false;
1070         }
1071
1072         /**
1073          * Parse flags with syntax -{FLAG| ... }-
1074          * @private
1075          */
1076         function parseFlags() {
1077                 $text = $this->mText;
1078                 $flags = array();
1079                 $variantFlags = array();
1080
1081                 $sepPos = strpos( $text, '|' );
1082                 if ( $sepPos !== false ) {
1083                         $validFlags = $this->mConverter->mFlags;
1084                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1085                         foreach ( $f as $ff ) {
1086                                 $ff = trim( $ff );
1087                                 if ( isset( $validFlags[$ff] ) ) {
1088                                         $flags[$validFlags[$ff]] = true;
1089                                 }
1090                         }
1091                         $text = strval( substr( $text, $sepPos + 1 ) );
1092                 }
1093
1094                 if ( !$flags ) {
1095                         $flags['S'] = true;
1096                 } elseif ( isset( $flags['R'] ) ) {
1097                         $flags = array( 'R' => true );// remove other flags
1098                 } elseif ( isset( $flags['N'] ) ) {
1099                         $flags = array( 'N' => true );// remove other flags
1100                 } elseif ( isset( $flags['-'] ) ) {
1101                         $flags = array( '-' => true );// remove other flags
1102                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1103                         $flags['H'] = true;
1104                 } elseif ( isset( $flags['H'] ) ) {
1105                         // replace A flag, and remove other flags except T
1106                         $temp = array( '+' => true, 'H' => true );
1107                         if ( isset( $flags['T'] ) ) {
1108                                 $temp['T'] = true;
1109                         }
1110                         if ( isset( $flags['D'] ) ) {
1111                                 $temp['D'] = true;
1112                         }
1113                         $flags = $temp;
1114                 } else {
1115                         if ( isset( $flags['A'] ) ) {
1116                                 $flags['+'] = true;
1117                                 $flags['S'] = true;
1118                         }
1119                         if ( isset( $flags['D'] ) ) {
1120                                 unset( $flags['S'] );
1121                         }
1122                         // try to find flags like "zh-hans", "zh-hant"
1123                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1124                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1125                         if ( $variantFlags ) {
1126                                 $variantFlags = array_flip( $variantFlags );
1127                                 $flags = array();
1128                         }
1129                 }
1130                 $this->mVariantFlags = $variantFlags;
1131                 $this->mRules = $text;
1132                 $this->mFlags = $flags;
1133         }
1134
1135         /**
1136          * Generate conversion table.
1137          * @private
1138          */
1139         function parseRules() {
1140                 $rules = $this->mRules;
1141                 $bidtable = array();
1142                 $unidtable = array();
1143                 $variants = $this->mConverter->mVariants;
1144                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1145
1146                 $choice = preg_split( $varsep_pattern, $rules );
1147
1148                 foreach ( $choice as $c ) {
1149                         $v  = explode( ':', $c, 2 );
1150                         if ( count( $v ) != 2 ) {
1151                                 // syntax error, skip
1152                                 continue;
1153                         }
1154                         $to = trim( $v[1] );
1155                         $v  = trim( $v[0] );
1156                         $u  = explode( '=>', $v, 2 );
1157                         // if $to is empty, strtr() could return a wrong result
1158                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1159                                 $bidtable[$v] = $to;
1160                         } elseif ( count( $u ) == 2 ) {
1161                                 $from = trim( $u[0] );
1162                                 $v      = trim( $u[1] );
1163                                 if ( array_key_exists( $v, $unidtable )
1164                                          && !is_array( $unidtable[$v] )
1165                                          && $to
1166                                          && in_array( $v, $variants ) ) {
1167                                         $unidtable[$v] = array( $from => $to );
1168                                 } elseif ( $to && in_array( $v, $variants ) ) {
1169                                         $unidtable[$v][$from] = $to;
1170                                 }
1171                         }
1172                         // syntax error, pass
1173                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1174                                 $bidtable = array();
1175                                 $unidtable = array();
1176                                 break;
1177                         }
1178                 }
1179                 $this->mBidtable = $bidtable;
1180                 $this->mUnidtable = $unidtable;
1181         }
1182
1183         /**
1184          * @private
1185          */
1186         function getRulesDesc() {
1187                 $codesep = $this->mConverter->mDescCodeSep;
1188                 $varsep = $this->mConverter->mDescVarSep;
1189                 $text = '';
1190                 foreach ( $this->mBidtable as $k => $v ) {
1191                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1192                 }
1193                 foreach ( $this->mUnidtable as $k => $a ) {
1194                         foreach ( $a as $from => $to ) {
1195                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1196                                         "$codesep$to$varsep";
1197                         }
1198                 }
1199                 return $text;
1200         }
1201
1202         /**
1203          * Parse rules conversion.
1204          * @private
1205          */
1206         function getRuleConvertedStr( $variant ) {
1207                 $bidtable = $this->mBidtable;
1208                 $unidtable = $this->mUnidtable;
1209
1210                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1211                         return $this->mRules;
1212                 } else {
1213                         // display current variant in bidirectional array
1214                         $disp = $this->getTextInBidtable( $variant );
1215                         // or display current variant in fallbacks
1216                         if ( !$disp ) {
1217                                 $disp = $this->getTextInBidtable(
1218                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1219                         }
1220                         // or display current variant in unidirectional array
1221                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1222                                 $disp = array_values( $unidtable[$variant] );
1223                                 $disp = $disp[0];
1224                         }
1225                         // or display frist text under disable manual convert
1226                         if ( !$disp
1227                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1228                                 if ( count( $bidtable ) > 0 ) {
1229                                         $disp = array_values( $bidtable );
1230                                         $disp = $disp[0];
1231                                 } else {
1232                                         $disp = array_values( $unidtable );
1233                                         $disp = array_values( $disp[0] );
1234                                         $disp = $disp[0];
1235                                 }
1236                         }
1237                         return $disp;
1238                 }
1239         }
1240
1241         /**
1242          * Generate conversion table for all text.
1243          * @private
1244          */
1245         function generateConvTable() {
1246                 // Special case optimisation
1247                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1248                         $this->mConvTable = array();
1249                         return;
1250                 }
1251
1252                 $bidtable = $this->mBidtable;
1253                 $unidtable = $this->mUnidtable;
1254                 $manLevel = $this->mConverter->mManualLevel;
1255
1256                 $vmarked = array();
1257                 foreach ( $this->mConverter->mVariants as $v ) {
1258                         /* for bidirectional array
1259                                 fill in the missing variants, if any,
1260                                 with fallbacks */
1261                         if ( !isset( $bidtable[$v] ) ) {
1262                                 $variantFallbacks =
1263                                         $this->mConverter->getVariantFallbacks( $v );
1264                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1265                                 if ( $vf ) {
1266                                         $bidtable[$v] = $vf;
1267                                 }
1268                         }
1269
1270                         if ( isset( $bidtable[$v] ) ) {
1271                                 foreach ( $vmarked as $vo ) {
1272                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1273                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1274                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1275                                         // to introduce a custom mapping between
1276                                         // words WordZh and WordTw in the whole text
1277                                         if ( $manLevel[$v] == 'bidirectional' ) {
1278                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1279                                         }
1280                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1281                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1282                                         }
1283                                 }
1284                                 $vmarked[] = $v;
1285                         }
1286                         /*for unidirectional array fill to convert tables */
1287                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1288                                 && isset( $unidtable[$v] ) )
1289                         {
1290                                 if ( isset( $this->mConvTable[$v] ) ) {
1291                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1292                                 } else {
1293                                         $this->mConvTable[$v] = $unidtable[$v];
1294                                 }
1295                         }
1296                 }
1297         }
1298
1299         /**
1300          * Parse rules and flags.
1301          * @public
1302          */
1303         function parse( $variant = NULL ) {
1304                 if ( !$variant ) {
1305                         $variant = $this->mConverter->getPreferredVariant();
1306                 }
1307
1308                 $this->parseFlags();
1309                 $flags = $this->mFlags;
1310
1311                 // convert to specified variant
1312                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1313                 if ( $this->mVariantFlags ) {
1314                         // check if current variant in flags
1315                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1316                                 // then convert <text to convert> to current language
1317                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1318                                                                                                                                 $variant );
1319                         } else { // if current variant no in flags,
1320                                    // then we check its fallback variants.
1321                                 $variantFallbacks =
1322                                         $this->mConverter->getVariantFallbacks( $variant );
1323                                 foreach ( $variantFallbacks as $variantFallback ) {
1324                                         // if current variant's fallback exist in flags
1325                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1326                                                 // then convert <text to convert> to fallback language
1327                                                 $this->mRules =
1328                                                         $this->mConverter->autoConvert( $this->mRules,
1329                                                                                                                         $variantFallback );
1330                                                 break;
1331                                         }
1332                                 }
1333                         }
1334                         $this->mFlags = $flags = array( 'R' => true );
1335                 }
1336
1337                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1338                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1339                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1340                         $this->parseRules();
1341                 }
1342                 $rules = $this->mRules;
1343
1344                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1345                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1346                                 // fill all variants if text in -{A/H/-|text} without rules
1347                                 foreach ( $this->mConverter->mVariants as $v ) {
1348                                         $this->mBidtable[$v] = $rules;
1349                                 }
1350                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1351                                 $this->mFlags = $flags = array( 'R' => true );
1352                         }
1353                 }
1354
1355                 $this->mRuleDisplay = false;
1356                 foreach ( $flags as $flag => $unused ) {
1357                         switch ( $flag ) {
1358                                 case 'R':
1359                                         // if we don't do content convert, still strip the -{}- tags
1360                                         $this->mRuleDisplay = $rules;
1361                                         break;
1362                                 case 'N':
1363                                         // process N flag: output current variant name
1364                                         $ruleVar = trim( $rules );
1365                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1366                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1367                                         } else {
1368                                                 $this->mRuleDisplay = '';
1369                                         }
1370                                         break;
1371                                 case 'D':
1372                                         // process D flag: output rules description
1373                                         $this->mRuleDisplay = $this->getRulesDesc();
1374                                         break;
1375                                 case 'H':
1376                                         // process H,- flag or T only: output nothing
1377                                         $this->mRuleDisplay = '';
1378                                         break;
1379                                 case '-':
1380                                         $this->mRulesAction = 'remove';
1381                                         $this->mRuleDisplay = '';
1382                                         break;
1383                                 case '+':
1384                                         $this->mRulesAction = 'add';
1385                                         $this->mRuleDisplay = '';
1386                                         break;
1387                                 case 'S':
1388                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1389                                         break;
1390                                 case 'T':
1391                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1392                                         $this->mRuleDisplay = '';
1393                                         break;
1394                                 default:
1395                                         // ignore unknown flags (but see error case below)
1396                         }
1397                 }
1398                 if ( $this->mRuleDisplay === false ) {
1399                         $this->mRuleDisplay = $this->mManualCodeError;
1400                 }
1401
1402                 $this->generateConvTable();
1403         }
1404
1405         /**
1406          * @public
1407          */
1408         function hasRules() {
1409                 // TODO:
1410         }
1411
1412         /**
1413          * Get display text on markup -{...}-
1414          * @public
1415          */
1416         function getDisplay() {
1417                 return $this->mRuleDisplay;
1418         }
1419
1420         /**
1421          * Get converted title.
1422          * @public
1423          */
1424         function getTitle() {
1425                 return $this->mRuleTitle;
1426         }
1427
1428         /**
1429          * Return how deal with conversion rules.
1430          * @public
1431          */
1432         function getRulesAction() {
1433                 return $this->mRulesAction;
1434         }
1435
1436         /**
1437          * Get conversion table. ( bidirectional and unidirectional
1438          * conversion table )
1439          * @public
1440          */
1441         function getConvTable() {
1442                 return $this->mConvTable;
1443         }
1444
1445         /**
1446          * Get conversion rules string.
1447          * @public
1448          */
1449         function getRules() {
1450                 return $this->mRules;
1451         }
1452
1453         /**
1454          * Get conversion flags.
1455          * @public
1456          */
1457         function getFlags() {
1458                 return $this->mFlags;
1459         }
1460 }