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                                 return $text;
 349                         }
 350                 }
 351
 352                 /* we convert everything except:
 353                    1. html markups (anything between < and >)
 354                    2. html entities
 355                    3. place holders created by the parser
 356                 */
 357                 global $wgParser;
 358                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 359                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 360                 } else {
 361                         $marker = '';
 362                 }
 363
 364                 // this one is needed when the text is inside an html markup
 365                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 366
 367                 // disable convert to variants between <code></code> tags
 368                 $codefix = '<code>.+?<\/code>|';
 369                 // disable convertsion of <script type="text/javascript"> ... </script>
 370                 $scriptfix = '<script.*?>.*?<\/script>|';
 371                 // disable conversion of <pre xxxx> ... </pre>
 372                 $prefix = '<pre.*?>.*?<\/pre>|';
 373
 374                 $reg = '/' . $codefix . $scriptfix . $prefix .
 375                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 376
 377                 $matches = preg_split( $reg, $text, - 1, PREG_SPLIT_OFFSET_CAPTURE );
 378
 379                 $m = array_shift( $matches );
 380
 381                 $ret = $this->translate( $m[0], $toVariant );
 382                 $mstart = $m[1] + strlen( $m[0] );
 383
 384                 // enable convertsion of '<img alt="xxxx" ... '
 385                 // or '<span title="xxxx" ... '
 386                 $captionpattern  = '/\s(title|alt)\s*=\s*"([\s\S]*?)"/';
 387
 388                 $trtext = '';
 389                 $trtextmark = "\0";
 390                 $notrtext = array();
 391                 foreach ( $matches as $m ) {
 392                         $mark = substr( $text, $mstart, $m[1] - $mstart );
 393                         $mark = preg_replace_callback( $captionpattern,
 394                                                                                    array( &$this, 'captionConvert' ),
 395                                                                                    $mark );
 396                         // Let's convert the trtext only once,
 397                         // it would give us more performance improvement
 398                         $notrtext[] = $mark;
 399                         $trtext .= $m[0] . $trtextmark;
 400                         $mstart = $m[1] + strlen( $m[0] );
 401                 }
 402                 $notrtext[] = '';
 403                 $trtext = $this->translate( $trtext, $toVariant );
 404                 $trtext = StringUtils::explode( $trtextmark, $trtext );
 405                 foreach ( $trtext as $t ) {
 406                         $ret .= array_shift( $notrtext );
 407                         $ret .= $t;
 408                 }
 409                 wfProfileOut( __METHOD__ );
 410                 return $ret;
 411         }
 412
 413         /**
 414          * Translate a string to a variant.
 415          * Doesn't parse rules or do any of that other stuff, for that use
 416          * convert() or convertTo().
 417          *
 418          * @param $text String: text to convert
 419          * @param $variant String: variant language code
 420          * @return String: translated text
 421          */
 422         protected function translate( $text, $variant ) {
 423                 wfProfileIn( __METHOD__ );
 424                 // If $text is empty or only includes spaces, do nothing
 425                 // Otherwise translate it
 426                 if ( trim( $text ) ) {
 427                         $this->loadTables();
 428                         $text = $this->mTables[$variant]->replace( $text );
 429                 }
 430                 wfProfileOut( __METHOD__ );
 431                 return $text;
 432         }
 433
 434         /**
 435          * Call translate() to convert text to all valid variants.
 436          *
 437          * @param $text String: the text to be converted
 438          * @return Array: variant => converted text
 439          */
 440         public function autoConvertToAllVariants( $text ) {
 441                 wfProfileIn( __METHOD__ );
 442                 $this->loadTables();
 443
 444                 $ret = array();
 445                 foreach ( $this->mVariants as $variant ) {
 446                         $ret[$variant] = $this->translate( $text, $variant );
 447                 }
 448
 449                 wfProfileOut( __METHOD__ );
 450                 return $ret;
 451         }
 452
 453         /**
 454          * Convert link text to all valid variants.
 455          * In the first, this function only convert text outside the
 456          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 457          * titles, the text will get all converted always.
 458          * So I removed this feature and deprecated the function.
 459          *
 460          * @param $text String: the text to be converted
 461          * @return Array: variant => converted text
 462          * @deprecated Use autoConvertToAllVariants() instead
 463          */
 464         public function convertLinkToAllVariants( $text ) {
 465                 return $this->autoConvertToAllVariants( $text );
 466         }
 467
 468         /**
 469          * Apply manual conversion rules.
 470          *
 471          * @param $convRule Object: Object of ConverterRule
 472          */
 473         protected function applyManualConv( $convRule ) {
 474                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 475                 // title conversion.
 476                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 477                 // rule(s) not for title, this breaks the title conversion.
 478                 $newConvRuleTitle = $convRule->getTitle();
 479                 if ( $newConvRuleTitle ) {
 480                         // So I add an empty check for getTitle()
 481                         $this->mConvRuleTitle = $newConvRuleTitle;
 482                 }
 483
 484                 // merge/remove manual conversion rules to/from global table
 485                 $convTable = $convRule->getConvTable();
 486                 $action = $convRule->getRulesAction();
 487                 foreach ( $convTable as $variant => $pair ) {
 488                         if ( !$this->validateVariant( $variant ) ) {
 489                                 continue;
 490                         }
 491
 492                         if ( $action == 'add' ) {
 493                                 foreach ( $pair as $from => $to ) {
 494                                         // to ensure that $from and $to not be left blank
 495                                         // so $this->translate() could always return a string
 496                                         if ( $from || $to ) {
 497                                                 // more efficient than array_merge(), about 2.5 times.
 498                                                 $this->mTables[$variant]->setPair( $from, $to );
 499                                         }
 500                                 }
 501                         } elseif ( $action == 'remove' ) {
 502                                 $this->mTables[$variant]->removeArray( $pair );
 503                         }
 504                 }
 505         }
 506
 507         /**
 508          * Auto convert a Title object to a readable string in the
 509          * preferred variant.
 510          *
 511          *@param $title Object: a object of Title
 512          *@return String: converted title text
 513          */
 514         public function convertTitle( $title ) {
 515                 $variant = $this->getPreferredVariant();
 516                 $index = $title->getNamespace();
 517                 if ( $index === NS_MAIN ) {
 518                         $text = '';
 519                 } else {
 520                         // first let's check if a message has given us a converted name
 521                         $nsConvKey = 'conversion-ns' . $index;
 522                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 523                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 524                         } else {
 525                                 // the message does not exist, try retrieve it from the current
 526                                 // variant's namespace names.
 527                                 $langObj = $this->mLangObj->factory( $variant );
 528                                 $text = $langObj->getFormattedNsText( $index );
 529                         }
 530                         $text .= ':';
 531                 }
 532                 $text .= $title->getText();
 533                 $text = $this->translate( $text, $variant );
 534                 return $text;
 535         }
 536
 537         /**
 538          * Convert text to different variants of a language. The automatic
 539          * conversion is done in autoConvert(). Here we parse the text
 540          * marked with -{}-, which specifies special conversions of the
 541          * text that can not be accomplished in autoConvert().
 542          *
 543          * Syntax of the markup:
 544          * -{code1:text1;code2:text2;...}-  or
 545          * -{flags|code1:text1;code2:text2;...}-  or
 546          * -{text}- in which case no conversion should take place for text
 547          *
 548          * @param $text String: text to be converted
 549          * @return String: converted text
 550          */
 551         public function convert( $text ) {
 552                 $variant = $this->getPreferredVariant();
 553                 return $this->convertTo( $text, $variant );
 554         }
 555
 556         /**
 557          * Same as convert() except a extra parameter to custom variant.
 558          *
 559          * @param $text String: text to be converted
 560          * @param $variant String: the target variant code
 561          * @return String: converted text
 562          */
 563         public function convertTo( $text, $variant ) {
 564                 global $wgDisableLangConversion;
 565                 if ( $wgDisableLangConversion ) return $text;
 566                 return $this->recursiveConvertTopLevel( $text, $variant );
 567         }
 568
 569         /**
 570          * Recursively convert text on the outside. Allow to use nested
 571          * markups to custom rules.
 572          *
 573          * @param $text String: text to be converted
 574          * @param $variant String: the target variant code
 575          * @param $depth Integer: depth of recursion
 576          * @return String: converted text
 577          */
 578         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 579                 $startPos = 0;
 580                 $out = '';
 581                 $length = strlen( $text );
 582                 while ( $startPos < $length ) {
 583                         $pos = strpos( $text, '-{', $startPos );
 584
 585                         if ( $pos === false ) {
 586                                 // No more markup, append final segment
 587                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 588                                 return $out;
 589                         }
 590
 591                         // Markup found
 592                         // Append initial segment
 593                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 594
 595                         // Advance position
 596                         $startPos = $pos;
 597
 598                         // Do recursive conversion
 599                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 600                 }
 601
 602                 return $out;
 603         }
 604
 605         /**
 606          * Recursively convert text on the inside.
 607          *
 608          * @param $text String: text to be converted
 609          * @param $variant String: the target variant code
 610          * @param $depth Integer: depth of recursion
 611          * @return String: converted text
 612          */
 613         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 614                 // Quick sanity check (no function calls)
 615                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 616                         throw new MWException( __METHOD__ . ': invalid input string' );
 617                 }
 618
 619                 $startPos += 2;
 620                 $inner = '';
 621                 $warningDone = false;
 622                 $length = strlen( $text );
 623
 624                 while ( $startPos < $length ) {
 625                         $m = false;
 626                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 627                         if ( !$m ) {
 628                                 // Unclosed rule
 629                                 break;
 630                         }
 631
 632                         $token = $m[0][0];
 633                         $pos = $m[0][1];
 634
 635                         // Markup found
 636                         // Append initial segment
 637                         $inner .= substr( $text, $startPos, $pos - $startPos );
 638
 639                         // Advance position
 640                         $startPos = $pos;
 641
 642                         switch ( $token ) {
 643                                 case '-{':
 644                                         // Check max depth
 645                                         if ( $depth >= $this->mMaxDepth ) {
 646                                                 $inner .= '-{';
 647                                                 if ( !$warningDone ) {
 648                                                         $inner .= '<span class="error">' .
 649                                                                 wfMsgForContent( 'language-converter-depth-warning',
 650                                                                         $this->mMaxDepth ) .
 651                                                                 '</span>';
 652                                                         $warningDone = true;
 653                                                 }
 654                                                 $startPos += 2;
 655                                                 continue;
 656                                         }
 657                                         // Recursively parse another rule
 658                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 659                                         break;
 660                                 case '}-':
 661                                         // Apply the rule
 662                                         $startPos += 2;
 663                                         $rule = new ConverterRule( $inner, $this );
 664                                         $rule->parse( $variant );
 665                                         $this->applyManualConv( $rule );
 666                                         return $rule->getDisplay();
 667                                 default:
 668                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 669                         }
 670                 }
 671
 672                 // Unclosed rule
 673                 if ( $startPos < $length ) {
 674                         $inner .= substr( $text, $startPos );
 675                 }
 676                 $startPos = $length;
 677                 return '-{' . $this->autoConvert( $inner, $variant );
 678         }
 679
 680         /**
 681          * If a language supports multiple variants, it is
 682          * possible that non-existing link in one variant
 683          * actually exists in another variant. This function
 684          * tries to find it. See e.g. LanguageZh.php
 685          *
 686          * @param $link String: the name of the link
 687          * @param $nt Mixed: the title object of the link
 688          * @param $ignoreOtherCond Boolean: to disable other conditions when
 689          *              we need to transclude a template or update a category's link
 690          * @return Null, the input parameters may be modified upon return
 691          */
 692         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 693                 # If the article has already existed, there is no need to
 694                 # check it again, otherwise it may cause a fault.
 695                 if ( is_object( $nt ) && $nt->exists() ) {
 696                         return;
 697                 }
 698
 699                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 700                         $wgUser;
 701                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 702                 $action = $wgRequest->getText( 'action' );
 703                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 704                 $disableLinkConversion = $wgDisableLangConversion
 705                         || $wgDisableTitleConversion;
 706                 $linkBatch = new LinkBatch();
 707
 708                 $ns = NS_MAIN;
 709
 710                 if ( $disableLinkConversion ||
 711                          ( !$ignoreOtherCond &&
 712                            ( $isredir == 'no'
 713                                  || $action == 'edit'
 714                                  || $action == 'submit'
 715                                  || $linkconvert == 'no'
 716                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 717                         return;
 718                 }
 719
 720                 if ( is_object( $nt ) ) {
 721                         $ns = $nt->getNamespace();
 722                 }
 723
 724                 $variants = $this->autoConvertToAllVariants( $link );
 725                 if ( !$variants ) { // give up
 726                         return;
 727                 }
 728
 729                 $titles = array();
 730
 731                 foreach ( $variants as $v ) {
 732                         if ( $v != $link ) {
 733                                 $varnt = Title::newFromText( $v, $ns );
 734                                 if ( !is_null( $varnt ) ) {
 735                                         $linkBatch->addObj( $varnt );
 736                                         $titles[] = $varnt;
 737                                 }
 738                         }
 739                 }
 740
 741                 // fetch all variants in single query
 742                 $linkBatch->execute();
 743
 744                 foreach ( $titles as $varnt ) {
 745                         if ( $varnt->getArticleID() > 0 ) {
 746                                 $nt = $varnt;
 747                                 $link = $varnt->getText();
 748                                 break;
 749                         }
 750                 }
 751         }
 752
 753         /**
 754          * Returns language specific hash options.
 755          */
 756         public function getExtraHashOptions() {
 757                 $variant = $this->getPreferredVariant();
 758                 return '!' . $variant ;
 759         }
 760
 761         /**
 762          * Load default conversion tables.
 763          * This method must be implemented in derived class.
 764          *
 765          * @private
 766          */
 767         function loadDefaultTables() {
 768                 $name = get_class( $this );
 769                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 770         }
 771
 772         /**
 773          * Load conversion tables either from the cache or the disk.
 774          * @private
 775          */
 776         function loadTables( $fromcache = true ) {
 777                 if ( $this->mTablesLoaded ) {
 778                         return;
 779                 }
 780                 global $wgMemc;
 781                 wfProfileIn( __METHOD__ );
 782                 $this->mTablesLoaded = true;
 783                 $this->mTables = false;
 784                 if ( $fromcache ) {
 785                         wfProfileIn( __METHOD__ . '-cache' );
 786                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 787                         wfProfileOut( __METHOD__ . '-cache' );
 788                 }
 789                 if ( !$this->mTables
 790                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 791                         wfProfileIn( __METHOD__ . '-recache' );
 792                         // not in cache, or we need a fresh reload.
 793                         // we will first load the default tables
 794                         // then update them using things in MediaWiki:Zhconversiontable/*
 795                         $this->loadDefaultTables();
 796                         foreach ( $this->mVariants as $var ) {
 797                                 $cached = $this->parseCachedTable( $var );
 798                                 $this->mTables[$var]->mergeArray( $cached );
 799                         }
 800
 801                         $this->postLoadTables();
 802                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 803
 804                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 805                         wfProfileOut( __METHOD__ . '-recache' );
 806                 }
 807                 wfProfileOut( __METHOD__ );
 808         }
 809
 810         /**
 811          * Hook for post processig after conversion tables are loaded.
 812          *
 813          */
 814         function postLoadTables() { }
 815
 816         /**
 817          * Reload the conversion tables.
 818          *
 819          * @private
 820          */
 821         function reloadTables() {
 822                 if ( $this->mTables ) {
 823                         unset( $this->mTables );
 824                 }
 825                 $this->mTablesLoaded = false;
 826                 $this->loadTables( false );
 827         }
 828
 829
 830         /**
 831          * Parse the conversion table stored in the cache.
 832          *
 833          * The tables should be in blocks of the following form:
 834          *              -{
 835          *                      word => word ;
 836          *                      word => word ;
 837          *                      ...
 838          *              }-
 839          *
 840          *      To make the tables more manageable, subpages are allowed
 841          *      and will be parsed recursively if $recursive == true.
 842          *
 843          */
 844         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 845                 global $wgMessageCache;
 846                 static $parsed = array();
 847
 848                 if ( !is_object( $wgMessageCache ) ) {
 849                         return array();
 850                 }
 851
 852                 $key = 'Conversiontable/' . $code;
 853                 if ( $subpage ) {
 854                         $key .= '/' . $subpage;
 855                 }
 856                 if ( array_key_exists( $key, $parsed ) ) {
 857                         return array();
 858                 }
 859
 860                 if ( strpos( $code, '/' ) === false ) {
 861                         $txt = $wgMessageCache->get( 'Conversiontable', true, $code );
 862                         if ( $txt === false ) {
 863                                 # FIXME: this method doesn't seem to be expecting
 864                                 # this possible outcome...
 865                                 $txt = '&lt;Conversiontable&gt;';
 866                         }
 867                 } else {
 868                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 869                                                                                    "Conversiontable/$code" );
 870                         if ( $title && $title->exists() ) {
 871                                 $article = new Article( $title );
 872                                 $txt = $article->getContents();
 873                         } else {
 874                                 $txt = '';
 875                         }
 876                 }
 877
 878                 // get all subpage links of the form
 879                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 880                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 881                         ':Conversiontable';
 882                 $subs = StringUtils::explode( '[[', $txt );
 883                 $sublinks = array();
 884                 foreach ( $subs as $sub ) {
 885                         $link = explode( ']]', $sub, 2 );
 886                         if ( count( $link ) != 2 ) {
 887                                 continue;
 888                         }
 889                         $b = explode( '|', $link[0], 2 );
 890                         $b = explode( '/', trim( $b[0] ), 3 );
 891                         if ( count( $b ) == 3 ) {
 892                                 $sublink = $b[2];
 893                         } else {
 894                                 $sublink = '';
 895                         }
 896
 897                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 898                                 $sublinks[] = $sublink;
 899                         }
 900                 }
 901
 902                 // parse the mappings in this page
 903                 $blocks = StringUtils::explode( '-{', $txt );
 904                 $ret = array();
 905                 $first = true;
 906                 foreach ( $blocks as $block ) {
 907                         if ( $first ) {
 908                                 // Skip the part before the first -{
 909                                 $first = false;
 910                                 continue;
 911                         }
 912                         $mappings = explode( '}-', $block, 2 );
 913                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 914                                                                          $mappings[0] );
 915                         $table = StringUtils::explode( ';', $stripped );
 916                         foreach ( $table as $t ) {
 917                                 $m = explode( '=>', $t, 3 );
 918                                 if ( count( $m ) != 2 )
 919                                         continue;
 920                                 // trim any trailling comments starting with '//'
 921                                 $tt = explode( '//', $m[1], 2 );
 922                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 923                         }
 924                 }
 925                 $parsed[$key] = true;
 926
 927                 // recursively parse the subpages
 928                 if ( $recursive ) {
 929                         foreach ( $sublinks as $link ) {
 930                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 931                                 $ret = array_merge( $ret, $s );
 932                         }
 933                 }
 934
 935                 if ( $this->mUcfirst ) {
 936                         foreach ( $ret as $k => $v ) {
 937                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 938                         }
 939                 }
 940                 return $ret;
 941         }
 942
 943         /**
 944          * Enclose a string with the "no conversion" tag. This is used by
 945          * various functions in the Parser.
 946          *
 947          * @param $text String: text to be tagged for no conversion
 948          * @param $noParse Unused (?)
 949          * @return String: the tagged text
 950          */
 951         public function markNoConversion( $text, $noParse = false ) {
 952                 # don't mark if already marked
 953                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 954                         return $text;
 955                 }
 956
 957                 $ret = "-{R|$text}-";
 958                 return $ret;
 959         }
 960
 961         /**
 962          * Convert the sorting key for category links. This should make different
 963          * keys that are variants of each other map to the same key.
 964          */
 965         function convertCategoryKey( $key ) {
 966                 return $key;
 967         }
 968
 969         /**
 970          * Hook to refresh the cache of conversion tables when
 971          * MediaWiki:conversiontable* is updated.
 972          * @private
 973          */
 974         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 975                         $iswatch, $section, $flags, $revision ) {
 976                 $titleobj = $article->getTitle();
 977                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 978                         $title = $titleobj->getDBkey();
 979                         $t = explode( '/', $title, 3 );
 980                         $c = count( $t );
 981                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 982                                 if ( $this->validateVariant( $t[1] ) ) {
 983                                         $this->reloadTables();
 984                                 }
 985                         }
 986                 }
 987                 return true;
 988         }
 989
 990         /**
 991          * Armour rendered math against conversion.
 992          * Escape special chars in parsed math text.(in most cases are img elements)
 993          */
 994         public function armourMath( $text ) {
 995                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
 996                 // any unwanted markup appearing in the math image tag.
 997                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
 998                 return $text;
 999         }
1000
1001         /**
1002          * Get the cached separator pattern for ConverterRule::parseRules()
1003          */
1004         function getVarSeparatorPattern() {
1005                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1006                         // varsep_pattern for preg_split:
1007                         // text should be splited by ";" only if a valid variant
1008                         // name exist after the markup, for example:
1009                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1010                         //      <span style="font-size:120%;">yyy</span>;}-
1011                         // we should split it as:
1012                         //  array(
1013                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1014                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1015                         //        [2] => ''
1016                         //       )
1017                         $pat = '/;\s*(?=';
1018                         foreach ( $this->mVariants as $variant ) {
1019                                 // zh-hans:xxx;zh-hant:yyy
1020                                 $pat .= $variant . '\s*:|';
1021                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1022                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1023                         }
1024                         $pat .= '\s*$)/';
1025                         $this->mVarSeparatorPattern = $pat;
1026                 }
1027                 return $this->mVarSeparatorPattern;
1028         }
1029 }
1030
1031 /**
1032  * Parser for rules of language conversion , parse rules in -{ }- tag.
1033  * @ingroup Language
1034  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1035  */
1036 class ConverterRule {
1037         var $mText; // original text in -{text}-
1038         var $mConverter; // LanguageConverter object
1039         var $mManualCodeError = '<strong class="error">code error!</strong>';
1040         var $mRuleDisplay = '';
1041         var $mRuleTitle = false;
1042         var $mRules = '';// string : the text of the rules
1043         var $mRulesAction = 'none';
1044         var $mFlags = array();
1045         var $mVariantFlags = array();
1046         var $mConvTable = array();
1047         var $mBidtable = array();// array of the translation in each variant
1048         var $mUnidtable = array();// array of the translation in each variant
1049
1050         /**
1051          * Constructor
1052          *
1053          * @param $text String: the text between -{ and }-
1054          * @param $converter LanguageConverter object
1055          */
1056         public function __construct( $text, $converter ) {
1057                 $this->mText = $text;
1058                 $this->mConverter = $converter;
1059         }
1060
1061         /**
1062          * Check if variants array in convert array.
1063          *
1064          * @param $variants Array or string: variant language code
1065          * @return String: translated text
1066          */
1067         public function getTextInBidtable( $variants ) {
1068                 $variants = (array)$variants;
1069                 if ( !$variants ) {
1070                         return false;
1071                 }
1072                 foreach ( $variants as $variant ) {
1073                         if ( isset( $this->mBidtable[$variant] ) ) {
1074                                 return $this->mBidtable[$variant];
1075                         }
1076                 }
1077                 return false;
1078         }
1079
1080         /**
1081          * Parse flags with syntax -{FLAG| ... }-
1082          * @private
1083          */
1084         function parseFlags() {
1085                 $text = $this->mText;
1086                 $flags = array();
1087                 $variantFlags = array();
1088
1089                 $sepPos = strpos( $text, '|' );
1090                 if ( $sepPos !== false ) {
1091                         $validFlags = $this->mConverter->mFlags;
1092                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1093                         foreach ( $f as $ff ) {
1094                                 $ff = trim( $ff );
1095                                 if ( isset( $validFlags[$ff] ) ) {
1096                                         $flags[$validFlags[$ff]] = true;
1097                                 }
1098                         }
1099                         $text = strval( substr( $text, $sepPos + 1 ) );
1100                 }
1101
1102                 if ( !$flags ) {
1103                         $flags['S'] = true;
1104                 } elseif ( isset( $flags['R'] ) ) {
1105                         $flags = array( 'R' => true );// remove other flags
1106                 } elseif ( isset( $flags['N'] ) ) {
1107                         $flags = array( 'N' => true );// remove other flags
1108                 } elseif ( isset( $flags['-'] ) ) {
1109                         $flags = array( '-' => true );// remove other flags
1110                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1111                         $flags['H'] = true;
1112                 } elseif ( isset( $flags['H'] ) ) {
1113                         // replace A flag, and remove other flags except T
1114                         $temp = array( '+' => true, 'H' => true );
1115                         if ( isset( $flags['T'] ) ) {
1116                                 $temp['T'] = true;
1117                         }
1118                         if ( isset( $flags['D'] ) ) {
1119                                 $temp['D'] = true;
1120                         }
1121                         $flags = $temp;
1122                 } else {
1123                         if ( isset( $flags['A'] ) ) {
1124                                 $flags['+'] = true;
1125                                 $flags['S'] = true;
1126                         }
1127                         if ( isset( $flags['D'] ) ) {
1128                                 unset( $flags['S'] );
1129                         }
1130                         // try to find flags like "zh-hans", "zh-hant"
1131                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1132                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1133                         if ( $variantFlags ) {
1134                                 $variantFlags = array_flip( $variantFlags );
1135                                 $flags = array();
1136                         }
1137                 }
1138                 $this->mVariantFlags = $variantFlags;
1139                 $this->mRules = $text;
1140                 $this->mFlags = $flags;
1141         }
1142
1143         /**
1144          * Generate conversion table.
1145          * @private
1146          */
1147         function parseRules() {
1148                 $rules = $this->mRules;
1149                 $bidtable = array();
1150                 $unidtable = array();
1151                 $variants = $this->mConverter->mVariants;
1152                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1153
1154                 $choice = preg_split( $varsep_pattern, $rules );
1155
1156                 foreach ( $choice as $c ) {
1157                         $v  = explode( ':', $c, 2 );
1158                         if ( count( $v ) != 2 ) {
1159                                 // syntax error, skip
1160                                 continue;
1161                         }
1162                         $to = trim( $v[1] );
1163                         $v  = trim( $v[0] );
1164                         $u  = explode( '=>', $v, 2 );
1165                         // if $to is empty, strtr() could return a wrong result
1166                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1167                                 $bidtable[$v] = $to;
1168                         } elseif ( count( $u ) == 2 ) {
1169                                 $from = trim( $u[0] );
1170                                 $v      = trim( $u[1] );
1171                                 if ( array_key_exists( $v, $unidtable )
1172                                          && !is_array( $unidtable[$v] )
1173                                          && $to
1174                                          && in_array( $v, $variants ) ) {
1175                                         $unidtable[$v] = array( $from => $to );
1176                                 } elseif ( $to && in_array( $v, $variants ) ) {
1177                                         $unidtable[$v][$from] = $to;
1178                                 }
1179                         }
1180                         // syntax error, pass
1181                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1182                                 $bidtable = array();
1183                                 $unidtable = array();
1184                                 break;
1185                         }
1186                 }
1187                 $this->mBidtable = $bidtable;
1188                 $this->mUnidtable = $unidtable;
1189         }
1190
1191         /**
1192          * @private
1193          */
1194         function getRulesDesc() {
1195                 $codesep = $this->mConverter->mDescCodeSep;
1196                 $varsep = $this->mConverter->mDescVarSep;
1197                 $text = '';
1198                 foreach ( $this->mBidtable as $k => $v ) {
1199                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1200                 }
1201                 foreach ( $this->mUnidtable as $k => $a ) {
1202                         foreach ( $a as $from => $to ) {
1203                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1204                                         "$codesep$to$varsep";
1205                         }
1206                 }
1207                 return $text;
1208         }
1209
1210         /**
1211          * Parse rules conversion.
1212          * @private
1213          */
1214         function getRuleConvertedStr( $variant ) {
1215                 $bidtable = $this->mBidtable;
1216                 $unidtable = $this->mUnidtable;
1217
1218                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1219                         return $this->mRules;
1220                 } else {
1221                         // display current variant in bidirectional array
1222                         $disp = $this->getTextInBidtable( $variant );
1223                         // or display current variant in fallbacks
1224                         if ( !$disp ) {
1225                                 $disp = $this->getTextInBidtable(
1226                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1227                         }
1228                         // or display current variant in unidirectional array
1229                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1230                                 $disp = array_values( $unidtable[$variant] );
1231                                 $disp = $disp[0];
1232                         }
1233                         // or display frist text under disable manual convert
1234                         if ( !$disp
1235                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1236                                 if ( count( $bidtable ) > 0 ) {
1237                                         $disp = array_values( $bidtable );
1238                                         $disp = $disp[0];
1239                                 } else {
1240                                         $disp = array_values( $unidtable );
1241                                         $disp = array_values( $disp[0] );
1242                                         $disp = $disp[0];
1243                                 }
1244                         }
1245                         return $disp;
1246                 }
1247         }
1248
1249         /**
1250          * Generate conversion table for all text.
1251          * @private
1252          */
1253         function generateConvTable() {
1254                 // Special case optimisation
1255                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1256                         $this->mConvTable = array();
1257                         return;
1258                 }
1259
1260                 $bidtable = $this->mBidtable;
1261                 $unidtable = $this->mUnidtable;
1262                 $manLevel = $this->mConverter->mManualLevel;
1263
1264                 $vmarked = array();
1265                 foreach ( $this->mConverter->mVariants as $v ) {
1266                         /* for bidirectional array
1267                                 fill in the missing variants, if any,
1268                                 with fallbacks */
1269                         if ( !isset( $bidtable[$v] ) ) {
1270                                 $variantFallbacks =
1271                                         $this->mConverter->getVariantFallbacks( $v );
1272                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1273                                 if ( $vf ) {
1274                                         $bidtable[$v] = $vf;
1275                                 }
1276                         }
1277
1278                         if ( isset( $bidtable[$v] ) ) {
1279                                 foreach ( $vmarked as $vo ) {
1280                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1281                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1282                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1283                                         // to introduce a custom mapping between
1284                                         // words WordZh and WordTw in the whole text
1285                                         if ( $manLevel[$v] == 'bidirectional' ) {
1286                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1287                                         }
1288                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1289                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1290                                         }
1291                                 }
1292                                 $vmarked[] = $v;
1293                         }
1294                         /*for unidirectional array fill to convert tables */
1295                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1296                                 && isset( $unidtable[$v] ) )
1297                         {
1298                                 if ( isset( $this->mConvTable[$v] ) ) {
1299                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1300                                 } else {
1301                                         $this->mConvTable[$v] = $unidtable[$v];
1302                                 }
1303                         }
1304                 }
1305         }
1306
1307         /**
1308          * Parse rules and flags.
1309          * @public
1310          */
1311         function parse( $variant = NULL ) {
1312                 if ( !$variant ) {
1313                         $variant = $this->mConverter->getPreferredVariant();
1314                 }
1315
1316                 $this->parseFlags();
1317                 $flags = $this->mFlags;
1318
1319                 // convert to specified variant
1320                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1321                 if ( $this->mVariantFlags ) {
1322                         // check if current variant in flags
1323                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1324                                 // then convert <text to convert> to current language
1325                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1326                                                                                                                                 $variant );
1327                         } else { // if current variant no in flags,
1328                                    // then we check its fallback variants.
1329                                 $variantFallbacks =
1330                                         $this->mConverter->getVariantFallbacks( $variant );
1331                                 foreach ( $variantFallbacks as $variantFallback ) {
1332                                         // if current variant's fallback exist in flags
1333                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1334                                                 // then convert <text to convert> to fallback language
1335                                                 $this->mRules =
1336                                                         $this->mConverter->autoConvert( $this->mRules,
1337                                                                                                                         $variantFallback );
1338                                                 break;
1339                                         }
1340                                 }
1341                         }
1342                         $this->mFlags = $flags = array( 'R' => true );
1343                 }
1344
1345                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1346                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1347                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1348                         $this->parseRules();
1349                 }
1350                 $rules = $this->mRules;
1351
1352                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1353                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1354                                 // fill all variants if text in -{A/H/-|text} without rules
1355                                 foreach ( $this->mConverter->mVariants as $v ) {
1356                                         $this->mBidtable[$v] = $rules;
1357                                 }
1358                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1359                                 $this->mFlags = $flags = array( 'R' => true );
1360                         }
1361                 }
1362
1363                 $this->mRuleDisplay = false;
1364                 foreach ( $flags as $flag => $unused ) {
1365                         switch ( $flag ) {
1366                                 case 'R':
1367                                         // if we don't do content convert, still strip the -{}- tags
1368                                         $this->mRuleDisplay = $rules;
1369                                         break;
1370                                 case 'N':
1371                                         // process N flag: output current variant name
1372                                         $ruleVar = trim( $rules );
1373                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1374                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1375                                         } else {
1376                                                 $this->mRuleDisplay = '';
1377                                         }
1378                                         break;
1379                                 case 'D':
1380                                         // process D flag: output rules description
1381                                         $this->mRuleDisplay = $this->getRulesDesc();
1382                                         break;
1383                                 case 'H':
1384                                         // process H,- flag or T only: output nothing
1385                                         $this->mRuleDisplay = '';
1386                                         break;
1387                                 case '-':
1388                                         $this->mRulesAction = 'remove';
1389                                         $this->mRuleDisplay = '';
1390                                         break;
1391                                 case '+':
1392                                         $this->mRulesAction = 'add';
1393                                         $this->mRuleDisplay = '';
1394                                         break;
1395                                 case 'S':
1396                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1397                                         break;
1398                                 case 'T':
1399                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1400                                         $this->mRuleDisplay = '';
1401                                         break;
1402                                 default:
1403                                         // ignore unknown flags (but see error case below)
1404                         }
1405                 }
1406                 if ( $this->mRuleDisplay === false ) {
1407                         $this->mRuleDisplay = $this->mManualCodeError;
1408                 }
1409
1410                 $this->generateConvTable();
1411         }
1412
1413         /**
1414          * @public
1415          */
1416         function hasRules() {
1417                 // TODO:
1418         }
1419
1420         /**
1421          * Get display text on markup -{...}-
1422          * @public
1423          */
1424         function getDisplay() {
1425                 return $this->mRuleDisplay;
1426         }
1427
1428         /**
1429          * Get converted title.
1430          * @public
1431          */
1432         function getTitle() {
1433                 return $this->mRuleTitle;
1434         }
1435
1436         /**
1437          * Return how deal with conversion rules.
1438          * @public
1439          */
1440         function getRulesAction() {
1441                 return $this->mRulesAction;
1442         }
1443
1444         /**
1445          * Get conversion table. ( bidirectional and unidirectional
1446          * conversion table )
1447          * @public
1448          */
1449         function getConvTable() {
1450                 return $this->mConvTable;
1451         }
1452
1453         /**
1454          * Get conversion rules string.
1455          * @public
1456          */
1457         function getRules() {
1458                 return $this->mRules;
1459         }
1460
1461         /**
1462          * Get conversion flags.
1463          * @public
1464          */
1465         function getFlags() {
1466                 return $this->mFlags;
1467         }
1468 }