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 Language 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          * Dictionary-based conversion.
 304          * This function would not parse the conversion rules.
 305          * If you want to parse rules, try to use convert() or
 306          * convertTo().
 307          *
 308          * @param $text String: the text to be converted
 309          * @param $toVariant String: the target language code
 310          * @return String: the converted text
 311          */
 312         public function autoConvert( $text, $toVariant = false ) {
 313                 wfProfileIn( __METHOD__ );
 314
 315                 $this->loadTables();
 316
 317                 if ( !$toVariant ) {
 318                         $toVariant = $this->getPreferredVariant();
 319                         if ( !$toVariant ) {
 320                                 wfProfileOut( __METHOD__ );
 321                                 return $text;
 322                         }
 323                 }
 324
 325                 /* we convert everything except:
 326                    1. html markups (anything between < and >)
 327                    2. html entities
 328                    3. place holders created by the parser
 329                 */
 330                 global $wgParser;
 331                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 332                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 333                 } else {
 334                         $marker = '';
 335                 }
 336
 337                 // this one is needed when the text is inside an html markup
 338                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 339
 340                 // disable convert to variants between <code></code> tags
 341                 $codefix = '<code>.+?<\/code>|';
 342                 // disable convertsion of <script type="text/javascript"> ... </script>
 343                 $scriptfix = '<script.*?>.*?<\/script>|';
 344                 // disable conversion of <pre xxxx> ... </pre>
 345                 $prefix = '<pre.*?>.*?<\/pre>|';
 346
 347                 $reg = '/' . $codefix . $scriptfix . $prefix .
 348                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 349                 $startPos = 0;
 350                 $sourceBlob = '';
 351                 $literalBlob = '';
 352
 353                 // Guard against delimiter nulls in the input
 354                 $text = str_replace( "\000", '', $text );
 355
 356                 $markupMatches = null;
 357                 $elementMatches = null;
 358                 while ( $startPos < strlen( $text ) ) {
 359                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 360                                 $elementPos = $markupMatches[0][1];
 361                                 $element = $markupMatches[0][0];
 362                         } else {
 363                                 $elementPos = strlen( $text );
 364                                 $element = '';
 365                         }
 366
 367                         // Queue the part before the markup for translation in a batch
 368                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 369
 370                         // Advance to the next position
 371                         $startPos = $elementPos + strlen( $element );
 372
 373                         // Translate any alt or title attributes inside the matched element
 374                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 375                                 $elementMatches  ) )
 376                         {
 377                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 378                                 $changed = false;
 379                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 380                                         if ( !isset( $attrs[$attrName] ) ) {
 381                                                 continue;
 382                                         }
 383                                         $attr = $attrs[$attrName];
 384                                         // Don't convert URLs
 385                                         if ( !strpos( $attr, '://' ) ) {
 386                                                 $attr = $this->translate( $attr, $toVariant );
 387                                         }
 388
 389                                         // Remove HTML tags to avoid disrupting the layout
 390                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 391                                         if ( $attr !== $attrs[$attrName] ) {
 392                                                 $attrs[$attrName] = $attr;
 393                                                 $changed = true;
 394                                         }
 395                                 }
 396                                 if ( $changed ) {
 397                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 398                                                 $elementMatches[3];
 399                                 }
 400                         }
 401                         $literalBlob .= $element . "\000";
 402                 }
 403
 404                 // Do the main translation batch
 405                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 406
 407                 // Put the output back together
 408                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 409                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 410                 $output = '';
 411                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 412                         $output .= $translatedIter->current();
 413                         $output .= $literalIter->current();
 414                         $translatedIter->next();
 415                         $literalIter->next();
 416                 }
 417
 418                 wfProfileOut( __METHOD__ );
 419                 return $output;
 420         }
 421
 422         /**
 423          * Translate a string to a variant.
 424          * Doesn't parse rules or do any of that other stuff, for that use
 425          * convert() or convertTo().
 426          *
 427          * @param $text String: text to convert
 428          * @param $variant String: variant language code
 429          * @return String: translated text
 430          */
 431         public function translate( $text, $variant ) {
 432                 wfProfileIn( __METHOD__ );
 433                 // If $text is empty or only includes spaces, do nothing
 434                 // Otherwise translate it
 435                 if ( trim( $text ) ) {
 436                         $this->loadTables();
 437                         $text = $this->mTables[$variant]->replace( $text );
 438                 }
 439                 wfProfileOut( __METHOD__ );
 440                 return $text;
 441         }
 442
 443         /**
 444          * Call translate() to convert text to all valid variants.
 445          *
 446          * @param $text String: the text to be converted
 447          * @return Array: variant => converted text
 448          */
 449         public function autoConvertToAllVariants( $text ) {
 450                 wfProfileIn( __METHOD__ );
 451                 $this->loadTables();
 452
 453                 $ret = array();
 454                 foreach ( $this->mVariants as $variant ) {
 455                         $ret[$variant] = $this->translate( $text, $variant );
 456                 }
 457
 458                 wfProfileOut( __METHOD__ );
 459                 return $ret;
 460         }
 461
 462         /**
 463          * Convert link text to all valid variants.
 464          * In the first, this function only convert text outside the
 465          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 466          * titles, the text will get all converted always.
 467          * So I removed this feature and deprecated the function.
 468          *
 469          * @param $text String: the text to be converted
 470          * @return Array: variant => converted text
 471          * @deprecated Use autoConvertToAllVariants() instead
 472          */
 473         public function convertLinkToAllVariants( $text ) {
 474                 return $this->autoConvertToAllVariants( $text );
 475         }
 476
 477         /**
 478          * Apply manual conversion rules.
 479          *
 480          * @param $convRule Object: Object of ConverterRule
 481          */
 482         protected function applyManualConv( $convRule ) {
 483                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 484                 // title conversion.
 485                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 486                 // rule(s) not for title, this breaks the title conversion.
 487                 $newConvRuleTitle = $convRule->getTitle();
 488                 if ( $newConvRuleTitle ) {
 489                         // So I add an empty check for getTitle()
 490                         $this->mConvRuleTitle = $newConvRuleTitle;
 491                 }
 492
 493                 // merge/remove manual conversion rules to/from global table
 494                 $convTable = $convRule->getConvTable();
 495                 $action = $convRule->getRulesAction();
 496                 foreach ( $convTable as $variant => $pair ) {
 497                         if ( !$this->validateVariant( $variant ) ) {
 498                                 continue;
 499                         }
 500
 501                         if ( $action == 'add' ) {
 502                                 foreach ( $pair as $from => $to ) {
 503                                         // to ensure that $from and $to not be left blank
 504                                         // so $this->translate() could always return a string
 505                                         if ( $from || $to ) {
 506                                                 // more efficient than array_merge(), about 2.5 times.
 507                                                 $this->mTables[$variant]->setPair( $from, $to );
 508                                         }
 509                                 }
 510                         } elseif ( $action == 'remove' ) {
 511                                 $this->mTables[$variant]->removeArray( $pair );
 512                         }
 513                 }
 514         }
 515
 516         /**
 517          * Auto convert a Title object to a readable string in the
 518          * preferred variant.
 519          *
 520          *@param $title Object: a object of Title
 521          *@return String: converted title text
 522          */
 523         public function convertTitle( $title ) {
 524                 $variant = $this->getPreferredVariant();
 525                 $index = $title->getNamespace();
 526                 if ( $index === NS_MAIN ) {
 527                         $text = '';
 528                 } else {
 529                         // first let's check if a message has given us a converted name
 530                         $nsConvKey = 'conversion-ns' . $index;
 531                         if ( !wfEmptyMsg( $nsConvKey ) ) {
 532                                 $text = wfMsgForContentNoTrans( $nsConvKey );
 533                         } else {
 534                                 // the message does not exist, try retrieve it from the current
 535                                 // variant's namespace names.
 536                                 $langObj = $this->mLangObj->factory( $variant );
 537                                 $text = $langObj->getFormattedNsText( $index );
 538                         }
 539                         $text .= ':';
 540                 }
 541                 $text .= $title->getText();
 542                 $text = $this->translate( $text, $variant );
 543                 return $text;
 544         }
 545
 546         /**
 547          * Convert text to different variants of a language. The automatic
 548          * conversion is done in autoConvert(). Here we parse the text
 549          * marked with -{}-, which specifies special conversions of the
 550          * text that can not be accomplished in autoConvert().
 551          *
 552          * Syntax of the markup:
 553          * -{code1:text1;code2:text2;...}-  or
 554          * -{flags|code1:text1;code2:text2;...}-  or
 555          * -{text}- in which case no conversion should take place for text
 556          *
 557          * @param $text String: text to be converted
 558          * @return String: converted text
 559          */
 560         public function convert( $text ) {
 561                 $variant = $this->getPreferredVariant();
 562                 return $this->convertTo( $text, $variant );
 563         }
 564
 565         /**
 566          * Same as convert() except a extra parameter to custom variant.
 567          *
 568          * @param $text String: text to be converted
 569          * @param $variant String: the target variant code
 570          * @return String: converted text
 571          */
 572         public function convertTo( $text, $variant ) {
 573                 global $wgDisableLangConversion;
 574                 if ( $wgDisableLangConversion ) return $text;
 575                 return $this->recursiveConvertTopLevel( $text, $variant );
 576         }
 577
 578         /**
 579          * Recursively convert text on the outside. Allow to use nested
 580          * markups to custom rules.
 581          *
 582          * @param $text String: text to be converted
 583          * @param $variant String: the target variant code
 584          * @param $depth Integer: depth of recursion
 585          * @return String: converted text
 586          */
 587         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 588                 $startPos = 0;
 589                 $out = '';
 590                 $length = strlen( $text );
 591                 while ( $startPos < $length ) {
 592                         $pos = strpos( $text, '-{', $startPos );
 593
 594                         if ( $pos === false ) {
 595                                 // No more markup, append final segment
 596                                 $out .= $this->autoConvert( substr( $text, $startPos ), $variant );
 597                                 return $out;
 598                         }
 599
 600                         // Markup found
 601                         // Append initial segment
 602                         $out .= $this->autoConvert( substr( $text, $startPos, $pos - $startPos ), $variant );
 603
 604                         // Advance position
 605                         $startPos = $pos;
 606
 607                         // Do recursive conversion
 608                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 609                 }
 610
 611                 return $out;
 612         }
 613
 614         /**
 615          * Recursively convert text on the inside.
 616          *
 617          * @param $text String: text to be converted
 618          * @param $variant String: the target variant code
 619          * @param $depth Integer: depth of recursion
 620          * @return String: converted text
 621          */
 622         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 623                 // Quick sanity check (no function calls)
 624                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 625                         throw new MWException( __METHOD__ . ': invalid input string' );
 626                 }
 627
 628                 $startPos += 2;
 629                 $inner = '';
 630                 $warningDone = false;
 631                 $length = strlen( $text );
 632
 633                 while ( $startPos < $length ) {
 634                         $m = false;
 635                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 636                         if ( !$m ) {
 637                                 // Unclosed rule
 638                                 break;
 639                         }
 640
 641                         $token = $m[0][0];
 642                         $pos = $m[0][1];
 643
 644                         // Markup found
 645                         // Append initial segment
 646                         $inner .= substr( $text, $startPos, $pos - $startPos );
 647
 648                         // Advance position
 649                         $startPos = $pos;
 650
 651                         switch ( $token ) {
 652                                 case '-{':
 653                                         // Check max depth
 654                                         if ( $depth >= $this->mMaxDepth ) {
 655                                                 $inner .= '-{';
 656                                                 if ( !$warningDone ) {
 657                                                         $inner .= '<span class="error">' .
 658                                                                 wfMsgForContent( 'language-converter-depth-warning',
 659                                                                         $this->mMaxDepth ) .
 660                                                                 '</span>';
 661                                                         $warningDone = true;
 662                                                 }
 663                                                 $startPos += 2;
 664                                                 continue;
 665                                         }
 666                                         // Recursively parse another rule
 667                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 668                                         break;
 669                                 case '}-':
 670                                         // Apply the rule
 671                                         $startPos += 2;
 672                                         $rule = new ConverterRule( $inner, $this );
 673                                         $rule->parse( $variant );
 674                                         $this->applyManualConv( $rule );
 675                                         return $rule->getDisplay();
 676                                 default:
 677                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 678                         }
 679                 }
 680
 681                 // Unclosed rule
 682                 if ( $startPos < $length ) {
 683                         $inner .= substr( $text, $startPos );
 684                 }
 685                 $startPos = $length;
 686                 return '-{' . $this->autoConvert( $inner, $variant );
 687         }
 688
 689         /**
 690          * If a language supports multiple variants, it is
 691          * possible that non-existing link in one variant
 692          * actually exists in another variant. This function
 693          * tries to find it. See e.g. LanguageZh.php
 694          *
 695          * @param $link String: the name of the link
 696          * @param $nt Mixed: the title object of the link
 697          * @param $ignoreOtherCond Boolean: to disable other conditions when
 698          *              we need to transclude a template or update a category's link
 699          * @return Null, the input parameters may be modified upon return
 700          */
 701         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 702                 # If the article has already existed, there is no need to
 703                 # check it again, otherwise it may cause a fault.
 704                 if ( is_object( $nt ) && $nt->exists() ) {
 705                         return;
 706                 }
 707
 708                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 709                         $wgUser;
 710                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 711                 $action = $wgRequest->getText( 'action' );
 712                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 713                 $disableLinkConversion = $wgDisableLangConversion
 714                         || $wgDisableTitleConversion;
 715                 $linkBatch = new LinkBatch();
 716
 717                 $ns = NS_MAIN;
 718
 719                 if ( $disableLinkConversion ||
 720                          ( !$ignoreOtherCond &&
 721                            ( $isredir == 'no'
 722                                  || $action == 'edit'
 723                                  || $action == 'submit'
 724                                  || $linkconvert == 'no'
 725                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 726                         return;
 727                 }
 728
 729                 if ( is_object( $nt ) ) {
 730                         $ns = $nt->getNamespace();
 731                 }
 732
 733                 $variants = $this->autoConvertToAllVariants( $link );
 734                 if ( !$variants ) { // give up
 735                         return;
 736                 }
 737
 738                 $titles = array();
 739
 740                 foreach ( $variants as $v ) {
 741                         if ( $v != $link ) {
 742                                 $varnt = Title::newFromText( $v, $ns );
 743                                 if ( !is_null( $varnt ) ) {
 744                                         $linkBatch->addObj( $varnt );
 745                                         $titles[] = $varnt;
 746                                 }
 747                         }
 748                 }
 749
 750                 // fetch all variants in single query
 751                 $linkBatch->execute();
 752
 753                 foreach ( $titles as $varnt ) {
 754                         if ( $varnt->getArticleID() > 0 ) {
 755                                 $nt = $varnt;
 756                                 $link = $varnt->getText();
 757                                 break;
 758                         }
 759                 }
 760         }
 761
 762         /**
 763          * Returns language specific hash options.
 764          */
 765         public function getExtraHashOptions() {
 766                 $variant = $this->getPreferredVariant();
 767                 return '!' . $variant ;
 768         }
 769
 770         /**
 771          * Load default conversion tables.
 772          * This method must be implemented in derived class.
 773          *
 774          * @private
 775          */
 776         function loadDefaultTables() {
 777                 $name = get_class( $this );
 778                 wfDie( "Must implement loadDefaultTables() method in class $name" );
 779         }
 780
 781         /**
 782          * Load conversion tables either from the cache or the disk.
 783          * @private
 784          */
 785         function loadTables( $fromcache = true ) {
 786                 if ( $this->mTablesLoaded ) {
 787                         return;
 788                 }
 789                 global $wgMemc;
 790                 wfProfileIn( __METHOD__ );
 791                 $this->mTablesLoaded = true;
 792                 $this->mTables = false;
 793                 if ( $fromcache ) {
 794                         wfProfileIn( __METHOD__ . '-cache' );
 795                         $this->mTables = $wgMemc->get( $this->mCacheKey );
 796                         wfProfileOut( __METHOD__ . '-cache' );
 797                 }
 798                 if ( !$this->mTables
 799                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 800                         wfProfileIn( __METHOD__ . '-recache' );
 801                         // not in cache, or we need a fresh reload.
 802                         // we will first load the default tables
 803                         // then update them using things in MediaWiki:Zhconversiontable/*
 804                         $this->loadDefaultTables();
 805                         foreach ( $this->mVariants as $var ) {
 806                                 $cached = $this->parseCachedTable( $var );
 807                                 $this->mTables[$var]->mergeArray( $cached );
 808                         }
 809
 810                         $this->postLoadTables();
 811                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 812
 813                         $wgMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 814                         wfProfileOut( __METHOD__ . '-recache' );
 815                 }
 816                 wfProfileOut( __METHOD__ );
 817         }
 818
 819         /**
 820          * Hook for post processig after conversion tables are loaded.
 821          *
 822          */
 823         function postLoadTables() { }
 824
 825         /**
 826          * Reload the conversion tables.
 827          *
 828          * @private
 829          */
 830         function reloadTables() {
 831                 if ( $this->mTables ) {
 832                         unset( $this->mTables );
 833                 }
 834                 $this->mTablesLoaded = false;
 835                 $this->loadTables( false );
 836         }
 837
 838
 839         /**
 840          * Parse the conversion table stored in the cache.
 841          *
 842          * The tables should be in blocks of the following form:
 843          *              -{
 844          *                      word => word ;
 845          *                      word => word ;
 846          *                      ...
 847          *              }-
 848          *
 849          *      To make the tables more manageable, subpages are allowed
 850          *      and will be parsed recursively if $recursive == true.
 851          *
 852          */
 853         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 854                 static $parsed = array();
 855
 856                 $key = 'Conversiontable/' . $code;
 857                 if ( $subpage ) {
 858                         $key .= '/' . $subpage;
 859                 }
 860                 if ( array_key_exists( $key, $parsed ) ) {
 861                         return array();
 862                 }
 863
 864                 if ( strpos( $code, '/' ) === false ) {
 865                         $txt = MessageCache::singleton()->get( 'Conversiontable', true, $code );
 866                         if ( $txt === false ) {
 867                                 # FIXME: this method doesn't seem to be expecting
 868                                 # this possible outcome...
 869                                 $txt = '&lt;Conversiontable&gt;';
 870                         }
 871                 } else {
 872                         $title = Title::makeTitleSafe( NS_MEDIAWIKI,
 873                                                                                    "Conversiontable/$code" );
 874                         if ( $title && $title->exists() ) {
 875                                 $article = new Article( $title );
 876                                 $txt = $article->getContents();
 877                         } else {
 878                                 $txt = '';
 879                         }
 880                 }
 881
 882                 // get all subpage links of the form
 883                 // [[MediaWiki:conversiontable/zh-xx/...|...]]
 884                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 885                         ':Conversiontable';
 886                 $subs = StringUtils::explode( '[[', $txt );
 887                 $sublinks = array();
 888                 foreach ( $subs as $sub ) {
 889                         $link = explode( ']]', $sub, 2 );
 890                         if ( count( $link ) != 2 ) {
 891                                 continue;
 892                         }
 893                         $b = explode( '|', $link[0], 2 );
 894                         $b = explode( '/', trim( $b[0] ), 3 );
 895                         if ( count( $b ) == 3 ) {
 896                                 $sublink = $b[2];
 897                         } else {
 898                                 $sublink = '';
 899                         }
 900
 901                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 902                                 $sublinks[] = $sublink;
 903                         }
 904                 }
 905
 906                 // parse the mappings in this page
 907                 $blocks = StringUtils::explode( '-{', $txt );
 908                 $ret = array();
 909                 $first = true;
 910                 foreach ( $blocks as $block ) {
 911                         if ( $first ) {
 912                                 // Skip the part before the first -{
 913                                 $first = false;
 914                                 continue;
 915                         }
 916                         $mappings = explode( '}-', $block, 2 );
 917                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 918                                                                          $mappings[0] );
 919                         $table = StringUtils::explode( ';', $stripped );
 920                         foreach ( $table as $t ) {
 921                                 $m = explode( '=>', $t, 3 );
 922                                 if ( count( $m ) != 2 )
 923                                         continue;
 924                                 // trim any trailling comments starting with '//'
 925                                 $tt = explode( '//', $m[1], 2 );
 926                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 927                         }
 928                 }
 929                 $parsed[$key] = true;
 930
 931                 // recursively parse the subpages
 932                 if ( $recursive ) {
 933                         foreach ( $sublinks as $link ) {
 934                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 935                                 $ret = array_merge( $ret, $s );
 936                         }
 937                 }
 938
 939                 if ( $this->mUcfirst ) {
 940                         foreach ( $ret as $k => $v ) {
 941                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 942                         }
 943                 }
 944                 return $ret;
 945         }
 946
 947         /**
 948          * Enclose a string with the "no conversion" tag. This is used by
 949          * various functions in the Parser.
 950          *
 951          * @param $text String: text to be tagged for no conversion
 952          * @param $noParse Unused (?)
 953          * @return String: the tagged text
 954          */
 955         public function markNoConversion( $text, $noParse = false ) {
 956                 # don't mark if already marked
 957                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
 958                         return $text;
 959                 }
 960
 961                 $ret = "-{R|$text}-";
 962                 return $ret;
 963         }
 964
 965         /**
 966          * Convert the sorting key for category links. This should make different
 967          * keys that are variants of each other map to the same key.
 968          */
 969         function convertCategoryKey( $key ) {
 970                 return $key;
 971         }
 972
 973         /**
 974          * Hook to refresh the cache of conversion tables when
 975          * MediaWiki:conversiontable* is updated.
 976          * @private
 977          */
 978         function OnArticleSaveComplete( $article, $user, $text, $summary, $isminor,
 979                         $iswatch, $section, $flags, $revision ) {
 980                 $titleobj = $article->getTitle();
 981                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
 982                         $title = $titleobj->getDBkey();
 983                         $t = explode( '/', $title, 3 );
 984                         $c = count( $t );
 985                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
 986                                 if ( $this->validateVariant( $t[1] ) ) {
 987                                         $this->reloadTables();
 988                                 }
 989                         }
 990                 }
 991                 return true;
 992         }
 993
 994         /**
 995          * Armour rendered math against conversion.
 996          * Escape special chars in parsed math text.(in most cases are img elements)
 997          */
 998         public function armourMath( $text ) {
 999                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1000                 // any unwanted markup appearing in the math image tag.
1001                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1002                 return $text;
1003         }
1004
1005         /**
1006          * Get the cached separator pattern for ConverterRule::parseRules()
1007          */
1008         function getVarSeparatorPattern() {
1009                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1010                         // varsep_pattern for preg_split:
1011                         // text should be splited by ";" only if a valid variant
1012                         // name exist after the markup, for example:
1013                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1014                         //      <span style="font-size:120%;">yyy</span>;}-
1015                         // we should split it as:
1016                         //  array(
1017                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1018                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1019                         //        [2] => ''
1020                         //       )
1021                         $pat = '/;\s*(?=';
1022                         foreach ( $this->mVariants as $variant ) {
1023                                 // zh-hans:xxx;zh-hant:yyy
1024                                 $pat .= $variant . '\s*:|';
1025                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1026                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1027                         }
1028                         $pat .= '\s*$)/';
1029                         $this->mVarSeparatorPattern = $pat;
1030                 }
1031                 return $this->mVarSeparatorPattern;
1032         }
1033 }
1034
1035 /**
1036  * Parser for rules of language conversion , parse rules in -{ }- tag.
1037  * @ingroup Language
1038  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1039  */
1040 class ConverterRule {
1041         var $mText; // original text in -{text}-
1042         var $mConverter; // LanguageConverter object
1043         var $mManualCodeError = '<strong class="error">code error!</strong>';
1044         var $mRuleDisplay = '';
1045         var $mRuleTitle = false;
1046         var $mRules = '';// string : the text of the rules
1047         var $mRulesAction = 'none';
1048         var $mFlags = array();
1049         var $mVariantFlags = array();
1050         var $mConvTable = array();
1051         var $mBidtable = array();// array of the translation in each variant
1052         var $mUnidtable = array();// array of the translation in each variant
1053
1054         /**
1055          * Constructor
1056          *
1057          * @param $text String: the text between -{ and }-
1058          * @param $converter LanguageConverter object
1059          */
1060         public function __construct( $text, $converter ) {
1061                 $this->mText = $text;
1062                 $this->mConverter = $converter;
1063         }
1064
1065         /**
1066          * Check if variants array in convert array.
1067          *
1068          * @param $variants Array or string: variant language code
1069          * @return String: translated text
1070          */
1071         public function getTextInBidtable( $variants ) {
1072                 $variants = (array)$variants;
1073                 if ( !$variants ) {
1074                         return false;
1075                 }
1076                 foreach ( $variants as $variant ) {
1077                         if ( isset( $this->mBidtable[$variant] ) ) {
1078                                 return $this->mBidtable[$variant];
1079                         }
1080                 }
1081                 return false;
1082         }
1083
1084         /**
1085          * Parse flags with syntax -{FLAG| ... }-
1086          * @private
1087          */
1088         function parseFlags() {
1089                 $text = $this->mText;
1090                 $flags = array();
1091                 $variantFlags = array();
1092
1093                 $sepPos = strpos( $text, '|' );
1094                 if ( $sepPos !== false ) {
1095                         $validFlags = $this->mConverter->mFlags;
1096                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1097                         foreach ( $f as $ff ) {
1098                                 $ff = trim( $ff );
1099                                 if ( isset( $validFlags[$ff] ) ) {
1100                                         $flags[$validFlags[$ff]] = true;
1101                                 }
1102                         }
1103                         $text = strval( substr( $text, $sepPos + 1 ) );
1104                 }
1105
1106                 if ( !$flags ) {
1107                         $flags['S'] = true;
1108                 } elseif ( isset( $flags['R'] ) ) {
1109                         $flags = array( 'R' => true );// remove other flags
1110                 } elseif ( isset( $flags['N'] ) ) {
1111                         $flags = array( 'N' => true );// remove other flags
1112                 } elseif ( isset( $flags['-'] ) ) {
1113                         $flags = array( '-' => true );// remove other flags
1114                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1115                         $flags['H'] = true;
1116                 } elseif ( isset( $flags['H'] ) ) {
1117                         // replace A flag, and remove other flags except T
1118                         $temp = array( '+' => true, 'H' => true );
1119                         if ( isset( $flags['T'] ) ) {
1120                                 $temp['T'] = true;
1121                         }
1122                         if ( isset( $flags['D'] ) ) {
1123                                 $temp['D'] = true;
1124                         }
1125                         $flags = $temp;
1126                 } else {
1127                         if ( isset( $flags['A'] ) ) {
1128                                 $flags['+'] = true;
1129                                 $flags['S'] = true;
1130                         }
1131                         if ( isset( $flags['D'] ) ) {
1132                                 unset( $flags['S'] );
1133                         }
1134                         // try to find flags like "zh-hans", "zh-hant"
1135                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1136                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1137                         if ( $variantFlags ) {
1138                                 $variantFlags = array_flip( $variantFlags );
1139                                 $flags = array();
1140                         }
1141                 }
1142                 $this->mVariantFlags = $variantFlags;
1143                 $this->mRules = $text;
1144                 $this->mFlags = $flags;
1145         }
1146
1147         /**
1148          * Generate conversion table.
1149          * @private
1150          */
1151         function parseRules() {
1152                 $rules = $this->mRules;
1153                 $bidtable = array();
1154                 $unidtable = array();
1155                 $variants = $this->mConverter->mVariants;
1156                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1157
1158                 $choice = preg_split( $varsep_pattern, $rules );
1159
1160                 foreach ( $choice as $c ) {
1161                         $v  = explode( ':', $c, 2 );
1162                         if ( count( $v ) != 2 ) {
1163                                 // syntax error, skip
1164                                 continue;
1165                         }
1166                         $to = trim( $v[1] );
1167                         $v  = trim( $v[0] );
1168                         $u  = explode( '=>', $v, 2 );
1169                         // if $to is empty, strtr() could return a wrong result
1170                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1171                                 $bidtable[$v] = $to;
1172                         } elseif ( count( $u ) == 2 ) {
1173                                 $from = trim( $u[0] );
1174                                 $v      = trim( $u[1] );
1175                                 if ( array_key_exists( $v, $unidtable )
1176                                          && !is_array( $unidtable[$v] )
1177                                          && $to
1178                                          && in_array( $v, $variants ) ) {
1179                                         $unidtable[$v] = array( $from => $to );
1180                                 } elseif ( $to && in_array( $v, $variants ) ) {
1181                                         $unidtable[$v][$from] = $to;
1182                                 }
1183                         }
1184                         // syntax error, pass
1185                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1186                                 $bidtable = array();
1187                                 $unidtable = array();
1188                                 break;
1189                         }
1190                 }
1191                 $this->mBidtable = $bidtable;
1192                 $this->mUnidtable = $unidtable;
1193         }
1194
1195         /**
1196          * @private
1197          */
1198         function getRulesDesc() {
1199                 $codesep = $this->mConverter->mDescCodeSep;
1200                 $varsep = $this->mConverter->mDescVarSep;
1201                 $text = '';
1202                 foreach ( $this->mBidtable as $k => $v ) {
1203                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1204                 }
1205                 foreach ( $this->mUnidtable as $k => $a ) {
1206                         foreach ( $a as $from => $to ) {
1207                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1208                                         "$codesep$to$varsep";
1209                         }
1210                 }
1211                 return $text;
1212         }
1213
1214         /**
1215          * Parse rules conversion.
1216          * @private
1217          */
1218         function getRuleConvertedStr( $variant ) {
1219                 $bidtable = $this->mBidtable;
1220                 $unidtable = $this->mUnidtable;
1221
1222                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1223                         return $this->mRules;
1224                 } else {
1225                         // display current variant in bidirectional array
1226                         $disp = $this->getTextInBidtable( $variant );
1227                         // or display current variant in fallbacks
1228                         if ( !$disp ) {
1229                                 $disp = $this->getTextInBidtable(
1230                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1231                         }
1232                         // or display current variant in unidirectional array
1233                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1234                                 $disp = array_values( $unidtable[$variant] );
1235                                 $disp = $disp[0];
1236                         }
1237                         // or display frist text under disable manual convert
1238                         if ( !$disp
1239                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1240                                 if ( count( $bidtable ) > 0 ) {
1241                                         $disp = array_values( $bidtable );
1242                                         $disp = $disp[0];
1243                                 } else {
1244                                         $disp = array_values( $unidtable );
1245                                         $disp = array_values( $disp[0] );
1246                                         $disp = $disp[0];
1247                                 }
1248                         }
1249                         return $disp;
1250                 }
1251         }
1252
1253         /**
1254          * Generate conversion table for all text.
1255          * @private
1256          */
1257         function generateConvTable() {
1258                 // Special case optimisation
1259                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1260                         $this->mConvTable = array();
1261                         return;
1262                 }
1263
1264                 $bidtable = $this->mBidtable;
1265                 $unidtable = $this->mUnidtable;
1266                 $manLevel = $this->mConverter->mManualLevel;
1267
1268                 $vmarked = array();
1269                 foreach ( $this->mConverter->mVariants as $v ) {
1270                         /* for bidirectional array
1271                                 fill in the missing variants, if any,
1272                                 with fallbacks */
1273                         if ( !isset( $bidtable[$v] ) ) {
1274                                 $variantFallbacks =
1275                                         $this->mConverter->getVariantFallbacks( $v );
1276                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1277                                 if ( $vf ) {
1278                                         $bidtable[$v] = $vf;
1279                                 }
1280                         }
1281
1282                         if ( isset( $bidtable[$v] ) ) {
1283                                 foreach ( $vmarked as $vo ) {
1284                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1285                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1286                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1287                                         // to introduce a custom mapping between
1288                                         // words WordZh and WordTw in the whole text
1289                                         if ( $manLevel[$v] == 'bidirectional' ) {
1290                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1291                                         }
1292                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1293                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1294                                         }
1295                                 }
1296                                 $vmarked[] = $v;
1297                         }
1298                         /*for unidirectional array fill to convert tables */
1299                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1300                                 && isset( $unidtable[$v] ) )
1301                         {
1302                                 if ( isset( $this->mConvTable[$v] ) ) {
1303                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1304                                 } else {
1305                                         $this->mConvTable[$v] = $unidtable[$v];
1306                                 }
1307                         }
1308                 }
1309         }
1310
1311         /**
1312          * Parse rules and flags.
1313          * @public
1314          */
1315         function parse( $variant = NULL ) {
1316                 if ( !$variant ) {
1317                         $variant = $this->mConverter->getPreferredVariant();
1318                 }
1319
1320                 $this->parseFlags();
1321                 $flags = $this->mFlags;
1322
1323                 // convert to specified variant
1324                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1325                 if ( $this->mVariantFlags ) {
1326                         // check if current variant in flags
1327                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1328                                 // then convert <text to convert> to current language
1329                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1330                                                                                                                                 $variant );
1331                         } else { // if current variant no in flags,
1332                                    // then we check its fallback variants.
1333                                 $variantFallbacks =
1334                                         $this->mConverter->getVariantFallbacks( $variant );
1335                                 foreach ( $variantFallbacks as $variantFallback ) {
1336                                         // if current variant's fallback exist in flags
1337                                         if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1338                                                 // then convert <text to convert> to fallback language
1339                                                 $this->mRules =
1340                                                         $this->mConverter->autoConvert( $this->mRules,
1341                                                                                                                         $variantFallback );
1342                                                 break;
1343                                         }
1344                                 }
1345                         }
1346                         $this->mFlags = $flags = array( 'R' => true );
1347                 }
1348
1349                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1350                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1351                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1352                         $this->parseRules();
1353                 }
1354                 $rules = $this->mRules;
1355
1356                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1357                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1358                                 // fill all variants if text in -{A/H/-|text} without rules
1359                                 foreach ( $this->mConverter->mVariants as $v ) {
1360                                         $this->mBidtable[$v] = $rules;
1361                                 }
1362                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1363                                 $this->mFlags = $flags = array( 'R' => true );
1364                         }
1365                 }
1366
1367                 $this->mRuleDisplay = false;
1368                 foreach ( $flags as $flag => $unused ) {
1369                         switch ( $flag ) {
1370                                 case 'R':
1371                                         // if we don't do content convert, still strip the -{}- tags
1372                                         $this->mRuleDisplay = $rules;
1373                                         break;
1374                                 case 'N':
1375                                         // process N flag: output current variant name
1376                                         $ruleVar = trim( $rules );
1377                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1378                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1379                                         } else {
1380                                                 $this->mRuleDisplay = '';
1381                                         }
1382                                         break;
1383                                 case 'D':
1384                                         // process D flag: output rules description
1385                                         $this->mRuleDisplay = $this->getRulesDesc();
1386                                         break;
1387                                 case 'H':
1388                                         // process H,- flag or T only: output nothing
1389                                         $this->mRuleDisplay = '';
1390                                         break;
1391                                 case '-':
1392                                         $this->mRulesAction = 'remove';
1393                                         $this->mRuleDisplay = '';
1394                                         break;
1395                                 case '+':
1396                                         $this->mRulesAction = 'add';
1397                                         $this->mRuleDisplay = '';
1398                                         break;
1399                                 case 'S':
1400                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1401                                         break;
1402                                 case 'T':
1403                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1404                                         $this->mRuleDisplay = '';
1405                                         break;
1406                                 default:
1407                                         // ignore unknown flags (but see error case below)
1408                         }
1409                 }
1410                 if ( $this->mRuleDisplay === false ) {
1411                         $this->mRuleDisplay = $this->mManualCodeError;
1412                 }
1413
1414                 $this->generateConvTable();
1415         }
1416
1417         /**
1418          * @public
1419          */
1420         function hasRules() {
1421                 // TODO:
1422         }
1423
1424         /**
1425          * Get display text on markup -{...}-
1426          * @public
1427          */
1428         function getDisplay() {
1429                 return $this->mRuleDisplay;
1430         }
1431
1432         /**
1433          * Get converted title.
1434          * @public
1435          */
1436         function getTitle() {
1437                 return $this->mRuleTitle;
1438         }
1439
1440         /**
1441          * Return how deal with conversion rules.
1442          * @public
1443          */
1444         function getRulesAction() {
1445                 return $this->mRulesAction;
1446         }
1447
1448         /**
1449          * Get conversion table. ( bidirectional and unidirectional
1450          * conversion table )
1451          * @public
1452          */
1453         function getConvTable() {
1454                 return $this->mConvTable;
1455         }
1456
1457         /**
1458          * Get conversion rules string.
1459          * @public
1460          */
1461         function getRules() {
1462                 return $this->mRules;
1463         }
1464
1465         /**
1466          * Get conversion flags.
1467          * @public
1468          */
1469         function getFlags() {
1470                 return $this->mFlags;
1471         }
1472 }