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
  33         /**
  34          * languages supporting variants
  35          * @since 1.20
  36          * @var array
  37          */
  38         static public $languagesWithVariants = array(
  39                 'gan',
  40                 'iu',
  41                 'kk',
  42                 'ku',
  43                 'shi',
  44                 'sr',
  45                 'tg',
  46                 'uz',
  47                 'zh',
  48         );
  49
  50         public $mMainLanguageCode;
  51         public $mVariants, $mVariantFallbacks, $mVariantNames;
  52         public $mTablesLoaded = false;
  53         public $mTables;
  54         // 'bidirectional' 'unidirectional' 'disable' for each variant
  55         public $mManualLevel;
  56
  57         /**
  58          * @var String: memcached key name
  59          */
  60         public $mCacheKey;
  61
  62         public $mLangObj;
  63         public $mFlags;
  64         public $mDescCodeSep = ':', $mDescVarSep = ';';
  65         public $mUcfirst = false;
  66         public $mConvRuleTitle = false;
  67         public $mURLVariant;
  68         public $mUserVariant;
  69         public $mHeaderVariant;
  70         public $mMaxDepth = 10;
  71         public $mVarSeparatorPattern;
  72
  73         const CACHE_VERSION_KEY = 'VERSION 6';
  74
  75         /**
  76          * Constructor
  77          *
  78          * @param $langobj Language: the Language Object
  79          * @param $maincode String: the main language code of this language
  80          * @param $variants Array: the supported variants of this language
  81          * @param $variantfallbacks Array: the fallback language of each variant
  82          * @param $flags Array: defining the custom strings that maps to the flags
  83          * @param $manualLevel Array: limit for supported variants
  84          */
  85         public function __construct( $langobj, $maincode, $variants = array(),
  86                                                                 $variantfallbacks = array(), $flags = array(),
  87                                                                 $manualLevel = array() ) {
  88                 global $wgDisabledVariants;
  89                 $this->mLangObj = $langobj;
  90                 $this->mMainLanguageCode = $maincode;
  91                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  92                 $this->mVariantFallbacks = $variantfallbacks;
  93                 $this->mVariantNames = Language::fetchLanguageNames();
  94                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  95                 $defaultflags = array(
  96                         // 'S' show converted text
  97                         // '+' add rules for alltext
  98                         // 'E' the gave flags is error
  99                         // these flags above are reserved for program
 100                         'A' => 'A',       // add rule for convert code (all text convert)
 101                         'T' => 'T',       // title convert
 102                         'R' => 'R',       // raw content
 103                         'D' => 'D',       // convert description (subclass implement)
 104                         '-' => '-',       // remove convert (not implement)
 105                         'H' => 'H',       // add rule for convert code
 106                                                   // (but no display in placed code)
 107                         'N' => 'N'        // current variant name
 108                 );
 109                 $this->mFlags = array_merge( $defaultflags, $flags );
 110                 foreach ( $this->mVariants as $v ) {
 111                         if ( array_key_exists( $v, $manualLevel ) ) {
 112                                 $this->mManualLevel[$v] = $manualLevel[$v];
 113                         } else {
 114                                 $this->mManualLevel[$v] = 'bidirectional';
 115                         }
 116                         $this->mFlags[$v] = $v;
 117                 }
 118         }
 119
 120         /**
 121          * Get all valid variants.
 122          * Call this instead of using $this->mVariants directly.
 123          *
 124          * @return Array: contains all valid variants
 125          */
 126         public function getVariants() {
 127                 return $this->mVariants;
 128         }
 129
 130         /**
 131          * In case some variant is not defined in the markup, we need
 132          * to have some fallback. For example, in zh, normally people
 133          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 134          * when zh-sg is preferred but not defined, we will pick zh-hans
 135          * in this case. Right now this is only used by zh.
 136          *
 137          * @param $variant String: the language code of the variant
 138          * @return String|array: The code of the fallback language or the
 139          *                               main code if there is no fallback
 140          */
 141         public function getVariantFallbacks( $variant ) {
 142                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 143                         return $this->mVariantFallbacks[$variant];
 144                 }
 145                 return $this->mMainLanguageCode;
 146         }
 147
 148         /**
 149          * Get the title produced by the conversion rule.
 150          * @return String: The converted title text
 151          */
 152         public function getConvRuleTitle() {
 153                 return $this->mConvRuleTitle;
 154         }
 155
 156         /**
 157          * Get preferred language variant.
 158          * @return String: the preferred language code
 159          */
 160         public function getPreferredVariant() {
 161                 global $wgDefaultLanguageVariant, $wgUser;
 162
 163                 $req = $this->getURLVariant();
 164
 165                 if ( $wgUser->isLoggedIn() && !$req ) {
 166                         $req = $this->getUserVariant();
 167                 } elseif ( !$req ) {
 168                         $req = $this->getHeaderVariant();
 169                 }
 170
 171                 if ( $wgDefaultLanguageVariant && !$req ) {
 172                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 173                 }
 174
 175                 // This function, unlike the other get*Variant functions, is
 176                 // not memoized (i.e. there return value is not cached) since
 177                 // new information might appear during processing after this
 178                 // is first called.
 179                 if ( $this->validateVariant( $req ) ) {
 180                         return $req;
 181                 }
 182                 return $this->mMainLanguageCode;
 183         }
 184
 185         /**
 186          * Get default variant.
 187          * This function would not be affected by user's settings
 188          * @return String: the default variant code
 189          */
 190         public function getDefaultVariant() {
 191                 global $wgDefaultLanguageVariant;
 192
 193                 $req = $this->getURLVariant();
 194
 195                 if ( !$req ) {
 196                         $req = $this->getHeaderVariant();
 197                 }
 198
 199                 if ( $wgDefaultLanguageVariant && !$req ) {
 200                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 201                 }
 202
 203                 if ( $req ) {
 204                         return $req;
 205                 }
 206                 return $this->mMainLanguageCode;
 207         }
 208
 209         /**
 210          * Validate the variant
 211          * @param $variant String: the variant to validate
 212          * @return Mixed: returns the variant if it is valid, null otherwise
 213          */
 214         public function validateVariant( $variant = null ) {
 215                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 216                         return $variant;
 217                 }
 218                 return null;
 219         }
 220
 221         /**
 222          * Get the variant specified in the URL
 223          *
 224          * @return Mixed: variant if one found, false otherwise.
 225          */
 226         public function getURLVariant() {
 227                 global $wgRequest;
 228
 229                 if ( $this->mURLVariant ) {
 230                         return $this->mURLVariant;
 231                 }
 232
 233                 // see if the preference is set in the request
 234                 $ret = $wgRequest->getText( 'variant' );
 235
 236                 if ( !$ret ) {
 237                         $ret = $wgRequest->getVal( 'uselang' );
 238                 }
 239
 240                 return $this->mURLVariant = $this->validateVariant( $ret );
 241         }
 242
 243         /**
 244          * Determine if the user has a variant set.
 245          *
 246          * @return Mixed: variant if one found, false otherwise.
 247          */
 248         protected function getUserVariant() {
 249                 global $wgUser;
 250
 251                 // memoizing this function wreaks havoc on parserTest.php
 252                 /*
 253                 if ( $this->mUserVariant ) {
 254                         return $this->mUserVariant;
 255                 }
 256                 */
 257
 258                 // Get language variant preference from logged in users
 259                 // Don't call this on stub objects because that causes infinite
 260                 // recursion during initialisation
 261                 if ( $wgUser->isLoggedIn() )  {
 262                         $ret = $wgUser->getOption( 'variant' );
 263                 } else {
 264                         // figure out user lang without constructing wgLang to avoid
 265                         // infinite recursion
 266                         $ret = $wgUser->getOption( 'language' );
 267                 }
 268
 269                 return $this->mUserVariant = $this->validateVariant( $ret );
 270         }
 271
 272         /**
 273          * Determine the language variant from the Accept-Language header.
 274          *
 275          * @return Mixed: variant if one found, false otherwise.
 276          */
 277         protected function getHeaderVariant() {
 278                 global $wgRequest;
 279
 280                 if ( $this->mHeaderVariant ) {
 281                         return $this->mHeaderVariant;
 282                 }
 283
 284                 // see if some supported language variant is set in the
 285                 // HTTP header.
 286                 $languages = array_keys( $wgRequest->getAcceptLang() );
 287                 if ( empty( $languages ) ) {
 288                         return null;
 289                 }
 290
 291                 $fallbackLanguages = array();
 292                 foreach ( $languages as $language ) {
 293                         $this->mHeaderVariant = $this->validateVariant( $language );
 294                         if ( $this->mHeaderVariant ) {
 295                                 break;
 296                         }
 297
 298                         // To see if there are fallbacks of current language.
 299                         // We record these fallback variants, and process
 300                         // them later.
 301                         $fallbacks = $this->getVariantFallbacks( $language );
 302                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 303                                 $fallbackLanguages[] = $fallbacks;
 304                         } elseif ( is_array( $fallbacks ) ) {
 305                                 $fallbackLanguages =
 306                                         array_merge( $fallbackLanguages, $fallbacks );
 307                         }
 308                 }
 309
 310                 if ( !$this->mHeaderVariant ) {
 311                         // process fallback languages now
 312                         $fallback_languages = array_unique( $fallbackLanguages );
 313                         foreach ( $fallback_languages as $language ) {
 314                                 $this->mHeaderVariant = $this->validateVariant( $language );
 315                                 if ( $this->mHeaderVariant ) {
 316                                         break;
 317                                 }
 318                         }
 319                 }
 320
 321                 return $this->mHeaderVariant;
 322         }
 323
 324         /**
 325          * Dictionary-based conversion.
 326          * This function would not parse the conversion rules.
 327          * If you want to parse rules, try to use convert() or
 328          * convertTo().
 329          *
 330          * @param $text String the text to be converted
 331          * @param $toVariant bool|string the target language code
 332          * @return String the converted text
 333          */
 334         public function autoConvert( $text, $toVariant = false ) {
 335                 wfProfileIn( __METHOD__ );
 336
 337                 $this->loadTables();
 338
 339                 if ( !$toVariant ) {
 340                         $toVariant = $this->getPreferredVariant();
 341                         if ( !$toVariant ) {
 342                                 wfProfileOut( __METHOD__ );
 343                                 return $text;
 344                         }
 345                 }
 346
 347                 if( $this->guessVariant( $text, $toVariant ) ) {
 348                         wfProfileOut( __METHOD__ );
 349                         return $text;
 350                 }
 351
 352                 /* we convert everything except:
 353                    1. HTML markups (anything between < and >)
 354                    2. HTML entities
 355                    3. placeholders 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                 $startPos = 0;
 377                 $sourceBlob = '';
 378                 $literalBlob = '';
 379
 380                 // Guard against delimiter nulls in the input
 381                 $text = str_replace( "\000", '', $text );
 382
 383                 $markupMatches = null;
 384                 $elementMatches = null;
 385                 while ( $startPos < strlen( $text ) ) {
 386                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 387                                 $elementPos = $markupMatches[0][1];
 388                                 $element = $markupMatches[0][0];
 389                         } else {
 390                                 $elementPos = strlen( $text );
 391                                 $element = '';
 392                         }
 393
 394                         // Queue the part before the markup for translation in a batch
 395                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 396
 397                         // Advance to the next position
 398                         $startPos = $elementPos + strlen( $element );
 399
 400                         // Translate any alt or title attributes inside the matched element
 401                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 402                                 $elementMatches ) )
 403                         {
 404                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 405                                 $changed = false;
 406                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 407                                         if ( !isset( $attrs[$attrName] ) ) {
 408                                                 continue;
 409                                         }
 410                                         $attr = $attrs[$attrName];
 411                                         // Don't convert URLs
 412                                         if ( !strpos( $attr, '://' ) ) {
 413                                                 $attr = $this->convertTo( $attr, $toVariant );
 414                                         }
 415
 416                                         // Remove HTML tags to avoid disrupting the layout
 417                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 418                                         if ( $attr !== $attrs[$attrName] ) {
 419                                                 $attrs[$attrName] = $attr;
 420                                                 $changed = true;
 421                                         }
 422                                 }
 423                                 if ( $changed ) {
 424                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 425                                                 $elementMatches[3];
 426                                 }
 427                         }
 428                         $literalBlob .= $element . "\000";
 429                 }
 430
 431                 // Do the main translation batch
 432                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 433
 434                 // Put the output back together
 435                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 436                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 437                 $output = '';
 438                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 439                         $output .= $translatedIter->current();
 440                         $output .= $literalIter->current();
 441                         $translatedIter->next();
 442                         $literalIter->next();
 443                 }
 444
 445                 wfProfileOut( __METHOD__ );
 446                 return $output;
 447         }
 448
 449         /**
 450          * Translate a string to a variant.
 451          * Doesn't parse rules or do any of that other stuff, for that use
 452          * convert() or convertTo().
 453          *
 454          * @param $text String: text to convert
 455          * @param $variant String: variant language code
 456          * @return String: translated text
 457          */
 458         public function translate( $text, $variant ) {
 459                 wfProfileIn( __METHOD__ );
 460                 // If $text is empty or only includes spaces, do nothing
 461                 // Otherwise translate it
 462                 if ( trim( $text ) ) {
 463                         $this->loadTables();
 464                         $text = $this->mTables[$variant]->replace( $text );
 465                 }
 466                 wfProfileOut( __METHOD__ );
 467                 return $text;
 468         }
 469
 470         /**
 471          * Call translate() to convert text to all valid variants.
 472          *
 473          * @param $text String: the text to be converted
 474          * @return Array: variant => converted text
 475          */
 476         public function autoConvertToAllVariants( $text ) {
 477                 wfProfileIn( __METHOD__ );
 478                 $this->loadTables();
 479
 480                 $ret = array();
 481                 foreach ( $this->mVariants as $variant ) {
 482                         $ret[$variant] = $this->translate( $text, $variant );
 483                 }
 484
 485                 wfProfileOut( __METHOD__ );
 486                 return $ret;
 487         }
 488
 489         /**
 490          * Convert link text to all valid variants.
 491          * In the first, this function only convert text outside the
 492          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 493          * titles, the text will get all converted always.
 494          * So I removed this feature and deprecated the function.
 495          *
 496          * @param $text String: the text to be converted
 497          * @return Array: variant => converted text
 498          * @deprecated since 1.17 Use autoConvertToAllVariants() instead
 499          */
 500         public function convertLinkToAllVariants( $text ) {
 501                 return $this->autoConvertToAllVariants( $text );
 502         }
 503
 504         /**
 505          * Apply manual conversion rules.
 506          *
 507          * @param $convRule ConverterRule Object of ConverterRule
 508          */
 509         protected function applyManualConv( $convRule ) {
 510                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 511                 // title conversion.
 512                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 513                 // rule(s) not for title, this breaks the title conversion.
 514                 $newConvRuleTitle = $convRule->getTitle();
 515                 if ( $newConvRuleTitle ) {
 516                         // So I add an empty check for getTitle()
 517                         $this->mConvRuleTitle = $newConvRuleTitle;
 518                 }
 519
 520                 // merge/remove manual conversion rules to/from global table
 521                 $convTable = $convRule->getConvTable();
 522                 $action = $convRule->getRulesAction();
 523                 foreach ( $convTable as $variant => $pair ) {
 524                         if ( !$this->validateVariant( $variant ) ) {
 525                                 continue;
 526                         }
 527
 528                         if ( $action == 'add' ) {
 529                                 foreach ( $pair as $from => $to ) {
 530                                         // to ensure that $from and $to not be left blank
 531                                         // so $this->translate() could always return a string
 532                                         if ( $from || $to ) {
 533                                                 // more efficient than array_merge(), about 2.5 times.
 534                                                 $this->mTables[$variant]->setPair( $from, $to );
 535                                         }
 536                                 }
 537                         } elseif ( $action == 'remove' ) {
 538                                 $this->mTables[$variant]->removeArray( $pair );
 539                         }
 540                 }
 541         }
 542
 543         /**
 544          * Auto convert a Title object to a readable string in the
 545          * preferred variant.
 546          *
 547          * @param $title Title a object of Title
 548          * @return String: converted title text
 549          */
 550         public function convertTitle( $title ) {
 551                 $variant = $this->getPreferredVariant();
 552                 $index = $title->getNamespace();
 553                 if ( $index !== NS_MAIN ) {
 554                         $text = $this->convertNamespace( $index ) . ':';
 555                 } else {
 556                         $text = '';
 557                 }
 558                 $text .= $this->translate( $title->getText(), $variant );
 559                 return $text;
 560         }
 561
 562         /**
 563          * Get the namespace display name in the preferred variant.
 564          *
 565          * @param $index int namespace id
 566          * @return String: namespace name for display
 567          */
 568         public function convertNamespace( $index ) {
 569                 $variant = $this->getPreferredVariant();
 570                 if ( $index === NS_MAIN ) {
 571                         return '';
 572                 } else {
 573                         // First check if a message gives a converted name in the target variant.
 574                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 575                         if ( $nsConvMsg->exists() ) {
 576                                 return $nsConvMsg->plain();
 577                         }
 578                         // Then check if a message gives a converted name in content language
 579                         // which needs extra translation to the target variant.
 580                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 581                         if ( $nsConvMsg->exists() ) {
 582                                 return $this->translate( $nsConvMsg->plain(), $variant );
 583                         }
 584                         // No message exists, retrieve it from the target variant's namespace names.
 585                         $langObj = $this->mLangObj->factory( $variant );
 586                         return $langObj->getFormattedNsText( $index );
 587                 }
 588         }
 589
 590         /**
 591          * Convert text to different variants of a language. The automatic
 592          * conversion is done in autoConvert(). Here we parse the text
 593          * marked with -{}-, which specifies special conversions of the
 594          * text that can not be accomplished in autoConvert().
 595          *
 596          * Syntax of the markup:
 597          * -{code1:text1;code2:text2;...}-  or
 598          * -{flags|code1:text1;code2:text2;...}-  or
 599          * -{text}- in which case no conversion should take place for text
 600          *
 601          * @param $text String: text to be converted
 602          * @return String: converted text
 603          */
 604         public function convert( $text ) {
 605                 $variant = $this->getPreferredVariant();
 606                 return $this->convertTo( $text, $variant );
 607         }
 608
 609         /**
 610          * Same as convert() except a extra parameter to custom variant.
 611          *
 612          * @param $text String: text to be converted
 613          * @param $variant String: the target variant code
 614          * @return String: converted text
 615          */
 616         public function convertTo( $text, $variant ) {
 617                 global $wgDisableLangConversion;
 618                 if ( $wgDisableLangConversion ) {
 619                         return $text;
 620                 }
 621                 return $this->recursiveConvertTopLevel( $text, $variant );
 622         }
 623
 624         /**
 625          * Recursively convert text on the outside. Allow to use nested
 626          * markups to custom rules.
 627          *
 628          * @param $text String: text to be converted
 629          * @param $variant String: the target variant code
 630          * @param $depth Integer: depth of recursion
 631          * @return String: converted text
 632          */
 633         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 634                 $startPos = 0;
 635                 $out = '';
 636                 $length = strlen( $text );
 637                 $shouldConvert = !$this->guessVariant( $text, $variant );
 638
 639                 while ( $startPos < $length ) {
 640                         $pos = strpos( $text, '-{', $startPos );
 641
 642                         if ( $pos === false ) {
 643                                 // No more markup, append final segment
 644                                 $fragment = substr( $text, $startPos );
 645                                 $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 646                                 return $out;
 647                         }
 648
 649                         // Markup found
 650                         // Append initial segment
 651                         $fragment = substr( $text, $startPos, $pos - $startPos );
 652                         $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 653
 654                         // Advance position
 655                         $startPos = $pos;
 656
 657                         // Do recursive conversion
 658                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 659                 }
 660
 661                 return $out;
 662         }
 663
 664         /**
 665          * Recursively convert text on the inside.
 666          *
 667          * @param $text String: text to be converted
 668          * @param $variant String: the target variant code
 669          * @param $startPos int
 670          * @param $depth Integer: depth of recursion
 671          *
 672          * @throws MWException
 673          * @return String: converted text
 674          */
 675         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 676                 // Quick sanity check (no function calls)
 677                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 678                         throw new MWException( __METHOD__ . ': invalid input string' );
 679                 }
 680
 681                 $startPos += 2;
 682                 $inner = '';
 683                 $warningDone = false;
 684                 $length = strlen( $text );
 685
 686                 while ( $startPos < $length ) {
 687                         $m = false;
 688                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 689                         if ( !$m ) {
 690                                 // Unclosed rule
 691                                 break;
 692                         }
 693
 694                         $token = $m[0][0];
 695                         $pos = $m[0][1];
 696
 697                         // Markup found
 698                         // Append initial segment
 699                         $inner .= substr( $text, $startPos, $pos - $startPos );
 700
 701                         // Advance position
 702                         $startPos = $pos;
 703
 704                         switch ( $token ) {
 705                                 case '-{':
 706                                         // Check max depth
 707                                         if ( $depth >= $this->mMaxDepth ) {
 708                                                 $inner .= '-{';
 709                                                 if ( !$warningDone ) {
 710                                                         $inner .= '<span class="error">' .
 711                                                                 wfMessage( 'language-converter-depth-warning' )
 712                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 713                                                                 '</span>';
 714                                                         $warningDone = true;
 715                                                 }
 716                                                 $startPos += 2;
 717                                                 continue;
 718                                         }
 719                                         // Recursively parse another rule
 720                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 721                                         break;
 722                                 case '}-':
 723                                         // Apply the rule
 724                                         $startPos += 2;
 725                                         $rule = new ConverterRule( $inner, $this );
 726                                         $rule->parse( $variant );
 727                                         $this->applyManualConv( $rule );
 728                                         return $rule->getDisplay();
 729                                 default:
 730                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 731                         }
 732                 }
 733
 734                 // Unclosed rule
 735                 if ( $startPos < $length ) {
 736                         $inner .= substr( $text, $startPos );
 737                 }
 738                 $startPos = $length;
 739                 return '-{' . $this->autoConvert( $inner, $variant );
 740         }
 741
 742         /**
 743          * If a language supports multiple variants, it is possible that
 744          * non-existing link in one variant actually exists in another variant.
 745          * This function tries to find it. See e.g. LanguageZh.php
 746          *
 747          * @param $link String: the name of the link
 748          * @param $nt Mixed: the title object of the link
 749          * @param $ignoreOtherCond Boolean: to disable other conditions when
 750          *              we need to transclude a template or update a category's link
 751          * @return Null, the input parameters may be modified upon return
 752          */
 753         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 754                 # If the article has already existed, there is no need to
 755                 # check it again, otherwise it may cause a fault.
 756                 if ( is_object( $nt ) && $nt->exists() ) {
 757                         return;
 758                 }
 759
 760                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 761                         $wgUser;
 762                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 763                 $action = $wgRequest->getText( 'action' );
 764                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 765                 $disableLinkConversion = $wgDisableLangConversion
 766                         || $wgDisableTitleConversion;
 767                 $linkBatch = new LinkBatch();
 768
 769                 $ns = NS_MAIN;
 770
 771                 if ( $disableLinkConversion ||
 772                          ( !$ignoreOtherCond &&
 773                            ( $isredir == 'no'
 774                                  || $action == 'edit'
 775                                  || $action == 'submit'
 776                                  || $linkconvert == 'no'
 777                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 778                         return;
 779                 }
 780
 781                 if ( is_object( $nt ) ) {
 782                         $ns = $nt->getNamespace();
 783                 }
 784
 785                 $variants = $this->autoConvertToAllVariants( $link );
 786                 if ( !$variants ) { // give up
 787                         return;
 788                 }
 789
 790                 $titles = array();
 791
 792                 foreach ( $variants as $v ) {
 793                         if ( $v != $link ) {
 794                                 $varnt = Title::newFromText( $v, $ns );
 795                                 if ( !is_null( $varnt ) ) {
 796                                         $linkBatch->addObj( $varnt );
 797                                         $titles[] = $varnt;
 798                                 }
 799                         }
 800                 }
 801
 802                 // fetch all variants in single query
 803                 $linkBatch->execute();
 804
 805                 foreach ( $titles as $varnt ) {
 806                         if ( $varnt->getArticleID() > 0 ) {
 807                                 $nt = $varnt;
 808                                 $link = $varnt->getText();
 809                                 break;
 810                         }
 811                 }
 812         }
 813
 814         /**
 815          * Returns language specific hash options.
 816          *
 817          * @return string
 818          */
 819         public function getExtraHashOptions() {
 820                 $variant = $this->getPreferredVariant();
 821                 return '!' . $variant;
 822         }
 823
 824         /**
 825          * Guess if a text is written in a variant. This should be implemented in subclasses.
 826          *
 827          * @param string        $text the text to be checked
 828          * @param string        $variant language code of the variant to be checked for
 829          * @return bool true if $text appears to be written in $variant, false if not
 830          *
 831          * @author Nikola Smolenski <smolensk@eunet.rs>
 832          * @since 1.19
 833          */
 834         public function guessVariant($text, $variant) {
 835                 return false;
 836         }
 837
 838         /**
 839          * Load default conversion tables.
 840          * This method must be implemented in derived class.
 841          *
 842          * @private
 843          * @throws MWException
 844          */
 845         function loadDefaultTables() {
 846                 $name = get_class( $this );
 847                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 848         }
 849
 850         /**
 851          * Load conversion tables either from the cache or the disk.
 852          * @private
 853          * @param $fromCache Boolean: load from memcached? Defaults to true.
 854          */
 855         function loadTables( $fromCache = true ) {
 856                 global $wgLangConvMemc;
 857
 858                 if ( $this->mTablesLoaded ) {
 859                         return;
 860                 }
 861
 862                 wfProfileIn( __METHOD__ );
 863                 $this->mTablesLoaded = true;
 864                 $this->mTables = false;
 865                 if ( $fromCache ) {
 866                         wfProfileIn( __METHOD__ . '-cache' );
 867                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 868                         wfProfileOut( __METHOD__ . '-cache' );
 869                 }
 870                 if ( !$this->mTables
 871                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 872                         wfProfileIn( __METHOD__ . '-recache' );
 873                         // not in cache, or we need a fresh reload.
 874                         // We will first load the default tables
 875                         // then update them using things in MediaWiki:Conversiontable/*
 876                         $this->loadDefaultTables();
 877                         foreach ( $this->mVariants as $var ) {
 878                                 $cached = $this->parseCachedTable( $var );
 879                                 $this->mTables[$var]->mergeArray( $cached );
 880                         }
 881
 882                         $this->postLoadTables();
 883                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 884
 885                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 886                         wfProfileOut( __METHOD__ . '-recache' );
 887                 }
 888                 wfProfileOut( __METHOD__ );
 889         }
 890
 891         /**
 892          * Hook for post processing after conversion tables are loaded.
 893          */
 894         function postLoadTables() { }
 895
 896         /**
 897          * Reload the conversion tables.
 898          *
 899          * @private
 900          */
 901         function reloadTables() {
 902                 if ( $this->mTables ) {
 903                         unset( $this->mTables );
 904                 }
 905                 $this->mTablesLoaded = false;
 906                 $this->loadTables( false );
 907         }
 908
 909         /**
 910          * Parse the conversion table stored in the cache.
 911          *
 912          * The tables should be in blocks of the following form:
 913          *              -{
 914          *                      word => word ;
 915          *                      word => word ;
 916          *                      ...
 917          *              }-
 918          *
 919          * To make the tables more manageable, subpages are allowed
 920          * and will be parsed recursively if $recursive == true.
 921          *
 922          * @param $code String: language code
 923          * @param $subpage String: subpage name
 924          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 925          *
 926          * @return array
 927          */
 928         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 929                 static $parsed = array();
 930
 931                 $key = 'Conversiontable/' . $code;
 932                 if ( $subpage ) {
 933                         $key .= '/' . $subpage;
 934                 }
 935                 if ( array_key_exists( $key, $parsed ) ) {
 936                         return array();
 937                 }
 938
 939                 $parsed[$key] = true;
 940
 941                 if ( $subpage === '' ) {
 942                         $txt = MessageCache::singleton()->get( 'conversiontable', true, $code );
 943                 } else {
 944                         $txt = false;
 945                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 946                         if ( $title && $title->exists() ) {
 947                                 $revision = Revision::newFromTitle( $title );
 948                                 if ( $revision ) {
 949                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 950                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 951                                         }
 952
 953                                         //@todo: in the future, use a specialized content model, perhaps based on json!
 954                                 }
 955                         }
 956                 }
 957
 958                 # Nothing to parse if there's no text
 959                 if ( $txt === false || $txt === null || $txt === '' ) {
 960                         return array();
 961                 }
 962
 963                 // get all subpage links of the form
 964                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 965                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 966                         ':Conversiontable';
 967                 $subs = StringUtils::explode( '[[', $txt );
 968                 $sublinks = array();
 969                 foreach ( $subs as $sub ) {
 970                         $link = explode( ']]', $sub, 2 );
 971                         if ( count( $link ) != 2 ) {
 972                                 continue;
 973                         }
 974                         $b = explode( '|', $link[0], 2 );
 975                         $b = explode( '/', trim( $b[0] ), 3 );
 976                         if ( count( $b ) == 3 ) {
 977                                 $sublink = $b[2];
 978                         } else {
 979                                 $sublink = '';
 980                         }
 981
 982                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 983                                 $sublinks[] = $sublink;
 984                         }
 985                 }
 986
 987                 // parse the mappings in this page
 988                 $blocks = StringUtils::explode( '-{', $txt );
 989                 $ret = array();
 990                 $first = true;
 991                 foreach ( $blocks as $block ) {
 992                         if ( $first ) {
 993                                 // Skip the part before the first -{
 994                                 $first = false;
 995                                 continue;
 996                         }
 997                         $mappings = explode( '}-', $block, 2 );
 998                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 999                                                                          $mappings[0] );
1000                         $table = StringUtils::explode( ';', $stripped );
1001                         foreach ( $table as $t ) {
1002                                 $m = explode( '=>', $t, 3 );
1003                                 if ( count( $m ) != 2 ) {
1004                                         continue;
1005                                 }
1006                                 // trim any trailling comments starting with '//'
1007                                 $tt = explode( '//', $m[1], 2 );
1008                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1009                         }
1010                 }
1011
1012                 // recursively parse the subpages
1013                 if ( $recursive ) {
1014                         foreach ( $sublinks as $link ) {
1015                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1016                                 $ret = array_merge( $ret, $s );
1017                         }
1018                 }
1019
1020                 if ( $this->mUcfirst ) {
1021                         foreach ( $ret as $k => $v ) {
1022                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1023                         }
1024                 }
1025                 return $ret;
1026         }
1027
1028         /**
1029          * Enclose a string with the "no conversion" tag. This is used by
1030          * various functions in the Parser.
1031          *
1032          * @param $text String: text to be tagged for no conversion
1033          * @param $noParse Boolean: unused
1034          * @return String: the tagged text
1035          */
1036         public function markNoConversion( $text, $noParse = false ) {
1037                 # don't mark if already marked
1038                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1039                         return $text;
1040                 }
1041
1042                 $ret = "-{R|$text}-";
1043                 return $ret;
1044         }
1045
1046         /**
1047          * Convert the sorting key for category links. This should make different
1048          * keys that are variants of each other map to the same key.
1049          *
1050          * @param $key string
1051          *
1052          * @return string
1053          */
1054         function convertCategoryKey( $key ) {
1055                 return $key;
1056         }
1057
1058         /**
1059          * Hook to refresh the cache of conversion tables when
1060          * MediaWiki:Conversiontable* is updated.
1061          * @private
1062          *
1063          * @param $page WikiPage object
1064          * @param $user Object: User object for the current user
1065          * @param $content Content: new page content
1066          * @param $summary String: edit summary of the edit
1067          * @param $isMinor Boolean: was the edit marked as minor?
1068          * @param $isWatch Boolean: did the user watch this page or not?
1069          * @param $section
1070          * @param $flags int Bitfield
1071          * @param $revision Object: new Revision object or null
1072          * @return Boolean: true
1073          */
1074         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1075                         $isWatch, $section, $flags, $revision ) {
1076                 $titleobj = $page->getTitle();
1077                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1078                         $title = $titleobj->getDBkey();
1079                         $t = explode( '/', $title, 3 );
1080                         $c = count( $t );
1081                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1082                                 if ( $this->validateVariant( $t[1] ) ) {
1083                                         $this->reloadTables();
1084                                 }
1085                         }
1086                 }
1087                 return true;
1088         }
1089
1090         /**
1091          * Armour rendered math against conversion.
1092          * Escape special chars in parsed math text. (in most cases are img elements)
1093          *
1094          * @param $text String: text to armour against conversion
1095          * @return String: armoured text where { and } have been converted to
1096          *                 &#123; and &#125;
1097          */
1098         public function armourMath( $text ) {
1099                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1100                 // any unwanted markup appearing in the math image tag.
1101                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1102                 return $text;
1103         }
1104
1105         /**
1106          * Get the cached separator pattern for ConverterRule::parseRules()
1107          */
1108         function getVarSeparatorPattern() {
1109                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1110                         // varsep_pattern for preg_split:
1111                         // text should be splited by ";" only if a valid variant
1112                         // name exist after the markup, for example:
1113                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1114                         //      <span style="font-size:120%;">yyy</span>;}-
1115                         // we should split it as:
1116                         //  array(
1117                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1118                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1119                         //        [2] => ''
1120                         //       )
1121                         $pat = '/;\s*(?=';
1122                         foreach ( $this->mVariants as $variant ) {
1123                                 // zh-hans:xxx;zh-hant:yyy
1124                                 $pat .= $variant . '\s*:|';
1125                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1126                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1127                         }
1128                         $pat .= '\s*$)/';
1129                         $this->mVarSeparatorPattern = $pat;
1130                 }
1131                 return $this->mVarSeparatorPattern;
1132         }
1133 }
1134
1135 /**
1136  * Parser for rules of language conversion , parse rules in -{ }- tag.
1137  * @ingroup Language
1138  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1139  */
1140 class ConverterRule {
1141         public $mText; // original text in -{text}-
1142         public $mConverter; // LanguageConverter object
1143         public $mRuleDisplay = '';
1144         public $mRuleTitle = false;
1145         public $mRules = '';// string : the text of the rules
1146         public $mRulesAction = 'none';
1147         public $mFlags = array();
1148         public $mVariantFlags = array();
1149         public $mConvTable = array();
1150         public $mBidtable = array();// array of the translation in each variant
1151         public $mUnidtable = array();// array of the translation in each variant
1152
1153         /**
1154          * Constructor
1155          *
1156          * @param $text String: the text between -{ and }-
1157          * @param $converter LanguageConverter object
1158          */
1159         public function __construct( $text, $converter ) {
1160                 $this->mText = $text;
1161                 $this->mConverter = $converter;
1162         }
1163
1164         /**
1165          * Check if variants array in convert array.
1166          *
1167          * @param $variants Array or string: variant language code
1168          * @return String: translated text
1169          */
1170         public function getTextInBidtable( $variants ) {
1171                 $variants = (array)$variants;
1172                 if ( !$variants ) {
1173                         return false;
1174                 }
1175                 foreach ( $variants as $variant ) {
1176                         if ( isset( $this->mBidtable[$variant] ) ) {
1177                                 return $this->mBidtable[$variant];
1178                         }
1179                 }
1180                 return false;
1181         }
1182
1183         /**
1184          * Parse flags with syntax -{FLAG| ... }-
1185          * @private
1186          */
1187         function parseFlags() {
1188                 $text = $this->mText;
1189                 $flags = array();
1190                 $variantFlags = array();
1191
1192                 $sepPos = strpos( $text, '|' );
1193                 if ( $sepPos !== false ) {
1194                         $validFlags = $this->mConverter->mFlags;
1195                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1196                         foreach ( $f as $ff ) {
1197                                 $ff = trim( $ff );
1198                                 if ( isset( $validFlags[$ff] ) ) {
1199                                         $flags[$validFlags[$ff]] = true;
1200                                 }
1201                         }
1202                         $text = strval( substr( $text, $sepPos + 1 ) );
1203                 }
1204
1205                 if ( !$flags ) {
1206                         $flags['S'] = true;
1207                 } elseif ( isset( $flags['R'] ) ) {
1208                         $flags = array( 'R' => true );// remove other flags
1209                 } elseif ( isset( $flags['N'] ) ) {
1210                         $flags = array( 'N' => true );// remove other flags
1211                 } elseif ( isset( $flags['-'] ) ) {
1212                         $flags = array( '-' => true );// remove other flags
1213                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1214                         $flags['H'] = true;
1215                 } elseif ( isset( $flags['H'] ) ) {
1216                         // replace A flag, and remove other flags except T
1217                         $temp = array( '+' => true, 'H' => true );
1218                         if ( isset( $flags['T'] ) ) {
1219                                 $temp['T'] = true;
1220                         }
1221                         if ( isset( $flags['D'] ) ) {
1222                                 $temp['D'] = true;
1223                         }
1224                         $flags = $temp;
1225                 } else {
1226                         if ( isset( $flags['A'] ) ) {
1227                                 $flags['+'] = true;
1228                                 $flags['S'] = true;
1229                         }
1230                         if ( isset( $flags['D'] ) ) {
1231                                 unset( $flags['S'] );
1232                         }
1233                         // try to find flags like "zh-hans", "zh-hant"
1234                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1235                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1236                         if ( $variantFlags ) {
1237                                 $variantFlags = array_flip( $variantFlags );
1238                                 $flags = array();
1239                         }
1240                 }
1241                 $this->mVariantFlags = $variantFlags;
1242                 $this->mRules = $text;
1243                 $this->mFlags = $flags;
1244         }
1245
1246         /**
1247          * Generate conversion table.
1248          * @private
1249          */
1250         function parseRules() {
1251                 $rules = $this->mRules;
1252                 $bidtable = array();
1253                 $unidtable = array();
1254                 $variants = $this->mConverter->mVariants;
1255                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1256
1257                 $choice = preg_split( $varsep_pattern, $rules );
1258
1259                 foreach ( $choice as $c ) {
1260                         $v  = explode( ':', $c, 2 );
1261                         if ( count( $v ) != 2 ) {
1262                                 // syntax error, skip
1263                                 continue;
1264                         }
1265                         $to = trim( $v[1] );
1266                         $v  = trim( $v[0] );
1267                         $u  = explode( '=>', $v, 2 );
1268                         // if $to is empty, strtr() could return a wrong result
1269                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1270                                 $bidtable[$v] = $to;
1271                         } elseif ( count( $u ) == 2 ) {
1272                                 $from = trim( $u[0] );
1273                                 $v      = trim( $u[1] );
1274                                 if ( array_key_exists( $v, $unidtable )
1275                                          && !is_array( $unidtable[$v] )
1276                                          && $to
1277                                          && in_array( $v, $variants ) ) {
1278                                         $unidtable[$v] = array( $from => $to );
1279                                 } elseif ( $to && in_array( $v, $variants ) ) {
1280                                         $unidtable[$v][$from] = $to;
1281                                 }
1282                         }
1283                         // syntax error, pass
1284                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1285                                 $bidtable = array();
1286                                 $unidtable = array();
1287                                 break;
1288                         }
1289                 }
1290                 $this->mBidtable = $bidtable;
1291                 $this->mUnidtable = $unidtable;
1292         }
1293
1294         /**
1295          * @private
1296          *
1297          * @return string
1298          */
1299         function getRulesDesc() {
1300                 $codesep = $this->mConverter->mDescCodeSep;
1301                 $varsep = $this->mConverter->mDescVarSep;
1302                 $text = '';
1303                 foreach ( $this->mBidtable as $k => $v ) {
1304                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1305                 }
1306                 foreach ( $this->mUnidtable as $k => $a ) {
1307                         foreach ( $a as $from => $to ) {
1308                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1309                                         "$codesep$to$varsep";
1310                         }
1311                 }
1312                 return $text;
1313         }
1314
1315         /**
1316          * Parse rules conversion.
1317          * @private
1318          *
1319          * @param $variant
1320          *
1321          * @return string
1322          */
1323         function getRuleConvertedStr( $variant ) {
1324                 $bidtable = $this->mBidtable;
1325                 $unidtable = $this->mUnidtable;
1326
1327                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1328                         return $this->mRules;
1329                 } else {
1330                         // display current variant in bidirectional array
1331                         $disp = $this->getTextInBidtable( $variant );
1332                         // or display current variant in fallbacks
1333                         if ( !$disp ) {
1334                                 $disp = $this->getTextInBidtable(
1335                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1336                         }
1337                         // or display current variant in unidirectional array
1338                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1339                                 $disp = array_values( $unidtable[$variant] );
1340                                 $disp = $disp[0];
1341                         }
1342                         // or display frist text under disable manual convert
1343                         if ( !$disp
1344                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1345                                 if ( count( $bidtable ) > 0 ) {
1346                                         $disp = array_values( $bidtable );
1347                                         $disp = $disp[0];
1348                                 } else {
1349                                         $disp = array_values( $unidtable );
1350                                         $disp = array_values( $disp[0] );
1351                                         $disp = $disp[0];
1352                                 }
1353                         }
1354                         return $disp;
1355                 }
1356         }
1357
1358         /**
1359          * Generate conversion table for all text.
1360          * @private
1361          */
1362         function generateConvTable() {
1363                 // Special case optimisation
1364                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1365                         $this->mConvTable = array();
1366                         return;
1367                 }
1368
1369                 $bidtable = $this->mBidtable;
1370                 $unidtable = $this->mUnidtable;
1371                 $manLevel = $this->mConverter->mManualLevel;
1372
1373                 $vmarked = array();
1374                 foreach ( $this->mConverter->mVariants as $v ) {
1375                         /* for bidirectional array
1376                                 fill in the missing variants, if any,
1377                                 with fallbacks */
1378                         if ( !isset( $bidtable[$v] ) ) {
1379                                 $variantFallbacks =
1380                                         $this->mConverter->getVariantFallbacks( $v );
1381                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1382                                 if ( $vf ) {
1383                                         $bidtable[$v] = $vf;
1384                                 }
1385                         }
1386
1387                         if ( isset( $bidtable[$v] ) ) {
1388                                 foreach ( $vmarked as $vo ) {
1389                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1390                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1391                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1392                                         // to introduce a custom mapping between
1393                                         // words WordZh and WordTw in the whole text
1394                                         if ( $manLevel[$v] == 'bidirectional' ) {
1395                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1396                                         }
1397                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1398                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1399                                         }
1400                                 }
1401                                 $vmarked[] = $v;
1402                         }
1403                         /* for unidirectional array fill to convert tables */
1404                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1405                                 && isset( $unidtable[$v] ) )
1406                         {
1407                                 if ( isset( $this->mConvTable[$v] ) ) {
1408                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1409                                 } else {
1410                                         $this->mConvTable[$v] = $unidtable[$v];
1411                                 }
1412                         }
1413                 }
1414         }
1415
1416         /**
1417          * Parse rules and flags.
1418          * @param $variant String: variant language code
1419          */
1420         public function parse( $variant = null ) {
1421                 if ( !$variant ) {
1422                         $variant = $this->mConverter->getPreferredVariant();
1423                 }
1424
1425                 $this->parseFlags();
1426                 $flags = $this->mFlags;
1427
1428                 // convert to specified variant
1429                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1430                 if ( $this->mVariantFlags ) {
1431                         // check if current variant in flags
1432                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1433                                 // then convert <text to convert> to current language
1434                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1435                                         $variant );
1436                         } else { // if current variant no in flags,
1437                                    // then we check its fallback variants.
1438                                 $variantFallbacks =
1439                                         $this->mConverter->getVariantFallbacks( $variant );
1440                                 if( is_array( $variantFallbacks ) ) {
1441                                         foreach ( $variantFallbacks as $variantFallback ) {
1442                                                 // if current variant's fallback exist in flags
1443                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1444                                                         // then convert <text to convert> to fallback language
1445                                                         $this->mRules =
1446                                                                 $this->mConverter->autoConvert( $this->mRules,
1447                                                                         $variantFallback );
1448                                                         break;
1449                                                 }
1450                                         }
1451                                 }
1452                         }
1453                         $this->mFlags = $flags = array( 'R' => true );
1454                 }
1455
1456                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1457                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1458                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1459                         $this->parseRules();
1460                 }
1461                 $rules = $this->mRules;
1462
1463                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1464                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1465                                 // fill all variants if text in -{A/H/-|text} without rules
1466                                 foreach ( $this->mConverter->mVariants as $v ) {
1467                                         $this->mBidtable[$v] = $rules;
1468                                 }
1469                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1470                                 $this->mFlags = $flags = array( 'R' => true );
1471                         }
1472                 }
1473
1474                 $this->mRuleDisplay = false;
1475                 foreach ( $flags as $flag => $unused ) {
1476                         switch ( $flag ) {
1477                                 case 'R':
1478                                         // if we don't do content convert, still strip the -{}- tags
1479                                         $this->mRuleDisplay = $rules;
1480                                         break;
1481                                 case 'N':
1482                                         // process N flag: output current variant name
1483                                         $ruleVar = trim( $rules );
1484                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1485                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1486                                         } else {
1487                                                 $this->mRuleDisplay = '';
1488                                         }
1489                                         break;
1490                                 case 'D':
1491                                         // process D flag: output rules description
1492                                         $this->mRuleDisplay = $this->getRulesDesc();
1493                                         break;
1494                                 case 'H':
1495                                         // process H,- flag or T only: output nothing
1496                                         $this->mRuleDisplay = '';
1497                                         break;
1498                                 case '-':
1499                                         $this->mRulesAction = 'remove';
1500                                         $this->mRuleDisplay = '';
1501                                         break;
1502                                 case '+':
1503                                         $this->mRulesAction = 'add';
1504                                         $this->mRuleDisplay = '';
1505                                         break;
1506                                 case 'S':
1507                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1508                                         break;
1509                                 case 'T':
1510                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1511                                         $this->mRuleDisplay = '';
1512                                         break;
1513                                 default:
1514                                         // ignore unknown flags (but see error case below)
1515                         }
1516                 }
1517                 if ( $this->mRuleDisplay === false ) {
1518                         $this->mRuleDisplay = '<span class="error">'
1519                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1520                                 . '</span>';
1521                 }
1522
1523                 $this->generateConvTable();
1524         }
1525
1526         /**
1527          * @todo FIXME: code this function :)
1528          */
1529         public function hasRules() {
1530                 // TODO:
1531         }
1532
1533         /**
1534          * Get display text on markup -{...}-
1535          * @return string
1536          */
1537         public function getDisplay() {
1538                 return $this->mRuleDisplay;
1539         }
1540
1541         /**
1542          * Get converted title.
1543          * @return string
1544          */
1545         public function getTitle() {
1546                 return $this->mRuleTitle;
1547         }
1548
1549         /**
1550          * Return how deal with conversion rules.
1551          * @return string
1552          */
1553         public function getRulesAction() {
1554                 return $this->mRulesAction;
1555         }
1556
1557         /**
1558          * Get conversion table. (bidirectional and unidirectional
1559          * conversion table)
1560          * @return array
1561          */
1562         public function getConvTable() {
1563                 return $this->mConvTable;
1564         }
1565
1566         /**
1567          * Get conversion rules string.
1568          * @return string
1569          */
1570         public function getRules() {
1571                 return $this->mRules;
1572         }
1573
1574         /**
1575          * Get conversion flags.
1576          * @return array
1577          */
1578         public function getFlags() {
1579                 return $this->mFlags;
1580         }
1581 }