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, $wgContLang;
 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                         if ( $this->mMainLanguageCode == $wgContLang->getCode() ) {
 263                                 $ret = $wgUser->getOption( 'variant' );
 264                         } else {
 265                                 $ret = $wgUser->getOption( 'variant-' . $this->mMainLanguageCode );
 266                         }
 267                 } else {
 268                         // figure out user lang without constructing wgLang to avoid
 269                         // infinite recursion
 270                         $ret = $wgUser->getOption( 'language' );
 271                 }
 272
 273                 return $this->mUserVariant = $this->validateVariant( $ret );
 274         }
 275
 276         /**
 277          * Determine the language variant from the Accept-Language header.
 278          *
 279          * @return Mixed: variant if one found, false otherwise.
 280          */
 281         protected function getHeaderVariant() {
 282                 global $wgRequest;
 283
 284                 if ( $this->mHeaderVariant ) {
 285                         return $this->mHeaderVariant;
 286                 }
 287
 288                 // see if some supported language variant is set in the
 289                 // HTTP header.
 290                 $languages = array_keys( $wgRequest->getAcceptLang() );
 291                 if ( empty( $languages ) ) {
 292                         return null;
 293                 }
 294
 295                 $fallbackLanguages = array();
 296                 foreach ( $languages as $language ) {
 297                         $this->mHeaderVariant = $this->validateVariant( $language );
 298                         if ( $this->mHeaderVariant ) {
 299                                 break;
 300                         }
 301
 302                         // To see if there are fallbacks of current language.
 303                         // We record these fallback variants, and process
 304                         // them later.
 305                         $fallbacks = $this->getVariantFallbacks( $language );
 306                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 307                                 $fallbackLanguages[] = $fallbacks;
 308                         } elseif ( is_array( $fallbacks ) ) {
 309                                 $fallbackLanguages =
 310                                         array_merge( $fallbackLanguages, $fallbacks );
 311                         }
 312                 }
 313
 314                 if ( !$this->mHeaderVariant ) {
 315                         // process fallback languages now
 316                         $fallback_languages = array_unique( $fallbackLanguages );
 317                         foreach ( $fallback_languages as $language ) {
 318                                 $this->mHeaderVariant = $this->validateVariant( $language );
 319                                 if ( $this->mHeaderVariant ) {
 320                                         break;
 321                                 }
 322                         }
 323                 }
 324
 325                 return $this->mHeaderVariant;
 326         }
 327
 328         /**
 329          * Dictionary-based conversion.
 330          * This function would not parse the conversion rules.
 331          * If you want to parse rules, try to use convert() or
 332          * convertTo().
 333          *
 334          * @param $text String the text to be converted
 335          * @param $toVariant bool|string the target language code
 336          * @return String the converted text
 337          */
 338         public function autoConvert( $text, $toVariant = false ) {
 339                 wfProfileIn( __METHOD__ );
 340
 341                 $this->loadTables();
 342
 343                 if ( !$toVariant ) {
 344                         $toVariant = $this->getPreferredVariant();
 345                         if ( !$toVariant ) {
 346                                 wfProfileOut( __METHOD__ );
 347                                 return $text;
 348                         }
 349                 }
 350
 351                 if ( $this->guessVariant( $text, $toVariant ) ) {
 352                         wfProfileOut( __METHOD__ );
 353                         return $text;
 354                 }
 355
 356                 /* we convert everything except:
 357                    1. HTML markups (anything between < and >)
 358                    2. HTML entities
 359                    3. placeholders created by the parser
 360                 */
 361                 global $wgParser;
 362                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 363                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 364                 } else {
 365                         $marker = '';
 366                 }
 367
 368                 // this one is needed when the text is inside an HTML markup
 369                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 370
 371                 // disable convert to variants between <code> tags
 372                 $codefix = '<code>.+?<\/code>|';
 373                 // disable conversion of <script> tags
 374                 $scriptfix = '<script.*?>.*?<\/script>|';
 375                 // disable conversion of <pre> tags
 376                 $prefix = '<pre.*?>.*?<\/pre>|';
 377
 378                 $reg = '/' . $codefix . $scriptfix . $prefix .
 379                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 380                 $startPos = 0;
 381                 $sourceBlob = '';
 382                 $literalBlob = '';
 383
 384                 // Guard against delimiter nulls in the input
 385                 $text = str_replace( "\000", '', $text );
 386
 387                 $markupMatches = null;
 388                 $elementMatches = null;
 389                 while ( $startPos < strlen( $text ) ) {
 390                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 391                                 $elementPos = $markupMatches[0][1];
 392                                 $element = $markupMatches[0][0];
 393                         } else {
 394                                 $elementPos = strlen( $text );
 395                                 $element = '';
 396                         }
 397
 398                         // Queue the part before the markup for translation in a batch
 399                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 400
 401                         // Advance to the next position
 402                         $startPos = $elementPos + strlen( $element );
 403
 404                         // Translate any alt or title attributes inside the matched element
 405                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 406                                 $elementMatches ) )
 407                         {
 408                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 409                                 $changed = false;
 410                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 411                                         if ( !isset( $attrs[$attrName] ) ) {
 412                                                 continue;
 413                                         }
 414                                         $attr = $attrs[$attrName];
 415                                         // Don't convert URLs
 416                                         if ( !strpos( $attr, '://' ) ) {
 417                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 418                                         }
 419
 420                                         // Remove HTML tags to avoid disrupting the layout
 421                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 422                                         if ( $attr !== $attrs[$attrName] ) {
 423                                                 $attrs[$attrName] = $attr;
 424                                                 $changed = true;
 425                                         }
 426                                 }
 427                                 if ( $changed ) {
 428                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 429                                                 $elementMatches[3];
 430                                 }
 431                         }
 432                         $literalBlob .= $element . "\000";
 433                 }
 434
 435                 // Do the main translation batch
 436                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 437
 438                 // Put the output back together
 439                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 440                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 441                 $output = '';
 442                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 443                         $output .= $translatedIter->current();
 444                         $output .= $literalIter->current();
 445                         $translatedIter->next();
 446                         $literalIter->next();
 447                 }
 448
 449                 wfProfileOut( __METHOD__ );
 450                 return $output;
 451         }
 452
 453         /**
 454          * Translate a string to a variant.
 455          * Doesn't parse rules or do any of that other stuff, for that use
 456          * convert() or convertTo().
 457          *
 458          * @param $text String: text to convert
 459          * @param $variant String: variant language code
 460          * @return String: translated text
 461          */
 462         public function translate( $text, $variant ) {
 463                 wfProfileIn( __METHOD__ );
 464                 // If $text is empty or only includes spaces, do nothing
 465                 // Otherwise translate it
 466                 if ( trim( $text ) ) {
 467                         $this->loadTables();
 468                         $text = $this->mTables[$variant]->replace( $text );
 469                 }
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Call translate() to convert text to all valid variants.
 476          *
 477          * @param $text String: the text to be converted
 478          * @return Array: variant => converted text
 479          */
 480         public function autoConvertToAllVariants( $text ) {
 481                 wfProfileIn( __METHOD__ );
 482                 $this->loadTables();
 483
 484                 $ret = array();
 485                 foreach ( $this->mVariants as $variant ) {
 486                         $ret[$variant] = $this->translate( $text, $variant );
 487                 }
 488
 489                 wfProfileOut( __METHOD__ );
 490                 return $ret;
 491         }
 492
 493         /**
 494          * Convert link text to all valid variants.
 495          * In the first, this function only convert text outside the
 496          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 497          * titles, the text will get all converted always.
 498          * So I removed this feature and deprecated the function.
 499          *
 500          * @param $text String: the text to be converted
 501          * @return Array: variant => converted text
 502          * @deprecated since 1.17 Use autoConvertToAllVariants() instead
 503          */
 504         public function convertLinkToAllVariants( $text ) {
 505                 return $this->autoConvertToAllVariants( $text );
 506         }
 507
 508         /**
 509          * Apply manual conversion rules.
 510          *
 511          * @param $convRule ConverterRule Object of ConverterRule
 512          */
 513         protected function applyManualConv( $convRule ) {
 514                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 515                 // title conversion.
 516                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 517                 // rule(s) not for title, this breaks the title conversion.
 518                 $newConvRuleTitle = $convRule->getTitle();
 519                 if ( $newConvRuleTitle ) {
 520                         // So I add an empty check for getTitle()
 521                         $this->mConvRuleTitle = $newConvRuleTitle;
 522                 }
 523
 524                 // merge/remove manual conversion rules to/from global table
 525                 $convTable = $convRule->getConvTable();
 526                 $action = $convRule->getRulesAction();
 527                 foreach ( $convTable as $variant => $pair ) {
 528                         if ( !$this->validateVariant( $variant ) ) {
 529                                 continue;
 530                         }
 531
 532                         if ( $action == 'add' ) {
 533                                 foreach ( $pair as $from => $to ) {
 534                                         // to ensure that $from and $to not be left blank
 535                                         // so $this->translate() could always return a string
 536                                         if ( $from || $to ) {
 537                                                 // more efficient than array_merge(), about 2.5 times.
 538                                                 $this->mTables[$variant]->setPair( $from, $to );
 539                                         }
 540                                 }
 541                         } elseif ( $action == 'remove' ) {
 542                                 $this->mTables[$variant]->removeArray( $pair );
 543                         }
 544                 }
 545         }
 546
 547         /**
 548          * Auto convert a Title object to a readable string in the
 549          * preferred variant.
 550          *
 551          * @param $title Title a object of Title
 552          * @return String: converted title text
 553          */
 554         public function convertTitle( $title ) {
 555                 $variant = $this->getPreferredVariant();
 556                 $index = $title->getNamespace();
 557                 if ( $index !== NS_MAIN ) {
 558                         $text = $this->convertNamespace( $index ) . ':';
 559                 } else {
 560                         $text = '';
 561                 }
 562                 $text .= $this->translate( $title->getText(), $variant );
 563                 return $text;
 564         }
 565
 566         /**
 567          * Get the namespace display name in the preferred variant.
 568          *
 569          * @param $index int namespace id
 570          * @return String: namespace name for display
 571          */
 572         public function convertNamespace( $index ) {
 573                 $variant = $this->getPreferredVariant();
 574                 if ( $index === NS_MAIN ) {
 575                         return '';
 576                 } else {
 577                         // First check if a message gives a converted name in the target variant.
 578                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 579                         if ( $nsConvMsg->exists() ) {
 580                                 return $nsConvMsg->plain();
 581                         }
 582                         // Then check if a message gives a converted name in content language
 583                         // which needs extra translation to the target variant.
 584                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 585                         if ( $nsConvMsg->exists() ) {
 586                                 return $this->translate( $nsConvMsg->plain(), $variant );
 587                         }
 588                         // No message exists, retrieve it from the target variant's namespace names.
 589                         $langObj = $this->mLangObj->factory( $variant );
 590                         return $langObj->getFormattedNsText( $index );
 591                 }
 592         }
 593
 594         /**
 595          * Convert text to different variants of a language. The automatic
 596          * conversion is done in autoConvert(). Here we parse the text
 597          * marked with -{}-, which specifies special conversions of the
 598          * text that can not be accomplished in autoConvert().
 599          *
 600          * Syntax of the markup:
 601          * -{code1:text1;code2:text2;...}-  or
 602          * -{flags|code1:text1;code2:text2;...}-  or
 603          * -{text}- in which case no conversion should take place for text
 604          *
 605          * @param $text String: text to be converted
 606          * @return String: converted text
 607          */
 608         public function convert( $text ) {
 609                 $variant = $this->getPreferredVariant();
 610                 return $this->convertTo( $text, $variant );
 611         }
 612
 613         /**
 614          * Same as convert() except a extra parameter to custom variant.
 615          *
 616          * @param $text String: text to be converted
 617          * @param $variant String: the target variant code
 618          * @return String: converted text
 619          */
 620         public function convertTo( $text, $variant ) {
 621                 global $wgDisableLangConversion;
 622                 if ( $wgDisableLangConversion ) {
 623                         return $text;
 624                 }
 625                 // Reset converter state for a new converter run.
 626                 $this->mConvRuleTitle = false;
 627                 return $this->recursiveConvertTopLevel( $text, $variant );
 628         }
 629
 630         /**
 631          * Recursively convert text on the outside. Allow to use nested
 632          * markups to custom rules.
 633          *
 634          * @param $text String: text to be converted
 635          * @param $variant String: the target variant code
 636          * @param $depth Integer: depth of recursion
 637          * @return String: converted text
 638          */
 639         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 640                 $startPos = 0;
 641                 $out = '';
 642                 $length = strlen( $text );
 643                 $shouldConvert = !$this->guessVariant( $text, $variant );
 644
 645                 while ( $startPos < $length ) {
 646                         $pos = strpos( $text, '-{', $startPos );
 647
 648                         if ( $pos === false ) {
 649                                 // No more markup, append final segment
 650                                 $fragment = substr( $text, $startPos );
 651                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 652                                 return $out;
 653                         }
 654
 655                         // Markup found
 656                         // Append initial segment
 657                         $fragment = substr( $text, $startPos, $pos - $startPos );
 658                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 659
 660                         // Advance position
 661                         $startPos = $pos;
 662
 663                         // Do recursive conversion
 664                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 665                 }
 666
 667                 return $out;
 668         }
 669
 670         /**
 671          * Recursively convert text on the inside.
 672          *
 673          * @param $text String: text to be converted
 674          * @param $variant String: the target variant code
 675          * @param $startPos int
 676          * @param $depth Integer: depth of recursion
 677          *
 678          * @throws MWException
 679          * @return String: converted text
 680          */
 681         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 682                 // Quick sanity check (no function calls)
 683                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 684                         throw new MWException( __METHOD__ . ': invalid input string' );
 685                 }
 686
 687                 $startPos += 2;
 688                 $inner = '';
 689                 $warningDone = false;
 690                 $length = strlen( $text );
 691
 692                 while ( $startPos < $length ) {
 693                         $m = false;
 694                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 695                         if ( !$m ) {
 696                                 // Unclosed rule
 697                                 break;
 698                         }
 699
 700                         $token = $m[0][0];
 701                         $pos = $m[0][1];
 702
 703                         // Markup found
 704                         // Append initial segment
 705                         $inner .= substr( $text, $startPos, $pos - $startPos );
 706
 707                         // Advance position
 708                         $startPos = $pos;
 709
 710                         switch ( $token ) {
 711                                 case '-{':
 712                                         // Check max depth
 713                                         if ( $depth >= $this->mMaxDepth ) {
 714                                                 $inner .= '-{';
 715                                                 if ( !$warningDone ) {
 716                                                         $inner .= '<span class="error">' .
 717                                                                 wfMessage( 'language-converter-depth-warning' )
 718                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 719                                                                 '</span>';
 720                                                         $warningDone = true;
 721                                                 }
 722                                                 $startPos += 2;
 723                                                 continue;
 724                                         }
 725                                         // Recursively parse another rule
 726                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 727                                         break;
 728                                 case '}-':
 729                                         // Apply the rule
 730                                         $startPos += 2;
 731                                         $rule = new ConverterRule( $inner, $this );
 732                                         $rule->parse( $variant );
 733                                         $this->applyManualConv( $rule );
 734                                         return $rule->getDisplay();
 735                                 default:
 736                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 737                         }
 738                 }
 739
 740                 // Unclosed rule
 741                 if ( $startPos < $length ) {
 742                         $inner .= substr( $text, $startPos );
 743                 }
 744                 $startPos = $length;
 745                 return '-{' . $this->autoConvert( $inner, $variant );
 746         }
 747
 748         /**
 749          * If a language supports multiple variants, it is possible that
 750          * non-existing link in one variant actually exists in another variant.
 751          * This function tries to find it. See e.g. LanguageZh.php
 752          *
 753          * @param $link String: the name of the link
 754          * @param $nt Mixed: the title object of the link
 755          * @param $ignoreOtherCond Boolean: to disable other conditions when
 756          *              we need to transclude a template or update a category's link
 757          * @return Null, the input parameters may be modified upon return
 758          */
 759         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 760                 # If the article has already existed, there is no need to
 761                 # check it again, otherwise it may cause a fault.
 762                 if ( is_object( $nt ) && $nt->exists() ) {
 763                         return;
 764                 }
 765
 766                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 767                         $wgUser;
 768                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 769                 $action = $wgRequest->getText( 'action' );
 770                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 771                 $disableLinkConversion = $wgDisableLangConversion
 772                         || $wgDisableTitleConversion;
 773                 $linkBatch = new LinkBatch();
 774
 775                 $ns = NS_MAIN;
 776
 777                 if ( $disableLinkConversion ||
 778                          ( !$ignoreOtherCond &&
 779                            ( $isredir == 'no'
 780                                  || $action == 'edit'
 781                                  || $action == 'submit'
 782                                  || $linkconvert == 'no'
 783                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 784                         return;
 785                 }
 786
 787                 if ( is_object( $nt ) ) {
 788                         $ns = $nt->getNamespace();
 789                 }
 790
 791                 $variants = $this->autoConvertToAllVariants( $link );
 792                 if ( !$variants ) { // give up
 793                         return;
 794                 }
 795
 796                 $titles = array();
 797
 798                 foreach ( $variants as $v ) {
 799                         if ( $v != $link ) {
 800                                 $varnt = Title::newFromText( $v, $ns );
 801                                 if ( !is_null( $varnt ) ) {
 802                                         $linkBatch->addObj( $varnt );
 803                                         $titles[] = $varnt;
 804                                 }
 805                         }
 806                 }
 807
 808                 // fetch all variants in single query
 809                 $linkBatch->execute();
 810
 811                 foreach ( $titles as $varnt ) {
 812                         if ( $varnt->getArticleID() > 0 ) {
 813                                 $nt = $varnt;
 814                                 $link = $varnt->getText();
 815                                 break;
 816                         }
 817                 }
 818         }
 819
 820         /**
 821          * Returns language specific hash options.
 822          *
 823          * @return string
 824          */
 825         public function getExtraHashOptions() {
 826                 $variant = $this->getPreferredVariant();
 827                 return '!' . $variant;
 828         }
 829
 830         /**
 831          * Guess if a text is written in a variant. This should be implemented in subclasses.
 832          *
 833          * @param string $text the text to be checked
 834          * @param string $variant language code of the variant to be checked for
 835          * @return bool true if $text appears to be written in $variant, false if not
 836          *
 837          * @author Nikola Smolenski <smolensk@eunet.rs>
 838          * @since 1.19
 839          */
 840         public function guessVariant( $text, $variant ) {
 841                 return false;
 842         }
 843
 844         /**
 845          * Load default conversion tables.
 846          * This method must be implemented in derived class.
 847          *
 848          * @private
 849          * @throws MWException
 850          */
 851         function loadDefaultTables() {
 852                 $name = get_class( $this );
 853                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 854         }
 855
 856         /**
 857          * Load conversion tables either from the cache or the disk.
 858          * @private
 859          * @param $fromCache Boolean: load from memcached? Defaults to true.
 860          */
 861         function loadTables( $fromCache = true ) {
 862                 global $wgLangConvMemc;
 863
 864                 if ( $this->mTablesLoaded ) {
 865                         return;
 866                 }
 867
 868                 wfProfileIn( __METHOD__ );
 869                 $this->mTablesLoaded = true;
 870                 $this->mTables = false;
 871                 if ( $fromCache ) {
 872                         wfProfileIn( __METHOD__ . '-cache' );
 873                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 874                         wfProfileOut( __METHOD__ . '-cache' );
 875                 }
 876                 if ( !$this->mTables
 877                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 878                         wfProfileIn( __METHOD__ . '-recache' );
 879                         // not in cache, or we need a fresh reload.
 880                         // We will first load the default tables
 881                         // then update them using things in MediaWiki:Conversiontable/*
 882                         $this->loadDefaultTables();
 883                         foreach ( $this->mVariants as $var ) {
 884                                 $cached = $this->parseCachedTable( $var );
 885                                 $this->mTables[$var]->mergeArray( $cached );
 886                         }
 887
 888                         $this->postLoadTables();
 889                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 890
 891                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 892                         wfProfileOut( __METHOD__ . '-recache' );
 893                 }
 894                 wfProfileOut( __METHOD__ );
 895         }
 896
 897         /**
 898          * Hook for post processing after conversion tables are loaded.
 899          */
 900         function postLoadTables() { }
 901
 902         /**
 903          * Reload the conversion tables.
 904          *
 905          * @private
 906          */
 907         function reloadTables() {
 908                 if ( $this->mTables ) {
 909                         unset( $this->mTables );
 910                 }
 911                 $this->mTablesLoaded = false;
 912                 $this->loadTables( false );
 913         }
 914
 915         /**
 916          * Parse the conversion table stored in the cache.
 917          *
 918          * The tables should be in blocks of the following form:
 919          *              -{
 920          *                      word => word ;
 921          *                      word => word ;
 922          *                      ...
 923          *              }-
 924          *
 925          * To make the tables more manageable, subpages are allowed
 926          * and will be parsed recursively if $recursive == true.
 927          *
 928          * @param $code String: language code
 929          * @param $subpage String: subpage name
 930          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 931          *
 932          * @return array
 933          */
 934         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 935                 static $parsed = array();
 936
 937                 $key = 'Conversiontable/' . $code;
 938                 if ( $subpage ) {
 939                         $key .= '/' . $subpage;
 940                 }
 941                 if ( array_key_exists( $key, $parsed ) ) {
 942                         return array();
 943                 }
 944
 945                 $parsed[$key] = true;
 946
 947                 if ( $subpage === '' ) {
 948                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 949                 } else {
 950                         $txt = false;
 951                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 952                         if ( $title && $title->exists() ) {
 953                                 $revision = Revision::newFromTitle( $title );
 954                                 if ( $revision ) {
 955                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 956                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 957                                         }
 958
 959                                         // @todo in the future, use a specialized content model, perhaps based on json!
 960                                 }
 961                         }
 962                 }
 963
 964                 # Nothing to parse if there's no text
 965                 if ( $txt === false || $txt === null || $txt === '' ) {
 966                         return array();
 967                 }
 968
 969                 // get all subpage links of the form
 970                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 971                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 972                         ':Conversiontable';
 973                 $subs = StringUtils::explode( '[[', $txt );
 974                 $sublinks = array();
 975                 foreach ( $subs as $sub ) {
 976                         $link = explode( ']]', $sub, 2 );
 977                         if ( count( $link ) != 2 ) {
 978                                 continue;
 979                         }
 980                         $b = explode( '|', $link[0], 2 );
 981                         $b = explode( '/', trim( $b[0] ), 3 );
 982                         if ( count( $b ) == 3 ) {
 983                                 $sublink = $b[2];
 984                         } else {
 985                                 $sublink = '';
 986                         }
 987
 988                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 989                                 $sublinks[] = $sublink;
 990                         }
 991                 }
 992
 993                 // parse the mappings in this page
 994                 $blocks = StringUtils::explode( '-{', $txt );
 995                 $ret = array();
 996                 $first = true;
 997                 foreach ( $blocks as $block ) {
 998                         if ( $first ) {
 999                                 // Skip the part before the first -{
1000                                 $first = false;
1001                                 continue;
1002                         }
1003                         $mappings = explode( '}-', $block, 2 );
1004                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
1005                                                                          $mappings[0] );
1006                         $table = StringUtils::explode( ';', $stripped );
1007                         foreach ( $table as $t ) {
1008                                 $m = explode( '=>', $t, 3 );
1009                                 if ( count( $m ) != 2 ) {
1010                                         continue;
1011                                 }
1012                                 // trim any trailling comments starting with '//'
1013                                 $tt = explode( '//', $m[1], 2 );
1014                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1015                         }
1016                 }
1017
1018                 // recursively parse the subpages
1019                 if ( $recursive ) {
1020                         foreach ( $sublinks as $link ) {
1021                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1022                                 $ret = array_merge( $ret, $s );
1023                         }
1024                 }
1025
1026                 if ( $this->mUcfirst ) {
1027                         foreach ( $ret as $k => $v ) {
1028                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1029                         }
1030                 }
1031                 return $ret;
1032         }
1033
1034         /**
1035          * Enclose a string with the "no conversion" tag. This is used by
1036          * various functions in the Parser.
1037          *
1038          * @param $text String: text to be tagged for no conversion
1039          * @param $noParse Boolean: unused
1040          * @return String: the tagged text
1041          */
1042         public function markNoConversion( $text, $noParse = false ) {
1043                 # don't mark if already marked
1044                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1045                         return $text;
1046                 }
1047
1048                 $ret = "-{R|$text}-";
1049                 return $ret;
1050         }
1051
1052         /**
1053          * Convert the sorting key for category links. This should make different
1054          * keys that are variants of each other map to the same key.
1055          *
1056          * @param $key string
1057          *
1058          * @return string
1059          */
1060         function convertCategoryKey( $key ) {
1061                 return $key;
1062         }
1063
1064         /**
1065          * Hook to refresh the cache of conversion tables when
1066          * MediaWiki:Conversiontable* is updated.
1067          * @private
1068          *
1069          * @param $page WikiPage object
1070          * @param $user Object: User object for the current user
1071          * @param $content Content: new page content
1072          * @param $summary String: edit summary of the edit
1073          * @param $isMinor Boolean: was the edit marked as minor?
1074          * @param $isWatch Boolean: did the user watch this page or not?
1075          * @param $section
1076          * @param $flags int Bitfield
1077          * @param $revision Object: new Revision object or null
1078          * @return Boolean: true
1079          */
1080         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1081                         $isWatch, $section, $flags, $revision ) {
1082                 $titleobj = $page->getTitle();
1083                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1084                         $title = $titleobj->getDBkey();
1085                         $t = explode( '/', $title, 3 );
1086                         $c = count( $t );
1087                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1088                                 if ( $this->validateVariant( $t[1] ) ) {
1089                                         $this->reloadTables();
1090                                 }
1091                         }
1092                 }
1093                 return true;
1094         }
1095
1096         /**
1097          * Armour rendered math against conversion.
1098          * Escape special chars in parsed math text. (in most cases are img elements)
1099          *
1100          * @param $text String: text to armour against conversion
1101          * @return String: armoured text where { and } have been converted to
1102          *                 &#123; and &#125;
1103          */
1104         public function armourMath( $text ) {
1105                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1106                 // any unwanted markup appearing in the math image tag.
1107                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1108                 return $text;
1109         }
1110
1111         /**
1112          * Get the cached separator pattern for ConverterRule::parseRules()
1113          */
1114         function getVarSeparatorPattern() {
1115                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1116                         // varsep_pattern for preg_split:
1117                         // text should be splited by ";" only if a valid variant
1118                         // name exist after the markup, for example:
1119                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1120                         //      <span style="font-size:120%;">yyy</span>;}-
1121                         // we should split it as:
1122                         //  array(
1123                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1124                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1125                         //        [2] => ''
1126                         //       )
1127                         $pat = '/;\s*(?=';
1128                         foreach ( $this->mVariants as $variant ) {
1129                                 // zh-hans:xxx;zh-hant:yyy
1130                                 $pat .= $variant . '\s*:|';
1131                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1132                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1133                         }
1134                         $pat .= '\s*$)/';
1135                         $this->mVarSeparatorPattern = $pat;
1136                 }
1137                 return $this->mVarSeparatorPattern;
1138         }
1139 }
1140
1141 /**
1142  * Parser for rules of language conversion , parse rules in -{ }- tag.
1143  * @ingroup Language
1144  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1145  */
1146 class ConverterRule {
1147         public $mText; // original text in -{text}-
1148         public $mConverter; // LanguageConverter object
1149         public $mRuleDisplay = '';
1150         public $mRuleTitle = false;
1151         public $mRules = '';// string : the text of the rules
1152         public $mRulesAction = 'none';
1153         public $mFlags = array();
1154         public $mVariantFlags = array();
1155         public $mConvTable = array();
1156         public $mBidtable = array();// array of the translation in each variant
1157         public $mUnidtable = array();// array of the translation in each variant
1158
1159         /**
1160          * Constructor
1161          *
1162          * @param $text String: the text between -{ and }-
1163          * @param $converter LanguageConverter object
1164          */
1165         public function __construct( $text, $converter ) {
1166                 $this->mText = $text;
1167                 $this->mConverter = $converter;
1168         }
1169
1170         /**
1171          * Check if variants array in convert array.
1172          *
1173          * @param $variants Array or string: variant language code
1174          * @return String: translated text
1175          */
1176         public function getTextInBidtable( $variants ) {
1177                 $variants = (array)$variants;
1178                 if ( !$variants ) {
1179                         return false;
1180                 }
1181                 foreach ( $variants as $variant ) {
1182                         if ( isset( $this->mBidtable[$variant] ) ) {
1183                                 return $this->mBidtable[$variant];
1184                         }
1185                 }
1186                 return false;
1187         }
1188
1189         /**
1190          * Parse flags with syntax -{FLAG| ... }-
1191          * @private
1192          */
1193         function parseFlags() {
1194                 $text = $this->mText;
1195                 $flags = array();
1196                 $variantFlags = array();
1197
1198                 $sepPos = strpos( $text, '|' );
1199                 if ( $sepPos !== false ) {
1200                         $validFlags = $this->mConverter->mFlags;
1201                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1202                         foreach ( $f as $ff ) {
1203                                 $ff = trim( $ff );
1204                                 if ( isset( $validFlags[$ff] ) ) {
1205                                         $flags[$validFlags[$ff]] = true;
1206                                 }
1207                         }
1208                         $text = strval( substr( $text, $sepPos + 1 ) );
1209                 }
1210
1211                 if ( !$flags ) {
1212                         $flags['S'] = true;
1213                 } elseif ( isset( $flags['R'] ) ) {
1214                         $flags = array( 'R' => true );// remove other flags
1215                 } elseif ( isset( $flags['N'] ) ) {
1216                         $flags = array( 'N' => true );// remove other flags
1217                 } elseif ( isset( $flags['-'] ) ) {
1218                         $flags = array( '-' => true );// remove other flags
1219                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1220                         $flags['H'] = true;
1221                 } elseif ( isset( $flags['H'] ) ) {
1222                         // replace A flag, and remove other flags except T
1223                         $temp = array( '+' => true, 'H' => true );
1224                         if ( isset( $flags['T'] ) ) {
1225                                 $temp['T'] = true;
1226                         }
1227                         if ( isset( $flags['D'] ) ) {
1228                                 $temp['D'] = true;
1229                         }
1230                         $flags = $temp;
1231                 } else {
1232                         if ( isset( $flags['A'] ) ) {
1233                                 $flags['+'] = true;
1234                                 $flags['S'] = true;
1235                         }
1236                         if ( isset( $flags['D'] ) ) {
1237                                 unset( $flags['S'] );
1238                         }
1239                         // try to find flags like "zh-hans", "zh-hant"
1240                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1241                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1242                         if ( $variantFlags ) {
1243                                 $variantFlags = array_flip( $variantFlags );
1244                                 $flags = array();
1245                         }
1246                 }
1247                 $this->mVariantFlags = $variantFlags;
1248                 $this->mRules = $text;
1249                 $this->mFlags = $flags;
1250         }
1251
1252         /**
1253          * Generate conversion table.
1254          * @private
1255          */
1256         function parseRules() {
1257                 $rules = $this->mRules;
1258                 $bidtable = array();
1259                 $unidtable = array();
1260                 $variants = $this->mConverter->mVariants;
1261                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1262
1263                 $choice = preg_split( $varsep_pattern, $rules );
1264
1265                 foreach ( $choice as $c ) {
1266                         $v = explode( ':', $c, 2 );
1267                         if ( count( $v ) != 2 ) {
1268                                 // syntax error, skip
1269                                 continue;
1270                         }
1271                         $to = trim( $v[1] );
1272                         $v = trim( $v[0] );
1273                         $u = explode( '=>', $v, 2 );
1274                         // if $to is empty, strtr() could return a wrong result
1275                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1276                                 $bidtable[$v] = $to;
1277                         } elseif ( count( $u ) == 2 ) {
1278                                 $from = trim( $u[0] );
1279                                 $v = trim( $u[1] );
1280                                 if ( array_key_exists( $v, $unidtable )
1281                                          && !is_array( $unidtable[$v] )
1282                                          && $to
1283                                          && in_array( $v, $variants ) ) {
1284                                         $unidtable[$v] = array( $from => $to );
1285                                 } elseif ( $to && in_array( $v, $variants ) ) {
1286                                         $unidtable[$v][$from] = $to;
1287                                 }
1288                         }
1289                         // syntax error, pass
1290                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1291                                 $bidtable = array();
1292                                 $unidtable = array();
1293                                 break;
1294                         }
1295                 }
1296                 $this->mBidtable = $bidtable;
1297                 $this->mUnidtable = $unidtable;
1298         }
1299
1300         /**
1301          * @private
1302          *
1303          * @return string
1304          */
1305         function getRulesDesc() {
1306                 $codesep = $this->mConverter->mDescCodeSep;
1307                 $varsep = $this->mConverter->mDescVarSep;
1308                 $text = '';
1309                 foreach ( $this->mBidtable as $k => $v ) {
1310                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1311                 }
1312                 foreach ( $this->mUnidtable as $k => $a ) {
1313                         foreach ( $a as $from => $to ) {
1314                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1315                                         "$codesep$to$varsep";
1316                         }
1317                 }
1318                 return $text;
1319         }
1320
1321         /**
1322          * Parse rules conversion.
1323          * @private
1324          *
1325          * @param $variant
1326          *
1327          * @return string
1328          */
1329         function getRuleConvertedStr( $variant ) {
1330                 $bidtable = $this->mBidtable;
1331                 $unidtable = $this->mUnidtable;
1332
1333                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1334                         return $this->mRules;
1335                 } else {
1336                         // display current variant in bidirectional array
1337                         $disp = $this->getTextInBidtable( $variant );
1338                         // or display current variant in fallbacks
1339                         if ( !$disp ) {
1340                                 $disp = $this->getTextInBidtable(
1341                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1342                         }
1343                         // or display current variant in unidirectional array
1344                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1345                                 $disp = array_values( $unidtable[$variant] );
1346                                 $disp = $disp[0];
1347                         }
1348                         // or display frist text under disable manual convert
1349                         if ( !$disp
1350                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1351                                 if ( count( $bidtable ) > 0 ) {
1352                                         $disp = array_values( $bidtable );
1353                                         $disp = $disp[0];
1354                                 } else {
1355                                         $disp = array_values( $unidtable );
1356                                         $disp = array_values( $disp[0] );
1357                                         $disp = $disp[0];
1358                                 }
1359                         }
1360                         return $disp;
1361                 }
1362         }
1363
1364         /**
1365          * Similar to getRuleConvertedStr(), but this prefers to use original
1366          * page title if $variant === $this->mConverter->mMainLanguageCode
1367          * and may return false in this case (so this title conversion rule
1368          * will be ignored and the original title is shown).
1369          *
1370          * @since 1.22
1371          * @param $variant The variant code to display page title in
1372          * @return String|false The converted title or false if just page name
1373          */
1374         function getRuleConvertedTitle( $variant ) {
1375                 if ( $variant === $this->mConverter->mMainLanguageCode ) {
1376                         // If a string targeting exactly this variant is set,
1377                         // use it. Otherwise, just return false, so the real
1378                         // page name can be shown (and because variant === main,
1379                         // there'll be no further automatic conversion).
1380                         $disp = $this->getTextInBidtable( $variant );
1381                         if ( $disp ) {
1382                                 return $disp;
1383                         }
1384                         if ( array_key_exists( $variant, $this->mUnidtable ) ) {
1385                                 $disp = array_values( $this->mUnidtable[$variant] );
1386                                 $disp = $disp[0];
1387                         }
1388                         // Assigned above or still false.
1389                         return $disp;
1390                 } else {
1391                         return $this->getRuleConvertedStr( $variant );
1392                 }
1393         }
1394
1395         /**
1396          * Generate conversion table for all text.
1397          * @private
1398          */
1399         function generateConvTable() {
1400                 // Special case optimisation
1401                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1402                         $this->mConvTable = array();
1403                         return;
1404                 }
1405
1406                 $bidtable = $this->mBidtable;
1407                 $unidtable = $this->mUnidtable;
1408                 $manLevel = $this->mConverter->mManualLevel;
1409
1410                 $vmarked = array();
1411                 foreach ( $this->mConverter->mVariants as $v ) {
1412                         /* for bidirectional array
1413                                 fill in the missing variants, if any,
1414                                 with fallbacks */
1415                         if ( !isset( $bidtable[$v] ) ) {
1416                                 $variantFallbacks =
1417                                         $this->mConverter->getVariantFallbacks( $v );
1418                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1419                                 if ( $vf ) {
1420                                         $bidtable[$v] = $vf;
1421                                 }
1422                         }
1423
1424                         if ( isset( $bidtable[$v] ) ) {
1425                                 foreach ( $vmarked as $vo ) {
1426                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1427                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1428                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1429                                         // to introduce a custom mapping between
1430                                         // words WordZh and WordTw in the whole text
1431                                         if ( $manLevel[$v] == 'bidirectional' ) {
1432                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1433                                         }
1434                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1435                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1436                                         }
1437                                 }
1438                                 $vmarked[] = $v;
1439                         }
1440                         /* for unidirectional array fill to convert tables */
1441                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1442                                 && isset( $unidtable[$v] ) )
1443                         {
1444                                 if ( isset( $this->mConvTable[$v] ) ) {
1445                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1446                                 } else {
1447                                         $this->mConvTable[$v] = $unidtable[$v];
1448                                 }
1449                         }
1450                 }
1451         }
1452
1453         /**
1454          * Parse rules and flags.
1455          * @param $variant String: variant language code
1456          */
1457         public function parse( $variant = null ) {
1458                 if ( !$variant ) {
1459                         $variant = $this->mConverter->getPreferredVariant();
1460                 }
1461
1462                 $this->parseFlags();
1463                 $flags = $this->mFlags;
1464
1465                 // convert to specified variant
1466                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1467                 if ( $this->mVariantFlags ) {
1468                         // check if current variant in flags
1469                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1470                                 // then convert <text to convert> to current language
1471                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1472                                         $variant );
1473                         } else { // if current variant no in flags,
1474                                    // then we check its fallback variants.
1475                                 $variantFallbacks =
1476                                         $this->mConverter->getVariantFallbacks( $variant );
1477                                 if ( is_array( $variantFallbacks ) ) {
1478                                         foreach ( $variantFallbacks as $variantFallback ) {
1479                                                 // if current variant's fallback exist in flags
1480                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1481                                                         // then convert <text to convert> to fallback language
1482                                                         $this->mRules =
1483                                                                 $this->mConverter->autoConvert( $this->mRules,
1484                                                                         $variantFallback );
1485                                                         break;
1486                                                 }
1487                                         }
1488                                 }
1489                         }
1490                         $this->mFlags = $flags = array( 'R' => true );
1491                 }
1492
1493                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1494                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1495                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1496                         $this->parseRules();
1497                 }
1498                 $rules = $this->mRules;
1499
1500                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1501                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1502                                 // fill all variants if text in -{A/H/-|text} without rules
1503                                 foreach ( $this->mConverter->mVariants as $v ) {
1504                                         $this->mBidtable[$v] = $rules;
1505                                 }
1506                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1507                                 $this->mFlags = $flags = array( 'R' => true );
1508                         }
1509                 }
1510
1511                 $this->mRuleDisplay = false;
1512                 foreach ( $flags as $flag => $unused ) {
1513                         switch ( $flag ) {
1514                                 case 'R':
1515                                         // if we don't do content convert, still strip the -{}- tags
1516                                         $this->mRuleDisplay = $rules;
1517                                         break;
1518                                 case 'N':
1519                                         // process N flag: output current variant name
1520                                         $ruleVar = trim( $rules );
1521                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1522                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1523                                         } else {
1524                                                 $this->mRuleDisplay = '';
1525                                         }
1526                                         break;
1527                                 case 'D':
1528                                         // process D flag: output rules description
1529                                         $this->mRuleDisplay = $this->getRulesDesc();
1530                                         break;
1531                                 case 'H':
1532                                         // process H,- flag or T only: output nothing
1533                                         $this->mRuleDisplay = '';
1534                                         break;
1535                                 case '-':
1536                                         $this->mRulesAction = 'remove';
1537                                         $this->mRuleDisplay = '';
1538                                         break;
1539                                 case '+':
1540                                         $this->mRulesAction = 'add';
1541                                         $this->mRuleDisplay = '';
1542                                         break;
1543                                 case 'S':
1544                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1545                                         break;
1546                                 case 'T':
1547                                         $this->mRuleTitle = $this->getRuleConvertedTitle( $variant );
1548                                         $this->mRuleDisplay = '';
1549                                         break;
1550                                 default:
1551                                         // ignore unknown flags (but see error case below)
1552                         }
1553                 }
1554                 if ( $this->mRuleDisplay === false ) {
1555                         $this->mRuleDisplay = '<span class="error">'
1556                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1557                                 . '</span>';
1558                 }
1559
1560                 $this->generateConvTable();
1561         }
1562
1563         /**
1564          * @todo FIXME: code this function :)
1565          */
1566         public function hasRules() {
1567                 // TODO:
1568         }
1569
1570         /**
1571          * Get display text on markup -{...}-
1572          * @return string
1573          */
1574         public function getDisplay() {
1575                 return $this->mRuleDisplay;
1576         }
1577
1578         /**
1579          * Get converted title.
1580          * @return string
1581          */
1582         public function getTitle() {
1583                 return $this->mRuleTitle;
1584         }
1585
1586         /**
1587          * Return how deal with conversion rules.
1588          * @return string
1589          */
1590         public function getRulesAction() {
1591                 return $this->mRulesAction;
1592         }
1593
1594         /**
1595          * Get conversion table. (bidirectional and unidirectional
1596          * conversion table)
1597          * @return array
1598          */
1599         public function getConvTable() {
1600                 return $this->mConvTable;
1601         }
1602
1603         /**
1604          * Get conversion rules string.
1605          * @return string
1606          */
1607         public function getRules() {
1608                 return $this->mRules;
1609         }
1610
1611         /**
1612          * Get conversion flags.
1613          * @return array
1614          */
1615         public function getFlags() {
1616                 return $this->mFlags;
1617         }
1618 }