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 7';
  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 (but no display in placed code)
 106                         'N' => 'N'        // current variant name
 107                 );
 108                 $this->mFlags = array_merge( $defaultflags, $flags );
 109                 foreach ( $this->mVariants as $v ) {
 110                         if ( array_key_exists( $v, $manualLevel ) ) {
 111                                 $this->mManualLevel[$v] = $manualLevel[$v];
 112                         } else {
 113                                 $this->mManualLevel[$v] = 'bidirectional';
 114                         }
 115                         $this->mFlags[$v] = $v;
 116                 }
 117         }
 118
 119         /**
 120          * Get all valid variants.
 121          * Call this instead of using $this->mVariants directly.
 122          *
 123          * @return Array: contains all valid variants
 124          */
 125         public function getVariants() {
 126                 return $this->mVariants;
 127         }
 128
 129         /**
 130          * In case some variant is not defined in the markup, we need
 131          * to have some fallback. For example, in zh, normally people
 132          * will define zh-hans and zh-hant, but less so for zh-sg or zh-hk.
 133          * when zh-sg is preferred but not defined, we will pick zh-hans
 134          * in this case. Right now this is only used by zh.
 135          *
 136          * @param $variant String: the language code of the variant
 137          * @return String|array: The code of the fallback language or the
 138          *                               main code if there is no fallback
 139          */
 140         public function getVariantFallbacks( $variant ) {
 141                 if ( isset( $this->mVariantFallbacks[$variant] ) ) {
 142                         return $this->mVariantFallbacks[$variant];
 143                 }
 144                 return $this->mMainLanguageCode;
 145         }
 146
 147         /**
 148          * Get the title produced by the conversion rule.
 149          * @return String: The converted title text
 150          */
 151         public function getConvRuleTitle() {
 152                 return $this->mConvRuleTitle;
 153         }
 154
 155         /**
 156          * Get preferred language variant.
 157          * @return String: the preferred language code
 158          */
 159         public function getPreferredVariant() {
 160                 global $wgDefaultLanguageVariant, $wgUser;
 161
 162                 $req = $this->getURLVariant();
 163
 164                 if ( $wgUser->isLoggedIn() && !$req ) {
 165                         $req = $this->getUserVariant();
 166                 } elseif ( !$req ) {
 167                         $req = $this->getHeaderVariant();
 168                 }
 169
 170                 if ( $wgDefaultLanguageVariant && !$req ) {
 171                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 172                 }
 173
 174                 // This function, unlike the other get*Variant functions, is
 175                 // not memoized (i.e. there return value is not cached) since
 176                 // new information might appear during processing after this
 177                 // is first called.
 178                 if ( $this->validateVariant( $req ) ) {
 179                         return $req;
 180                 }
 181                 return $this->mMainLanguageCode;
 182         }
 183
 184         /**
 185          * Get default variant.
 186          * This function would not be affected by user's settings
 187          * @return String: the default variant code
 188          */
 189         public function getDefaultVariant() {
 190                 global $wgDefaultLanguageVariant;
 191
 192                 $req = $this->getURLVariant();
 193
 194                 if ( !$req ) {
 195                         $req = $this->getHeaderVariant();
 196                 }
 197
 198                 if ( $wgDefaultLanguageVariant && !$req ) {
 199                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 200                 }
 201
 202                 if ( $req ) {
 203                         return $req;
 204                 }
 205                 return $this->mMainLanguageCode;
 206         }
 207
 208         /**
 209          * Validate the variant
 210          * @param $variant String: the variant to validate
 211          * @return Mixed: returns the variant if it is valid, null otherwise
 212          */
 213         public function validateVariant( $variant = null ) {
 214                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 215                         return $variant;
 216                 }
 217                 return null;
 218         }
 219
 220         /**
 221          * Get the variant specified in the URL
 222          *
 223          * @return Mixed: variant if one found, false otherwise.
 224          */
 225         public function getURLVariant() {
 226                 global $wgRequest;
 227
 228                 if ( $this->mURLVariant ) {
 229                         return $this->mURLVariant;
 230                 }
 231
 232                 // see if the preference is set in the request
 233                 $ret = $wgRequest->getText( 'variant' );
 234
 235                 if ( !$ret ) {
 236                         $ret = $wgRequest->getVal( 'uselang' );
 237                 }
 238
 239                 $this->mURLVariant = $this->validateVariant( $ret );
 240                 return $this->mURLVariant;
 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                 $this->mUserVariant = $this->validateVariant( $ret );
 274                 return $this->mUserVariant;
 275         }
 276
 277         /**
 278          * Determine the language variant from the Accept-Language header.
 279          *
 280          * @return Mixed: variant if one found, false otherwise.
 281          */
 282         protected function getHeaderVariant() {
 283                 global $wgRequest;
 284
 285                 if ( $this->mHeaderVariant ) {
 286                         return $this->mHeaderVariant;
 287                 }
 288
 289                 // see if some supported language variant is set in the
 290                 // HTTP header.
 291                 $languages = array_keys( $wgRequest->getAcceptLang() );
 292                 if ( empty( $languages ) ) {
 293                         return null;
 294                 }
 295
 296                 $fallbackLanguages = array();
 297                 foreach ( $languages as $language ) {
 298                         $this->mHeaderVariant = $this->validateVariant( $language );
 299                         if ( $this->mHeaderVariant ) {
 300                                 break;
 301                         }
 302
 303                         // To see if there are fallbacks of current language.
 304                         // We record these fallback variants, and process
 305                         // them later.
 306                         $fallbacks = $this->getVariantFallbacks( $language );
 307                         if ( is_string( $fallbacks ) && $fallbacks !== $this->mMainLanguageCode ) {
 308                                 $fallbackLanguages[] = $fallbacks;
 309                         } elseif ( is_array( $fallbacks ) ) {
 310                                 $fallbackLanguages =
 311                                         array_merge( $fallbackLanguages, $fallbacks );
 312                         }
 313                 }
 314
 315                 if ( !$this->mHeaderVariant ) {
 316                         // process fallback languages now
 317                         $fallback_languages = array_unique( $fallbackLanguages );
 318                         foreach ( $fallback_languages as $language ) {
 319                                 $this->mHeaderVariant = $this->validateVariant( $language );
 320                                 if ( $this->mHeaderVariant ) {
 321                                         break;
 322                                 }
 323                         }
 324                 }
 325
 326                 return $this->mHeaderVariant;
 327         }
 328
 329         /**
 330          * Dictionary-based conversion.
 331          * This function would not parse the conversion rules.
 332          * If you want to parse rules, try to use convert() or
 333          * convertTo().
 334          *
 335          * @param $text String the text to be converted
 336          * @param $toVariant bool|string the target language code
 337          * @return String the converted text
 338          */
 339         public function autoConvert( $text, $toVariant = false ) {
 340                 wfProfileIn( __METHOD__ );
 341
 342                 $this->loadTables();
 343
 344                 if ( !$toVariant ) {
 345                         $toVariant = $this->getPreferredVariant();
 346                         if ( !$toVariant ) {
 347                                 wfProfileOut( __METHOD__ );
 348                                 return $text;
 349                         }
 350                 }
 351
 352                 if ( $this->guessVariant( $text, $toVariant ) ) {
 353                         wfProfileOut( __METHOD__ );
 354                         return $text;
 355                 }
 356
 357                 /* we convert everything except:
 358                    1. HTML markups (anything between < and >)
 359                    2. HTML entities
 360                    3. placeholders created by the parser
 361                 */
 362                 global $wgParser;
 363                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 364                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 365                 } else {
 366                         $marker = '';
 367                 }
 368
 369                 // this one is needed when the text is inside an HTML markup
 370                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 371
 372                 // disable convert to variants between <code> tags
 373                 $codefix = '<code>.+?<\/code>|';
 374                 // disable conversion of <script> tags
 375                 $scriptfix = '<script.*?>.*?<\/script>|';
 376                 // disable conversion of <pre> tags
 377                 $prefix = '<pre.*?>.*?<\/pre>|';
 378
 379                 $reg = '/' . $codefix . $scriptfix . $prefix .
 380                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 381                 $startPos = 0;
 382                 $sourceBlob = '';
 383                 $literalBlob = '';
 384
 385                 // Guard against delimiter nulls in the input
 386                 $text = str_replace( "\000", '', $text );
 387
 388                 $markupMatches = null;
 389                 $elementMatches = null;
 390                 while ( $startPos < strlen( $text ) ) {
 391                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 392                                 $elementPos = $markupMatches[0][1];
 393                                 $element = $markupMatches[0][0];
 394                         } else {
 395                                 $elementPos = strlen( $text );
 396                                 $element = '';
 397                         }
 398
 399                         // Queue the part before the markup for translation in a batch
 400                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 401
 402                         // Advance to the next position
 403                         $startPos = $elementPos + strlen( $element );
 404
 405                         // Translate any alt or title attributes inside the matched element
 406                         if ( $element !== ''
 407                                 && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element, $elementMatches )
 408                         ) {
 409                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 410                                 $changed = false;
 411                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 412                                         if ( !isset( $attrs[$attrName] ) ) {
 413                                                 continue;
 414                                         }
 415                                         $attr = $attrs[$attrName];
 416                                         // Don't convert URLs
 417                                         if ( !strpos( $attr, '://' ) ) {
 418                                                 $attr = $this->recursiveConvertTopLevel( $attr, $toVariant );
 419                                         }
 420
 421                                         // Remove HTML tags to avoid disrupting the layout
 422                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 423                                         if ( $attr !== $attrs[$attrName] ) {
 424                                                 $attrs[$attrName] = $attr;
 425                                                 $changed = true;
 426                                         }
 427                                 }
 428                                 if ( $changed ) {
 429                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 430                                                 $elementMatches[3];
 431                                 }
 432                         }
 433                         $literalBlob .= $element . "\000";
 434                 }
 435
 436                 // Do the main translation batch
 437                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 438
 439                 // Put the output back together
 440                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 441                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 442                 $output = '';
 443                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 444                         $output .= $translatedIter->current();
 445                         $output .= $literalIter->current();
 446                         $translatedIter->next();
 447                         $literalIter->next();
 448                 }
 449
 450                 wfProfileOut( __METHOD__ );
 451                 return $output;
 452         }
 453
 454         /**
 455          * Translate a string to a variant.
 456          * Doesn't parse rules or do any of that other stuff, for that use
 457          * convert() or convertTo().
 458          *
 459          * @param $text String: text to convert
 460          * @param $variant String: variant language code
 461          * @return String: translated text
 462          */
 463         public function translate( $text, $variant ) {
 464                 wfProfileIn( __METHOD__ );
 465                 // If $text is empty or only includes spaces, do nothing
 466                 // Otherwise translate it
 467                 if ( trim( $text ) ) {
 468                         $this->loadTables();
 469                         $text = $this->mTables[$variant]->replace( $text );
 470                 }
 471                 wfProfileOut( __METHOD__ );
 472                 return $text;
 473         }
 474
 475         /**
 476          * Call translate() to convert text to all valid variants.
 477          *
 478          * @param $text String: the text to be converted
 479          * @return Array: variant => converted text
 480          */
 481         public function autoConvertToAllVariants( $text ) {
 482                 wfProfileIn( __METHOD__ );
 483                 $this->loadTables();
 484
 485                 $ret = array();
 486                 foreach ( $this->mVariants as $variant ) {
 487                         $ret[$variant] = $this->translate( $text, $variant );
 488                 }
 489
 490                 wfProfileOut( __METHOD__ );
 491                 return $ret;
 492         }
 493
 494         /**
 495          * Apply manual conversion rules.
 496          *
 497          * @param $convRule ConverterRule Object of ConverterRule
 498          */
 499         protected function applyManualConv( $convRule ) {
 500                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 501                 // title conversion.
 502                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 503                 // rule(s) not for title, this breaks the title conversion.
 504                 $newConvRuleTitle = $convRule->getTitle();
 505                 if ( $newConvRuleTitle ) {
 506                         // So I add an empty check for getTitle()
 507                         $this->mConvRuleTitle = $newConvRuleTitle;
 508                 }
 509
 510                 // merge/remove manual conversion rules to/from global table
 511                 $convTable = $convRule->getConvTable();
 512                 $action = $convRule->getRulesAction();
 513                 foreach ( $convTable as $variant => $pair ) {
 514                         if ( !$this->validateVariant( $variant ) ) {
 515                                 continue;
 516                         }
 517
 518                         if ( $action == 'add' ) {
 519                                 foreach ( $pair as $from => $to ) {
 520                                         // to ensure that $from and $to not be left blank
 521                                         // so $this->translate() could always return a string
 522                                         if ( $from || $to ) {
 523                                                 // more efficient than array_merge(), about 2.5 times.
 524                                                 $this->mTables[$variant]->setPair( $from, $to );
 525                                         }
 526                                 }
 527                         } elseif ( $action == 'remove' ) {
 528                                 $this->mTables[$variant]->removeArray( $pair );
 529                         }
 530                 }
 531         }
 532
 533         /**
 534          * Auto convert a Title object to a readable string in the
 535          * preferred variant.
 536          *
 537          * @param $title Title a object of Title
 538          * @return String: converted title text
 539          */
 540         public function convertTitle( $title ) {
 541                 $variant = $this->getPreferredVariant();
 542                 $index = $title->getNamespace();
 543                 if ( $index !== NS_MAIN ) {
 544                         $text = $this->convertNamespace( $index, $variant ) . ':';
 545                 } else {
 546                         $text = '';
 547                 }
 548                 $text .= $this->translate( $title->getText(), $variant );
 549                 return $text;
 550         }
 551
 552         /**
 553          * Get the namespace display name in the preferred variant.
 554          *
 555          * @param $index int namespace id
 556          * @param $variant string|null variant code or null for preferred variant
 557          * @return String: namespace name for display
 558          */
 559         public function convertNamespace( $index, $variant = null ) {
 560                 if ( $variant === null ) {
 561                         $variant = $this->getPreferredVariant();
 562                 }
 563                 if ( $index === NS_MAIN ) {
 564                         return '';
 565                 } else {
 566                         // First check if a message gives a converted name in the target variant.
 567                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inLanguage( $variant );
 568                         if ( $nsConvMsg->exists() ) {
 569                                 return $nsConvMsg->plain();
 570                         }
 571                         // Then check if a message gives a converted name in content language
 572                         // which needs extra translation to the target variant.
 573                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 574                         if ( $nsConvMsg->exists() ) {
 575                                 return $this->translate( $nsConvMsg->plain(), $variant );
 576                         }
 577                         // No message exists, retrieve it from the target variant's namespace names.
 578                         $langObj = $this->mLangObj->factory( $variant );
 579                         return $langObj->getFormattedNsText( $index );
 580                 }
 581         }
 582
 583         /**
 584          * Convert text to different variants of a language. The automatic
 585          * conversion is done in autoConvert(). Here we parse the text
 586          * marked with -{}-, which specifies special conversions of the
 587          * text that can not be accomplished in autoConvert().
 588          *
 589          * Syntax of the markup:
 590          * -{code1:text1;code2:text2;...}-  or
 591          * -{flags|code1:text1;code2:text2;...}-  or
 592          * -{text}- in which case no conversion should take place for text
 593          *
 594          * @param $text String: text to be converted
 595          * @return String: converted text
 596          */
 597         public function convert( $text ) {
 598                 $variant = $this->getPreferredVariant();
 599                 return $this->convertTo( $text, $variant );
 600         }
 601
 602         /**
 603          * Same as convert() except a extra parameter to custom variant.
 604          *
 605          * @param $text String: text to be converted
 606          * @param $variant String: the target variant code
 607          * @return String: converted text
 608          */
 609         public function convertTo( $text, $variant ) {
 610                 global $wgDisableLangConversion;
 611                 if ( $wgDisableLangConversion ) {
 612                         return $text;
 613                 }
 614                 // Reset converter state for a new converter run.
 615                 $this->mConvRuleTitle = false;
 616                 return $this->recursiveConvertTopLevel( $text, $variant );
 617         }
 618
 619         /**
 620          * Recursively convert text on the outside. Allow to use nested
 621          * markups to custom rules.
 622          *
 623          * @param $text String: text to be converted
 624          * @param $variant String: the target variant code
 625          * @param $depth Integer: depth of recursion
 626          * @return String: converted text
 627          */
 628         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 629                 $startPos = 0;
 630                 $out = '';
 631                 $length = strlen( $text );
 632                 $shouldConvert = !$this->guessVariant( $text, $variant );
 633
 634                 while ( $startPos < $length ) {
 635                         $pos = strpos( $text, '-{', $startPos );
 636
 637                         if ( $pos === false ) {
 638                                 // No more markup, append final segment
 639                                 $fragment = substr( $text, $startPos );
 640                                 $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 641                                 return $out;
 642                         }
 643
 644                         // Markup found
 645                         // Append initial segment
 646                         $fragment = substr( $text, $startPos, $pos - $startPos );
 647                         $out .= $shouldConvert ? $this->autoConvert( $fragment, $variant ) : $fragment;
 648
 649                         // Advance position
 650                         $startPos = $pos;
 651
 652                         // Do recursive conversion
 653                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 654                 }
 655
 656                 return $out;
 657         }
 658
 659         /**
 660          * Recursively convert text on the inside.
 661          *
 662          * @param $text String: text to be converted
 663          * @param $variant String: the target variant code
 664          * @param $startPos int
 665          * @param $depth Integer: depth of recursion
 666          *
 667          * @throws MWException
 668          * @return String: converted text
 669          */
 670         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 671                 // Quick sanity check (no function calls)
 672                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 673                         throw new MWException( __METHOD__ . ': invalid input string' );
 674                 }
 675
 676                 $startPos += 2;
 677                 $inner = '';
 678                 $warningDone = false;
 679                 $length = strlen( $text );
 680
 681                 while ( $startPos < $length ) {
 682                         $m = false;
 683                         preg_match( '/-\{|\}-/', $text, $m, PREG_OFFSET_CAPTURE, $startPos );
 684                         if ( !$m ) {
 685                                 // Unclosed rule
 686                                 break;
 687                         }
 688
 689                         $token = $m[0][0];
 690                         $pos = $m[0][1];
 691
 692                         // Markup found
 693                         // Append initial segment
 694                         $inner .= substr( $text, $startPos, $pos - $startPos );
 695
 696                         // Advance position
 697                         $startPos = $pos;
 698
 699                         switch ( $token ) {
 700                                 case '-{':
 701                                         // Check max depth
 702                                         if ( $depth >= $this->mMaxDepth ) {
 703                                                 $inner .= '-{';
 704                                                 if ( !$warningDone ) {
 705                                                         $inner .= '<span class="error">' .
 706                                                                 wfMessage( 'language-converter-depth-warning' )
 707                                                                         ->numParams( $this->mMaxDepth )->inContentLanguage()->text() .
 708                                                                 '</span>';
 709                                                         $warningDone = true;
 710                                                 }
 711                                                 $startPos += 2;
 712                                                 continue;
 713                                         }
 714                                         // Recursively parse another rule
 715                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 716                                         break;
 717                                 case '}-':
 718                                         // Apply the rule
 719                                         $startPos += 2;
 720                                         $rule = new ConverterRule( $inner, $this );
 721                                         $rule->parse( $variant );
 722                                         $this->applyManualConv( $rule );
 723                                         return $rule->getDisplay();
 724                                 default:
 725                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 726                         }
 727                 }
 728
 729                 // Unclosed rule
 730                 if ( $startPos < $length ) {
 731                         $inner .= substr( $text, $startPos );
 732                 }
 733                 $startPos = $length;
 734                 return '-{' . $this->autoConvert( $inner, $variant );
 735         }
 736
 737         /**
 738          * If a language supports multiple variants, it is possible that
 739          * non-existing link in one variant actually exists in another variant.
 740          * This function tries to find it. See e.g. LanguageZh.php
 741          *
 742          * @param $link String: the name of the link
 743          * @param $nt Mixed: the title object of the link
 744          * @param $ignoreOtherCond Boolean: to disable other conditions when
 745          *              we need to transclude a template or update a category's link
 746          * @return Null, the input parameters may be modified upon return
 747          */
 748         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 749                 # If the article has already existed, there is no need to
 750                 # check it again, otherwise it may cause a fault.
 751                 if ( is_object( $nt ) && $nt->exists() ) {
 752                         return;
 753                 }
 754
 755                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 756                         $wgUser;
 757                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 758                 $action = $wgRequest->getText( 'action' );
 759                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 760                 $disableLinkConversion = $wgDisableLangConversion
 761                         || $wgDisableTitleConversion;
 762                 $linkBatch = new LinkBatch();
 763
 764                 $ns = NS_MAIN;
 765
 766                 if ( $disableLinkConversion ||
 767                         ( !$ignoreOtherCond &&
 768                                 ( $isredir == 'no'
 769                                         || $action == 'edit'
 770                                         || $action == 'submit'
 771                                         || $linkconvert == 'no'
 772                                         || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 773                         return;
 774                 }
 775
 776                 if ( is_object( $nt ) ) {
 777                         $ns = $nt->getNamespace();
 778                 }
 779
 780                 $variants = $this->autoConvertToAllVariants( $link );
 781                 if ( !$variants ) { // give up
 782                         return;
 783                 }
 784
 785                 $titles = array();
 786
 787                 foreach ( $variants as $v ) {
 788                         if ( $v != $link ) {
 789                                 $varnt = Title::newFromText( $v, $ns );
 790                                 if ( !is_null( $varnt ) ) {
 791                                         $linkBatch->addObj( $varnt );
 792                                         $titles[] = $varnt;
 793                                 }
 794                         }
 795                 }
 796
 797                 // fetch all variants in single query
 798                 $linkBatch->execute();
 799
 800                 foreach ( $titles as $varnt ) {
 801                         if ( $varnt->getArticleID() > 0 ) {
 802                                 $nt = $varnt;
 803                                 $link = $varnt->getText();
 804                                 break;
 805                         }
 806                 }
 807         }
 808
 809         /**
 810          * Returns language specific hash options.
 811          *
 812          * @return string
 813          */
 814         public function getExtraHashOptions() {
 815                 $variant = $this->getPreferredVariant();
 816                 return '!' . $variant;
 817         }
 818
 819         /**
 820          * Guess if a text is written in a variant. This should be implemented in subclasses.
 821          *
 822          * @param string $text the text to be checked
 823          * @param string $variant language code of the variant to be checked for
 824          * @return bool true if $text appears to be written in $variant, false if not
 825          *
 826          * @author Nikola Smolenski <smolensk@eunet.rs>
 827          * @since 1.19
 828          */
 829         public function guessVariant( $text, $variant ) {
 830                 return false;
 831         }
 832
 833         /**
 834          * Load default conversion tables.
 835          * This method must be implemented in derived class.
 836          *
 837          * @private
 838          * @throws MWException
 839          */
 840         function loadDefaultTables() {
 841                 $name = get_class( $this );
 842                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 843         }
 844
 845         /**
 846          * Load conversion tables either from the cache or the disk.
 847          * @private
 848          * @param $fromCache Boolean: load from memcached? Defaults to true.
 849          */
 850         function loadTables( $fromCache = true ) {
 851                 global $wgLangConvMemc;
 852
 853                 if ( $this->mTablesLoaded ) {
 854                         return;
 855                 }
 856
 857                 wfProfileIn( __METHOD__ );
 858                 $this->mTablesLoaded = true;
 859                 $this->mTables = false;
 860                 if ( $fromCache ) {
 861                         wfProfileIn( __METHOD__ . '-cache' );
 862                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 863                         wfProfileOut( __METHOD__ . '-cache' );
 864                 }
 865                 if ( !$this->mTables || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 866                         wfProfileIn( __METHOD__ . '-recache' );
 867                         // not in cache, or we need a fresh reload.
 868                         // We will first load the default tables
 869                         // then update them using things in MediaWiki:Conversiontable/*
 870                         $this->loadDefaultTables();
 871                         foreach ( $this->mVariants as $var ) {
 872                                 $cached = $this->parseCachedTable( $var );
 873                                 $this->mTables[$var]->mergeArray( $cached );
 874                         }
 875
 876                         $this->postLoadTables();
 877                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 878
 879                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 880                         wfProfileOut( __METHOD__ . '-recache' );
 881                 }
 882                 wfProfileOut( __METHOD__ );
 883         }
 884
 885         /**
 886          * Hook for post processing after conversion tables are loaded.
 887          */
 888         function postLoadTables() { }
 889
 890         /**
 891          * Reload the conversion tables.
 892          *
 893          * @private
 894          */
 895         function reloadTables() {
 896                 if ( $this->mTables ) {
 897                         unset( $this->mTables );
 898                 }
 899                 $this->mTablesLoaded = false;
 900                 $this->loadTables( false );
 901         }
 902
 903         /**
 904          * Parse the conversion table stored in the cache.
 905          *
 906          * The tables should be in blocks of the following form:
 907          *              -{
 908          *                      word => word ;
 909          *                      word => word ;
 910          *                      ...
 911          *              }-
 912          *
 913          * To make the tables more manageable, subpages are allowed
 914          * and will be parsed recursively if $recursive == true.
 915          *
 916          * @param $code String: language code
 917          * @param $subpage String: subpage name
 918          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 919          *
 920          * @return array
 921          */
 922         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 923                 static $parsed = array();
 924
 925                 $key = 'Conversiontable/' . $code;
 926                 if ( $subpage ) {
 927                         $key .= '/' . $subpage;
 928                 }
 929                 if ( array_key_exists( $key, $parsed ) ) {
 930                         return array();
 931                 }
 932
 933                 $parsed[$key] = true;
 934
 935                 if ( $subpage === '' ) {
 936                         $txt = MessageCache::singleton()->getMsgFromNamespace( $key, $code );
 937                 } else {
 938                         $txt = false;
 939                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 940                         if ( $title && $title->exists() ) {
 941                                 $revision = Revision::newFromTitle( $title );
 942                                 if ( $revision ) {
 943                                         if ( $revision->getContentModel() == CONTENT_MODEL_WIKITEXT ) {
 944                                                 $txt = $revision->getContent( Revision::RAW )->getNativeData();
 945                                         }
 946
 947                                         // @todo in the future, use a specialized content model, perhaps based on json!
 948                                 }
 949                         }
 950                 }
 951
 952                 # Nothing to parse if there's no text
 953                 if ( $txt === false || $txt === null || $txt === '' ) {
 954                         return array();
 955                 }
 956
 957                 // get all subpage links of the form
 958                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 959                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 960                         ':Conversiontable';
 961                 $subs = StringUtils::explode( '[[', $txt );
 962                 $sublinks = array();
 963                 foreach ( $subs as $sub ) {
 964                         $link = explode( ']]', $sub, 2 );
 965                         if ( count( $link ) != 2 ) {
 966                                 continue;
 967                         }
 968                         $b = explode( '|', $link[0], 2 );
 969                         $b = explode( '/', trim( $b[0] ), 3 );
 970                         if ( count( $b ) == 3 ) {
 971                                 $sublink = $b[2];
 972                         } else {
 973                                 $sublink = '';
 974                         }
 975
 976                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 977                                 $sublinks[] = $sublink;
 978                         }
 979                 }
 980
 981                 // parse the mappings in this page
 982                 $blocks = StringUtils::explode( '-{', $txt );
 983                 $ret = array();
 984                 $first = true;
 985                 foreach ( $blocks as $block ) {
 986                         if ( $first ) {
 987                                 // Skip the part before the first -{
 988                                 $first = false;
 989                                 continue;
 990                         }
 991                         $mappings = explode( '}-', $block, 2 );
 992                         $stripped = str_replace( array( "'", '"', '*', '#' ), '', $mappings[0] );
 993                         $table = StringUtils::explode( ';', $stripped );
 994                         foreach ( $table as $t ) {
 995                                 $m = explode( '=>', $t, 3 );
 996                                 if ( count( $m ) != 2 ) {
 997                                         continue;
 998                                 }
 999                                 // trim any trailling comments starting with '//'
1000                                 $tt = explode( '//', $m[1], 2 );
1001                                 $ret[trim( $m[0] )] = trim( $tt[0] );
1002                         }
1003                 }
1004
1005                 // recursively parse the subpages
1006                 if ( $recursive ) {
1007                         foreach ( $sublinks as $link ) {
1008                                 $s = $this->parseCachedTable( $code, $link, $recursive );
1009                                 $ret = array_merge( $ret, $s );
1010                         }
1011                 }
1012
1013                 if ( $this->mUcfirst ) {
1014                         foreach ( $ret as $k => $v ) {
1015                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
1016                         }
1017                 }
1018                 return $ret;
1019         }
1020
1021         /**
1022          * Enclose a string with the "no conversion" tag. This is used by
1023          * various functions in the Parser.
1024          *
1025          * @param $text String: text to be tagged for no conversion
1026          * @param $noParse Boolean: unused
1027          * @return String: the tagged text
1028          */
1029         public function markNoConversion( $text, $noParse = false ) {
1030                 # don't mark if already marked
1031                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1032                         return $text;
1033                 }
1034
1035                 $ret = "-{R|$text}-";
1036                 return $ret;
1037         }
1038
1039         /**
1040          * Convert the sorting key for category links. This should make different
1041          * keys that are variants of each other map to the same key.
1042          *
1043          * @param $key string
1044          *
1045          * @return string
1046          */
1047         function convertCategoryKey( $key ) {
1048                 return $key;
1049         }
1050
1051         /**
1052          * Hook to refresh the cache of conversion tables when
1053          * MediaWiki:Conversiontable* is updated.
1054          * @private
1055          *
1056          * @param $page WikiPage object
1057          * @param $user Object: User object for the current user
1058          * @param $content Content: new page content
1059          * @param $summary String: edit summary of the edit
1060          * @param $isMinor Boolean: was the edit marked as minor?
1061          * @param $isWatch Boolean: did the user watch this page or not?
1062          * @param $section
1063          * @param $flags int Bitfield
1064          * @param $revision Object: new Revision object or null
1065          * @return Boolean: true
1066          */
1067         function OnPageContentSaveComplete( $page, $user, $content, $summary, $isMinor,
1068                         $isWatch, $section, $flags, $revision ) {
1069                 $titleobj = $page->getTitle();
1070                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1071                         $title = $titleobj->getDBkey();
1072                         $t = explode( '/', $title, 3 );
1073                         $c = count( $t );
1074                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1075                                 if ( $this->validateVariant( $t[1] ) ) {
1076                                         $this->reloadTables();
1077                                 }
1078                         }
1079                 }
1080                 return true;
1081         }
1082
1083         /**
1084          * Armour rendered math against conversion.
1085          * Escape special chars in parsed math text. (in most cases are img elements)
1086          *
1087          * @param $text String: text to armour against conversion
1088          * @return String: armoured text where { and } have been converted to
1089          *                 &#123; and &#125;
1090          * @deprecated since 1.22 is no longer used
1091          */
1092         public function armourMath( $text ) {
1093                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1094                 // any unwanted markup appearing in the math image tag.
1095                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1096                 return $text;
1097         }
1098
1099         /**
1100          * Get the cached separator pattern for ConverterRule::parseRules()
1101          */
1102         function getVarSeparatorPattern() {
1103                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1104                         // varsep_pattern for preg_split:
1105                         // text should be splited by ";" only if a valid variant
1106                         // name exist after the markup, for example:
1107                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1108                         //      <span style="font-size:120%;">yyy</span>;}-
1109                         // we should split it as:
1110                         //  array(
1111                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1112                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1113                         //        [2] => ''
1114                         //       )
1115                         $pat = '/;\s*(?=';
1116                         foreach ( $this->mVariants as $variant ) {
1117                                 // zh-hans:xxx;zh-hant:yyy
1118                                 $pat .= $variant . '\s*:|';
1119                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1120                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1121                         }
1122                         $pat .= '\s*$)/';
1123                         $this->mVarSeparatorPattern = $pat;
1124                 }
1125                 return $this->mVarSeparatorPattern;
1126         }
1127 }
1128
1129 /**
1130  * Parser for rules of language conversion , parse rules in -{ }- tag.
1131  * @ingroup Language
1132  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1133  */
1134 class ConverterRule {
1135         public $mText; // original text in -{text}-
1136         public $mConverter; // LanguageConverter object
1137         public $mRuleDisplay = '';
1138         public $mRuleTitle = false;
1139         public $mRules = '';// string : the text of the rules
1140         public $mRulesAction = 'none';
1141         public $mFlags = array();
1142         public $mVariantFlags = array();
1143         public $mConvTable = array();
1144         public $mBidtable = array();// array of the translation in each variant
1145         public $mUnidtable = array();// array of the translation in each variant
1146
1147         /**
1148          * Constructor
1149          *
1150          * @param $text String: the text between -{ and }-
1151          * @param $converter LanguageConverter object
1152          */
1153         public function __construct( $text, $converter ) {
1154                 $this->mText = $text;
1155                 $this->mConverter = $converter;
1156         }
1157
1158         /**
1159          * Check if variants array in convert array.
1160          *
1161          * @param $variants Array or string: variant language code
1162          * @return String: translated text
1163          */
1164         public function getTextInBidtable( $variants ) {
1165                 $variants = (array)$variants;
1166                 if ( !$variants ) {
1167                         return false;
1168                 }
1169                 foreach ( $variants as $variant ) {
1170                         if ( isset( $this->mBidtable[$variant] ) ) {
1171                                 return $this->mBidtable[$variant];
1172                         }
1173                 }
1174                 return false;
1175         }
1176
1177         /**
1178          * Parse flags with syntax -{FLAG| ... }-
1179          * @private
1180          */
1181         function parseFlags() {
1182                 $text = $this->mText;
1183                 $flags = array();
1184                 $variantFlags = array();
1185
1186                 $sepPos = strpos( $text, '|' );
1187                 if ( $sepPos !== false ) {
1188                         $validFlags = $this->mConverter->mFlags;
1189                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1190                         foreach ( $f as $ff ) {
1191                                 $ff = trim( $ff );
1192                                 if ( isset( $validFlags[$ff] ) ) {
1193                                         $flags[$validFlags[$ff]] = true;
1194                                 }
1195                         }
1196                         $text = strval( substr( $text, $sepPos + 1 ) );
1197                 }
1198
1199                 if ( !$flags ) {
1200                         $flags['S'] = true;
1201                 } elseif ( isset( $flags['R'] ) ) {
1202                         $flags = array( 'R' => true );// remove other flags
1203                 } elseif ( isset( $flags['N'] ) ) {
1204                         $flags = array( 'N' => true );// remove other flags
1205                 } elseif ( isset( $flags['-'] ) ) {
1206                         $flags = array( '-' => true );// remove other flags
1207                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1208                         $flags['H'] = true;
1209                 } elseif ( isset( $flags['H'] ) ) {
1210                         // replace A flag, and remove other flags except T
1211                         $temp = array( '+' => true, 'H' => true );
1212                         if ( isset( $flags['T'] ) ) {
1213                                 $temp['T'] = true;
1214                         }
1215                         if ( isset( $flags['D'] ) ) {
1216                                 $temp['D'] = true;
1217                         }
1218                         $flags = $temp;
1219                 } else {
1220                         if ( isset( $flags['A'] ) ) {
1221                                 $flags['+'] = true;
1222                                 $flags['S'] = true;
1223                         }
1224                         if ( isset( $flags['D'] ) ) {
1225                                 unset( $flags['S'] );
1226                         }
1227                         // try to find flags like "zh-hans", "zh-hant"
1228                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1229                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1230                         if ( $variantFlags ) {
1231                                 $variantFlags = array_flip( $variantFlags );
1232                                 $flags = array();
1233                         }
1234                 }
1235                 $this->mVariantFlags = $variantFlags;
1236                 $this->mRules = $text;
1237                 $this->mFlags = $flags;
1238         }
1239
1240         /**
1241          * Generate conversion table.
1242          * @private
1243          */
1244         function parseRules() {
1245                 $rules = $this->mRules;
1246                 $bidtable = array();
1247                 $unidtable = array();
1248                 $variants = $this->mConverter->mVariants;
1249                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1250
1251                 // Split according to $varsep_pattern, but ignore semicolons from HTML entities
1252                 $rules = preg_replace( '/(&[#a-zA-Z0-9]+);/', "$1\x01", $rules );
1253                 $choice = preg_split( $varsep_pattern, $rules );
1254                 $choice = str_replace( "\x01", ';', $choice );
1255
1256                 foreach ( $choice as $c ) {
1257                         $v = explode( ':', $c, 2 );
1258                         if ( count( $v ) != 2 ) {
1259                                 // syntax error, skip
1260                                 continue;
1261                         }
1262                         $to = trim( $v[1] );
1263                         $v = trim( $v[0] );
1264                         $u = explode( '=>', $v, 2 );
1265                         // if $to is empty, strtr() could return a wrong result
1266                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1267                                 $bidtable[$v] = $to;
1268                         } elseif ( count( $u ) == 2 ) {
1269                                 $from = trim( $u[0] );
1270                                 $v = trim( $u[1] );
1271                                 if ( array_key_exists( $v, $unidtable )
1272                                         && !is_array( $unidtable[$v] )
1273                                         && $to
1274                                         && in_array( $v, $variants ) ) {
1275                                         $unidtable[$v] = array( $from => $to );
1276                                 } elseif ( $to && in_array( $v, $variants ) ) {
1277                                         $unidtable[$v][$from] = $to;
1278                                 }
1279                         }
1280                         // syntax error, pass
1281                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1282                                 $bidtable = array();
1283                                 $unidtable = array();
1284                                 break;
1285                         }
1286                 }
1287                 $this->mBidtable = $bidtable;
1288                 $this->mUnidtable = $unidtable;
1289         }
1290
1291         /**
1292          * @private
1293          *
1294          * @return string
1295          */
1296         function getRulesDesc() {
1297                 $codesep = $this->mConverter->mDescCodeSep;
1298                 $varsep = $this->mConverter->mDescVarSep;
1299                 $text = '';
1300                 foreach ( $this->mBidtable as $k => $v ) {
1301                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1302                 }
1303                 foreach ( $this->mUnidtable as $k => $a ) {
1304                         foreach ( $a as $from => $to ) {
1305                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1306                                         "$codesep$to$varsep";
1307                         }
1308                 }
1309                 return $text;
1310         }
1311
1312         /**
1313          * Parse rules conversion.
1314          * @private
1315          *
1316          * @param $variant
1317          *
1318          * @return string
1319          */
1320         function getRuleConvertedStr( $variant ) {
1321                 $bidtable = $this->mBidtable;
1322                 $unidtable = $this->mUnidtable;
1323
1324                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1325                         return $this->mRules;
1326                 } else {
1327                         // display current variant in bidirectional array
1328                         $disp = $this->getTextInBidtable( $variant );
1329                         // or display current variant in fallbacks
1330                         if ( !$disp ) {
1331                                 $disp = $this->getTextInBidtable(
1332                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1333                         }
1334                         // or display current variant in unidirectional array
1335                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1336                                 $disp = array_values( $unidtable[$variant] );
1337                                 $disp = $disp[0];
1338                         }
1339                         // or display frist text under disable manual convert
1340                         if ( !$disp && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1341                                 if ( count( $bidtable ) > 0 ) {
1342                                         $disp = array_values( $bidtable );
1343                                         $disp = $disp[0];
1344                                 } else {
1345                                         $disp = array_values( $unidtable );
1346                                         $disp = array_values( $disp[0] );
1347                                         $disp = $disp[0];
1348                                 }
1349                         }
1350                         return $disp;
1351                 }
1352         }
1353
1354         /**
1355          * Similar to getRuleConvertedStr(), but this prefers to use original
1356          * page title if $variant === $this->mConverter->mMainLanguageCode
1357          * and may return false in this case (so this title conversion rule
1358          * will be ignored and the original title is shown).
1359          *
1360          * @since 1.22
1361          * @param $variant The variant code to display page title in
1362          * @return String|false The converted title or false if just page name
1363          */
1364         function getRuleConvertedTitle( $variant ) {
1365                 if ( $variant === $this->mConverter->mMainLanguageCode ) {
1366                         // If a string targeting exactly this variant is set,
1367                         // use it. Otherwise, just return false, so the real
1368                         // page name can be shown (and because variant === main,
1369                         // there'll be no further automatic conversion).
1370                         $disp = $this->getTextInBidtable( $variant );
1371                         if ( $disp ) {
1372                                 return $disp;
1373                         }
1374                         if ( array_key_exists( $variant, $this->mUnidtable ) ) {
1375                                 $disp = array_values( $this->mUnidtable[$variant] );
1376                                 $disp = $disp[0];
1377                         }
1378                         // Assigned above or still false.
1379                         return $disp;
1380                 } else {
1381                         return $this->getRuleConvertedStr( $variant );
1382                 }
1383         }
1384
1385         /**
1386          * Generate conversion table for all text.
1387          * @private
1388          */
1389         function generateConvTable() {
1390                 // Special case optimisation
1391                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1392                         $this->mConvTable = array();
1393                         return;
1394                 }
1395
1396                 $bidtable = $this->mBidtable;
1397                 $unidtable = $this->mUnidtable;
1398                 $manLevel = $this->mConverter->mManualLevel;
1399
1400                 $vmarked = array();
1401                 foreach ( $this->mConverter->mVariants as $v ) {
1402                         /* for bidirectional array
1403                                 fill in the missing variants, if any,
1404                                 with fallbacks */
1405                         if ( !isset( $bidtable[$v] ) ) {
1406                                 $variantFallbacks =
1407                                         $this->mConverter->getVariantFallbacks( $v );
1408                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1409                                 if ( $vf ) {
1410                                         $bidtable[$v] = $vf;
1411                                 }
1412                         }
1413
1414                         if ( isset( $bidtable[$v] ) ) {
1415                                 foreach ( $vmarked as $vo ) {
1416                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1417                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1418                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1419                                         // to introduce a custom mapping between
1420                                         // words WordZh and WordTw in the whole text
1421                                         if ( $manLevel[$v] == 'bidirectional' ) {
1422                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1423                                         }
1424                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1425                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1426                                         }
1427                                 }
1428                                 $vmarked[] = $v;
1429                         }
1430                         /* for unidirectional array fill to convert tables */
1431                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1432                                 && isset( $unidtable[$v] )
1433                         ) {
1434                                 if ( isset( $this->mConvTable[$v] ) ) {
1435                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1436                                 } else {
1437                                         $this->mConvTable[$v] = $unidtable[$v];
1438                                 }
1439                         }
1440                 }
1441         }
1442
1443         /**
1444          * Parse rules and flags.
1445          * @param $variant String: variant language code
1446          */
1447         public function parse( $variant = null ) {
1448                 if ( !$variant ) {
1449                         $variant = $this->mConverter->getPreferredVariant();
1450                 }
1451
1452                 $this->parseFlags();
1453                 $flags = $this->mFlags;
1454
1455                 // convert to specified variant
1456                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1457                 if ( $this->mVariantFlags ) {
1458                         // check if current variant in flags
1459                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1460                                 // then convert <text to convert> to current language
1461                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1462                                         $variant );
1463                         } else {
1464                                 // if current variant no in flags,
1465                                 // then we check its fallback variants.
1466                                 $variantFallbacks =
1467                                         $this->mConverter->getVariantFallbacks( $variant );
1468                                 if ( is_array( $variantFallbacks ) ) {
1469                                         foreach ( $variantFallbacks as $variantFallback ) {
1470                                                 // if current variant's fallback exist in flags
1471                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1472                                                         // then convert <text to convert> to fallback language
1473                                                         $this->mRules =
1474                                                                 $this->mConverter->autoConvert( $this->mRules,
1475                                                                         $variantFallback );
1476                                                         break;
1477                                                 }
1478                                         }
1479                                 }
1480                         }
1481                         $this->mFlags = $flags = array( 'R' => true );
1482                 }
1483
1484                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1485                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1486                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1487                         $this->parseRules();
1488                 }
1489                 $rules = $this->mRules;
1490
1491                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1492                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1493                                 // fill all variants if text in -{A/H/-|text} without rules
1494                                 foreach ( $this->mConverter->mVariants as $v ) {
1495                                         $this->mBidtable[$v] = $rules;
1496                                 }
1497                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1498                                 $this->mFlags = $flags = array( 'R' => true );
1499                         }
1500                 }
1501
1502                 $this->mRuleDisplay = false;
1503                 foreach ( $flags as $flag => $unused ) {
1504                         switch ( $flag ) {
1505                                 case 'R':
1506                                         // if we don't do content convert, still strip the -{}- tags
1507                                         $this->mRuleDisplay = $rules;
1508                                         break;
1509                                 case 'N':
1510                                         // process N flag: output current variant name
1511                                         $ruleVar = trim( $rules );
1512                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1513                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1514                                         } else {
1515                                                 $this->mRuleDisplay = '';
1516                                         }
1517                                         break;
1518                                 case 'D':
1519                                         // process D flag: output rules description
1520                                         $this->mRuleDisplay = $this->getRulesDesc();
1521                                         break;
1522                                 case 'H':
1523                                         // process H,- flag or T only: output nothing
1524                                         $this->mRuleDisplay = '';
1525                                         break;
1526                                 case '-':
1527                                         $this->mRulesAction = 'remove';
1528                                         $this->mRuleDisplay = '';
1529                                         break;
1530                                 case '+':
1531                                         $this->mRulesAction = 'add';
1532                                         $this->mRuleDisplay = '';
1533                                         break;
1534                                 case 'S':
1535                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1536                                         break;
1537                                 case 'T':
1538                                         $this->mRuleTitle = $this->getRuleConvertedTitle( $variant );
1539                                         $this->mRuleDisplay = '';
1540                                         break;
1541                                 default:
1542                                         // ignore unknown flags (but see error case below)
1543                         }
1544                 }
1545                 if ( $this->mRuleDisplay === false ) {
1546                         $this->mRuleDisplay = '<span class="error">'
1547                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1548                                 . '</span>';
1549                 }
1550
1551                 $this->generateConvTable();
1552         }
1553
1554         /**
1555          * @todo FIXME: code this function :)
1556          */
1557         public function hasRules() {
1558                 // TODO:
1559         }
1560
1561         /**
1562          * Get display text on markup -{...}-
1563          * @return string
1564          */
1565         public function getDisplay() {
1566                 return $this->mRuleDisplay;
1567         }
1568
1569         /**
1570          * Get converted title.
1571          * @return string
1572          */
1573         public function getTitle() {
1574                 return $this->mRuleTitle;
1575         }
1576
1577         /**
1578          * Return how deal with conversion rules.
1579          * @return string
1580          */
1581         public function getRulesAction() {
1582                 return $this->mRulesAction;
1583         }
1584
1585         /**
1586          * Get conversion table. (bidirectional and unidirectional
1587          * conversion table)
1588          * @return array
1589          */
1590         public function getConvTable() {
1591                 return $this->mConvTable;
1592         }
1593
1594         /**
1595          * Get conversion rules string.
1596          * @return string
1597          */
1598         public function getRules() {
1599                 return $this->mRules;
1600         }
1601
1602         /**
1603          * Get conversion flags.
1604          * @return array
1605          */
1606         public function getFlags() {
1607                 return $this->mFlags;
1608         }
1609 }