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                 'zh',
  47         );
  48
  49         var $mMainLanguageCode;
  50         var $mVariants, $mVariantFallbacks, $mVariantNames;
  51         var $mTablesLoaded = false;
  52         var $mTables;
  53         // 'bidirectional' 'unidirectional' 'disable' for each variant
  54         var $mManualLevel;
  55
  56         /**
  57          * @var String: memcached key name
  58          */
  59         var $mCacheKey;
  60
  61         var $mLangObj;
  62         var $mFlags;
  63         var $mDescCodeSep = ':', $mDescVarSep = ';';
  64         var $mUcfirst = false;
  65         var $mConvRuleTitle = false;
  66         var $mURLVariant;
  67         var $mUserVariant;
  68         var $mHeaderVariant;
  69         var $mMaxDepth = 10;
  70         var $mVarSeparatorPattern;
  71
  72         const CACHE_VERSION_KEY = 'VERSION 6';
  73
  74         /**
  75          * Constructor
  76          *
  77          * @param $langobj Language: the Language Object
  78          * @param $maincode String: the main language code of this language
  79          * @param $variants Array: the supported variants of this language
  80          * @param $variantfallbacks Array: the fallback language of each variant
  81          * @param $flags Array: defining the custom strings that maps to the flags
  82          * @param $manualLevel Array: limit for supported variants
  83          */
  84         public function __construct( $langobj, $maincode, $variants = array(),
  85                                                                 $variantfallbacks = array(), $flags = array(),
  86                                                                 $manualLevel = array() ) {
  87                 global $wgDisabledVariants;
  88                 $this->mLangObj = $langobj;
  89                 $this->mMainLanguageCode = $maincode;
  90                 $this->mVariants = array_diff( $variants, $wgDisabledVariants );
  91                 $this->mVariantFallbacks = $variantfallbacks;
  92                 $this->mVariantNames = Language::fetchLanguageNames();
  93                 $this->mCacheKey = wfMemcKey( 'conversiontables', $maincode );
  94                 $defaultflags = array(
  95                         // 'S' show converted text
  96                         // '+' add rules for alltext
  97                         // 'E' the gave flags is error
  98                         // these flags above are reserved for program
  99                         'A' => 'A',       // add rule for convert code (all text convert)
 100                         'T' => 'T',       // title convert
 101                         'R' => 'R',       // raw content
 102                         'D' => 'D',       // convert description (subclass implement)
 103                         '-' => '-',       // remove convert (not implement)
 104                         'H' => 'H',       // add rule for convert code
 105                                                   // (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 or headers
 187          * @return String: the default variant code
 188          */
 189         public function getDefaultVariant() {
 190                 global $wgDefaultLanguageVariant;
 191
 192                 $req = $this->getURLVariant();
 193
 194                 if ( $wgDefaultLanguageVariant && !$req ) {
 195                         $req = $this->validateVariant( $wgDefaultLanguageVariant );
 196                 }
 197
 198                 if ( $req ) {
 199                         return $req;
 200                 }
 201                 return $this->mMainLanguageCode;
 202         }
 203
 204         /**
 205          * Validate the variant
 206          * @param $variant String: the variant to validate
 207          * @return Mixed: returns the variant if it is valid, null otherwise
 208          */
 209         public function validateVariant( $variant = null ) {
 210                 if ( $variant !== null && in_array( $variant, $this->mVariants ) ) {
 211                         return $variant;
 212                 }
 213                 return null;
 214         }
 215
 216         /**
 217          * Get the variant specified in the URL
 218          *
 219          * @return Mixed: variant if one found, false otherwise.
 220          */
 221         public function getURLVariant() {
 222                 global $wgRequest;
 223
 224                 if ( $this->mURLVariant ) {
 225                         return $this->mURLVariant;
 226                 }
 227
 228                 // see if the preference is set in the request
 229                 $ret = $wgRequest->getText( 'variant' );
 230
 231                 if ( !$ret ) {
 232                         $ret = $wgRequest->getVal( 'uselang' );
 233                 }
 234
 235                 return $this->mURLVariant = $this->validateVariant( $ret );
 236         }
 237
 238         /**
 239          * Determine if the user has a variant set.
 240          *
 241          * @return Mixed: variant if one found, false otherwise.
 242          */
 243         protected function getUserVariant() {
 244                 global $wgUser;
 245
 246                 // memoizing this function wreaks havoc on parserTest.php
 247                 /*
 248                 if ( $this->mUserVariant ) {
 249                         return $this->mUserVariant;
 250                 }
 251                 */
 252
 253                 // Get language variant preference from logged in users
 254                 // Don't call this on stub objects because that causes infinite
 255                 // recursion during initialisation
 256                 if ( $wgUser->isLoggedIn() )  {
 257                         $ret = $wgUser->getOption( 'variant' );
 258                 } else {
 259                         // figure out user lang without constructing wgLang to avoid
 260                         // infinite recursion
 261                         $ret = $wgUser->getOption( 'language' );
 262                 }
 263
 264                 return $this->mUserVariant = $this->validateVariant( $ret );
 265         }
 266
 267         /**
 268          * Determine the language variant from the Accept-Language header.
 269          *
 270          * @return Mixed: variant if one found, false otherwise.
 271          */
 272         protected function getHeaderVariant() {
 273                 global $wgRequest;
 274
 275                 if ( $this->mHeaderVariant ) {
 276                         return $this->mHeaderVariant;
 277                 }
 278
 279                 // see if some supported language variant is set in the
 280                 // HTTP header.
 281                 $languages = array_keys( $wgRequest->getAcceptLang() );
 282                 if ( empty( $languages ) ) {
 283                         return null;
 284                 }
 285
 286                 $fallbackLanguages = array();
 287                 foreach ( $languages as $language ) {
 288                         $this->mHeaderVariant = $this->validateVariant( $language );
 289                         if ( $this->mHeaderVariant ) {
 290                                 break;
 291                         }
 292
 293                         // To see if there are fallbacks of current language.
 294                         // We record these fallback variants, and process
 295                         // them later.
 296                         $fallbacks = $this->getVariantFallbacks( $language );
 297                         if ( is_string( $fallbacks ) ) {
 298                                 $fallbackLanguages[] = $fallbacks;
 299                         } elseif ( is_array( $fallbacks ) ) {
 300                                 $fallbackLanguages =
 301                                         array_merge( $fallbackLanguages, $fallbacks );
 302                         }
 303                 }
 304
 305                 if ( !$this->mHeaderVariant ) {
 306                         // process fallback languages now
 307                         $fallback_languages = array_unique( $fallbackLanguages );
 308                         foreach ( $fallback_languages as $language ) {
 309                                 $this->mHeaderVariant = $this->validateVariant( $language );
 310                                 if ( $this->mHeaderVariant ) {
 311                                         break;
 312                                 }
 313                         }
 314                 }
 315
 316                 return $this->mHeaderVariant;
 317         }
 318
 319         /**
 320          * Dictionary-based conversion.
 321          * This function would not parse the conversion rules.
 322          * If you want to parse rules, try to use convert() or
 323          * convertTo().
 324          *
 325          * @param $text String the text to be converted
 326          * @param $toVariant bool|string the target language code
 327          * @return String the converted text
 328          */
 329         public function autoConvert( $text, $toVariant = false ) {
 330                 wfProfileIn( __METHOD__ );
 331
 332                 $this->loadTables();
 333
 334                 if ( !$toVariant ) {
 335                         $toVariant = $this->getPreferredVariant();
 336                         if ( !$toVariant ) {
 337                                 wfProfileOut( __METHOD__ );
 338                                 return $text;
 339                         }
 340                 }
 341
 342                 if( $this->guessVariant( $text, $toVariant ) ) {
 343                         wfProfileOut( __METHOD__ );
 344                         return $text;
 345                 }
 346
 347                 /* we convert everything except:
 348                    1. HTML markups (anything between < and >)
 349                    2. HTML entities
 350                    3. placeholders created by the parser
 351                 */
 352                 global $wgParser;
 353                 if ( isset( $wgParser ) && $wgParser->UniqPrefix() != '' ) {
 354                         $marker = '|' . $wgParser->UniqPrefix() . '[\-a-zA-Z0-9]+';
 355                 } else {
 356                         $marker = '';
 357                 }
 358
 359                 // this one is needed when the text is inside an HTML markup
 360                 $htmlfix = '|<[^>]+$|^[^<>]*>';
 361
 362                 // disable convert to variants between <code></code> tags
 363                 $codefix = '<code>.+?<\/code>|';
 364                 // disable convertsion of <script type="text/javascript"> ... </script>
 365                 $scriptfix = '<script.*?>.*?<\/script>|';
 366                 // disable conversion of <pre xxxx> ... </pre>
 367                 $prefix = '<pre.*?>.*?<\/pre>|';
 368
 369                 $reg = '/' . $codefix . $scriptfix . $prefix .
 370                         '<[^>]+>|&[a-zA-Z#][a-z0-9]+;' . $marker . $htmlfix . '/s';
 371                 $startPos = 0;
 372                 $sourceBlob = '';
 373                 $literalBlob = '';
 374
 375                 // Guard against delimiter nulls in the input
 376                 $text = str_replace( "\000", '', $text );
 377
 378                 $markupMatches = null;
 379                 $elementMatches = null;
 380                 while ( $startPos < strlen( $text ) ) {
 381                         if ( preg_match( $reg, $text, $markupMatches, PREG_OFFSET_CAPTURE, $startPos ) ) {
 382                                 $elementPos = $markupMatches[0][1];
 383                                 $element = $markupMatches[0][0];
 384                         } else {
 385                                 $elementPos = strlen( $text );
 386                                 $element = '';
 387                         }
 388
 389                         // Queue the part before the markup for translation in a batch
 390                         $sourceBlob .= substr( $text, $startPos, $elementPos - $startPos ) . "\000";
 391
 392                         // Advance to the next position
 393                         $startPos = $elementPos + strlen( $element );
 394
 395                         // Translate any alt or title attributes inside the matched element
 396                         if ( $element !== '' && preg_match( '/^(<[^>\s]*)\s([^>]*)(.*)$/', $element,
 397                                 $elementMatches ) )
 398                         {
 399                                 $attrs = Sanitizer::decodeTagAttributes( $elementMatches[2] );
 400                                 $changed = false;
 401                                 foreach ( array( 'title', 'alt' ) as $attrName ) {
 402                                         if ( !isset( $attrs[$attrName] ) ) {
 403                                                 continue;
 404                                         }
 405                                         $attr = $attrs[$attrName];
 406                                         // Don't convert URLs
 407                                         if ( !strpos( $attr, '://' ) ) {
 408                                                 $attr = $this->translate( $attr, $toVariant );
 409                                         }
 410
 411                                         // Remove HTML tags to avoid disrupting the layout
 412                                         $attr = preg_replace( '/<[^>]+>/', '', $attr );
 413                                         if ( $attr !== $attrs[$attrName] ) {
 414                                                 $attrs[$attrName] = $attr;
 415                                                 $changed = true;
 416                                         }
 417                                 }
 418                                 if ( $changed ) {
 419                                         $element = $elementMatches[1] . Html::expandAttributes( $attrs ) .
 420                                                 $elementMatches[3];
 421                                 }
 422                         }
 423                         $literalBlob .= $element . "\000";
 424                 }
 425
 426                 // Do the main translation batch
 427                 $translatedBlob = $this->translate( $sourceBlob, $toVariant );
 428
 429                 // Put the output back together
 430                 $translatedIter = StringUtils::explode( "\000", $translatedBlob );
 431                 $literalIter = StringUtils::explode( "\000", $literalBlob );
 432                 $output = '';
 433                 while ( $translatedIter->valid() && $literalIter->valid() ) {
 434                         $output .= $translatedIter->current();
 435                         $output .= $literalIter->current();
 436                         $translatedIter->next();
 437                         $literalIter->next();
 438                 }
 439
 440                 wfProfileOut( __METHOD__ );
 441                 return $output;
 442         }
 443
 444         /**
 445          * Translate a string to a variant.
 446          * Doesn't parse rules or do any of that other stuff, for that use
 447          * convert() or convertTo().
 448          *
 449          * @param $text String: text to convert
 450          * @param $variant String: variant language code
 451          * @return String: translated text
 452          */
 453         public function translate( $text, $variant ) {
 454                 wfProfileIn( __METHOD__ );
 455                 // If $text is empty or only includes spaces, do nothing
 456                 // Otherwise translate it
 457                 if ( trim( $text ) ) {
 458                         $this->loadTables();
 459                         $text = $this->mTables[$variant]->replace( $text );
 460                 }
 461                 wfProfileOut( __METHOD__ );
 462                 return $text;
 463         }
 464
 465         /**
 466          * Call translate() to convert text to all valid variants.
 467          *
 468          * @param $text String: the text to be converted
 469          * @return Array: variant => converted text
 470          */
 471         public function autoConvertToAllVariants( $text ) {
 472                 wfProfileIn( __METHOD__ );
 473                 $this->loadTables();
 474
 475                 $ret = array();
 476                 foreach ( $this->mVariants as $variant ) {
 477                         $ret[$variant] = $this->translate( $text, $variant );
 478                 }
 479
 480                 wfProfileOut( __METHOD__ );
 481                 return $ret;
 482         }
 483
 484         /**
 485          * Convert link text to all valid variants.
 486          * In the first, this function only convert text outside the
 487          * "-{" "}-" markups. Since the "{" and "}" are not allowed in
 488          * titles, the text will get all converted always.
 489          * So I removed this feature and deprecated the function.
 490          *
 491          * @param $text String: the text to be converted
 492          * @return Array: variant => converted text
 493          * @deprecated since 1.17 Use autoConvertToAllVariants() instead
 494          */
 495         public function convertLinkToAllVariants( $text ) {
 496                 return $this->autoConvertToAllVariants( $text );
 497         }
 498
 499         /**
 500          * Apply manual conversion rules.
 501          *
 502          * @param $convRule ConverterRule Object of ConverterRule
 503          */
 504         protected function applyManualConv( $convRule ) {
 505                 // Use syntax -{T|zh-cn:TitleCN; zh-tw:TitleTw}- to custom
 506                 // title conversion.
 507                 // Bug 24072: $mConvRuleTitle was overwritten by other manual
 508                 // rule(s) not for title, this breaks the title conversion.
 509                 $newConvRuleTitle = $convRule->getTitle();
 510                 if ( $newConvRuleTitle ) {
 511                         // So I add an empty check for getTitle()
 512                         $this->mConvRuleTitle = $newConvRuleTitle;
 513                 }
 514
 515                 // merge/remove manual conversion rules to/from global table
 516                 $convTable = $convRule->getConvTable();
 517                 $action = $convRule->getRulesAction();
 518                 foreach ( $convTable as $variant => $pair ) {
 519                         if ( !$this->validateVariant( $variant ) ) {
 520                                 continue;
 521                         }
 522
 523                         if ( $action == 'add' ) {
 524                                 foreach ( $pair as $from => $to ) {
 525                                         // to ensure that $from and $to not be left blank
 526                                         // so $this->translate() could always return a string
 527                                         if ( $from || $to ) {
 528                                                 // more efficient than array_merge(), about 2.5 times.
 529                                                 $this->mTables[$variant]->setPair( $from, $to );
 530                                         }
 531                                 }
 532                         } elseif ( $action == 'remove' ) {
 533                                 $this->mTables[$variant]->removeArray( $pair );
 534                         }
 535                 }
 536         }
 537
 538         /**
 539          * Auto convert a Title object to a readable string in the
 540          * preferred variant.
 541          *
 542          * @param $title Title a object of Title
 543          * @return String: converted title text
 544          */
 545         public function convertTitle( $title ) {
 546                 $variant = $this->getPreferredVariant();
 547                 $index = $title->getNamespace();
 548                 if ( $index === NS_MAIN ) {
 549                         $text = '';
 550                 } else {
 551                         // first let's check if a message has given us a converted name
 552                         $nsConvMsg = wfMessage( 'conversion-ns' . $index )->inContentLanguage();
 553                         if ( $nsConvMsg->exists() ) {
 554                                 $text = $nsConvMsg->plain();
 555                         } else {
 556                                 // the message does not exist, try retrieve it from the current
 557                                 // variant's namespace names.
 558                                 $langObj = $this->mLangObj->factory( $variant );
 559                                 $text = $langObj->getFormattedNsText( $index );
 560                         }
 561                         $text .= ':';
 562                 }
 563                 $text .= $title->getText();
 564                 $text = $this->translate( $text, $variant );
 565                 return $text;
 566         }
 567
 568         /**
 569          * Convert text to different variants of a language. The automatic
 570          * conversion is done in autoConvert(). Here we parse the text
 571          * marked with -{}-, which specifies special conversions of the
 572          * text that can not be accomplished in autoConvert().
 573          *
 574          * Syntax of the markup:
 575          * -{code1:text1;code2:text2;...}-  or
 576          * -{flags|code1:text1;code2:text2;...}-  or
 577          * -{text}- in which case no conversion should take place for text
 578          *
 579          * @param $text String: text to be converted
 580          * @return String: converted text
 581          */
 582         public function convert( $text ) {
 583                 $variant = $this->getPreferredVariant();
 584                 return $this->convertTo( $text, $variant );
 585         }
 586
 587         /**
 588          * Same as convert() except a extra parameter to custom variant.
 589          *
 590          * @param $text String: text to be converted
 591          * @param $variant String: the target variant code
 592          * @return String: converted text
 593          */
 594         public function convertTo( $text, $variant ) {
 595                 global $wgDisableLangConversion;
 596                 if ( $wgDisableLangConversion ) {
 597                         return $text;
 598                 }
 599                 return $this->recursiveConvertTopLevel( $text, $variant );
 600         }
 601
 602         /**
 603          * Recursively convert text on the outside. Allow to use nested
 604          * markups to custom rules.
 605          *
 606          * @param $text String: text to be converted
 607          * @param $variant String: the target variant code
 608          * @param $depth Integer: depth of recursion
 609          * @return String: converted text
 610          */
 611         protected function recursiveConvertTopLevel( $text, $variant, $depth = 0 ) {
 612                 $startPos = 0;
 613                 $out = '';
 614                 $length = strlen( $text );
 615                 $shouldConvert = !$this->guessVariant( $text, $variant );
 616
 617                 while ( $startPos < $length ) {
 618                         $pos = strpos( $text, '-{', $startPos );
 619
 620                         if ( $pos === false ) {
 621                                 // No more markup, append final segment
 622                                 $fragment = substr( $text, $startPos );
 623                                 $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 624                                 return $out;
 625                         }
 626
 627                         // Markup found
 628                         // Append initial segment
 629                         $fragment = substr( $text, $startPos, $pos - $startPos );
 630                         $out .= $shouldConvert? $this->autoConvert( $fragment, $variant ): $fragment;
 631
 632                         // Advance position
 633                         $startPos = $pos;
 634
 635                         // Do recursive conversion
 636                         $out .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 637                 }
 638
 639                 return $out;
 640         }
 641
 642         /**
 643          * Recursively convert text on the inside.
 644          *
 645          * @param $text String: text to be converted
 646          * @param $variant String: the target variant code
 647          * @param $startPos int
 648          * @param $depth Integer: depth of recursion
 649          *
 650          * @throws MWException
 651          * @return String: converted text
 652          */
 653         protected function recursiveConvertRule( $text, $variant, &$startPos, $depth = 0 ) {
 654                 // Quick sanity check (no function calls)
 655                 if ( $text[$startPos] !== '-' || $text[$startPos + 1] !== '{' ) {
 656                         throw new MWException( __METHOD__ . ': invalid input string' );
 657                 }
 658
 659                 $startPos += 2;
 660                 $inner = '';
 661                 $warningDone = false;
 662                 $length = strlen( $text );
 663
 664                 while ( $startPos < $length ) {
 665                         $m = false;
 666                         preg_match( '/-\{|\}-/', $text, $m,  PREG_OFFSET_CAPTURE, $startPos );
 667                         if ( !$m ) {
 668                                 // Unclosed rule
 669                                 break;
 670                         }
 671
 672                         $token = $m[0][0];
 673                         $pos = $m[0][1];
 674
 675                         // Markup found
 676                         // Append initial segment
 677                         $inner .= substr( $text, $startPos, $pos - $startPos );
 678
 679                         // Advance position
 680                         $startPos = $pos;
 681
 682                         switch ( $token ) {
 683                                 case '-{':
 684                                         // Check max depth
 685                                         if ( $depth >= $this->mMaxDepth ) {
 686                                                 $inner .= '-{';
 687                                                 if ( !$warningDone ) {
 688                                                         $inner .= '<span class="error">' .
 689                                                                 wfMsgForContent( 'language-converter-depth-warning',
 690                                                                         $this->mMaxDepth ) .
 691                                                                 '</span>';
 692                                                         $warningDone = true;
 693                                                 }
 694                                                 $startPos += 2;
 695                                                 continue;
 696                                         }
 697                                         // Recursively parse another rule
 698                                         $inner .= $this->recursiveConvertRule( $text, $variant, $startPos, $depth + 1 );
 699                                         break;
 700                                 case '}-':
 701                                         // Apply the rule
 702                                         $startPos += 2;
 703                                         $rule = new ConverterRule( $inner, $this );
 704                                         $rule->parse( $variant );
 705                                         $this->applyManualConv( $rule );
 706                                         return $rule->getDisplay();
 707                                 default:
 708                                         throw new MWException( __METHOD__ . ': invalid regex match' );
 709                         }
 710                 }
 711
 712                 // Unclosed rule
 713                 if ( $startPos < $length ) {
 714                         $inner .= substr( $text, $startPos );
 715                 }
 716                 $startPos = $length;
 717                 return '-{' . $this->autoConvert( $inner, $variant );
 718         }
 719
 720         /**
 721          * If a language supports multiple variants, it is possible that
 722          * non-existing link in one variant actually exists in another variant.
 723          * This function tries to find it. See e.g. LanguageZh.php
 724          *
 725          * @param $link String: the name of the link
 726          * @param $nt Mixed: the title object of the link
 727          * @param $ignoreOtherCond Boolean: to disable other conditions when
 728          *              we need to transclude a template or update a category's link
 729          * @return Null, the input parameters may be modified upon return
 730          */
 731         public function findVariantLink( &$link, &$nt, $ignoreOtherCond = false ) {
 732                 # If the article has already existed, there is no need to
 733                 # check it again, otherwise it may cause a fault.
 734                 if ( is_object( $nt ) && $nt->exists() ) {
 735                         return;
 736                 }
 737
 738                 global $wgDisableLangConversion, $wgDisableTitleConversion, $wgRequest,
 739                         $wgUser;
 740                 $isredir = $wgRequest->getText( 'redirect', 'yes' );
 741                 $action = $wgRequest->getText( 'action' );
 742                 $linkconvert = $wgRequest->getText( 'linkconvert', 'yes' );
 743                 $disableLinkConversion = $wgDisableLangConversion
 744                         || $wgDisableTitleConversion;
 745                 $linkBatch = new LinkBatch();
 746
 747                 $ns = NS_MAIN;
 748
 749                 if ( $disableLinkConversion ||
 750                          ( !$ignoreOtherCond &&
 751                            ( $isredir == 'no'
 752                                  || $action == 'edit'
 753                                  || $action == 'submit'
 754                                  || $linkconvert == 'no'
 755                                  || $wgUser->getOption( 'noconvertlink' ) == 1 ) ) ) {
 756                         return;
 757                 }
 758
 759                 if ( is_object( $nt ) ) {
 760                         $ns = $nt->getNamespace();
 761                 }
 762
 763                 $variants = $this->autoConvertToAllVariants( $link );
 764                 if ( !$variants ) { // give up
 765                         return;
 766                 }
 767
 768                 $titles = array();
 769
 770                 foreach ( $variants as $v ) {
 771                         if ( $v != $link ) {
 772                                 $varnt = Title::newFromText( $v, $ns );
 773                                 if ( !is_null( $varnt ) ) {
 774                                         $linkBatch->addObj( $varnt );
 775                                         $titles[] = $varnt;
 776                                 }
 777                         }
 778                 }
 779
 780                 // fetch all variants in single query
 781                 $linkBatch->execute();
 782
 783                 foreach ( $titles as $varnt ) {
 784                         if ( $varnt->getArticleID() > 0 ) {
 785                                 $nt = $varnt;
 786                                 $link = $varnt->getText();
 787                                 break;
 788                         }
 789                 }
 790         }
 791
 792         /**
 793          * Returns language specific hash options.
 794          *
 795          * @return string
 796          */
 797         public function getExtraHashOptions() {
 798                 $variant = $this->getPreferredVariant();
 799                 return '!' . $variant;
 800         }
 801
 802         /**
 803          * Guess if a text is written in a variant. This should be implemented in subclasses.
 804          *
 805          * @param string        $text the text to be checked
 806          * @param string        $variant language code of the variant to be checked for
 807          * @return bool true if $text appears to be written in $variant, false if not
 808          *
 809          * @author Nikola Smolenski <smolensk@eunet.rs>
 810          * @since 1.19
 811          */
 812         public function guessVariant($text, $variant) {
 813                 return false;
 814         }
 815
 816         /**
 817          * Load default conversion tables.
 818          * This method must be implemented in derived class.
 819          *
 820          * @private
 821          * @throws MWException
 822          */
 823         function loadDefaultTables() {
 824                 $name = get_class( $this );
 825                 throw new MWException( "Must implement loadDefaultTables() method in class $name" );
 826         }
 827
 828         /**
 829          * Load conversion tables either from the cache or the disk.
 830          * @private
 831          * @param $fromCache Boolean: load from memcached? Defaults to true.
 832          */
 833         function loadTables( $fromCache = true ) {
 834                 global $wgLangConvMemc;
 835
 836                 if ( $this->mTablesLoaded ) {
 837                         return;
 838                 }
 839
 840                 wfProfileIn( __METHOD__ );
 841                 $this->mTablesLoaded = true;
 842                 $this->mTables = false;
 843                 if ( $fromCache ) {
 844                         wfProfileIn( __METHOD__ . '-cache' );
 845                         $this->mTables = $wgLangConvMemc->get( $this->mCacheKey );
 846                         wfProfileOut( __METHOD__ . '-cache' );
 847                 }
 848                 if ( !$this->mTables
 849                          || !array_key_exists( self::CACHE_VERSION_KEY, $this->mTables ) ) {
 850                         wfProfileIn( __METHOD__ . '-recache' );
 851                         // not in cache, or we need a fresh reload.
 852                         // We will first load the default tables
 853                         // then update them using things in MediaWiki:Conversiontable/*
 854                         $this->loadDefaultTables();
 855                         foreach ( $this->mVariants as $var ) {
 856                                 $cached = $this->parseCachedTable( $var );
 857                                 $this->mTables[$var]->mergeArray( $cached );
 858                         }
 859
 860                         $this->postLoadTables();
 861                         $this->mTables[self::CACHE_VERSION_KEY] = true;
 862
 863                         $wgLangConvMemc->set( $this->mCacheKey, $this->mTables, 43200 );
 864                         wfProfileOut( __METHOD__ . '-recache' );
 865                 }
 866                 wfProfileOut( __METHOD__ );
 867         }
 868
 869         /**
 870          * Hook for post processing after conversion tables are loaded.
 871          */
 872         function postLoadTables() { }
 873
 874         /**
 875          * Reload the conversion tables.
 876          *
 877          * @private
 878          */
 879         function reloadTables() {
 880                 if ( $this->mTables ) {
 881                         unset( $this->mTables );
 882                 }
 883                 $this->mTablesLoaded = false;
 884                 $this->loadTables( false );
 885         }
 886
 887         /**
 888          * Parse the conversion table stored in the cache.
 889          *
 890          * The tables should be in blocks of the following form:
 891          *              -{
 892          *                      word => word ;
 893          *                      word => word ;
 894          *                      ...
 895          *              }-
 896          *
 897          * To make the tables more manageable, subpages are allowed
 898          * and will be parsed recursively if $recursive == true.
 899          *
 900          * @param $code String: language code
 901          * @param $subpage String: subpage name
 902          * @param $recursive Boolean: parse subpages recursively? Defaults to true.
 903          *
 904          * @return array
 905          */
 906         function parseCachedTable( $code, $subpage = '', $recursive = true ) {
 907                 static $parsed = array();
 908
 909                 $key = 'Conversiontable/' . $code;
 910                 if ( $subpage ) {
 911                         $key .= '/' . $subpage;
 912                 }
 913                 if ( array_key_exists( $key, $parsed ) ) {
 914                         return array();
 915                 }
 916
 917                 $parsed[$key] = true;
 918
 919                 if ( $subpage === '' ) {
 920                         $txt = MessageCache::singleton()->get( 'conversiontable', true, $code );
 921                 } else {
 922                         $txt = false;
 923                         $title = Title::makeTitleSafe( NS_MEDIAWIKI, $key );
 924                         if ( $title && $title->exists() ) {
 925                                 $revision = Revision::newFromTitle( $title );
 926                                 if ( $revision ) {
 927                                         $txt = $revision->getRawText();
 928                                 }
 929                         }
 930                 }
 931
 932                 # Nothing to parse if there's no text
 933                 if ( $txt === false || $txt === null || $txt === '' ) {
 934                         return array();
 935                 }
 936
 937                 // get all subpage links of the form
 938                 // [[MediaWiki:Conversiontable/zh-xx/...|...]]
 939                 $linkhead = $this->mLangObj->getNsText( NS_MEDIAWIKI ) .
 940                         ':Conversiontable';
 941                 $subs = StringUtils::explode( '[[', $txt );
 942                 $sublinks = array();
 943                 foreach ( $subs as $sub ) {
 944                         $link = explode( ']]', $sub, 2 );
 945                         if ( count( $link ) != 2 ) {
 946                                 continue;
 947                         }
 948                         $b = explode( '|', $link[0], 2 );
 949                         $b = explode( '/', trim( $b[0] ), 3 );
 950                         if ( count( $b ) == 3 ) {
 951                                 $sublink = $b[2];
 952                         } else {
 953                                 $sublink = '';
 954                         }
 955
 956                         if ( $b[0] == $linkhead && $b[1] == $code ) {
 957                                 $sublinks[] = $sublink;
 958                         }
 959                 }
 960
 961                 // parse the mappings in this page
 962                 $blocks = StringUtils::explode( '-{', $txt );
 963                 $ret = array();
 964                 $first = true;
 965                 foreach ( $blocks as $block ) {
 966                         if ( $first ) {
 967                                 // Skip the part before the first -{
 968                                 $first = false;
 969                                 continue;
 970                         }
 971                         $mappings = explode( '}-', $block, 2 );
 972                         $stripped = str_replace( array( "'", '"', '*', '#' ), '',
 973                                                                          $mappings[0] );
 974                         $table = StringUtils::explode( ';', $stripped );
 975                         foreach ( $table as $t ) {
 976                                 $m = explode( '=>', $t, 3 );
 977                                 if ( count( $m ) != 2 ) {
 978                                         continue;
 979                                 }
 980                                 // trim any trailling comments starting with '//'
 981                                 $tt = explode( '//', $m[1], 2 );
 982                                 $ret[trim( $m[0] )] = trim( $tt[0] );
 983                         }
 984                 }
 985
 986                 // recursively parse the subpages
 987                 if ( $recursive ) {
 988                         foreach ( $sublinks as $link ) {
 989                                 $s = $this->parseCachedTable( $code, $link, $recursive );
 990                                 $ret = array_merge( $ret, $s );
 991                         }
 992                 }
 993
 994                 if ( $this->mUcfirst ) {
 995                         foreach ( $ret as $k => $v ) {
 996                                 $ret[$this->mLangObj->ucfirst( $k )] = $this->mLangObj->ucfirst( $v );
 997                         }
 998                 }
 999                 return $ret;
1000         }
1001
1002         /**
1003          * Enclose a string with the "no conversion" tag. This is used by
1004          * various functions in the Parser.
1005          *
1006          * @param $text String: text to be tagged for no conversion
1007          * @param $noParse Boolean: unused
1008          * @return String: the tagged text
1009          */
1010         public function markNoConversion( $text, $noParse = false ) {
1011                 # don't mark if already marked
1012                 if ( strpos( $text, '-{' ) || strpos( $text, '}-' ) ) {
1013                         return $text;
1014                 }
1015
1016                 $ret = "-{R|$text}-";
1017                 return $ret;
1018         }
1019
1020         /**
1021          * Convert the sorting key for category links. This should make different
1022          * keys that are variants of each other map to the same key.
1023          *
1024          * @param $key string
1025          *
1026          * @return string
1027          */
1028         function convertCategoryKey( $key ) {
1029                 return $key;
1030         }
1031
1032         /**
1033          * Hook to refresh the cache of conversion tables when
1034          * MediaWiki:Conversiontable* is updated.
1035          * @private
1036          *
1037          * @param $article Article object
1038          * @param $user Object: User object for the current user
1039          * @param $text String: article text (?)
1040          * @param $summary String: edit summary of the edit
1041          * @param $isMinor Boolean: was the edit marked as minor?
1042          * @param $isWatch Boolean: did the user watch this page or not?
1043          * @param $section
1044          * @param $flags int Bitfield
1045          * @param $revision Object: new Revision object or null
1046          * @return Boolean: true
1047          */
1048         function OnArticleSaveComplete( $article, $user, $text, $summary, $isMinor,
1049                         $isWatch, $section, $flags, $revision ) {
1050                 $titleobj = $article->getTitle();
1051                 if ( $titleobj->getNamespace() == NS_MEDIAWIKI ) {
1052                         $title = $titleobj->getDBkey();
1053                         $t = explode( '/', $title, 3 );
1054                         $c = count( $t );
1055                         if ( $c > 1 && $t[0] == 'Conversiontable' ) {
1056                                 if ( $this->validateVariant( $t[1] ) ) {
1057                                         $this->reloadTables();
1058                                 }
1059                         }
1060                 }
1061                 return true;
1062         }
1063
1064         /**
1065          * Armour rendered math against conversion.
1066          * Escape special chars in parsed math text. (in most cases are img elements)
1067          *
1068          * @param $text String: text to armour against conversion
1069          * @return String: armoured text where { and } have been converted to
1070          *                 &#123; and &#125;
1071          */
1072         public function armourMath( $text ) {
1073                 // convert '-{' and '}-' to '-&#123;' and '&#125;-' to prevent
1074                 // any unwanted markup appearing in the math image tag.
1075                 $text = strtr( $text, array( '-{' => '-&#123;', '}-' => '&#125;-' ) );
1076                 return $text;
1077         }
1078
1079         /**
1080          * Get the cached separator pattern for ConverterRule::parseRules()
1081          */
1082         function getVarSeparatorPattern() {
1083                 if ( is_null( $this->mVarSeparatorPattern ) ) {
1084                         // varsep_pattern for preg_split:
1085                         // text should be splited by ";" only if a valid variant
1086                         // name exist after the markup, for example:
1087                         //  -{zh-hans:<span style="font-size:120%;">xxx</span>;zh-hant:\
1088                         //      <span style="font-size:120%;">yyy</span>;}-
1089                         // we should split it as:
1090                         //  array(
1091                         //        [0] => 'zh-hans:<span style="font-size:120%;">xxx</span>'
1092                         //        [1] => 'zh-hant:<span style="font-size:120%;">yyy</span>'
1093                         //        [2] => ''
1094                         //       )
1095                         $pat = '/;\s*(?=';
1096                         foreach ( $this->mVariants as $variant ) {
1097                                 // zh-hans:xxx;zh-hant:yyy
1098                                 $pat .= $variant . '\s*:|';
1099                                 // xxx=>zh-hans:yyy; xxx=>zh-hant:zzz
1100                                 $pat .= '[^;]*?=>\s*' . $variant . '\s*:|';
1101                         }
1102                         $pat .= '\s*$)/';
1103                         $this->mVarSeparatorPattern = $pat;
1104                 }
1105                 return $this->mVarSeparatorPattern;
1106         }
1107 }
1108
1109 /**
1110  * Parser for rules of language conversion , parse rules in -{ }- tag.
1111  * @ingroup Language
1112  * @author fdcn <fdcn64@gmail.com>, PhiLiP <philip.npc@gmail.com>
1113  */
1114 class ConverterRule {
1115         var $mText; // original text in -{text}-
1116         var $mConverter; // LanguageConverter object
1117         var $mRuleDisplay = '';
1118         var $mRuleTitle = false;
1119         var $mRules = '';// string : the text of the rules
1120         var $mRulesAction = 'none';
1121         var $mFlags = array();
1122         var $mVariantFlags = array();
1123         var $mConvTable = array();
1124         var $mBidtable = array();// array of the translation in each variant
1125         var $mUnidtable = array();// array of the translation in each variant
1126
1127         /**
1128          * Constructor
1129          *
1130          * @param $text String: the text between -{ and }-
1131          * @param $converter LanguageConverter object
1132          */
1133         public function __construct( $text, $converter ) {
1134                 $this->mText = $text;
1135                 $this->mConverter = $converter;
1136         }
1137
1138         /**
1139          * Check if variants array in convert array.
1140          *
1141          * @param $variants Array or string: variant language code
1142          * @return String: translated text
1143          */
1144         public function getTextInBidtable( $variants ) {
1145                 $variants = (array)$variants;
1146                 if ( !$variants ) {
1147                         return false;
1148                 }
1149                 foreach ( $variants as $variant ) {
1150                         if ( isset( $this->mBidtable[$variant] ) ) {
1151                                 return $this->mBidtable[$variant];
1152                         }
1153                 }
1154                 return false;
1155         }
1156
1157         /**
1158          * Parse flags with syntax -{FLAG| ... }-
1159          * @private
1160          */
1161         function parseFlags() {
1162                 $text = $this->mText;
1163                 $flags = array();
1164                 $variantFlags = array();
1165
1166                 $sepPos = strpos( $text, '|' );
1167                 if ( $sepPos !== false ) {
1168                         $validFlags = $this->mConverter->mFlags;
1169                         $f = StringUtils::explode( ';', substr( $text, 0, $sepPos ) );
1170                         foreach ( $f as $ff ) {
1171                                 $ff = trim( $ff );
1172                                 if ( isset( $validFlags[$ff] ) ) {
1173                                         $flags[$validFlags[$ff]] = true;
1174                                 }
1175                         }
1176                         $text = strval( substr( $text, $sepPos + 1 ) );
1177                 }
1178
1179                 if ( !$flags ) {
1180                         $flags['S'] = true;
1181                 } elseif ( isset( $flags['R'] ) ) {
1182                         $flags = array( 'R' => true );// remove other flags
1183                 } elseif ( isset( $flags['N'] ) ) {
1184                         $flags = array( 'N' => true );// remove other flags
1185                 } elseif ( isset( $flags['-'] ) ) {
1186                         $flags = array( '-' => true );// remove other flags
1187                 } elseif ( count( $flags ) == 1 && isset( $flags['T'] ) ) {
1188                         $flags['H'] = true;
1189                 } elseif ( isset( $flags['H'] ) ) {
1190                         // replace A flag, and remove other flags except T
1191                         $temp = array( '+' => true, 'H' => true );
1192                         if ( isset( $flags['T'] ) ) {
1193                                 $temp['T'] = true;
1194                         }
1195                         if ( isset( $flags['D'] ) ) {
1196                                 $temp['D'] = true;
1197                         }
1198                         $flags = $temp;
1199                 } else {
1200                         if ( isset( $flags['A'] ) ) {
1201                                 $flags['+'] = true;
1202                                 $flags['S'] = true;
1203                         }
1204                         if ( isset( $flags['D'] ) ) {
1205                                 unset( $flags['S'] );
1206                         }
1207                         // try to find flags like "zh-hans", "zh-hant"
1208                         // allow syntaxes like "-{zh-hans;zh-hant|XXXX}-"
1209                         $variantFlags = array_intersect( array_keys( $flags ), $this->mConverter->mVariants );
1210                         if ( $variantFlags ) {
1211                                 $variantFlags = array_flip( $variantFlags );
1212                                 $flags = array();
1213                         }
1214                 }
1215                 $this->mVariantFlags = $variantFlags;
1216                 $this->mRules = $text;
1217                 $this->mFlags = $flags;
1218         }
1219
1220         /**
1221          * Generate conversion table.
1222          * @private
1223          */
1224         function parseRules() {
1225                 $rules = $this->mRules;
1226                 $bidtable = array();
1227                 $unidtable = array();
1228                 $variants = $this->mConverter->mVariants;
1229                 $varsep_pattern = $this->mConverter->getVarSeparatorPattern();
1230
1231                 $choice = preg_split( $varsep_pattern, $rules );
1232
1233                 foreach ( $choice as $c ) {
1234                         $v  = explode( ':', $c, 2 );
1235                         if ( count( $v ) != 2 ) {
1236                                 // syntax error, skip
1237                                 continue;
1238                         }
1239                         $to = trim( $v[1] );
1240                         $v  = trim( $v[0] );
1241                         $u  = explode( '=>', $v, 2 );
1242                         // if $to is empty, strtr() could return a wrong result
1243                         if ( count( $u ) == 1 && $to && in_array( $v, $variants ) ) {
1244                                 $bidtable[$v] = $to;
1245                         } elseif ( count( $u ) == 2 ) {
1246                                 $from = trim( $u[0] );
1247                                 $v      = trim( $u[1] );
1248                                 if ( array_key_exists( $v, $unidtable )
1249                                          && !is_array( $unidtable[$v] )
1250                                          && $to
1251                                          && in_array( $v, $variants ) ) {
1252                                         $unidtable[$v] = array( $from => $to );
1253                                 } elseif ( $to && in_array( $v, $variants ) ) {
1254                                         $unidtable[$v][$from] = $to;
1255                                 }
1256                         }
1257                         // syntax error, pass
1258                         if ( !isset( $this->mConverter->mVariantNames[$v] ) ) {
1259                                 $bidtable = array();
1260                                 $unidtable = array();
1261                                 break;
1262                         }
1263                 }
1264                 $this->mBidtable = $bidtable;
1265                 $this->mUnidtable = $unidtable;
1266         }
1267
1268         /**
1269          * @private
1270          *
1271          * @return string
1272          */
1273         function getRulesDesc() {
1274                 $codesep = $this->mConverter->mDescCodeSep;
1275                 $varsep = $this->mConverter->mDescVarSep;
1276                 $text = '';
1277                 foreach ( $this->mBidtable as $k => $v ) {
1278                         $text .= $this->mConverter->mVariantNames[$k] . "$codesep$v$varsep";
1279                 }
1280                 foreach ( $this->mUnidtable as $k => $a ) {
1281                         foreach ( $a as $from => $to ) {
1282                                 $text .= $from . '⇒' . $this->mConverter->mVariantNames[$k] .
1283                                         "$codesep$to$varsep";
1284                         }
1285                 }
1286                 return $text;
1287         }
1288
1289         /**
1290          * Parse rules conversion.
1291          * @private
1292          *
1293          * @param $variant
1294          *
1295          * @return string
1296          */
1297         function getRuleConvertedStr( $variant ) {
1298                 $bidtable = $this->mBidtable;
1299                 $unidtable = $this->mUnidtable;
1300
1301                 if ( count( $bidtable ) + count( $unidtable ) == 0 ) {
1302                         return $this->mRules;
1303                 } else {
1304                         // display current variant in bidirectional array
1305                         $disp = $this->getTextInBidtable( $variant );
1306                         // or display current variant in fallbacks
1307                         if ( !$disp ) {
1308                                 $disp = $this->getTextInBidtable(
1309                                                 $this->mConverter->getVariantFallbacks( $variant ) );
1310                         }
1311                         // or display current variant in unidirectional array
1312                         if ( !$disp && array_key_exists( $variant, $unidtable ) ) {
1313                                 $disp = array_values( $unidtable[$variant] );
1314                                 $disp = $disp[0];
1315                         }
1316                         // or display frist text under disable manual convert
1317                         if ( !$disp
1318                                  && $this->mConverter->mManualLevel[$variant] == 'disable' ) {
1319                                 if ( count( $bidtable ) > 0 ) {
1320                                         $disp = array_values( $bidtable );
1321                                         $disp = $disp[0];
1322                                 } else {
1323                                         $disp = array_values( $unidtable );
1324                                         $disp = array_values( $disp[0] );
1325                                         $disp = $disp[0];
1326                                 }
1327                         }
1328                         return $disp;
1329                 }
1330         }
1331
1332         /**
1333          * Generate conversion table for all text.
1334          * @private
1335          */
1336         function generateConvTable() {
1337                 // Special case optimisation
1338                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1339                         $this->mConvTable = array();
1340                         return;
1341                 }
1342
1343                 $bidtable = $this->mBidtable;
1344                 $unidtable = $this->mUnidtable;
1345                 $manLevel = $this->mConverter->mManualLevel;
1346
1347                 $vmarked = array();
1348                 foreach ( $this->mConverter->mVariants as $v ) {
1349                         /* for bidirectional array
1350                                 fill in the missing variants, if any,
1351                                 with fallbacks */
1352                         if ( !isset( $bidtable[$v] ) ) {
1353                                 $variantFallbacks =
1354                                         $this->mConverter->getVariantFallbacks( $v );
1355                                 $vf = $this->getTextInBidtable( $variantFallbacks );
1356                                 if ( $vf ) {
1357                                         $bidtable[$v] = $vf;
1358                                 }
1359                         }
1360
1361                         if ( isset( $bidtable[$v] ) ) {
1362                                 foreach ( $vmarked as $vo ) {
1363                                         // use syntax: -{A|zh:WordZh;zh-tw:WordTw}-
1364                                         // or -{H|zh:WordZh;zh-tw:WordTw}-
1365                                         // or -{-|zh:WordZh;zh-tw:WordTw}-
1366                                         // to introduce a custom mapping between
1367                                         // words WordZh and WordTw in the whole text
1368                                         if ( $manLevel[$v] == 'bidirectional' ) {
1369                                                 $this->mConvTable[$v][$bidtable[$vo]] = $bidtable[$v];
1370                                         }
1371                                         if ( $manLevel[$vo] == 'bidirectional' ) {
1372                                                 $this->mConvTable[$vo][$bidtable[$v]] = $bidtable[$vo];
1373                                         }
1374                                 }
1375                                 $vmarked[] = $v;
1376                         }
1377                         /* for unidirectional array fill to convert tables */
1378                         if ( ( $manLevel[$v] == 'bidirectional' || $manLevel[$v] == 'unidirectional' )
1379                                 && isset( $unidtable[$v] ) )
1380                         {
1381                                 if ( isset( $this->mConvTable[$v] ) ) {
1382                                         $this->mConvTable[$v] = array_merge( $this->mConvTable[$v], $unidtable[$v] );
1383                                 } else {
1384                                         $this->mConvTable[$v] = $unidtable[$v];
1385                                 }
1386                         }
1387                 }
1388         }
1389
1390         /**
1391          * Parse rules and flags.
1392          * @param $variant String: variant language code
1393          */
1394         public function parse( $variant = null ) {
1395                 if ( !$variant ) {
1396                         $variant = $this->mConverter->getPreferredVariant();
1397                 }
1398
1399                 $this->parseFlags();
1400                 $flags = $this->mFlags;
1401
1402                 // convert to specified variant
1403                 // syntax: -{zh-hans;zh-hant[;...]|<text to convert>}-
1404                 if ( $this->mVariantFlags ) {
1405                         // check if current variant in flags
1406                         if ( isset( $this->mVariantFlags[$variant] ) ) {
1407                                 // then convert <text to convert> to current language
1408                                 $this->mRules = $this->mConverter->autoConvert( $this->mRules,
1409                                         $variant );
1410                         } else { // if current variant no in flags,
1411                                    // then we check its fallback variants.
1412                                 $variantFallbacks =
1413                                         $this->mConverter->getVariantFallbacks( $variant );
1414                                 if( is_array( $variantFallbacks ) ) {
1415                                         foreach ( $variantFallbacks as $variantFallback ) {
1416                                                 // if current variant's fallback exist in flags
1417                                                 if ( isset( $this->mVariantFlags[$variantFallback] ) ) {
1418                                                         // then convert <text to convert> to fallback language
1419                                                         $this->mRules =
1420                                                                 $this->mConverter->autoConvert( $this->mRules,
1421                                                                         $variantFallback );
1422                                                         break;
1423                                                 }
1424                                         }
1425                                 }
1426                         }
1427                         $this->mFlags = $flags = array( 'R' => true );
1428                 }
1429
1430                 if ( !isset( $flags['R'] ) && !isset( $flags['N'] ) ) {
1431                         // decode => HTML entities modified by Sanitizer::removeHTMLtags
1432                         $this->mRules = str_replace( '=&gt;', '=>', $this->mRules );
1433                         $this->parseRules();
1434                 }
1435                 $rules = $this->mRules;
1436
1437                 if ( !$this->mBidtable && !$this->mUnidtable ) {
1438                         if ( isset( $flags['+'] ) || isset( $flags['-'] ) ) {
1439                                 // fill all variants if text in -{A/H/-|text} without rules
1440                                 foreach ( $this->mConverter->mVariants as $v ) {
1441                                         $this->mBidtable[$v] = $rules;
1442                                 }
1443                         } elseif ( !isset( $flags['N'] ) && !isset( $flags['T'] ) ) {
1444                                 $this->mFlags = $flags = array( 'R' => true );
1445                         }
1446                 }
1447
1448                 $this->mRuleDisplay = false;
1449                 foreach ( $flags as $flag => $unused ) {
1450                         switch ( $flag ) {
1451                                 case 'R':
1452                                         // if we don't do content convert, still strip the -{}- tags
1453                                         $this->mRuleDisplay = $rules;
1454                                         break;
1455                                 case 'N':
1456                                         // process N flag: output current variant name
1457                                         $ruleVar = trim( $rules );
1458                                         if ( isset( $this->mConverter->mVariantNames[$ruleVar] ) ) {
1459                                                 $this->mRuleDisplay = $this->mConverter->mVariantNames[$ruleVar];
1460                                         } else {
1461                                                 $this->mRuleDisplay = '';
1462                                         }
1463                                         break;
1464                                 case 'D':
1465                                         // process D flag: output rules description
1466                                         $this->mRuleDisplay = $this->getRulesDesc();
1467                                         break;
1468                                 case 'H':
1469                                         // process H,- flag or T only: output nothing
1470                                         $this->mRuleDisplay = '';
1471                                         break;
1472                                 case '-':
1473                                         $this->mRulesAction = 'remove';
1474                                         $this->mRuleDisplay = '';
1475                                         break;
1476                                 case '+':
1477                                         $this->mRulesAction = 'add';
1478                                         $this->mRuleDisplay = '';
1479                                         break;
1480                                 case 'S':
1481                                         $this->mRuleDisplay = $this->getRuleConvertedStr( $variant );
1482                                         break;
1483                                 case 'T':
1484                                         $this->mRuleTitle = $this->getRuleConvertedStr( $variant );
1485                                         $this->mRuleDisplay = '';
1486                                         break;
1487                                 default:
1488                                         // ignore unknown flags (but see error case below)
1489                         }
1490                 }
1491                 if ( $this->mRuleDisplay === false ) {
1492                         $this->mRuleDisplay = '<span class="error">'
1493                                 . wfMessage( 'converter-manual-rule-error' )->inContentLanguage()->escaped()
1494                                 . '</span>';
1495                 }
1496
1497                 $this->generateConvTable();
1498         }
1499
1500         /**
1501          * @todo FIXME: code this function :)
1502          */
1503         public function hasRules() {
1504                 // TODO:
1505         }
1506
1507         /**
1508          * Get display text on markup -{...}-
1509          * @return string
1510          */
1511         public function getDisplay() {
1512                 return $this->mRuleDisplay;
1513         }
1514
1515         /**
1516          * Get converted title.
1517          * @return string
1518          */
1519         public function getTitle() {
1520                 return $this->mRuleTitle;
1521         }
1522
1523         /**
1524          * Return how deal with conversion rules.
1525          * @return string
1526          */
1527         public function getRulesAction() {
1528                 return $this->mRulesAction;
1529         }
1530
1531         /**
1532          * Get conversion table. (bidirectional and unidirectional
1533          * conversion table)
1534          * @return array
1535          */
1536         public function getConvTable() {
1537                 return $this->mConvTable;
1538         }
1539
1540         /**
1541          * Get conversion rules string.
1542          * @return string
1543          */
1544         public function getRules() {
1545                 return $this->mRules;
1546         }
1547
1548         /**
1549          * Get conversion flags.
1550          * @return array
1551          */
1552         public function getFlags() {
1553                 return $this->mFlags;
1554         }
1555 }